From 8160a375038d5c590033c84c81a769ae0057a9ff Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Wed, 14 Oct 2026 22:56:28 +0000
Subject: [PATCH 01/81] Slack replies without a bot token say so; user text is
 escaped; booleans take yes/no; graph=False means no charts
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- Slack: link previews are turned off in the request body, where Slack reads unfurl_links/unfurl_media (they went in the headers
  and never worked). A reply without slackBotToken logs "ERROR slack: a reply to channel … needs slackBotToken", slack_api raises
  the same instead of a TypeError, and the bot warns once at startup when Slack commands are configured without the token.
- Text a user wrote is shown as text: a .plan entry with a < made Telegram refuse every .plan reply after it, and a first name
  is escaped where it arrives. Slack's already escaped text is not escaped twice.
- A true/false setting takes true/false, yes/no, on/off or 1/0; anything else is a config error (scheduler=yes and graph=Ture
  used to mean False). A blank value means the default, as the README says of finbot.env.
- graph=False turns off every chart, not only the price lines.
- performance names each portfolio from the portfolio list: one with no holdings crashed the report.
- Yahoo: the chart request sends the cookie of its crumb; cookie and crumb are fetched together under the lock; price_history no
  longer writes its PNG cache with the cache off.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 README.md               |  4 +--
 bot.py                  |  9 ++++--
 lib/config.py           | 33 +++++++++++++++++----
 lib/webhook.py          | 25 +++++++++++-----
 lib/worker.py           |  6 +++-
 lib/yahoo.py            | 20 ++++++++-----
 performance.py          |  9 +++---
 tests/test_config.py    | 19 ++++++++++++
 tests/test_edges.py     | 26 +++++++++++++++++
 tests/test_providers.py | 65 +++++++++++++++++++++++++++++++++++++++++
 tests/test_reports.py   |  6 ++++
 11 files changed, 191 insertions(+), 31 deletions(-)

diff --git a/README.md b/README.md
index 978cbb2a..614b376f 100644
--- a/README.md
+++ b/README.md
@@ -480,7 +480,7 @@ Copies preferences.conf and metadata files into the data directory (never overwr
 util/sync_defaults.py [-d DEST] --apply
 ```
 
-All 26 keys, read by the bot itself. A change takes effect on a reload.
+All 26 keys, read by the bot itself. A change takes effect on a reload. A true/false preference takes `true`/`false`, `yes`/`no`, `on`/`off` or `1`/`0`; anything else is a mistake, reported like any other. A blank value means the default.
 
 | Key | Default | Description |
 |---|---|---|
@@ -716,7 +716,7 @@ Visit https://api.slack.com/apps/ to create a new Slack app from scratch (if you
     * Adding scopes to an app that is already installed takes effect only after _Reinstall to Workspace_.
   * Scroll up to _OAuth Tokens for Your Workspace_:
     * If _Bot User OAuth Token_ is not visible, hit _Install to Workspace > Allow_
-    * Copy _Bot User OAuth Token_ into `finbot.env` as `slackBotToken`. This allows finbot to authenticate with Slack so it can send replies.
+    * Copy _Bot User OAuth Token_ into `finbot.env` as `slackBotToken`. This allows finbot to authenticate with Slack so it can send replies. Without it Slack commands get no reply: the log says so at startup, and again for each reply.
     * Restart `bot.py`
 
 ### Daemonize (systemd)
diff --git a/bot.py b/bot.py
index 0b5cf865..d33dc9a1 100755
--- a/bot.py
+++ b/bot.py
@@ -5,7 +5,7 @@
 	signal.signal(signal.SIGHUP, signal.SIG_IGN) # ignored until startup is done (the reload handler is installed last): its default action kills the bot
 import gevent.monkey
 gevent.monkey.patch_all()
-import json, os, re, sys, traceback
+import html, json, os, re, sys, traceback
 from pathlib import Path
 from gevent import pywsgi
 import threading
@@ -97,11 +97,12 @@ def reply(status, body=b''):
 		sender = msg["from"]
 		message_id = str(msg["message_id"])
 		chat_id = str(msg["chat"]["id"])
+		first_name = html.escape(sender["first_name"], quote=False) # replies are Telegram HTML: a name is text, not markup
 		if "username" in sender:
 			user = '@' + sender["username"]
-			userRealName = sender["first_name"] or user # what to call them: the handle when the first name is blank
+			userRealName = first_name or user # what to call them: the handle when the first name is blank
 		else:
-			user = userRealName = '@' + sender["first_name"]
+			user = userRealName = '@' + first_name
 		if "text" in msg:
 			message = msg["text"]
 			print("[Telegram]:", user, message, file=sys.stderr) if debug else None
@@ -205,6 +206,8 @@ def reload_all(*args):
 		sys.exit(1)
 	for line in config.startup_report():
 		print('config:', line, file=sys.stderr)
+	for line in config.startup_warnings():
+		print('ERROR config:', line, file=sys.stderr)
 	if os.getuid() == 0:
 		print("Running as superuser. This is not recommended.", file=sys.stderr)
 	if config_scheduler: # read before anything else happens: an impossible time in it stops the start
diff --git a/lib/config.py b/lib/config.py
index 45c2f8c1..3b73061d 100644
--- a/lib/config.py
+++ b/lib/config.py
@@ -110,20 +110,34 @@ def read_startup_file(path):
 	return values
 
 
+TRUE_WORDS = ('true', '1', 't', 'yes', 'y', 'on')
+FALSE_WORDS = ('false', '0', 'f', 'no', 'n', 'off')
+
+
 class _Reader:
-	"""Typed reads from a getenv(name, default); a bad value is remembered, the default used, and all problems raised together."""
+	"""Typed reads from a getenv(name, default); a bad value is remembered, the default used, and all problems raised together.
+	A blank value means unset: the default."""
 	def __init__(self, getenv):
 		self.getenv = getenv
 		self.errors = []
 
+	def value(self, name, default):
+		value = self.getenv(name, default)
+		return default if value == '' else value
+
 	def str(self, name, default=None):
-		return self.getenv(name, default)
+		return self.value(name, default)
 
 	def bool(self, name, default):
-		return str(self.getenv(name, default)).lower() in ('true', '1', 't')
+		value = self.value(name, default)
+		word = str(value).strip().lower()
+		if word in TRUE_WORDS or word in FALSE_WORDS:
+			return word in TRUE_WORDS
+		self.errors.append(f"{name}={value!r} is not true or false")
+		return default
 
 	def number(self, kind, name, default):
-		value = self.getenv(name, default)
+		value = self.value(name, default)
 		try:
 			return kind(value)
 		except ValueError:
@@ -327,6 +341,14 @@ def startup_report():
 	return lines
 
 
+def startup_warnings():
+	"""Problems in finbot.env that do not stop the start, a line each: a setting that leaves part of the bot unable to work."""
+	warnings = []
+	if config_slackOutgoingWebhook and not config_slackBotToken:
+		warnings.append('slackOutgoingWebhook is set but slackBotToken is not: Slack commands will get no reply')
+	return warnings
+
+
 adjectives = [
 	'absolutely',
 	'amazingly',
@@ -515,5 +537,6 @@ def startup_report():
 # `from lib.config import *` is how the other modules get their config_* names; keep the loader machinery out of their namespaces
 _INTERNAL = {'difflib', 're', 'sys', 'ConfigError', 'key_problem', 'parse_assignments', 'read_text', 'read_preferences', 'read_startup_file',
 	'startup_values', 'preference_values', 'reload', 'child_environment', 'print_errors', 'check_startup_environment', 'startup_report', 'preferences_path',
-	'STARTUP_KEYS', 'PREFERENCE_KEYS', 'SECRET_KEYS', 'PREFERENCES_FILE', 'PASSED', 'PREFERENCE_NAMES', 'OLD_ENV'}
+	'startup_warnings', 'STARTUP_KEYS', 'PREFERENCE_KEYS', 'SECRET_KEYS', 'PREFERENCES_FILE', 'PASSED', 'PREFERENCE_NAMES', 'OLD_ENV',
+	'TRUE_WORDS', 'FALSE_WORDS'}
 __all__ = [name for name in dir() if not name.startswith('_') and name not in _INTERNAL]
diff --git a/lib/webhook.py b/lib/webhook.py
index a612e596..52d82810 100644
--- a/lib/webhook.py
+++ b/lib/webhook.py
@@ -15,11 +15,11 @@ def write(service, url, payload_string, chat_id=None, message_id=None, reply_mar
 	headers = {'Content-type': 'application/json'}
 	payload = {'text': payload_string}
 	if 'slack.com' in url:
-		headers['unfurl_links'] = 'false'
-		headers['unfurl_media'] = 'false'
+		payload['unfurl_links'] = False # no link previews: Slack reads these from the body, not the headers
+		payload['unfurl_media'] = False
 		if chat_id:
 			if not config_slackBotToken:
-				print("Slack: replying in a channel needs slackBotToken in finbot.env", file=sys.stderr)
+				print(f"ERROR slack: a reply to channel {chat_id} needs slackBotToken in finbot.env; not sent", file=sys.stderr)
 				return None
 			headers['Authorization'] = 'Bearer ' + config_slackBotToken
 			payload['channel'] = chat_id
@@ -154,6 +154,13 @@ def payload_wrapper(service, url, payload, chat_id=None, message_id=None, reply_
 	'strike': {'telegram': ('<s>', '</s>'), 'slack': ('~', '~'), 'discord': ('~~', '~~')},
 }
 
+def escape(text, service):
+	"""Text a user wrote (a .plan entry, a name) made safe to show: Telegram's HTML and Slack's markup both take &amp; &lt; &gt;.
+	It is unescaped first because it may arrive escaped already (Slack escapes what it sends; bot.py escapes Telegram names):
+	escaping twice would show &amp;amp;. Discord's markdown shows entities as they are, so it gets the plain text."""
+	text = html.unescape(str(text))
+	return text if service == 'discord' else html.escape(text, quote=False)
+
 def _styled(style, message, service):
 	if service not in MARKUP[style]:
 		return message
@@ -189,10 +196,10 @@ def visible(text):
 MAYBE_SENT = object() # a read timeout on the request that posts a chart: it may have arrived, so no text copy follows
 
 def takes_charts(service, reply=True):
-	"""Whether a report to `service` can carry a chart image: on Telegram always; on Slack in a reply to a chat (`reply`), uploaded
-	into its channel with the bot token, but not in a scheduled report, which goes through the incoming webhook and takes text only;
-	never on Discord. Elsewhere a report goes out as text."""
-	return service == 'telegram' or (service == 'slack' and reply and bool(config_slackBotToken))
+	"""Whether a report to `service` can carry a chart image: never with the graph preference off; on Telegram always; on Slack in
+	a reply to a chat (`reply`), uploaded into its channel with the bot token, but not in a scheduled report, which goes through the
+	incoming webhook and takes text only; never on Discord. Elsewhere a report goes out as text."""
+	return config_graph and (service == 'telegram' or (service == 'slack' and reply and bool(config_slackBotToken)))
 
 def telegram_photo(chat_id, image_data, caption, message_id=None, reply_markup=None):
 	"""sendPhoto, the report as its caption (trimmed to fit). Telegram's answer, MAYBE_SENT, or None if it was refused."""
@@ -229,6 +236,8 @@ def telegram_photo(chat_id, image_data, caption, message_id=None, reply_markup=N
 
 def slack_api(method, **data):
 	"""POST a Slack Web API method as the bot, form-encoded, and return its answer. Raises RuntimeError with Slack's error if not ok."""
+	if not config_slackBotToken:
+		raise RuntimeError(f"Slack {method}: needs slackBotToken in finbot.env")
 	r = net.post('https://slack.com/api/' + method, headers={'Authorization': 'Bearer ' + config_slackBotToken}, data=data, timeout=config_http_timeout)
 	try:
 		output = r.json()
@@ -270,7 +279,7 @@ def sendPhoto(chat_id, image_data, caption, service, message_id=None, reply_mark
 	with its chart; but not after a read timeout on the request that posts it, when the chart may have arrived after all."""
 	output = None
 	chartable = takes_charts(service, reply=bool(chat_id))
-	if service == 'telegram':
+	if chartable and service == 'telegram':
 		output = telegram_photo(chat_id, image_data, caption, message_id, reply_markup)
 	elif chartable:
 		output = slack_photo(chat_id, image_data, caption, message_id)
diff --git a/lib/worker.py b/lib/worker.py
index 02943c2b..acf456fe 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -350,7 +350,11 @@ def on_plan(request, m):
 	if m.group(1) and not m.group(1).startswith('@'):
 		plan[request.user] = m.group(1)
 		util.json_write('finbot_plan.json', plan, persist=True)
-	request.say([f"{webhook.bold(k.removeprefix('@'), request.service)}: {webhook.italic(v, request.service)}\n" for k, v in plan.items()])
+	def who(user): # a Slack user is a mention (<@U123>), which is markup to keep; elsewhere a handle is text
+		user = user.removeprefix('@')
+		return user if request.service == 'slack' else webhook.escape(user, request.service)
+	request.say([f"{webhook.bold(who(k), request.service)}: {webhook.italic(webhook.escape(v, request.service), request.service)}\n"
+		for k, v in plan.items()])
 
 def on_who(request, m):
 	if not m.group('ticker'):
diff --git a/lib/yahoo.py b/lib/yahoo.py
index 83720238..236532b9 100644
--- a/lib/yahoo.py
+++ b/lib/yahoo.py
@@ -56,6 +56,13 @@ def getCrumb(seconds=1209600): # 14 days
 		print("Got a new Yahoo crumb", file=sys.stderr)
 		return r.text
 
+def credentials():
+	"""(cookie, crumb), taken together under the lock: a crumb is only good with the cookie it was issued for, and a renewal
+	between two separate calls could pair a new cookie with an old crumb."""
+	with _auth:
+		crumb = getCrumb() # renews the cookie too, when due
+		return getCookie(), crumb
+
 def rmCrumb():
 	"""Set the cached crumb aside (as .old): it belongs to the cookie being replaced."""
 	crumb = Path(util.cache_path("finbot_yahoo_crumb.json"))
@@ -141,8 +148,7 @@ def fetch(tickers):
 	cacheFile = "finbot_yahoo_" + hashlib.sha256("_".join(tickers).encode()).hexdigest() + '.json'
 	if config_cache and (cache := util.read_cache(cacheFile, 300)):
 		return cache
-	cookie = getCookie()
-	crumb = getCrumb()
+	cookie, crumb = credentials()
 	r = _get_first(_hosts('/v7/finance/quote?crumb=' + crumb + '&symbols=' + ','.join(tickers)), {**HEADERS, 'Content-type': 'application/json', 'Cookie': cookie})
 	if r is None:
 		print("Exhausted Yahoo API attempts. Giving up", file=sys.stderr)
@@ -226,8 +232,7 @@ def fetch_detail(ticker, seconds=UNSET):
 	if seconds is UNSET:
 		seconds = config_cache_seconds
 	now = datetime.datetime.now()
-	cookie = getCookie()
-	crumb = getCrumb()
+	cookie, crumb = credentials()
 	cacheFile = "finbot_yahoo_detail_" + ticker + '.json'
 	data = util.read_cache(cacheFile, seconds) if config_cache else None
 	if data:
@@ -362,7 +367,8 @@ def price_history(ticker, days=None, seconds=UNSET, graph=UNSET, graphCache=True
 	if config_cache and graphCache and (image := util.read_binary_cache(image_cache_file, seconds)):
 		return percent_dict, image
 	image = util.graph(df, title, stock.get('currency'))
-	util.write_binary_cache(image_cache_file, image)
+	if config_cache:
+		util.write_binary_cache(image_cache_file, image)
 	return percent_dict, image
 
 def price_series(ticker):
@@ -439,12 +445,12 @@ def fetch_chart_json(ticker, days=3665, seconds=UNSET, full=False):
 	cacheFile = "finbot_yahoo_history_" + ticker + ("_max" if full else "") + ".json"
 	if config_cache and (cache := util.read_cache(cacheFile, seconds)):
 		return cache
-	crumb = getCrumb() # renews the cookie too, when due
+	cookie, crumb = credentials()
 	start = '0' if full else str(int((now - datetime.timedelta(days=days)).timestamp()))
 	url = f'https://query1.finance.yahoo.com/v8/finance/chart/{ticker}?period1={start}&period2={int(now.timestamp())}&interval=1d&crumb={crumb}'
 	print("Fetching", net.redact(url), file=sys.stderr) if debug else None
 	try:
-		r = net.get(url, headers={**HEADERS, 'Content-type': 'application/json'}, timeout=config_http_timeout)
+		r = net.get(url, headers={**HEADERS, 'Content-type': 'application/json', 'Cookie': cookie}, timeout=config_http_timeout)
 	except Exception as e:
 		raise RuntimeError(f"Yahoo unreachable for {ticker}: {net.redact(e)}") from e
 	if r.status_code != 200:
diff --git a/performance.py b/performance.py
index 23864308..5fd25df4 100755
--- a/performance.py
+++ b/performance.py
@@ -12,10 +12,10 @@
 BENCHMARKS = (('SPY', 'NYSEARCA', 'S&P 500'), ('QQQ', 'NasdaqGM', 'NASDAQ 100'))
 
 def gather(past_days, portfolio_select=None):
-	"""{portfolio id: Sharesight's performance report over past_days} of every portfolio, or of the one named (in any case). Raises
-	sharesight.PortfolioNotFound for a name that is none of them, and RuntimeError when there is no report."""
+	"""{portfolio id: (its name, Sharesight's performance report over past_days)} of every portfolio, or of the one named (in any
+	case). Raises sharesight.PortfolioNotFound for a name that is none of them, and RuntimeError when there is no report."""
 	portfolios = sharesight.pick_portfolios(portfolio_select)
-	performance = {portfolio_id: sharesight.get_performance(portfolio_id, past_days) for portfolio_id in portfolios.values()}
+	performance = {portfolio_id: (name, sharesight.get_performance(portfolio_id, past_days)) for name, portfolio_id in portfolios.items()}
 	if not performance:
 		print("Error: no Sharesight data found", file=sys.stderr)
 		raise RuntimeError("no Sharesight data found")
@@ -40,8 +40,7 @@ def report(service, past_days=UNSET, portfolio_select=None, interactive=False, p
 		performance = gather(past_days, portfolio_select)
 	lines = [] # (percent, text), sorted into the payload below
 	chart_rows = []
-	for portfolio_id, data in performance.items():
-		name = data['report']['holdings'][0]['portfolio']['name']
+	for portfolio_id, (name, data) in performance.items(): # the name from the portfolio list: one with no holdings names none
 		link = util.link("https://portfolio.sharesight.com/portfolios/" + str(portfolio_id), name, service)
 		percent = float(data['report']['capital_gain_percent'])
 		lines.append((percent, f"{util.get_emoji(percent)} {link} {util.signed_percent(percent, 2)}"))
diff --git a/tests/test_config.py b/tests/test_config.py
index ba4ade60..6f9703f8 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -98,6 +98,18 @@ def test_invalid_values_are_named(self):
 		self.assertEqual(len(caught.exception.errors), 2)
 		self.assertIn("future_days='abc' is not a whole number", caught.exception.errors[0])
 
+	def test_booleans_take_the_usual_words_and_refuse_a_typo(self):
+		for word, value in (('True', True), ('yes', True), ('ON', True), ('1', True), ('false', False), ('No', False), ('off', False), ('0', False)):
+			with self.subTest(word=word):
+				self.assertIs(config.preference_values(config._getenv_for({'graph': word}))['config_graph'], value)
+		with self.assertRaises(config.ConfigError) as caught: # it used to mean False without a word
+			config.preference_values(config._getenv_for({'graph': 'Ture'}))
+		self.assertIn("graph='Ture' is not true or false", caught.exception.errors[0])
+
+	def test_a_blank_value_means_the_default(self):
+		values = config.preference_values(config._getenv_for({'cache': '', 'future_days': '', 'country_code': ''}))
+		self.assertEqual((values['config_cache'], values['config_future_days'], values['config_country_code']), (True, 30, 'AU'))
+
 	def test_unreadable_file_is_an_error_naming_it(self):
 		path = Path(tempfile.mkdtemp(), 'preferences.conf')
 		path.mkdir() # reading a directory fails like a permission error would
@@ -323,6 +335,13 @@ def test_a_leftover_env_in_the_checkout_refuses_to_start(self):
 		self.assertIn(f'{old} is the old single config file, which finbot no longer reads', caught.exception.errors[0])
 		self.assertIn('/util/split_env.py, which writes finbot.env.new and preferences.conf.new', caught.exception.errors[1])
 
+	def test_slack_commands_without_a_bot_token_are_warned_about(self):
+		with mock.patch.object(config, 'config_slackOutgoingWebhook', 'https://example.com/slack'):
+			with mock.patch.object(config, 'config_slackBotToken', False):
+				self.assertEqual(config.startup_warnings(), ['slackOutgoingWebhook is set but slackBotToken is not: Slack commands will get no reply'])
+			with mock.patch.object(config, 'config_slackBotToken', 'xoxb-1'):
+				self.assertEqual(config.startup_warnings(), [])
+
 	def test_startup_report_is_a_short_count_never_names_or_values(self):
 		with mock.patch.dict(os.environ, {'slack_webhook': 'https://secret.example/abc', 'port': '5000'}, clear=True):
 			lines = config.startup_report()
diff --git a/tests/test_edges.py b/tests/test_edges.py
index d6776e93..43edf3cd 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -264,6 +264,32 @@ def test_slack_does_not_change_the_telegram_bot_name(self):
         self.assertEqual(bot.botName, '@finbot_tg')
 
 
+class UserTextIsEscaped(FinbotCase):
+    """Replies go out as Telegram HTML: text a user wrote is shown, never read as markup. One stray < in a .plan entry made Telegram
+    refuse that reply and every .plan reply after it."""
+
+    def test_a_plan_is_shown_as_text(self):
+        reply = self.command('.plan buy <cheap> & hold')[0][1]
+        self.assertIn('<i>buy &lt;cheap&gt; &amp; hold</i>', reply[0])
+        self.assertEqual(self.command('.plan')[0][1], reply) # and the same way again
+        self.sent.clear()
+        worker.process_request('slack', 'C1', '<@U1>', '.plan sell &lt;dear&gt;', BOT, 'U', '1') # Slack escapes what it sends
+        text = '\n'.join(self.sent[0][1])
+        self.assertIn('*<@U1>*', text) # a Slack mention stays markup
+        self.assertIn('_sell &lt;dear&gt;_', text) # escaped once, not twice
+        self.assertIn('_buy &lt;cheap&gt; &amp; hold_', text)
+
+    def test_a_telegram_name_is_escaped_where_it_arrives(self):
+        requests = []
+        self.patch(worker, 'process_request', lambda *a: requests.append(a))
+        body = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A<b>&'}, 'text': '.hello'}}
+        env = {'PATH_INFO': '/telegram', 'wsgi.input': io.BytesIO(json.dumps(body).encode()), 'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'tok'}
+        with contextlib.redirect_stderr(io.StringIO()):
+            bot.main(env, lambda *a, **k: None)
+        gevent.sleep(0.05) # the worker runs in a (gevent) thread
+        self.assertEqual((requests[0][2], requests[0][5]), ('@A&lt;b&gt;&amp;', '@A&lt;b&gt;&amp;')) # the handle and the name
+
+
 class SecretsStayOutOfLogs(FinbotCase):
     """A failed request's error message carries its URL: the Telegram token, a Slack or Discord webhook path, a Yahoo crumb."""
     SECRET_URL = 'https://api.telegram.org/bot123456:SECRET-tok_en/sendMessage?chat_id=1'
diff --git a/tests/test_providers.py b/tests/test_providers.py
index fdbbf8f6..9f66cf30 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -201,6 +201,20 @@ def slack(text): # Slack: the reports come as text
         self.assertEqual(len(self.errors), 1) # the one ticker asked for: one line saying why
         self.assertTrue(self.errors[0].startswith('⚠️ .price: RuntimeError: 404 error for T3'))
 
+    def test_the_chart_request_carries_the_cookie_of_its_crumb(self):
+        headers = []
+        self.patch(yahoo.net, 'get', lambda url, **kwargs: headers.append(kwargs.get('headers', {})) or Resp(self.chart([100.0] * 5)))
+        yahoo.fetch_chart_json('T1', full=True)
+        self.assertEqual(headers[0].get('Cookie'), 'A=B') # the crumb used to go without its cookie
+
+    def test_with_the_cache_off_no_chart_is_written(self):
+        self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
+        self.patch(yahoo, 'config_cache', False)
+        self.http(yahoo, Resp(self.chart([100.0] * 30 + [110.0])))
+        table, image = yahoo.price_history('T1', 7, graph=True)
+        self.assertIsNotNone(image)
+        self.assertEqual([name for name in os.listdir(util.config_cache_dir) if name.startswith('finbot_graph_')], [])
+
     def test_price_series_raises_on_failure_and_on_empty_history(self):
         self.http(yahoo, Resp(status=500))
         with self.assertRaises(RuntimeError):
@@ -545,6 +559,21 @@ def refuse(*a, **k):
                 self.assertTrue(self.sent)
                 self.assertEqual({kind for kind, *_ in self.sent}, {'text'})
 
+    def test_the_graph_preference_off_sends_no_chart_anywhere(self):
+        """graph=False means no charts, as the README says; it used to hold back only the price-line charts."""
+        self.patch(webhook, 'config_graph', False)
+        self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
+        posts = []
+        self.patch(webhook.net, 'post', lambda url, **k: posts.append(url) or Resp({'ok': True, 'result': True}))
+        for service, chat in (('telegram', '55'), ('slack', 'C1')):
+            for cmd in ('.price', '.price top', '.performance 7d', '.shorts', '.buy', '.peg', '.beta', '.compare T1 T2', '.pie Rob'):
+                with self.subTest(service=service, cmd=cmd):
+                    self.sent.clear(); self.errors.clear()
+                    worker.process_request(service, chat, '@u', cmd, BOT, 'U', '1')
+                    self.assertEqual(self.errors, [])
+                    self.assertEqual({kind for kind, *_ in self.sent}, {'text'})
+        self.assertEqual([url for url in posts if 'sendPhoto' in url or 'files.' in url], [])
+
     def test_a_chart_telegram_refuses_goes_out_as_text(self):
         def post(outcome):
             def fake(url, **kwargs):
@@ -565,6 +594,42 @@ def fake(url, **kwargs):
                 self.assertEqual(self.sent, expect)
 
 
+class SlackSending(FinbotCase):
+    """Slack replies go through chat.postMessage with the bot token; scheduled reports through the incoming webhook."""
+
+    def setUp(self):
+        super().setUp()
+        self.patch(webhook, 'payload_wrapper', REAL_PAYLOAD_WRAPPER)
+        self.patch(webhook, 'sendPhoto', REAL_SEND_PHOTO)
+        self.posts = []
+        self.patch(webhook.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or Resp({'ok': True}))
+
+    def test_link_previews_are_turned_off_in_the_body(self):
+        self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
+        webhook.write('slack', 'https://slack.com/api/chat.postMessage', 'hi', 'C1')
+        webhook.write('slack', 'https://hooks.slack.com/services/T/B/x', 'hi')
+        self.assertEqual(len(self.posts), 2)
+        for url, kwargs in self.posts:
+            with self.subTest(url=url): # they used to go in the headers, where Slack does not look
+                self.assertEqual((kwargs['json']['unfurl_links'], kwargs['json']['unfurl_media']), (False, False))
+                self.assertNotIn('unfurl_links', kwargs['headers'])
+
+    def test_a_reply_without_a_bot_token_is_logged_not_sent(self):
+        self.patch(webhook, 'config_slackBotToken', False)
+        err = io.StringIO()
+        with contextlib.redirect_stderr(err), contextlib.redirect_stdout(io.StringIO()):
+            for cmd in ('.help', '.price', '.compare T1 T2'):
+                worker.process_request('slack', 'C1', '<@U1>', cmd, BOT, 'U', '1')
+        self.assertEqual((self.posts, self.errors), ([], []))
+        self.assertEqual(err.getvalue().count('ERROR slack: a reply to channel C1 needs slackBotToken in finbot.env; not sent'), 3)
+
+    def test_the_slack_api_says_what_is_missing(self):
+        self.patch(webhook, 'config_slackBotToken', False)
+        with self.assertRaisesRegex(RuntimeError, 'Slack files.getUploadURLExternal: needs slackBotToken in finbot.env'):
+            webhook.slack_api('files.getUploadURLExternal', filename='chart.png', length=1) # it used to be a TypeError
+        self.assertEqual(self.posts, [])
+
+
 class TelegramWebhook(FinbotCase):
     """bot.py registers the webhook once as it starts. Importing lib.telegram, as every job process does, touches nothing."""
 
diff --git a/tests/test_reports.py b/tests/test_reports.py
index df019833..34869a2a 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -60,6 +60,12 @@ class ReportFunctions(FinbotCase):
     def test_prepare_help(self):
         self.assertTrue(reports.prepare_help('telegram', BOT))
 
+    def test_a_portfolio_with_no_holdings_is_reported_by_name(self):
+        """Its name used to be read from its first holding: an empty portfolio crashed .performance."""
+        self.patch(sharesight, 'get_performance', lambda pid, d, *a: {'report': {'holdings': [], 'capital_gain_percent': '0.0'}})
+        payload, _ = performance.report('telegram', 7)
+        self.assertTrue(any('>Rob</a>' in line for line in payload), payload)
+
     def test_prepare_holdings(self):
         self.assertTrue(reports.prepare_holdings_payload(None, 'telegram'))
 

From d71c1b142206f593fb6e1dcc42a37a8f2968920a Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Wed, 14 Oct 2026 23:01:54 +0000
Subject: [PATCH 02/81] Dead plumbing removed, quiet tests, a chat-networks
 table and a code map for CLAUDE.md

- Replies never quoted the asking message (the worker passed message_id=None even before the command table): the whole chain
  goes, from bot.py through process_request, the report scripts and write/sendPhoto, with Slack's thread_ts/reply_broadcast.
  So do the chat_id=telegramChatID defaults only an interactive run reads, the "make google cloud happy" returns,
  hbar_chart(threshold=) that only a test used, sharesight.warm_once (the tests call warm()[0]), a fontweight DejaVu Sans
  does not have, a dead emoji, an unreachable except, a stranded unittest.main() and a misnamed test.
- A passing test run prints 0.4 KB instead of 23.7 KB: -b in CLAUDE.md and CI, bot imported before requests so gevent patches
  ssl first, a real job child's output sent to /dev/null, and the two chart warnings fixed at their cause (all-zero bars
  divided by zero; a 2-day span made the date locator complain).
- README: a table of what each chat network can do, near the top; the command list as it is (.value and .rating never
  existed; .beta .buy .sell .compare .pe .fpe .peg .plan were missing); 12-hour cache default; no key counts to keep in step.
- CLAUDE.md: a code map, a test map, conventions and token tips.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 .github/workflows/ci.yml |  2 +-
 CLAUDE.md                | 36 +++++++++++++++++++++--
 README.md                | 63 ++++++++++++++++++++--------------------
 bot.py                   |  4 +--
 cal.py                   |  7 ++---
 lib/reports.py           | 50 +++++++++++++++----------------
 lib/sharesight.py        |  4 ---
 lib/util.py              | 12 ++++----
 lib/webhook.py           | 35 +++++++++-------------
 lib/worker.py            |  2 +-
 milestone.py             |  4 ---
 performance.py           |  4 +--
 price.py                 |  2 +-
 rating.py                |  5 +---
 reminder.py              |  3 --
 shorts.py                |  5 +---
 tests/common.py          |  4 +--
 tests/test_charts.py     |  3 +-
 tests/test_edges.py      | 12 ++++----
 tests/test_jobs.py       |  4 ++-
 tests/test_providers.py  | 34 +++++++++++-----------
 tests/test_reports.py    | 10 +++----
 trades.py                |  7 ++---
 23 files changed, 152 insertions(+), 160 deletions(-)

diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index e9a0f38b..d171e059 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -48,7 +48,7 @@ jobs:
         run: |
           pip install -r requirements.txt
           python -m compileall -q *.py lib util
-          python -m unittest discover -s tests -v
+          python -m unittest discover -s tests -v -b
 
   notify:
     # Telegram alert when the weekly run fails. Needs repository secrets TELEGRAM_BOT_TOKEN and TELEGRAM_CHAT_ID.
diff --git a/CLAUDE.md b/CLAUDE.md
index 9b1ea0ab..3089ef0b 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -2,7 +2,7 @@
 
 ## Iterating (auto mode)
 The owner chats back and forth; every message that asks for a change gets a change.
-1. Make the change, run the tests (`.venv/bin/python -m unittest discover -s tests`), fix what breaks.
+1. Make the change, run the tests (`.venv/bin/python -m unittest discover -s tests -b`), fix what breaks.
 2. Commit (with the attribution trailer) and **push to the work branch** (`git push -u origin <branch>`). Do this every iteration, so the branch always holds the latest working state. If a PR from the work branch is open, update its title and description in the same iteration (see below).
 3. Reply briefly: what changed, test result, and the pushed commit's short hash.
 
@@ -20,5 +20,37 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 ## Facts
 - Python 3.13.5; tests are offline and need no network (`tests/`). CI runs them on the latest 3.x and 3.13.5, except for pull requests that change only README.md, CLAUDE.md or `img/*.png` (install, compile and tests are skipped; the allowlist in `.github/workflows/ci.yml` must never include a path a test reads).
 - Production runs the Debian system packages (including `python3-croniter`, for the scheduler in `lib/jobs.py`); `requirements.txt` is for pip/uv users and CI.
-- A new preference goes in five places: `PREFERENCE_KEYS` and `preference_values` (`lib/config.py`), `defaults/preferences.conf`, the README's preferences table and its key count, and `DEFAULTS` plus the key count in `tests/test_config.py`. The tests catch all but the README.
+- A new preference goes in five places: `PREFERENCE_KEYS` and `preference_values` (`lib/config.py`), `defaults/preferences.conf`, the README's preferences table, and `DEFAULTS` plus the key count in `tests/test_config.py`. The tests catch all but the README (no test may read README.md: CI skips the tests for a README-only change).
 - If Yahoo and Sharesight are unreachable (the cloud environment's network policy can block them), check behaviour against the offline fixtures in `tests/common.py`. For a refactor, compare old and new output: load `git show HEAD:<file>` as a module beside the new one and diff what both produce for the same inputs.
+
+## Code map (open the one file you need)
+- `bot.py`: WSGI entry; routes /telegram and /slack, authenticates, hands each message to `lib/worker.py` in a thread; SIGHUP reload.
+- `lib/worker.py`: `COMMANDS` (regex → `on_*` handler, first match wins, `.SYMBOL` last); `Request`; the typing indicator.
+- `lib/reports.py`: `prepare_*` builders for the chat-only reports (lines, plus a chart spec).
+- `price.py performance.py trades.py shorts.py cal.py rating.py milestone.py reminder.py podcasts.py cleanup.py`: reports that also
+  run on a schedule; each has `lambda_handler()` and `job(argv)` (its schedule-line argument parser).
+- `lib/webhook.py`: sending (chunking, photos, Slack upload), markup, escaping, error reporting. `lib/telegram.py`: Bot API calls.
+- `lib/charts.py`: Telegram period buttons and the rebuild behind them. `lib/util.py`: formatting, links, market tables, cache and
+  state files, all matplotlib rendering.
+- `lib/yahoo.py lib/sharesight.py lib/shortman.py lib/simplywallst.py`: data sources, cached in `DATA_DIR/cache`.
+- `lib/config.py`: finbot.env + preferences.conf → `config_*` names (star-imported), reload. `lib/jobs.py` + `lib/jobrun.py`: the
+  scheduler; each run is a child process.
+- `tests/common.py`: `FinbotCase` (offline fixtures). `self.command('.price')` returns what was sent: `('text', lines)` or
+  `('photo', size, caption)`.
+- Which test file: commands and scheduled reports `test_reports`; failures, bad input, WSGI and Slack events `test_edges`;
+  Yahoo/Sharesight parsing, sending, Telegram API `test_providers`; scheduler `test_jobs`; config `test_config`; charts
+  `test_charts` (refresh goldens with `UPDATE_GOLDEN=1`); pure helpers `test_util`; data dir `test_datadir`; chunking, git_version
+  `test_finbot`.
+
+## Conventions
+- A report is a list of lines. Markup through `webhook.bold/italic(text, service)`, user-written text through
+  `webhook.escape(text, service)`, links through `util.link`/`util.finance_link`.
+- Tabs in code; in tests match the file. New and changed lines stay under 120 characters (an edit quotes whole lines).
+- stdout is discarded under systemd (`StandardOutput=null`): anything worth logging goes to stderr.
+- A new command: `COMMANDS`, `prepare_help`, the README command list, and `command_keyboard` if it gets a button.
+
+## Saving tokens
+- Run the tests quietly: `.venv/bin/python -m unittest discover -s tests -b` prints only failures (about 0.4 KB when green). One
+  module: `.venv/bin/python -m unittest tests.test_reports -b`.
+- README.md is 38 KB: grep for the section instead of reading it whole. `doc/` holds large API samples that the tests replay:
+  inspect them with python, don't read them.
diff --git a/README.md b/README.md
index 614b376f..7f2e802c 100644
--- a/README.md
+++ b/README.md
@@ -22,6 +22,20 @@
   * Shared watch list
   * List stocks (holdings + watchlist) by P/E, Forward P/E, PEG, market cap, or Yahoo analyst consensus
 
+### Chat networks
+What finbot can do on each network. Telegram takes everything; the others take what their APIs allow.
+
+| Feature | Telegram | Slack | Discord |
+|---|---|---|---|
+| Chat commands | ✓ | ✓ | ✗ (a webhook only posts) |
+| "typing…" while a command works | ✓ | ✗ (Slack has no typing notice for bots) | ✗ |
+| Charts in replies | ✓ | ✓ with `slackBotToken` | n/a |
+| Charts in scheduled reports | ✓ | ✗ (the incoming webhook takes text only) | ✗ |
+| Period and job buttons | ✓ | ✗ | ✗ |
+| No link previews | ✓ | ✓ | ✗ (set `hyperlink=False`) |
+| Linked tickers | ✓ | ✓ | ✓ |
+| Charts updated in place by their buttons | ✓ | ✗ | ✗ |
+
 ![Screenshot of showing trade notifications on Slack](img/screenshot.png?raw=true "Screenshot showing trade notifications on Slack")
 
 ### Requirements
@@ -37,53 +51,38 @@ croniter python-dateutil requests gevent pandas matplotlib numpy
 ### Supported commands
 
 Where:
-* _percent_ is a number followed by '%'
-* _period_ is a number followed by 'd' 'w' 'm' or 'y'
+* _percent_ is a number, optionally followed by '%'
+* _period_ is a number followed by 'd', 'w', 'm' or 'y', or 'ytd'
 * _portfolio_ is a Sharesight portfolio name
 * _symbol_ is a Yahoo Finance symbol
 ```
 .symbol
+.beta
+.buy
+.sell
+.compare symbol symbol [symbol...] [period|max]
 .dividend [period|symbol]
 .earnings [period|symbol]
+.forward pe [symbol|top|bottom|negative]
 .history symbol
 .holdings portfolio
 .jobs [run number|command]
 .marketcap [symbol|top|bottom]
+.pe [symbol|top|bottom]
+.peg [symbol|top|bottom|negative]
 .performance [period] [portfolio]
 .pie portfolio [sector]
+.plan [your plan]
 .premarket [percent|symbol]
-.price [percent|symbol] [period]
-.rating [strong buy|buy|hold|underperform|sell]
+.price [percent|symbol|top|bottom] [period]
 .session [percent|symbol]
 .shorts [percent|symbol]
 .trades [period] [portfolio]
-.value [pe|forward pe|peg|bottom pe|bottom forward pe|bottom peg|negative forward pe|negative peg]
 .watchlist [add|del symbol]
 .who symbol
 ```
 
-Alternative syntax:
-
-```
-@botname <symbol>
-@botname dividend [period|symbol]
-@botname earnings [period|symbol]
-@botname history symbol
-@botname holdings [portfolio]
-@botname jobs [run number|command]
-@botname marketcap [symbol|top|bottom]
-@botname performance [period] [portfolio]
-@botname pie portfolio [sector]
-@botname premarket [percent|symbol]
-@botname price [percent|symbol] [period]
-@botname rating [buy|hold|underperform|sell]
-@botname session [percent|symbol]
-@botname shorts [percent|symbol]
-@botname trades [period] [portfolio]
-@botname value [pe|forward pe|peg]
-@botname watchlist [add|del symbol]
-@botname who symbol
-```
+`!` or `/` work in place of the dot, and so does the bot's name: `@botname price 7d`, `@botname AAPL`. `.help` lists the commands in chat.
 
 Trade notifications are peformed by polling the Sharesight trades API on a schedule (see [Scheduling](#scheduling)), and notifying your configured chat networks of any new trades. Thus, it works best if your trades are auto-imported into Sharesight through its broker integrations, and if your environment has persistent storage so that the bot can keep track of known trade IDs between runs.
 
@@ -176,7 +175,7 @@ You can also specify a portfolio name to get trades for just that portfolio:
 ![Price chart triggered from Slack](img/finbot_price.png?raw=true "Price chart triggered from Slack")
 ![Price alert triggered by Cron](img/price.png?raw=true "Price alert triggered by Cron")
 
-`prices.py` sends premarket, midsession, and interday price movements if any holdings moved over a percentage threshold. This data is sourced from Yahoo! Finance and Sharesight. The default threshold is 5% but you can change it by setting `price_percent` in the `preferences.conf`, or by providing a number as argument when triggered through the chat bot. Decimal fractions are accepted.
+`price.py` sends premarket, midsession, and interday price movements if any holdings moved over a percentage threshold. This data is sourced from Yahoo! Finance and Sharesight. The default threshold is 5% but you can change it by setting `price_percent` in the `preferences.conf`, or by providing a number as argument when triggered through the chat bot. Decimal fractions are accepted.
 
 If `demote_volatile` is `true`, stocks with a [market cap](https://www.investopedia.com/terms/m/marketcapitalization.asp) under US$150M, or a [beta](https://www.investopedia.com/terms/b/beta.asp) of > 1.5 and a market cap of under US$1B will have their alert threshold (`price_percent`) doubled. A market cap in another currency is converted at Yahoo's exchange rate.
 
@@ -451,7 +450,7 @@ sudo cp ~/finbot/defaults/finbot.env /opt/finbot.env && sudo chmod 600 /opt/finb
 * **init.d / shell:** `util/check_config.py /opt/finbot.env && set -a && . /opt/finbot.env && set +a && ./bot.py` (`set -a` is what exports the values to the bot).
 * **Scheduled jobs:** they are started by `bot.py` (see [Scheduling](#scheduling)), so they run on the same settings. For a one-off run, see [Running a report by hand](#scheduling).
 
-All 17 keys, read by your launcher. A change needs a restart. Blank means unset.
+The startup keys, read by your launcher. A change needs a restart. Blank means unset.
 
 | Key | Default | Description |
 |---|---|---|
@@ -480,7 +479,7 @@ Copies preferences.conf and metadata files into the data directory (never overwr
 util/sync_defaults.py [-d DEST] --apply
 ```
 
-All 26 keys, read by the bot itself. A change takes effect on a reload. A true/false preference takes `true`/`false`, `yes`/`no`, `on`/`off` or `1`/`0`; anything else is a mistake, reported like any other. A blank value means the default.
+The preferences, read by the bot itself. A change takes effect on a reload. A true/false preference takes `true`/`false`, `yes`/`no`, `on`/`off` or `1`/`0`; anything else is a mistake, reported like any other. A blank value means the default.
 
 | Key | Default | Description |
 |---|---|---|
@@ -583,10 +582,10 @@ include_portfolios=100001 100002
 ```
 
 ### Caching
-Many objects are cached for just under one day by default. Cache is controlled by the settings below. Trades IDs are stored on disk, but trade queries generally don't result in a cache hit for functional reasons.
+Many objects are cached for 12 hours by default (`cache_seconds`). Cache is controlled by the settings below. Trades IDs are stored on disk, but trade queries generally don't result in a cache hit for functional reasons.
 ```
 cache=True
-cache_seconds=82800
+cache_seconds=43200
 performance_cache_seconds=1800
 ```
 
diff --git a/bot.py b/bot.py
index d33dc9a1..59cd22f3 100755
--- a/bot.py
+++ b/bot.py
@@ -95,7 +95,6 @@ def reply(status, body=b''):
 			return reply('200 OK', b'Unsupported')
 		msg = inbound["message"]
 		sender = msg["from"]
-		message_id = str(msg["message_id"])
 		chat_id = str(msg["chat"]["id"])
 		first_name = html.escape(sender["first_name"], quote=False) # replies are Telegram HTML: a name is text, not markup
 		if "username" in sender:
@@ -141,7 +140,6 @@ def reply(status, body=b''):
 			if event["type"] not in ('message', 'app_mention') or "text" not in event:
 				print(f"[{service}]: unhandled event callback type", event["type"], file=sys.stderr)
 				return reply('200 OK', b'<h1>Unhandled</h1>')
-			message_id = str(event["ts"])
 			message = event['text']
 			message = re.sub(r'<http://.*\|([\w\.]+)>', r'\g<1>', message) # <http://dub.ax|dub.ax> becomes dub.ax
 			message = re.sub(r'<(@[\w\.]+)>', r'\g<1>', message) # <@QWERTY> becomes @QWERTY
@@ -155,7 +153,7 @@ def reply(status, body=b''):
 			return reply('200 OK', b'<h1>Unhandled</h1>')
 
 	# process in a background thread so we don't keep the requesting client waiting
-	threading.Thread(target=worker.process_request, args=(service, chat_id, user, message, bot_name, userRealName, message_id)).start()
+	threading.Thread(target=worker.process_request, args=(service, chat_id, user, message, bot_name, userRealName)).start()
 
 	# Return an empty response to the client
 	return reply('200 OK')
diff --git a/cal.py b/cal.py
index 1a63e487..10b9dd0f 100755
--- a/cal.py
+++ b/cal.py
@@ -11,7 +11,7 @@
 
 EVENTS = {'earnings': ('earnings_date', 'Upcoming earnings'), 'ex-dividend': ('ex_dividend_date', 'Upcoming ex-dividends')} # -> (field, heading)
 
-def lambda_handler(chat_id=config_telegramChatID, days=UNSET, service=None, specific_stock=None, message_id=None, interactive=False, event='earnings'):
+def lambda_handler(chat_id=None, days=UNSET, service=None, specific_stock=None, interactive=False, event='earnings'):
 	"""The tracked stocks' (or one stock's) upcoming earnings or ex-dividend dates, as `event` says. A scheduled earnings report
 	is pinned on Telegram."""
 	if days is UNSET:
@@ -78,10 +78,7 @@ def pin_if_earnings(service, results):
 		print("")
 
 	for service, url, chat in webhook.destinations(interactive, service, chat_id):
-		pin_if_earnings(service, webhook.send(service, url, chat, prepare_payload(service, market_data), message_id=message_id))
-
-	# make google cloud happy
-	return True
+		pin_if_earnings(service, webhook.send(service, url, chat, prepare_payload(service, market_data)))
 
 def job(argv):
 	"""Arguments of a scheduled run: (label, lambda_handler kwargs). Raises ValueError with the usage on a bad argument."""
diff --git a/lib/reports.py b/lib/reports.py
index fb6fe5a7..d42c06bd 100755
--- a/lib/reports.py
+++ b/lib/reports.py
@@ -423,36 +423,32 @@ def prepare_value_payload(service, action='pe', ticker_select=None, length=15):
 		forward_ratio = None
 		peg_ratio = None
 		ratio = None
-		try:
-			if action in ('pe', 'bottom pe', 'forward pe', 'bottom forward pe'):
-				trailing_ratio = market_data.get(ticker).get('price_to_earnings_trailing')
-				forward_ratio = market_data.get(ticker).get('price_to_earnings_forward')
-				if action in ('forward pe', 'bottom forward pe'):
-					if not ticker_select and forward_ratio is not None and forward_ratio < 0:
-						continue
-					ratio = forward_ratio
-				else:
-					if not ticker_select and trailing_ratio is not None and trailing_ratio < 0:
-						continue
-					ratio = trailing_ratio
-			elif action == 'negative forward pe':
-				forward_ratio = market_data.get(ticker).get('price_to_earnings_forward')
-				if not ticker_select and (forward_ratio is None or forward_ratio >= 0):
+		if action in ('pe', 'bottom pe', 'forward pe', 'bottom forward pe'):
+			trailing_ratio = market_data.get(ticker).get('price_to_earnings_trailing')
+			forward_ratio = market_data.get(ticker).get('price_to_earnings_forward')
+			if action in ('forward pe', 'bottom forward pe'):
+				if not ticker_select and forward_ratio is not None and forward_ratio < 0:
 					continue
 				ratio = forward_ratio
-			elif action in ('peg', 'bottom peg'):
-				peg_ratio = market_data.get(ticker).get('price_to_earnings_peg')
-				if not ticker_select and peg_ratio is not None and peg_ratio < 0:
-					continue
-				ratio = peg_ratio
-			elif action == 'negative peg':
-				peg_ratio = market_data.get(ticker).get('price_to_earnings_peg')
-				if not ticker_select and (peg_ratio is None or peg_ratio >= 0):
+			else:
+				if not ticker_select and trailing_ratio is not None and trailing_ratio < 0:
 					continue
-				ratio = peg_ratio
-		except KeyError:
-			print(ticker, action, "value not found", file=sys.stderr)
-			continue
+				ratio = trailing_ratio
+		elif action == 'negative forward pe':
+			forward_ratio = market_data.get(ticker).get('price_to_earnings_forward')
+			if not ticker_select and (forward_ratio is None or forward_ratio >= 0):
+				continue
+			ratio = forward_ratio
+		elif action in ('peg', 'bottom peg'):
+			peg_ratio = market_data.get(ticker).get('price_to_earnings_peg')
+			if not ticker_select and peg_ratio is not None and peg_ratio < 0:
+				continue
+			ratio = peg_ratio
+		elif action == 'negative peg':
+			peg_ratio = market_data.get(ticker).get('price_to_earnings_peg')
+			if not ticker_select and (peg_ratio is None or peg_ratio >= 0):
+				continue
+			ratio = peg_ratio
 		if ratio is None:
 			continue
 		profile_title = market_data.get(ticker).get('profile_title')
diff --git a/lib/sharesight.py b/lib/sharesight.py
index d41a4895..824b9257 100644
--- a/lib/sharesight.py
+++ b/lib/sharesight.py
@@ -220,10 +220,6 @@ def warm(margin=0.9):
 			fetched += 1
 	return fetched, portfolios
 
-def warm_once(margin=0.9):
-	"""warm(), returning only the number of fetches made."""
-	return warm(margin)[0]
-
 def next_due(margin=0.9, portfolios=None):
 	"""Seconds until the soonest warmed cache needs refreshing (0 if one already does). `portfolios` is the list warm() just
 	used; without it the cached list is read."""
diff --git a/lib/util.py b/lib/util.py
index 26c8f44c..22726183 100644
--- a/lib/util.py
+++ b/lib/util.py
@@ -353,11 +353,9 @@ def style_date_axis(ax, start, end, ink):
 	import matplotlib.dates as mdates
 	days = (end - start).days
 	fmt = '%Y' if days > 1460 else '%b %Y' if days > 150 else '%d %b'
-	ax.xaxis.set_major_locator(mdates.AutoDateLocator(minticks=3, maxticks=5))
+	ax.xaxis.set_major_locator(mdates.AutoDateLocator(minticks=2, maxticks=5)) # minticks=3 warns on a 2-day span
 	ax.xaxis.set_major_formatter(mdates.DateFormatter(fmt))
 	ax.tick_params(axis='x', colors=ink, labelsize=10, length=3, width=0.8, pad=6)
-	for label in ax.get_xticklabels():
-		label.set_fontweight('medium')
 
 PALETTE = ('#14181f', '#eef1f5', '#9aa4b2', '#2a313c') # charts: dark surface, ink, muted ink, grid lines
 
@@ -496,7 +494,7 @@ def tinted(bg, color, amount=0.07):
 	mix = lambda i: round(int(bg[i:i + 2], 16) * (1 - amount) + int(color[i:i + 2], 16) * amount)
 	return '#{:02x}{:02x}{:02x}'.format(mix(1), mix(3), mix(5))
 
-def hbar_chart(rows, title, subtitle='', value_fmt=None, threshold=None, xlim=None, session=None):
+def hbar_chart(rows, title, subtitle='', value_fmt=None, xlim=None, session=None):
 	"""Horizontal bars that mirror the caption list, first row on top, filling the whole 1280x800 canvas.
 	Blue for >= 0, red below. Each row's label sits at the bar's base and its value at the tip.
 	rows: [(label, value)]; a (label, None) row is a blank spacer. xlim zooms the value axis (bars then start at its left edge).
@@ -517,11 +515,11 @@ def hbar_chart(rows, title, subtitle='', value_fmt=None, threshold=None, xlim=No
 	base = max(xlim[0], 0) if xlim else 0
 	if not xlim:
 		ax.axvline(0, color=ink2, linewidth=1.0, zorder=4)
-	if threshold is not None:
-		ax.axvline(threshold, color=ink2, linewidth=0.9, linestyle=(0, (4, 3)), zorder=4)
-	top, bottom = max(max(values), 0, threshold or 0), min(min(values), 0)
+	top, bottom = max(max(values), 0), min(min(values), 0)
 	lo0 = xlim[0] if xlim else bottom
 	hi0 = xlim[1] if xlim else top
+	if hi0 <= lo0: # every value zero: give the axis a width, or the scale below divides by zero
+		hi0 = lo0 + 1
 	ax.set_xlim(lo0, hi0)
 	ax.set_ylim(n - 0.4, -0.6) # first row on top
 	ax.set_yticks([])
diff --git a/lib/webhook.py b/lib/webhook.py
index 52d82810..3dc706ab 100644
--- a/lib/webhook.py
+++ b/lib/webhook.py
@@ -11,7 +11,7 @@
 from lib import net
 from lib import telegram
 
-def write(service, url, payload_string, chat_id=None, message_id=None, reply_markup=None):
+def write(service, url, payload_string, chat_id=None, reply_markup=None):
 	headers = {'Content-type': 'application/json'}
 	payload = {'text': payload_string}
 	if 'slack.com' in url:
@@ -23,15 +23,10 @@ def write(service, url, payload_string, chat_id=None, message_id=None, reply_mar
 				return None
 			headers['Authorization'] = 'Bearer ' + config_slackBotToken
 			payload['channel'] = chat_id
-			if message_id:
-				payload['thread_ts'] = message_id
-				payload['reply_broadcast'] = 'true'
 	elif 'api.telegram.org' in url:
 		payload['parse_mode'] = 'HTML'
 		payload['disable_web_page_preview'] = 'true'
 		payload['disable_notification'] = 'true'
-		payload['allow_sending_without_reply'] = 'true'
-		payload['reply_to_message_id'] = message_id
 		if reply_markup:
 			payload['reply_markup'] = reply_markup # keyboard (Telegram only)
 	try:
@@ -70,12 +65,12 @@ def destinations(interactive, service, chat_id):
 	"""Where a report goes: back to the chat that asked for it (interactive), else to the default chats."""
 	return [(service, chat_url(service, chat_id), chat_id)] if interactive else default_chats()
 
-def send(service, url, chat_id, payload, image=None, reply_markup=None, message_id=None):
+def send(service, url, chat_id, payload, image=None, reply_markup=None):
 	"""A report: its chart with the text as caption when there is one (sendPhoto falls back to the text where a chart cannot go),
 	else the text. reply_markup is the chart's buttons. Returns what payload_wrapper or sendPhoto did."""
 	if image:
-		return sendPhoto(chat_id, image, '\n'.join(payload), service, message_id=message_id, reply_markup=reply_markup)
-	return payload_wrapper(service, url, payload, chat_id, message_id)
+		return sendPhoto(chat_id, image, '\n'.join(payload), service, reply_markup=reply_markup)
+	return payload_wrapper(service, url, payload, chat_id)
 
 def error_line(e, context=None, limit=200):
 	"""One-line, secret-free summary of an exception (or string)."""
@@ -126,7 +121,7 @@ def cli(script, job, handler, argv=None):
 		sys.exit(1)
 	guarded(label, handler, **kwargs)
 
-def payload_wrapper(service, url, payload, chat_id=None, message_id=None, reply_markup=None):
+def payload_wrapper(service, url, payload, chat_id=None, reply_markup=None):
 	"""Returns a list with the response of each message sent (several if the payload was chunked)."""
 	if not payload:
 		print(service + ": Nothing to send") # informational
@@ -137,7 +132,7 @@ def payload_wrapper(service, url, payload, chat_id=None, message_id=None, reply_
 	print("Payload: " + payload_string) if debug else None
 	limit = 2000 if service == 'discord' else 4096
 	if len(payload_string) <= limit:
-		return [write(service, url, payload_string, chat_id, message_id, reply_markup)]
+		return [write(service, url, payload_string, chat_id, reply_markup)]
 	print(service, f"payload is over {limit:,} bytes. Splitting.")
 	chunks = list(batched(payload, config_chunk_maxlines))
 	results = []
@@ -201,7 +196,7 @@ def takes_charts(service, reply=True):
 	incoming webhook and takes text only; never on Discord. Elsewhere a report goes out as text."""
 	return config_graph and (service == 'telegram' or (service == 'slack' and reply and bool(config_slackBotToken)))
 
-def telegram_photo(chat_id, image_data, caption, message_id=None, reply_markup=None):
+def telegram_photo(chat_id, image_data, caption, reply_markup=None):
 	"""sendPhoto, the report as its caption (trimmed to fit). Telegram's answer, MAYBE_SENT, or None if it was refused."""
 	url = webhooks['telegram'] + "sendPhoto?chat_id=" + str(chat_id)
 	data = {
@@ -209,9 +204,7 @@ def telegram_photo(chat_id, image_data, caption, message_id=None, reply_markup=N
 		'chat_id': chat_id,
 		'caption': fit_caption(caption),
 		'parse_mode': 'HTML',
-		'disable_web_page_preview': True,
-		'allow_sending_without_reply': True,
-		'reply_to_message_id': message_id}
+		'disable_web_page_preview': True}
 	if reply_markup:
 		data['reply_markup'] = json.dumps(reply_markup) # inline keyboard (Telegram only)
 	if hasattr(image_data, 'seek'):
@@ -247,7 +240,7 @@ def slack_api(method, **data):
 		raise RuntimeError(f"Slack {method}: {output.get('error') or r.status_code}")
 	return output
 
-def slack_photo(chat_id, image_data, caption, message_id=None):
+def slack_photo(chat_id, image_data, caption):
 	"""Upload the chart into a Slack channel with the report as its message, the way Slack takes files since it retired files.upload:
 	ask for an upload URL, send the bytes there, then complete the upload into the channel (the bot needs the files:write and
 	files:read scopes). Slack's answer, MAYBE_SENT, or None if a step failed."""
@@ -260,8 +253,6 @@ def slack_photo(chat_id, image_data, caption, message_id=None):
 		if r.status_code != 200:
 			raise RuntimeError(f"Slack upload: {r.status_code} {r.text[:200]}")
 		share = {'files': json.dumps([{'id': ticket['file_id']}]), 'channel_id': chat_id, 'initial_comment': caption}
-		if message_id:
-			share['thread_ts'] = message_id
 		try:
 			output = slack_api('files.completeUploadExternal', **share)
 		except requests.exceptions.ReadTimeout as e:
@@ -273,23 +264,23 @@ def slack_photo(chat_id, image_data, caption, message_id=None):
 	print("OK slack chart upload to", chat_id, file=sys.stderr)
 	return output
 
-def sendPhoto(chat_id, image_data, caption, service, message_id=None, reply_markup=None):
+def sendPhoto(chat_id, image_data, caption, service, reply_markup=None):
 	"""The chart, with the report as its caption: on Telegram, or uploaded into the Slack channel of a reply. Where a chart cannot go
 	(takes_charts), or the service refuses it, the report is sent as text instead, without the chart's buttons, so it does not vanish
 	with its chart; but not after a read timeout on the request that posts it, when the chart may have arrived after all."""
 	output = None
 	chartable = takes_charts(service, reply=bool(chat_id))
 	if chartable and service == 'telegram':
-		output = telegram_photo(chat_id, image_data, caption, message_id, reply_markup)
+		output = telegram_photo(chat_id, image_data, caption, reply_markup)
 	elif chartable:
-		output = slack_photo(chat_id, image_data, caption, message_id)
+		output = slack_photo(chat_id, image_data, caption)
 	if output is MAYBE_SENT:
 		return None
 	if output:
 		return output
 	if chartable:
 		print("chart not sent, sending the report as text", file=sys.stderr)
-	return payload_wrapper(service, chat_url(service, chat_id) if chat_id else webhooks.get(service), caption.split('\n'), chat_id, message_id)
+	return payload_wrapper(service, chat_url(service, chat_id) if chat_id else webhooks.get(service), caption.split('\n'), chat_id)
 
 def editMessageText(chat_id, message_id, text, reply_markup=None):
 	"""Telegram: replace a text message's text and buttons in place. A failure is only logged."""
diff --git a/lib/worker.py b/lib/worker.py
index acf456fe..193cf716 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -129,7 +129,7 @@ def process_callback(service, callback_id, chat_id, message_id, data):
 	except Exception as e:
 		webhook.report_error(e, service, chat_id, context='chart')
 
-def process_request(service, chat_id, user, message, botName, userRealName, message_id):
+def process_request(service, chat_id, user, message, botName, userRealName):
 	"""Entry point for inbound chat requests: the first command in COMMANDS that fits the message answers it (a message that fits
 	none gets no answer). Any failure becomes a one-line reply in the originating chat, after the typing stops."""
 	request = Request(service, chat_id, user, userRealName, botName)
diff --git a/milestone.py b/milestone.py
index 4c778d5b..5ccbd3d6 100755
--- a/milestone.py
+++ b/milestone.py
@@ -13,7 +13,6 @@
 
 def lambda_handler():
 	def prepare_milestone_payload(service, market_data):
-		emoji = '⭐'
 		payload = []
 		payloadhigh = []
 		payloadlow = []
@@ -68,9 +67,6 @@ def prepare_milestone_payload(service, market_data):
 	if records:
 		util.json_write('finbot_milestone.json', records, persist=True)
 
-	# make google cloud happy
-	return True
-
 def job(argv):
 	return 'milestone.py', {}
 
diff --git a/performance.py b/performance.py
index 5fd25df4..dc684740 100755
--- a/performance.py
+++ b/performance.py
@@ -61,14 +61,14 @@ def report(service, past_days=UNSET, portfolio_select=None, interactive=False, p
 		graph = util.rows_chart(chart_rows, "Performance", f"% change {'' if period == 'today' else 'over '}{period}", value_fmt=lambda v: util.signed_percent(v, 2))
 	return payload, graph
 
-def lambda_handler(chat_id=config_telegramChatID, past_days=UNSET, service=None, user='', portfolio_select=None, message_id=None, interactive=False):
+def lambda_handler(chat_id=None, past_days=UNSET, service=None, user='', portfolio_select=None, interactive=False):
 	if past_days is UNSET:
 		past_days = config_past_days
 	performance = gather(past_days, portfolio_select)
 	for service, url, chat in webhook.destinations(interactive, service, chat_id):
 		payload, graph = report(service, past_days, portfolio_select, interactive, performance)
 		buttons = charts.keyboard('f', [portfolio_select or ''], charts.period_for_days(past_days)) if graph and service == 'telegram' else None
-		webhook.send(service, url, chat, payload, graph, reply_markup=buttons, message_id=message_id)
+		webhook.send(service, url, chat, payload, graph, reply_markup=buttons)
 
 def job(argv):
 	"""Arguments of a scheduled run: (label, lambda_handler kwargs). Raises ValueError with the usage on a bad argument."""
diff --git a/price.py b/price.py
index d1b27426..7fad0339 100755
--- a/price.py
+++ b/price.py
@@ -154,7 +154,7 @@ def report(service, threshold=UNSET, user='', specific_stock=None, interactive=F
 			payload = [f"{user}, no price movements meet threshold {threshold:g}%"]
 	return payload, graph
 
-def lambda_handler(chat_id=config_telegramChatID, threshold=UNSET, service=None, user='', specific_stock=None, interactive=False, midsession=False,
+def lambda_handler(chat_id=None, threshold=UNSET, service=None, user='', specific_stock=None, interactive=False, midsession=False,
 		premarket=False, interday=False, days=None, close=False, top=None):
 	if threshold is UNSET:
 		threshold = config_price_percent
diff --git a/rating.py b/rating.py
index 28476edd..34c0877d 100755
--- a/rating.py
+++ b/rating.py
@@ -6,7 +6,7 @@
 from lib import util
 from lib import yahoo
 
-def lambda_handler(chat_id=config_telegramChatID, specific_stock=None, service=None, interactive=False):
+def lambda_handler(chat_id=None, specific_stock=None, service=None, interactive=False):
 	"""Analysts' consensus ratings that changed since the last scheduled run, which records the ratings it saw."""
 	def prepare_rating_payload(service, market_data):
 		def get_emoji(old, new):
@@ -66,9 +66,6 @@ def get_emoji(old, new):
 	if new and not interactive: # added to what is recorded: a ticker missing from this run keeps its last rating
 		util.json_write('finbot_rating.json', old | new, persist=True)
 
-	# make google cloud happy
-	return True
-
 def job(argv):
 	return 'rating.py', {}
 
diff --git a/reminder.py b/reminder.py
index 09b043b2..7205c783 100755
--- a/reminder.py
+++ b/reminder.py
@@ -71,9 +71,6 @@ def prepare_finance_calendar_payload(service):
 	for service, url, chat in webhook.default_chats():
 		webhook.send(service, url, chat, prepare_finance_calendar_payload(service))
 
-	# make google cloud happy
-	return True
-
 def job(argv):
 	return 'reminder.py', {}
 
diff --git a/shorts.py b/shorts.py
index 674694ca..2e44772b 100755
--- a/shorts.py
+++ b/shorts.py
@@ -7,7 +7,7 @@
 from lib import yahoo
 from lib import shortman
 
-def lambda_handler(chat_id=config_telegramChatID, threshold=UNSET, specific_stock=None, service=None, interactive=False):
+def lambda_handler(chat_id=None, threshold=UNSET, specific_stock=None, service=None, interactive=False):
 	if threshold is UNSET:
 		threshold = config_shorts_percent
 	def prepare_shorts_payload(service, market_data):
@@ -48,9 +48,6 @@ def prepare_shorts_payload(service, market_data):
 		payload, graph = prepare_shorts_payload(service, market_data)
 		webhook.send(service, url, chat, payload, graph)
 
-	# make google cloud happy
-	return True
-
 def job(argv):
 	return 'shorts.py', {}
 
diff --git a/tests/common.py b/tests/common.py
index d6a0e971..7df12171 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -9,9 +9,9 @@
 os.chdir(ROOT)
 os.environ.update(telegramOutgoingWebhook='https://example.com/telegram', telegramOutgoingToken='tok', telegramChatID='1')
 
+import bot # first, and once: it monkey-patches with gevent, which must happen before requests imports ssl and never mid-test
 from lib import config, webhook, util, yahoo, sharesight, shortman, worker, charts, telegram, reports
 import performance, shorts, price, reminder, cal, trades, milestone, rating
-import bot # imported once up front: it monkey-patches with gevent, which must not happen mid-test
 
 REAL = {(m.__name__, n): getattr(m, n) for m, n in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (yahoo, 'price_history'),
     (sharesight, 'get_portfolios'), (sharesight, 'get_trades'), (sharesight, 'get_performance'), (sharesight, 'get_performance_wrapper'))}
@@ -103,6 +103,6 @@ def command(self, text, chat='55'):
         self.sent.clear()
         self.markups.clear()
         self.errors.clear()
-        worker.process_request('telegram', chat, '@u', text, BOT, 'U', '1')
+        worker.process_request('telegram', chat, '@u', text, BOT, 'U')
         self.assertEqual(self.errors, [], f'{text} crashed') # the worker reports crashes instead of raising
         return list(self.sent)
diff --git a/tests/test_charts.py b/tests/test_charts.py
index 44822e20..a9d33208 100644
--- a/tests/test_charts.py
+++ b/tests/test_charts.py
@@ -87,11 +87,10 @@ def test_edge_case_rows(self):
                 self.check_image(util.hbar_chart(rows, 'Title', 'sub'), name)
                 self.check_image(util.rows_chart(rows, 'Title'), name + '/rows_chart')
 
-    def test_zoomed_axis_and_threshold(self):
+    def test_zoomed_axis(self):
         self.check_image(util.hbar_chart([('AAA', 1.2), ('BBB', 3.4)], 'Ratings', xlim=(0.9, 4.0)), 'zoomed')
         self.check_image(util.hbar_chart([('WWWWWWWW', 1.2), ('MMM', 3.4)], 'Ratings', xlim=(0.9, 4.0)), 'zoomed wide labels')
         self.check_image(util.hbar_chart([(f'T{i}', 1 + i / 10) for i in range(15)], 'Ratings', xlim=(0.9, 3.0)), 'zoomed 15 rows')
-        self.check_image(util.hbar_chart(MIXED, 'Threshold', threshold=10), 'threshold')
 
     def test_shorts_chart_sizes_and_verdicts(self):
         for n in (1, 5, 45):
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 43edf3cd..0635972a 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -19,7 +19,7 @@ def run_all(self):
             with self.subTest(cmd=cmd):
                 if cmd in self.ONE_LINE_ERROR:
                     self.sent.clear(); self.errors.clear()
-                    worker.process_request('telegram', '55', '@u', cmd, BOT, 'U', '1')
+                    worker.process_request('telegram', '55', '@u', cmd, BOT, 'U')
                     self.assertLessEqual(len(self.errors), 1, cmd)
                 else:
                     self.command(cmd)
@@ -37,7 +37,7 @@ def down(*a, **k):
         for cmd in ('.price', '.peg', '.T1', '.compare T1 T2'):
             with self.subTest(cmd=cmd):
                 self.sent.clear(); self.errors.clear()
-                worker.process_request('telegram', '55', '@u', cmd, BOT, 'U', '1')
+                worker.process_request('telegram', '55', '@u', cmd, BOT, 'U')
                 self.assertEqual(len(self.errors), 1, cmd) # exactly one one-line report, not a traceback to the user
                 self.assertTrue(self.errors[0].startswith('⚠️'))
 
@@ -48,7 +48,7 @@ def down(*a, **k):
         self.patch(sharesight, 'get_performance', down)
         self.patch(sharesight, 'get_portfolios', down)
         self.sent.clear(); self.errors.clear()
-        worker.process_request('telegram', '55', '@u', '.performance 7d', BOT, 'U', '1')
+        worker.process_request('telegram', '55', '@u', '.performance 7d', BOT, 'U')
         self.assertEqual(len(self.errors), 1)
 
     def test_a_failed_command_stops_typing_before_saying_why(self):
@@ -57,7 +57,7 @@ def down(*a, **k):
         self.patch(sharesight, 'get_portfolios', down)
         typing_when_reported = []
         self.patch(webhook, 'report_error', lambda e, *a, **k: typing_when_reported.append([t.is_active() for t in self.typing]))
-        worker.process_request('telegram', '55', '@u', '.performance 7d', BOT, 'U', '1')
+        worker.process_request('telegram', '55', '@u', '.performance 7d', BOT, 'U')
         self.assertEqual(typing_when_reported, [[False]]) # its one indicator had stopped when the error line went out
 
     def test_no_holdings_and_empty_watchlist(self):
@@ -91,7 +91,7 @@ def no_history(t):
             raise RuntimeError(f'no price history for {t}')
         self.patch(yahoo, 'price_series', no_history)
         self.sent.clear(); self.errors.clear()
-        worker.process_request('telegram', '55', '@u', '.compare T1 T2', BOT, 'U', '1')
+        worker.process_request('telegram', '55', '@u', '.compare T1 T2', BOT, 'U')
         self.assertEqual(len(self.errors), 1)
 
     def test_history_without_a_ticker_or_with_extra_argument_gives_usage(self):
@@ -273,7 +273,7 @@ def test_a_plan_is_shown_as_text(self):
         self.assertIn('<i>buy &lt;cheap&gt; &amp; hold</i>', reply[0])
         self.assertEqual(self.command('.plan')[0][1], reply) # and the same way again
         self.sent.clear()
-        worker.process_request('slack', 'C1', '<@U1>', '.plan sell &lt;dear&gt;', BOT, 'U', '1') # Slack escapes what it sends
+        worker.process_request('slack', 'C1', '<@U1>', '.plan sell &lt;dear&gt;', BOT, 'U') # Slack escapes what it sends
         text = '\n'.join(self.sent[0][1])
         self.assertIn('*<@U1>*', text) # a Slack mention stays markup
         self.assertIn('_sell &lt;dear&gt;_', text) # escaped once, not twice
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
index 072f3c10..4ae1cfe1 100644
--- a/tests/test_jobs.py
+++ b/tests/test_jobs.py
@@ -483,7 +483,7 @@ def test_telegram_gets_a_button_per_command_and_slack_does_not(self):
     def test_the_chat_command_passes_the_keyboard_on(self):
         seen = []
         self.patch(webhook, 'payload_wrapper', lambda svc, url, payload, *a, **k: seen.append(k.get('reply_markup')) or [])
-        worker.process_request('telegram', '55', '@u', '.jobs', BOT, 'U', '1')
+        worker.process_request('telegram', '55', '@u', '.jobs', BOT, 'U')
         self.assertEqual(len(seen[0]['inline_keyboard'][0]), 2)
 
     def test_button_press_runs_that_command(self):
@@ -580,6 +580,8 @@ def test_every_command_can_be_handed_to_a_child(self):
                 self.assertEqual(json.loads(json.dumps(kwargs)), kwargs, label)
 
     def test_a_real_child_runs_the_handler_and_reports_its_exit_status(self):
+        real = jobs.subprocess.Popen
+        self.patch(jobs.subprocess, 'Popen', lambda args, **kw: real(args, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, **kw))
         with tempfile.TemporaryDirectory() as data:
             env = {'DATA_DIR': data, 'telegramBotToken': '', 'slack_webhook': '', 'discord_webhook': ''}
             with mock.patch.dict(os.environ, env):
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 9f66cf30..ddd3f8dd 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -183,7 +183,7 @@ def get(url, **kwargs):
         self.patch(yahoo.net, 'get', get)
         def slack(text): # Slack: the reports come as text
             self.sent.clear(); self.errors.clear()
-            worker.process_request('slack', 'C1', '<@U1>', text, BOT, 'U', '1')
+            worker.process_request('slack', 'C1', '<@U1>', text, BOT, 'U')
             self.assertEqual(self.errors, [], text)
             return '\n'.join(map(str, self.sent[-1][1]))
         price_list = slack('.price 7d') # T3 is left out, the rest are listed
@@ -197,7 +197,7 @@ def slack(text): # Slack: the reports come as text
         self.assertIn('NASDAQ 100', report)
         self.command('.T3') # the profile goes out without its history lines
         self.sent.clear(); self.errors.clear()
-        worker.process_request('telegram', '55', '@u', '.price T3 7d', BOT, 'U', '1')
+        worker.process_request('telegram', '55', '@u', '.price T3 7d', BOT, 'U')
         self.assertEqual(len(self.errors), 1) # the one ticker asked for: one line saying why
         self.assertTrue(self.errors[0].startswith('⚠️ .price: RuntimeError: 404 error for T3'))
 
@@ -303,28 +303,28 @@ def age(self, cache_file, seconds):
 
     def test_cold_cache_is_filled_then_left_alone(self):
         calls = self.serve_sharesight()
-        self.assertEqual(sharesight.warm_once(), 3) # portfolio list + one holdings report per portfolio (2)
+        self.assertEqual(sharesight.warm()[0], 3) # portfolio list + one holdings report per portfolio (2)
         self.assertEqual(sum('/performance' in c for c in calls), 2)
         self.assertTrue(all('start_date=' + datetime.date.today().isoformat() in c for c in calls if '/performance' in c)) # days=0
-        self.assertEqual(sharesight.warm_once(), 0)
+        self.assertEqual(sharesight.warm()[0], 0)
         self.assertEqual(len(calls), 3) # nothing refetched while fresh
 
     def test_refreshes_before_the_ttl_runs_out(self):
         calls = self.serve_sharesight()
-        sharesight.warm_once()
+        sharesight.warm()[0]
         calls.clear()
         self.age(sharesight.performance_cache_file(684141, 0), 0.95 * sharesight.config_performance_cache_seconds) # 95% of its lifetime
-        self.assertEqual(sharesight.warm_once(), 1)
+        self.assertEqual(sharesight.warm()[0], 1)
         self.assertEqual(len(calls), 1)
         self.assertIn('684141/performance', calls[0])
         calls.clear()
         self.age(sharesight.PORTFOLIOS_CACHE, 0.95 * sharesight.config_cache_seconds)
-        self.assertEqual(sharesight.warm_once(), 1)
+        self.assertEqual(sharesight.warm()[0], 1)
         self.assertTrue(calls[0].endswith('/portfolios'))
 
     def test_a_request_never_waits_on_a_cache_the_job_has_warmed(self):
         calls = self.serve_sharesight()
-        sharesight.warm_once()
+        sharesight.warm()[0]
         calls.clear()
         self.assertEqual(sharesight.get_portfolios(), {'Rob': 445825, 'RobSMSF': 684141})
         self.assertEqual(len(sharesight.get_performance(684141, 0)['report']['holdings']), 18)
@@ -332,7 +332,7 @@ def test_a_request_never_waits_on_a_cache_the_job_has_warmed(self):
 
     def test_holdings_wrapper_is_served_entirely_from_the_warmed_cache(self):
         calls = self.serve_sharesight()
-        sharesight.warm_once()
+        sharesight.warm()[0]
         calls.clear()
         tickers = sharesight.get_holdings_wrapper() # get_portfolios() + get_holdings() per portfolio, no cache of its own
         self.assertIn('ARM', tickers)
@@ -340,11 +340,11 @@ def test_holdings_wrapper_is_served_entirely_from_the_warmed_cache(self):
 
     def test_failed_refresh_keeps_the_old_cache(self):
         self.serve_sharesight()
-        sharesight.warm_once()
+        sharesight.warm()[0]
         self.age(sharesight.PORTFOLIOS_CACHE, 0.95 * sharesight.config_cache_seconds)
         self.http(sharesight, ConnectionError('down'))
         with self.assertRaises(RuntimeError):
-            sharesight.warm_once()
+            sharesight.warm()[0]
         self.assertEqual(sharesight.get_portfolios(), {'Rob': 445825, 'RobSMSF': 684141}) # still served from the cache
 
     def test_loop_survives_failures_and_backs_off(self):
@@ -382,7 +382,7 @@ def wait(self, seconds):
 
     def test_a_warming_pass_reads_the_cached_portfolio_list_once(self):
         self.serve_sharesight()
-        sharesight.warm_once() # fill the caches
+        sharesight.warm()[0] # fill the caches
         reads = []
         real = util.read_cache
         self.patch(util, 'read_cache', lambda name, *a, **k: (reads.append(name), real(name, *a, **k))[1])
@@ -395,7 +395,7 @@ def test_a_warming_pass_reads_the_cached_portfolio_list_once(self):
     def test_next_due_is_90_percent_of_the_ttl(self):
         self.serve_sharesight()
         self.assertEqual(sharesight.next_due(), 0) # cold cache
-        sharesight.warm_once()
+        sharesight.warm()[0]
         due = sharesight.next_due()
         self.assertAlmostEqual(due, 0.9 * min(sharesight.config_cache_seconds, sharesight.config_performance_cache_seconds), delta=5)
 
@@ -504,7 +504,7 @@ def test_slack_replies_upload_their_charts(self):
             with self.subTest(cmd=cmd):
                 calls = self.slack_answers(*self.upload_steps())
                 self.sent.clear(); self.errors.clear()
-                worker.process_request('slack', 'C1', '<@U1>', cmd, BOT, 'U', '1')
+                worker.process_request('slack', 'C1', '<@U1>', cmd, BOT, 'U')
                 self.assertEqual(self.errors, [])
                 self.assertEqual([url for url, _ in calls], ['https://slack.com/api/files.getUploadURLExternal', self.UPLOAD_URL,
                                                              'https://slack.com/api/files.completeUploadExternal'])
@@ -554,7 +554,7 @@ def refuse(*a, **k):
         for cmd in ('.beta', '.pie Rob', '.buy', '.sell', '.peg', '.price', '.price top', '.price T1 7d', '.performance 7d', '.shorts', '.history T1', '.compare T1 T2'):
             with self.subTest(cmd=cmd):
                 self.sent.clear(); self.errors.clear()
-                worker.process_request('slack', 'C1', '<@U1>', cmd, BOT, 'U', '1')
+                worker.process_request('slack', 'C1', '<@U1>', cmd, BOT, 'U')
                 self.assertEqual(self.errors, [])
                 self.assertTrue(self.sent)
                 self.assertEqual({kind for kind, *_ in self.sent}, {'text'})
@@ -569,7 +569,7 @@ def test_the_graph_preference_off_sends_no_chart_anywhere(self):
             for cmd in ('.price', '.price top', '.performance 7d', '.shorts', '.buy', '.peg', '.beta', '.compare T1 T2', '.pie Rob'):
                 with self.subTest(service=service, cmd=cmd):
                     self.sent.clear(); self.errors.clear()
-                    worker.process_request(service, chat, '@u', cmd, BOT, 'U', '1')
+                    worker.process_request(service, chat, '@u', cmd, BOT, 'U')
                     self.assertEqual(self.errors, [])
                     self.assertEqual({kind for kind, *_ in self.sent}, {'text'})
         self.assertEqual([url for url in posts if 'sendPhoto' in url or 'files.' in url], [])
@@ -619,7 +619,7 @@ def test_a_reply_without_a_bot_token_is_logged_not_sent(self):
         err = io.StringIO()
         with contextlib.redirect_stderr(err), contextlib.redirect_stdout(io.StringIO()):
             for cmd in ('.help', '.price', '.compare T1 T2'):
-                worker.process_request('slack', 'C1', '<@U1>', cmd, BOT, 'U', '1')
+                worker.process_request('slack', 'C1', '<@U1>', cmd, BOT, 'U')
         self.assertEqual((self.posts, self.errors), ([], []))
         self.assertEqual(err.getvalue().count('ERROR slack: a reply to channel C1 needs slackBotToken in finbot.env; not sent'), 3)
 
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 34869a2a..516db938 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -47,7 +47,7 @@ def test_history_and_compare(self):
             with self.subTest(cmd=cmd):
                 self.assertTrue(self.command(cmd), f'{cmd} sent nothing')
 
-    def test_trades_and_rating_report_when_nothing_is_stored(self):
+    def test_trades_and_watchlist_report_when_nothing_is_stored(self):
         # first run: no state files exist yet, which used to crash these
         for cmd in ('.trades', '.watchlist'):
             self.assertTrue(self.command(cmd), cmd)
@@ -376,10 +376,6 @@ def item(n, date):
         self.assertEqual(self.sent, []) # all seen now
 
 
-if __name__ == '__main__':
-    unittest.main()
-
-
 class PieCommand(FinbotCase):
     def setUp(self):
         super().setUp()
@@ -417,3 +413,7 @@ def test_unknown_or_missing_portfolio_lists_known(self):
     def test_text_only_service(self):
         worker.Request('discord', '1', '@u', 'U', BOT).deliver(['x'], {'pie': True, 'rows': [('a', 1.0)], 'title': 't'})
         self.assertEqual(self.sent[-1][0], 'text')
+
+
+if __name__ == '__main__':
+    unittest.main()
diff --git a/trades.py b/trades.py
index fe5fa02a..7e0fdcd1 100755
--- a/trades.py
+++ b/trades.py
@@ -9,7 +9,7 @@
 from lib import util
 from lib import yahoo
 
-def lambda_handler(chat_id=config_telegramChatID, days=UNSET, service=None, user='', portfolio_select=None, message_id=None, interactive=False):
+def lambda_handler(chat_id=None, days=UNSET, service=None, user='', portfolio_select=None, interactive=False):
 	"""The buys and sells of the past `days`, in every portfolio or the one named: on a scheduled run only those not reported
 	before. Raises sharesight.PortfolioNotFound for a name that is none of them."""
 	if days is UNSET:
@@ -95,16 +95,13 @@ def prepare_trade_payload(service, trades):
 		payload = prepare_trade_payload(service, trades)
 		if interactive and not payload: # no trades, or none of them a buy or a sell
 			payload = [f"{user} No trades {util.days_english(days, 'in the past ')}. {random.choice(noTradesVerb)}"]
-		webhook.send(service, url, chat, payload, message_id=message_id)
+		webhook.send(service, url, chat, payload)
 
 	# write state file
 	if newtrades and not interactive:
 		known_trades.update(newtrades)
 		util.json_write(state_file, list(known_trades), persist=True)
 
-	# make google cloud happy
-	return True
-
 def job(argv):
 	return 'trades.py', {}
 

From 40c51b7a60517897f539263ab1b7b0d22af51191 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Wed, 14 Oct 2026 23:04:22 +0000
Subject: [PATCH 03/81] SIGUSR1 flushes the cache: systemctl kill -s USR1
 finbot, or systemctl start finbot-flush

- bot.py deletes its cache files on SIGUSR1 and keeps its state (watchlist, seen trades, ratings, schedule). Yahoo's cookie and
  crumb go together under their lock, and the Sharesight caches refill at once (keep_warm can now be woken). The signal is
  ignored until startup is done, like SIGHUP, and job processes ignore it, since systemctl kill signals every process of the unit.
- finbot-flush.service, a oneshot unit, gives it a systemctl verb; finbot.service needs no change. README: how to flush under
  systemd, Docker or by hand, and cleanup.py 0 with the bot stopped.
- cleanup leaves a temp file written in the last minute (an atomic write in flight) and still deletes a crash's old leftovers.
- A cache file deleted between the age check and the read is a miss, not an error.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md               |  3 ++-
 README.md               | 12 ++++++++++++
 bot.py                  | 15 +++++++++++++++
 cleanup.py              |  8 ++++++--
 finbot-flush.service    |  7 +++++++
 lib/jobrun.py           |  1 +
 lib/sharesight.py       |  9 ++++++++-
 lib/util.py             | 16 ++++++++++++----
 tests/test_edges.py     | 35 +++++++++++++++++++++++++++++++++++
 tests/test_jobs.py      | 10 ++++++++++
 tests/test_providers.py | 13 +++++++++++++
 tests/test_util.py      | 11 +++++++++++
 12 files changed, 132 insertions(+), 8 deletions(-)
 create mode 100644 finbot-flush.service

diff --git a/CLAUDE.md b/CLAUDE.md
index 3089ef0b..dd018293 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -24,7 +24,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - If Yahoo and Sharesight are unreachable (the cloud environment's network policy can block them), check behaviour against the offline fixtures in `tests/common.py`. For a refactor, compare old and new output: load `git show HEAD:<file>` as a module beside the new one and diff what both produce for the same inputs.
 
 ## Code map (open the one file you need)
-- `bot.py`: WSGI entry; routes /telegram and /slack, authenticates, hands each message to `lib/worker.py` in a thread; SIGHUP reload.
+- `bot.py`: WSGI entry; routes /telegram and /slack, authenticates, hands each message to `lib/worker.py` in a thread; SIGHUP reload, SIGUSR1
+  cache flush.
 - `lib/worker.py`: `COMMANDS` (regex → `on_*` handler, first match wins, `.SYMBOL` last); `Request`; the typing indicator.
 - `lib/reports.py`: `prepare_*` builders for the chat-only reports (lines, plus a chart spec).
 - `price.py performance.py trades.py shorts.py cal.py rating.py milestone.py reminder.py podcasts.py cleanup.py`: reports that also
diff --git a/README.md b/README.md
index 7f2e802c..40eab733 100644
--- a/README.md
+++ b/README.md
@@ -591,6 +591,18 @@ performance_cache_seconds=1800
 
 `performance_cache_seconds` (default 30 minutes) is the cache lifetime of Sharesight performance reports.
 
+To flush the cache, so that everything is fetched afresh, signal the running bot: it deletes its cache files (Yahoo and Sharesight
+answers, charts, the Yahoo cookie) and keeps its state (watchlist, seen trades, ratings, schedule), then refills the Sharesight caches
+at once. There is no `systemctl flush`, since systemctl takes no custom verbs; any of these does it:
+```
+sudo systemctl kill -s USR1 finbot          # systemd
+sudo systemctl start finbot-flush           # the same, once finbot-flush.service is installed (see below)
+docker kill -s USR1 <container>             # Docker
+kill -USR1 <pid of bot.py>                  # anything else
+```
+The log says `cache: flushed N files`. To install the `finbot-flush` shortcut: `sudo cp -v finbot-flush.service /etc/systemd/system/ &&
+sudo systemctl daemon-reload`. `finbot.service` itself needs no change. With the bot stopped, `cleanup.py 0` deletes the same files.
+
 While `bot.py` is running with Sharesight configured and `cache=True` (a reload that switches `cache` pauses or resumes it), a background job keeps the Sharesight portfolio list (`cache_seconds`) and each portfolio's holdings report (1 day, `performance_cache_seconds` cache) refreshed at 90% of their cache lifetime, so chat requests do not wait on Sharesight. If a refresh fails, the previous cache keeps being served and the job retries after 5 minutes.
 
 ### Hyperlinks
diff --git a/bot.py b/bot.py
index 59cd22f3..58a06fef 100755
--- a/bot.py
+++ b/bot.py
@@ -3,6 +3,7 @@
 import signal
 if __name__ == '__main__':
 	signal.signal(signal.SIGHUP, signal.SIG_IGN) # ignored until startup is done (the reload handler is installed last): its default action kills the bot
+	signal.signal(signal.SIGUSR1, signal.SIG_IGN) # likewise the cache flush
 import gevent.monkey
 gevent.monkey.patch_all()
 import html, json, os, re, sys, traceback
@@ -20,6 +21,8 @@
 from lib import net
 from lib import telegram
 from lib import webhook
+from lib import yahoo
+import cleanup
 
 botName = None # the Telegram bot's @username, from telegram.register() at start
 AUTHENTICATED = 'finbot.authenticated' # set in a request's WSGI environ once it has passed its service's check
@@ -195,6 +198,17 @@ def reload_all(*args):
 	if scheduler:
 		scheduler.reload()
 
+def flush_cache(*args):
+	"""SIGUSR1 (`systemctl kill -s USR1 finbot`): delete the cache and keep the state (watchlist, seen trades, ratings, schedule).
+	Yahoo's cookie and crumb belong together, so they go while nothing is renewing them; the Sharesight caches refill at once."""
+	try:
+		with yahoo._auth:
+			removed = cleanup.lambda_handler(days=0)
+		sharesight.wake()
+		print(f"cache: flushed {removed} file{'' if removed == 1 else 's'} from {config_cache_dir}", file=sys.stderr)
+	except Exception as e:
+		print('ERROR', webhook.error_line(e, 'cache flush'), file=sys.stderr)
+
 if __name__ == '__main__':
 	print(f'Starting finbot {git_version()}', file=sys.stderr)
 	try:
@@ -237,6 +251,7 @@ def reload_all(*args):
 		jobs.current = scheduler # for the .jobs chat command
 		threading.Thread(target=scheduler.run_forever, daemon=True).start() # the scheduled reports, formerly cron jobs
 	gevent.signal_handler(signal.SIGHUP, reload_all) # startup is done. systemctl reload finbot: pick up edited preferences.conf and schedule.cron
+	gevent.signal_handler(signal.SIGUSR1, flush_cache) # systemctl kill -s USR1 finbot: delete the cache
 	try:
 		httpd.serve_forever()
 	except OSError as e:
diff --git a/cleanup.py b/cleanup.py
index 95a6bf57..40bafe80 100755
--- a/cleanup.py
+++ b/cleanup.py
@@ -5,8 +5,10 @@
 from lib import webhook
 
 def lambda_handler(days=30):
-	"""Delete the cache files (finbot_*) that were last written more than `days` days ago."""
-	cutoff = time.time() - days * 86400
+	"""Delete the cache files (finbot_*) that were last written more than `days` days ago (days=0: all of them, a flush), but not
+	a temp file written in the last minute: that is an atomic write in flight (util.write_file)."""
+	now = time.time()
+	cutoff = now - days * 86400
 	removed = 0
 	try:
 		entries = list(os.scandir(config_cache_dir))
@@ -14,6 +16,8 @@ def lambda_handler(days=30):
 		return 0
 	for entry in entries:
 		if entry.is_file(follow_symlinks=False) and entry.name.startswith('finbot_') and entry.stat().st_mtime < cutoff:
+			if entry.name.endswith('.tmp') and entry.stat().st_mtime > now - 60:
+				continue
 			try:
 				os.unlink(entry.path)
 				removed += 1
diff --git a/finbot-flush.service b/finbot-flush.service
new file mode 100644
index 00000000..d79c3580
--- /dev/null
+++ b/finbot-flush.service
@@ -0,0 +1,7 @@
+[Unit]
+Description=Flush finbot's cache (sudo systemctl start finbot-flush)
+
+[Service]
+Type=oneshot
+# finbot deletes its cache files on SIGUSR1 and keeps its state; its job processes ignore the signal
+ExecStart=/bin/systemctl kill --signal=USR1 finbot.service
diff --git a/lib/jobrun.py b/lib/jobrun.py
index 1139f4d9..c3bc0d29 100644
--- a/lib/jobrun.py
+++ b/lib/jobrun.py
@@ -5,6 +5,7 @@
 import signal
 if __name__ == '__main__':
 	signal.signal(signal.SIGHUP, signal.SIG_IGN) # a reload is for the bot, but `systemctl kill -s HUP` signals every process of the unit
+	signal.signal(signal.SIGUSR1, signal.SIG_IGN) # likewise a flush (systemctl kill -s USR1); its default action would kill the job
 import importlib, json, re, sys
 
 from lib import webhook
diff --git a/lib/sharesight.py b/lib/sharesight.py
index 824b9257..a39caf34 100644
--- a/lib/sharesight.py
+++ b/lib/sharesight.py
@@ -231,6 +231,12 @@ def next_due(margin=0.9, portfolios=None):
 		waits.append(0 if age is None else margin * config_performance_cache_seconds - age)
 	return max(0, min(waits))
 
+_wake = threading.Event() # set by wake(): the caches were flushed, refill them now
+
+def wake():
+	"""Have keep_warm refill the caches now instead of when they would next be due."""
+	_wake.set()
+
 def keep_warm(min_wait=5, retry=300, idle=60, stop=None):
 	"""Background loop: keep the portfolio list and holdings report cached, sleeping until the next one is due (90% of its TTL).
 	While the cache preference is off (a reload can switch it either way) it fetches nothing, since nothing would be written and
@@ -248,4 +254,5 @@ def keep_warm(min_wait=5, retry=300, idle=60, stop=None):
 		if stop:
 			stop.wait(wait)
 		else:
-			time.sleep(wait)
+			_wake.wait(wait)
+			_wake.clear()
diff --git a/lib/util.py b/lib/util.py
index 22726183..e2b5132b 100644
--- a/lib/util.py
+++ b/lib/util.py
@@ -119,10 +119,19 @@ def _fresh_cache(cacheFile, maxSeconds):
 	print("cache hit:", path, "TTL:", td_to_human(datetime.timedelta(seconds=maxSeconds - age)), file=sys.stderr) if debug else None
 	return path
 
+def _read_fresh(cacheFile, maxSeconds, read):
+	"""read(path) of the cache file if it is fresh, else None; also None if the file goes (cleanup, a flush) after the check."""
+	path = _fresh_cache(cacheFile, maxSeconds)
+	if not path:
+		return None
+	try:
+		return read(Path(path))
+	except FileNotFoundError:
+		return None
+
 def read_cache(cacheFile, maxSeconds=UNSET):
 	"""A cache file's JSON, or None when it is missing or older than maxSeconds."""
-	path = _fresh_cache(cacheFile, maxSeconds)
-	return json.loads(Path(path).read_text(encoding="utf-8")) if path else None
+	return _read_fresh(cacheFile, maxSeconds, lambda path: json.loads(path.read_text(encoding="utf-8")))
 
 def cache_age(cacheFile):
 	"""Seconds since the cache file was written, or None when it does not exist."""
@@ -165,8 +174,7 @@ def json_load(filename, persist=False):
 
 def read_binary_cache(cacheFile, maxSeconds=UNSET):
 	"""A cache file's bytes (as a file object), or None when it is missing or older than maxSeconds."""
-	path = _fresh_cache(cacheFile, maxSeconds)
-	return io.BytesIO(Path(path).read_bytes()) if path else None
+	return _read_fresh(cacheFile, maxSeconds, lambda path: io.BytesIO(path.read_bytes()))
 
 def write_binary_cache(cacheFile, data):
 	data.seek(0)
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 0635972a..7c9c28b3 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -366,6 +366,41 @@ def test_sighup_during_startup_is_ignored_and_works_afterwards(self):
             self.assertIsNone(proc.poll(), ''.join(lines))
             self.assertIn('reload requested but', ''.join(lines))
 
+    def test_usr1_flushes_the_cache_and_keeps_the_state(self):
+        skip_if_old_env(self)
+        with tempfile.TemporaryDirectory() as data:
+            cache = Path(data, 'cache')
+            (cache / 'matplotlib').mkdir(parents=True)
+            for name in ('finbot_yahoo_cookie.json', 'finbot_graph_T1_None.png', 'finbot_x.json.abc.tmp'): # the last: a write in flight
+                (cache / name).write_text('{}')
+            Path(data, 'finbot_watchlist.json').write_text('["T1"]') # state, not cache
+            env = {k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k not in config.STARTUP_KEYS}
+            env.update(DATA_DIR=data, port='0', scheduler='False')
+            proc = subprocess.Popen([sys.executable, '-u', 'bot.py'], cwd=ROOT, env=env, stderr=subprocess.PIPE, stdout=subprocess.DEVNULL, text=True)
+            self.addCleanup(proc.wait)
+            self.addCleanup(proc.kill)
+            deadline = time.monotonic() + 30
+            while not int(Path(f'/proc/{proc.pid}/status').read_text().split('SigIgn:')[1].split()[0], 16) & (1 << (signal.SIGUSR1 - 1)):
+                self.assertLess(time.monotonic(), deadline)
+                time.sleep(0.001)
+            proc.send_signal(signal.SIGUSR1) # still starting: ignored, where its default action would kill the bot
+            lines = []
+            for line in proc.stderr:
+                lines.append(line)
+                if line.startswith('Opening socket'):
+                    break
+            self.assertIsNone(proc.poll(), ''.join(lines))
+            self.assertEqual(len(os.listdir(cache)), 4) # nothing flushed yet
+            time.sleep(1) # the handler goes in just before serving
+            proc.send_signal(signal.SIGUSR1)
+            for line in proc.stderr:
+                lines.append(line)
+                if 'cache: flushed' in line:
+                    break
+            self.assertIn('cache: flushed 2 files', ''.join(lines))
+            self.assertEqual(sorted(os.listdir(cache)), ['finbot_x.json.abc.tmp', 'matplotlib'])
+            self.assertTrue(Path(data, 'finbot_watchlist.json').exists())
+
 
 if __name__ == '__main__':
     unittest.main()
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
index 4ae1cfe1..7526ca57 100644
--- a/tests/test_jobs.py
+++ b/tests/test_jobs.py
@@ -161,6 +161,16 @@ def test_removes_only_old_finbot_cache_files(self):
         self.assertEqual(cleanup.lambda_handler(30), 2)
         self.assertEqual(sorted(os.listdir(folder)), ['finbot_dir', 'finbot_new.json', 'other_old.json'])
 
+    def test_a_temp_file_being_written_is_left_alone(self):
+        folder = tempfile.mkdtemp()
+        self.addCleanup(lambda: __import__('shutil').rmtree(folder, ignore_errors=True))
+        self.patch(cleanup, 'config_cache_dir', folder)
+        for name, stamp in (('finbot_a.json.x1.tmp', time.time() - 5), ('finbot_b.json.x2.tmp', time.time() - 40 * 86400)):
+            Path(folder, name).write_text('x')
+            os.utime(Path(folder, name), (stamp, stamp))
+        self.assertEqual(cleanup.lambda_handler(0), 1) # a flush: the stale one goes, the write in flight stays
+        self.assertEqual(os.listdir(folder), ['finbot_a.json.x1.tmp'])
+
     def test_missing_cache_dir_is_fine(self):
         self.patch(cleanup, 'config_cache_dir', '/nonexistent/finbot-cache')
         self.assertEqual(cleanup.lambda_handler(30), 0)
diff --git a/tests/test_providers.py b/tests/test_providers.py
index ddd3f8dd..b7084738 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -1,6 +1,7 @@
 """Yahoo, Sharesight and Telegram parsing and failure handling, driven by the real API responses saved under doc/.
 HTTP is mocked at lib.net, so the production parsing code runs unchanged."""
 import contextlib, copy, datetime, io, json, os, subprocess, unittest
+import gevent
 from unittest import mock
 from tests.common import *
 
@@ -380,6 +381,18 @@ def wait(self, seconds):
         sharesight.keep_warm(idle=60, stop=stop)
         self.assertEqual((stop.waits, len(passes)), ([60, 60, 3000, 3000], 2)) # idle while off: no Sharesight request every 5 seconds
 
+    def test_a_flush_wakes_the_loop_to_refill_at_once(self):
+        passes = []
+        self.patch(sharesight, 'warm', lambda: (passes.append(1), (0, {}))[1])
+        self.patch(sharesight, 'next_due', lambda portfolios=None: 3000)
+        loop = gevent.spawn(sharesight.keep_warm)
+        self.addCleanup(loop.kill)
+        gevent.sleep(0.05)
+        self.assertEqual(len(passes), 1) # then it sleeps until the next cache is due, 50 minutes away
+        sharesight.wake()
+        gevent.sleep(0.05)
+        self.assertEqual(len(passes), 2)
+
     def test_a_warming_pass_reads_the_cached_portfolio_list_once(self):
         self.serve_sharesight()
         sharesight.warm()[0] # fill the caches
diff --git a/tests/test_util.py b/tests/test_util.py
index f84e2b5c..4b7e03f7 100644
--- a/tests/test_util.py
+++ b/tests/test_util.py
@@ -119,6 +119,17 @@ def test_read_cache_missing_or_expired_is_falsy(self):
         util.json_write('old.json', {'a': 1})
         self.assertFalse(util.read_cache('old.json', -1)) # already expired
 
+    def test_a_cache_file_that_goes_while_it_is_read_is_a_miss(self):
+        util.json_write('gone.json', {'a': 1})
+        real = Path.read_text
+        def vanish(path, *a, **k): # cleanup or a flush deleted it between the age check and the read
+            if path.name == 'gone.json':
+                path.unlink()
+                raise FileNotFoundError(path)
+            return real(path, *a, **k)
+        with mock.patch.object(Path, 'read_text', vanish):
+            self.assertIsNone(util.read_cache('gone.json', 60))
+
     def test_binary_cache_roundtrip(self):
         self.assertFalse(util.read_binary_cache('absent.png'))
         util.write_binary_cache('img.png', io.BytesIO(b'png-bytes'))

From 551caa792469011f8d8e9c13a305e97855b79dc6 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Wed, 14 Oct 2026 23:08:16 +0000
Subject: [PATCH 04/81] Charts drawn off the gevent hub; a retried update runs
 once; report scripts load less; cache-first Yahoo details
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- The bot draws charts in one native thread (util.render_off_hub, from bot.py only). Drawn in the hub, the shorts pictogram froze
  every chat and incoming webhook for 1.3 s; now the hub's worst pause is about 0.1 s. One thread, because several at once
  deadlock matplotlib under gevent's patched locks (tested). matplotlib is imported at start instead of by the first chart.
- A Telegram update or a Slack event that arrives again (Slack resends after 3 s without an answer) runs once, by update_id
  and event_id.
- reports imports pandas and the scheduler where it uses them: a price or performance job process no longer loads pandas and
  every script through charts → reports → jobs (import price: 0.53 s → 0.14 s).
- fetch_detail reads its cache before taking Yahoo's auth lock; data files (the simplywall.st lists, 73 KB, read on every
  profile) are parsed again only when they change.
- A scheduled price report fetches the betas of the small caps that moved together, not one at a time.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md               |  2 ++
 bot.py                  | 24 +++++++++++++++++++++++-
 lib/config.py           | 15 ++++++++++++---
 lib/reports.py          |  6 ++++--
 lib/util.py             | 28 ++++++++++++++++++++++++++++
 lib/yahoo.py            |  2 +-
 price.py                | 19 +++++++++++++------
 tests/common.py         |  3 ++-
 tests/test_charts.py    | 33 ++++++++++++++++++++++++++++++++-
 tests/test_datadir.py   | 13 ++++++++++++-
 tests/test_edges.py     | 19 +++++++++++++++++--
 tests/test_jobs.py      |  8 ++++++++
 tests/test_providers.py |  8 ++++++++
 tests/test_reports.py   | 11 +++++++++++
 14 files changed, 173 insertions(+), 18 deletions(-)

diff --git a/CLAUDE.md b/CLAUDE.md
index dd018293..388aa1cb 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -48,6 +48,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   `webhook.escape(text, service)`, links through `util.link`/`util.finance_link`.
 - Tabs in code; in tests match the file. New and changed lines stay under 120 characters (an edit quotes whole lines).
 - stdout is discarded under systemd (`StandardOutput=null`): anything worth logging goes to stderr.
+- A chart renderer in `lib/util.py` gets `@_drawn` (the bot draws in one native thread); it must not call another `@_drawn`
+  function, or that thread waits on itself.
 - A new command: `COMMANDS`, `prepare_help`, the README command list, and `command_keyboard` if it gets a button.
 
 ## Saving tokens
diff --git a/bot.py b/bot.py
index 58a06fef..da1c69ee 100755
--- a/bot.py
+++ b/bot.py
@@ -6,7 +6,7 @@
 	signal.signal(signal.SIGUSR1, signal.SIG_IGN) # likewise the cache flush
 import gevent.monkey
 gevent.monkey.patch_all()
-import html, json, os, re, sys, traceback
+import collections, html, json, os, re, sys, traceback
 from pathlib import Path
 from gevent import pywsgi
 import threading
@@ -22,11 +22,26 @@
 from lib import telegram
 from lib import webhook
 from lib import yahoo
+from lib import util
 import cleanup
 
 botName = None # the Telegram bot's @username, from telegram.register() at start
 AUTHENTICATED = 'finbot.authenticated' # set in a request's WSGI environ once it has passed its service's check
 SLACK_USER_SUBTYPES = (None, 'thread_broadcast') # Slack message subtypes a person writes; the others are bot posts, edits, joins and the like
+_updates_seen = collections.OrderedDict() # (service, update id) of the last updates taken, oldest first
+
+def seen_before(service, update_id, keep=1000):
+	"""Whether this update was taken already: Telegram's update_id, Slack's event_id. Slack resends an event it got no answer to
+	within 3 seconds and Telegram one that failed, and a command must not run twice. An update without an id is never a repeat."""
+	if update_id is None:
+		return False
+	key = (service, update_id)
+	if key in _updates_seen:
+		return True
+	_updates_seen[key] = None
+	if len(_updates_seen) > keep:
+		_updates_seen.popitem(last=False)
+	return False
 
 def main(environ, start_response):
 	"""WSGI entry point. Never lets an exception escape. A failure handling an authenticated request is reported as one line to the
@@ -84,6 +99,9 @@ def reply(status, body=b''):
 		print(f"[{service}]: request body is not a JSON object", file=sys.stderr)
 		return reply('400 Bad Request', b'<h1>Bad Request</h1>')
 	if service == 'telegram':
+		if seen_before(service, inbound.get('update_id')):
+			print("[telegram]: update", inbound['update_id'], "seen already, not run again", file=sys.stderr)
+			return reply('200 OK')
 		bot_name = botName
 		if "callback_query" in inbound: # chart period buttons
 			cq = inbound["callback_query"]
@@ -136,6 +154,9 @@ def reply(status, body=b''):
 			print("replying with", response, file=sys.stderr)
 			return reply('200 OK', bytes(response, "utf-8"))
 		if inbound['type'] == 'event_callback':
+			if seen_before(service, inbound.get('event_id')):
+				print("[slack]: event", inbound['event_id'], "seen already (a retry), not run again", file=sys.stderr)
+				return reply('200 OK')
 			event = inbound['event']
 			bot_user = inbound['authorizations'][0]['user_id'] # QWERTY
 			if event.get('bot_id') or event.get('subtype') not in SLACK_USER_SUBTYPES or not event.get('user') or event['user'] == bot_user:
@@ -234,6 +255,7 @@ def flush_cache(*args):
 			sys.exit(1)
 	else:
 		print('Scheduled reports are off (scheduler=False)', file=sys.stderr)
+	util.render_off_hub() # charts drawn in a thread of their own: no chat waits while another's chart is drawn
 	httpd = pywsgi.WSGIServer((config_ip, config_port), main)
 	httpd.secure_repr = False if debug else None
 	print(f'Opening socket on http://{config_ip}:{config_port}', file=sys.stderr)
diff --git a/lib/config.py b/lib/config.py
index 3b73061d..3ec3e88b 100644
--- a/lib/config.py
+++ b/lib/config.py
@@ -275,13 +275,22 @@ def data_file(name):
 	user = config_var_dir + '/' + name
 	return user if os.path.isfile(user) else str(REPO_DIR / 'defaults' / name)
 
+_data_files = {} # path -> (modification time, parsed JSON)
+
 def data_json(name, default=None):
-	"""A data file's JSON (the user's copy or the shipped default, as data_file has it), or `default` when there is neither."""
+	"""A data file's JSON (the user's copy or the shipped default, as data_file has it), or `default` when there is neither. Parsed
+	again only when the file changes; callers must not modify what they get."""
 	path = data_file(name)
-	if not os.path.isfile(path):
+	try:
+		mtime = os.path.getmtime(path)
+	except OSError:
 		return default
+	if (cached := _data_files.get(path)) and cached[0] == mtime:
+		return cached[1]
 	with open(path, encoding="utf-8") as f:
-		return json.load(f)
+		value = json.load(f)
+	_data_files[path] = (mtime, value)
+	return value
 
 
 def reload():
diff --git a/lib/reports.py b/lib/reports.py
index d42c06bd..01f8dc45 100755
--- a/lib/reports.py
+++ b/lib/reports.py
@@ -2,9 +2,7 @@
 import hashlib, re
 import datetime
 import sys
-import pandas as pd
 from lib.config import *
-from lib import jobs
 from lib import sharesight
 from lib import util
 from lib import webhook
@@ -146,6 +144,7 @@ def profile_title_sort(e): # disregards markup in sort command
 def prepare_compare(service, args):
 	"""Compare 2-6 tickers as rebased % change. A trailing time period (3m/1y/90d/ytd) is optional;
 	without one, the span is the earliest date covered by every ticker. Returns (caption lines, image, tickers, period days)."""
+	import pandas as pd
 	usage = ['Usage: .compare SYMBOL SYMBOL [SYMBOL...up to 6] [period|max]  (default: 10 years)']
 	period_days = None
 	full_history = False
@@ -219,6 +218,7 @@ def prepare_compare(service, args):
 def prepare_jobs_payload(service, args):
 	"""`.jobs` lists the scheduled commands, each once; `.jobs run <number|command>` runs one now, as cron would.
 	Returns (payload, reply_markup): Telegram gets a button per command."""
+	from lib import jobs # here: jobs imports every script, and some scripts import this module
 	scheduler = jobs.current
 	if scheduler is None:
 		return ['Scheduled jobs are off (scheduler=False)'], None
@@ -243,12 +243,14 @@ def prepare_jobs_payload(service, args):
 
 def run_job_line(scheduler, text):
 	"""Run a scheduled command now and say how it went in one line."""
+	from lib import jobs # here: jobs imports every script, and some scripts import this module
 	try:
 		return job_line(scheduler, *scheduler.run_now(text))
 	except jobs.ScheduleError as e:
 		return str(e)
 
 def job_line(scheduler, label, outcome):
+	from lib import jobs # here: jobs imports every script, and some scripts import this module
 	number = [command[0] for command in scheduler.commands()].index(label) + 1
 	return f"{number}. {jobs.display(label)}: {outcome}"
 
diff --git a/lib/util.py b/lib/util.py
index e2b5132b..7a7f6d47 100644
--- a/lib/util.py
+++ b/lib/util.py
@@ -1,3 +1,4 @@
+import functools
 import os
 import io
 import datetime
@@ -367,6 +368,28 @@ def style_date_axis(ax, start, end, ink):
 
 PALETTE = ('#14181f', '#eef1f5', '#9aa4b2', '#2a313c') # charts: dark surface, ink, muted ink, grid lines
 
+_render_pool = None # set by render_off_hub(): the bot draws its charts in one native thread
+
+def render_off_hub():
+	"""For the bot: draw charts in one native thread from now on, so that the gevent hub keeps serving every other chat while one
+	is drawn (the shorts pictogram takes over a second; in the hub that froze everything). One thread, because several at once
+	deadlock matplotlib under gevent's patched locks. matplotlib is imported here, at start, instead of by the first chart. Job
+	processes and the tests draw in place."""
+	global _render_pool
+	from gevent.threadpool import ThreadPool
+	import matplotlib.backends.backend_agg, matplotlib.dates, matplotlib.figure, matplotlib.patches, matplotlib.path, matplotlib.ticker # noqa: F401
+	_render_pool = ThreadPool(1)
+
+def _drawn(draw):
+	"""A chart function that draws in the render pool when there is one. Only for a function that calls none of the others: the
+	pool's one thread waiting on itself would never finish."""
+	@functools.wraps(draw)
+	def in_pool(*args, **kwargs):
+		if _render_pool is None:
+			return draw(*args, **kwargs)
+		return _render_pool.apply(draw, args, kwargs)
+	return in_pool
+
 def _figure(facecolor):
 	"""A 1280x800 figure (Telegram's photo size, so it is never resampled) and its Agg canvas."""
 	from matplotlib.figure import Figure
@@ -399,6 +422,7 @@ def _png(fig, facecolor):
 	buf.seek(0)
 	return buf
 
+@_drawn
 def graph(df, title, ylabel):
 	"""Render a price chart as PNG bytes, sized for Telegram (1280px wide, no server-side resampling)."""
 	from matplotlib.ticker import MaxNLocator, FuncFormatter
@@ -449,6 +473,7 @@ def annotate(i, name, above):
 
 	return _png(fig, bg)
 
+@_drawn
 def compare_graph(series, title, subtitle=''):
 	"""Rebased % change lines for 2-6 tickers. series: list of (label, pandas Series of closes), already cropped to the period.
 	Returns PNG bytes (1280x800, Telegram's photo size)."""
@@ -502,6 +527,7 @@ def tinted(bg, color, amount=0.07):
 	mix = lambda i: round(int(bg[i:i + 2], 16) * (1 - amount) + int(color[i:i + 2], 16) * amount)
 	return '#{:02x}{:02x}{:02x}'.format(mix(1), mix(3), mix(5))
 
+@_drawn
 def hbar_chart(rows, title, subtitle='', value_fmt=None, xlim=None, session=None):
 	"""Horizontal bars that mirror the caption list, first row on top, filling the whole 1280x800 canvas.
 	Blue for >= 0, red below. Each row's label sits at the bar's base and its value at the tip.
@@ -596,6 +622,7 @@ def shorts_verdict(percent):
 	"""The more shorted, the shorter the shorts."""
 	return next(name for limit, name in SHORTS_VERDICTS if percent < limit)
 
+@_drawn
 def shorts_chart(rows, threshold, step=5):
 	"""Short interest as a pictogram: each row is a run of cartoon shorts, one pair per `step` percent (the last pair cropped).
 	rows: [(label, percent)], largest first. Loud on purpose and drawn from patches, since matplotlib cannot render colour emoji.
@@ -711,6 +738,7 @@ def rows_chart(rows, title, subtitle='', **kw):
 		rows = rows[:20] + rows[-20:]
 	return hbar_chart(rows, title, subtitle, **kw)
 
+@_drawn
 def pie_chart(rows, title, subtitle=''):
 	"""Pie of (label, value) rows, largest first, with a legend of name and share. At most 8 named slices
 	(the dark categorical slots, in fixed order); a slice labelled 'Other' is gray. Returns PNG bytes (1280x800)."""
diff --git a/lib/yahoo.py b/lib/yahoo.py
index 236532b9..743a2978 100644
--- a/lib/yahoo.py
+++ b/lib/yahoo.py
@@ -232,13 +232,13 @@ def fetch_detail(ticker, seconds=UNSET):
 	if seconds is UNSET:
 		seconds = config_cache_seconds
 	now = datetime.datetime.now()
-	cookie, crumb = credentials()
 	cacheFile = "finbot_yahoo_detail_" + ticker + '.json'
 	data = util.read_cache(cacheFile, seconds) if config_cache else None
 	if data:
 		print('.', sep=' ', end='', flush=True)
 	else:
 		print('↓', sep=' ', end='', flush=True)
+		cookie, crumb = credentials() # only on a miss: a cached detail needs neither, nor the lock
 		modules = 'calendarEvents,defaultKeyStatistics,balanceSheetHistoryQuarterly,financialData,summaryProfile,summaryDetail,price,earnings,earningsTrend,insiderTransactions'
 		r = _get_first(_hosts('/v10/finance/quoteSummary/' + ticker + '?modules=' + modules + '&crumb=' + crumb), {**HEADERS, 'Cookie': cookie})
 		if r is None:
diff --git a/price.py b/price.py
index 7fad0339..37be6dae 100755
--- a/price.py
+++ b/price.py
@@ -30,9 +30,15 @@ def gather(specific_stock=None, days=None):
 			yahoo.prefetch_history([t for t in market_data if 'percent_change_period' not in market_data[t]])
 	return tickers, market_data
 
-def trigger_bar(ticker, quote, threshold):
+def needs_beta(quote):
+	"""Whether trigger_bar must know this ticker's beta: a small cap between US$150M and US$1B, which counts as volatile when its
+	beta is over 1.5. The beta is only in Yahoo's details."""
+	return config_demote_volatile and 1.5e8 <= quote.get('market_cap_usd', float('inf')) < 1e9
+
+def trigger_bar(ticker, quote, threshold, details=None):
 	"""The move this ticker needs before it alone sends a scheduled report: more for leveraged ETFs (demote_leveraged) and volatile
-	small caps (demote_volatile; their market caps in US dollars), else the threshold itself."""
+	small caps (demote_volatile; their market caps in US dollars), else the threshold itself. details: Yahoo's details of the
+	tickers that need_beta, fetched together; without it this ticker's are fetched here."""
 	title = quote['profile_title'].lower()
 	if config_demote_leveraged and ('2x' in title or '3x' in title):
 		return threshold * 1.3
@@ -42,7 +48,7 @@ def trigger_bar(ticker, quote, threshold):
 			return threshold * config_volatility_multiplier * 1.3
 		if cap < 1.5e8:
 			return threshold * config_volatility_multiplier
-		beta = yahoo.fetch_detail(ticker).get(ticker, {}).get('beta')
+		beta = (yahoo.fetch_detail(ticker) if details is None else details).get(ticker, {}).get('beta')
 		if beta is None or beta > 1.5:
 			return threshold * config_volatility_multiplier
 	return threshold
@@ -100,9 +106,10 @@ def report(service, threshold=UNSET, user='', specific_stock=None, interactive=F
 			exchange = quote['profile_exchange']
 			exchanges.add(util.exchange_human(exchange))
 			rows.append((ticker, percent, f"{util.get_emoji(percent)} {quote['profile_title']} ({util.quote_link(ticker, exchange, service, days=days)})"))
-			triggered = triggered or abs(percent) >= trigger_bar(ticker, quote, threshold)
-	if not triggered:
-		rows = []
+	if not triggered: # a scheduled report: is there a move big enough for its ticker? Small caps' betas are fetched together
+		details = yahoo.fetch_details([ticker for ticker, _, _ in rows if needs_beta(market_data[ticker])])
+		if not any(abs(percent) >= trigger_bar(ticker, market_data[ticker], threshold, details) for ticker, percent, _ in rows):
+			rows = []
 	rows.sort(key=lambda row: row[1], reverse=True)
 	payload = [f"{line} {util.signed_percent(round(percent))}" for _, percent, line in rows]
 	chart_rows = [(ticker, percent) for ticker, percent, _ in rows]
diff --git a/tests/common.py b/tests/common.py
index 7df12171..39fc46a2 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -1,6 +1,6 @@
 """Shared fixture: points finbot at a temp cache dir and replaces every external service
 (Telegram, Yahoo, Sharesight, Short Man) with canned data. Nothing here touches the network."""
-import io, os, sys, tempfile, time, unittest
+import collections, io, os, sys, tempfile, time, unittest
 import pandas as pd
 from unittest import mock
 
@@ -61,6 +61,7 @@ def setUp(self):
         self.patch(util, 'config_cache_dir', tmp.name)
         self.patch(config, 'config_var_dir', tmp.name)
         self.patch(webhook, 'webhooks', {'telegram': 'https://api.telegram.org/botX/'})
+        self.patch(bot, '_updates_seen', collections.OrderedDict()) # each test starts with no update taken
         for m in (performance, shorts, price, cal, trades, milestone, rating, worker, reminder, telegram):
             self.patch(m, 'webhooks', webhook.webhooks)
         self.sent = []
diff --git a/tests/test_charts.py b/tests/test_charts.py
index a9d33208..7c09862b 100644
--- a/tests/test_charts.py
+++ b/tests/test_charts.py
@@ -4,7 +4,8 @@
   3. one golden image per chart type, compared with a tolerance. Refresh after an intended restyle with:
          UPDATE_GOLDEN=1 python -m unittest tests.test_charts
 """
-import datetime, io, os, unittest
+import datetime, io, os, time, unittest
+import gevent
 from unittest import mock
 import numpy as np
 import pandas as pd
@@ -112,6 +113,36 @@ def test_flat_and_tiny_price_series(self):
         self.check_image(util.graph(two, 'Two points', 'USD'), 'two points')
 
 
+class RenderPool(unittest.TestCase):
+    """The bot draws charts in one native thread (util.render_off_hub), so the gevent hub keeps serving while a chart is drawn."""
+
+    def setUp(self):
+        util.render_off_hub()
+        self.addCleanup(setattr, util, '_render_pool', None)
+
+    def test_charts_asked_for_at_once_all_finish(self):
+        jobs = [gevent.spawn(util.pie_chart, [('a', 1.0), ('b', 2.0)], f't{i}') for i in range(4)]
+        jobs += [gevent.spawn(util.hbar_chart, MIXED, 'T'), gevent.spawn(util.rows_chart, MIXED, 'T')]
+        gevent.joinall(jobs, timeout=120)
+        self.assertTrue(all(job.successful() for job in jobs), 'a chart did not finish (several native threads deadlock matplotlib)')
+        self.assertTrue(all(decode(job.value).size == (1280, 800) for job in jobs))
+
+    def test_the_hub_keeps_serving_while_a_chart_is_drawn(self):
+        gaps, done = [], []
+        def ticker():
+            last = time.monotonic()
+            while not done:
+                gevent.sleep(0.01)
+                now = time.monotonic()
+                gaps.append(now - last)
+                last = now
+        tick = gevent.spawn(ticker)
+        util.shorts_chart([(f'T{i}', 60.0 - 2 * i) for i in range(24)], 15) # about a second of drawing
+        done.append(True)
+        tick.join()
+        self.assertLess(max(gaps), 0.5) # drawn in the hub, the ticker stopped for the whole chart
+
+
 class Contracts(unittest.TestCase):
     def capture_bars(self, rows, **kwargs):
         calls = []
diff --git a/tests/test_datadir.py b/tests/test_datadir.py
index 417b4b05..f59b65af 100644
--- a/tests/test_datadir.py
+++ b/tests/test_datadir.py
@@ -1,4 +1,5 @@
-import os, shutil, subprocess, sys, tempfile, unittest
+import os, shutil, subprocess, sys, tempfile, time, unittest
+from unittest import mock
 from pathlib import Path
 from lib import config
 from util import sync_defaults
@@ -51,6 +52,16 @@ def test_data_file_override_and_fallback(self):
 		Path(config.config_var_dir, name).write_text('{}')
 		self.assertEqual(config.data_file(name), config.config_var_dir + '/' + name)
 
+	def test_a_data_file_is_parsed_again_only_when_it_changes(self):
+		path = Path(config.config_var_dir, 'finbot_adr.json')
+		path.write_text('{"A": "B"}')
+		self.assertEqual(config.data_json('finbot_adr.json'), {'A': 'B'})
+		with mock.patch('json.load', side_effect=AssertionError('parsed again')):
+			self.assertEqual(config.data_json('finbot_adr.json'), {'A': 'B'})
+		path.write_text('{"A": "C"}')
+		os.utime(path, (time.time() + 5, time.time() + 5)) # a later modification time
+		self.assertEqual(config.data_json('finbot_adr.json'), {'A': 'C'})
+
 	def test_sync_defaults(self):
 		self.assertIn(('finbot_adr.json', 'would copy'), sync_defaults.sync())
 		self.assertFalse(Path(config.config_var_dir, 'finbot_adr.json').exists()) # dry run
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 7c9c28b3..76c06df8 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -212,6 +212,15 @@ def test_telegram_edge_updates(self):
         odd = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'new_chat_members': []}}
         self.assertIn(b'Unhandled', self.request(body=odd, headers=secret)[0])
 
+    def test_a_telegram_update_sent_again_runs_once(self):
+        calls = []
+        self.patch(worker, 'process_request', lambda *a: calls.append(a))
+        body = {'update_id': 7, 'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.price'}}
+        for _ in range(2): # Telegram sends an update again when it got no answer
+            self.assertEqual(self.request(body=body, headers=self.SECRET)[1], '200 OK')
+        gevent.sleep(0.05)
+        self.assertEqual(len(calls), 1)
+
     def test_button_presses_are_logged(self):
         calls = []
         self.patch(worker, 'process_callback', lambda *a: calls.append(a))
@@ -234,8 +243,8 @@ def setUp(self):
         self.requests = []
         self.patch(worker, 'process_request', lambda *a: self.requests.append(a))
 
-    def event(self, **event):
-        body = json.dumps({'token': 'stok', 'type': 'event_callback', 'authorizations': [{'user_id': 'UBOT'}],
+    def event(self, event_id=None, **event):
+        body = json.dumps({'token': 'stok', 'type': 'event_callback', 'authorizations': [{'user_id': 'UBOT'}], 'event_id': event_id,
             'event': dict({'type': 'message', 'channel': 'C1', 'ts': '1.2'}, **event)}).encode()
         status = []
         with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
@@ -258,6 +267,12 @@ def test_only_people_are_answered(self):
         self.assertEqual([(r[0], r[1], r[2], r[3], r[4]) for r in self.requests],
                          [('slack', 'C1', '<@U1>', '.price @UBOT', '@UBOT'), ('slack', 'C1', '<@U2>', '.trades', '@UBOT')])
 
+    def test_a_retried_event_runs_once(self):
+        for _ in range(2): # Slack resends an event it got no answer to within 3 seconds, with the same event_id
+            self.assertEqual(self.event('Ev1', user='U1', text='.price'), '200 OK')
+        self.assertEqual(self.event('Ev2', user='U1', text='.price'), '200 OK')
+        self.assertEqual(len(self.requests), 2)
+
     def test_slack_does_not_change_the_telegram_bot_name(self):
         self.patch(bot, 'botName', '@finbot_tg')
         self.event(user='U1', text='.price')
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
index 7526ca57..e95b1b7e 100644
--- a/tests/test_jobs.py
+++ b/tests/test_jobs.py
@@ -630,6 +630,14 @@ def test_scheduler_reports_a_child_that_ended_without_saying_why(self):
                 self.assertEqual(self.errors, [f'⚠️ trades.py: RuntimeError: {reason}'] if reason else [])
                 self.assertEqual(s.busy, set())
 
+    def test_a_report_script_loads_neither_pandas_nor_the_scheduler(self):
+        """A job process imports only its own script: price used to pull in pandas and, through charts, reports and jobs, every
+        script there is."""
+        code = 'import sys, price, performance, trades, shorts, cal; print(sorted(m for m in ("pandas", "lib.jobs", "matplotlib") if m in sys.modules))'
+        with tempfile.TemporaryDirectory() as data:
+            out = subprocess.run([sys.executable, '-c', code], cwd=ROOT, capture_output=True, text=True, env={**os.environ, 'DATA_DIR': data})
+        self.assertEqual(out.stdout.strip().splitlines()[-1:], ['[]'], out.stderr)
+
     def test_the_event_loop_keeps_running_while_a_child_works(self):
         self.sleeper(1)
         ticks = []
diff --git a/tests/test_providers.py b/tests/test_providers.py
index b7084738..5f718171 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -117,6 +117,14 @@ def test_parses_real_quote_summary(self):
         self.assertEqual(data['profile_title'], 'NVIDIA')
         self.assertTrue(data['profile_bio'])
 
+    def test_a_cached_detail_needs_no_cookie(self):
+        self.http(yahoo, Resp(self.combined()))
+        yahoo.fetch_detail('NVDA')
+        def refuse():
+            raise AssertionError('a cached detail took the Yahoo auth lock')
+        self.patch(yahoo, 'credentials', refuse)
+        self.assertIn('NVDA', yahoo.fetch_detail('NVDA'))
+
     def test_http_failure_gives_empty(self):
         for failure in (Resp(status=404), ConnectionError('down')):
             with self.subTest(failure=failure):
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 516db938..d668ff3f 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -100,6 +100,17 @@ def test_a_small_cap_quoted_in_won_needs_the_bigger_move_too(self):
         self.assertAlmostEqual(quote['market_cap_usd'], 1e11 / 1380.0) # US$72M
         self.assertEqual(price.trigger_bar('T1', quote, 5), 5 * price.config_volatility_multiplier)
 
+    def test_the_small_caps_betas_are_fetched_together(self):
+        """A scheduled price report asks for the beta of each small cap that moved (US$150M to US$1B): together, not one by one."""
+        calls = []
+        self.patch(yahoo, 'fetch_details', lambda tickers, strict=False: calls.append(sorted(tickers)) or {t: market_entry(t) for t in tickers})
+        self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: self.fail('a beta fetched on its own'))
+        self.patch(price, 'config_demote_volatile', True)
+        self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | {'market_cap': 5e8} for t in ts})
+        price.lambda_handler()
+        self.assertEqual(len(calls), 1)
+        self.assertTrue(calls[0])
+
     def test_forward_pe_lists_need_as_many_analysts_as_buy(self):
         """A forward P/E is analysts' estimate: with few of them it says little, as a perfect 1.0 buy from two brokers does."""
         analysts = {'T1': 7, 'T2': 6, 'T3': 12, 'T4': 2}

From 8e0989333501379d423065cc15d62839d702d34c Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Wed, 14 Oct 2026 23:11:15 +0000
Subject: [PATCH 05/81] Discord: the native webhook, with charts, masked links
 and no link previews; reports split by size as well as lines

- Discord reports go to the webhook's own endpoint instead of its Slack-compatible one (a /slack suffix in finbot.env is dropped):
  flags=4 turns link previews off (the README's "garbage link previews" limitation is gone), mentions ping nobody, and
  ?wait=true makes Discord answer 200 instead of an empty 204.
- Scheduled reports to Discord carry their charts, uploaded with the report as the message (up to Discord's 2,000 characters).
- Links on Discord are masked links, [text](<url>), with the brackets and parentheses that would end one escaped.
- A long report is split into messages of at most chunk_maxlines lines and the service's character limit, whichever comes
  first: twenty linked lines made a 3,000-character Discord message, which Discord refuses.
- README: the Discord setup and the chat-networks table say what the webhook now does.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 README.md               | 14 ++++----
 lib/config.py           |  2 +-
 lib/util.py             | 18 ++++++----
 lib/webhook.py          | 79 ++++++++++++++++++++++++++++++++++-------
 tests/test_config.py    |  5 +++
 tests/test_finbot.py    |  7 ++++
 tests/test_providers.py | 45 +++++++++++++++++++++--
 tests/test_reports.py   |  5 +--
 8 files changed, 143 insertions(+), 32 deletions(-)

diff --git a/README.md b/README.md
index 40eab733..08764927 100644
--- a/README.md
+++ b/README.md
@@ -30,9 +30,9 @@ What finbot can do on each network. Telegram takes everything; the others take w
 | Chat commands | ✓ | ✓ | ✗ (a webhook only posts) |
 | "typing…" while a command works | ✓ | ✗ (Slack has no typing notice for bots) | ✗ |
 | Charts in replies | ✓ | ✓ with `slackBotToken` | n/a |
-| Charts in scheduled reports | ✓ | ✗ (the incoming webhook takes text only) | ✗ |
+| Charts in scheduled reports | ✓ | ✗ (the incoming webhook takes text only) | ✓ |
 | Period and job buttons | ✓ | ✗ | ✗ |
-| No link previews | ✓ | ✓ | ✗ (set `hyperlink=False`) |
+| No link previews | ✓ | ✓ | ✓ |
 | Linked tickers | ✓ | ✓ | ✓ |
 | Charts updated in place by their buttons | ✓ | ✗ | ✗ |
 
@@ -541,12 +541,13 @@ sharesight_client_secret=01d692d4de7 samplefc64bc2e2f01d692d4de72986ea808f6e9981
 ```
 
 ### Discord
-Discord support is currently only for trade notifications and scheduled reports.
-The Discord webhook can be provisioned in Discord under _Server Settings > Integrations > Webhooks_.
+A Discord webhook takes the trade notifications and scheduled reports, with their charts attached and without link previews.
+Provision it in Discord under _Server Settings > Integrations > Webhooks_.
 Example:
 ```
 discord_webhook=https://discord.com/api/webhooks/1009998000000000000/aaaaaaaaaaaaaaaaaaaaaaaaaa-bbbbbbbbbbbbbb
 ```
+A URL ending in `/slack`, from older setups, works as it is: finbot uses Discord's own endpoint either way.
 
 ### Slack
 A Slack webhook can be provisioned by creating a new app from scratch at https://api.slack.com/apps/ then navigating to _Incoming Webhooks_. Once the link is generated, add it to `finbot.env`. Example:
@@ -608,7 +609,7 @@ While `bot.py` is running with Sharesight configured and `cache=True` (a reload
 ### Hyperlinks
 These `preferences.conf` options relate to hyperlinks:
 
-* `hyperlink`, if set to `False`, disables all hyperlinks. Note: it can not disable syntax highlighting and false positives by chat clients that try to detect domains. Setting it to `False` can be useful for chat networks like Discord that can't easily disable link previews. Defaults to `True`.
+* `hyperlink`, if set to `False`, disables all hyperlinks. Note: it can not disable syntax highlighting and false positives by chat clients that try to detect domains. Defaults to `True`.
 ```
 hyperlink=True
 ```
@@ -750,9 +751,6 @@ You can now monitor the bot's stderr with :
 journalctl -fu finbot
 ```
 
-## Limitations
-* Discord shows garbage link previews from Sharesight. Set `hyperlink=False`, or disable this for each Discord client under _Settings > Text & Images > Embeds and link previews_ (must be repeated for each participating Discord client).
-
 ## Suggestions
 * Is my code is doing something the hard way?
 * Something important is missing from this README?
diff --git a/lib/config.py b/lib/config.py
index 3ec3e88b..e7e3259c 100644
--- a/lib/config.py
+++ b/lib/config.py
@@ -184,7 +184,7 @@ def startup_values(getenv):
 	if getenv('slack_webhook'):
 		hooks['slack'] = getenv('slack_webhook').rstrip('/')
 	if getenv('discord_webhook'):
-		hooks['discord'] = getenv('discord_webhook').rstrip('/').removesuffix('/slack') + '/slack'
+		hooks['discord'] = getenv('discord_webhook').rstrip('/').removesuffix('/slack') # the native endpoint, not the Slack-compatible one
 	if getenv('telegramBotToken'):
 		hooks['telegram'] = 'https://api.telegram.org/bot' + getenv('telegramBotToken').rstrip('/') + '/'
 	values['webhooks'] = hooks
diff --git a/lib/util.py b/lib/util.py
index 7a7f6d47..8a65736a 100644
--- a/lib/util.py
+++ b/lib/util.py
@@ -199,13 +199,17 @@ def security_line(flag, title, link):
 	return f"{flag} {title} ({link})" if flag else f"{title} ({link})"
 
 def link(url, text, service='telegram'):
-	if service == 'telegram' and config_hyperlink:
-		hyperlink = '<a href="' + url + '">' + text + '</a>'
-	elif service in {'discord', 'slack'} and config_hyperlink:
-		hyperlink = '<' + url + '|' + text + '>'
-	else:
-		hyperlink = text
-	return hyperlink
+	"""text linking to url in the service's markup: HTML on Telegram, <url|text> on Slack, a masked link on Discord (the angle
+	brackets keep its preview away; brackets in the text and parentheses in the url would end it early)."""
+	if not config_hyperlink:
+		return text
+	if service == 'telegram':
+		return '<a href="' + url + '">' + text + '</a>'
+	if service == 'slack':
+		return '<' + url + '|' + text + '>'
+	if service == 'discord':
+		return '[' + text.replace('[', '\\[').replace(']', '\\]') + '](<' + url.replace('(', '%28').replace(')', '%29') + '>)'
+	return text
 
 def finance_link(symbol, exchange, service='telegram', days=1, brief=True, text=None):
 	if config_hyperlinkProvider == 'google':
diff --git a/lib/webhook.py b/lib/webhook.py
index 3dc706ab..ba5ce1bc 100644
--- a/lib/webhook.py
+++ b/lib/webhook.py
@@ -3,7 +3,6 @@
 import time
 import requests
 import re
-from itertools import batched
 import sys
 import traceback
 
@@ -11,10 +10,21 @@
 from lib import net
 from lib import telegram
 
+DISCORD_LIMIT = 2000 # characters in a Discord message
+SUPPRESS_EMBEDS = 4 # Discord message flag: no link previews
+NO_MENTIONS = {'parse': []} # Discord: text that looks like @everyone or a mention pings nobody
+
+def discord_url(url):
+	"""A Discord webhook URL that answers with the message posted (200) instead of an empty 204."""
+	return url + ('&' if '?' in url else '?') + 'wait=true'
+
 def write(service, url, payload_string, chat_id=None, reply_markup=None):
 	headers = {'Content-type': 'application/json'}
 	payload = {'text': payload_string}
-	if 'slack.com' in url:
+	if service == 'discord': # the native webhook endpoint: Discord's Slack-compatible one takes no flags
+		payload = {'content': payload_string, 'flags': SUPPRESS_EMBEDS, 'allowed_mentions': NO_MENTIONS}
+		url = discord_url(url)
+	elif 'slack.com' in url:
 		payload['unfurl_links'] = False # no link previews: Slack reads these from the body, not the headers
 		payload['unfurl_media'] = False
 		if chat_id:
@@ -130,11 +140,11 @@ def payload_wrapper(service, url, payload, chat_id=None, reply_markup=None):
 	payload_string = '\n'.join(payload)
 	print("Preparing outbound to", service, str(len(payload_string)), "bytes")
 	print("Payload: " + payload_string) if debug else None
-	limit = 2000 if service == 'discord' else 4096
+	limit = DISCORD_LIMIT if service == 'discord' else 4096
 	if len(payload_string) <= limit:
 		return [write(service, url, payload_string, chat_id, reply_markup)]
 	print(service, f"payload is over {limit:,} bytes. Splitting.")
-	chunks = list(batched(payload, config_chunk_maxlines))
+	chunks = chunked(payload, limit, config_chunk_maxlines)
 	results = []
 	for idx, chunk in enumerate(chunks):
 		if idx:
@@ -143,6 +153,22 @@ def payload_wrapper(service, url, payload, chat_id=None, reply_markup=None):
 		results.append(write(service, url, '\n'.join(chunk), chat_id, reply_markup=reply_markup if last else None))
 	return results
 
+def chunked(lines, limit, max_lines):
+	"""The lines in messages of at most max_lines lines and limit characters each, whichever comes first; a line longer than limit
+	on its own is cut short."""
+	chunks, current, size = [], [], 0
+	for line in lines:
+		if len(line) > limit:
+			line = line[:limit - 1] + '…'
+		if current and (len(current) >= max_lines or size + 1 + len(line) > limit):
+			chunks.append(current)
+			current, size = [], 0
+		size += len(line) + (1 if current else 0)
+		current.append(line)
+	if current:
+		chunks.append(current)
+	return chunks
+
 MARKUP = { # how each service marks text up: style -> service -> (before, after)
 	'bold': {'telegram': ('<b>', '</b>'), 'slack': ('*', '*'), 'discord': ('**', '**')},
 	'italic': {'telegram': ('<i>', '</i>'), 'slack': ('_', '_'), 'discord': ('_', '_')},
@@ -172,10 +198,12 @@ def strike(message, service):
 	return _styled('strike', message, service)
 
 
-def fit_caption(caption, limit=1024):
-	"""Telegram rejects photo captions over 1024 visible characters. Keep whole lines that fit and say how many were left out."""
-	def visible(text):
-		return len(re.sub(r'<[^>]+>', '', text))
+def fit_caption(caption, limit=1024, visible=None):
+	"""Telegram rejects photo captions over 1024 visible characters; Discord takes 2,000 characters, markup included
+	(visible=len). Keep whole lines that fit and say how many were left out."""
+	if visible is None:
+		def visible(text):
+			return len(re.sub(r'<[^>]+>', '', text))
 	if visible(caption) <= limit:
 		return caption
 	lines = caption.split('\n')
@@ -191,10 +219,10 @@ def visible(text):
 MAYBE_SENT = object() # a read timeout on the request that posts a chart: it may have arrived, so no text copy follows
 
 def takes_charts(service, reply=True):
-	"""Whether a report to `service` can carry a chart image: never with the graph preference off; on Telegram always; on Slack in
-	a reply to a chat (`reply`), uploaded into its channel with the bot token, but not in a scheduled report, which goes through the
-	incoming webhook and takes text only; never on Discord. Elsewhere a report goes out as text."""
-	return config_graph and (service == 'telegram' or (service == 'slack' and reply and bool(config_slackBotToken)))
+	"""Whether a report to `service` can carry a chart image: never with the graph preference off; on Telegram and Discord always;
+	on Slack in a reply to a chat (`reply`), uploaded into its channel with the bot token, but not in a scheduled report, which goes
+	through the incoming webhook and takes text only. Elsewhere a report goes out as text."""
+	return config_graph and (service in ('telegram', 'discord') or (service == 'slack' and reply and bool(config_slackBotToken)))
 
 def telegram_photo(chat_id, image_data, caption, reply_markup=None):
 	"""sendPhoto, the report as its caption (trimmed to fit). Telegram's answer, MAYBE_SENT, or None if it was refused."""
@@ -227,6 +255,31 @@ def telegram_photo(chat_id, image_data, caption, reply_markup=None):
 	print(r.status_code, "error telegram sendPhoto", output.get('error_code'), output.get('description') or r.text[:300], file=sys.stderr)
 	return None
 
+def discord_photo(image_data, caption):
+	"""The chart on the Discord webhook, with the report as its message (trimmed to Discord's 2,000 characters), no link previews.
+	Discord's answer, MAYBE_SENT, or None if it was refused."""
+	if hasattr(image_data, 'seek'):
+		image_data.seek(0)
+	payload = {'content': fit_caption(caption, DISCORD_LIMIT, len), 'flags': SUPPRESS_EMBEDS, 'allowed_mentions': NO_MENTIONS,
+		'attachments': [{'id': 0, 'filename': 'chart.png'}]}
+	try:
+		r = net.post(discord_url(webhooks['discord']), data={'payload_json': json.dumps(payload)},
+			files={'files[0]': ('chart.png', image_data, 'image/png')}, timeout=config_http_timeout)
+	except requests.exceptions.ReadTimeout as e:
+		print("Discord chart timed out, it may still arrive:", net.redact(e), file=sys.stderr)
+		return MAYBE_SENT
+	except Exception as e:
+		print("Discord chart failed:", net.redact(e), file=sys.stderr)
+		return None
+	if r.status_code == 200:
+		print(r.status_code, "OK discord chart", file=sys.stderr)
+		try:
+			return r.json() or {}
+		except ValueError:
+			return {}
+	print(r.status_code, "error discord chart", net.redact(r.text[:300]), file=sys.stderr)
+	return None
+
 def slack_api(method, **data):
 	"""POST a Slack Web API method as the bot, form-encoded, and return its answer. Raises RuntimeError with Slack's error if not ok."""
 	if not config_slackBotToken:
@@ -272,6 +325,8 @@ def sendPhoto(chat_id, image_data, caption, service, reply_markup=None):
 	chartable = takes_charts(service, reply=bool(chat_id))
 	if chartable and service == 'telegram':
 		output = telegram_photo(chat_id, image_data, caption, reply_markup)
+	elif chartable and service == 'discord':
+		output = discord_photo(image_data, caption)
 	elif chartable:
 		output = slack_photo(chat_id, image_data, caption)
 	if output is MAYBE_SENT:
diff --git a/tests/test_config.py b/tests/test_config.py
index 6f9703f8..431ab57a 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -335,6 +335,11 @@ def test_a_leftover_env_in_the_checkout_refuses_to_start(self):
 		self.assertIn(f'{old} is the old single config file, which finbot no longer reads', caught.exception.errors[0])
 		self.assertIn('/util/split_env.py, which writes finbot.env.new and preferences.conf.new', caught.exception.errors[1])
 
+	def test_a_discord_webhook_uses_the_native_endpoint(self):
+		for url in ('https://discord.com/api/webhooks/1/abc', 'https://discord.com/api/webhooks/1/abc/slack/'): # the old setup's suffix
+			values = config.startup_values(lambda name, default=None: url if name == 'discord_webhook' else default)
+			self.assertEqual(values['webhooks']['discord'], 'https://discord.com/api/webhooks/1/abc')
+
 	def test_slack_commands_without_a_bot_token_are_warned_about(self):
 		with mock.patch.object(config, 'config_slackOutgoingWebhook', 'https://example.com/slack'):
 			with mock.patch.object(config, 'config_slackBotToken', False):
diff --git a/tests/test_finbot.py b/tests/test_finbot.py
index df4c8614..5eec41e3 100644
--- a/tests/test_finbot.py
+++ b/tests/test_finbot.py
@@ -25,6 +25,13 @@ def post(url, **kwargs):
         self.assertIn('reply_markup', posted[-1])
         self.assertTrue(all('reply_markup' not in p for p in posted[:-1]))
 
+    def test_chunks_respect_both_the_line_count_and_the_size(self):
+        chunks = webhook.chunked(['a' * 30] * 10 + ['b' * 500], limit=100, max_lines=4)
+        self.assertEqual([len(chunk) for chunk in chunks], [3, 3, 3, 1, 1]) # 3 x 30 plus newlines is 92; a fourth would pass 100
+        self.assertTrue(all(len('\n'.join(chunk)) <= 100 for chunk in chunks))
+        self.assertEqual(chunks[-1], ['b' * 99 + '…']) # a line over the limit on its own is cut short
+        self.assertEqual(webhook.chunked(['x'] * 9, limit=4096, max_lines=4), [['x'] * 4, ['x'] * 4, ['x']])
+
     def test_zoneinfo_offset_is_not_lmt(self):
         tz = zoneinfo.ZoneInfo('Australia/Melbourne')
         self.assertEqual(datetime.datetime(2022, 8, 18, tzinfo=tz).utcoffset(), datetime.timedelta(hours=10))
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 5f718171..6af35bda 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -514,8 +514,8 @@ def upload_steps(self):
 
     def test_where_charts_can_go(self):
         self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
-        cases = (('telegram',), ('telegram', False), ('slack',), ('slack', False), ('discord',))
-        self.assertEqual([webhook.takes_charts(*case) for case in cases], [True, True, True, False, False]) # Slack: replies, not scheduled reports
+        cases = (('telegram',), ('telegram', False), ('slack',), ('slack', False), ('discord', False))
+        self.assertEqual([webhook.takes_charts(*case) for case in cases], [True, True, True, False, True]) # Slack: replies, not scheduled reports
         self.patch(webhook, 'config_slackBotToken', False)
         self.assertFalse(webhook.takes_charts('slack'))
 
@@ -615,6 +615,47 @@ def fake(url, **kwargs):
                 self.assertEqual(self.sent, expect)
 
 
+class DiscordWebhook(FinbotCase):
+    """Discord's own webhook endpoint (the Slack-compatible one takes no flags): no link previews, no pings, charts attached."""
+    URL = 'https://discord.com/api/webhooks/1/abc'
+
+    def setUp(self):
+        super().setUp()
+        self.patch(webhook, 'webhooks', {'discord': self.URL})
+        self.patch(webhook, 'payload_wrapper', REAL_PAYLOAD_WRAPPER)
+        self.patch(webhook, 'sendPhoto', REAL_SEND_PHOTO)
+        self.posts = []
+        self.patch(webhook.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or Resp({'id': '1'}))
+
+    def test_text_goes_to_the_native_endpoint_without_previews_or_pings(self):
+        webhook.write('discord', self.URL, 'hello @everyone')
+        url, kwargs = self.posts[0]
+        self.assertEqual(url, self.URL + '?wait=true') # without wait Discord answers 204, which counted as a failure
+        self.assertEqual(kwargs['json'], {'content': 'hello @everyone', 'flags': 4, 'allowed_mentions': {'parse': []}})
+
+    def test_a_scheduled_report_carries_its_chart(self):
+        price.lambda_handler()
+        (url, kwargs), = self.posts
+        self.assertEqual(url, self.URL + '?wait=true')
+        payload = json.loads(kwargs['data']['payload_json'])
+        self.assertEqual((payload['flags'], payload['allowed_mentions'], payload['attachments'][0]['filename']), (4, {'parse': []}, 'chart.png'))
+        self.assertTrue(payload['content'].startswith('**5% at close'), payload['content'][:60])
+        self.assertLessEqual(len(payload['content']), 2000)
+        name, image, kind = kwargs['files']['files[0]']
+        self.assertEqual((name, kind), ('chart.png', 'image/png'))
+        self.assertTrue(image.read().startswith(b'\x89PNG'))
+
+    def test_a_long_report_is_split_by_size(self):
+        line = 'x' * 150
+        webhook.payload_wrapper('discord', self.URL, [line] * 30)
+        sizes = [len(kwargs['json']['content']) for _, kwargs in self.posts]
+        self.assertGreater(len(sizes), 2)
+        self.assertTrue(all(size <= 2000 for size in sizes), sizes) # 20 lines a chunk used to make 3,019 characters
+
+    def test_links_are_masked_links(self):
+        self.assertEqual(util.link('https://x.example/a(b)', 'T[1]', 'discord'), '[T\\[1\\]](<https://x.example/a%28b%29>)')
+
+
 class SlackSending(FinbotCase):
     """Slack replies go through chat.postMessage with the bot token; scheduled reports through the incoming webhook."""
 
diff --git a/tests/test_reports.py b/tests/test_reports.py
index d668ff3f..49c06a3d 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -421,8 +421,9 @@ def test_unknown_or_missing_portfolio_lists_known(self):
             self.assertTrue(sent[0][1][0].startswith(first), text)
             self.assertIn('Rob', sent[0][1])
 
-    def test_text_only_service(self):
-        worker.Request('discord', '1', '@u', 'U', BOT).deliver(['x'], {'pie': True, 'rows': [('a', 1.0)], 'title': 't'})
+    def test_a_service_that_takes_no_images_gets_text(self):
+        self.patch(webhook, 'config_slackBotToken', False) # Slack replies carry charts only with the bot token
+        worker.Request('slack', 'C1', '@u', 'U', BOT).deliver(['x'], {'pie': True, 'rows': [('a', 1.0)], 'title': 't'})
         self.assertEqual(self.sent[-1][0], 'text')
 
 

From e8b635df818bc5d8e0e227e2713d543cf5d03410 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Wed, 14 Oct 2026 23:13:02 +0000
Subject: [PATCH 06/81] Tabs everywhere: the tests, shortman.py and sws_gen.py
 re-indented; .editorconfig

A whitespace-only change (git diff -w shows nothing but .editorconfig, .gitignore and CLAUDE.md): the 4-space files now indent
with tabs like the rest of the code. Lines inside multi-line strings that are data were left exactly as they were. One style
means an edit's quoted indentation always matches.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 .editorconfig           |   14 +
 .gitignore              |    1 +
 CLAUDE.md               |    2 +-
 lib/shortman.py         |   48 +-
 tests/common.py         |  142 ++--
 tests/test_charts.py    |  508 +++++++-------
 tests/test_edges.py     |  748 ++++++++++----------
 tests/test_finbot.py    |  176 ++---
 tests/test_jobs.py      | 1232 ++++++++++++++++-----------------
 tests/test_providers.py | 1428 +++++++++++++++++++--------------------
 tests/test_reports.py   |  812 +++++++++++-----------
 tests/test_util.py      |  264 ++++----
 util/sws_gen.py         |   84 +--
 13 files changed, 2737 insertions(+), 2722 deletions(-)
 create mode 100644 .editorconfig

diff --git a/.editorconfig b/.editorconfig
new file mode 100644
index 00000000..1951c484
--- /dev/null
+++ b/.editorconfig
@@ -0,0 +1,14 @@
+root = true
+
+[*]
+charset = utf-8
+end_of_line = lf
+insert_final_newline = true
+
+[*.py]
+indent_style = tab
+indent_size = 4
+
+[*.{yml,yaml}]
+indent_style = space
+indent_size = 2
diff --git a/.gitignore b/.gitignore
index 4e27a261..dea55888 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,5 +1,6 @@
 .*
 !.gitignore
+!.editorconfig
 !.python-version
 !.claude
 !.github
diff --git a/CLAUDE.md b/CLAUDE.md
index 388aa1cb..b0ce1f7e 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -46,7 +46,7 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 ## Conventions
 - A report is a list of lines. Markup through `webhook.bold/italic(text, service)`, user-written text through
   `webhook.escape(text, service)`, links through `util.link`/`util.finance_link`.
-- Tabs in code; in tests match the file. New and changed lines stay under 120 characters (an edit quotes whole lines).
+- Tabs everywhere, tests included (`.editorconfig`). New and changed lines stay under 120 characters (an edit quotes whole lines).
 - stdout is discarded under systemd (`StandardOutput=null`): anything worth logging goes to stderr.
 - A chart renderer in `lib/util.py` gets `@_drawn` (the bot draws in one native thread); it must not call another `@_drawn`
   function, or that thread waits on itself.
diff --git a/lib/shortman.py b/lib/shortman.py
index 46ca8727..e1f1a4c4 100644
--- a/lib/shortman.py
+++ b/lib/shortman.py
@@ -3,27 +3,27 @@
 from lib import net
 
 def fetch(market_data):
-    local_market_data = market_data.copy()
-    print("Fetching ASX shorts from Shortman")
-    url = 'https://www.shortman.com.au/downloadeddata/latest.csv'
-    try:
-        r = net.get(url, timeout=config_http_timeout)
-    except Exception as e:
-        print("Failure fetching", url, net.redact(e), file=sys.stderr)
-        return market_data # the ASX short interest is missing, the rest of the report is not
-    if r.status_code == 200:
-        print(r.status_code, "success shortman")
-    else:
-        print(r.status_code, "error communicating with", url, file=sys.stderr)
-        return market_data
-    csv = r.content.decode('utf-8')
-    csv = csv.split('\r\n')
-    csv.pop(0) # remove header
-    del csv[-1] # remove junk
-    for line in csv:
-        cells = line.split(',') # title, code, positions, on issue, percent shorted
-        ticker = cells[1] + '.AX'
-        short_percent = float(cells[4]) # every row, so a malformed one still fails the fetch
-        if ticker in market_data:
-            local_market_data[ticker]['short_percent'] = short_percent
-    return local_market_data
+	local_market_data = market_data.copy()
+	print("Fetching ASX shorts from Shortman")
+	url = 'https://www.shortman.com.au/downloadeddata/latest.csv'
+	try:
+		r = net.get(url, timeout=config_http_timeout)
+	except Exception as e:
+		print("Failure fetching", url, net.redact(e), file=sys.stderr)
+		return market_data # the ASX short interest is missing, the rest of the report is not
+	if r.status_code == 200:
+		print(r.status_code, "success shortman")
+	else:
+		print(r.status_code, "error communicating with", url, file=sys.stderr)
+		return market_data
+	csv = r.content.decode('utf-8')
+	csv = csv.split('\r\n')
+	csv.pop(0) # remove header
+	del csv[-1] # remove junk
+	for line in csv:
+		cells = line.split(',') # title, code, positions, on issue, percent shorted
+		ticker = cells[1] + '.AX'
+		short_percent = float(cells[4]) # every row, so a malformed one still fails the fetch
+		if ticker in market_data:
+			local_market_data[ticker]['short_percent'] = short_percent
+	return local_market_data
diff --git a/tests/common.py b/tests/common.py
index 39fc46a2..915e647b 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -14,7 +14,7 @@
 import performance, shorts, price, reminder, cal, trades, milestone, rating
 
 REAL = {(m.__name__, n): getattr(m, n) for m, n in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (yahoo, 'price_history'),
-    (sharesight, 'get_portfolios'), (sharesight, 'get_trades'), (sharesight, 'get_performance'), (sharesight, 'get_performance_wrapper'))}
+	(sharesight, 'get_portfolios'), (sharesight, 'get_trades'), (sharesight, 'get_performance'), (sharesight, 'get_performance_wrapper'))}
 REAL_GET_HOLDINGS = util.get_holdings_and_watchlist
 REAL_PAYLOAD_WRAPPER = webhook.payload_wrapper # FinbotCase replaces it; tests of the real thing use this
 REAL_REPORT_ERROR = webhook.report_error # likewise
@@ -24,86 +24,86 @@
 
 
 def skip_if_old_env(case):
-    """For the tests that start bot.py: it will not start while the checkout has a leftover .env (config.check_startup_environment)."""
-    if config.OLD_ENV.exists():
-        case.skipTest(f'bot.py will not start while {config.OLD_ENV} exists')
+	"""For the tests that start bot.py: it will not start while the checkout has a leftover .env (config.check_startup_environment)."""
+	if config.OLD_ENV.exists():
+		case.skipTest(f'bot.py will not start while {config.OLD_ENV} exists')
 
 
 STATES = ('REGULAR', 'PRE', 'POST', 'CLOSED')
 
 
 def market_entry(ticker):
-    i = int(ticker[1:]) if ticker[1:].isdigit() else 0 # varied data so threshold/sort/sell branches all have rows
-    return {'marketState': STATES[i % 4], 'regularMarketTime': time.time(), 'quoteType': 'EQUITY',
-            'profile_title': 'Co ' + ticker, 'profile_exchange': 'NasdaqGS', 'percent_change': (i - 12) * 1.5, 'percent_change_premarket': (i - 12) * 1.5, 'percent_change_postmarket': (i - 12) * 1.5,
-            'market_cap': 1e9, 'price_to_earnings_trailing': 9, 'price_to_earnings_forward': 9,
-            'beta': 2.0, 'short_percent': 20, 'fiftyTwoWeekHigh': 120.0, 'fiftyTwoWeekLow': 80.0,
-            'regularMarketPrice': 100.0, 'earnings_date': time.time() + 86400 * 3,
-            'ex_dividend_date': time.time() + 86400 * 3, 'recommend': 'buy' if i % 2 else 'sell', 'recommend_index': 1.2 if i % 2 else 4.2,
-            'recommend_analysts': 8, 'price_to_earnings_peg': 1.2, 'currency': 'USD'}
+	i = int(ticker[1:]) if ticker[1:].isdigit() else 0 # varied data so threshold/sort/sell branches all have rows
+	return {'marketState': STATES[i % 4], 'regularMarketTime': time.time(), 'quoteType': 'EQUITY',
+			'profile_title': 'Co ' + ticker, 'profile_exchange': 'NasdaqGS', 'percent_change': (i - 12) * 1.5, 'percent_change_premarket': (i - 12) * 1.5, 'percent_change_postmarket': (i - 12) * 1.5,
+			'market_cap': 1e9, 'price_to_earnings_trailing': 9, 'price_to_earnings_forward': 9,
+			'beta': 2.0, 'short_percent': 20, 'fiftyTwoWeekHigh': 120.0, 'fiftyTwoWeekLow': 80.0,
+			'regularMarketPrice': 100.0, 'earnings_date': time.time() + 86400 * 3,
+			'ex_dividend_date': time.time() + 86400 * 3, 'recommend': 'buy' if i % 2 else 'sell', 'recommend_index': 1.2 if i % 2 else 4.2,
+			'recommend_analysts': 8, 'price_to_earnings_peg': 1.2, 'currency': 'USD'}
 
 
 class FinbotCase(unittest.TestCase):
-    """Base class: self.sent collects ('photo'|'text', ...) tuples for everything the bot tried to send."""
+	"""Base class: self.sent collects ('photo'|'text', ...) tuples for everything the bot tried to send."""
 
-    def patch(self, target, attr, value):
-        p = mock.patch.object(target, attr, value)
-        p.start()
-        self.addCleanup(p.stop)
+	def patch(self, target, attr, value):
+		p = mock.patch.object(target, attr, value)
+		p.start()
+		self.addCleanup(p.stop)
 
-    def setUp(self):
-        fake = mock.Mock(status_code=200, text='{}')
-        fake.json.return_value = {'ok': True, 'result': {'message_id': 1, 'chat': {'id': 5}}}
-        self.patch(__import__('requests').Session, 'request', mock.Mock(return_value=fake)) # no real network traffic
-        tmp = tempfile.TemporaryDirectory()
-        self.addCleanup(tmp.cleanup)
-        self.patch(util, 'config_var_dir', tmp.name)
-        self.patch(util, 'config_cache_dir', tmp.name)
-        self.patch(config, 'config_var_dir', tmp.name)
-        self.patch(webhook, 'webhooks', {'telegram': 'https://api.telegram.org/botX/'})
-        self.patch(bot, '_updates_seen', collections.OrderedDict()) # each test starts with no update taken
-        for m in (performance, shorts, price, cal, trades, milestone, rating, worker, reminder, telegram):
-            self.patch(m, 'webhooks', webhook.webhooks)
-        self.sent = []
-        self.markups = [] # reply_markup of each photo sent
-        self.errors = []
-        self.patch(webhook, 'report_error', lambda e, *a, **k: self.errors.append(webhook.error_line(e, k.get('context'))))
-        self.patch(webhook, 'sendPhoto', lambda chat, img, cap, svc, **k: (self.sent.append(('photo', len(img.read()), cap.split('\n')[0])), self.markups.append(k.get('reply_markup'))))
-        self.patch(webhook, 'payload_wrapper', lambda svc, url, payload, *a, **k: self.sent.append(('text', payload)) or [])
-        self.patch(util, 'get_holdings_and_watchlist', lambda: TICKERS)
-        self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts})
-        self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t)})
-        self.patch(yahoo, 'price_history', lambda t, days=None, **k: ({days: 1.5} if days else {'Max': 5, '1Y': 3}, None))
-        self.patch(yahoo, 'prefetch_history', lambda ts: None)
-        self.patch(yahoo, 'historic_high', lambda t: (100.0, 50.0))
-        dates = pd.date_range(end=pd.Timestamp.today().normalize(), periods=400)
-        self.patch(yahoo, 'price_series', lambda t: ('Co ' + t, 'NASDAQ', pd.Series(range(100, 500), index=dates, name=t, dtype=float)))
-        self.patch(shortman, 'fetch', lambda md: md)
-        holdings = [{'id': i, 'instrument': {'code': t, 'market_code': 'NASDAQ'}, 'portfolio': {'name': 'Rob'},
-                     'capital_gain_percent': str(i * 4 - 40), 'value': '1000'} for i, t in enumerate(TICKERS)]
-        self.patch(sharesight, 'get_performance_wrapper', lambda d: {1: {'report': {'holdings': holdings}}})
-        self.patch(sharesight, 'get_portfolios', lambda: {'Rob': 1})
-        self.patch(sharesight, 'get_performance', lambda pid, d, *a: {'report': {
-            'holdings': holdings, 'currency_gain_percent': '0', 'capital_gain_percent': '2.5', 'total_gain_percent': '0'}})
-        self.patch(sharesight, 'get_trades', lambda name, pid, days=7: [{
-            'id': 1, 'transaction_type': 'BUY', 'symbol': 'T1', 'market': 'NASDAQ', 'portfolio': name,
-            'transaction_date': time.strftime('%Y-%m-%d'), 'quantity': '10', 'price': '5.0', 'value': '50', 'brokerage_currency_code': 'USD', 'holding_id': 7}])
-        self.typing = [] # every typing indicator a test starts; none may still be running when the test ends
-        real_start = worker.TypingIndicator.start
-        self.patch(worker.TypingIndicator, 'start', lambda indicator: (self.typing.append(indicator), real_start(indicator)))
-        self.addCleanup(self.no_typing_left)
+	def setUp(self):
+		fake = mock.Mock(status_code=200, text='{}')
+		fake.json.return_value = {'ok': True, 'result': {'message_id': 1, 'chat': {'id': 5}}}
+		self.patch(__import__('requests').Session, 'request', mock.Mock(return_value=fake)) # no real network traffic
+		tmp = tempfile.TemporaryDirectory()
+		self.addCleanup(tmp.cleanup)
+		self.patch(util, 'config_var_dir', tmp.name)
+		self.patch(util, 'config_cache_dir', tmp.name)
+		self.patch(config, 'config_var_dir', tmp.name)
+		self.patch(webhook, 'webhooks', {'telegram': 'https://api.telegram.org/botX/'})
+		self.patch(bot, '_updates_seen', collections.OrderedDict()) # each test starts with no update taken
+		for m in (performance, shorts, price, cal, trades, milestone, rating, worker, reminder, telegram):
+			self.patch(m, 'webhooks', webhook.webhooks)
+		self.sent = []
+		self.markups = [] # reply_markup of each photo sent
+		self.errors = []
+		self.patch(webhook, 'report_error', lambda e, *a, **k: self.errors.append(webhook.error_line(e, k.get('context'))))
+		self.patch(webhook, 'sendPhoto', lambda chat, img, cap, svc, **k: (self.sent.append(('photo', len(img.read()), cap.split('\n')[0])), self.markups.append(k.get('reply_markup'))))
+		self.patch(webhook, 'payload_wrapper', lambda svc, url, payload, *a, **k: self.sent.append(('text', payload)) or [])
+		self.patch(util, 'get_holdings_and_watchlist', lambda: TICKERS)
+		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts})
+		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t)})
+		self.patch(yahoo, 'price_history', lambda t, days=None, **k: ({days: 1.5} if days else {'Max': 5, '1Y': 3}, None))
+		self.patch(yahoo, 'prefetch_history', lambda ts: None)
+		self.patch(yahoo, 'historic_high', lambda t: (100.0, 50.0))
+		dates = pd.date_range(end=pd.Timestamp.today().normalize(), periods=400)
+		self.patch(yahoo, 'price_series', lambda t: ('Co ' + t, 'NASDAQ', pd.Series(range(100, 500), index=dates, name=t, dtype=float)))
+		self.patch(shortman, 'fetch', lambda md: md)
+		holdings = [{'id': i, 'instrument': {'code': t, 'market_code': 'NASDAQ'}, 'portfolio': {'name': 'Rob'},
+					 'capital_gain_percent': str(i * 4 - 40), 'value': '1000'} for i, t in enumerate(TICKERS)]
+		self.patch(sharesight, 'get_performance_wrapper', lambda d: {1: {'report': {'holdings': holdings}}})
+		self.patch(sharesight, 'get_portfolios', lambda: {'Rob': 1})
+		self.patch(sharesight, 'get_performance', lambda pid, d, *a: {'report': {
+			'holdings': holdings, 'currency_gain_percent': '0', 'capital_gain_percent': '2.5', 'total_gain_percent': '0'}})
+		self.patch(sharesight, 'get_trades', lambda name, pid, days=7: [{
+			'id': 1, 'transaction_type': 'BUY', 'symbol': 'T1', 'market': 'NASDAQ', 'portfolio': name,
+			'transaction_date': time.strftime('%Y-%m-%d'), 'quantity': '10', 'price': '5.0', 'value': '50', 'brokerage_currency_code': 'USD', 'holding_id': 7}])
+		self.typing = [] # every typing indicator a test starts; none may still be running when the test ends
+		real_start = worker.TypingIndicator.start
+		self.patch(worker.TypingIndicator, 'start', lambda indicator: (self.typing.append(indicator), real_start(indicator)))
+		self.addCleanup(self.no_typing_left)
 
-    def no_typing_left(self):
-        left = [indicator for indicator in self.typing if indicator.is_active()]
-        for indicator in left:
-            indicator.stop()
-        self.assertEqual(left, [], 'a typing indicator was left running')
+	def no_typing_left(self):
+		left = [indicator for indicator in self.typing if indicator.is_active()]
+		for indicator in left:
+			indicator.stop()
+		self.assertEqual(left, [], 'a typing indicator was left running')
 
-    def command(self, text, chat='55'):
-        """Run a chat message through the real command parser and handlers."""
-        self.sent.clear()
-        self.markups.clear()
-        self.errors.clear()
-        worker.process_request('telegram', chat, '@u', text, BOT, 'U')
-        self.assertEqual(self.errors, [], f'{text} crashed') # the worker reports crashes instead of raising
-        return list(self.sent)
+	def command(self, text, chat='55'):
+		"""Run a chat message through the real command parser and handlers."""
+		self.sent.clear()
+		self.markups.clear()
+		self.errors.clear()
+		worker.process_request('telegram', chat, '@u', text, BOT, 'U')
+		self.assertEqual(self.errors, [], f'{text} crashed') # the worker reports crashes instead of raising
+		return list(self.sent)
diff --git a/tests/test_charts.py b/tests/test_charts.py
index 7c09862b..0af42bd1 100644
--- a/tests/test_charts.py
+++ b/tests/test_charts.py
@@ -2,7 +2,7 @@
   1. image sanity: a real 1280x800 PNG, drawn, with nothing touching the edges (no clipped text)
   2. contracts: blue/red by sign, row order, titles and sort order, captured from what is handed to matplotlib
   3. one golden image per chart type, compared with a tolerance. Refresh after an intended restyle with:
-         UPDATE_GOLDEN=1 python -m unittest tests.test_charts
+		 UPDATE_GOLDEN=1 python -m unittest tests.test_charts
 """
 import datetime, io, os, time, unittest
 import gevent
@@ -24,287 +24,287 @@
 
 
 def decode(buf):
-    buf.seek(0)
-    return Image.open(buf).convert('RGB')
+	buf.seek(0)
+	return Image.open(buf).convert('RGB')
 
 
 def price_frame(days=120):
-    dates = [datetime.date(2026, 1, 1) + datetime.timedelta(days=i) for i in range(days)]
-    closes = [100 + 12 * np.sin(i / 9) + i * 0.15 for i in range(days)]
-    return pd.DataFrame({'Date': dates, 'Close': closes})
+	dates = [datetime.date(2026, 1, 1) + datetime.timedelta(days=i) for i in range(days)]
+	closes = [100 + 12 * np.sin(i / 9) + i * 0.15 for i in range(days)]
+	return pd.DataFrame({'Date': dates, 'Close': closes})
 
 
 def compare_series():
-    index = pd.date_range('2026-01-01', periods=120)
-    return [(name, pd.Series(100 + np.cumsum(np.sin(np.arange(120) / k) + drift), index=index))
-            for name, k, drift in (('AAA', 7, 0.2), ('BBB', 11, -0.1), ('CCC', 5, 0.05))]
+	index = pd.date_range('2026-01-01', periods=120)
+	return [(name, pd.Series(100 + np.cumsum(np.sin(np.arange(120) / k) + drift), index=index))
+			for name, k, drift in (('AAA', 7, 0.2), ('BBB', 11, -0.1), ('CCC', 5, 0.05))]
 
 
 CHARTS = {
-    'hbar_mixed': lambda: util.hbar_chart(MIXED, 'Price movement 5.0%, 5 Sep - 5 Oct 2026', 'subtitle here'),
-    'price_line': lambda: util.graph(price_frame(), 'Apple Inc (AAPL)', 'USD'),
-    'compare_lines': lambda: util.compare_graph(compare_series(), 'Compare 120 days', 'subtitle here'),
-    'shorts': lambda: util.shorts_chart([('GME', 47.3), ('AMC', 32.1), ('TSLA', 21.6), ('AAPL', 15.2)], 15),
+	'hbar_mixed': lambda: util.hbar_chart(MIXED, 'Price movement 5.0%, 5 Sep - 5 Oct 2026', 'subtitle here'),
+	'price_line': lambda: util.graph(price_frame(), 'Apple Inc (AAPL)', 'USD'),
+	'compare_lines': lambda: util.compare_graph(compare_series(), 'Compare 120 days', 'subtitle here'),
+	'shorts': lambda: util.shorts_chart([('GME', 47.3), ('AMC', 32.1), ('TSLA', 21.6), ('AAPL', 15.2)], 15),
 }
 # inputs that have broken chart code before or are likely to: sizes, signs, long names, zeros, a single row
 EDGE_ROWS = {
-    'one_row': [('AAA', 5)],
-    'one_negative': [('AAA', -5)],
-    'all_positive': [(f'T{i}', i + 1) for i in range(15)],
-    'all_negative': [(f'T{i}', -(i + 1)) for i in range(15)],
-    'all_zero': [(f'T{i}', 0) for i in range(5)],
-    'forty_rows': [(f'T{i}', (i - 20) * 1.5) for i in range(40)],
-    'long_labels': [('A very long company name indeed (TICKER)', 12), ('Another equally long label here ok', -8)],
-    'huge_values': [('AAA', 12345), ('BBB', -9876)],
-    'tiny_values': [('AAA', 0.01), ('BBB', -0.02)],
+	'one_row': [('AAA', 5)],
+	'one_negative': [('AAA', -5)],
+	'all_positive': [(f'T{i}', i + 1) for i in range(15)],
+	'all_negative': [(f'T{i}', -(i + 1)) for i in range(15)],
+	'all_zero': [(f'T{i}', 0) for i in range(5)],
+	'forty_rows': [(f'T{i}', (i - 20) * 1.5) for i in range(40)],
+	'long_labels': [('A very long company name indeed (TICKER)', 12), ('Another equally long label here ok', -8)],
+	'huge_values': [('AAA', 12345), ('BBB', -9876)],
+	'tiny_values': [('AAA', 0.01), ('BBB', -0.02)],
 }
 
 
 class ImageSanity(unittest.TestCase):
-    def check_image(self, buf, name, bg=BG, edge_clean=True):
-        image = decode(buf)
-        self.assertEqual(image.size, (1280, 800), name)
-        pixels = np.asarray(image)
-        self.assertGreater(len({tuple(p) for p in pixels.reshape(-1, 3)[::97]}), 8, f'{name} looks blank')
-        self.assertGreater(pixels.std(), 5, f'{name} looks blank')
-        allowed = bg if isinstance(bg[0], tuple) else (bg,)
-        self.assertIn(tuple(int(c) for c in pixels[0, 0]), allowed, name)
-        if not edge_clean: # the shorts beach draws its sea right to the edge
-            return
-        edge = np.concatenate([pixels[:3].reshape(-1, 3), pixels[-3:].reshape(-1, 3), pixels[:, :3].reshape(-1, 3), pixels[:, -3:].reshape(-1, 3)])
-        self.assertTrue((edge == bg).all(), f'{name}: something is drawn on the image edge (clipped text or bars?)')
-
-    def test_each_chart_type(self):
-        for name, make in CHARTS.items():
-            with self.subTest(chart=name):
-                if name == 'shorts':
-                    self.check_image(make(), name, bg=SHORTS_BG, edge_clean=False)
-                else:
-                    self.check_image(make(), name)
-
-    def test_edge_case_rows(self):
-        for name, rows in EDGE_ROWS.items():
-            with self.subTest(rows=name):
-                self.check_image(util.hbar_chart(rows, 'Title', 'sub'), name)
-                self.check_image(util.rows_chart(rows, 'Title'), name + '/rows_chart')
-
-    def test_zoomed_axis(self):
-        self.check_image(util.hbar_chart([('AAA', 1.2), ('BBB', 3.4)], 'Ratings', xlim=(0.9, 4.0)), 'zoomed')
-        self.check_image(util.hbar_chart([('WWWWWWWW', 1.2), ('MMM', 3.4)], 'Ratings', xlim=(0.9, 4.0)), 'zoomed wide labels')
-        self.check_image(util.hbar_chart([(f'T{i}', 1 + i / 10) for i in range(15)], 'Ratings', xlim=(0.9, 3.0)), 'zoomed 15 rows')
-
-    def test_shorts_chart_sizes_and_verdicts(self):
-        for n in (1, 5, 45):
-            rows = [(f'T{i}', 60.0 - i) for i in range(n)]
-            self.check_image(util.shorts_chart(rows, 15), f'shorts {n} rows', bg=SHORTS_BG, edge_clean=False)
-        self.check_image(util.shorts_chart([('A VERY LONG LABEL', 15.0), ('B', 0.4)], 15), 'shorts long label and tiny value', bg=SHORTS_BG, edge_clean=False)
-        tiers = [(0, 'board shorts'), (4.9, 'board shorts'), (5, 'cargo shorts'), (10, 'boxers'), (15, 'gym shorts'), (20, 'cutoffs'),
-                 (30, 'short shorts'), (40, 'daisy dukes'), (50, 'hot pants'), (60, 'speedos'), (70, 'budgie smugglers'),
-                 (80, 'just a belt'), (90, 'never nudes'), (200, 'never nudes')]
-        self.assertEqual([util.shorts_verdict(p) for p, _ in tiers], [name for _, name in tiers])
-
-    def test_bar_graph_has_top_and_bottom(self):
-        self.check_image(util.bar_graph([('AAA', 9)], [('ZZZ', -9)], 'Top and bottom'), 'bar_graph')
-
-    def test_flat_and_tiny_price_series(self):
-        flat = pd.DataFrame({'Date': [datetime.date(2026, 1, 1) + datetime.timedelta(days=i) for i in range(10)], 'Close': [5.0] * 10})
-        self.check_image(util.graph(flat, 'Flat', 'USD'), 'flat price')
-        two = price_frame(2)
-        self.check_image(util.graph(two, 'Two points', 'USD'), 'two points')
+	def check_image(self, buf, name, bg=BG, edge_clean=True):
+		image = decode(buf)
+		self.assertEqual(image.size, (1280, 800), name)
+		pixels = np.asarray(image)
+		self.assertGreater(len({tuple(p) for p in pixels.reshape(-1, 3)[::97]}), 8, f'{name} looks blank')
+		self.assertGreater(pixels.std(), 5, f'{name} looks blank')
+		allowed = bg if isinstance(bg[0], tuple) else (bg,)
+		self.assertIn(tuple(int(c) for c in pixels[0, 0]), allowed, name)
+		if not edge_clean: # the shorts beach draws its sea right to the edge
+			return
+		edge = np.concatenate([pixels[:3].reshape(-1, 3), pixels[-3:].reshape(-1, 3), pixels[:, :3].reshape(-1, 3), pixels[:, -3:].reshape(-1, 3)])
+		self.assertTrue((edge == bg).all(), f'{name}: something is drawn on the image edge (clipped text or bars?)')
+
+	def test_each_chart_type(self):
+		for name, make in CHARTS.items():
+			with self.subTest(chart=name):
+				if name == 'shorts':
+					self.check_image(make(), name, bg=SHORTS_BG, edge_clean=False)
+				else:
+					self.check_image(make(), name)
+
+	def test_edge_case_rows(self):
+		for name, rows in EDGE_ROWS.items():
+			with self.subTest(rows=name):
+				self.check_image(util.hbar_chart(rows, 'Title', 'sub'), name)
+				self.check_image(util.rows_chart(rows, 'Title'), name + '/rows_chart')
+
+	def test_zoomed_axis(self):
+		self.check_image(util.hbar_chart([('AAA', 1.2), ('BBB', 3.4)], 'Ratings', xlim=(0.9, 4.0)), 'zoomed')
+		self.check_image(util.hbar_chart([('WWWWWWWW', 1.2), ('MMM', 3.4)], 'Ratings', xlim=(0.9, 4.0)), 'zoomed wide labels')
+		self.check_image(util.hbar_chart([(f'T{i}', 1 + i / 10) for i in range(15)], 'Ratings', xlim=(0.9, 3.0)), 'zoomed 15 rows')
+
+	def test_shorts_chart_sizes_and_verdicts(self):
+		for n in (1, 5, 45):
+			rows = [(f'T{i}', 60.0 - i) for i in range(n)]
+			self.check_image(util.shorts_chart(rows, 15), f'shorts {n} rows', bg=SHORTS_BG, edge_clean=False)
+		self.check_image(util.shorts_chart([('A VERY LONG LABEL', 15.0), ('B', 0.4)], 15), 'shorts long label and tiny value', bg=SHORTS_BG, edge_clean=False)
+		tiers = [(0, 'board shorts'), (4.9, 'board shorts'), (5, 'cargo shorts'), (10, 'boxers'), (15, 'gym shorts'), (20, 'cutoffs'),
+				 (30, 'short shorts'), (40, 'daisy dukes'), (50, 'hot pants'), (60, 'speedos'), (70, 'budgie smugglers'),
+				 (80, 'just a belt'), (90, 'never nudes'), (200, 'never nudes')]
+		self.assertEqual([util.shorts_verdict(p) for p, _ in tiers], [name for _, name in tiers])
+
+	def test_bar_graph_has_top_and_bottom(self):
+		self.check_image(util.bar_graph([('AAA', 9)], [('ZZZ', -9)], 'Top and bottom'), 'bar_graph')
+
+	def test_flat_and_tiny_price_series(self):
+		flat = pd.DataFrame({'Date': [datetime.date(2026, 1, 1) + datetime.timedelta(days=i) for i in range(10)], 'Close': [5.0] * 10})
+		self.check_image(util.graph(flat, 'Flat', 'USD'), 'flat price')
+		two = price_frame(2)
+		self.check_image(util.graph(two, 'Two points', 'USD'), 'two points')
 
 
 class RenderPool(unittest.TestCase):
-    """The bot draws charts in one native thread (util.render_off_hub), so the gevent hub keeps serving while a chart is drawn."""
-
-    def setUp(self):
-        util.render_off_hub()
-        self.addCleanup(setattr, util, '_render_pool', None)
-
-    def test_charts_asked_for_at_once_all_finish(self):
-        jobs = [gevent.spawn(util.pie_chart, [('a', 1.0), ('b', 2.0)], f't{i}') for i in range(4)]
-        jobs += [gevent.spawn(util.hbar_chart, MIXED, 'T'), gevent.spawn(util.rows_chart, MIXED, 'T')]
-        gevent.joinall(jobs, timeout=120)
-        self.assertTrue(all(job.successful() for job in jobs), 'a chart did not finish (several native threads deadlock matplotlib)')
-        self.assertTrue(all(decode(job.value).size == (1280, 800) for job in jobs))
-
-    def test_the_hub_keeps_serving_while_a_chart_is_drawn(self):
-        gaps, done = [], []
-        def ticker():
-            last = time.monotonic()
-            while not done:
-                gevent.sleep(0.01)
-                now = time.monotonic()
-                gaps.append(now - last)
-                last = now
-        tick = gevent.spawn(ticker)
-        util.shorts_chart([(f'T{i}', 60.0 - 2 * i) for i in range(24)], 15) # about a second of drawing
-        done.append(True)
-        tick.join()
-        self.assertLess(max(gaps), 0.5) # drawn in the hub, the ticker stopped for the whole chart
+	"""The bot draws charts in one native thread (util.render_off_hub), so the gevent hub keeps serving while a chart is drawn."""
+
+	def setUp(self):
+		util.render_off_hub()
+		self.addCleanup(setattr, util, '_render_pool', None)
+
+	def test_charts_asked_for_at_once_all_finish(self):
+		jobs = [gevent.spawn(util.pie_chart, [('a', 1.0), ('b', 2.0)], f't{i}') for i in range(4)]
+		jobs += [gevent.spawn(util.hbar_chart, MIXED, 'T'), gevent.spawn(util.rows_chart, MIXED, 'T')]
+		gevent.joinall(jobs, timeout=120)
+		self.assertTrue(all(job.successful() for job in jobs), 'a chart did not finish (several native threads deadlock matplotlib)')
+		self.assertTrue(all(decode(job.value).size == (1280, 800) for job in jobs))
+
+	def test_the_hub_keeps_serving_while_a_chart_is_drawn(self):
+		gaps, done = [], []
+		def ticker():
+			last = time.monotonic()
+			while not done:
+				gevent.sleep(0.01)
+				now = time.monotonic()
+				gaps.append(now - last)
+				last = now
+		tick = gevent.spawn(ticker)
+		util.shorts_chart([(f'T{i}', 60.0 - 2 * i) for i in range(24)], 15) # about a second of drawing
+		done.append(True)
+		tick.join()
+		self.assertLess(max(gaps), 0.5) # drawn in the hub, the ticker stopped for the whole chart
 
 
 class Contracts(unittest.TestCase):
-    def capture_bars(self, rows, **kwargs):
-        calls = []
-        real = Axes.barh
-        def spy(ax, y, width, *a, **k):
-            calls.append({'y': list(y), 'width': list(width), 'color': list(k.get('color', []))})
-            return real(ax, y, width, *a, **k)
-        with mock.patch.object(Axes, 'barh', spy):
-            util.hbar_chart(rows, 'T', **kwargs)
-        return calls[0]
-
-    def capture_text(self, make):
-        texts = []
-        real = Figure.text
-        def spy(fig, x, y, s, *a, **k):
-            texts.append(s)
-            return real(fig, x, y, s, *a, **k)
-        with mock.patch.object(Figure, 'text', spy):
-            make()
-        return texts
-
-    def test_bars_are_blue_when_positive_and_red_when_negative(self):
-        bars = self.capture_bars([('A', 3), ('B', 0), ('C', -2)])
-        self.assertEqual(bars['color'], ['#3987e5', '#3987e5', '#fb7185'])
-
-    def test_first_row_is_on_top_and_spacers_are_skipped(self):
-        bars = self.capture_bars(MIXED)
-        self.assertEqual(bars['width'], [42, 17, 3, -5, -30])
-        self.assertEqual(bars['y'], [0, 1, 2, 4, 5]) # row 3 is the blank spacer
-        calls = []
-        real = Axes.set_ylim
-        with mock.patch.object(Axes, 'set_ylim', lambda ax, *a, **k: (calls.append(a), real(ax, *a, **k))[1]):
-            util.hbar_chart(MIXED, 'T')
-        self.assertTrue(any(len(a) == 2 and a[0] > a[1] for a in calls), 'y axis must be inverted so row 0 is drawn first (top)')
-
-    def test_pixels_follow_the_sign_of_the_data(self):
-        def count(rows):
-            pixels = np.asarray(decode(util.hbar_chart(rows, 'T'))).reshape(-1, 3)
-            return (pixels == BLUE).all(axis=1).sum(), (pixels == RED).all(axis=1).sum()
-        blue, red = count([('A', 5), ('B', 3)])
-        self.assertGreater(blue, 1000); self.assertEqual(red, 0)
-        blue, red = count([('A', -5), ('B', -3)])
-        self.assertEqual(blue, 0); self.assertGreater(red, 1000)
-        blue, red = count([('A', 5), ('B', -3)])
-        self.assertGreater(blue, 1000); self.assertGreater(red, 1000)
-
-    def test_session_tints_the_background(self):
-        corner = lambda session: tuple(np.asarray(decode(util.hbar_chart(MIXED, 'T', session=session)))[2, 2][:3])
-        plain, close, mid, pre = corner(None), corner('close'), corner('midsession'), corner('premarket')
-        self.assertEqual(plain, (0x14, 0x18, 0x1f))
-        self.assertEqual(close, plain) # at close stays untinted
-        self.assertTrue(mid[1] > mid[0] and mid[1] > plain[1]) # green
-        self.assertTrue(pre[0] > plain[0] and pre[1] > plain[1] and pre[0] > pre[2] and pre[1] > pre[2]) # yellow: red and green channels both lead blue
-        self.assertLess(max(abs(int(a) - int(b)) for a, b in zip(pre, plain)), 20) # subtle
-
-    def test_longer_bar_is_drawn_longer(self):
-        pixels = np.asarray(decode(util.hbar_chart([('A', 40), ('B', 10)], 'T')))
-        blue_rows = np.where((pixels == BLUE).all(axis=2).any(axis=1))[0]
-        top_band, bottom_band = pixels[blue_rows[0]], pixels[blue_rows[-1]]
-        self.assertGreater((top_band == BLUE).all(axis=1).sum(), 2 * (bottom_band == BLUE).all(axis=1).sum())
-
-    def test_titles_are_drawn(self):
-        texts = self.capture_text(lambda: util.hbar_chart(MIXED, 'My Title', 'My subtitle'))
-        self.assertIn('My Title', texts)
-        self.assertIn('My subtitle', texts)
-        texts = self.capture_text(lambda: util.graph(price_frame(), 'Apple Inc (AAPL)', 'USD'))
-        self.assertIn('Apple Inc (AAPL)', texts)
-        self.assertTrue(any('over period' in t for t in texts))
-
-    def test_price_line_colour_follows_direction(self):
-        def line_colour(closes):
-            frame = pd.DataFrame({'Date': [datetime.date(2026, 1, 1) + datetime.timedelta(days=i) for i in range(len(closes))], 'Close': closes})
-            texts = []
-            real = Figure.text
-            with mock.patch.object(Figure, 'text', lambda fig, x, y, s, *a, **k: (texts.append((s, k.get('color'))), real(fig, x, y, s, *a, **k))[1]):
-                util.graph(frame, 'T', 'USD')
-            return next(color for s, color in texts if 'over period' in s)
-        self.assertEqual(line_colour([1, 2, 3]), '#34d399') # up: green
-        self.assertEqual(line_colour([3, 2, 1]), '#fb7185') # down: red
-        self.assertEqual(line_colour([2, 3, 2]), '#9aa4b2') # flat
+	def capture_bars(self, rows, **kwargs):
+		calls = []
+		real = Axes.barh
+		def spy(ax, y, width, *a, **k):
+			calls.append({'y': list(y), 'width': list(width), 'color': list(k.get('color', []))})
+			return real(ax, y, width, *a, **k)
+		with mock.patch.object(Axes, 'barh', spy):
+			util.hbar_chart(rows, 'T', **kwargs)
+		return calls[0]
+
+	def capture_text(self, make):
+		texts = []
+		real = Figure.text
+		def spy(fig, x, y, s, *a, **k):
+			texts.append(s)
+			return real(fig, x, y, s, *a, **k)
+		with mock.patch.object(Figure, 'text', spy):
+			make()
+		return texts
+
+	def test_bars_are_blue_when_positive_and_red_when_negative(self):
+		bars = self.capture_bars([('A', 3), ('B', 0), ('C', -2)])
+		self.assertEqual(bars['color'], ['#3987e5', '#3987e5', '#fb7185'])
+
+	def test_first_row_is_on_top_and_spacers_are_skipped(self):
+		bars = self.capture_bars(MIXED)
+		self.assertEqual(bars['width'], [42, 17, 3, -5, -30])
+		self.assertEqual(bars['y'], [0, 1, 2, 4, 5]) # row 3 is the blank spacer
+		calls = []
+		real = Axes.set_ylim
+		with mock.patch.object(Axes, 'set_ylim', lambda ax, *a, **k: (calls.append(a), real(ax, *a, **k))[1]):
+			util.hbar_chart(MIXED, 'T')
+		self.assertTrue(any(len(a) == 2 and a[0] > a[1] for a in calls), 'y axis must be inverted so row 0 is drawn first (top)')
+
+	def test_pixels_follow_the_sign_of_the_data(self):
+		def count(rows):
+			pixels = np.asarray(decode(util.hbar_chart(rows, 'T'))).reshape(-1, 3)
+			return (pixels == BLUE).all(axis=1).sum(), (pixels == RED).all(axis=1).sum()
+		blue, red = count([('A', 5), ('B', 3)])
+		self.assertGreater(blue, 1000); self.assertEqual(red, 0)
+		blue, red = count([('A', -5), ('B', -3)])
+		self.assertEqual(blue, 0); self.assertGreater(red, 1000)
+		blue, red = count([('A', 5), ('B', -3)])
+		self.assertGreater(blue, 1000); self.assertGreater(red, 1000)
+
+	def test_session_tints_the_background(self):
+		corner = lambda session: tuple(np.asarray(decode(util.hbar_chart(MIXED, 'T', session=session)))[2, 2][:3])
+		plain, close, mid, pre = corner(None), corner('close'), corner('midsession'), corner('premarket')
+		self.assertEqual(plain, (0x14, 0x18, 0x1f))
+		self.assertEqual(close, plain) # at close stays untinted
+		self.assertTrue(mid[1] > mid[0] and mid[1] > plain[1]) # green
+		self.assertTrue(pre[0] > plain[0] and pre[1] > plain[1] and pre[0] > pre[2] and pre[1] > pre[2]) # yellow: red and green channels both lead blue
+		self.assertLess(max(abs(int(a) - int(b)) for a, b in zip(pre, plain)), 20) # subtle
+
+	def test_longer_bar_is_drawn_longer(self):
+		pixels = np.asarray(decode(util.hbar_chart([('A', 40), ('B', 10)], 'T')))
+		blue_rows = np.where((pixels == BLUE).all(axis=2).any(axis=1))[0]
+		top_band, bottom_band = pixels[blue_rows[0]], pixels[blue_rows[-1]]
+		self.assertGreater((top_band == BLUE).all(axis=1).sum(), 2 * (bottom_band == BLUE).all(axis=1).sum())
+
+	def test_titles_are_drawn(self):
+		texts = self.capture_text(lambda: util.hbar_chart(MIXED, 'My Title', 'My subtitle'))
+		self.assertIn('My Title', texts)
+		self.assertIn('My subtitle', texts)
+		texts = self.capture_text(lambda: util.graph(price_frame(), 'Apple Inc (AAPL)', 'USD'))
+		self.assertIn('Apple Inc (AAPL)', texts)
+		self.assertTrue(any('over period' in t for t in texts))
+
+	def test_price_line_colour_follows_direction(self):
+		def line_colour(closes):
+			frame = pd.DataFrame({'Date': [datetime.date(2026, 1, 1) + datetime.timedelta(days=i) for i in range(len(closes))], 'Close': closes})
+			texts = []
+			real = Figure.text
+			with mock.patch.object(Figure, 'text', lambda fig, x, y, s, *a, **k: (texts.append((s, k.get('color'))), real(fig, x, y, s, *a, **k))[1]):
+				util.graph(frame, 'T', 'USD')
+			return next(color for s, color in texts if 'over period' in s)
+		self.assertEqual(line_colour([1, 2, 3]), '#34d399') # up: green
+		self.assertEqual(line_colour([3, 2, 1]), '#fb7185') # down: red
+		self.assertEqual(line_colour([2, 3, 2]), '#9aa4b2') # flat
 
 
 class ReportChartInputs(FinbotCase):
-    """What the reports hand to the chart: sort order, titles and dates, taken from the real report code."""
-
-    def run_report(self, call):
-        captured = []
-        real = util.hbar_chart
-        def spy(rows, title, subtitle='', **kw):
-            captured.append({'rows': list(rows), 'title': title, 'subtitle': subtitle})
-            return real(rows, title, subtitle, **kw)
-        self.patch(util, 'hbar_chart', spy)
-        call()
-        self.assertTrue(captured, 'no chart was drawn')
-        return captured[0]
-
-    def test_price_top_lists_best_first_then_worst(self):
-        chart = self.run_report(lambda: self.command('.price top 7d'))
-        values = [v for _, v in chart['rows'] if v is not None]
-        top, bottom = values[:10], values[10:]
-        self.assertEqual(top, sorted(top, reverse=True))
-        self.assertEqual(bottom, sorted(bottom)) # the worst performer comes first, like the caption's second list
-        self.assertGreater(min(top), max(bottom))
-        self.assertIn(None, [v for _, v in chart['rows']]) # spacer between the two lists
-
-    def test_price_threshold_chart_is_sorted_and_titled_with_the_threshold(self):
-        chart = self.run_report(lambda: self.command('.price 5%'))
-        values = [v for _, v in chart['rows'] if v is not None]
-        self.assertEqual(values, sorted(values, reverse=True))
-        self.assertIn('5', chart['title'])
-
-    def test_price_over_a_period_has_a_dated_title(self):
-        chart = self.run_report(lambda: self.command('.price 7d'))
-        self.assertRegex(chart['title'], r'\d{1,2} \w{3} - \d{1,2} \w{3} \d{4}') # 28 Sep - 5 Oct 2026
-
-    def test_performance_is_sorted_by_percent(self):
-        chart = self.run_report(lambda: self.command('.performance 7d'))
-        values = [v for _, v in chart['rows'] if v is not None]
-        self.assertEqual(values, sorted(values, reverse=True))
-
-    def test_shorts_is_sorted_and_carries_the_threshold(self):
-        captured = []
-        real = util.shorts_chart
-        def spy(rows, threshold, *a, **k):
-            captured.append((list(rows), threshold))
-            return real(rows, threshold, *a, **k)
-        self.patch(util, 'shorts_chart', spy)
-        self.command('.shorts')
-        self.assertTrue(captured, 'no chart was drawn')
-        rows, threshold = captured[0]
-        values = [v for _, v in rows]
-        self.assertEqual(values, sorted(values, reverse=True))
-        self.assertTrue(all(v > threshold for v in values))
+	"""What the reports hand to the chart: sort order, titles and dates, taken from the real report code."""
+
+	def run_report(self, call):
+		captured = []
+		real = util.hbar_chart
+		def spy(rows, title, subtitle='', **kw):
+			captured.append({'rows': list(rows), 'title': title, 'subtitle': subtitle})
+			return real(rows, title, subtitle, **kw)
+		self.patch(util, 'hbar_chart', spy)
+		call()
+		self.assertTrue(captured, 'no chart was drawn')
+		return captured[0]
+
+	def test_price_top_lists_best_first_then_worst(self):
+		chart = self.run_report(lambda: self.command('.price top 7d'))
+		values = [v for _, v in chart['rows'] if v is not None]
+		top, bottom = values[:10], values[10:]
+		self.assertEqual(top, sorted(top, reverse=True))
+		self.assertEqual(bottom, sorted(bottom)) # the worst performer comes first, like the caption's second list
+		self.assertGreater(min(top), max(bottom))
+		self.assertIn(None, [v for _, v in chart['rows']]) # spacer between the two lists
+
+	def test_price_threshold_chart_is_sorted_and_titled_with_the_threshold(self):
+		chart = self.run_report(lambda: self.command('.price 5%'))
+		values = [v for _, v in chart['rows'] if v is not None]
+		self.assertEqual(values, sorted(values, reverse=True))
+		self.assertIn('5', chart['title'])
+
+	def test_price_over_a_period_has_a_dated_title(self):
+		chart = self.run_report(lambda: self.command('.price 7d'))
+		self.assertRegex(chart['title'], r'\d{1,2} \w{3} - \d{1,2} \w{3} \d{4}') # 28 Sep - 5 Oct 2026
+
+	def test_performance_is_sorted_by_percent(self):
+		chart = self.run_report(lambda: self.command('.performance 7d'))
+		values = [v for _, v in chart['rows'] if v is not None]
+		self.assertEqual(values, sorted(values, reverse=True))
+
+	def test_shorts_is_sorted_and_carries_the_threshold(self):
+		captured = []
+		real = util.shorts_chart
+		def spy(rows, threshold, *a, **k):
+			captured.append((list(rows), threshold))
+			return real(rows, threshold, *a, **k)
+		self.patch(util, 'shorts_chart', spy)
+		self.command('.shorts')
+		self.assertTrue(captured, 'no chart was drawn')
+		rows, threshold = captured[0]
+		values = [v for _, v in rows]
+		self.assertEqual(values, sorted(values, reverse=True))
+		self.assertTrue(all(v > threshold for v in values))
 
 
 class GoldenImages(unittest.TestCase):
-    """One reference image per chart type. Tolerant on purpose: compares 160x100 thumbnails, so font anti-aliasing and
-    minor matplotlib differences pass, while changed colours, layout, bar sizes or missing text do not."""
-    TOLERANCE = 2.0 # mean absolute difference, in 0-255 grey levels, of the 160x100 thumbnails
-
-    @staticmethod
-    def thumb(image):
-        return image.resize((160, 100), Image.LANCZOS) # goldens are stored at this size too: a few KB each, not 100 KB
-
-    def check(self, name):
-        thumb = self.thumb(decode(CHARTS[name]()))
-        path = os.path.join(GOLDEN, name + '.png')
-        if os.environ.get('UPDATE_GOLDEN') or not os.path.exists(path):
-            os.makedirs(GOLDEN, exist_ok=True)
-            thumb.save(path, optimize=True)
-            if not os.environ.get('UPDATE_GOLDEN'):
-                self.fail(f'{name}: no golden image existed, so one was written to {path}. Review it and commit it.')
-            return
-        diff = np.abs(np.asarray(thumb, dtype=float) - np.asarray(Image.open(path).convert('RGB'), dtype=float)).mean()
-        self.assertLess(diff, self.TOLERANCE, f'{name} no longer matches tests/golden/{name}.png (diff {diff:.2f}). '
-                        f'If the change is intended, run: UPDATE_GOLDEN=1 python -m unittest tests.test_charts')
-
-    def test_hbar_mixed(self): self.check('hbar_mixed')
-    def test_price_line(self): self.check('price_line')
-    def test_shorts(self): self.check('shorts')
-    def test_compare_lines(self): self.check('compare_lines')
+	"""One reference image per chart type. Tolerant on purpose: compares 160x100 thumbnails, so font anti-aliasing and
+	minor matplotlib differences pass, while changed colours, layout, bar sizes or missing text do not."""
+	TOLERANCE = 2.0 # mean absolute difference, in 0-255 grey levels, of the 160x100 thumbnails
+
+	@staticmethod
+	def thumb(image):
+		return image.resize((160, 100), Image.LANCZOS) # goldens are stored at this size too: a few KB each, not 100 KB
+
+	def check(self, name):
+		thumb = self.thumb(decode(CHARTS[name]()))
+		path = os.path.join(GOLDEN, name + '.png')
+		if os.environ.get('UPDATE_GOLDEN') or not os.path.exists(path):
+			os.makedirs(GOLDEN, exist_ok=True)
+			thumb.save(path, optimize=True)
+			if not os.environ.get('UPDATE_GOLDEN'):
+				self.fail(f'{name}: no golden image existed, so one was written to {path}. Review it and commit it.')
+			return
+		diff = np.abs(np.asarray(thumb, dtype=float) - np.asarray(Image.open(path).convert('RGB'), dtype=float)).mean()
+		self.assertLess(diff, self.TOLERANCE, f'{name} no longer matches tests/golden/{name}.png (diff {diff:.2f}). '
+						f'If the change is intended, run: UPDATE_GOLDEN=1 python -m unittest tests.test_charts')
+
+	def test_hbar_mixed(self): self.check('hbar_mixed')
+	def test_price_line(self): self.check('price_line')
+	def test_shorts(self): self.check('shorts')
+	def test_compare_lines(self): self.check('compare_lines')
 
 
 if __name__ == '__main__':
-    unittest.main()
+	unittest.main()
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 76c06df8..46992e06 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -5,417 +5,417 @@
 from tests.common import *
 
 COMMANDS = ('.price', '.price 7d', '.price top', '.session', '.premarket', '.performance 7d', '.shorts', '.peg', '.buy',
-            '.sell', '.beta', '.pe', '.marketcap', '.earnings', '.dividend', '.trades', '.holdings', '.watchlist',
-            '.T1', '.history T1', '.compare T1 T2')
+			'.sell', '.beta', '.pe', '.marketcap', '.earnings', '.dividend', '.trades', '.holdings', '.watchlist',
+			'.T1', '.history T1', '.compare T1 T2')
 
 
 class UpstreamEmpty(FinbotCase):
-    """Yahoo answers, but with nothing usable. Every command must reply sanely and never crash."""
-
-    ONE_LINE_ERROR = ('.history T1', '.compare T1 T2') # these say so with a single error line when the ticker has no data
-
-    def run_all(self):
-        for cmd in COMMANDS:
-            with self.subTest(cmd=cmd):
-                if cmd in self.ONE_LINE_ERROR:
-                    self.sent.clear(); self.errors.clear()
-                    worker.process_request('telegram', '55', '@u', cmd, BOT, 'U')
-                    self.assertLessEqual(len(self.errors), 1, cmd)
-                else:
-                    self.command(cmd)
-
-    def test_yahoo_returns_empty_dict(self):
-        self.patch(yahoo, 'fetch', lambda ts: {})
-        self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {})
-        self.run_all()
-
-    def test_yahoo_raises(self):
-        def down(*a, **k):
-            raise RuntimeError('Yahoo API unavailable')
-        for name in ('fetch', 'fetch_detail', 'price_series'):
-            self.patch(yahoo, name, down)
-        for cmd in ('.price', '.peg', '.T1', '.compare T1 T2'):
-            with self.subTest(cmd=cmd):
-                self.sent.clear(); self.errors.clear()
-                worker.process_request('telegram', '55', '@u', cmd, BOT, 'U')
-                self.assertEqual(len(self.errors), 1, cmd) # exactly one one-line report, not a traceback to the user
-                self.assertTrue(self.errors[0].startswith('⚠️'))
-
-    def test_sharesight_down_is_reported_once(self):
-        def down(*a, **k):
-            raise RuntimeError('Sharesight performance error: boom')
-        self.patch(sharesight, 'get_performance_wrapper', down)
-        self.patch(sharesight, 'get_performance', down)
-        self.patch(sharesight, 'get_portfolios', down)
-        self.sent.clear(); self.errors.clear()
-        worker.process_request('telegram', '55', '@u', '.performance 7d', BOT, 'U')
-        self.assertEqual(len(self.errors), 1)
-
-    def test_a_failed_command_stops_typing_before_saying_why(self):
-        def down(*a, **k):
-            raise RuntimeError('Sharesight performance error: boom')
-        self.patch(sharesight, 'get_portfolios', down)
-        typing_when_reported = []
-        self.patch(webhook, 'report_error', lambda e, *a, **k: typing_when_reported.append([t.is_active() for t in self.typing]))
-        worker.process_request('telegram', '55', '@u', '.performance 7d', BOT, 'U')
-        self.assertEqual(typing_when_reported, [[False]]) # its one indicator had stopped when the error line went out
-
-    def test_no_holdings_and_empty_watchlist(self):
-        self.patch(util, 'get_holdings_and_watchlist', lambda: [])
-        for cmd in ('.watchlist', '.price', '.earnings'):
-            with self.subTest(cmd=cmd):
-                self.command(cmd)
+	"""Yahoo answers, but with nothing usable. Every command must reply sanely and never crash."""
+
+	ONE_LINE_ERROR = ('.history T1', '.compare T1 T2') # these say so with a single error line when the ticker has no data
+
+	def run_all(self):
+		for cmd in COMMANDS:
+			with self.subTest(cmd=cmd):
+				if cmd in self.ONE_LINE_ERROR:
+					self.sent.clear(); self.errors.clear()
+					worker.process_request('telegram', '55', '@u', cmd, BOT, 'U')
+					self.assertLessEqual(len(self.errors), 1, cmd)
+				else:
+					self.command(cmd)
+
+	def test_yahoo_returns_empty_dict(self):
+		self.patch(yahoo, 'fetch', lambda ts: {})
+		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {})
+		self.run_all()
+
+	def test_yahoo_raises(self):
+		def down(*a, **k):
+			raise RuntimeError('Yahoo API unavailable')
+		for name in ('fetch', 'fetch_detail', 'price_series'):
+			self.patch(yahoo, name, down)
+		for cmd in ('.price', '.peg', '.T1', '.compare T1 T2'):
+			with self.subTest(cmd=cmd):
+				self.sent.clear(); self.errors.clear()
+				worker.process_request('telegram', '55', '@u', cmd, BOT, 'U')
+				self.assertEqual(len(self.errors), 1, cmd) # exactly one one-line report, not a traceback to the user
+				self.assertTrue(self.errors[0].startswith('⚠️'))
+
+	def test_sharesight_down_is_reported_once(self):
+		def down(*a, **k):
+			raise RuntimeError('Sharesight performance error: boom')
+		self.patch(sharesight, 'get_performance_wrapper', down)
+		self.patch(sharesight, 'get_performance', down)
+		self.patch(sharesight, 'get_portfolios', down)
+		self.sent.clear(); self.errors.clear()
+		worker.process_request('telegram', '55', '@u', '.performance 7d', BOT, 'U')
+		self.assertEqual(len(self.errors), 1)
+
+	def test_a_failed_command_stops_typing_before_saying_why(self):
+		def down(*a, **k):
+			raise RuntimeError('Sharesight performance error: boom')
+		self.patch(sharesight, 'get_portfolios', down)
+		typing_when_reported = []
+		self.patch(webhook, 'report_error', lambda e, *a, **k: typing_when_reported.append([t.is_active() for t in self.typing]))
+		worker.process_request('telegram', '55', '@u', '.performance 7d', BOT, 'U')
+		self.assertEqual(typing_when_reported, [[False]]) # its one indicator had stopped when the error line went out
+
+	def test_no_holdings_and_empty_watchlist(self):
+		self.patch(util, 'get_holdings_and_watchlist', lambda: [])
+		for cmd in ('.watchlist', '.price', '.earnings'):
+			with self.subTest(cmd=cmd):
+				self.command(cmd)
 
 
 class WatchlistActions(FinbotCase):
-    def test_add_duplicate_delete_and_unknown_action(self):
-        add = self.command('.watchlist add MSFT')
-        self.assertTrue(add)
-        self.assertEqual(util.json_load('finbot_watchlist.json', persist=True), ['MSFT'])
-        again = self.command('.watchlist add MSFT')
-        self.assertNotEqual(add, again) # a duplicate is reported differently from a first add
-        self.assertEqual(util.json_load('finbot_watchlist.json', persist=True), ['MSFT'])
-        self.command('.watchlist rm MSFT')
-        self.assertEqual(util.json_load('finbot_watchlist.json', persist=True), [])
-        self.assertIn('not a valid watchlist action', str(self.command('.watchlist frobnicate MSFT')))
+	def test_add_duplicate_delete_and_unknown_action(self):
+		add = self.command('.watchlist add MSFT')
+		self.assertTrue(add)
+		self.assertEqual(util.json_load('finbot_watchlist.json', persist=True), ['MSFT'])
+		again = self.command('.watchlist add MSFT')
+		self.assertNotEqual(add, again) # a duplicate is reported differently from a first add
+		self.assertEqual(util.json_load('finbot_watchlist.json', persist=True), ['MSFT'])
+		self.command('.watchlist rm MSFT')
+		self.assertEqual(util.json_load('finbot_watchlist.json', persist=True), [])
+		self.assertIn('not a valid watchlist action', str(self.command('.watchlist frobnicate MSFT')))
 
 
 class BadArguments(FinbotCase):
-    def test_compare_argument_errors_give_usage(self):
-        for cmd in ('.compare', '.compare T1', '.compare T1 T2 T3 T4 T5 T6 T7', '.compare T1 T1'):
-            with self.subTest(cmd=cmd):
-                self.assertIn('Usage', str(self.command(cmd)))
+	def test_compare_argument_errors_give_usage(self):
+		for cmd in ('.compare', '.compare T1', '.compare T1 T2 T3 T4 T5 T6 T7', '.compare T1 T1'):
+			with self.subTest(cmd=cmd):
+				self.assertIn('Usage', str(self.command(cmd)))
 
-    def test_compare_with_a_ticker_that_has_no_history(self):
-        def no_history(t):
-            raise RuntimeError(f'no price history for {t}')
-        self.patch(yahoo, 'price_series', no_history)
-        self.sent.clear(); self.errors.clear()
-        worker.process_request('telegram', '55', '@u', '.compare T1 T2', BOT, 'U')
-        self.assertEqual(len(self.errors), 1)
+	def test_compare_with_a_ticker_that_has_no_history(self):
+		def no_history(t):
+			raise RuntimeError(f'no price history for {t}')
+		self.patch(yahoo, 'price_series', no_history)
+		self.sent.clear(); self.errors.clear()
+		worker.process_request('telegram', '55', '@u', '.compare T1 T2', BOT, 'U')
+		self.assertEqual(len(self.errors), 1)
 
-    def test_history_without_a_ticker_or_with_extra_argument_gives_usage(self):
-        self.assertEqual(self.command('.history'), [('text', ['Usage: .history SYMBOL'])]) # not the profile of a ticker called HISTORY
-        self.assertEqual(self.command('.history T1 1y'), [('text', ['Usage: .history SYMBOL'])])
+	def test_history_without_a_ticker_or_with_extra_argument_gives_usage(self):
+		self.assertEqual(self.command('.history'), [('text', ['Usage: .history SYMBOL'])]) # not the profile of a ticker called HISTORY
+		self.assertEqual(self.command('.history T1 1y'), [('text', ['Usage: .history SYMBOL'])])
 
 
 class PortfolioNames(FinbotCase):
-    def test_an_unknown_portfolio_gets_the_same_answer_everywhere(self):
-        for cmd in ('.trades 7d Nope', '.performance Nope', '.holdings Nope', '.pie Nope'):
-            with self.subTest(cmd=cmd): # it used to be a ValueError, "no Sharesight data found", or two wordings of this
-                self.assertEqual(self.command(cmd), [('text', ['Nope portfolio not found. I only know about:', 'Rob'])])
+	def test_an_unknown_portfolio_gets_the_same_answer_everywhere(self):
+		for cmd in ('.trades 7d Nope', '.performance Nope', '.holdings Nope', '.pie Nope'):
+			with self.subTest(cmd=cmd): # it used to be a ValueError, "no Sharesight data found", or two wordings of this
+				self.assertEqual(self.command(cmd), [('text', ['Nope portfolio not found. I only know about:', 'Rob'])])
 
-    def test_a_portfolio_name_in_any_case(self):
-        self.patch(sharesight, 'get_holdings_new', lambda name, pid: {'T1:NASDAQ': {'code': 'T1', 'market_code': 'NASDAQ', 'name': 'Co T1', 'holding_id': 1}})
-        self.assertIn('Holdings for Rob (1)', str(self.command('.holdings rob')))
+	def test_a_portfolio_name_in_any_case(self):
+		self.patch(sharesight, 'get_holdings_new', lambda name, pid: {'T1:NASDAQ': {'code': 'T1', 'market_code': 'NASDAQ', 'name': 'Co T1', 'holding_id': 1}})
+		self.assertIn('Holdings for Rob (1)', str(self.command('.holdings rob')))
 
-    def test_trades_with_no_buy_or_sell_still_get_an_answer(self):
-        self.patch(sharesight, 'get_trades', lambda name, pid, days=7: [{'id': 2, 'transaction_type': 'DIVIDEND', 'portfolio': name}])
-        self.assertIn('No trades', str(self.command('.trades 7d'))) # it used to say nothing at all
+	def test_trades_with_no_buy_or_sell_still_get_an_answer(self):
+		self.patch(sharesight, 'get_trades', lambda name, pid, days=7: [{'id': 2, 'transaction_type': 'DIVIDEND', 'portfolio': name}])
+		self.assertIn('No trades', str(self.command('.trades 7d'))) # it used to say nothing at all
 
 
 class CommandArguments(FinbotCase):
-    def test_period_and_portfolio_in_either_order(self):
-        for text, expected in (('', (30, None)), ('7d', (7, None)), ('Rob', (30, 'Rob')), ('7d Rob', (7, 'Rob')), ('Rob 7d', (7, 'Rob')),
-                               ('My Portfolio', (30, 'My Portfolio')), ('My Portfolio 3m', (90, 'My Portfolio')), ('Rob ', (30, 'Rob'))):
-            with self.subTest(text=text):
-                self.assertEqual(worker.period_and_portfolio(text, 30), expected)
-
-    def test_a_percentage_is_a_number_and_nan_a_ticker(self):
-        for arg, expected in (('', (5, None)), ('2.5', (2.5, None)), ('3%', (3.0, None)), ('T1', (5, 'T1')), ('nan', (5, 'NAN')), ('inf', (5, 'INF'))):
-            with self.subTest(arg=arg):
-                self.assertEqual(worker.percent_or_ticker(arg, 5), expected)
-        self.assertIn('2.5%', str(self.command('.session 2.5'))) # a threshold, not a ticker called 2.5
-
-    def test_a_bare_number_on_price_is_a_percent(self):
-        """As on .session, .premarket and .shorts, and as the README has it; a period carries its unit."""
-        self.assertIn('3% at close', str(self.command('.price 3'))) # it used to be a 3-day report
-        self.assertIn('Price movement 3%', str(self.command('.price 3 7d'))) # the 3 used to be dropped
-        self.assertIn('Price movement 5%', str(self.command('.price 7d')))
-
-    def test_top_is_gone(self):
-        self.assertIn('Co TOP', str(self.command('.top 7d'))) # .price top 7d is the command; .top is a ticker now, like any word
-
-    def test_a_ratio_command_reads_its_ticker_as_the_other_commands_do(self):
-        self.assertIn('Co T1 ', str(self.command('.pe t1')))
-        self.assertIn('Top 15 tracked stocks by pe ratio', str(self.command('.pe top '))) # trailing space and all
-
-    def test_hello_leaves_the_word_list_as_it_was(self):
-        before = list(worker.verb)
-        for _ in range(30):
-            line = self.command('.hello')[0][1][0]
-            self.assertNotIn("('", line) # a verb, not a tuple of them
-            self.assertTrue(line.endswith(' you, U! 😇'), line)
-        self.assertEqual(worker.verb, before)
-        self.assertEqual(self.command('hi there'), []) # hello by name only: "hi @bot"
+	def test_period_and_portfolio_in_either_order(self):
+		for text, expected in (('', (30, None)), ('7d', (7, None)), ('Rob', (30, 'Rob')), ('7d Rob', (7, 'Rob')), ('Rob 7d', (7, 'Rob')),
+							   ('My Portfolio', (30, 'My Portfolio')), ('My Portfolio 3m', (90, 'My Portfolio')), ('Rob ', (30, 'Rob'))):
+			with self.subTest(text=text):
+				self.assertEqual(worker.period_and_portfolio(text, 30), expected)
+
+	def test_a_percentage_is_a_number_and_nan_a_ticker(self):
+		for arg, expected in (('', (5, None)), ('2.5', (2.5, None)), ('3%', (3.0, None)), ('T1', (5, 'T1')), ('nan', (5, 'NAN')), ('inf', (5, 'INF'))):
+			with self.subTest(arg=arg):
+				self.assertEqual(worker.percent_or_ticker(arg, 5), expected)
+		self.assertIn('2.5%', str(self.command('.session 2.5'))) # a threshold, not a ticker called 2.5
+
+	def test_a_bare_number_on_price_is_a_percent(self):
+		"""As on .session, .premarket and .shorts, and as the README has it; a period carries its unit."""
+		self.assertIn('3% at close', str(self.command('.price 3'))) # it used to be a 3-day report
+		self.assertIn('Price movement 3%', str(self.command('.price 3 7d'))) # the 3 used to be dropped
+		self.assertIn('Price movement 5%', str(self.command('.price 7d')))
+
+	def test_top_is_gone(self):
+		self.assertIn('Co TOP', str(self.command('.top 7d'))) # .price top 7d is the command; .top is a ticker now, like any word
+
+	def test_a_ratio_command_reads_its_ticker_as_the_other_commands_do(self):
+		self.assertIn('Co T1 ', str(self.command('.pe t1')))
+		self.assertIn('Top 15 tracked stocks by pe ratio', str(self.command('.pe top '))) # trailing space and all
+
+	def test_hello_leaves_the_word_list_as_it_was(self):
+		before = list(worker.verb)
+		for _ in range(30):
+			line = self.command('.hello')[0][1][0]
+			self.assertNotIn("('", line) # a verb, not a tuple of them
+			self.assertTrue(line.endswith(' you, U! 😇'), line)
+		self.assertEqual(worker.verb, before)
+		self.assertEqual(self.command('hi there'), []) # hello by name only: "hi @bot"
 
 
 class WsgiRejections(FinbotCase):
-    SECRET = {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'tok'}
-
-    def request(self, path='/telegram', body=None, headers=None):
-        body = body if isinstance(body, bytes) else json.dumps(body if body is not None else {}).encode()
-        env = {'PATH_INFO': path, 'wsgi.input': io.BytesIO(body), **(headers or {})}
-        status = []
-        with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
-            out = bot.main(env, lambda s, *a, **k: status.append(s))
-        return b''.join(out), status[-1]
-
-    def test_telegram_requires_the_secret_header(self):
-        message = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.help'}}
-        self.assertIn(b'Unauthorized', self.request(body=message)[0])
-        self.assertIn(b'Unauthorized', self.request(body=message, headers={'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'wrong'})[0])
-        self.assertEqual(self.sent, []) # nothing was processed
-
-    def test_unknown_path_is_404(self):
-        self.assertEqual(self.request(path='/nope')[1], '404 Not Found')
-
-    def test_malformed_body_is_refused_not_a_crash(self):
-        self.assertEqual(self.request(body=b'not json')[1], '403 Forbidden') # no secret: turned away before the body is read
-        for body in (b'not json', b'', b'[1, 2]', b'7'):
-            self.assertEqual(self.request(body=body, headers=self.SECRET)[1], '400 Bad Request', body)
-        self.assertEqual(self.errors, [])
-
-    def test_strangers_cannot_post_into_the_chats(self):
-        self.patch(bot, 'config_slackOutgoingWebhook', 'https://example.com/slack')
-        self.patch(bot, 'config_slackOutgoingToken', 'stok')
-        for path, body, headers, status in (('/', b'', {}, '404 Not Found'), ('/wp-login.php', b'log=x&pwd=y', {}, '404 Not Found'),
-                ('/telegram', b'log=x', {}, '403 Forbidden'), ('/telegram', {'message': {}}, {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'wrong'}, '403 Forbidden'),
-                ('/slack', b'', {}, '400 Bad Request'), ('/slack', b'token=stok', {}, '400 Bad Request'), ('/slack', {'type': 'event_callback'}, {}, '403 Forbidden'),
-                ('/slack', {'token': 'wrong', 'type': 'event_callback'}, {}, '403 Forbidden'), ('/slack', {'token': 'stok'}, {}, '200 OK')):
-            with self.subTest(path=path, body=body):
-                self.assertEqual(self.request(path, body, headers)[1], status)
-        self.assertEqual(self.errors, []) # nothing reached the chats
-        self.assertEqual(self.sent, [])
-
-    def test_a_failure_before_authentication_is_only_logged(self):
-        def broken(*a, **k):
-            raise RuntimeError('boom')
-        self.patch(bot, 'urlparse', broken) # fails while routing, before the secret is even looked at
-        env = {'PATH_INFO': '/telegram', 'wsgi.input': io.BytesIO(b'{}'), **self.SECRET}
-        status = []
-        with contextlib.redirect_stderr(io.StringIO()) as err:
-            bot.main(env, lambda s, *a, **k: status.append(s))
-        self.assertEqual((status[-1][:3], self.errors), ('500', []))
-        self.assertIn('boom', err.getvalue())
-
-    def test_a_failure_after_authentication_is_reported_once_and_not_retried(self):
-        broken = {'message': {'message_id': 1, 'from': {'first_name': 'A'}, 'text': '.help'}} # no chat
-        self.assertEqual(self.request(body=broken, headers=self.SECRET)[1], '200 OK') # a retry would only fail and report again
-        self.assertEqual(len(self.errors), 1)
-        self.assertIn("KeyError: 'chat'", self.errors[0])
-
-    def test_telegram_edge_updates(self):
-        secret = self.SECRET
-        self.assertEqual(self.request(body={'edited_message': {}}, headers=secret)[0], b'Unsupported')
-        self.assertEqual(self.request(body={'callback_query': {'id': '1'}}, headers=secret)[0], b'Unsupported')
-        odd = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'new_chat_members': []}}
-        self.assertIn(b'Unhandled', self.request(body=odd, headers=secret)[0])
-
-    def test_a_telegram_update_sent_again_runs_once(self):
-        calls = []
-        self.patch(worker, 'process_request', lambda *a: calls.append(a))
-        body = {'update_id': 7, 'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.price'}}
-        for _ in range(2): # Telegram sends an update again when it got no answer
-            self.assertEqual(self.request(body=body, headers=self.SECRET)[1], '200 OK')
-        gevent.sleep(0.05)
-        self.assertEqual(len(calls), 1)
-
-    def test_button_presses_are_logged(self):
-        calls = []
-        self.patch(worker, 'process_callback', lambda *a: calls.append(a))
-        secret = {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'tok'}
-        for sender, who in (({'username': 'rob', 'first_name': 'Rob'}, '@rob'), ({'first_name': 'Ann'}, 'Ann'), ({}, 'unknown')):
-            body = json.dumps({'callback_query': {'id': '1', 'from': sender, 'data': 'j|7|abcd', 'message': {'message_id': 9, 'chat': {'id': 55}}}}).encode()
-            err = io.StringIO()
-            with contextlib.redirect_stderr(err):
-                bot.main({'PATH_INFO': '/telegram', 'wsgi.input': io.BytesIO(body), **secret}, lambda *a, **k: None)
-            self.assertIn(f'[Telegram button]: {who} chat 55 j|7|abcd', err.getvalue())
+	SECRET = {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'tok'}
+
+	def request(self, path='/telegram', body=None, headers=None):
+		body = body if isinstance(body, bytes) else json.dumps(body if body is not None else {}).encode()
+		env = {'PATH_INFO': path, 'wsgi.input': io.BytesIO(body), **(headers or {})}
+		status = []
+		with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
+			out = bot.main(env, lambda s, *a, **k: status.append(s))
+		return b''.join(out), status[-1]
+
+	def test_telegram_requires_the_secret_header(self):
+		message = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.help'}}
+		self.assertIn(b'Unauthorized', self.request(body=message)[0])
+		self.assertIn(b'Unauthorized', self.request(body=message, headers={'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'wrong'})[0])
+		self.assertEqual(self.sent, []) # nothing was processed
+
+	def test_unknown_path_is_404(self):
+		self.assertEqual(self.request(path='/nope')[1], '404 Not Found')
+
+	def test_malformed_body_is_refused_not_a_crash(self):
+		self.assertEqual(self.request(body=b'not json')[1], '403 Forbidden') # no secret: turned away before the body is read
+		for body in (b'not json', b'', b'[1, 2]', b'7'):
+			self.assertEqual(self.request(body=body, headers=self.SECRET)[1], '400 Bad Request', body)
+		self.assertEqual(self.errors, [])
+
+	def test_strangers_cannot_post_into_the_chats(self):
+		self.patch(bot, 'config_slackOutgoingWebhook', 'https://example.com/slack')
+		self.patch(bot, 'config_slackOutgoingToken', 'stok')
+		for path, body, headers, status in (('/', b'', {}, '404 Not Found'), ('/wp-login.php', b'log=x&pwd=y', {}, '404 Not Found'),
+				('/telegram', b'log=x', {}, '403 Forbidden'), ('/telegram', {'message': {}}, {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'wrong'}, '403 Forbidden'),
+				('/slack', b'', {}, '400 Bad Request'), ('/slack', b'token=stok', {}, '400 Bad Request'), ('/slack', {'type': 'event_callback'}, {}, '403 Forbidden'),
+				('/slack', {'token': 'wrong', 'type': 'event_callback'}, {}, '403 Forbidden'), ('/slack', {'token': 'stok'}, {}, '200 OK')):
+			with self.subTest(path=path, body=body):
+				self.assertEqual(self.request(path, body, headers)[1], status)
+		self.assertEqual(self.errors, []) # nothing reached the chats
+		self.assertEqual(self.sent, [])
+
+	def test_a_failure_before_authentication_is_only_logged(self):
+		def broken(*a, **k):
+			raise RuntimeError('boom')
+		self.patch(bot, 'urlparse', broken) # fails while routing, before the secret is even looked at
+		env = {'PATH_INFO': '/telegram', 'wsgi.input': io.BytesIO(b'{}'), **self.SECRET}
+		status = []
+		with contextlib.redirect_stderr(io.StringIO()) as err:
+			bot.main(env, lambda s, *a, **k: status.append(s))
+		self.assertEqual((status[-1][:3], self.errors), ('500', []))
+		self.assertIn('boom', err.getvalue())
+
+	def test_a_failure_after_authentication_is_reported_once_and_not_retried(self):
+		broken = {'message': {'message_id': 1, 'from': {'first_name': 'A'}, 'text': '.help'}} # no chat
+		self.assertEqual(self.request(body=broken, headers=self.SECRET)[1], '200 OK') # a retry would only fail and report again
+		self.assertEqual(len(self.errors), 1)
+		self.assertIn("KeyError: 'chat'", self.errors[0])
+
+	def test_telegram_edge_updates(self):
+		secret = self.SECRET
+		self.assertEqual(self.request(body={'edited_message': {}}, headers=secret)[0], b'Unsupported')
+		self.assertEqual(self.request(body={'callback_query': {'id': '1'}}, headers=secret)[0], b'Unsupported')
+		odd = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'new_chat_members': []}}
+		self.assertIn(b'Unhandled', self.request(body=odd, headers=secret)[0])
+
+	def test_a_telegram_update_sent_again_runs_once(self):
+		calls = []
+		self.patch(worker, 'process_request', lambda *a: calls.append(a))
+		body = {'update_id': 7, 'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.price'}}
+		for _ in range(2): # Telegram sends an update again when it got no answer
+			self.assertEqual(self.request(body=body, headers=self.SECRET)[1], '200 OK')
+		gevent.sleep(0.05)
+		self.assertEqual(len(calls), 1)
+
+	def test_button_presses_are_logged(self):
+		calls = []
+		self.patch(worker, 'process_callback', lambda *a: calls.append(a))
+		secret = {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'tok'}
+		for sender, who in (({'username': 'rob', 'first_name': 'Rob'}, '@rob'), ({'first_name': 'Ann'}, 'Ann'), ({}, 'unknown')):
+			body = json.dumps({'callback_query': {'id': '1', 'from': sender, 'data': 'j|7|abcd', 'message': {'message_id': 9, 'chat': {'id': 55}}}}).encode()
+			err = io.StringIO()
+			with contextlib.redirect_stderr(err):
+				bot.main({'PATH_INFO': '/telegram', 'wsgi.input': io.BytesIO(body), **secret}, lambda *a, **k: None)
+			self.assertIn(f'[Telegram button]: {who} chat 55 j|7|abcd', err.getvalue())
 
 
 class SlackEvents(FinbotCase):
-    """Slack sends every message in a subscribed channel, the bot's own posts included."""
-
-    def setUp(self):
-        super().setUp()
-        self.patch(bot, 'config_slackOutgoingWebhook', 'https://example.com/slack')
-        self.patch(bot, 'config_slackOutgoingToken', 'stok')
-        self.requests = []
-        self.patch(worker, 'process_request', lambda *a: self.requests.append(a))
-
-    def event(self, event_id=None, **event):
-        body = json.dumps({'token': 'stok', 'type': 'event_callback', 'authorizations': [{'user_id': 'UBOT'}], 'event_id': event_id,
-            'event': dict({'type': 'message', 'channel': 'C1', 'ts': '1.2'}, **event)}).encode()
-        status = []
-        with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
-            bot.main({'PATH_INFO': '/slack', 'wsgi.input': io.BytesIO(body)}, lambda s, *a, **k: status.append(s))
-        gevent.sleep(0.05) # the worker runs in a (gevent) thread
-        return status[-1]
-
-    def test_only_people_are_answered(self):
-        for event in ({'subtype': 'bot_message', 'bot_id': 'B1', 'text': "⚠️ /slack: KeyError: 'user'"}, # a webhook report
-                      {'user': 'UBOT', 'bot_id': 'B2', 'text': '.holdings: Please try again'}, # this bot's own reply
-                      {'user': 'UBOT', 'text': '.holdings'},
-                      {'subtype': 'channel_join', 'user': 'U1', 'text': '<@U1> has joined the channel'},
-                      {'subtype': 'message_changed', 'message': {'text': '.price'}},
-                      {'text': '.price'}): # no user at all
-            with self.subTest(event=event):
-                self.assertEqual(self.event(**event), '200 OK')
-        self.assertEqual((self.requests, self.errors), ([], []))
-        self.assertEqual(self.event(user='U1', text='.price <@UBOT>'), '200 OK')
-        self.assertEqual(self.event(user='U2', subtype='thread_broadcast', text='.trades'), '200 OK')
-        self.assertEqual([(r[0], r[1], r[2], r[3], r[4]) for r in self.requests],
-                         [('slack', 'C1', '<@U1>', '.price @UBOT', '@UBOT'), ('slack', 'C1', '<@U2>', '.trades', '@UBOT')])
-
-    def test_a_retried_event_runs_once(self):
-        for _ in range(2): # Slack resends an event it got no answer to within 3 seconds, with the same event_id
-            self.assertEqual(self.event('Ev1', user='U1', text='.price'), '200 OK')
-        self.assertEqual(self.event('Ev2', user='U1', text='.price'), '200 OK')
-        self.assertEqual(len(self.requests), 2)
-
-    def test_slack_does_not_change_the_telegram_bot_name(self):
-        self.patch(bot, 'botName', '@finbot_tg')
-        self.event(user='U1', text='.price')
-        self.assertEqual(bot.botName, '@finbot_tg')
+	"""Slack sends every message in a subscribed channel, the bot's own posts included."""
+
+	def setUp(self):
+		super().setUp()
+		self.patch(bot, 'config_slackOutgoingWebhook', 'https://example.com/slack')
+		self.patch(bot, 'config_slackOutgoingToken', 'stok')
+		self.requests = []
+		self.patch(worker, 'process_request', lambda *a: self.requests.append(a))
+
+	def event(self, event_id=None, **event):
+		body = json.dumps({'token': 'stok', 'type': 'event_callback', 'authorizations': [{'user_id': 'UBOT'}], 'event_id': event_id,
+			'event': dict({'type': 'message', 'channel': 'C1', 'ts': '1.2'}, **event)}).encode()
+		status = []
+		with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
+			bot.main({'PATH_INFO': '/slack', 'wsgi.input': io.BytesIO(body)}, lambda s, *a, **k: status.append(s))
+		gevent.sleep(0.05) # the worker runs in a (gevent) thread
+		return status[-1]
+
+	def test_only_people_are_answered(self):
+		for event in ({'subtype': 'bot_message', 'bot_id': 'B1', 'text': "⚠️ /slack: KeyError: 'user'"}, # a webhook report
+					  {'user': 'UBOT', 'bot_id': 'B2', 'text': '.holdings: Please try again'}, # this bot's own reply
+					  {'user': 'UBOT', 'text': '.holdings'},
+					  {'subtype': 'channel_join', 'user': 'U1', 'text': '<@U1> has joined the channel'},
+					  {'subtype': 'message_changed', 'message': {'text': '.price'}},
+					  {'text': '.price'}): # no user at all
+			with self.subTest(event=event):
+				self.assertEqual(self.event(**event), '200 OK')
+		self.assertEqual((self.requests, self.errors), ([], []))
+		self.assertEqual(self.event(user='U1', text='.price <@UBOT>'), '200 OK')
+		self.assertEqual(self.event(user='U2', subtype='thread_broadcast', text='.trades'), '200 OK')
+		self.assertEqual([(r[0], r[1], r[2], r[3], r[4]) for r in self.requests],
+						 [('slack', 'C1', '<@U1>', '.price @UBOT', '@UBOT'), ('slack', 'C1', '<@U2>', '.trades', '@UBOT')])
+
+	def test_a_retried_event_runs_once(self):
+		for _ in range(2): # Slack resends an event it got no answer to within 3 seconds, with the same event_id
+			self.assertEqual(self.event('Ev1', user='U1', text='.price'), '200 OK')
+		self.assertEqual(self.event('Ev2', user='U1', text='.price'), '200 OK')
+		self.assertEqual(len(self.requests), 2)
+
+	def test_slack_does_not_change_the_telegram_bot_name(self):
+		self.patch(bot, 'botName', '@finbot_tg')
+		self.event(user='U1', text='.price')
+		self.assertEqual(bot.botName, '@finbot_tg')
 
 
 class UserTextIsEscaped(FinbotCase):
-    """Replies go out as Telegram HTML: text a user wrote is shown, never read as markup. One stray < in a .plan entry made Telegram
-    refuse that reply and every .plan reply after it."""
-
-    def test_a_plan_is_shown_as_text(self):
-        reply = self.command('.plan buy <cheap> & hold')[0][1]
-        self.assertIn('<i>buy &lt;cheap&gt; &amp; hold</i>', reply[0])
-        self.assertEqual(self.command('.plan')[0][1], reply) # and the same way again
-        self.sent.clear()
-        worker.process_request('slack', 'C1', '<@U1>', '.plan sell &lt;dear&gt;', BOT, 'U') # Slack escapes what it sends
-        text = '\n'.join(self.sent[0][1])
-        self.assertIn('*<@U1>*', text) # a Slack mention stays markup
-        self.assertIn('_sell &lt;dear&gt;_', text) # escaped once, not twice
-        self.assertIn('_buy &lt;cheap&gt; &amp; hold_', text)
-
-    def test_a_telegram_name_is_escaped_where_it_arrives(self):
-        requests = []
-        self.patch(worker, 'process_request', lambda *a: requests.append(a))
-        body = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A<b>&'}, 'text': '.hello'}}
-        env = {'PATH_INFO': '/telegram', 'wsgi.input': io.BytesIO(json.dumps(body).encode()), 'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'tok'}
-        with contextlib.redirect_stderr(io.StringIO()):
-            bot.main(env, lambda *a, **k: None)
-        gevent.sleep(0.05) # the worker runs in a (gevent) thread
-        self.assertEqual((requests[0][2], requests[0][5]), ('@A&lt;b&gt;&amp;', '@A&lt;b&gt;&amp;')) # the handle and the name
+	"""Replies go out as Telegram HTML: text a user wrote is shown, never read as markup. One stray < in a .plan entry made Telegram
+	refuse that reply and every .plan reply after it."""
+
+	def test_a_plan_is_shown_as_text(self):
+		reply = self.command('.plan buy <cheap> & hold')[0][1]
+		self.assertIn('<i>buy &lt;cheap&gt; &amp; hold</i>', reply[0])
+		self.assertEqual(self.command('.plan')[0][1], reply) # and the same way again
+		self.sent.clear()
+		worker.process_request('slack', 'C1', '<@U1>', '.plan sell &lt;dear&gt;', BOT, 'U') # Slack escapes what it sends
+		text = '\n'.join(self.sent[0][1])
+		self.assertIn('*<@U1>*', text) # a Slack mention stays markup
+		self.assertIn('_sell &lt;dear&gt;_', text) # escaped once, not twice
+		self.assertIn('_buy &lt;cheap&gt; &amp; hold_', text)
+
+	def test_a_telegram_name_is_escaped_where_it_arrives(self):
+		requests = []
+		self.patch(worker, 'process_request', lambda *a: requests.append(a))
+		body = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A<b>&'}, 'text': '.hello'}}
+		env = {'PATH_INFO': '/telegram', 'wsgi.input': io.BytesIO(json.dumps(body).encode()), 'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'tok'}
+		with contextlib.redirect_stderr(io.StringIO()):
+			bot.main(env, lambda *a, **k: None)
+		gevent.sleep(0.05) # the worker runs in a (gevent) thread
+		self.assertEqual((requests[0][2], requests[0][5]), ('@A&lt;b&gt;&amp;', '@A&lt;b&gt;&amp;')) # the handle and the name
 
 
 class SecretsStayOutOfLogs(FinbotCase):
-    """A failed request's error message carries its URL: the Telegram token, a Slack or Discord webhook path, a Yahoo crumb."""
-    SECRET_URL = 'https://api.telegram.org/bot123456:SECRET-tok_en/sendMessage?chat_id=1'
-
-    def stderr_of(self, call):
-        err = io.StringIO()
-        with contextlib.redirect_stderr(err), contextlib.redirect_stdout(io.StringIO()):
-            call()
-        return err.getvalue()
-
-    def test_a_failed_send_logs_no_token(self):
-        def refuse(url, **kwargs):
-            raise ConnectionError(f"Max retries exceeded with url: {url} (headers {kwargs.get('headers')})")
-        self.patch(webhook.net, 'post', refuse)
-        self.patch(webhook, 'config_slackBotToken', 'xoxb-SECRET')
-        log = self.stderr_of(lambda: webhook.write('telegram', self.SECRET_URL, 'hi', '1'))
-        log += self.stderr_of(lambda: webhook.write('slack', 'https://hooks.slack.com/services/T0/B0/SECRETpath', 'hi', 'C1'))
-        self.assertIn('Failure sending', log)
-        for secret in ('SECRET-tok_en', 'SECRETpath', 'xoxb-SECRET'):
-            self.assertNotIn(secret, log)
-
-    def test_the_traceback_of_a_reported_error_is_redacted(self):
-        self.patch(webhook, 'report_error', REAL_REPORT_ERROR)
-        self.patch(webhook, 'write', lambda *a, **k: None)
-        try:
-            raise ConnectionError('failed: ' + self.SECRET_URL)
-        except ConnectionError as e:
-            log = self.stderr_of(lambda: webhook.report_error(e, 'telegram', '55', context='x'))
-        self.assertIn('Traceback', log)
-        self.assertIn('bot<redacted>', log)
-        self.assertNotIn('SECRET-tok_en', log)
+	"""A failed request's error message carries its URL: the Telegram token, a Slack or Discord webhook path, a Yahoo crumb."""
+	SECRET_URL = 'https://api.telegram.org/bot123456:SECRET-tok_en/sendMessage?chat_id=1'
+
+	def stderr_of(self, call):
+		err = io.StringIO()
+		with contextlib.redirect_stderr(err), contextlib.redirect_stdout(io.StringIO()):
+			call()
+		return err.getvalue()
+
+	def test_a_failed_send_logs_no_token(self):
+		def refuse(url, **kwargs):
+			raise ConnectionError(f"Max retries exceeded with url: {url} (headers {kwargs.get('headers')})")
+		self.patch(webhook.net, 'post', refuse)
+		self.patch(webhook, 'config_slackBotToken', 'xoxb-SECRET')
+		log = self.stderr_of(lambda: webhook.write('telegram', self.SECRET_URL, 'hi', '1'))
+		log += self.stderr_of(lambda: webhook.write('slack', 'https://hooks.slack.com/services/T0/B0/SECRETpath', 'hi', 'C1'))
+		self.assertIn('Failure sending', log)
+		for secret in ('SECRET-tok_en', 'SECRETpath', 'xoxb-SECRET'):
+			self.assertNotIn(secret, log)
+
+	def test_the_traceback_of_a_reported_error_is_redacted(self):
+		self.patch(webhook, 'report_error', REAL_REPORT_ERROR)
+		self.patch(webhook, 'write', lambda *a, **k: None)
+		try:
+			raise ConnectionError('failed: ' + self.SECRET_URL)
+		except ConnectionError as e:
+			log = self.stderr_of(lambda: webhook.report_error(e, 'telegram', '55', context='x'))
+		self.assertIn('Traceback', log)
+		self.assertIn('bot<redacted>', log)
+		self.assertNotIn('SECRET-tok_en', log)
 
 
 class SharedSession(unittest.TestCase):
-    def test_the_session_keeps_no_cookies(self):
-        """Yahoo's cookie goes in an explicit header: a stored one would be sent to fc.yahoo.com, which then sets no new one."""
-        import email.message, urllib.request
-        from lib import net
-        headers = email.message.Message()
-        headers['Set-Cookie'] = 'A3=d=AQAB; Max-Age=31557600; Domain=.yahoo.com; Path=/; Secure'
-        response = type('Response', (), {'info': lambda self: headers})()
-        net.session.cookies.extract_cookies(response, urllib.request.Request('https://fc.yahoo.com/'))
-        self.assertEqual(len(net.session.cookies), 0)
+	def test_the_session_keeps_no_cookies(self):
+		"""Yahoo's cookie goes in an explicit header: a stored one would be sent to fc.yahoo.com, which then sets no new one."""
+		import email.message, urllib.request
+		from lib import net
+		headers = email.message.Message()
+		headers['Set-Cookie'] = 'A3=d=AQAB; Max-Age=31557600; Domain=.yahoo.com; Path=/; Secure'
+		response = type('Response', (), {'info': lambda self: headers})()
+		net.session.cookies.extract_cookies(response, urllib.request.Request('https://fc.yahoo.com/'))
+		self.assertEqual(len(net.session.cookies), 0)
 
 
 class StartupSignals(unittest.TestCase):
-    """A reload (SIGHUP) that arrives while the bot is still starting must not kill it."""
-
-    def test_sighup_during_startup_is_ignored_and_works_afterwards(self):
-        skip_if_old_env(self)
-        with tempfile.TemporaryDirectory() as data:
-            env = {k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k not in config.STARTUP_KEYS}
-            env.update(DATA_DIR=data, port='0', scheduler='False')
-            proc = subprocess.Popen([sys.executable, '-u', 'bot.py'], cwd=ROOT, env=env, stderr=subprocess.PIPE, stdout=subprocess.DEVNULL, text=True)
-            self.addCleanup(proc.wait)
-            self.addCleanup(proc.kill)
-            deadline = time.monotonic() + 30
-            while not int(Path(f'/proc/{proc.pid}/status').read_text().split('SigIgn:')[1].split()[0], 16) & 1: # until SIGHUP is ignored
-                self.assertLess(time.monotonic(), deadline)
-                time.sleep(0.001)
-            proc.send_signal(signal.SIGHUP) # still importing: the default action would kill it
-            lines = []
-            for line in proc.stderr:
-                lines.append(line)
-                if line.startswith('Opening socket'):
-                    break
-            self.assertIsNone(proc.poll(), ''.join(lines))
-            time.sleep(1) # the handler goes in just before serving
-            proc.send_signal(signal.SIGHUP)
-            for line in proc.stderr: # a reload now runs (there is no preferences.conf here, so it says so)
-                lines.append(line)
-                if 'reload requested' in line:
-                    break
-            self.assertIsNone(proc.poll(), ''.join(lines))
-            self.assertIn('reload requested but', ''.join(lines))
-
-    def test_usr1_flushes_the_cache_and_keeps_the_state(self):
-        skip_if_old_env(self)
-        with tempfile.TemporaryDirectory() as data:
-            cache = Path(data, 'cache')
-            (cache / 'matplotlib').mkdir(parents=True)
-            for name in ('finbot_yahoo_cookie.json', 'finbot_graph_T1_None.png', 'finbot_x.json.abc.tmp'): # the last: a write in flight
-                (cache / name).write_text('{}')
-            Path(data, 'finbot_watchlist.json').write_text('["T1"]') # state, not cache
-            env = {k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k not in config.STARTUP_KEYS}
-            env.update(DATA_DIR=data, port='0', scheduler='False')
-            proc = subprocess.Popen([sys.executable, '-u', 'bot.py'], cwd=ROOT, env=env, stderr=subprocess.PIPE, stdout=subprocess.DEVNULL, text=True)
-            self.addCleanup(proc.wait)
-            self.addCleanup(proc.kill)
-            deadline = time.monotonic() + 30
-            while not int(Path(f'/proc/{proc.pid}/status').read_text().split('SigIgn:')[1].split()[0], 16) & (1 << (signal.SIGUSR1 - 1)):
-                self.assertLess(time.monotonic(), deadline)
-                time.sleep(0.001)
-            proc.send_signal(signal.SIGUSR1) # still starting: ignored, where its default action would kill the bot
-            lines = []
-            for line in proc.stderr:
-                lines.append(line)
-                if line.startswith('Opening socket'):
-                    break
-            self.assertIsNone(proc.poll(), ''.join(lines))
-            self.assertEqual(len(os.listdir(cache)), 4) # nothing flushed yet
-            time.sleep(1) # the handler goes in just before serving
-            proc.send_signal(signal.SIGUSR1)
-            for line in proc.stderr:
-                lines.append(line)
-                if 'cache: flushed' in line:
-                    break
-            self.assertIn('cache: flushed 2 files', ''.join(lines))
-            self.assertEqual(sorted(os.listdir(cache)), ['finbot_x.json.abc.tmp', 'matplotlib'])
-            self.assertTrue(Path(data, 'finbot_watchlist.json').exists())
+	"""A reload (SIGHUP) that arrives while the bot is still starting must not kill it."""
+
+	def test_sighup_during_startup_is_ignored_and_works_afterwards(self):
+		skip_if_old_env(self)
+		with tempfile.TemporaryDirectory() as data:
+			env = {k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k not in config.STARTUP_KEYS}
+			env.update(DATA_DIR=data, port='0', scheduler='False')
+			proc = subprocess.Popen([sys.executable, '-u', 'bot.py'], cwd=ROOT, env=env, stderr=subprocess.PIPE, stdout=subprocess.DEVNULL, text=True)
+			self.addCleanup(proc.wait)
+			self.addCleanup(proc.kill)
+			deadline = time.monotonic() + 30
+			while not int(Path(f'/proc/{proc.pid}/status').read_text().split('SigIgn:')[1].split()[0], 16) & 1: # until SIGHUP is ignored
+				self.assertLess(time.monotonic(), deadline)
+				time.sleep(0.001)
+			proc.send_signal(signal.SIGHUP) # still importing: the default action would kill it
+			lines = []
+			for line in proc.stderr:
+				lines.append(line)
+				if line.startswith('Opening socket'):
+					break
+			self.assertIsNone(proc.poll(), ''.join(lines))
+			time.sleep(1) # the handler goes in just before serving
+			proc.send_signal(signal.SIGHUP)
+			for line in proc.stderr: # a reload now runs (there is no preferences.conf here, so it says so)
+				lines.append(line)
+				if 'reload requested' in line:
+					break
+			self.assertIsNone(proc.poll(), ''.join(lines))
+			self.assertIn('reload requested but', ''.join(lines))
+
+	def test_usr1_flushes_the_cache_and_keeps_the_state(self):
+		skip_if_old_env(self)
+		with tempfile.TemporaryDirectory() as data:
+			cache = Path(data, 'cache')
+			(cache / 'matplotlib').mkdir(parents=True)
+			for name in ('finbot_yahoo_cookie.json', 'finbot_graph_T1_None.png', 'finbot_x.json.abc.tmp'): # the last: a write in flight
+				(cache / name).write_text('{}')
+			Path(data, 'finbot_watchlist.json').write_text('["T1"]') # state, not cache
+			env = {k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k not in config.STARTUP_KEYS}
+			env.update(DATA_DIR=data, port='0', scheduler='False')
+			proc = subprocess.Popen([sys.executable, '-u', 'bot.py'], cwd=ROOT, env=env, stderr=subprocess.PIPE, stdout=subprocess.DEVNULL, text=True)
+			self.addCleanup(proc.wait)
+			self.addCleanup(proc.kill)
+			deadline = time.monotonic() + 30
+			while not int(Path(f'/proc/{proc.pid}/status').read_text().split('SigIgn:')[1].split()[0], 16) & (1 << (signal.SIGUSR1 - 1)):
+				self.assertLess(time.monotonic(), deadline)
+				time.sleep(0.001)
+			proc.send_signal(signal.SIGUSR1) # still starting: ignored, where its default action would kill the bot
+			lines = []
+			for line in proc.stderr:
+				lines.append(line)
+				if line.startswith('Opening socket'):
+					break
+			self.assertIsNone(proc.poll(), ''.join(lines))
+			self.assertEqual(len(os.listdir(cache)), 4) # nothing flushed yet
+			time.sleep(1) # the handler goes in just before serving
+			proc.send_signal(signal.SIGUSR1)
+			for line in proc.stderr:
+				lines.append(line)
+				if 'cache: flushed' in line:
+					break
+			self.assertIn('cache: flushed 2 files', ''.join(lines))
+			self.assertEqual(sorted(os.listdir(cache)), ['finbot_x.json.abc.tmp', 'matplotlib'])
+			self.assertTrue(Path(data, 'finbot_watchlist.json').exists())
 
 
 if __name__ == '__main__':
-    unittest.main()
+	unittest.main()
diff --git a/tests/test_finbot.py b/tests/test_finbot.py
index 5eec41e3..fa8c9b2f 100644
--- a/tests/test_finbot.py
+++ b/tests/test_finbot.py
@@ -6,97 +6,97 @@
 
 
 class Plumbing(FinbotCase):
-    def test_chunked_send_keeps_keyboard_on_last_chunk(self):
-        posted = []
-
-        class Response:
-            status_code = 200
-            def json(self): return {'ok': True, 'result': {'message_id': len(posted), 'chat': {'id': 5}}}
-
-        def post(url, **kwargs):
-            posted.append(kwargs['json'])
-            return Response()
-
-        with mock.patch.object(webhook.net, 'post', post), mock.patch.object(webhook.time, 'sleep', lambda s: None):
-            results = REAL_PAYLOAD_WRAPPER('telegram', 'https://api.telegram.org/botX/sendMessage?chat_id=5',
-                ['x' * 60] * 300, '5', reply_markup={'keyboard': []})
-        self.assertGreater(len(results), 1)
-        self.assertEqual(len(posted), len(results))
-        self.assertIn('reply_markup', posted[-1])
-        self.assertTrue(all('reply_markup' not in p for p in posted[:-1]))
-
-    def test_chunks_respect_both_the_line_count_and_the_size(self):
-        chunks = webhook.chunked(['a' * 30] * 10 + ['b' * 500], limit=100, max_lines=4)
-        self.assertEqual([len(chunk) for chunk in chunks], [3, 3, 3, 1, 1]) # 3 x 30 plus newlines is 92; a fourth would pass 100
-        self.assertTrue(all(len('\n'.join(chunk)) <= 100 for chunk in chunks))
-        self.assertEqual(chunks[-1], ['b' * 99 + '…']) # a line over the limit on its own is cut short
-        self.assertEqual(webhook.chunked(['x'] * 9, limit=4096, max_lines=4), [['x'] * 4, ['x'] * 4, ['x']])
-
-    def test_zoneinfo_offset_is_not_lmt(self):
-        tz = zoneinfo.ZoneInfo('Australia/Melbourne')
-        self.assertEqual(datetime.datetime(2022, 8, 18, tzinfo=tz).utcoffset(), datetime.timedelta(hours=10))
-
-    def test_reminder_runs(self):
-        with mock.patch.object(reminder, 'data_json', lambda name, default=None: []):
-            reminder.lambda_handler()
-
-    def test_bot_handler(self):
-        def post(message):
-            body = json.dumps({'update_id': 1, 'message': dict({'message_id': 5, 'chat': {'id': 55, 'type': 'private'},
-                'from': {'id': 9, 'first_name': 'Rob'}, 'date': 1}, **message)}).encode()
-            env = {'PATH_INFO': '/telegram', 'wsgi.input': io.BytesIO(body), 'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'tok'}
-            err = io.StringIO()
-            with contextlib.redirect_stderr(err):
-                out = bot.main(env, lambda *a, **k: None)
-            return out, err.getvalue()
-        self.assertEqual(post({'sticker': {'file_id': 'x'}}), ([b''], '')) # stickers: silent, not logged
-        self.assertGreaterEqual(len(bot.git_version()), 7)
-        head = subprocess.run(['git', 'rev-parse', 'HEAD'], cwd=bot.__file__.rsplit('/', 1)[0], capture_output=True, text=True).stdout.strip()
-        self.assertEqual(bot.git_version(), head[:7]) # matches what git says
+	def test_chunked_send_keeps_keyboard_on_last_chunk(self):
+		posted = []
+
+		class Response:
+			status_code = 200
+			def json(self): return {'ok': True, 'result': {'message_id': len(posted), 'chat': {'id': 5}}}
+
+		def post(url, **kwargs):
+			posted.append(kwargs['json'])
+			return Response()
+
+		with mock.patch.object(webhook.net, 'post', post), mock.patch.object(webhook.time, 'sleep', lambda s: None):
+			results = REAL_PAYLOAD_WRAPPER('telegram', 'https://api.telegram.org/botX/sendMessage?chat_id=5',
+				['x' * 60] * 300, '5', reply_markup={'keyboard': []})
+		self.assertGreater(len(results), 1)
+		self.assertEqual(len(posted), len(results))
+		self.assertIn('reply_markup', posted[-1])
+		self.assertTrue(all('reply_markup' not in p for p in posted[:-1]))
+
+	def test_chunks_respect_both_the_line_count_and_the_size(self):
+		chunks = webhook.chunked(['a' * 30] * 10 + ['b' * 500], limit=100, max_lines=4)
+		self.assertEqual([len(chunk) for chunk in chunks], [3, 3, 3, 1, 1]) # 3 x 30 plus newlines is 92; a fourth would pass 100
+		self.assertTrue(all(len('\n'.join(chunk)) <= 100 for chunk in chunks))
+		self.assertEqual(chunks[-1], ['b' * 99 + '…']) # a line over the limit on its own is cut short
+		self.assertEqual(webhook.chunked(['x'] * 9, limit=4096, max_lines=4), [['x'] * 4, ['x'] * 4, ['x']])
+
+	def test_zoneinfo_offset_is_not_lmt(self):
+		tz = zoneinfo.ZoneInfo('Australia/Melbourne')
+		self.assertEqual(datetime.datetime(2022, 8, 18, tzinfo=tz).utcoffset(), datetime.timedelta(hours=10))
+
+	def test_reminder_runs(self):
+		with mock.patch.object(reminder, 'data_json', lambda name, default=None: []):
+			reminder.lambda_handler()
+
+	def test_bot_handler(self):
+		def post(message):
+			body = json.dumps({'update_id': 1, 'message': dict({'message_id': 5, 'chat': {'id': 55, 'type': 'private'},
+				'from': {'id': 9, 'first_name': 'Rob'}, 'date': 1}, **message)}).encode()
+			env = {'PATH_INFO': '/telegram', 'wsgi.input': io.BytesIO(body), 'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'tok'}
+			err = io.StringIO()
+			with contextlib.redirect_stderr(err):
+				out = bot.main(env, lambda *a, **k: None)
+			return out, err.getvalue()
+		self.assertEqual(post({'sticker': {'file_id': 'x'}}), ([b''], '')) # stickers: silent, not logged
+		self.assertGreaterEqual(len(bot.git_version()), 7)
+		head = subprocess.run(['git', 'rev-parse', 'HEAD'], cwd=bot.__file__.rsplit('/', 1)[0], capture_output=True, text=True).stdout.strip()
+		self.assertEqual(bot.git_version(), head[:7]) # matches what git says
 
 
 class GitVersion(unittest.TestCase):
-    SHA = '0123456789abcdef0123456789abcdef01234567'
-
-    def repo(self, tmp, head, refs=None, packed=None):
-        git = Path(tmp, '.git')
-        git.mkdir(parents=True)
-        (git / 'HEAD').write_text(head + '\n')
-        for ref, sha in (refs or {}).items():
-            (git / ref).parent.mkdir(parents=True, exist_ok=True)
-            (git / ref).write_text(sha + '\n')
-        if packed:
-            (git / 'packed-refs').write_text('# pack-refs with: peeled fully-peeled sorted\n' + packed)
-        return git
-
-    def test_loose_ref_detached_and_packed(self):
-        with tempfile.TemporaryDirectory() as a, tempfile.TemporaryDirectory() as b, tempfile.TemporaryDirectory() as c:
-            self.repo(a, 'ref: refs/heads/work', {'refs/heads/work': self.SHA})
-            self.repo(b, self.SHA)
-            self.repo(c, 'ref: refs/heads/work', packed=f'{self.SHA} refs/heads/work\n{"f" * 40} refs/heads/other\n')
-            self.assertEqual([bot.git_version(d) for d in (a, b, c)], ['0123456'] * 3)
-
-    def test_linked_worktree(self):
-        with tempfile.TemporaryDirectory() as main, tempfile.TemporaryDirectory() as tree:
-            git = self.repo(main, 'ref: refs/heads/main', {'refs/heads/main': self.SHA})
-            linked = git / 'worktrees' / 'w'
-            linked.mkdir(parents=True)
-            (linked / 'HEAD').write_text('ref: refs/heads/main\n')
-            (linked / 'commondir').write_text('../..\n')
-            Path(tree, '.git').write_text(f'gitdir: {linked}\n')
-            self.assertEqual(bot.git_version(tree), '0123456')
-
-    def test_unreadable_is_unknown(self):
-        with tempfile.TemporaryDirectory() as empty, tempfile.TemporaryDirectory() as odd:
-            self.repo(odd, 'ref: refs/heads/gone')
-            self.assertEqual([bot.git_version(empty), bot.git_version(odd), bot.git_version('/nonexistent')], ['unknown'] * 3)
-            self.repo(empty, 'garbage')
-            self.assertEqual(bot.git_version(empty), 'unknown')
-
-    def test_does_not_run_git(self):
-        with mock.patch('subprocess.run', side_effect=AssertionError('git was run')):
-            self.assertGreaterEqual(len(bot.git_version()), 7)
+	SHA = '0123456789abcdef0123456789abcdef01234567'
+
+	def repo(self, tmp, head, refs=None, packed=None):
+		git = Path(tmp, '.git')
+		git.mkdir(parents=True)
+		(git / 'HEAD').write_text(head + '\n')
+		for ref, sha in (refs or {}).items():
+			(git / ref).parent.mkdir(parents=True, exist_ok=True)
+			(git / ref).write_text(sha + '\n')
+		if packed:
+			(git / 'packed-refs').write_text('# pack-refs with: peeled fully-peeled sorted\n' + packed)
+		return git
+
+	def test_loose_ref_detached_and_packed(self):
+		with tempfile.TemporaryDirectory() as a, tempfile.TemporaryDirectory() as b, tempfile.TemporaryDirectory() as c:
+			self.repo(a, 'ref: refs/heads/work', {'refs/heads/work': self.SHA})
+			self.repo(b, self.SHA)
+			self.repo(c, 'ref: refs/heads/work', packed=f'{self.SHA} refs/heads/work\n{"f" * 40} refs/heads/other\n')
+			self.assertEqual([bot.git_version(d) for d in (a, b, c)], ['0123456'] * 3)
+
+	def test_linked_worktree(self):
+		with tempfile.TemporaryDirectory() as main, tempfile.TemporaryDirectory() as tree:
+			git = self.repo(main, 'ref: refs/heads/main', {'refs/heads/main': self.SHA})
+			linked = git / 'worktrees' / 'w'
+			linked.mkdir(parents=True)
+			(linked / 'HEAD').write_text('ref: refs/heads/main\n')
+			(linked / 'commondir').write_text('../..\n')
+			Path(tree, '.git').write_text(f'gitdir: {linked}\n')
+			self.assertEqual(bot.git_version(tree), '0123456')
+
+	def test_unreadable_is_unknown(self):
+		with tempfile.TemporaryDirectory() as empty, tempfile.TemporaryDirectory() as odd:
+			self.repo(odd, 'ref: refs/heads/gone')
+			self.assertEqual([bot.git_version(empty), bot.git_version(odd), bot.git_version('/nonexistent')], ['unknown'] * 3)
+			self.repo(empty, 'garbage')
+			self.assertEqual(bot.git_version(empty), 'unknown')
+
+	def test_does_not_run_git(self):
+		with mock.patch('subprocess.run', side_effect=AssertionError('git was run')):
+			self.assertGreaterEqual(len(bot.git_version()), 7)
 
 
 if __name__ == '__main__':
-    unittest.main()
+	unittest.main()
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
index e95b1b7e..c65df849 100644
--- a/tests/test_jobs.py
+++ b/tests/test_jobs.py
@@ -12,657 +12,657 @@
 
 
 def at(day, hour, minute, second=0):
-    return datetime.datetime(2026, 10, day, hour, minute, second, tzinfo=UTC) # 13 Oct 2026 is a Tuesday
+	return datetime.datetime(2026, 10, day, hour, minute, second, tzinfo=UTC) # 13 Oct 2026 is a Tuesday
 
 
 class Clock:
-    def __init__(self, now):
-        self.now = now
+	def __init__(self, now):
+		self.now = now
 
-    def __call__(self):
-        return self.now
+	def __call__(self):
+		return self.now
 
 
 class ParseTests(unittest.TestCase):
-    def test_shipped_schedule_has_no_errors(self):
-        parsed, errors = jobs.parse_schedule(DEFAULT_SCHEDULE.read_text(), UTC)
-        self.assertEqual(errors, [])
-        self.assertEqual(len(parsed), 9)
-        weekly = next(j for j in parsed if j.expr == '28 21 * * Sat')
-        self.assertEqual([label for label, _, _ in weekly.commands],
-                         ['cal.py earnings', 'cal.py ex-dividend', 'price.py 7', 'performance.py 7', 'rating.py'])
-
-    def test_pasted_crontab_lines(self):
-        text = ('MAILTO=me\n'
-                '*/20 * * * Mon-Fri ~/finbot/trades.py > ~/finbot.trades 2>&1\n'
-                '28  21 * * Sat { ./cal.py earnings; ./price.py 7 ;} &> /dev/null # weekly (UTC)\n'
-                '@daily /opt/finbot/reminder.py\n')
-        parsed, errors = jobs.parse_schedule(text, UTC)
-        self.assertEqual(errors, [])
-        self.assertEqual([[label for label, _, _ in j.commands] for j in parsed],
-                         [['trades.py'], ['cal.py earnings', 'price.py 7'], ['reminder.py']])
-
-    def test_quoted_arguments_may_hold_anything(self):
-        url = 'https://example.com/feed?a=1&b=2;c>d $x |y'
-        parsed, errors = jobs.parse_schedule(f'01 0-10,23 * * Mon-Fri podcasts.py "{url}" > ~/out 2>&1 # hourly (UTC)\n', UTC)
-        self.assertEqual(errors, [])
-        self.assertEqual(parsed[0].commands[0][2], {'feed_url': url})
-        parsed, errors = jobs.parse_schedule(f"5 4 * * * podcasts.py '{url}'; shorts.py\n", UTC)
-        self.assertEqual([(label[:9], kw) for label, _, kw in parsed[0].commands], [('podcasts.', {'feed_url': url}), ('shorts.py', {})])
-
-    def test_unquoted_shell_syntax_is_still_refused(self):
-        for line in ('0 1 * * * trades.py | tee x', '0 1 * * * trades.py && shorts.py & ', '0 1 * * * trades.py $(date)', '0 1 * * * trades.py `x`',
-                     '0 1 * * * trades.py < in', '0 1 * * * (trades.py)', '0 1 * * * trades.py "unterminated'):
-            parsed, errors = jobs.parse_schedule(line + '\n', UTC)
-            self.assertEqual((parsed, len(errors)), ([], 1), line)
-
-    def test_redirects_and_braces_are_dropped(self):
-        for command in ('trades.py > /dev/null', 'trades.py >/dev/null 2>&1', 'trades.py &> /dev/null', 'trades.py >> ~/x 2>&1',
-                        '{ trades.py ;} > ~/x', '/opt/finbot/trades.py # note (x)'):
-            parsed, errors = jobs.parse_schedule(f'0 1 * * * {command}\n', UTC)
-            self.assertEqual(errors, [], command)
-            self.assertEqual([label for label, _, _ in parsed[0].commands], ['trades.py'], command)
-
-    def test_bin_run_wrapper_is_no_longer_accepted(self):
-        for command in ('bin/run trades.py', 'cd /opt/finbot; trades.py'):
-            parsed, errors = jobs.parse_schedule(f'0 1 * * * {command}\n', UTC)
-            self.assertEqual((parsed, len(errors)), ([], 1), command)
-
-    def test_bad_lines_are_skipped_and_named(self):
-        text = '0 1 * * * ls | wc -l\n0 1 * * * price.py bogus\n0 1 * * * nothere.py\n61 1 * * * trades.py\n0 1 * *\nFOO=bar\n5 4 * * * shorts.py\n'
-        parsed, errors = jobs.parse_schedule(text, UTC)
-        self.assertEqual([j.expr for j in parsed], ['5 4 * * *'])
-        self.assertEqual([number for number, _, _ in errors], [1, 2, 3, 4, 5, 6])
-
-    def test_impossible_times_are_marked_and_reboot_is_just_unsupported(self):
-        for line in ('0 0 30 2 * trades.py', '0 9 31 4,6,9,11 * trades.py', '61 1 * * * trades.py', '0 0 * 13 * trades.py', '@bogus trades.py'):
-            parsed, errors = jobs.parse_schedule(line + '\n', UTC)
-            self.assertEqual(parsed, [], line)
-            self.assertIsInstance(errors[0][2], jobs.ImpossibleTime, line)
-        parsed, errors = jobs.parse_schedule('@reboot trades.py\n0 0 29 2 * trades.py\n', UTC)
-        self.assertEqual([j.expr for j in parsed], ['0 0 29 2 *']) # a leap day does come
-        self.assertNotIsInstance(errors[0][2], jobs.ImpossibleTime)
-
-    def test_cron_tz_applies_to_the_lines_below(self):
-        parsed, _ = jobs.parse_schedule('1 1 * * * trades.py\nCRON_TZ=Australia/Melbourne\n2 2 * * * trades.py\nCRON_TZ=Nowhere/Land\n', UTC)
-        self.assertEqual([j.zone.key for j in parsed], ['UTC', 'Australia/Melbourne'])
-
-    def test_local_zone_follows_tz(self):
-        with mock.patch.dict(os.environ, {'TZ': 'Australia/Melbourne'}):
-            self.assertEqual(jobs.local_zone().key, 'Australia/Melbourne')
+	def test_shipped_schedule_has_no_errors(self):
+		parsed, errors = jobs.parse_schedule(DEFAULT_SCHEDULE.read_text(), UTC)
+		self.assertEqual(errors, [])
+		self.assertEqual(len(parsed), 9)
+		weekly = next(j for j in parsed if j.expr == '28 21 * * Sat')
+		self.assertEqual([label for label, _, _ in weekly.commands],
+						 ['cal.py earnings', 'cal.py ex-dividend', 'price.py 7', 'performance.py 7', 'rating.py'])
+
+	def test_pasted_crontab_lines(self):
+		text = ('MAILTO=me\n'
+				'*/20 * * * Mon-Fri ~/finbot/trades.py > ~/finbot.trades 2>&1\n'
+				'28  21 * * Sat { ./cal.py earnings; ./price.py 7 ;} &> /dev/null # weekly (UTC)\n'
+				'@daily /opt/finbot/reminder.py\n')
+		parsed, errors = jobs.parse_schedule(text, UTC)
+		self.assertEqual(errors, [])
+		self.assertEqual([[label for label, _, _ in j.commands] for j in parsed],
+						 [['trades.py'], ['cal.py earnings', 'price.py 7'], ['reminder.py']])
+
+	def test_quoted_arguments_may_hold_anything(self):
+		url = 'https://example.com/feed?a=1&b=2;c>d $x |y'
+		parsed, errors = jobs.parse_schedule(f'01 0-10,23 * * Mon-Fri podcasts.py "{url}" > ~/out 2>&1 # hourly (UTC)\n', UTC)
+		self.assertEqual(errors, [])
+		self.assertEqual(parsed[0].commands[0][2], {'feed_url': url})
+		parsed, errors = jobs.parse_schedule(f"5 4 * * * podcasts.py '{url}'; shorts.py\n", UTC)
+		self.assertEqual([(label[:9], kw) for label, _, kw in parsed[0].commands], [('podcasts.', {'feed_url': url}), ('shorts.py', {})])
+
+	def test_unquoted_shell_syntax_is_still_refused(self):
+		for line in ('0 1 * * * trades.py | tee x', '0 1 * * * trades.py && shorts.py & ', '0 1 * * * trades.py $(date)', '0 1 * * * trades.py `x`',
+					 '0 1 * * * trades.py < in', '0 1 * * * (trades.py)', '0 1 * * * trades.py "unterminated'):
+			parsed, errors = jobs.parse_schedule(line + '\n', UTC)
+			self.assertEqual((parsed, len(errors)), ([], 1), line)
+
+	def test_redirects_and_braces_are_dropped(self):
+		for command in ('trades.py > /dev/null', 'trades.py >/dev/null 2>&1', 'trades.py &> /dev/null', 'trades.py >> ~/x 2>&1',
+						'{ trades.py ;} > ~/x', '/opt/finbot/trades.py # note (x)'):
+			parsed, errors = jobs.parse_schedule(f'0 1 * * * {command}\n', UTC)
+			self.assertEqual(errors, [], command)
+			self.assertEqual([label for label, _, _ in parsed[0].commands], ['trades.py'], command)
+
+	def test_bin_run_wrapper_is_no_longer_accepted(self):
+		for command in ('bin/run trades.py', 'cd /opt/finbot; trades.py'):
+			parsed, errors = jobs.parse_schedule(f'0 1 * * * {command}\n', UTC)
+			self.assertEqual((parsed, len(errors)), ([], 1), command)
+
+	def test_bad_lines_are_skipped_and_named(self):
+		text = '0 1 * * * ls | wc -l\n0 1 * * * price.py bogus\n0 1 * * * nothere.py\n61 1 * * * trades.py\n0 1 * *\nFOO=bar\n5 4 * * * shorts.py\n'
+		parsed, errors = jobs.parse_schedule(text, UTC)
+		self.assertEqual([j.expr for j in parsed], ['5 4 * * *'])
+		self.assertEqual([number for number, _, _ in errors], [1, 2, 3, 4, 5, 6])
+
+	def test_impossible_times_are_marked_and_reboot_is_just_unsupported(self):
+		for line in ('0 0 30 2 * trades.py', '0 9 31 4,6,9,11 * trades.py', '61 1 * * * trades.py', '0 0 * 13 * trades.py', '@bogus trades.py'):
+			parsed, errors = jobs.parse_schedule(line + '\n', UTC)
+			self.assertEqual(parsed, [], line)
+			self.assertIsInstance(errors[0][2], jobs.ImpossibleTime, line)
+		parsed, errors = jobs.parse_schedule('@reboot trades.py\n0 0 29 2 * trades.py\n', UTC)
+		self.assertEqual([j.expr for j in parsed], ['0 0 29 2 *']) # a leap day does come
+		self.assertNotIsInstance(errors[0][2], jobs.ImpossibleTime)
+
+	def test_cron_tz_applies_to_the_lines_below(self):
+		parsed, _ = jobs.parse_schedule('1 1 * * * trades.py\nCRON_TZ=Australia/Melbourne\n2 2 * * * trades.py\nCRON_TZ=Nowhere/Land\n', UTC)
+		self.assertEqual([j.zone.key for j in parsed], ['UTC', 'Australia/Melbourne'])
+
+	def test_local_zone_follows_tz(self):
+		with mock.patch.dict(os.environ, {'TZ': 'Australia/Melbourne'}):
+			self.assertEqual(jobs.local_zone().key, 'Australia/Melbourne')
 
 
 class JobArguments(unittest.TestCase):
-    def test_price(self):
-        self.assertEqual(price.job(['midsession']), ('price.py midsession', {'midsession': True}))
-        self.assertEqual(price.job(['premarket'])[1], {'premarket': True})
-        self.assertEqual(price.job(['interday'])[1], {'interday': True})
-        self.assertEqual(price.job(['close'])[1], {'close': True})
-        self.assertEqual(price.job(['28']), ('price.py 28', {'threshold': 0, 'days': 28, 'top': 10}))
-        for bad in ([], ['soon'], ['7', '8']):
-            with self.assertRaises(ValueError):
-                price.job(bad)
-
-    def test_cal_and_performance(self):
-        self.assertEqual(cal.job(['earnings']), ('cal.py earnings', {'event': 'earnings'}))
-        self.assertEqual(cal.job(['ex-dividend']), ('cal.py ex-dividend', {'event': 'ex-dividend'}))
-        self.assertEqual(performance.job([]), ('performance.py', {}))
-        self.assertEqual(performance.job(['7']), ('performance.py 7', {'past_days': 7}))
-        for bad, module in ((['x'], cal), ([], cal), (['x'], performance)):
-            with self.assertRaises(ValueError):
-                module.job(bad)
-
-    def test_podcasts(self):
-        url = 'https://example.com/podcast.rss'
-        label, kwargs = podcasts.job([url])
-        self.assertEqual(kwargs, {'feed_url': url})
-        self.assertIn('finbot_podcast_seen_' + label.split()[1], podcasts.get_seen_filename(url)) # the label is the start of the state file's hash
-        self.assertEqual(label, 'podcasts.py ' + hashlib.md5(url.encode()).hexdigest()[:8])
-        for bad in ([], [url, url]):
-            with self.assertRaises(ValueError):
-                podcasts.job(bad)
-
-    def test_cleanup_arguments(self):
-        self.assertEqual(cleanup.job([]), ('cleanup.py', {}))
-        self.assertEqual(cleanup.job(['7']), ('cleanup.py 7', {'days': 7}))
-        for bad in (['x'], ['1', '2']):
-            with self.assertRaises(ValueError):
-                cleanup.job(bad)
-
-    def test_plain_scripts(self):
-        for name, module in jobs.SCRIPTS.items():
-            if name not in ('cal', 'cleanup', 'performance', 'podcasts', 'price'):
-                self.assertEqual(module.job([]), (name + '.py', {}))
-
-    def test_cli_prints_usage_and_exits(self):
-        with mock.patch('sys.stderr', new_callable=io.StringIO) as err, self.assertRaises(SystemExit) as exit_:
-            webhook.cli('price.py', price.job, lambda **k: None, ['bogus'])
-        self.assertEqual(exit_.exception.code, 1)
-        self.assertIn('Usage: price.py [midsession', err.getvalue())
-
-    def test_cli_runs_the_handler(self):
-        calls = []
-        webhook.cli('price.py', price.job, lambda **k: calls.append(k), ['midsession'])
-        self.assertEqual(calls, [{'midsession': True}])
+	def test_price(self):
+		self.assertEqual(price.job(['midsession']), ('price.py midsession', {'midsession': True}))
+		self.assertEqual(price.job(['premarket'])[1], {'premarket': True})
+		self.assertEqual(price.job(['interday'])[1], {'interday': True})
+		self.assertEqual(price.job(['close'])[1], {'close': True})
+		self.assertEqual(price.job(['28']), ('price.py 28', {'threshold': 0, 'days': 28, 'top': 10}))
+		for bad in ([], ['soon'], ['7', '8']):
+			with self.assertRaises(ValueError):
+				price.job(bad)
+
+	def test_cal_and_performance(self):
+		self.assertEqual(cal.job(['earnings']), ('cal.py earnings', {'event': 'earnings'}))
+		self.assertEqual(cal.job(['ex-dividend']), ('cal.py ex-dividend', {'event': 'ex-dividend'}))
+		self.assertEqual(performance.job([]), ('performance.py', {}))
+		self.assertEqual(performance.job(['7']), ('performance.py 7', {'past_days': 7}))
+		for bad, module in ((['x'], cal), ([], cal), (['x'], performance)):
+			with self.assertRaises(ValueError):
+				module.job(bad)
+
+	def test_podcasts(self):
+		url = 'https://example.com/podcast.rss'
+		label, kwargs = podcasts.job([url])
+		self.assertEqual(kwargs, {'feed_url': url})
+		self.assertIn('finbot_podcast_seen_' + label.split()[1], podcasts.get_seen_filename(url)) # the label is the start of the state file's hash
+		self.assertEqual(label, 'podcasts.py ' + hashlib.md5(url.encode()).hexdigest()[:8])
+		for bad in ([], [url, url]):
+			with self.assertRaises(ValueError):
+				podcasts.job(bad)
+
+	def test_cleanup_arguments(self):
+		self.assertEqual(cleanup.job([]), ('cleanup.py', {}))
+		self.assertEqual(cleanup.job(['7']), ('cleanup.py 7', {'days': 7}))
+		for bad in (['x'], ['1', '2']):
+			with self.assertRaises(ValueError):
+				cleanup.job(bad)
+
+	def test_plain_scripts(self):
+		for name, module in jobs.SCRIPTS.items():
+			if name not in ('cal', 'cleanup', 'performance', 'podcasts', 'price'):
+				self.assertEqual(module.job([]), (name + '.py', {}))
+
+	def test_cli_prints_usage_and_exits(self):
+		with mock.patch('sys.stderr', new_callable=io.StringIO) as err, self.assertRaises(SystemExit) as exit_:
+			webhook.cli('price.py', price.job, lambda **k: None, ['bogus'])
+		self.assertEqual(exit_.exception.code, 1)
+		self.assertIn('Usage: price.py [midsession', err.getvalue())
+
+	def test_cli_runs_the_handler(self):
+		calls = []
+		webhook.cli('price.py', price.job, lambda **k: calls.append(k), ['midsession'])
+		self.assertEqual(calls, [{'midsession': True}])
 
 
 class CleanupJob(FinbotCase):
-    def test_removes_only_old_finbot_cache_files(self):
-        folder = tempfile.mkdtemp()
-        self.addCleanup(lambda: __import__('shutil').rmtree(folder, ignore_errors=True))
-        self.patch(cleanup, 'config_cache_dir', folder)
-        old, new = time.time() - 40 * 86400, time.time() - 5 * 86400
-        for name, stamp in (('finbot_old.json', old), ('finbot_new.json', new), ('other_old.json', old), ('finbot_old.png', old)):
-            path = os.path.join(folder, name)
-            Path(path).write_text('x')
-            os.utime(path, (stamp, stamp))
-        os.mkdir(os.path.join(folder, 'finbot_dir'))
-        os.utime(os.path.join(folder, 'finbot_dir'), (old, old))
-        self.assertEqual(cleanup.lambda_handler(30), 2)
-        self.assertEqual(sorted(os.listdir(folder)), ['finbot_dir', 'finbot_new.json', 'other_old.json'])
-
-    def test_a_temp_file_being_written_is_left_alone(self):
-        folder = tempfile.mkdtemp()
-        self.addCleanup(lambda: __import__('shutil').rmtree(folder, ignore_errors=True))
-        self.patch(cleanup, 'config_cache_dir', folder)
-        for name, stamp in (('finbot_a.json.x1.tmp', time.time() - 5), ('finbot_b.json.x2.tmp', time.time() - 40 * 86400)):
-            Path(folder, name).write_text('x')
-            os.utime(Path(folder, name), (stamp, stamp))
-        self.assertEqual(cleanup.lambda_handler(0), 1) # a flush: the stale one goes, the write in flight stays
-        self.assertEqual(os.listdir(folder), ['finbot_a.json.x1.tmp'])
-
-    def test_missing_cache_dir_is_fine(self):
-        self.patch(cleanup, 'config_cache_dir', '/nonexistent/finbot-cache')
-        self.assertEqual(cleanup.lambda_handler(30), 0)
+	def test_removes_only_old_finbot_cache_files(self):
+		folder = tempfile.mkdtemp()
+		self.addCleanup(lambda: __import__('shutil').rmtree(folder, ignore_errors=True))
+		self.patch(cleanup, 'config_cache_dir', folder)
+		old, new = time.time() - 40 * 86400, time.time() - 5 * 86400
+		for name, stamp in (('finbot_old.json', old), ('finbot_new.json', new), ('other_old.json', old), ('finbot_old.png', old)):
+			path = os.path.join(folder, name)
+			Path(path).write_text('x')
+			os.utime(path, (stamp, stamp))
+		os.mkdir(os.path.join(folder, 'finbot_dir'))
+		os.utime(os.path.join(folder, 'finbot_dir'), (old, old))
+		self.assertEqual(cleanup.lambda_handler(30), 2)
+		self.assertEqual(sorted(os.listdir(folder)), ['finbot_dir', 'finbot_new.json', 'other_old.json'])
+
+	def test_a_temp_file_being_written_is_left_alone(self):
+		folder = tempfile.mkdtemp()
+		self.addCleanup(lambda: __import__('shutil').rmtree(folder, ignore_errors=True))
+		self.patch(cleanup, 'config_cache_dir', folder)
+		for name, stamp in (('finbot_a.json.x1.tmp', time.time() - 5), ('finbot_b.json.x2.tmp', time.time() - 40 * 86400)):
+			Path(folder, name).write_text('x')
+			os.utime(Path(folder, name), (stamp, stamp))
+		self.assertEqual(cleanup.lambda_handler(0), 1) # a flush: the stale one goes, the write in flight stays
+		self.assertEqual(os.listdir(folder), ['finbot_a.json.x1.tmp'])
+
+	def test_missing_cache_dir_is_fine(self):
+		self.patch(cleanup, 'config_cache_dir', '/nonexistent/finbot-cache')
+		self.assertEqual(cleanup.lambda_handler(30), 0)
 
 
 class SchedulerFlag(unittest.TestCase):
-    def flag(self, **env):
-        code = 'from lib import config; print(config.config_scheduler)'
-        e = {k: v for k, v in os.environ.items() if k != 'scheduler'}
-        e.update(env)
-        repo = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
-        return subprocess.run([sys.executable, '-c', code], cwd=repo, env=e, capture_output=True, text=True, check=True).stdout.split()[-1]
+	def flag(self, **env):
+		code = 'from lib import config; print(config.config_scheduler)'
+		e = {k: v for k, v in os.environ.items() if k != 'scheduler'}
+		e.update(env)
+		repo = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+		return subprocess.run([sys.executable, '-c', code], cwd=repo, env=e, capture_output=True, text=True, check=True).stdout.split()[-1]
 
-    def test_on_by_default_and_off_for_false(self):
-        self.assertEqual(self.flag(), 'True')
-        for value in ('True', 'true', '1', 't'):
-            self.assertEqual(self.flag(scheduler=value), 'True', value)
-        for value in ('False', 'false', '0', 'no'):
-            self.assertEqual(self.flag(scheduler=value), 'False', value)
+	def test_on_by_default_and_off_for_false(self):
+		self.assertEqual(self.flag(), 'True')
+		for value in ('True', 'true', '1', 't'):
+			self.assertEqual(self.flag(scheduler=value), 'True', value)
+		for value in ('False', 'false', '0', 'no'):
+			self.assertEqual(self.flag(scheduler=value), 'False', value)
 
 
 class SchedulerBase(FinbotCase):
-    def setUp(self):
-        super().setUp()
-        tmp = tempfile.TemporaryDirectory()
-        self.addCleanup(tmp.cleanup)
-        self.file = Path(tmp.name, 'schedule.cron')
-        self.file.write_text('*/20 * * * Mon-Fri trades.py\n')
-        self.clock = Clock(at(13, 21, 3))
-        self.ran = []
-        self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', lambda **k: self.ran.append(('trades', k)))
-        self.patch(jobs.SCRIPTS['shorts'], 'lambda_handler', lambda **k: self.ran.append(('shorts', k)))
-
-    def scheduler(self, catchup=5):
-        s = jobs.Scheduler(lambda: self.file, catchup=catchup, clock=self.clock, sleep=lambda seconds: None,
-                           spawn=lambda f, *a: f(*a), zone=UTC, runner=jobs.run_in_process)
-        self.assertTrue(s.load())
-        return s
+	def setUp(self):
+		super().setUp()
+		tmp = tempfile.TemporaryDirectory()
+		self.addCleanup(tmp.cleanup)
+		self.file = Path(tmp.name, 'schedule.cron')
+		self.file.write_text('*/20 * * * Mon-Fri trades.py\n')
+		self.clock = Clock(at(13, 21, 3))
+		self.ran = []
+		self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', lambda **k: self.ran.append(('trades', k)))
+		self.patch(jobs.SCRIPTS['shorts'], 'lambda_handler', lambda **k: self.ran.append(('shorts', k)))
+
+	def scheduler(self, catchup=5):
+		s = jobs.Scheduler(lambda: self.file, catchup=catchup, clock=self.clock, sleep=lambda seconds: None,
+						   spawn=lambda f, *a: f(*a), zone=UTC, runner=jobs.run_in_process)
+		self.assertTrue(s.load())
+		return s
 
 
 class SchedulerTests(SchedulerBase):
-    def test_fires_on_the_minute_on_weekdays_only(self):
-        s = self.scheduler()
-        self.clock.now = at(13, 21, 19, 59)
-        self.assertEqual(s.due(self.clock.now), [])
-        self.clock.now = at(13, 21, 20)
-        self.assertEqual([t for _, t in s.due(self.clock.now)], [at(13, 21, 20)])
-        self.assertEqual(s.due(self.clock.now), []) # not twice
-        self.clock.now = at(13, 21, 40, 30)
-        self.assertEqual([t for _, t in s.due(self.clock.now)], [at(13, 21, 40)])
-        self.clock.now = at(16, 23, 50) # the rest of the week, seen at once: the many missed runs coalesce into one
-        self.assertEqual([t for _, t in s.due(self.clock.now)], [at(16, 23, 40)])
-        self.clock.now = at(17, 21, 20) # a Saturday
-        self.assertEqual(s.due(self.clock.now), [])
-
-    def test_tick_runs_the_job_and_returns_the_wait(self):
-        s = self.scheduler()
-        self.clock.now = at(13, 21, 20)
-        wait = s.tick()
-        self.assertEqual(self.ran, [('trades', {})])
-        self.assertEqual(wait, 30) # the next run is 20 minutes away: wake at most every 30 seconds
-        self.clock.now = at(13, 21, 39, 50)
-        self.assertEqual(s.tick(), 10)
-
-    def test_a_run_missed_within_the_catch_up_window_fires_once(self):
-        first = self.scheduler()
-        first.last = {first.jobs[0].key: at(13, 21, 0).isoformat()}
-        util.json_write(jobs.STATE_FILE, first.last, persist=True)
-        self.clock.now = at(13, 21, 43)
-        s = self.scheduler()
-        self.assertEqual([t for _, t in s.due(self.clock.now)], [at(13, 21, 40)]) # 21:20 is older than 5 minutes: skipped
-
-    def test_a_corrupt_state_file_is_reported_and_the_schedule_starts_afresh(self):
-        Path(util.config_var_dir, jobs.STATE_FILE).write_text('{"*/20 * * * Mon-Fri|UTC|trades.py": "2026-10-13T21:') # cut off mid-write
-        s = self.scheduler()
-        self.assertEqual((s.last, len(self.errors)), ({}, 1))
-        self.assertIn('cannot read finbot_jobs.json', self.errors[0])
-        self.clock.now = at(13, 21, 20)
-        self.assertEqual(len(s.due(self.clock.now)), 1) # runs as usual, and the record is written afresh
-        self.assertEqual(util.json_load(jobs.STATE_FILE, persist=True), {s.jobs[0].key: at(13, 21, 20).isoformat()})
-        s.reload() # and a reload is no trouble either
-        self.assertEqual(len(self.errors), 1)
-
-    def test_no_record_means_nothing_is_missed(self):
-        self.clock.now = at(13, 21, 43)
-        s = self.scheduler()
-        self.assertEqual(s.due(self.clock.now), [])
-
-    def test_a_restart_just_after_a_run_does_not_repeat_it(self):
-        s = self.scheduler()
-        self.clock.now = at(13, 21, 20, 5)
-        self.assertEqual(len(s.due(self.clock.now)), 1)
-        self.clock.now = at(13, 21, 21)
-        again = self.scheduler()
-        self.assertEqual(again.due(self.clock.now), [])
-
-    def test_an_overlapping_run_is_skipped(self):
-        s = self.scheduler()
-        s.jobs[0].running = True
-        s.run_job(s.jobs[0], at(13, 21, 20))
-        self.assertEqual(self.ran, [])
-        s.jobs[0].running = False
-        s.run_job(s.jobs[0], at(13, 21, 20))
-        self.assertEqual(self.ran, [('trades', {})])
-        self.assertFalse(s.jobs[0].running)
-
-    def test_a_failing_job_is_reported_and_the_next_command_still_runs(self):
-        self.file.write_text('*/20 * * * * trades.py; shorts.py\n')
-        def boom(**k):
-            raise RuntimeError('boom')
-        self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', boom)
-        s = self.scheduler()
-        s.run_job(s.jobs[0], at(13, 21, 20))
-        self.assertEqual(self.ran, [('shorts', {})])
-        self.assertTrue(any('boom' in e for e in self.errors))
-
-    def test_system_exit_inside_a_job_is_reported_not_raised(self):
-        def quit_(**k):
-            raise SystemExit(3)
-        self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', quit_)
-        s = self.scheduler()
-        s.run_job(s.jobs[0], at(13, 21, 20))
-        self.assertEqual(len(self.errors), 1)
-        self.assertFalse(s.jobs[0].running)
-
-    def test_a_job_that_hangs_is_abandoned_and_reported(self):
-        self.patch(jobs, 'JOB_TIMEOUT', 0.05)
-        self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', lambda **k: gevent.sleep(5))
-        s = self.scheduler()
-        s.run_job(s.jobs[0], at(13, 21, 20))
-        self.assertTrue(any('still running' in e for e in self.errors))
-        self.assertFalse(s.jobs[0].running)
-
-    def test_reload_applies_edits_keeps_state_and_survives_a_bad_file(self):
-        s = self.scheduler()
-        old_key = s.jobs[0].key
-        base = s.base[old_key]
-        self.file.write_text('*/20 * * * Mon-Fri trades.py\n5 4 * * * shorts.py\n')
-        self.assertTrue(s.load())
-        self.assertEqual(len(s.jobs), 2)
-        self.assertEqual(s.base[old_key], base)
-        self.file.write_text('# nothing usable\nnot a job\n')
-        self.assertFalse(s.load())
-        self.assertEqual(len(s.jobs), 2)
-
-    def test_an_impossible_time_refuses_the_whole_file(self):
-        s = self.scheduler()
-        self.file.write_text('*/20 * * * Mon-Fri trades.py\n0 9 30 2 * shorts.py\n5 4 * * * nothere.py\n')
-        with self.assertRaises(jobs.ImpossibleTime) as caught:
-            s.load()
-        self.assertEqual(len(caught.exception.lines), 1)
-        self.assertIn("line 2: '0 9 30 2 *' never comes", caught.exception.lines[0])
-        self.assertEqual(self.errors, []) # load reports nothing for a refused file...
-        s.reload() # ...a reload says why it was refused, and the running schedule carries on
-        self.assertEqual(len(self.errors), 1)
-        self.assertIn('not reloaded, the running schedule is unchanged', self.errors[0])
-        self.assertEqual([j.text for j in s.jobs], ['*/20 * * * Mon-Fri trades.py'])
-        self.clock.now = at(13, 21, 20)
-        self.assertEqual(len(s.due(self.clock.now)), 1)
-
-    def test_load_reports_bad_lines_once_and_runs_the_rest(self):
-        self.file.write_text('0 1 * * * nothere.py\n*/20 * * * * trades.py\n')
-        s = self.scheduler()
-        self.assertEqual(len(s.jobs), 1)
-        self.assertEqual(len(self.errors), 1)
-
-    def test_run_forever_survives_an_error(self):
-        s = self.scheduler()
-        calls = []
-        def sleep(seconds):
-            calls.append(seconds)
-            if len(calls) == 3:
-                raise KeyboardInterrupt
-        s.sleep = sleep
-        with mock.patch.object(s, 'tick', side_effect=[RuntimeError('tick failed'), 30, 30]):
-            with self.assertRaises(KeyboardInterrupt):
-                s.run_forever()
-        self.assertEqual(calls, [30, 30, 30])
-        self.assertTrue(any('tick failed' in e for e in self.errors))
+	def test_fires_on_the_minute_on_weekdays_only(self):
+		s = self.scheduler()
+		self.clock.now = at(13, 21, 19, 59)
+		self.assertEqual(s.due(self.clock.now), [])
+		self.clock.now = at(13, 21, 20)
+		self.assertEqual([t for _, t in s.due(self.clock.now)], [at(13, 21, 20)])
+		self.assertEqual(s.due(self.clock.now), []) # not twice
+		self.clock.now = at(13, 21, 40, 30)
+		self.assertEqual([t for _, t in s.due(self.clock.now)], [at(13, 21, 40)])
+		self.clock.now = at(16, 23, 50) # the rest of the week, seen at once: the many missed runs coalesce into one
+		self.assertEqual([t for _, t in s.due(self.clock.now)], [at(16, 23, 40)])
+		self.clock.now = at(17, 21, 20) # a Saturday
+		self.assertEqual(s.due(self.clock.now), [])
+
+	def test_tick_runs_the_job_and_returns_the_wait(self):
+		s = self.scheduler()
+		self.clock.now = at(13, 21, 20)
+		wait = s.tick()
+		self.assertEqual(self.ran, [('trades', {})])
+		self.assertEqual(wait, 30) # the next run is 20 minutes away: wake at most every 30 seconds
+		self.clock.now = at(13, 21, 39, 50)
+		self.assertEqual(s.tick(), 10)
+
+	def test_a_run_missed_within_the_catch_up_window_fires_once(self):
+		first = self.scheduler()
+		first.last = {first.jobs[0].key: at(13, 21, 0).isoformat()}
+		util.json_write(jobs.STATE_FILE, first.last, persist=True)
+		self.clock.now = at(13, 21, 43)
+		s = self.scheduler()
+		self.assertEqual([t for _, t in s.due(self.clock.now)], [at(13, 21, 40)]) # 21:20 is older than 5 minutes: skipped
+
+	def test_a_corrupt_state_file_is_reported_and_the_schedule_starts_afresh(self):
+		Path(util.config_var_dir, jobs.STATE_FILE).write_text('{"*/20 * * * Mon-Fri|UTC|trades.py": "2026-10-13T21:') # cut off mid-write
+		s = self.scheduler()
+		self.assertEqual((s.last, len(self.errors)), ({}, 1))
+		self.assertIn('cannot read finbot_jobs.json', self.errors[0])
+		self.clock.now = at(13, 21, 20)
+		self.assertEqual(len(s.due(self.clock.now)), 1) # runs as usual, and the record is written afresh
+		self.assertEqual(util.json_load(jobs.STATE_FILE, persist=True), {s.jobs[0].key: at(13, 21, 20).isoformat()})
+		s.reload() # and a reload is no trouble either
+		self.assertEqual(len(self.errors), 1)
+
+	def test_no_record_means_nothing_is_missed(self):
+		self.clock.now = at(13, 21, 43)
+		s = self.scheduler()
+		self.assertEqual(s.due(self.clock.now), [])
+
+	def test_a_restart_just_after_a_run_does_not_repeat_it(self):
+		s = self.scheduler()
+		self.clock.now = at(13, 21, 20, 5)
+		self.assertEqual(len(s.due(self.clock.now)), 1)
+		self.clock.now = at(13, 21, 21)
+		again = self.scheduler()
+		self.assertEqual(again.due(self.clock.now), [])
+
+	def test_an_overlapping_run_is_skipped(self):
+		s = self.scheduler()
+		s.jobs[0].running = True
+		s.run_job(s.jobs[0], at(13, 21, 20))
+		self.assertEqual(self.ran, [])
+		s.jobs[0].running = False
+		s.run_job(s.jobs[0], at(13, 21, 20))
+		self.assertEqual(self.ran, [('trades', {})])
+		self.assertFalse(s.jobs[0].running)
+
+	def test_a_failing_job_is_reported_and_the_next_command_still_runs(self):
+		self.file.write_text('*/20 * * * * trades.py; shorts.py\n')
+		def boom(**k):
+			raise RuntimeError('boom')
+		self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', boom)
+		s = self.scheduler()
+		s.run_job(s.jobs[0], at(13, 21, 20))
+		self.assertEqual(self.ran, [('shorts', {})])
+		self.assertTrue(any('boom' in e for e in self.errors))
+
+	def test_system_exit_inside_a_job_is_reported_not_raised(self):
+		def quit_(**k):
+			raise SystemExit(3)
+		self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', quit_)
+		s = self.scheduler()
+		s.run_job(s.jobs[0], at(13, 21, 20))
+		self.assertEqual(len(self.errors), 1)
+		self.assertFalse(s.jobs[0].running)
+
+	def test_a_job_that_hangs_is_abandoned_and_reported(self):
+		self.patch(jobs, 'JOB_TIMEOUT', 0.05)
+		self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', lambda **k: gevent.sleep(5))
+		s = self.scheduler()
+		s.run_job(s.jobs[0], at(13, 21, 20))
+		self.assertTrue(any('still running' in e for e in self.errors))
+		self.assertFalse(s.jobs[0].running)
+
+	def test_reload_applies_edits_keeps_state_and_survives_a_bad_file(self):
+		s = self.scheduler()
+		old_key = s.jobs[0].key
+		base = s.base[old_key]
+		self.file.write_text('*/20 * * * Mon-Fri trades.py\n5 4 * * * shorts.py\n')
+		self.assertTrue(s.load())
+		self.assertEqual(len(s.jobs), 2)
+		self.assertEqual(s.base[old_key], base)
+		self.file.write_text('# nothing usable\nnot a job\n')
+		self.assertFalse(s.load())
+		self.assertEqual(len(s.jobs), 2)
+
+	def test_an_impossible_time_refuses_the_whole_file(self):
+		s = self.scheduler()
+		self.file.write_text('*/20 * * * Mon-Fri trades.py\n0 9 30 2 * shorts.py\n5 4 * * * nothere.py\n')
+		with self.assertRaises(jobs.ImpossibleTime) as caught:
+			s.load()
+		self.assertEqual(len(caught.exception.lines), 1)
+		self.assertIn("line 2: '0 9 30 2 *' never comes", caught.exception.lines[0])
+		self.assertEqual(self.errors, []) # load reports nothing for a refused file...
+		s.reload() # ...a reload says why it was refused, and the running schedule carries on
+		self.assertEqual(len(self.errors), 1)
+		self.assertIn('not reloaded, the running schedule is unchanged', self.errors[0])
+		self.assertEqual([j.text for j in s.jobs], ['*/20 * * * Mon-Fri trades.py'])
+		self.clock.now = at(13, 21, 20)
+		self.assertEqual(len(s.due(self.clock.now)), 1)
+
+	def test_load_reports_bad_lines_once_and_runs_the_rest(self):
+		self.file.write_text('0 1 * * * nothere.py\n*/20 * * * * trades.py\n')
+		s = self.scheduler()
+		self.assertEqual(len(s.jobs), 1)
+		self.assertEqual(len(self.errors), 1)
+
+	def test_run_forever_survives_an_error(self):
+		s = self.scheduler()
+		calls = []
+		def sleep(seconds):
+			calls.append(seconds)
+			if len(calls) == 3:
+				raise KeyboardInterrupt
+		s.sleep = sleep
+		with mock.patch.object(s, 'tick', side_effect=[RuntimeError('tick failed'), 30, 30]):
+			with self.assertRaises(KeyboardInterrupt):
+				s.run_forever()
+		self.assertEqual(calls, [30, 30, 30])
+		self.assertTrue(any('tick failed' in e for e in self.errors))
 
 
 class DaylightSaving(unittest.TestCase):
-    """Melbourne's clocks go back an hour at 03:00 on 5 April 2026 and forward at 02:00 on 4 October 2026. The fire times must be
-    right on any croniter: the distro packages (up to 6.0) got zoneinfo's changes wrong."""
-    MEL = zoneinfo.ZoneInfo('Australia/Melbourne')
-
-    def fires(self, expr, start, end):
-        job = jobs.parse_schedule(f'{expr} trades.py\n', self.MEL)[0][0]
-        s = jobs.Scheduler(lambda: None, zone=self.MEL)
-        times, t = [], start
-        while (t := s.next_fire(job, t)) <= end:
-            times.append(t.astimezone(UTC).astimezone(self.MEL).strftime('%d %H:%M %Z')) # by way of UTC: the real local time
-        return times
-
-    def mel(self, *args, **kw):
-        return datetime.datetime(*args, tzinfo=self.MEL, **kw)
-
-    def test_a_daily_job_fires_once_at_its_time(self):
-        self.assertEqual(self.fires('0 9 * * *', self.mel(2026, 10, 2), self.mel(2026, 10, 6)), ['02 09:00 AEST', '03 09:00 AEST', '04 09:00 AEDT', '05 09:00 AEDT'])
-        self.assertEqual(self.fires('0 9 * * *', self.mel(2026, 4, 3), self.mel(2026, 4, 7)), ['03 09:00 AEDT', '04 09:00 AEDT', '05 09:00 AEST', '06 09:00 AEST'])
-
-    def test_a_skipped_or_repeated_time_fires_once(self):
-        self.assertEqual(self.fires('30 2 * * *', self.mel(2026, 10, 3), self.mel(2026, 10, 6)), ['03 02:30 AEST', '04 03:30 AEDT', '05 02:30 AEDT'])
-        self.assertEqual(self.fires('30 2 * * *', self.mel(2026, 4, 4), self.mel(2026, 4, 7)), ['04 02:30 AEDT', '05 02:30 AEDT', '06 02:30 AEST'])
-        self.assertEqual(self.fires('*/20 * * * *', self.mel(2026, 10, 4, 1, 30), self.mel(2026, 10, 4, 3, 50)), ['04 01:40 AEST', '04 03:00 AEDT', '04 03:20 AEDT', '04 03:40 AEDT'])
-        self.assertEqual(self.fires('*/20 * * * *', self.mel(2026, 4, 5, 1, 30), self.mel(2026, 4, 5, 3, 30)),
-                         ['05 01:40 AEDT', '05 02:00 AEDT', '05 02:20 AEDT', '05 02:40 AEDT', '05 03:00 AEST', '05 03:20 AEST'])
-
-    def test_starting_in_the_repeated_hour_runs_nothing_twice(self):
-        second_time_round = self.mel(2026, 4, 5, 2, 10, fold=1).astimezone(UTC)
-        self.assertEqual(self.fires('*/20 * * * *', second_time_round, self.mel(2026, 4, 5, 3, 0, 30)), ['05 03:00 AEST'])
-        self.assertEqual(self.fires('30 2 * * *', second_time_round, self.mel(2026, 4, 6, 12)), ['06 02:30 AEST'])
-
-    def test_the_scheduler_runs_a_daily_job_once_across_the_change(self):
-        file = Path(tempfile.mkdtemp(), 'schedule.cron')
-        file.write_text('CRON_TZ=Australia/Melbourne\n0 9 * * * trades.py\n')
-        clock = Clock(self.mel(2026, 10, 3, 12).astimezone(UTC))
-        s = jobs.Scheduler(lambda: file, clock=clock, zone=UTC, runner=lambda *a: 0)
-        with mock.patch.object(util, 'config_var_dir', tempfile.mkdtemp()), contextlib.redirect_stderr(io.StringIO()):
-            self.assertTrue(s.load())
-            fired = []
-            while clock.now < self.mel(2026, 10, 5, 12):
-                fired += [t.astimezone(UTC).astimezone(self.MEL).strftime('%d %H:%M') for _, t in s.due(clock.now)]
-                clock.now += datetime.timedelta(minutes=10)
-        self.assertEqual(fired, ['04 09:00', '05 09:00'])
+	"""Melbourne's clocks go back an hour at 03:00 on 5 April 2026 and forward at 02:00 on 4 October 2026. The fire times must be
+	right on any croniter: the distro packages (up to 6.0) got zoneinfo's changes wrong."""
+	MEL = zoneinfo.ZoneInfo('Australia/Melbourne')
+
+	def fires(self, expr, start, end):
+		job = jobs.parse_schedule(f'{expr} trades.py\n', self.MEL)[0][0]
+		s = jobs.Scheduler(lambda: None, zone=self.MEL)
+		times, t = [], start
+		while (t := s.next_fire(job, t)) <= end:
+			times.append(t.astimezone(UTC).astimezone(self.MEL).strftime('%d %H:%M %Z')) # by way of UTC: the real local time
+		return times
+
+	def mel(self, *args, **kw):
+		return datetime.datetime(*args, tzinfo=self.MEL, **kw)
+
+	def test_a_daily_job_fires_once_at_its_time(self):
+		self.assertEqual(self.fires('0 9 * * *', self.mel(2026, 10, 2), self.mel(2026, 10, 6)), ['02 09:00 AEST', '03 09:00 AEST', '04 09:00 AEDT', '05 09:00 AEDT'])
+		self.assertEqual(self.fires('0 9 * * *', self.mel(2026, 4, 3), self.mel(2026, 4, 7)), ['03 09:00 AEDT', '04 09:00 AEDT', '05 09:00 AEST', '06 09:00 AEST'])
+
+	def test_a_skipped_or_repeated_time_fires_once(self):
+		self.assertEqual(self.fires('30 2 * * *', self.mel(2026, 10, 3), self.mel(2026, 10, 6)), ['03 02:30 AEST', '04 03:30 AEDT', '05 02:30 AEDT'])
+		self.assertEqual(self.fires('30 2 * * *', self.mel(2026, 4, 4), self.mel(2026, 4, 7)), ['04 02:30 AEDT', '05 02:30 AEDT', '06 02:30 AEST'])
+		self.assertEqual(self.fires('*/20 * * * *', self.mel(2026, 10, 4, 1, 30), self.mel(2026, 10, 4, 3, 50)), ['04 01:40 AEST', '04 03:00 AEDT', '04 03:20 AEDT', '04 03:40 AEDT'])
+		self.assertEqual(self.fires('*/20 * * * *', self.mel(2026, 4, 5, 1, 30), self.mel(2026, 4, 5, 3, 30)),
+						 ['05 01:40 AEDT', '05 02:00 AEDT', '05 02:20 AEDT', '05 02:40 AEDT', '05 03:00 AEST', '05 03:20 AEST'])
+
+	def test_starting_in_the_repeated_hour_runs_nothing_twice(self):
+		second_time_round = self.mel(2026, 4, 5, 2, 10, fold=1).astimezone(UTC)
+		self.assertEqual(self.fires('*/20 * * * *', second_time_round, self.mel(2026, 4, 5, 3, 0, 30)), ['05 03:00 AEST'])
+		self.assertEqual(self.fires('30 2 * * *', second_time_round, self.mel(2026, 4, 6, 12)), ['06 02:30 AEST'])
+
+	def test_the_scheduler_runs_a_daily_job_once_across_the_change(self):
+		file = Path(tempfile.mkdtemp(), 'schedule.cron')
+		file.write_text('CRON_TZ=Australia/Melbourne\n0 9 * * * trades.py\n')
+		clock = Clock(self.mel(2026, 10, 3, 12).astimezone(UTC))
+		s = jobs.Scheduler(lambda: file, clock=clock, zone=UTC, runner=lambda *a: 0)
+		with mock.patch.object(util, 'config_var_dir', tempfile.mkdtemp()), contextlib.redirect_stderr(io.StringIO()):
+			self.assertTrue(s.load())
+			fired = []
+			while clock.now < self.mel(2026, 10, 5, 12):
+				fired += [t.astimezone(UTC).astimezone(self.MEL).strftime('%d %H:%M') for _, t in s.due(clock.now)]
+				clock.now += datetime.timedelta(minutes=10)
+		self.assertEqual(fired, ['04 09:00', '05 09:00'])
 
 
 class StartupRefusal(unittest.TestCase):
-    def test_the_bot_will_not_start_with_an_impossible_time(self):
-        skip_if_old_env(self)
-        with tempfile.TemporaryDirectory() as data:
-            Path(data, 'schedule.cron').write_text('*/20 * * * Mon-Fri trades.py\n0 9 31 6 * price.py close\n')
-            env = {k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k not in config.STARTUP_KEYS}
-            env.update(DATA_DIR=data, port='0')
-            out = subprocess.run([sys.executable, 'bot.py'], cwd=ROOT, env=env, capture_output=True, text=True, timeout=120)
-        self.assertEqual(out.returncode, 1, out.stderr)
-        self.assertIn(f"ERROR schedule: {data}/schedule.cron line 2: '0 9 31 6 *' never comes", out.stderr)
+	def test_the_bot_will_not_start_with_an_impossible_time(self):
+		skip_if_old_env(self)
+		with tempfile.TemporaryDirectory() as data:
+			Path(data, 'schedule.cron').write_text('*/20 * * * Mon-Fri trades.py\n0 9 31 6 * price.py close\n')
+			env = {k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k not in config.STARTUP_KEYS}
+			env.update(DATA_DIR=data, port='0')
+			out = subprocess.run([sys.executable, 'bot.py'], cwd=ROOT, env=env, capture_output=True, text=True, timeout=120)
+		self.assertEqual(out.returncode, 1, out.stderr)
+		self.assertIn(f"ERROR schedule: {data}/schedule.cron line 2: '0 9 31 6 *' never comes", out.stderr)
 
 
 class ManualRuns(SchedulerBase):
-    def setUp(self):
-        super().setUp()
-        self.file.write_text('# a comment\n*/20 * * * Mon-Fri trades.py > /dev/null\n28 21 * * Sat price.py close; trades.py # weekly\n'
-                             '5 4 * * * { price.py close ;}\n6 4 * * * price.py 7\n'
-                             '0 3 * * * performance.py 365; performance.py 7; performance.py 90; performance.py 30; performance.py 180\n'
-                             '0 2 * * * cleanup.py\n1 * * * * podcasts.py "https://example.com/feed"\n')
-        self.s = self.scheduler()
-        self.patch(jobs, 'current', self.s)
-        self.patch(jobs.SCRIPTS['price'], 'lambda_handler', lambda **k: self.ran.append(('price', k)))
-
-    def test_commands_are_unique_clean_and_in_order(self):
-        self.assertEqual([label for label, _, _ in self.s.commands()], ['performance.py 7', 'performance.py 30', 'performance.py 90', 'performance.py 180',
-                                                                       'performance.py 365', 'price.py 7', 'price.py close', 'trades.py'])
-
-    def test_select_by_number_or_command_text(self):
-        self.assertEqual(self.s.select('7')[0], 'price.py close')
-        self.assertEqual(self.s.select('8')[0], 'trades.py')
-        self.assertEqual(self.s.select('  "price.py" close > /dev/null')[0], 'price.py close')
-        self.assertEqual(self.s.select('~/finbot/price.py 7')[0], 'price.py 7')
-
-    def test_select_refuses_what_is_not_offered(self):
-        for text in ('0', '9', 'cleanup.py', 'podcasts.py "https://example.com/feed"', 'price.py 99', 'cal.py earnings', 'nonsense', 'ls | wc', '', 'price.py close; trades.py'):
-            with self.assertRaises(jobs.ScheduleError, msg=text):
-                self.s.select(text)
-
-    def test_run_now_runs_one_command_and_leaves_the_schedule_alone(self):
-        base, last = dict(self.s.base), dict(self.s.last)
-        self.assertEqual(self.s.run_now('7'), ('price.py close', 'done'))
-        self.assertEqual(self.ran, [('price', {'close': True})])
-        self.assertEqual((self.s.base, self.s.last), (base, last))
-        self.assertTrue(all(not job.running for job in self.s.jobs))
-        self.assertEqual(self.s.busy, set())
-
-    def test_run_now_refuses_while_the_command_is_running_and_reports_failures(self):
-        self.s.busy.add('trades.py')
-        self.assertEqual(self.s.run_now('8'), ('trades.py', 'already running'))
-        self.assertEqual(self.ran, [])
-        self.s.busy.clear()
-        def boom(**k):
-            raise RuntimeError('boom')
-        self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', boom)
-        self.assertEqual(self.s.run_now('8'), ('trades.py', 'failed'))
-        self.assertTrue(any('boom' in e for e in self.errors))
-        self.assertEqual(self.s.busy, set())
-
-    def test_a_scheduled_run_skips_a_command_that_a_manual_run_is_in_the_middle_of(self):
-        self.s.busy.add('trades.py')
-        self.assertEqual(self.s.run_job(self.s.jobs[1], at(13, 21, 28)), 'already running')
-        self.assertEqual(self.ran, [('price', {'close': True})]) # price.py close ran, trades.py was skipped
-
-    def test_run_now_times_out(self):
-        self.patch(jobs, 'JOB_TIMEOUT', 0.05)
-        self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', lambda **k: gevent.sleep(5))
-        self.assertEqual(self.s.run_now('8'), ('trades.py', 'timed out'))
-        self.assertEqual(self.s.busy, set())
-
-    def list_in_chat(self, text='.jobs'):
-        return self.command(text)[0][1]
-
-    def test_list_in_chat_is_just_the_commands(self):
-        for alias in ('.jobs', '.cron', '.crons', '.schedule', '@' + BOT.lstrip('@') + ' jobs'):
-            text = '\n'.join(map(str, self.list_in_chat(alias)))
-            for line in ('1. performance 7', '2. performance 30', '5. performance 365', '6. price 7', '7. price close', '8. trades'):
-                self.assertIn(line, text, alias)
-            for noise in ('*/20', '#', 'Mon-Fri', 'next', 'last', '/dev/null', 'weekly', '.py', 'cleanup', 'podcasts'):
-                self.assertNotIn(noise, text, alias)
-        self.s.busy.add('price.py 7')
-        self.assertIn('6. price 7 (running)', '\n'.join(map(str, self.list_in_chat())))
-
-    def test_telegram_gets_a_button_per_command_and_slack_does_not(self):
-        payload, markup = reports.prepare_jobs_payload('telegram', '')
-        buttons = [b for row in markup['inline_keyboard'] for b in row]
-        self.assertEqual([b['text'] for b in buttons], ['1. performance 7', '2. performance 30', '3. performance 90', '4. performance 180',
-                                                      '5. performance 365', '6. price 7', '7. price close', '8. trades'])
-        for b in buttons:
-            self.assertLessEqual(len(b['callback_data'].encode()), 64)
-        payload, markup = reports.prepare_jobs_payload('slack', '')
-        self.assertIsNone(markup)
-
-    def test_the_chat_command_passes_the_keyboard_on(self):
-        seen = []
-        self.patch(webhook, 'payload_wrapper', lambda svc, url, payload, *a, **k: seen.append(k.get('reply_markup')) or [])
-        worker.process_request('telegram', '55', '@u', '.jobs', BOT, 'U')
-        self.assertEqual(len(seen[0]['inline_keyboard'][0]), 2)
-
-    def test_button_press_runs_that_command(self):
-        sent = []
-        self.patch(telegram, 'answerCallbackQuery', lambda cid, text=None: sent.append(text))
-        markup = reports.prepare_jobs_payload('telegram', '')[1]
-        data = next(b['callback_data'] for row in markup['inline_keyboard'] for b in row if b['text'].startswith('7.'))
-        self.sent.clear()
-        worker.process_callback('telegram', 'cb1', '55', '9', data)
-        self.assertEqual(self.ran, [('price', {'close': True})])
-        self.assertEqual(self.sent, [('text', ['7. price close: done'])])
-        self.assertEqual(sent, ['Running price close…'])
-
-    def stale_buttons(self):
-        """Capture the toast and the in-place edit that a refused button press produces."""
-        self.toasts, self.edits = [], []
-        self.patch(telegram, 'answerCallbackQuery', lambda cid, text=None: self.toasts.append(text))
-        self.patch(webhook, 'editMessageText', lambda chat, mid, text, markup=None: self.edits.append((chat, mid, text, markup)))
-
-    def test_forged_button_data_never_runs_anything_else(self):
-        self.stale_buttons()
-        good = jobs.checksum('trades.py')
-        for data in ('j|-1|' + good, 'j|0|' + good, 'j|9|' + good, 'j|8|0000', 'j|8|', 'j|8', 'j|8|' + good + '|x', 'j||' + good,
-                     'j|٨|' + good, 'j| 8|' + good, 'j|8.0|' + good, 'j|trades.py|' + good, 'j|1|' + good, 'j|8|price.py close',
-                     'j|8|' + good.upper() + 'x', 'j|99999999999999999999|' + good):
-            self.sent.clear()
-            self.toasts.clear(); self.edits.clear()
-            worker.process_callback('telegram', 'cb', '55', '9', data)
-            self.assertEqual(self.ran, [], data)
-            self.assertEqual(self.toasts, ['Buttons updated, try again'], data)
-            self.assertEqual(len(self.edits), 1, data)
-            self.assertEqual(self.sent, [], data) # no new message, the old one is edited
-        # cleanup and podcasts are scheduled but cannot be picked, by number or by their checksum
-        for label in ('cleanup.py', 'podcasts.py'):
-            for n in range(1, 12):
-                with self.assertRaises(jobs.ScheduleError):
-                    self.s.pick(str(n), jobs.checksum(label))
-        worker.process_callback('telegram', 'cb', '55', '9', 'j|8|' + good) # the genuine button still works
-        self.assertEqual(self.ran, [('trades', {})])
-
-    def test_button_from_an_old_schedule_updates_the_message(self):
-        self.stale_buttons()
-        worker.process_callback('telegram', 'cb1', '55', '9', 'j|7|0000')
-        worker.process_callback('telegram', 'cb1', '55', '9', 'j|9|0000')
-        self.assertEqual(self.ran, [])
-        self.assertEqual(self.sent, [])
-        self.assertEqual(self.toasts, ['Buttons updated, try again'] * 2)
-        chat, mid, text, markup = self.edits[0]
-        self.assertEqual((chat, mid), ('55', '9')) # the message with the old buttons is edited in place
-        self.assertIn('7. price close', text)
-        self.assertEqual(markup, reports.prepare_jobs_payload('telegram', '')[1]) # with the current buttons
-        self.patch(jobs, 'current', None) # scheduler off: the list is replaced by the notice and the buttons removed
-        worker.process_callback('telegram', 'cb1', '55', '9', 'j|7|0000')
-        self.assertEqual(self.edits[-1][2:], ('Scheduled jobs are off (scheduler=False)', {'inline_keyboard': []}))
-
-    def test_run_in_chat(self):
-        self.assertEqual(self.command('.jobs run 8'), [('text', ['8. trades: done'])])
-        self.assertEqual(self.ran, [('trades', {})])
-        self.ran.clear()
-        self.assertEqual(self.command('.cron run price close')[0][1], ['7. price close: done'])
-        self.assertEqual(self.ran, [('price', {'close': True})])
-
-    def test_run_in_chat_refusals(self):
-        self.assertIn('not in the schedule', self.list_in_chat('.jobs run cal.py earnings')[0])
-        self.assertIn('no command 99', self.list_in_chat('.jobs run 99')[0])
-        self.assertIn('one command at a time', self.list_in_chat('.jobs run price.py close; trades.py')[0])
-        self.assertIn('Usage', self.list_in_chat('.jobs run')[0])
-        self.assertIn('Usage', self.list_in_chat('.jobs bogus')[0])
-        self.assertEqual(self.ran, [])
-
-    def test_scheduler_off(self):
-        self.patch(jobs, 'current', None)
-        self.assertIn('off', self.list_in_chat('.jobs')[0])
-        self.assertIn('off', self.list_in_chat('.jobs run 8')[0])
-
-    def test_job_is_still_a_ticker(self):
-        self.assertNotIn('Scheduled', str(self.command('.job')))
+	def setUp(self):
+		super().setUp()
+		self.file.write_text('# a comment\n*/20 * * * Mon-Fri trades.py > /dev/null\n28 21 * * Sat price.py close; trades.py # weekly\n'
+							 '5 4 * * * { price.py close ;}\n6 4 * * * price.py 7\n'
+							 '0 3 * * * performance.py 365; performance.py 7; performance.py 90; performance.py 30; performance.py 180\n'
+							 '0 2 * * * cleanup.py\n1 * * * * podcasts.py "https://example.com/feed"\n')
+		self.s = self.scheduler()
+		self.patch(jobs, 'current', self.s)
+		self.patch(jobs.SCRIPTS['price'], 'lambda_handler', lambda **k: self.ran.append(('price', k)))
+
+	def test_commands_are_unique_clean_and_in_order(self):
+		self.assertEqual([label for label, _, _ in self.s.commands()], ['performance.py 7', 'performance.py 30', 'performance.py 90', 'performance.py 180',
+																	   'performance.py 365', 'price.py 7', 'price.py close', 'trades.py'])
+
+	def test_select_by_number_or_command_text(self):
+		self.assertEqual(self.s.select('7')[0], 'price.py close')
+		self.assertEqual(self.s.select('8')[0], 'trades.py')
+		self.assertEqual(self.s.select('  "price.py" close > /dev/null')[0], 'price.py close')
+		self.assertEqual(self.s.select('~/finbot/price.py 7')[0], 'price.py 7')
+
+	def test_select_refuses_what_is_not_offered(self):
+		for text in ('0', '9', 'cleanup.py', 'podcasts.py "https://example.com/feed"', 'price.py 99', 'cal.py earnings', 'nonsense', 'ls | wc', '', 'price.py close; trades.py'):
+			with self.assertRaises(jobs.ScheduleError, msg=text):
+				self.s.select(text)
+
+	def test_run_now_runs_one_command_and_leaves_the_schedule_alone(self):
+		base, last = dict(self.s.base), dict(self.s.last)
+		self.assertEqual(self.s.run_now('7'), ('price.py close', 'done'))
+		self.assertEqual(self.ran, [('price', {'close': True})])
+		self.assertEqual((self.s.base, self.s.last), (base, last))
+		self.assertTrue(all(not job.running for job in self.s.jobs))
+		self.assertEqual(self.s.busy, set())
+
+	def test_run_now_refuses_while_the_command_is_running_and_reports_failures(self):
+		self.s.busy.add('trades.py')
+		self.assertEqual(self.s.run_now('8'), ('trades.py', 'already running'))
+		self.assertEqual(self.ran, [])
+		self.s.busy.clear()
+		def boom(**k):
+			raise RuntimeError('boom')
+		self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', boom)
+		self.assertEqual(self.s.run_now('8'), ('trades.py', 'failed'))
+		self.assertTrue(any('boom' in e for e in self.errors))
+		self.assertEqual(self.s.busy, set())
+
+	def test_a_scheduled_run_skips_a_command_that_a_manual_run_is_in_the_middle_of(self):
+		self.s.busy.add('trades.py')
+		self.assertEqual(self.s.run_job(self.s.jobs[1], at(13, 21, 28)), 'already running')
+		self.assertEqual(self.ran, [('price', {'close': True})]) # price.py close ran, trades.py was skipped
+
+	def test_run_now_times_out(self):
+		self.patch(jobs, 'JOB_TIMEOUT', 0.05)
+		self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', lambda **k: gevent.sleep(5))
+		self.assertEqual(self.s.run_now('8'), ('trades.py', 'timed out'))
+		self.assertEqual(self.s.busy, set())
+
+	def list_in_chat(self, text='.jobs'):
+		return self.command(text)[0][1]
+
+	def test_list_in_chat_is_just_the_commands(self):
+		for alias in ('.jobs', '.cron', '.crons', '.schedule', '@' + BOT.lstrip('@') + ' jobs'):
+			text = '\n'.join(map(str, self.list_in_chat(alias)))
+			for line in ('1. performance 7', '2. performance 30', '5. performance 365', '6. price 7', '7. price close', '8. trades'):
+				self.assertIn(line, text, alias)
+			for noise in ('*/20', '#', 'Mon-Fri', 'next', 'last', '/dev/null', 'weekly', '.py', 'cleanup', 'podcasts'):
+				self.assertNotIn(noise, text, alias)
+		self.s.busy.add('price.py 7')
+		self.assertIn('6. price 7 (running)', '\n'.join(map(str, self.list_in_chat())))
+
+	def test_telegram_gets_a_button_per_command_and_slack_does_not(self):
+		payload, markup = reports.prepare_jobs_payload('telegram', '')
+		buttons = [b for row in markup['inline_keyboard'] for b in row]
+		self.assertEqual([b['text'] for b in buttons], ['1. performance 7', '2. performance 30', '3. performance 90', '4. performance 180',
+													  '5. performance 365', '6. price 7', '7. price close', '8. trades'])
+		for b in buttons:
+			self.assertLessEqual(len(b['callback_data'].encode()), 64)
+		payload, markup = reports.prepare_jobs_payload('slack', '')
+		self.assertIsNone(markup)
+
+	def test_the_chat_command_passes_the_keyboard_on(self):
+		seen = []
+		self.patch(webhook, 'payload_wrapper', lambda svc, url, payload, *a, **k: seen.append(k.get('reply_markup')) or [])
+		worker.process_request('telegram', '55', '@u', '.jobs', BOT, 'U')
+		self.assertEqual(len(seen[0]['inline_keyboard'][0]), 2)
+
+	def test_button_press_runs_that_command(self):
+		sent = []
+		self.patch(telegram, 'answerCallbackQuery', lambda cid, text=None: sent.append(text))
+		markup = reports.prepare_jobs_payload('telegram', '')[1]
+		data = next(b['callback_data'] for row in markup['inline_keyboard'] for b in row if b['text'].startswith('7.'))
+		self.sent.clear()
+		worker.process_callback('telegram', 'cb1', '55', '9', data)
+		self.assertEqual(self.ran, [('price', {'close': True})])
+		self.assertEqual(self.sent, [('text', ['7. price close: done'])])
+		self.assertEqual(sent, ['Running price close…'])
+
+	def stale_buttons(self):
+		"""Capture the toast and the in-place edit that a refused button press produces."""
+		self.toasts, self.edits = [], []
+		self.patch(telegram, 'answerCallbackQuery', lambda cid, text=None: self.toasts.append(text))
+		self.patch(webhook, 'editMessageText', lambda chat, mid, text, markup=None: self.edits.append((chat, mid, text, markup)))
+
+	def test_forged_button_data_never_runs_anything_else(self):
+		self.stale_buttons()
+		good = jobs.checksum('trades.py')
+		for data in ('j|-1|' + good, 'j|0|' + good, 'j|9|' + good, 'j|8|0000', 'j|8|', 'j|8', 'j|8|' + good + '|x', 'j||' + good,
+					 'j|٨|' + good, 'j| 8|' + good, 'j|8.0|' + good, 'j|trades.py|' + good, 'j|1|' + good, 'j|8|price.py close',
+					 'j|8|' + good.upper() + 'x', 'j|99999999999999999999|' + good):
+			self.sent.clear()
+			self.toasts.clear(); self.edits.clear()
+			worker.process_callback('telegram', 'cb', '55', '9', data)
+			self.assertEqual(self.ran, [], data)
+			self.assertEqual(self.toasts, ['Buttons updated, try again'], data)
+			self.assertEqual(len(self.edits), 1, data)
+			self.assertEqual(self.sent, [], data) # no new message, the old one is edited
+		# cleanup and podcasts are scheduled but cannot be picked, by number or by their checksum
+		for label in ('cleanup.py', 'podcasts.py'):
+			for n in range(1, 12):
+				with self.assertRaises(jobs.ScheduleError):
+					self.s.pick(str(n), jobs.checksum(label))
+		worker.process_callback('telegram', 'cb', '55', '9', 'j|8|' + good) # the genuine button still works
+		self.assertEqual(self.ran, [('trades', {})])
+
+	def test_button_from_an_old_schedule_updates_the_message(self):
+		self.stale_buttons()
+		worker.process_callback('telegram', 'cb1', '55', '9', 'j|7|0000')
+		worker.process_callback('telegram', 'cb1', '55', '9', 'j|9|0000')
+		self.assertEqual(self.ran, [])
+		self.assertEqual(self.sent, [])
+		self.assertEqual(self.toasts, ['Buttons updated, try again'] * 2)
+		chat, mid, text, markup = self.edits[0]
+		self.assertEqual((chat, mid), ('55', '9')) # the message with the old buttons is edited in place
+		self.assertIn('7. price close', text)
+		self.assertEqual(markup, reports.prepare_jobs_payload('telegram', '')[1]) # with the current buttons
+		self.patch(jobs, 'current', None) # scheduler off: the list is replaced by the notice and the buttons removed
+		worker.process_callback('telegram', 'cb1', '55', '9', 'j|7|0000')
+		self.assertEqual(self.edits[-1][2:], ('Scheduled jobs are off (scheduler=False)', {'inline_keyboard': []}))
+
+	def test_run_in_chat(self):
+		self.assertEqual(self.command('.jobs run 8'), [('text', ['8. trades: done'])])
+		self.assertEqual(self.ran, [('trades', {})])
+		self.ran.clear()
+		self.assertEqual(self.command('.cron run price close')[0][1], ['7. price close: done'])
+		self.assertEqual(self.ran, [('price', {'close': True})])
+
+	def test_run_in_chat_refusals(self):
+		self.assertIn('not in the schedule', self.list_in_chat('.jobs run cal.py earnings')[0])
+		self.assertIn('no command 99', self.list_in_chat('.jobs run 99')[0])
+		self.assertIn('one command at a time', self.list_in_chat('.jobs run price.py close; trades.py')[0])
+		self.assertIn('Usage', self.list_in_chat('.jobs run')[0])
+		self.assertIn('Usage', self.list_in_chat('.jobs bogus')[0])
+		self.assertEqual(self.ran, [])
+
+	def test_scheduler_off(self):
+		self.patch(jobs, 'current', None)
+		self.assertIn('off', self.list_in_chat('.jobs')[0])
+		self.assertIn('off', self.list_in_chat('.jobs run 8')[0])
+
+	def test_job_is_still_a_ticker(self):
+		self.assertNotIn('Scheduled', str(self.command('.job')))
 
 class ChildProcess(FinbotCase):
-    """Jobs run in their own process, so heavy work cannot hold up the bot's web server."""
-
-    def sleeper(self, seconds):
-        real = jobs.subprocess.Popen
-        self.children = []
-        def popen(args, **kw):
-            self.children.append(real([sys.executable, '-c', f'import time; time.sleep({seconds})']))
-            return self.children[-1]
-        self.patch(jobs.subprocess, 'Popen', popen)
-
-    def test_every_command_can_be_handed_to_a_child(self):
-        parsed, _ = jobs.parse_schedule(DEFAULT_SCHEDULE.read_text(), UTC)
-        for job in parsed:
-            for label, module, kwargs in job.commands:
-                self.assertEqual(json.loads(json.dumps(kwargs)), kwargs, label)
-
-    def test_a_real_child_runs_the_handler_and_reports_its_exit_status(self):
-        real = jobs.subprocess.Popen
-        self.patch(jobs.subprocess, 'Popen', lambda args, **kw: real(args, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, **kw))
-        with tempfile.TemporaryDirectory() as data:
-            env = {'DATA_DIR': data, 'telegramBotToken': '', 'slack_webhook': '', 'discord_webhook': ''}
-            with mock.patch.dict(os.environ, env):
-                self.assertEqual(jobs.run_child('cleanup.py', cleanup, {}), 0)
-                self.assertEqual(jobs.run_child('cleanup.py 0', cleanup, {'days': 'not a number'}), jobs.jobrun.REPORTED)
-
-    def test_a_job_process_survives_a_reload_signal(self):
-        with tempfile.TemporaryDirectory() as scripts, tempfile.TemporaryDirectory() as data:
-            Path(scripts, 'sleeper.py').write_text('import time\ndef lambda_handler():\n    print("started", flush=True)\n    time.sleep(1)\n')
-            env = {**os.environ, 'PYTHONPATH': scripts, 'DATA_DIR': data, 'telegramBotToken': '', 'slack_webhook': '', 'discord_webhook': ''}
-            proc = subprocess.Popen([sys.executable, '-m', 'lib.jobrun', 'sleeper', '{}'], cwd=ROOT, env=env, stdout=subprocess.PIPE, text=True)
-            self.addCleanup(proc.kill)
-            for line in proc.stdout:
-                if line.strip() == 'started':
-                    break
-            proc.send_signal(signal.SIGHUP) # what `systemctl kill -s HUP finbot` does to every process of the unit
-            self.assertEqual(proc.wait(timeout=30), 0)
-
-    def test_a_script_that_cannot_be_imported_is_reported(self):
-        with tempfile.TemporaryDirectory() as data:
-            env = {**os.environ, 'DATA_DIR': data, 'telegramBotToken': '', 'slack_webhook': '', 'discord_webhook': ''}
-            out = subprocess.run([sys.executable, '-m', 'lib.jobrun', 'nosuchscript', '{}'], cwd=ROOT, env=env, capture_output=True, text=True)
-        self.assertEqual(out.returncode, jobs.jobrun.REPORTED, out.stderr)
-        self.assertIn('⚠️ nosuchscript.py: ModuleNotFoundError', out.stderr)
-
-    def test_scheduler_reports_a_child_that_ended_without_saying_why(self):
-        file = Path(tempfile.mkdtemp(), 'schedule.cron')
-        file.write_text('* * * * * trades.py\n')
-        for status, reason in ((-signal.SIGHUP, 'killed by SIGHUP'), (-signal.SIGKILL, 'killed by SIGKILL'), (1, 'exited with status 1 without saying why'),
-                               (jobs.jobrun.REPORTED, None)): # the child reported this one itself
-            with self.subTest(status=status):
-                self.errors.clear()
-                s = jobs.Scheduler(lambda: file, zone=UTC, runner=lambda label, module, kwargs: status)
-                self.assertTrue(s.load())
-                self.assertEqual(s.run_now('1'), ('trades.py', 'failed'))
-                self.assertEqual(self.errors, [f'⚠️ trades.py: RuntimeError: {reason}'] if reason else [])
-                self.assertEqual(s.busy, set())
-
-    def test_a_report_script_loads_neither_pandas_nor_the_scheduler(self):
-        """A job process imports only its own script: price used to pull in pandas and, through charts, reports and jobs, every
-        script there is."""
-        code = 'import sys, price, performance, trades, shorts, cal; print(sorted(m for m in ("pandas", "lib.jobs", "matplotlib") if m in sys.modules))'
-        with tempfile.TemporaryDirectory() as data:
-            out = subprocess.run([sys.executable, '-c', code], cwd=ROOT, capture_output=True, text=True, env={**os.environ, 'DATA_DIR': data})
-        self.assertEqual(out.stdout.strip().splitlines()[-1:], ['[]'], out.stderr)
-
-    def test_the_event_loop_keeps_running_while_a_child_works(self):
-        self.sleeper(1)
-        ticks = []
-        def ticker():
-            for _ in range(5):
-                gevent.sleep(0.1)
-                ticks.append(time.monotonic())
-        greenlet = gevent.spawn(ticker)
-        started = time.monotonic()
-        self.assertEqual(jobs.run_child('cleanup.py', cleanup, {}), 0)
-        greenlet.join()
-        self.assertGreater(time.monotonic() - started, 0.9)
-        self.assertEqual(len(ticks), 5) # the loop was not blocked by the child
-        self.assertLess(ticks[-1] - started, 0.8)
-
-    def test_a_timeout_kills_the_child(self):
-        self.sleeper(30)
-        started = time.monotonic()
-        with self.assertRaises(gevent.Timeout):
-            with gevent.Timeout(0.3):
-                jobs.run_child('cleanup.py', cleanup, {})
-        self.assertLess(time.monotonic() - started, 5)
-        self.assertIsNotNone(self.children[0].poll()) # the child was stopped, not left running
+	"""Jobs run in their own process, so heavy work cannot hold up the bot's web server."""
+
+	def sleeper(self, seconds):
+		real = jobs.subprocess.Popen
+		self.children = []
+		def popen(args, **kw):
+			self.children.append(real([sys.executable, '-c', f'import time; time.sleep({seconds})']))
+			return self.children[-1]
+		self.patch(jobs.subprocess, 'Popen', popen)
+
+	def test_every_command_can_be_handed_to_a_child(self):
+		parsed, _ = jobs.parse_schedule(DEFAULT_SCHEDULE.read_text(), UTC)
+		for job in parsed:
+			for label, module, kwargs in job.commands:
+				self.assertEqual(json.loads(json.dumps(kwargs)), kwargs, label)
+
+	def test_a_real_child_runs_the_handler_and_reports_its_exit_status(self):
+		real = jobs.subprocess.Popen
+		self.patch(jobs.subprocess, 'Popen', lambda args, **kw: real(args, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, **kw))
+		with tempfile.TemporaryDirectory() as data:
+			env = {'DATA_DIR': data, 'telegramBotToken': '', 'slack_webhook': '', 'discord_webhook': ''}
+			with mock.patch.dict(os.environ, env):
+				self.assertEqual(jobs.run_child('cleanup.py', cleanup, {}), 0)
+				self.assertEqual(jobs.run_child('cleanup.py 0', cleanup, {'days': 'not a number'}), jobs.jobrun.REPORTED)
+
+	def test_a_job_process_survives_a_reload_signal(self):
+		with tempfile.TemporaryDirectory() as scripts, tempfile.TemporaryDirectory() as data:
+			Path(scripts, 'sleeper.py').write_text('import time\ndef lambda_handler():\n    print("started", flush=True)\n    time.sleep(1)\n')
+			env = {**os.environ, 'PYTHONPATH': scripts, 'DATA_DIR': data, 'telegramBotToken': '', 'slack_webhook': '', 'discord_webhook': ''}
+			proc = subprocess.Popen([sys.executable, '-m', 'lib.jobrun', 'sleeper', '{}'], cwd=ROOT, env=env, stdout=subprocess.PIPE, text=True)
+			self.addCleanup(proc.kill)
+			for line in proc.stdout:
+				if line.strip() == 'started':
+					break
+			proc.send_signal(signal.SIGHUP) # what `systemctl kill -s HUP finbot` does to every process of the unit
+			self.assertEqual(proc.wait(timeout=30), 0)
+
+	def test_a_script_that_cannot_be_imported_is_reported(self):
+		with tempfile.TemporaryDirectory() as data:
+			env = {**os.environ, 'DATA_DIR': data, 'telegramBotToken': '', 'slack_webhook': '', 'discord_webhook': ''}
+			out = subprocess.run([sys.executable, '-m', 'lib.jobrun', 'nosuchscript', '{}'], cwd=ROOT, env=env, capture_output=True, text=True)
+		self.assertEqual(out.returncode, jobs.jobrun.REPORTED, out.stderr)
+		self.assertIn('⚠️ nosuchscript.py: ModuleNotFoundError', out.stderr)
+
+	def test_scheduler_reports_a_child_that_ended_without_saying_why(self):
+		file = Path(tempfile.mkdtemp(), 'schedule.cron')
+		file.write_text('* * * * * trades.py\n')
+		for status, reason in ((-signal.SIGHUP, 'killed by SIGHUP'), (-signal.SIGKILL, 'killed by SIGKILL'), (1, 'exited with status 1 without saying why'),
+							   (jobs.jobrun.REPORTED, None)): # the child reported this one itself
+			with self.subTest(status=status):
+				self.errors.clear()
+				s = jobs.Scheduler(lambda: file, zone=UTC, runner=lambda label, module, kwargs: status)
+				self.assertTrue(s.load())
+				self.assertEqual(s.run_now('1'), ('trades.py', 'failed'))
+				self.assertEqual(self.errors, [f'⚠️ trades.py: RuntimeError: {reason}'] if reason else [])
+				self.assertEqual(s.busy, set())
+
+	def test_a_report_script_loads_neither_pandas_nor_the_scheduler(self):
+		"""A job process imports only its own script: price used to pull in pandas and, through charts, reports and jobs, every
+		script there is."""
+		code = 'import sys, price, performance, trades, shorts, cal; print(sorted(m for m in ("pandas", "lib.jobs", "matplotlib") if m in sys.modules))'
+		with tempfile.TemporaryDirectory() as data:
+			out = subprocess.run([sys.executable, '-c', code], cwd=ROOT, capture_output=True, text=True, env={**os.environ, 'DATA_DIR': data})
+		self.assertEqual(out.stdout.strip().splitlines()[-1:], ['[]'], out.stderr)
+
+	def test_the_event_loop_keeps_running_while_a_child_works(self):
+		self.sleeper(1)
+		ticks = []
+		def ticker():
+			for _ in range(5):
+				gevent.sleep(0.1)
+				ticks.append(time.monotonic())
+		greenlet = gevent.spawn(ticker)
+		started = time.monotonic()
+		self.assertEqual(jobs.run_child('cleanup.py', cleanup, {}), 0)
+		greenlet.join()
+		self.assertGreater(time.monotonic() - started, 0.9)
+		self.assertEqual(len(ticks), 5) # the loop was not blocked by the child
+		self.assertLess(ticks[-1] - started, 0.8)
+
+	def test_a_timeout_kills_the_child(self):
+		self.sleeper(30)
+		started = time.monotonic()
+		with self.assertRaises(gevent.Timeout):
+			with gevent.Timeout(0.3):
+				jobs.run_child('cleanup.py', cleanup, {})
+		self.assertLess(time.monotonic() - started, 5)
+		self.assertIsNotNone(self.children[0].poll()) # the child was stopped, not left running
 
 
 
 if __name__ == '__main__':
-    unittest.main()
+	unittest.main()
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 6af35bda..43736735 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -9,759 +9,759 @@
 
 
 def sample(*path):
-    with open(os.path.join(DOC, *path), encoding='utf-8') as f:
-        return json.load(f)
+	with open(os.path.join(DOC, *path), encoding='utf-8') as f:
+		return json.load(f)
 
 
 class Resp:
-    def __init__(self, payload=None, status=200, text=''):
-        self.payload, self.status_code, self.text = payload, status, text or json.dumps(payload)
-    def json(self):
-        if self.payload is None:
-            raise ValueError('no json')
-        return self.payload
+	def __init__(self, payload=None, status=200, text=''):
+		self.payload, self.status_code, self.text = payload, status, text or json.dumps(payload)
+	def json(self):
+		if self.payload is None:
+			raise ValueError('no json')
+		return self.payload
 
 
 def serve(*responses):
-    """net.get replacement: replays the responses in order (exceptions are raised); the last one repeats."""
-    queue = list(responses)
-    calls = []
-    def get(url, **kwargs):
-        calls.append(url)
-        item = queue.pop(0) if len(queue) > 1 else queue[0]
-        if isinstance(item, Exception):
-            raise item
-        return item
-    get.calls = calls
-    return get
+	"""net.get replacement: replays the responses in order (exceptions are raised); the last one repeats."""
+	queue = list(responses)
+	calls = []
+	def get(url, **kwargs):
+		calls.append(url)
+		item = queue.pop(0) if len(queue) > 1 else queue[0]
+		if isinstance(item, Exception):
+			raise item
+		return item
+	get.calls = calls
+	return get
 
 
 class ProviderCase(FinbotCase):
-    def setUp(self):
-        super().setUp()
-        # these tests exercise the real parsers, so undo the canned data from the base fixture
-        for module, name in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (sharesight, 'get_portfolios'),
-                             (sharesight, 'get_trades'), (sharesight, 'get_performance'), (sharesight, 'get_performance_wrapper')):
-            self.patch(module, name, REAL[(module.__name__, name)])
-        self.patch(yahoo, 'getCookie', lambda *a, **k: 'A=B')
-        self.patch(yahoo, 'getCrumb', lambda *a, **k: 'crumb')
-        self.patch(sharesight, 'get_token', lambda: 'token')
-        self.patch(sharesight.time, 'sleep', lambda s: None)
-
-    def http(self, module, *responses):
-        getter = serve(*responses)
-        self.patch(module.net, 'get', getter)
-        return getter
+	def setUp(self):
+		super().setUp()
+		# these tests exercise the real parsers, so undo the canned data from the base fixture
+		for module, name in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (sharesight, 'get_portfolios'),
+							 (sharesight, 'get_trades'), (sharesight, 'get_performance'), (sharesight, 'get_performance_wrapper')):
+			self.patch(module, name, REAL[(module.__name__, name)])
+		self.patch(yahoo, 'getCookie', lambda *a, **k: 'A=B')
+		self.patch(yahoo, 'getCrumb', lambda *a, **k: 'crumb')
+		self.patch(sharesight, 'get_token', lambda: 'token')
+		self.patch(sharesight.time, 'sleep', lambda s: None)
+
+	def http(self, module, *responses):
+		getter = serve(*responses)
+		self.patch(module.net, 'get', getter)
+		return getter
 
 
 class YahooQuotes(ProviderCase):
-    def test_parses_real_v7_quote(self):
-        self.http(yahoo, Resp(sample('yahoo', 'sample_yahoo_v7.json')))
-        data = yahoo.fetch(['AAPL'])['AAPL']
-        self.assertEqual(data['profile_title'], 'Apple')
-        self.assertEqual(data['percent_change'], 0.09)
-        self.assertEqual(data['percent_change_postmarket'], -0.13)
-        self.assertEqual(data['quoteType'], 'EQUITY')
-        self.assertEqual(data['marketState'], 'CLOSED')
-        self.assertEqual(data['profile_exchange'], 'NasdaqGS')
-        self.assertEqual(data['exchangeTimezoneName'], 'America/New_York')
-        self.assertGreater(data['market_cap'], 1e12)
-
-    def test_second_call_is_served_from_cache(self):
-        getter = self.http(yahoo, Resp(sample('yahoo', 'sample_yahoo_v7.json')))
-        yahoo.fetch(['AAPL'])
-        yahoo.fetch(['AAPL'])
-        self.assertEqual(len(getter.calls), 1)
-
-    def test_falls_back_to_second_host(self):
-        getter = self.http(yahoo, Resp(status=500), Resp(sample('yahoo', 'sample_yahoo_v7.json')))
-        self.assertIn('AAPL', yahoo.fetch(['AAPL']))
-        self.assertIn('query1', getter.calls[1])
-
-    def test_all_hosts_failing_raises(self):
-        for failure in (Resp(status=500), ConnectionError('down')):
-            with self.subTest(failure=failure):
-                self.http(yahoo, failure)
-                with self.assertRaises(RuntimeError):
-                    yahoo.fetch(['AAPL'])
-
-    def test_empty_result_is_none(self):
-        self.http(yahoo, Resp({'quoteResponse': {'result': None, 'error': {'code': 'x'}}}))
-        self.assertEqual(yahoo.fetch(['NOPE']), {})
-
-    def test_no_tickers_is_none(self):
-        self.assertEqual(yahoo.fetch([]), {})
-
-    def test_incomplete_quotes_are_skipped(self):
-        body = sample('yahoo', 'sample_yahoo_v7.json')
-        good = body['quoteResponse']['result'][0]
-        no_price = {k: v for k, v in good.items() if k != 'regularMarketPrice'} | {'symbol': 'NOPRICE'}
-        no_name = {k: v for k, v in good.items() if k not in ('longName', 'shortName')} | {'symbol': 'NONAME'}
-        body['quoteResponse']['result'] = [good, no_price, no_name]
-        self.http(yahoo, Resp(body))
-        self.assertEqual(list(yahoo.fetch(['AAPL', 'NOPRICE', 'NONAME'])), ['AAPL'])
+	def test_parses_real_v7_quote(self):
+		self.http(yahoo, Resp(sample('yahoo', 'sample_yahoo_v7.json')))
+		data = yahoo.fetch(['AAPL'])['AAPL']
+		self.assertEqual(data['profile_title'], 'Apple')
+		self.assertEqual(data['percent_change'], 0.09)
+		self.assertEqual(data['percent_change_postmarket'], -0.13)
+		self.assertEqual(data['quoteType'], 'EQUITY')
+		self.assertEqual(data['marketState'], 'CLOSED')
+		self.assertEqual(data['profile_exchange'], 'NasdaqGS')
+		self.assertEqual(data['exchangeTimezoneName'], 'America/New_York')
+		self.assertGreater(data['market_cap'], 1e12)
+
+	def test_second_call_is_served_from_cache(self):
+		getter = self.http(yahoo, Resp(sample('yahoo', 'sample_yahoo_v7.json')))
+		yahoo.fetch(['AAPL'])
+		yahoo.fetch(['AAPL'])
+		self.assertEqual(len(getter.calls), 1)
+
+	def test_falls_back_to_second_host(self):
+		getter = self.http(yahoo, Resp(status=500), Resp(sample('yahoo', 'sample_yahoo_v7.json')))
+		self.assertIn('AAPL', yahoo.fetch(['AAPL']))
+		self.assertIn('query1', getter.calls[1])
+
+	def test_all_hosts_failing_raises(self):
+		for failure in (Resp(status=500), ConnectionError('down')):
+			with self.subTest(failure=failure):
+				self.http(yahoo, failure)
+				with self.assertRaises(RuntimeError):
+					yahoo.fetch(['AAPL'])
+
+	def test_empty_result_is_none(self):
+		self.http(yahoo, Resp({'quoteResponse': {'result': None, 'error': {'code': 'x'}}}))
+		self.assertEqual(yahoo.fetch(['NOPE']), {})
+
+	def test_no_tickers_is_none(self):
+		self.assertEqual(yahoo.fetch([]), {})
+
+	def test_incomplete_quotes_are_skipped(self):
+		body = sample('yahoo', 'sample_yahoo_v7.json')
+		good = body['quoteResponse']['result'][0]
+		no_price = {k: v for k, v in good.items() if k != 'regularMarketPrice'} | {'symbol': 'NOPRICE'}
+		no_name = {k: v for k, v in good.items() if k not in ('longName', 'shortName')} | {'symbol': 'NONAME'}
+		body['quoteResponse']['result'] = [good, no_price, no_name]
+		self.http(yahoo, Resp(body))
+		self.assertEqual(list(yahoo.fetch(['AAPL', 'NOPRICE', 'NONAME'])), ['AAPL'])
 
 
 class YahooDetail(ProviderCase):
-    @staticmethod
-    def combined():
-        """quoteSummary as Yahoo returns it for several modules at once, assembled from the per-module samples."""
-        merged = {}
-        for module in ('price', 'summaryProfile', 'summaryDetail', 'defaultKeyStatistics', 'financialData', 'calendarEvents', 'earnings'):
-            merged.update(sample('yahoo', f'sample_yahoo_v10_{module}.json')['quoteSummary']['result'][0])
-        return {'quoteSummary': {'result': [merged], 'error': None}}
-
-    def test_parses_real_quote_summary(self):
-        self.http(yahoo, Resp(self.combined()))
-        data = yahoo.fetch_detail('NVDA')['NVDA']
-        self.assertEqual(data['profile_title'], 'NVIDIA')
-        self.assertTrue(data['profile_bio'])
-
-    def test_a_cached_detail_needs_no_cookie(self):
-        self.http(yahoo, Resp(self.combined()))
-        yahoo.fetch_detail('NVDA')
-        def refuse():
-            raise AssertionError('a cached detail took the Yahoo auth lock')
-        self.patch(yahoo, 'credentials', refuse)
-        self.assertIn('NVDA', yahoo.fetch_detail('NVDA'))
-
-    def test_http_failure_gives_empty(self):
-        for failure in (Resp(status=404), ConnectionError('down')):
-            with self.subTest(failure=failure):
-                self.http(yahoo, failure)
-                self.assertEqual(yahoo.fetch_detail('DRNA'), {})
-
-    def test_missing_names_give_empty(self):
-        body = self.combined()
-        body['quoteSummary']['result'][0]['price'].update(longName=None, shortName=None)
-        self.http(yahoo, Resp(body))
-        self.assertEqual(yahoo.fetch_detail('DUB'), {})
-        self.http(yahoo, Resp({'quoteSummary': {'result': None, 'error': {}}}))
-        self.assertEqual(yahoo.fetch_detail('GONE'), {})
+	@staticmethod
+	def combined():
+		"""quoteSummary as Yahoo returns it for several modules at once, assembled from the per-module samples."""
+		merged = {}
+		for module in ('price', 'summaryProfile', 'summaryDetail', 'defaultKeyStatistics', 'financialData', 'calendarEvents', 'earnings'):
+			merged.update(sample('yahoo', f'sample_yahoo_v10_{module}.json')['quoteSummary']['result'][0])
+		return {'quoteSummary': {'result': [merged], 'error': None}}
+
+	def test_parses_real_quote_summary(self):
+		self.http(yahoo, Resp(self.combined()))
+		data = yahoo.fetch_detail('NVDA')['NVDA']
+		self.assertEqual(data['profile_title'], 'NVIDIA')
+		self.assertTrue(data['profile_bio'])
+
+	def test_a_cached_detail_needs_no_cookie(self):
+		self.http(yahoo, Resp(self.combined()))
+		yahoo.fetch_detail('NVDA')
+		def refuse():
+			raise AssertionError('a cached detail took the Yahoo auth lock')
+		self.patch(yahoo, 'credentials', refuse)
+		self.assertIn('NVDA', yahoo.fetch_detail('NVDA'))
+
+	def test_http_failure_gives_empty(self):
+		for failure in (Resp(status=404), ConnectionError('down')):
+			with self.subTest(failure=failure):
+				self.http(yahoo, failure)
+				self.assertEqual(yahoo.fetch_detail('DRNA'), {})
+
+	def test_missing_names_give_empty(self):
+		body = self.combined()
+		body['quoteSummary']['result'][0]['price'].update(longName=None, shortName=None)
+		self.http(yahoo, Resp(body))
+		self.assertEqual(yahoo.fetch_detail('DUB'), {})
+		self.http(yahoo, Resp({'quoteSummary': {'result': None, 'error': {}}}))
+		self.assertEqual(yahoo.fetch_detail('GONE'), {})
 
 
 class YahooChart(ProviderCase):
-    def test_real_chart_to_dataframe_and_basics(self):
-        chart = sample('yahoo', 'sample_yahoo_v8_chart.json')
-        df = yahoo.chart_json_to_df(chart)
-        self.assertEqual(len(df), 2)
-        self.assertTrue({'Date', 'Close'} <= set(df.columns))
-        basics = yahoo.chart_json_to_stock_basics(chart)
-        self.assertEqual((basics['symbol'], basics['currency']), ('CRWD', 'USD'))
-
-    def test_fetch_chart_json_failures_raise(self):
-        for failure in (Resp(status=404), ConnectionError('down')):
-            with self.subTest(failure=failure):
-                self.http(yahoo, failure)
-                with self.assertRaises(RuntimeError):
-                    yahoo.fetch_chart_json('AAPL', full=True)
-
-    @staticmethod
-    def chart(closes, failing=()):
-        """A v8 chart response with one daily bar per close, the last one today (14:30 UTC, New York's open)."""
-        today = datetime.datetime.now(datetime.timezone.utc).replace(hour=14, minute=30, second=0, microsecond=0)
-        stamps = [int((today - datetime.timedelta(days=len(closes) - 1 - i)).timestamp()) for i in range(len(closes))]
-        meta = {'currency': 'USD', 'symbol': 'T1', 'exchangeTimezoneName': 'America/New_York', 'regularMarketTime': stamps[-1],
-                'regularMarketPrice': closes[-1], 'longName': 'Co T1'}
-        quote = {'open': closes, 'high': closes, 'low': closes, 'close': closes, 'volume': [1] * len(closes)}
-        return {'chart': {'result': [{'meta': meta, 'timestamp': stamps, 'indicators': {'quote': [quote]}}], 'error': None}}
-
-    def test_1d_is_the_change_since_the_previous_close(self):
-        self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
-        self.http(yahoo, Resp(self.chart([90] * 30 + [100, 105, 110, 115, 120, 116.4]))) # up 20% in a week, down 3% today
-        table, image = yahoo.price_history('T1', graph=False)
-        self.assertEqual(table['1D'], -3)
-        self.assertGreater(table['5D'], 10) # the week, not the day
-
-    def test_a_failed_chart_fetch_raises(self):
-        self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
-        for failure in (Resp(status=429), ConnectionError('down')):
-            with self.subTest(failure=failure):
-                self.http(yahoo, failure)
-                with self.assertRaises(RuntimeError):
-                    yahoo.price_history('AAPL', 7, graph=False)
-
-    def test_one_failed_ticker_does_not_take_the_report_down(self):
-        self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
-        self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts}) # quotes as canned in the base fixture, charts over HTTP
-        self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t)})
-        self.patch(sharesight, 'get_performance_wrapper', lambda d: {}) # no Sharesight figures: every ticker needs its Yahoo history
-        good = Resp(self.chart([100.0] * 20 + [110.0]))
-        def get(url, **kwargs):
-            return Resp(status=404) if '/chart/T3?' in url or '/chart/SPY?' in url else good
-        self.patch(yahoo.net, 'get', get)
-        def slack(text): # Slack: the reports come as text
-            self.sent.clear(); self.errors.clear()
-            worker.process_request('slack', 'C1', '<@U1>', text, BOT, 'U')
-            self.assertEqual(self.errors, [], text)
-            return '\n'.join(map(str, self.sent[-1][1]))
-        price_list = slack('.price 7d') # T3 is left out, the rest are listed
-        self.assertNotIn('Co T3 (', price_list)
-        self.assertIn('Co T4 (', price_list)
-        self.patch(sharesight, 'get_portfolios', lambda: {'Rob': 1})
-        self.patch(sharesight, 'get_performance', lambda pid, d, *a: {'report': {'holdings': [{'portfolio': {'name': 'Rob'}}],
-            'currency_gain_percent': '0', 'capital_gain_percent': '2.5', 'total_gain_percent': '0'}})
-        report = slack('.performance 7d') # the S&P 500 benchmark has no history
-        self.assertIn('⚠️ S&P 500: no price history', report)
-        self.assertIn('NASDAQ 100', report)
-        self.command('.T3') # the profile goes out without its history lines
-        self.sent.clear(); self.errors.clear()
-        worker.process_request('telegram', '55', '@u', '.price T3 7d', BOT, 'U')
-        self.assertEqual(len(self.errors), 1) # the one ticker asked for: one line saying why
-        self.assertTrue(self.errors[0].startswith('⚠️ .price: RuntimeError: 404 error for T3'))
-
-    def test_the_chart_request_carries_the_cookie_of_its_crumb(self):
-        headers = []
-        self.patch(yahoo.net, 'get', lambda url, **kwargs: headers.append(kwargs.get('headers', {})) or Resp(self.chart([100.0] * 5)))
-        yahoo.fetch_chart_json('T1', full=True)
-        self.assertEqual(headers[0].get('Cookie'), 'A=B') # the crumb used to go without its cookie
-
-    def test_with_the_cache_off_no_chart_is_written(self):
-        self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
-        self.patch(yahoo, 'config_cache', False)
-        self.http(yahoo, Resp(self.chart([100.0] * 30 + [110.0])))
-        table, image = yahoo.price_history('T1', 7, graph=True)
-        self.assertIsNotNone(image)
-        self.assertEqual([name for name in os.listdir(util.config_cache_dir) if name.startswith('finbot_graph_')], [])
-
-    def test_price_series_raises_on_failure_and_on_empty_history(self):
-        self.http(yahoo, Resp(status=500))
-        with self.assertRaises(RuntimeError):
-            yahoo.price_series('AAPL')
-        empty = sample('yahoo', 'sample_yahoo_v8_chart.json')
-        empty['chart']['result'][0]['indicators']['quote'][0]['close'] = [None, None]
-        self.http(yahoo, Resp(empty))
-        with self.assertRaises(RuntimeError):
-            yahoo.price_series('EMPTY')
+	def test_real_chart_to_dataframe_and_basics(self):
+		chart = sample('yahoo', 'sample_yahoo_v8_chart.json')
+		df = yahoo.chart_json_to_df(chart)
+		self.assertEqual(len(df), 2)
+		self.assertTrue({'Date', 'Close'} <= set(df.columns))
+		basics = yahoo.chart_json_to_stock_basics(chart)
+		self.assertEqual((basics['symbol'], basics['currency']), ('CRWD', 'USD'))
+
+	def test_fetch_chart_json_failures_raise(self):
+		for failure in (Resp(status=404), ConnectionError('down')):
+			with self.subTest(failure=failure):
+				self.http(yahoo, failure)
+				with self.assertRaises(RuntimeError):
+					yahoo.fetch_chart_json('AAPL', full=True)
+
+	@staticmethod
+	def chart(closes, failing=()):
+		"""A v8 chart response with one daily bar per close, the last one today (14:30 UTC, New York's open)."""
+		today = datetime.datetime.now(datetime.timezone.utc).replace(hour=14, minute=30, second=0, microsecond=0)
+		stamps = [int((today - datetime.timedelta(days=len(closes) - 1 - i)).timestamp()) for i in range(len(closes))]
+		meta = {'currency': 'USD', 'symbol': 'T1', 'exchangeTimezoneName': 'America/New_York', 'regularMarketTime': stamps[-1],
+				'regularMarketPrice': closes[-1], 'longName': 'Co T1'}
+		quote = {'open': closes, 'high': closes, 'low': closes, 'close': closes, 'volume': [1] * len(closes)}
+		return {'chart': {'result': [{'meta': meta, 'timestamp': stamps, 'indicators': {'quote': [quote]}}], 'error': None}}
+
+	def test_1d_is_the_change_since_the_previous_close(self):
+		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
+		self.http(yahoo, Resp(self.chart([90] * 30 + [100, 105, 110, 115, 120, 116.4]))) # up 20% in a week, down 3% today
+		table, image = yahoo.price_history('T1', graph=False)
+		self.assertEqual(table['1D'], -3)
+		self.assertGreater(table['5D'], 10) # the week, not the day
+
+	def test_a_failed_chart_fetch_raises(self):
+		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
+		for failure in (Resp(status=429), ConnectionError('down')):
+			with self.subTest(failure=failure):
+				self.http(yahoo, failure)
+				with self.assertRaises(RuntimeError):
+					yahoo.price_history('AAPL', 7, graph=False)
+
+	def test_one_failed_ticker_does_not_take_the_report_down(self):
+		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
+		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts}) # quotes as canned in the base fixture, charts over HTTP
+		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t)})
+		self.patch(sharesight, 'get_performance_wrapper', lambda d: {}) # no Sharesight figures: every ticker needs its Yahoo history
+		good = Resp(self.chart([100.0] * 20 + [110.0]))
+		def get(url, **kwargs):
+			return Resp(status=404) if '/chart/T3?' in url or '/chart/SPY?' in url else good
+		self.patch(yahoo.net, 'get', get)
+		def slack(text): # Slack: the reports come as text
+			self.sent.clear(); self.errors.clear()
+			worker.process_request('slack', 'C1', '<@U1>', text, BOT, 'U')
+			self.assertEqual(self.errors, [], text)
+			return '\n'.join(map(str, self.sent[-1][1]))
+		price_list = slack('.price 7d') # T3 is left out, the rest are listed
+		self.assertNotIn('Co T3 (', price_list)
+		self.assertIn('Co T4 (', price_list)
+		self.patch(sharesight, 'get_portfolios', lambda: {'Rob': 1})
+		self.patch(sharesight, 'get_performance', lambda pid, d, *a: {'report': {'holdings': [{'portfolio': {'name': 'Rob'}}],
+			'currency_gain_percent': '0', 'capital_gain_percent': '2.5', 'total_gain_percent': '0'}})
+		report = slack('.performance 7d') # the S&P 500 benchmark has no history
+		self.assertIn('⚠️ S&P 500: no price history', report)
+		self.assertIn('NASDAQ 100', report)
+		self.command('.T3') # the profile goes out without its history lines
+		self.sent.clear(); self.errors.clear()
+		worker.process_request('telegram', '55', '@u', '.price T3 7d', BOT, 'U')
+		self.assertEqual(len(self.errors), 1) # the one ticker asked for: one line saying why
+		self.assertTrue(self.errors[0].startswith('⚠️ .price: RuntimeError: 404 error for T3'))
+
+	def test_the_chart_request_carries_the_cookie_of_its_crumb(self):
+		headers = []
+		self.patch(yahoo.net, 'get', lambda url, **kwargs: headers.append(kwargs.get('headers', {})) or Resp(self.chart([100.0] * 5)))
+		yahoo.fetch_chart_json('T1', full=True)
+		self.assertEqual(headers[0].get('Cookie'), 'A=B') # the crumb used to go without its cookie
+
+	def test_with_the_cache_off_no_chart_is_written(self):
+		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
+		self.patch(yahoo, 'config_cache', False)
+		self.http(yahoo, Resp(self.chart([100.0] * 30 + [110.0])))
+		table, image = yahoo.price_history('T1', 7, graph=True)
+		self.assertIsNotNone(image)
+		self.assertEqual([name for name in os.listdir(util.config_cache_dir) if name.startswith('finbot_graph_')], [])
+
+	def test_price_series_raises_on_failure_and_on_empty_history(self):
+		self.http(yahoo, Resp(status=500))
+		with self.assertRaises(RuntimeError):
+			yahoo.price_series('AAPL')
+		empty = sample('yahoo', 'sample_yahoo_v8_chart.json')
+		empty['chart']['result'][0]['indicators']['quote'][0]['close'] = [None, None]
+		self.http(yahoo, Resp(empty))
+		with self.assertRaises(RuntimeError):
+			yahoo.price_series('EMPTY')
 
 
 class SharesightApi(ProviderCase):
-    def test_portfolios_from_real_response(self):
-        self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_portfolios.json')))
-        self.assertEqual(sharesight.get_portfolios(), {'Rob': 445825, 'RobSMSF': 684141})
-
-    def test_portfolio_exclusion_and_inclusion(self):
-        self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_portfolios.json')))
-        self.patch(sharesight, 'config_exclude_portfolios', ['445825'])
-        self.assertEqual(sharesight.get_portfolios(), {'RobSMSF': 684141})
-        self.patch(sharesight, 'config_exclude_portfolios', [])
-        self.patch(sharesight, 'config_include_portfolios', ['445825'])
-        self.assertEqual(sharesight.get_portfolios(), {'Rob': 445825})
-
-    def test_portfolio_failures_raise(self):
-        for failure in (Resp({'error': 'nope', 'error_code': 'x'}, 401), ConnectionError('down'), Resp({'portfolios': []})):
-            with self.subTest(failure=failure):
-                self.http(sharesight, failure)
-                with self.assertRaises(RuntimeError):
-                    sharesight.get_portfolios()
-
-    def test_trades_from_real_response(self):
-        self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_trades_684141_1.json')))
-        trades = sharesight.get_trades('RobSMSF', 684141, 1)
-        self.assertEqual(len(trades), 2)
-        self.assertEqual(trades[0]['portfolio'], 'RobSMSF')
-        self.assertEqual(trades[0]['symbol'], 'FANG')
-
-    def test_trade_failures_raise(self):
-        self.http(sharesight, ConnectionError('down'))
-        with self.assertRaises(RuntimeError):
-            sharesight.get_trades('RobSMSF', 684141, 1)
-        self.http(sharesight, Resp({'trades': [], 'error': 'bad', 'error_code': 'x'}, 400))
-        with self.assertRaises(RuntimeError):
-            sharesight.get_trades('RobSMSF', 684141, 2)
-
-    def test_performance_and_holdings_from_real_response(self):
-        self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_performance_684141_0.json')))
-        self.assertEqual(len(sharesight.get_performance(684141, 0)['report']['holdings']), 18)
-        tickers = sharesight.get_holdings('RobSMSF', 684141)
-        self.assertIn('ARM', tickers) # US venues carry no Yahoo suffix
-        self.assertEqual(tickers, sorted(tickers))
-
-    def test_performance_retries_when_sharesight_is_busy(self):
-        busy = Resp({'error': 'Too many parallel requests', 'error_code': 'x'}, 429)
-        getter = self.http(sharesight, busy, busy, Resp(sample('sharesight', 'finbot_sharesight_performance_684141_0.json')))
-        self.assertIn('report', sharesight.get_performance(684141, 0))
-        self.assertEqual(len(getter.calls), 3)
-
-    def test_performance_gives_up_after_repeated_busy_or_hard_errors(self):
-        for failure in (Resp({'error': 'Too many parallel requests', 'error_code': 'x'}, 429),
-                        Resp({'error': 'forbidden', 'error_code': 'x'}, 403), ConnectionError('down')):
-            with self.subTest(failure=failure):
-                self.http(sharesight, failure)
-                with self.assertRaises(RuntimeError):
-                    sharesight.get_performance(684141, 7)
+	def test_portfolios_from_real_response(self):
+		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_portfolios.json')))
+		self.assertEqual(sharesight.get_portfolios(), {'Rob': 445825, 'RobSMSF': 684141})
+
+	def test_portfolio_exclusion_and_inclusion(self):
+		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_portfolios.json')))
+		self.patch(sharesight, 'config_exclude_portfolios', ['445825'])
+		self.assertEqual(sharesight.get_portfolios(), {'RobSMSF': 684141})
+		self.patch(sharesight, 'config_exclude_portfolios', [])
+		self.patch(sharesight, 'config_include_portfolios', ['445825'])
+		self.assertEqual(sharesight.get_portfolios(), {'Rob': 445825})
+
+	def test_portfolio_failures_raise(self):
+		for failure in (Resp({'error': 'nope', 'error_code': 'x'}, 401), ConnectionError('down'), Resp({'portfolios': []})):
+			with self.subTest(failure=failure):
+				self.http(sharesight, failure)
+				with self.assertRaises(RuntimeError):
+					sharesight.get_portfolios()
+
+	def test_trades_from_real_response(self):
+		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_trades_684141_1.json')))
+		trades = sharesight.get_trades('RobSMSF', 684141, 1)
+		self.assertEqual(len(trades), 2)
+		self.assertEqual(trades[0]['portfolio'], 'RobSMSF')
+		self.assertEqual(trades[0]['symbol'], 'FANG')
+
+	def test_trade_failures_raise(self):
+		self.http(sharesight, ConnectionError('down'))
+		with self.assertRaises(RuntimeError):
+			sharesight.get_trades('RobSMSF', 684141, 1)
+		self.http(sharesight, Resp({'trades': [], 'error': 'bad', 'error_code': 'x'}, 400))
+		with self.assertRaises(RuntimeError):
+			sharesight.get_trades('RobSMSF', 684141, 2)
+
+	def test_performance_and_holdings_from_real_response(self):
+		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_performance_684141_0.json')))
+		self.assertEqual(len(sharesight.get_performance(684141, 0)['report']['holdings']), 18)
+		tickers = sharesight.get_holdings('RobSMSF', 684141)
+		self.assertIn('ARM', tickers) # US venues carry no Yahoo suffix
+		self.assertEqual(tickers, sorted(tickers))
+
+	def test_performance_retries_when_sharesight_is_busy(self):
+		busy = Resp({'error': 'Too many parallel requests', 'error_code': 'x'}, 429)
+		getter = self.http(sharesight, busy, busy, Resp(sample('sharesight', 'finbot_sharesight_performance_684141_0.json')))
+		self.assertIn('report', sharesight.get_performance(684141, 0))
+		self.assertEqual(len(getter.calls), 3)
+
+	def test_performance_gives_up_after_repeated_busy_or_hard_errors(self):
+		for failure in (Resp({'error': 'Too many parallel requests', 'error_code': 'x'}, 429),
+						Resp({'error': 'forbidden', 'error_code': 'x'}, 403), ConnectionError('down')):
+			with self.subTest(failure=failure):
+				self.http(sharesight, failure)
+				with self.assertRaises(RuntimeError):
+					sharesight.get_performance(684141, 7)
 
 
 class SharesightKeepWarm(ProviderCase):
-    """The background refresher keeps the portfolio list and the holdings report (days=0) cached ahead of their TTLs."""
-
-    def serve_sharesight(self):
-        portfolios = sample('sharesight', 'finbot_sharesight_portfolios.json')
-        performance = sample('sharesight', 'finbot_sharesight_performance_684141_0.json')
-        calls = []
-        def get(url, **kwargs):
-            calls.append(url)
-            return Resp(performance if '/performance' in url else portfolios)
-        self.patch(sharesight.net, 'get', get)
-        return calls
-
-    def age(self, cache_file, seconds):
-        path = os.path.join(util.config_cache_dir, cache_file)
-        old = os.path.getmtime(path) - seconds
-        os.utime(path, (old, old))
-
-    def test_cold_cache_is_filled_then_left_alone(self):
-        calls = self.serve_sharesight()
-        self.assertEqual(sharesight.warm()[0], 3) # portfolio list + one holdings report per portfolio (2)
-        self.assertEqual(sum('/performance' in c for c in calls), 2)
-        self.assertTrue(all('start_date=' + datetime.date.today().isoformat() in c for c in calls if '/performance' in c)) # days=0
-        self.assertEqual(sharesight.warm()[0], 0)
-        self.assertEqual(len(calls), 3) # nothing refetched while fresh
-
-    def test_refreshes_before_the_ttl_runs_out(self):
-        calls = self.serve_sharesight()
-        sharesight.warm()[0]
-        calls.clear()
-        self.age(sharesight.performance_cache_file(684141, 0), 0.95 * sharesight.config_performance_cache_seconds) # 95% of its lifetime
-        self.assertEqual(sharesight.warm()[0], 1)
-        self.assertEqual(len(calls), 1)
-        self.assertIn('684141/performance', calls[0])
-        calls.clear()
-        self.age(sharesight.PORTFOLIOS_CACHE, 0.95 * sharesight.config_cache_seconds)
-        self.assertEqual(sharesight.warm()[0], 1)
-        self.assertTrue(calls[0].endswith('/portfolios'))
-
-    def test_a_request_never_waits_on_a_cache_the_job_has_warmed(self):
-        calls = self.serve_sharesight()
-        sharesight.warm()[0]
-        calls.clear()
-        self.assertEqual(sharesight.get_portfolios(), {'Rob': 445825, 'RobSMSF': 684141})
-        self.assertEqual(len(sharesight.get_performance(684141, 0)['report']['holdings']), 18)
-        self.assertEqual(calls, [])
-
-    def test_holdings_wrapper_is_served_entirely_from_the_warmed_cache(self):
-        calls = self.serve_sharesight()
-        sharesight.warm()[0]
-        calls.clear()
-        tickers = sharesight.get_holdings_wrapper() # get_portfolios() + get_holdings() per portfolio, no cache of its own
-        self.assertIn('ARM', tickers)
-        self.assertEqual(calls, [])
-
-    def test_failed_refresh_keeps_the_old_cache(self):
-        self.serve_sharesight()
-        sharesight.warm()[0]
-        self.age(sharesight.PORTFOLIOS_CACHE, 0.95 * sharesight.config_cache_seconds)
-        self.http(sharesight, ConnectionError('down'))
-        with self.assertRaises(RuntimeError):
-            sharesight.warm()[0]
-        self.assertEqual(sharesight.get_portfolios(), {'Rob': 445825, 'RobSMSF': 684141}) # still served from the cache
-
-    def test_loop_survives_failures_and_backs_off(self):
-        class Stop:
-            def __init__(self): self.waits = []
-            def is_set(self): return len(self.waits) >= 2
-            def wait(self, seconds): self.waits.append(seconds)
-        outcomes = [RuntimeError('boom'), 0]
-        def warm():
-            outcome = outcomes.pop(0)
-            if isinstance(outcome, Exception):
-                raise outcome
-            return outcome, {}
-        self.patch(sharesight, 'warm', warm)
-        self.patch(sharesight, 'next_due', lambda portfolios=None: 3000)
-        stop = Stop()
-        sharesight.keep_warm(retry=300, stop=stop)
-        self.assertEqual(stop.waits, [300, 3000]) # slow retry after the failure, then sleep until the next cache is due
-
-    def test_with_cache_off_nothing_is_fetched_until_a_reload_turns_it_on(self):
-        class Stop:
-            def __init__(self): self.waits = []
-            def is_set(self): return len(self.waits) >= 4
-            def wait(self, seconds):
-                self.waits.append(seconds)
-                if len(self.waits) == 2:
-                    sharesight.config_cache = True # what a reload that switches it on does
-        passes = []
-        self.patch(sharesight, 'warm', lambda: (passes.append(1), (0, {}))[1])
-        self.patch(sharesight, 'next_due', lambda portfolios=None: 3000)
-        self.patch(sharesight, 'config_cache', False)
-        stop = Stop()
-        sharesight.keep_warm(idle=60, stop=stop)
-        self.assertEqual((stop.waits, len(passes)), ([60, 60, 3000, 3000], 2)) # idle while off: no Sharesight request every 5 seconds
-
-    def test_a_flush_wakes_the_loop_to_refill_at_once(self):
-        passes = []
-        self.patch(sharesight, 'warm', lambda: (passes.append(1), (0, {}))[1])
-        self.patch(sharesight, 'next_due', lambda portfolios=None: 3000)
-        loop = gevent.spawn(sharesight.keep_warm)
-        self.addCleanup(loop.kill)
-        gevent.sleep(0.05)
-        self.assertEqual(len(passes), 1) # then it sleeps until the next cache is due, 50 minutes away
-        sharesight.wake()
-        gevent.sleep(0.05)
-        self.assertEqual(len(passes), 2)
-
-    def test_a_warming_pass_reads_the_cached_portfolio_list_once(self):
-        self.serve_sharesight()
-        sharesight.warm()[0] # fill the caches
-        reads = []
-        real = util.read_cache
-        self.patch(util, 'read_cache', lambda name, *a, **k: (reads.append(name), real(name, *a, **k))[1])
-        fetched, portfolios = sharesight.warm() # what one pass of keep_warm does
-        due = sharesight.next_due(portfolios=portfolios)
-        self.assertEqual((fetched, reads), (0, [sharesight.PORTFOLIOS_CACHE]))
-        self.assertAlmostEqual(due, sharesight.next_due(), delta=1) # the same answer as reading the list again
-        self.assertEqual(portfolios, {'Rob': 445825, 'RobSMSF': 684141})
-
-    def test_next_due_is_90_percent_of_the_ttl(self):
-        self.serve_sharesight()
-        self.assertEqual(sharesight.next_due(), 0) # cold cache
-        sharesight.warm()[0]
-        due = sharesight.next_due()
-        self.assertAlmostEqual(due, 0.9 * min(sharesight.config_cache_seconds, sharesight.config_performance_cache_seconds), delta=5)
-
-    def test_refresh_flag_bypasses_a_fresh_cache(self):
-        calls = self.serve_sharesight()
-        sharesight.get_portfolios()
-        sharesight.get_portfolios()
-        self.assertEqual(len(calls), 1)
-        sharesight.get_portfolios(refresh=True)
-        self.assertEqual(len(calls), 2)
+	"""The background refresher keeps the portfolio list and the holdings report (days=0) cached ahead of their TTLs."""
+
+	def serve_sharesight(self):
+		portfolios = sample('sharesight', 'finbot_sharesight_portfolios.json')
+		performance = sample('sharesight', 'finbot_sharesight_performance_684141_0.json')
+		calls = []
+		def get(url, **kwargs):
+			calls.append(url)
+			return Resp(performance if '/performance' in url else portfolios)
+		self.patch(sharesight.net, 'get', get)
+		return calls
+
+	def age(self, cache_file, seconds):
+		path = os.path.join(util.config_cache_dir, cache_file)
+		old = os.path.getmtime(path) - seconds
+		os.utime(path, (old, old))
+
+	def test_cold_cache_is_filled_then_left_alone(self):
+		calls = self.serve_sharesight()
+		self.assertEqual(sharesight.warm()[0], 3) # portfolio list + one holdings report per portfolio (2)
+		self.assertEqual(sum('/performance' in c for c in calls), 2)
+		self.assertTrue(all('start_date=' + datetime.date.today().isoformat() in c for c in calls if '/performance' in c)) # days=0
+		self.assertEqual(sharesight.warm()[0], 0)
+		self.assertEqual(len(calls), 3) # nothing refetched while fresh
+
+	def test_refreshes_before_the_ttl_runs_out(self):
+		calls = self.serve_sharesight()
+		sharesight.warm()[0]
+		calls.clear()
+		self.age(sharesight.performance_cache_file(684141, 0), 0.95 * sharesight.config_performance_cache_seconds) # 95% of its lifetime
+		self.assertEqual(sharesight.warm()[0], 1)
+		self.assertEqual(len(calls), 1)
+		self.assertIn('684141/performance', calls[0])
+		calls.clear()
+		self.age(sharesight.PORTFOLIOS_CACHE, 0.95 * sharesight.config_cache_seconds)
+		self.assertEqual(sharesight.warm()[0], 1)
+		self.assertTrue(calls[0].endswith('/portfolios'))
+
+	def test_a_request_never_waits_on_a_cache_the_job_has_warmed(self):
+		calls = self.serve_sharesight()
+		sharesight.warm()[0]
+		calls.clear()
+		self.assertEqual(sharesight.get_portfolios(), {'Rob': 445825, 'RobSMSF': 684141})
+		self.assertEqual(len(sharesight.get_performance(684141, 0)['report']['holdings']), 18)
+		self.assertEqual(calls, [])
+
+	def test_holdings_wrapper_is_served_entirely_from_the_warmed_cache(self):
+		calls = self.serve_sharesight()
+		sharesight.warm()[0]
+		calls.clear()
+		tickers = sharesight.get_holdings_wrapper() # get_portfolios() + get_holdings() per portfolio, no cache of its own
+		self.assertIn('ARM', tickers)
+		self.assertEqual(calls, [])
+
+	def test_failed_refresh_keeps_the_old_cache(self):
+		self.serve_sharesight()
+		sharesight.warm()[0]
+		self.age(sharesight.PORTFOLIOS_CACHE, 0.95 * sharesight.config_cache_seconds)
+		self.http(sharesight, ConnectionError('down'))
+		with self.assertRaises(RuntimeError):
+			sharesight.warm()[0]
+		self.assertEqual(sharesight.get_portfolios(), {'Rob': 445825, 'RobSMSF': 684141}) # still served from the cache
+
+	def test_loop_survives_failures_and_backs_off(self):
+		class Stop:
+			def __init__(self): self.waits = []
+			def is_set(self): return len(self.waits) >= 2
+			def wait(self, seconds): self.waits.append(seconds)
+		outcomes = [RuntimeError('boom'), 0]
+		def warm():
+			outcome = outcomes.pop(0)
+			if isinstance(outcome, Exception):
+				raise outcome
+			return outcome, {}
+		self.patch(sharesight, 'warm', warm)
+		self.patch(sharesight, 'next_due', lambda portfolios=None: 3000)
+		stop = Stop()
+		sharesight.keep_warm(retry=300, stop=stop)
+		self.assertEqual(stop.waits, [300, 3000]) # slow retry after the failure, then sleep until the next cache is due
+
+	def test_with_cache_off_nothing_is_fetched_until_a_reload_turns_it_on(self):
+		class Stop:
+			def __init__(self): self.waits = []
+			def is_set(self): return len(self.waits) >= 4
+			def wait(self, seconds):
+				self.waits.append(seconds)
+				if len(self.waits) == 2:
+					sharesight.config_cache = True # what a reload that switches it on does
+		passes = []
+		self.patch(sharesight, 'warm', lambda: (passes.append(1), (0, {}))[1])
+		self.patch(sharesight, 'next_due', lambda portfolios=None: 3000)
+		self.patch(sharesight, 'config_cache', False)
+		stop = Stop()
+		sharesight.keep_warm(idle=60, stop=stop)
+		self.assertEqual((stop.waits, len(passes)), ([60, 60, 3000, 3000], 2)) # idle while off: no Sharesight request every 5 seconds
+
+	def test_a_flush_wakes_the_loop_to_refill_at_once(self):
+		passes = []
+		self.patch(sharesight, 'warm', lambda: (passes.append(1), (0, {}))[1])
+		self.patch(sharesight, 'next_due', lambda portfolios=None: 3000)
+		loop = gevent.spawn(sharesight.keep_warm)
+		self.addCleanup(loop.kill)
+		gevent.sleep(0.05)
+		self.assertEqual(len(passes), 1) # then it sleeps until the next cache is due, 50 minutes away
+		sharesight.wake()
+		gevent.sleep(0.05)
+		self.assertEqual(len(passes), 2)
+
+	def test_a_warming_pass_reads_the_cached_portfolio_list_once(self):
+		self.serve_sharesight()
+		sharesight.warm()[0] # fill the caches
+		reads = []
+		real = util.read_cache
+		self.patch(util, 'read_cache', lambda name, *a, **k: (reads.append(name), real(name, *a, **k))[1])
+		fetched, portfolios = sharesight.warm() # what one pass of keep_warm does
+		due = sharesight.next_due(portfolios=portfolios)
+		self.assertEqual((fetched, reads), (0, [sharesight.PORTFOLIOS_CACHE]))
+		self.assertAlmostEqual(due, sharesight.next_due(), delta=1) # the same answer as reading the list again
+		self.assertEqual(portfolios, {'Rob': 445825, 'RobSMSF': 684141})
+
+	def test_next_due_is_90_percent_of_the_ttl(self):
+		self.serve_sharesight()
+		self.assertEqual(sharesight.next_due(), 0) # cold cache
+		sharesight.warm()[0]
+		due = sharesight.next_due()
+		self.assertAlmostEqual(due, 0.9 * min(sharesight.config_cache_seconds, sharesight.config_performance_cache_seconds), delta=5)
+
+	def test_refresh_flag_bypasses_a_fresh_cache(self):
+		calls = self.serve_sharesight()
+		sharesight.get_portfolios()
+		sharesight.get_portfolios()
+		self.assertEqual(len(calls), 1)
+		sharesight.get_portfolios(refresh=True)
+		self.assertEqual(len(calls), 2)
 
 
 class TelegramSending(FinbotCase):
-    """webhook.write against failing endpoints. The real send path, not the base fixture's recorder."""
-
-    def post(self, outcome):
-        def fake(url, **kwargs):
-            if isinstance(outcome, Exception):
-                raise outcome
-            return outcome
-        self.patch(webhook.net, 'post', fake)
-        return webhook.write('telegram', 'https://api.telegram.org/botX/sendMessage?chat_id=5', 'hi', '5')
-
-    def test_success_returns_json(self):
-        self.assertEqual(self.post(Resp({'ok': True})), {'ok': True})
-
-    def test_http_error_and_exception_return_none(self):
-        self.assertIsNone(self.post(Resp(status=429)))
-        self.assertIsNone(self.post(ConnectionError('down')))
-
-    def test_non_json_success_returns_empty_dict(self):
-        self.assertEqual(self.post(Resp(None)), {})
-
-    def test_error_line_redacts_secrets_and_truncates(self):
-        line = webhook.error_line(RuntimeError('GET /bot123456:ABC-def_1/send?crumb=SECRET&x=1 failed'), 'ctx')
-        self.assertNotIn('ABC-def_1', line)
-        self.assertNotIn('SECRET', line)
-        self.assertTrue(line.startswith('⚠️ ctx: '))
-        self.assertLessEqual(len(webhook.error_line('x' * 1000)), 205)
-
-    def test_error_reports_are_escaped_for_telegram_and_slack(self):
-        sent = []
-        self.patch(webhook, 'write', lambda service, url, text, *a, **k: sent.append((service, text)))
-        self.patch(webhook, 'webhooks', {'telegram': 'https://api.telegram.org/botX/', 'slack': 'https://hooks.slack.com/x', 'discord': 'https://discord.com/x/slack'})
-        self.patch(webhook, 'config_telegramChatID', '5')
-        error = RuntimeError("404 for ZZ at https://x/?a=1&crumb=abc; '<' not supported between <urllib3.connection>")
-        with contextlib.redirect_stderr(io.StringIO()):
-            REAL_REPORT_ERROR(error, context='.compare <b>')
-            REAL_REPORT_ERROR(error, service='telegram', chat_id='55', context='.compare')
-        texts = dict(sent[:3])
-        self.assertIn('crumb=&lt;redacted&gt;', texts['telegram']) # Telegram's HTML parser rejects a bare <redacted>
-        self.assertNotIn('<', texts['telegram'] + sent[3][1])
-        self.assertIn('.compare &lt;b&gt;: RuntimeError', texts['telegram'])
-        self.assertEqual(texts['slack'], texts['telegram'])
-        self.assertIn('crumb=<redacted>', texts['discord'])
-
-    def test_report_error_never_raises(self):
-        self.patch(webhook, 'write', mock.Mock(side_effect=RuntimeError('send failed')))
-        webhook.report_error(ValueError('boom'), service='telegram', chat_id='5')
-        webhook.report_error(ValueError('boom'))
-
-    def test_guarded_reports_and_exits_nonzero(self):
-        reported = []
-        self.patch(webhook, 'report_error', lambda e, **k: reported.append(e))
-        with self.assertRaises(SystemExit) as raised:
-            webhook.guarded('job', lambda: 1 / 0)
-        self.assertEqual(raised.exception.code, 1)
-        self.assertEqual(len(reported), 1)
-        self.assertEqual(webhook.guarded('job', lambda: 'fine'), 'fine')
+	"""webhook.write against failing endpoints. The real send path, not the base fixture's recorder."""
+
+	def post(self, outcome):
+		def fake(url, **kwargs):
+			if isinstance(outcome, Exception):
+				raise outcome
+			return outcome
+		self.patch(webhook.net, 'post', fake)
+		return webhook.write('telegram', 'https://api.telegram.org/botX/sendMessage?chat_id=5', 'hi', '5')
+
+	def test_success_returns_json(self):
+		self.assertEqual(self.post(Resp({'ok': True})), {'ok': True})
+
+	def test_http_error_and_exception_return_none(self):
+		self.assertIsNone(self.post(Resp(status=429)))
+		self.assertIsNone(self.post(ConnectionError('down')))
+
+	def test_non_json_success_returns_empty_dict(self):
+		self.assertEqual(self.post(Resp(None)), {})
+
+	def test_error_line_redacts_secrets_and_truncates(self):
+		line = webhook.error_line(RuntimeError('GET /bot123456:ABC-def_1/send?crumb=SECRET&x=1 failed'), 'ctx')
+		self.assertNotIn('ABC-def_1', line)
+		self.assertNotIn('SECRET', line)
+		self.assertTrue(line.startswith('⚠️ ctx: '))
+		self.assertLessEqual(len(webhook.error_line('x' * 1000)), 205)
+
+	def test_error_reports_are_escaped_for_telegram_and_slack(self):
+		sent = []
+		self.patch(webhook, 'write', lambda service, url, text, *a, **k: sent.append((service, text)))
+		self.patch(webhook, 'webhooks', {'telegram': 'https://api.telegram.org/botX/', 'slack': 'https://hooks.slack.com/x', 'discord': 'https://discord.com/x/slack'})
+		self.patch(webhook, 'config_telegramChatID', '5')
+		error = RuntimeError("404 for ZZ at https://x/?a=1&crumb=abc; '<' not supported between <urllib3.connection>")
+		with contextlib.redirect_stderr(io.StringIO()):
+			REAL_REPORT_ERROR(error, context='.compare <b>')
+			REAL_REPORT_ERROR(error, service='telegram', chat_id='55', context='.compare')
+		texts = dict(sent[:3])
+		self.assertIn('crumb=&lt;redacted&gt;', texts['telegram']) # Telegram's HTML parser rejects a bare <redacted>
+		self.assertNotIn('<', texts['telegram'] + sent[3][1])
+		self.assertIn('.compare &lt;b&gt;: RuntimeError', texts['telegram'])
+		self.assertEqual(texts['slack'], texts['telegram'])
+		self.assertIn('crumb=<redacted>', texts['discord'])
+
+	def test_report_error_never_raises(self):
+		self.patch(webhook, 'write', mock.Mock(side_effect=RuntimeError('send failed')))
+		webhook.report_error(ValueError('boom'), service='telegram', chat_id='5')
+		webhook.report_error(ValueError('boom'))
+
+	def test_guarded_reports_and_exits_nonzero(self):
+		reported = []
+		self.patch(webhook, 'report_error', lambda e, **k: reported.append(e))
+		with self.assertRaises(SystemExit) as raised:
+			webhook.guarded('job', lambda: 1 / 0)
+		self.assertEqual(raised.exception.code, 1)
+		self.assertEqual(len(reported), 1)
+		self.assertEqual(webhook.guarded('job', lambda: 'fine'), 'fine')
 
 
 class ChartDelivery(FinbotCase):
-    """Charts go to Telegram, and to Slack in replies (uploaded into the channel). A report whose chart cannot be sent goes out as
-    text instead of vanishing."""
-    UPLOAD_URL = 'https://files.slack.com/upload/v1/abc'
-
-    def setUp(self):
-        super().setUp()
-        self.patch(webhook, 'sendPhoto', REAL_SEND_PHOTO)
-
-    def slack_answers(self, *outcomes):
-        """net.post for the upload steps: the outcomes in turn (an exception is raised). Returns the calls made."""
-        calls = []
-        def post(url, **kwargs):
-            calls.append((url, kwargs))
-            outcome = outcomes[len(calls) - 1]
-            if isinstance(outcome, Exception):
-                raise outcome
-            return outcome
-        self.patch(webhook.net, 'post', post)
-        return calls
-
-    def upload_steps(self):
-        return [Resp({'ok': True, 'upload_url': self.UPLOAD_URL, 'file_id': 'F1'}), Resp(None, text='OK - 3'), Resp({'ok': True, 'files': [{'id': 'F1'}]})]
-
-    def test_where_charts_can_go(self):
-        self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
-        cases = (('telegram',), ('telegram', False), ('slack',), ('slack', False), ('discord', False))
-        self.assertEqual([webhook.takes_charts(*case) for case in cases], [True, True, True, False, True]) # Slack: replies, not scheduled reports
-        self.patch(webhook, 'config_slackBotToken', False)
-        self.assertFalse(webhook.takes_charts('slack'))
-
-    def test_slack_replies_upload_their_charts(self):
-        self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
-        for cmd, heading in (('.beta', '*Beta over 1.5'), ('.price', '*5% at close'), ('.compare T1 T2', '*')):
-            with self.subTest(cmd=cmd):
-                calls = self.slack_answers(*self.upload_steps())
-                self.sent.clear(); self.errors.clear()
-                worker.process_request('slack', 'C1', '<@U1>', cmd, BOT, 'U')
-                self.assertEqual(self.errors, [])
-                self.assertEqual([url for url, _ in calls], ['https://slack.com/api/files.getUploadURLExternal', self.UPLOAD_URL,
-                                                             'https://slack.com/api/files.completeUploadExternal'])
-                (_, ticket), (_, upload), (_, share) = calls
-                self.assertEqual(ticket['headers'], {'Authorization': 'Bearer xoxb-test'})
-                self.assertTrue(upload['data'].startswith(b'\x89PNG')) # the chart's bytes, as the body
-                self.assertEqual((ticket['data']['filename'], ticket['data']['length']), ('chart.png', len(upload['data'])))
-                self.assertEqual((share['data']['channel_id'], json.loads(share['data']['files'])), ('C1', [{'id': 'F1'}]))
-                self.assertTrue(share['data']['initial_comment'].startswith(heading), share['data']['initial_comment'][:80]) # the report
-                self.assertNotIn(share['data']['initial_comment'].split('\n'), [payload for _, payload in self.sent]) # not sent again as text
-
-    def test_a_slack_upload_that_fails_goes_out_as_text(self):
-        self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
-        good = self.upload_steps()
-        text = [('text', ['*Heading*', 'line 2'])]
-        for outcomes, expect in (([Resp({'ok': False, 'error': 'missing_scope'})], text), ([ConnectionError('down')], text),
-                                 ([good[0], Resp(None, status=500, text='nope')], text),
-                                 ([good[0], good[1], Resp({'ok': False, 'error': 'not_in_channel'})], text),
-                                 ([good[0], good[1], webhook.requests.exceptions.ReadTimeout('slow')], []), # it may have been shared after all
-                                 (good, [])):
-            with self.subTest(outcomes=outcomes):
-                self.slack_answers(*outcomes)
-                self.sent.clear()
-                with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
-                    webhook.sendPhoto('C1', io.BytesIO(b'png'), '*Heading*\nline 2', 'slack')
-                self.assertEqual(self.sent, expect)
-
-    def test_scheduled_reports_to_slack_stay_text(self):
-        self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
-        webhook.webhooks['slack'] = 'https://hooks.slack.com/services/x' # an incoming webhook cannot upload
-        sent = []
-        self.patch(webhook, 'sendPhoto', lambda chat, image, caption, service, **k: sent.append((service, 'chart')))
-        self.patch(webhook, 'payload_wrapper', lambda service, url, payload, *a, **k: sent.append((service, 'text')) or [])
-        for report in (price.lambda_handler, lambda: performance.lambda_handler(past_days=7), shorts.lambda_handler):
-            with self.subTest(report=report):
-                sent.clear()
-                report()
-                self.assertEqual(sorted(set(sent)), [('slack', 'text'), ('telegram', 'chart')])
-
-    def test_slack_without_a_bot_token_gets_text_and_no_chart_is_drawn(self):
-        self.patch(webhook, 'config_slackBotToken', False)
-        def refuse(*a, **k):
-            raise AssertionError('a chart was drawn for Slack')
-        for name in ('rows_chart', 'pie_chart', 'bar_graph', 'shorts_chart'):
-            self.patch(util, name, refuse)
-        webhook.webhooks['slack'] = 'https://hooks.slack.com/services/x'
-        for cmd in ('.beta', '.pie Rob', '.buy', '.sell', '.peg', '.price', '.price top', '.price T1 7d', '.performance 7d', '.shorts', '.history T1', '.compare T1 T2'):
-            with self.subTest(cmd=cmd):
-                self.sent.clear(); self.errors.clear()
-                worker.process_request('slack', 'C1', '<@U1>', cmd, BOT, 'U')
-                self.assertEqual(self.errors, [])
-                self.assertTrue(self.sent)
-                self.assertEqual({kind for kind, *_ in self.sent}, {'text'})
-
-    def test_the_graph_preference_off_sends_no_chart_anywhere(self):
-        """graph=False means no charts, as the README says; it used to hold back only the price-line charts."""
-        self.patch(webhook, 'config_graph', False)
-        self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
-        posts = []
-        self.patch(webhook.net, 'post', lambda url, **k: posts.append(url) or Resp({'ok': True, 'result': True}))
-        for service, chat in (('telegram', '55'), ('slack', 'C1')):
-            for cmd in ('.price', '.price top', '.performance 7d', '.shorts', '.buy', '.peg', '.beta', '.compare T1 T2', '.pie Rob'):
-                with self.subTest(service=service, cmd=cmd):
-                    self.sent.clear(); self.errors.clear()
-                    worker.process_request(service, chat, '@u', cmd, BOT, 'U')
-                    self.assertEqual(self.errors, [])
-                    self.assertEqual({kind for kind, *_ in self.sent}, {'text'})
-        self.assertEqual([url for url in posts if 'sendPhoto' in url or 'files.' in url], [])
-
-    def test_a_chart_telegram_refuses_goes_out_as_text(self):
-        def post(outcome):
-            def fake(url, **kwargs):
-                if isinstance(outcome, Exception):
-                    raise outcome
-                return outcome
-            return fake
-        text = [('text', ['Heading', 'line 2'])]
-        for outcome, expect in ((Resp({'ok': False, 'error_code': 400, 'description': "Bad Request: can't parse entities"}, status=400), text),
-                                (Resp(None, status=502), text), (ConnectionError('down'), text),
-                                (webhook.requests.exceptions.ReadTimeout('slow'), []), # it may have arrived after all: no duplicate
-                                (Resp({'ok': True, 'result': {'message_id': 1}}), [])):
-            with self.subTest(outcome=outcome):
-                self.sent.clear()
-                self.patch(webhook.net, 'post', post(outcome))
-                with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
-                    webhook.sendPhoto('5', io.BytesIO(b'png'), 'Heading\nline 2', 'telegram', reply_markup={'inline_keyboard': []})
-                self.assertEqual(self.sent, expect)
+	"""Charts go to Telegram, and to Slack in replies (uploaded into the channel). A report whose chart cannot be sent goes out as
+	text instead of vanishing."""
+	UPLOAD_URL = 'https://files.slack.com/upload/v1/abc'
+
+	def setUp(self):
+		super().setUp()
+		self.patch(webhook, 'sendPhoto', REAL_SEND_PHOTO)
+
+	def slack_answers(self, *outcomes):
+		"""net.post for the upload steps: the outcomes in turn (an exception is raised). Returns the calls made."""
+		calls = []
+		def post(url, **kwargs):
+			calls.append((url, kwargs))
+			outcome = outcomes[len(calls) - 1]
+			if isinstance(outcome, Exception):
+				raise outcome
+			return outcome
+		self.patch(webhook.net, 'post', post)
+		return calls
+
+	def upload_steps(self):
+		return [Resp({'ok': True, 'upload_url': self.UPLOAD_URL, 'file_id': 'F1'}), Resp(None, text='OK - 3'), Resp({'ok': True, 'files': [{'id': 'F1'}]})]
+
+	def test_where_charts_can_go(self):
+		self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
+		cases = (('telegram',), ('telegram', False), ('slack',), ('slack', False), ('discord', False))
+		self.assertEqual([webhook.takes_charts(*case) for case in cases], [True, True, True, False, True]) # Slack: replies, not scheduled reports
+		self.patch(webhook, 'config_slackBotToken', False)
+		self.assertFalse(webhook.takes_charts('slack'))
+
+	def test_slack_replies_upload_their_charts(self):
+		self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
+		for cmd, heading in (('.beta', '*Beta over 1.5'), ('.price', '*5% at close'), ('.compare T1 T2', '*')):
+			with self.subTest(cmd=cmd):
+				calls = self.slack_answers(*self.upload_steps())
+				self.sent.clear(); self.errors.clear()
+				worker.process_request('slack', 'C1', '<@U1>', cmd, BOT, 'U')
+				self.assertEqual(self.errors, [])
+				self.assertEqual([url for url, _ in calls], ['https://slack.com/api/files.getUploadURLExternal', self.UPLOAD_URL,
+															 'https://slack.com/api/files.completeUploadExternal'])
+				(_, ticket), (_, upload), (_, share) = calls
+				self.assertEqual(ticket['headers'], {'Authorization': 'Bearer xoxb-test'})
+				self.assertTrue(upload['data'].startswith(b'\x89PNG')) # the chart's bytes, as the body
+				self.assertEqual((ticket['data']['filename'], ticket['data']['length']), ('chart.png', len(upload['data'])))
+				self.assertEqual((share['data']['channel_id'], json.loads(share['data']['files'])), ('C1', [{'id': 'F1'}]))
+				self.assertTrue(share['data']['initial_comment'].startswith(heading), share['data']['initial_comment'][:80]) # the report
+				self.assertNotIn(share['data']['initial_comment'].split('\n'), [payload for _, payload in self.sent]) # not sent again as text
+
+	def test_a_slack_upload_that_fails_goes_out_as_text(self):
+		self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
+		good = self.upload_steps()
+		text = [('text', ['*Heading*', 'line 2'])]
+		for outcomes, expect in (([Resp({'ok': False, 'error': 'missing_scope'})], text), ([ConnectionError('down')], text),
+								 ([good[0], Resp(None, status=500, text='nope')], text),
+								 ([good[0], good[1], Resp({'ok': False, 'error': 'not_in_channel'})], text),
+								 ([good[0], good[1], webhook.requests.exceptions.ReadTimeout('slow')], []), # it may have been shared after all
+								 (good, [])):
+			with self.subTest(outcomes=outcomes):
+				self.slack_answers(*outcomes)
+				self.sent.clear()
+				with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
+					webhook.sendPhoto('C1', io.BytesIO(b'png'), '*Heading*\nline 2', 'slack')
+				self.assertEqual(self.sent, expect)
+
+	def test_scheduled_reports_to_slack_stay_text(self):
+		self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
+		webhook.webhooks['slack'] = 'https://hooks.slack.com/services/x' # an incoming webhook cannot upload
+		sent = []
+		self.patch(webhook, 'sendPhoto', lambda chat, image, caption, service, **k: sent.append((service, 'chart')))
+		self.patch(webhook, 'payload_wrapper', lambda service, url, payload, *a, **k: sent.append((service, 'text')) or [])
+		for report in (price.lambda_handler, lambda: performance.lambda_handler(past_days=7), shorts.lambda_handler):
+			with self.subTest(report=report):
+				sent.clear()
+				report()
+				self.assertEqual(sorted(set(sent)), [('slack', 'text'), ('telegram', 'chart')])
+
+	def test_slack_without_a_bot_token_gets_text_and_no_chart_is_drawn(self):
+		self.patch(webhook, 'config_slackBotToken', False)
+		def refuse(*a, **k):
+			raise AssertionError('a chart was drawn for Slack')
+		for name in ('rows_chart', 'pie_chart', 'bar_graph', 'shorts_chart'):
+			self.patch(util, name, refuse)
+		webhook.webhooks['slack'] = 'https://hooks.slack.com/services/x'
+		for cmd in ('.beta', '.pie Rob', '.buy', '.sell', '.peg', '.price', '.price top', '.price T1 7d', '.performance 7d', '.shorts', '.history T1', '.compare T1 T2'):
+			with self.subTest(cmd=cmd):
+				self.sent.clear(); self.errors.clear()
+				worker.process_request('slack', 'C1', '<@U1>', cmd, BOT, 'U')
+				self.assertEqual(self.errors, [])
+				self.assertTrue(self.sent)
+				self.assertEqual({kind for kind, *_ in self.sent}, {'text'})
+
+	def test_the_graph_preference_off_sends_no_chart_anywhere(self):
+		"""graph=False means no charts, as the README says; it used to hold back only the price-line charts."""
+		self.patch(webhook, 'config_graph', False)
+		self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
+		posts = []
+		self.patch(webhook.net, 'post', lambda url, **k: posts.append(url) or Resp({'ok': True, 'result': True}))
+		for service, chat in (('telegram', '55'), ('slack', 'C1')):
+			for cmd in ('.price', '.price top', '.performance 7d', '.shorts', '.buy', '.peg', '.beta', '.compare T1 T2', '.pie Rob'):
+				with self.subTest(service=service, cmd=cmd):
+					self.sent.clear(); self.errors.clear()
+					worker.process_request(service, chat, '@u', cmd, BOT, 'U')
+					self.assertEqual(self.errors, [])
+					self.assertEqual({kind for kind, *_ in self.sent}, {'text'})
+		self.assertEqual([url for url in posts if 'sendPhoto' in url or 'files.' in url], [])
+
+	def test_a_chart_telegram_refuses_goes_out_as_text(self):
+		def post(outcome):
+			def fake(url, **kwargs):
+				if isinstance(outcome, Exception):
+					raise outcome
+				return outcome
+			return fake
+		text = [('text', ['Heading', 'line 2'])]
+		for outcome, expect in ((Resp({'ok': False, 'error_code': 400, 'description': "Bad Request: can't parse entities"}, status=400), text),
+								(Resp(None, status=502), text), (ConnectionError('down'), text),
+								(webhook.requests.exceptions.ReadTimeout('slow'), []), # it may have arrived after all: no duplicate
+								(Resp({'ok': True, 'result': {'message_id': 1}}), [])):
+			with self.subTest(outcome=outcome):
+				self.sent.clear()
+				self.patch(webhook.net, 'post', post(outcome))
+				with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
+					webhook.sendPhoto('5', io.BytesIO(b'png'), 'Heading\nline 2', 'telegram', reply_markup={'inline_keyboard': []})
+				self.assertEqual(self.sent, expect)
 
 
 class DiscordWebhook(FinbotCase):
-    """Discord's own webhook endpoint (the Slack-compatible one takes no flags): no link previews, no pings, charts attached."""
-    URL = 'https://discord.com/api/webhooks/1/abc'
-
-    def setUp(self):
-        super().setUp()
-        self.patch(webhook, 'webhooks', {'discord': self.URL})
-        self.patch(webhook, 'payload_wrapper', REAL_PAYLOAD_WRAPPER)
-        self.patch(webhook, 'sendPhoto', REAL_SEND_PHOTO)
-        self.posts = []
-        self.patch(webhook.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or Resp({'id': '1'}))
-
-    def test_text_goes_to_the_native_endpoint_without_previews_or_pings(self):
-        webhook.write('discord', self.URL, 'hello @everyone')
-        url, kwargs = self.posts[0]
-        self.assertEqual(url, self.URL + '?wait=true') # without wait Discord answers 204, which counted as a failure
-        self.assertEqual(kwargs['json'], {'content': 'hello @everyone', 'flags': 4, 'allowed_mentions': {'parse': []}})
-
-    def test_a_scheduled_report_carries_its_chart(self):
-        price.lambda_handler()
-        (url, kwargs), = self.posts
-        self.assertEqual(url, self.URL + '?wait=true')
-        payload = json.loads(kwargs['data']['payload_json'])
-        self.assertEqual((payload['flags'], payload['allowed_mentions'], payload['attachments'][0]['filename']), (4, {'parse': []}, 'chart.png'))
-        self.assertTrue(payload['content'].startswith('**5% at close'), payload['content'][:60])
-        self.assertLessEqual(len(payload['content']), 2000)
-        name, image, kind = kwargs['files']['files[0]']
-        self.assertEqual((name, kind), ('chart.png', 'image/png'))
-        self.assertTrue(image.read().startswith(b'\x89PNG'))
-
-    def test_a_long_report_is_split_by_size(self):
-        line = 'x' * 150
-        webhook.payload_wrapper('discord', self.URL, [line] * 30)
-        sizes = [len(kwargs['json']['content']) for _, kwargs in self.posts]
-        self.assertGreater(len(sizes), 2)
-        self.assertTrue(all(size <= 2000 for size in sizes), sizes) # 20 lines a chunk used to make 3,019 characters
-
-    def test_links_are_masked_links(self):
-        self.assertEqual(util.link('https://x.example/a(b)', 'T[1]', 'discord'), '[T\\[1\\]](<https://x.example/a%28b%29>)')
+	"""Discord's own webhook endpoint (the Slack-compatible one takes no flags): no link previews, no pings, charts attached."""
+	URL = 'https://discord.com/api/webhooks/1/abc'
+
+	def setUp(self):
+		super().setUp()
+		self.patch(webhook, 'webhooks', {'discord': self.URL})
+		self.patch(webhook, 'payload_wrapper', REAL_PAYLOAD_WRAPPER)
+		self.patch(webhook, 'sendPhoto', REAL_SEND_PHOTO)
+		self.posts = []
+		self.patch(webhook.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or Resp({'id': '1'}))
+
+	def test_text_goes_to_the_native_endpoint_without_previews_or_pings(self):
+		webhook.write('discord', self.URL, 'hello @everyone')
+		url, kwargs = self.posts[0]
+		self.assertEqual(url, self.URL + '?wait=true') # without wait Discord answers 204, which counted as a failure
+		self.assertEqual(kwargs['json'], {'content': 'hello @everyone', 'flags': 4, 'allowed_mentions': {'parse': []}})
+
+	def test_a_scheduled_report_carries_its_chart(self):
+		price.lambda_handler()
+		(url, kwargs), = self.posts
+		self.assertEqual(url, self.URL + '?wait=true')
+		payload = json.loads(kwargs['data']['payload_json'])
+		self.assertEqual((payload['flags'], payload['allowed_mentions'], payload['attachments'][0]['filename']), (4, {'parse': []}, 'chart.png'))
+		self.assertTrue(payload['content'].startswith('**5% at close'), payload['content'][:60])
+		self.assertLessEqual(len(payload['content']), 2000)
+		name, image, kind = kwargs['files']['files[0]']
+		self.assertEqual((name, kind), ('chart.png', 'image/png'))
+		self.assertTrue(image.read().startswith(b'\x89PNG'))
+
+	def test_a_long_report_is_split_by_size(self):
+		line = 'x' * 150
+		webhook.payload_wrapper('discord', self.URL, [line] * 30)
+		sizes = [len(kwargs['json']['content']) for _, kwargs in self.posts]
+		self.assertGreater(len(sizes), 2)
+		self.assertTrue(all(size <= 2000 for size in sizes), sizes) # 20 lines a chunk used to make 3,019 characters
+
+	def test_links_are_masked_links(self):
+		self.assertEqual(util.link('https://x.example/a(b)', 'T[1]', 'discord'), '[T\\[1\\]](<https://x.example/a%28b%29>)')
 
 
 class SlackSending(FinbotCase):
-    """Slack replies go through chat.postMessage with the bot token; scheduled reports through the incoming webhook."""
-
-    def setUp(self):
-        super().setUp()
-        self.patch(webhook, 'payload_wrapper', REAL_PAYLOAD_WRAPPER)
-        self.patch(webhook, 'sendPhoto', REAL_SEND_PHOTO)
-        self.posts = []
-        self.patch(webhook.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or Resp({'ok': True}))
-
-    def test_link_previews_are_turned_off_in_the_body(self):
-        self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
-        webhook.write('slack', 'https://slack.com/api/chat.postMessage', 'hi', 'C1')
-        webhook.write('slack', 'https://hooks.slack.com/services/T/B/x', 'hi')
-        self.assertEqual(len(self.posts), 2)
-        for url, kwargs in self.posts:
-            with self.subTest(url=url): # they used to go in the headers, where Slack does not look
-                self.assertEqual((kwargs['json']['unfurl_links'], kwargs['json']['unfurl_media']), (False, False))
-                self.assertNotIn('unfurl_links', kwargs['headers'])
-
-    def test_a_reply_without_a_bot_token_is_logged_not_sent(self):
-        self.patch(webhook, 'config_slackBotToken', False)
-        err = io.StringIO()
-        with contextlib.redirect_stderr(err), contextlib.redirect_stdout(io.StringIO()):
-            for cmd in ('.help', '.price', '.compare T1 T2'):
-                worker.process_request('slack', 'C1', '<@U1>', cmd, BOT, 'U')
-        self.assertEqual((self.posts, self.errors), ([], []))
-        self.assertEqual(err.getvalue().count('ERROR slack: a reply to channel C1 needs slackBotToken in finbot.env; not sent'), 3)
-
-    def test_the_slack_api_says_what_is_missing(self):
-        self.patch(webhook, 'config_slackBotToken', False)
-        with self.assertRaisesRegex(RuntimeError, 'Slack files.getUploadURLExternal: needs slackBotToken in finbot.env'):
-            webhook.slack_api('files.getUploadURLExternal', filename='chart.png', length=1) # it used to be a TypeError
-        self.assertEqual(self.posts, [])
+	"""Slack replies go through chat.postMessage with the bot token; scheduled reports through the incoming webhook."""
+
+	def setUp(self):
+		super().setUp()
+		self.patch(webhook, 'payload_wrapper', REAL_PAYLOAD_WRAPPER)
+		self.patch(webhook, 'sendPhoto', REAL_SEND_PHOTO)
+		self.posts = []
+		self.patch(webhook.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or Resp({'ok': True}))
+
+	def test_link_previews_are_turned_off_in_the_body(self):
+		self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
+		webhook.write('slack', 'https://slack.com/api/chat.postMessage', 'hi', 'C1')
+		webhook.write('slack', 'https://hooks.slack.com/services/T/B/x', 'hi')
+		self.assertEqual(len(self.posts), 2)
+		for url, kwargs in self.posts:
+			with self.subTest(url=url): # they used to go in the headers, where Slack does not look
+				self.assertEqual((kwargs['json']['unfurl_links'], kwargs['json']['unfurl_media']), (False, False))
+				self.assertNotIn('unfurl_links', kwargs['headers'])
+
+	def test_a_reply_without_a_bot_token_is_logged_not_sent(self):
+		self.patch(webhook, 'config_slackBotToken', False)
+		err = io.StringIO()
+		with contextlib.redirect_stderr(err), contextlib.redirect_stdout(io.StringIO()):
+			for cmd in ('.help', '.price', '.compare T1 T2'):
+				worker.process_request('slack', 'C1', '<@U1>', cmd, BOT, 'U')
+		self.assertEqual((self.posts, self.errors), ([], []))
+		self.assertEqual(err.getvalue().count('ERROR slack: a reply to channel C1 needs slackBotToken in finbot.env; not sent'), 3)
+
+	def test_the_slack_api_says_what_is_missing(self):
+		self.patch(webhook, 'config_slackBotToken', False)
+		with self.assertRaisesRegex(RuntimeError, 'Slack files.getUploadURLExternal: needs slackBotToken in finbot.env'):
+			webhook.slack_api('files.getUploadURLExternal', filename='chart.png', length=1) # it used to be a TypeError
+		self.assertEqual(self.posts, [])
 
 
 class TelegramWebhook(FinbotCase):
-    """bot.py registers the webhook once as it starts. Importing lib.telegram, as every job process does, touches nothing."""
-
-    def setUp(self):
-        super().setUp()
-        self.patch(telegram, 'webhooks', {'telegram': 'https://api.telegram.org/botX/'})
-        self.patch(telegram, 'config_telegramOutgoingWebhook', 'https://example.com/telegram')
-
-    def test_register_unregisters_registers_and_names_the_bot(self):
-        calls = []
-        def post(url, **kwargs):
-            calls.append((url.rsplit('/', 1)[1], kwargs.get('params', {}).get('url')))
-            return Resp({'ok': True, 'result': {'username': 'finbot'} if url.endswith('getMe') else True})
-        self.patch(telegram.net, 'post', post)
-        with contextlib.redirect_stderr(io.StringIO()):
-            self.assertEqual(telegram.register(), '@finbot')
-        self.assertEqual(calls, [('setWebhook', ''), ('setWebhook', 'https://example.com/telegram'), ('getMe', None)])
-
-    def test_a_refused_call_raises(self):
-        self.patch(telegram.net, 'post', lambda url, **k: Resp({'ok': False, 'description': 'Bad Request: bad webhook: HTTPS url must be provided'}))
-        with contextlib.redirect_stderr(io.StringIO()), self.assertRaises(RuntimeError) as caught:
-            telegram.register()
-        self.assertIn('HTTPS url must be provided', str(caught.exception))
-
-    def test_edits_toasts_and_typing_go_through_call(self):
-        """A chart edit that Telegram refuses is an error (unless nothing changed); a toast, a text edit or a typing notice that
-        fails, or an edit lost on the way, is only logged."""
-        def answer(outcome):
-            def post(url, **kwargs):
-                if isinstance(outcome, Exception):
-                    raise outcome
-                return outcome
-            return post
-        refused = Resp({'ok': False, 'description': 'Bad Request: message to edit not found'})
-        unchanged = Resp({'ok': False, 'description': 'Bad Request: message is not modified: specified new message content is the same'})
-        for outcome, raises in ((refused, True), (unchanged, False), (ConnectionError('down'), False), (Resp({'ok': True, 'result': True}), False)):
-            with self.subTest(outcome=outcome):
-                self.patch(telegram.net, 'post', answer(outcome))
-                with contextlib.redirect_stderr(io.StringIO()):
-                    if raises:
-                        self.assertRaisesRegex(RuntimeError, 'not found', webhook.editMessageMedia, '5', '9', io.BytesIO(b'png'), 'caption')
-                    else:
-                        webhook.editMessageMedia('5', '9', io.BytesIO(b'png'), 'caption')
-                    telegram.answerCallbackQuery('1', 'Loading…')
-                    webhook.editMessageText('5', '9', 'text')
-                    webhook.pleaseHold('typing', 'telegram', '5')
-
-    def test_importing_never_calls_telegram(self):
-        code = ('import requests\n'
-                'def refuse(*a, **k): raise SystemExit("network call on import")\n'
-                'requests.Session.request = refuse\n'
-                'from lib import telegram, yahoo, worker, jobrun\n'
-                'print("quiet")')
-        with tempfile.TemporaryDirectory() as data:
-            out = subprocess.run([sys.executable, '-c', code], cwd=ROOT, capture_output=True, text=True,
-                env={**os.environ, 'DATA_DIR': data, 'telegramBotToken': '123:abc', 'telegramOutgoingWebhook': 'https://example.com/telegram'})
-        self.assertEqual(out.stdout.split()[-1:], ['quiet'], out.stderr)
-
-    def test_the_bot_will_not_start_unregistered_and_never_logs_the_token(self):
-        skip_if_old_env(self)
-        with tempfile.TemporaryDirectory() as data:
-            env = {k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k not in config.STARTUP_KEYS and 'proxy' not in k.lower()}
-            env.update(DATA_DIR=data, port='0', scheduler='False', telegramBotToken='123456:SECRET-token', telegramOutgoingToken='tok',
-                       telegramOutgoingWebhook='https://example.com/telegram',
-                       HTTPS_PROXY='http://127.0.0.1:9', https_proxy='http://127.0.0.1:9') # unreachable, so the registration fails at once and offline
-            out = subprocess.run([sys.executable, 'bot.py'], cwd=ROOT, env=env, capture_output=True, text=True, timeout=120)
-        self.assertEqual(out.returncode, 1, out.stderr)
-        self.assertIn('ERROR ⚠️ telegram webhook:', out.stderr)
-        self.assertNotIn('SECRET-token', out.stderr + out.stdout)
+	"""bot.py registers the webhook once as it starts. Importing lib.telegram, as every job process does, touches nothing."""
+
+	def setUp(self):
+		super().setUp()
+		self.patch(telegram, 'webhooks', {'telegram': 'https://api.telegram.org/botX/'})
+		self.patch(telegram, 'config_telegramOutgoingWebhook', 'https://example.com/telegram')
+
+	def test_register_unregisters_registers_and_names_the_bot(self):
+		calls = []
+		def post(url, **kwargs):
+			calls.append((url.rsplit('/', 1)[1], kwargs.get('params', {}).get('url')))
+			return Resp({'ok': True, 'result': {'username': 'finbot'} if url.endswith('getMe') else True})
+		self.patch(telegram.net, 'post', post)
+		with contextlib.redirect_stderr(io.StringIO()):
+			self.assertEqual(telegram.register(), '@finbot')
+		self.assertEqual(calls, [('setWebhook', ''), ('setWebhook', 'https://example.com/telegram'), ('getMe', None)])
+
+	def test_a_refused_call_raises(self):
+		self.patch(telegram.net, 'post', lambda url, **k: Resp({'ok': False, 'description': 'Bad Request: bad webhook: HTTPS url must be provided'}))
+		with contextlib.redirect_stderr(io.StringIO()), self.assertRaises(RuntimeError) as caught:
+			telegram.register()
+		self.assertIn('HTTPS url must be provided', str(caught.exception))
+
+	def test_edits_toasts_and_typing_go_through_call(self):
+		"""A chart edit that Telegram refuses is an error (unless nothing changed); a toast, a text edit or a typing notice that
+		fails, or an edit lost on the way, is only logged."""
+		def answer(outcome):
+			def post(url, **kwargs):
+				if isinstance(outcome, Exception):
+					raise outcome
+				return outcome
+			return post
+		refused = Resp({'ok': False, 'description': 'Bad Request: message to edit not found'})
+		unchanged = Resp({'ok': False, 'description': 'Bad Request: message is not modified: specified new message content is the same'})
+		for outcome, raises in ((refused, True), (unchanged, False), (ConnectionError('down'), False), (Resp({'ok': True, 'result': True}), False)):
+			with self.subTest(outcome=outcome):
+				self.patch(telegram.net, 'post', answer(outcome))
+				with contextlib.redirect_stderr(io.StringIO()):
+					if raises:
+						self.assertRaisesRegex(RuntimeError, 'not found', webhook.editMessageMedia, '5', '9', io.BytesIO(b'png'), 'caption')
+					else:
+						webhook.editMessageMedia('5', '9', io.BytesIO(b'png'), 'caption')
+					telegram.answerCallbackQuery('1', 'Loading…')
+					webhook.editMessageText('5', '9', 'text')
+					webhook.pleaseHold('typing', 'telegram', '5')
+
+	def test_importing_never_calls_telegram(self):
+		code = ('import requests\n'
+				'def refuse(*a, **k): raise SystemExit("network call on import")\n'
+				'requests.Session.request = refuse\n'
+				'from lib import telegram, yahoo, worker, jobrun\n'
+				'print("quiet")')
+		with tempfile.TemporaryDirectory() as data:
+			out = subprocess.run([sys.executable, '-c', code], cwd=ROOT, capture_output=True, text=True,
+				env={**os.environ, 'DATA_DIR': data, 'telegramBotToken': '123:abc', 'telegramOutgoingWebhook': 'https://example.com/telegram'})
+		self.assertEqual(out.stdout.split()[-1:], ['quiet'], out.stderr)
+
+	def test_the_bot_will_not_start_unregistered_and_never_logs_the_token(self):
+		skip_if_old_env(self)
+		with tempfile.TemporaryDirectory() as data:
+			env = {k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k not in config.STARTUP_KEYS and 'proxy' not in k.lower()}
+			env.update(DATA_DIR=data, port='0', scheduler='False', telegramBotToken='123456:SECRET-token', telegramOutgoingToken='tok',
+					   telegramOutgoingWebhook='https://example.com/telegram',
+					   HTTPS_PROXY='http://127.0.0.1:9', https_proxy='http://127.0.0.1:9') # unreachable, so the registration fails at once and offline
+			out = subprocess.run([sys.executable, 'bot.py'], cwd=ROOT, env=env, capture_output=True, text=True, timeout=120)
+		self.assertEqual(out.returncode, 1, out.stderr)
+		self.assertIn('ERROR ⚠️ telegram webhook:', out.stderr)
+		self.assertNotIn('SECRET-token', out.stderr + out.stdout)
 
 
 if __name__ == '__main__':
-    unittest.main()
+	unittest.main()
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 49c06a3d..084ea115 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -5,427 +5,427 @@
 
 
 class Commands(FinbotCase):
-    def test_text_commands(self):
-        for cmd in ('.help', '/start', '.hello', '.thanks', '.who', '.plan', '.holdings', '.marketcap',
-                    '.pe', '.fpe', '.peg', '.beta', '.buy', '.sell', '.earnings', '.dividend',
-                    '.earnings T1', '.dividend T1', '.trades', '.trades 30d', '.price T1', '.session T1',
-                    '.premarket T1', '.shorts T1', '.watchlist', '.T1'):
-            with self.subTest(cmd=cmd):
-                self.assertTrue(self.command(cmd), f'{cmd} sent nothing')
-
-    def test_chart_commands(self):
-        for cmd in ('.price', '.price 7d', '.price top', '.price bottom', '.session', '.premarket', '.postmarket',
-                    '.performance', '.performance 7d', '.performance 1y', '.shorts', '.peg', '.buy', '.sell', '.beta', '.price top 7d',
-                    '.forward pe', '.fpe bottom'):
-            with self.subTest(cmd=cmd):
-                out = self.command(cmd)
-                self.assertTrue(out, f'{cmd} sent nothing')
-                self.assertTrue(any(kind == 'photo' for kind, *_ in out), f'{cmd} sent no chart: {out}')
-
-    def test_one_button_asks_for_both_ratings_reports(self):
-        for text in ('.buy & sell', '.buy sell', '.buy/sell', '.buy and sell'):
-            with self.subTest(text=text):
-                captions = [caption for kind, _, caption in self.command(text) if kind == 'photo']
-                self.assertEqual(len(captions), 2, captions)
-                self.assertIn('analyst buy ratings', captions[0])
-                self.assertIn('analyst sell ratings', captions[1])
-        self.assertIn('analyst sell ratings', str(self.command('.sell & buy')[0])) # in the order asked
-        self.assertEqual(len(self.command('.buy & buy')), 1) # each report once
-
-    def test_the_keyboard_is_alphabetical_and_every_button_is_a_command(self):
-        buttons = [button['text'] for row in charts.command_keyboard()['keyboard'] for button in row]
-        self.assertEqual(buttons, sorted(buttons))
-        self.assertEqual(len(buttons), 15)
-        self.assertIn('.buy & sell', buttons)
-        self.assertIn('.forward pe', buttons)
-        for text in buttons: # not taken for a ticker by the catch-all .SYMBOL
-            with self.subTest(text=text):
-                self.assertIsNot(next(handler for pattern, handler in worker.COMMANDS if worker.match(pattern, text, BOT)), worker.on_profile)
-
-    def test_history_and_compare(self):
-        for cmd in ('.history T1', '.compare T1 T2', '.compare T1 T2 3m'):
-            with self.subTest(cmd=cmd):
-                self.assertTrue(self.command(cmd), f'{cmd} sent nothing')
-
-    def test_trades_and_watchlist_report_when_nothing_is_stored(self):
-        # first run: no state files exist yet, which used to crash these
-        for cmd in ('.trades', '.watchlist'):
-            self.assertTrue(self.command(cmd), cmd)
-
-    def test_unknown_stickers_and_noise_are_ignored(self):
-        self.assertEqual(self.command('lorem ipsum dolor sit amet'), [])
+	def test_text_commands(self):
+		for cmd in ('.help', '/start', '.hello', '.thanks', '.who', '.plan', '.holdings', '.marketcap',
+					'.pe', '.fpe', '.peg', '.beta', '.buy', '.sell', '.earnings', '.dividend',
+					'.earnings T1', '.dividend T1', '.trades', '.trades 30d', '.price T1', '.session T1',
+					'.premarket T1', '.shorts T1', '.watchlist', '.T1'):
+			with self.subTest(cmd=cmd):
+				self.assertTrue(self.command(cmd), f'{cmd} sent nothing')
+
+	def test_chart_commands(self):
+		for cmd in ('.price', '.price 7d', '.price top', '.price bottom', '.session', '.premarket', '.postmarket',
+					'.performance', '.performance 7d', '.performance 1y', '.shorts', '.peg', '.buy', '.sell', '.beta', '.price top 7d',
+					'.forward pe', '.fpe bottom'):
+			with self.subTest(cmd=cmd):
+				out = self.command(cmd)
+				self.assertTrue(out, f'{cmd} sent nothing')
+				self.assertTrue(any(kind == 'photo' for kind, *_ in out), f'{cmd} sent no chart: {out}')
+
+	def test_one_button_asks_for_both_ratings_reports(self):
+		for text in ('.buy & sell', '.buy sell', '.buy/sell', '.buy and sell'):
+			with self.subTest(text=text):
+				captions = [caption for kind, _, caption in self.command(text) if kind == 'photo']
+				self.assertEqual(len(captions), 2, captions)
+				self.assertIn('analyst buy ratings', captions[0])
+				self.assertIn('analyst sell ratings', captions[1])
+		self.assertIn('analyst sell ratings', str(self.command('.sell & buy')[0])) # in the order asked
+		self.assertEqual(len(self.command('.buy & buy')), 1) # each report once
+
+	def test_the_keyboard_is_alphabetical_and_every_button_is_a_command(self):
+		buttons = [button['text'] for row in charts.command_keyboard()['keyboard'] for button in row]
+		self.assertEqual(buttons, sorted(buttons))
+		self.assertEqual(len(buttons), 15)
+		self.assertIn('.buy & sell', buttons)
+		self.assertIn('.forward pe', buttons)
+		for text in buttons: # not taken for a ticker by the catch-all .SYMBOL
+			with self.subTest(text=text):
+				self.assertIsNot(next(handler for pattern, handler in worker.COMMANDS if worker.match(pattern, text, BOT)), worker.on_profile)
+
+	def test_history_and_compare(self):
+		for cmd in ('.history T1', '.compare T1 T2', '.compare T1 T2 3m'):
+			with self.subTest(cmd=cmd):
+				self.assertTrue(self.command(cmd), f'{cmd} sent nothing')
+
+	def test_trades_and_watchlist_report_when_nothing_is_stored(self):
+		# first run: no state files exist yet, which used to crash these
+		for cmd in ('.trades', '.watchlist'):
+			self.assertTrue(self.command(cmd), cmd)
+
+	def test_unknown_stickers_and_noise_are_ignored(self):
+		self.assertEqual(self.command('lorem ipsum dolor sit amet'), [])
 
 
 class ReportFunctions(FinbotCase):
-    def test_prepare_help(self):
-        self.assertTrue(reports.prepare_help('telegram', BOT))
-
-    def test_a_portfolio_with_no_holdings_is_reported_by_name(self):
-        """Its name used to be read from its first holding: an empty portfolio crashed .performance."""
-        self.patch(sharesight, 'get_performance', lambda pid, d, *a: {'report': {'holdings': [], 'capital_gain_percent': '0.0'}})
-        payload, _ = performance.report('telegram', 7)
-        self.assertTrue(any('>Rob</a>' in line for line in payload), payload)
-
-    def test_prepare_holdings(self):
-        self.assertTrue(reports.prepare_holdings_payload(None, 'telegram'))
-
-    def test_prepare_marketcap(self):
-        self.assertTrue(reports.prepare_marketcap_payload('telegram', 'top', length=5))
-
-    def test_prepare_rating(self):
-        for action in ('buy', 'sell'):
-            self.assertTrue(reports.prepare_rating_payload('telegram', action, length=5)[0])
-
-    def test_ratings_list_the_strongest_first(self):
-        """1 is a strong buy and 5 a strong sell: .sell used to list the mildest sells first. On equal ratings, more analysts first;
-        fewer than 7 analysts and the stock is left out (T7, T8: the ASX small caps that are always a perfect 1.0)."""
-        ratings = {'T1': ('buy', 1.5, 7), 'T2': ('strong_buy', 1.2, 12), 'T3': ('buy', 1.5, 9), 'T7': ('strong_buy', 1.0, 6),
-                   'T4': ('sell', 4.2, 7), 'T5': ('underperform', 4.6, 10), 'T6': ('sell', 4.2, 8), 'T8': ('strong_buy', 1.0, 2)}
-        self.patch(util, 'get_holdings_and_watchlist', lambda: list(ratings))
-        self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | dict(zip(('recommend', 'recommend_index', 'recommend_analysts'), ratings[t]))})
-        def order(action):
-            payload, chart = reports.prepare_rating_payload('telegram', action)
-            return [line.split('Co ')[1].split(' ')[0] for line in payload[1:]], [label for label, _ in chart['rows']]
-        self.assertEqual(order('buy'), (['T2', 'T3', 'T1'], ['T2', 'T3', 'T1']))
-        self.assertEqual(order('sell'), (['T5', 'T6', 'T4'], ['T5', 'T6', 'T4']))
-        self.assertIn('(7+ analysts)', reports.prepare_rating_payload('telegram', 'buy')[0][0])
-        self.patch(util, 'get_holdings_and_watchlist', lambda: ['T7', 'T8'])
-        self.assertEqual(reports.prepare_rating_payload('telegram', 'buy'), (['No tracked stocks rated buy by 7+ analysts'], None))
-
-    def test_a_small_cap_quoted_in_won_needs_the_bigger_move_too(self):
-        """demote_volatile compares market caps in US dollars: unconverted, a small cap quoted in won was never small."""
-        self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | ({'regularMarketPrice': 1380.0} if t == 'KRW=X' else {'currency': 'KRW', 'market_cap': 1e11}) for t in ts})
-        self.patch(price, 'config_demote_volatile', True)
-        quote = price.gather()[1]['T1']
-        self.assertAlmostEqual(quote['market_cap_usd'], 1e11 / 1380.0) # US$72M
-        self.assertEqual(price.trigger_bar('T1', quote, 5), 5 * price.config_volatility_multiplier)
-
-    def test_the_small_caps_betas_are_fetched_together(self):
-        """A scheduled price report asks for the beta of each small cap that moved (US$150M to US$1B): together, not one by one."""
-        calls = []
-        self.patch(yahoo, 'fetch_details', lambda tickers, strict=False: calls.append(sorted(tickers)) or {t: market_entry(t) for t in tickers})
-        self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: self.fail('a beta fetched on its own'))
-        self.patch(price, 'config_demote_volatile', True)
-        self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | {'market_cap': 5e8} for t in ts})
-        price.lambda_handler()
-        self.assertEqual(len(calls), 1)
-        self.assertTrue(calls[0])
-
-    def test_forward_pe_lists_need_as_many_analysts_as_buy(self):
-        """A forward P/E is analysts' estimate: with few of them it says little, as a perfect 1.0 buy from two brokers does."""
-        analysts = {'T1': 7, 'T2': 6, 'T3': 12, 'T4': 2}
-        self.patch(util, 'get_holdings_and_watchlist', lambda: list(analysts))
-        self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | {'price_to_earnings_forward': -5 if t == 'T3' else 20} for t in ts})
-        self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'recommend_analysts': analysts[t]}})
-        def listed(action):
-            payload, chart = reports.prepare_value_payload('telegram', action)
-            return payload[0], [line.split('Co ')[1].split(' ')[0] for line in payload[1:]], chart
-        heading, names, chart = listed('forward pe')
-        self.assertEqual((names, [label for label, _ in chart['rows']]), (['T1'], ['T1'])) # T2 and T4: too few; T3: negative
-        self.assertIn('forward pe ratio (7+ analysts)', heading)
-        self.assertEqual(listed('negative forward pe')[1], ['T3'])
-        self.assertIn('Co T2', str(reports.prepare_value_payload('telegram', 'forward pe', 'T2'))) # one ticker asked for: shown
-        self.patch(reports, 'config_min_analysts', 13) # the one setting moves .buy and .forward pe together
-        self.assertEqual(reports.prepare_value_payload('telegram', 'forward pe')[0], ['No tracked stocks with 13+ analysts have a forward pe ratio'])
-        self.assertEqual(reports.prepare_rating_payload('telegram', 'buy')[0], ['No tracked stocks rated buy by 13+ analysts'])
-
-    def test_market_caps_compare_in_us_dollars(self):
-        """Yahoo gives a market cap in the stock's own currency: unconverted, a won market cap topped every list."""
-        stocks = {'K1.KS': ('KRW', 1.01e15), 'K2.KS': ('KRW', 5e11), 'U1': ('USD', 5.39e12), 'L1.L': ('GBp', 6.25e10), 'Z1': ('XYZ', 1e12)}
-        fx = {'KRW=X': 1380.0, 'GBP=X': 0.79} # units per US dollar; nothing for XYZ
-        def fetch(tickers):
-            return {t: market_entry(t) | {'regularMarketPrice': fx[t]} if t in fx else market_entry(t) | {'currency': stocks[t][0], 'market_cap': stocks[t][1], 'beta': 2.0}
-                    for t in tickers if t in fx or t in stocks}
-        self.patch(yahoo, 'fetch', fetch)
-        self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: fetch([t]))
-        self.patch(util, 'get_holdings_and_watchlist', lambda: list(stocks))
-        payload = reports.prepare_marketcap_payload('telegram')
-        self.assertIn('by market cap (USD)', payload[0])
-        self.assertEqual([line.split('Co ')[1].split(' ')[0] for line in payload[1:]], ['U1', 'K1.KS', 'L1.L', 'K2.KS']) # Z1: no rate, left out
-        self.assertIn('mkt cap: 731.88 B', payload[2]) # 1.01 Q won
-        self.assertIn('mkt cap: KRW 1.01 Q (USD 731.88 B)', reports.prepare_marketcap_payload('telegram', 'K1.KS')[0])
-        self.assertIn('mkt cap: GBP 62.50 B (USD 79.11 B)', reports.prepare_marketcap_payload('telegram', 'L1.L')[0]) # pence quote, pound cap
-        beta = reports.prepare_beta_payload('telegram')[0]
-        self.assertIn('Co K2.KS', str(beta)) # 500 billion won is US$362M: a small cap, which it never was unconverted
-        self.assertNotIn('Co K1.KS', str(beta))
-
-    def test_prepare_value(self):
-        for action, heading in (('pe', 'Top 5 tracked stocks by pe ratio'), ('bottom forward pe', 'Bottom 5 tracked stocks by forward pe ratio'),
-                                ('peg', 'Top 5 tracked stocks by peg ratio'), ('negative peg', 'negative peg: no data found')):
-            with self.subTest(action=action):
-                payload, chart = reports.prepare_value_payload('telegram', action, length=5)
-                self.assertIn(heading, payload[0])
-
-    def test_prepare_beta(self):
-        payload, chart = reports.prepare_beta_payload('telegram')
-        self.assertIn('Beta over 1.5', payload[0])
-        self.assertEqual(len(chart['rows']), len(payload) - 1) # a bar per stock listed
-        self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'beta': 1.0}})
-        self.assertEqual(reports.prepare_beta_payload('telegram'), (['No tracked stocks have a beta over 1.5 and mkt cap under US$2B'], None))
-
-    def test_prepare_profile(self):
-        self.assertTrue(reports.prepare_profile_payload('telegram', '@u', 'T1'))
-
-    def test_a_profile_leaves_out_what_yahoo_did_not_send(self):
-        """Each of these fields missing (or a zero equity) used to crash the whole profile."""
-        full = market_entry('T1') | {'insiderBuy': 1, 'insiderSell': 0, 'insiderBuyValue': 5e6, 'insiderSellValue': 0, 'netIncomeToCommon': 1e6,
-                                     'sharesOutstanding': 1e6, 'shareholder_equity': 1e9, 'total_debt': 9e8, 'total_cash': 1e8, 'profile_industry': 'Tools'}
-        for missing in ('marketState', 'profile_exchange', 'market_cap', 'currency', 'regularMarketPrice', 'sharesOutstanding', 'percent_change'):
-            with self.subTest(missing=missing):
-                self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: {k: v for k, v in full.items() if k != missing}})
-                self.assertIn('Co T1', reports.prepare_profile_payload('telegram', '@u', 'T1')[0])
-        self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: full | {'shareholder_equity': 0}})
-        self.assertNotIn('Net debt/equity', str(reports.prepare_profile_payload('telegram', '@u', 'T1')))
-
-    def test_a_profile_warns_of_a_ps_ratio_only_above_15(self):
-        for ps, warned in ((12.3, False), (15.1, True)): # forward P/E 99 warns (above 30): its warning used to show on the PS line too
-            with self.subTest(ps=ps):
-                self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'price_to_sales': ps, 'price_to_earnings_forward': 99, 'profile_industry': 'Tools'}})
-                line = next(line for line in reports.prepare_profile_payload('telegram', '@u', 'T1') if 'PS ratio' in line)
-                self.assertEqual('⚠️' in line, warned, line)
-
-    def test_prepare_watchlist(self):
-        self.assertTrue(reports.prepare_watchlist('telegram', '@u'))
-
-    def test_prepare_compare_usage(self):
-        caption = reports.prepare_compare('telegram', ['T1'])[0]
-        self.assertIn('Usage', ' '.join(caption))
+	def test_prepare_help(self):
+		self.assertTrue(reports.prepare_help('telegram', BOT))
+
+	def test_a_portfolio_with_no_holdings_is_reported_by_name(self):
+		"""Its name used to be read from its first holding: an empty portfolio crashed .performance."""
+		self.patch(sharesight, 'get_performance', lambda pid, d, *a: {'report': {'holdings': [], 'capital_gain_percent': '0.0'}})
+		payload, _ = performance.report('telegram', 7)
+		self.assertTrue(any('>Rob</a>' in line for line in payload), payload)
+
+	def test_prepare_holdings(self):
+		self.assertTrue(reports.prepare_holdings_payload(None, 'telegram'))
+
+	def test_prepare_marketcap(self):
+		self.assertTrue(reports.prepare_marketcap_payload('telegram', 'top', length=5))
+
+	def test_prepare_rating(self):
+		for action in ('buy', 'sell'):
+			self.assertTrue(reports.prepare_rating_payload('telegram', action, length=5)[0])
+
+	def test_ratings_list_the_strongest_first(self):
+		"""1 is a strong buy and 5 a strong sell: .sell used to list the mildest sells first. On equal ratings, more analysts first;
+		fewer than 7 analysts and the stock is left out (T7, T8: the ASX small caps that are always a perfect 1.0)."""
+		ratings = {'T1': ('buy', 1.5, 7), 'T2': ('strong_buy', 1.2, 12), 'T3': ('buy', 1.5, 9), 'T7': ('strong_buy', 1.0, 6),
+				   'T4': ('sell', 4.2, 7), 'T5': ('underperform', 4.6, 10), 'T6': ('sell', 4.2, 8), 'T8': ('strong_buy', 1.0, 2)}
+		self.patch(util, 'get_holdings_and_watchlist', lambda: list(ratings))
+		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | dict(zip(('recommend', 'recommend_index', 'recommend_analysts'), ratings[t]))})
+		def order(action):
+			payload, chart = reports.prepare_rating_payload('telegram', action)
+			return [line.split('Co ')[1].split(' ')[0] for line in payload[1:]], [label for label, _ in chart['rows']]
+		self.assertEqual(order('buy'), (['T2', 'T3', 'T1'], ['T2', 'T3', 'T1']))
+		self.assertEqual(order('sell'), (['T5', 'T6', 'T4'], ['T5', 'T6', 'T4']))
+		self.assertIn('(7+ analysts)', reports.prepare_rating_payload('telegram', 'buy')[0][0])
+		self.patch(util, 'get_holdings_and_watchlist', lambda: ['T7', 'T8'])
+		self.assertEqual(reports.prepare_rating_payload('telegram', 'buy'), (['No tracked stocks rated buy by 7+ analysts'], None))
+
+	def test_a_small_cap_quoted_in_won_needs_the_bigger_move_too(self):
+		"""demote_volatile compares market caps in US dollars: unconverted, a small cap quoted in won was never small."""
+		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | ({'regularMarketPrice': 1380.0} if t == 'KRW=X' else {'currency': 'KRW', 'market_cap': 1e11}) for t in ts})
+		self.patch(price, 'config_demote_volatile', True)
+		quote = price.gather()[1]['T1']
+		self.assertAlmostEqual(quote['market_cap_usd'], 1e11 / 1380.0) # US$72M
+		self.assertEqual(price.trigger_bar('T1', quote, 5), 5 * price.config_volatility_multiplier)
+
+	def test_the_small_caps_betas_are_fetched_together(self):
+		"""A scheduled price report asks for the beta of each small cap that moved (US$150M to US$1B): together, not one by one."""
+		calls = []
+		self.patch(yahoo, 'fetch_details', lambda tickers, strict=False: calls.append(sorted(tickers)) or {t: market_entry(t) for t in tickers})
+		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: self.fail('a beta fetched on its own'))
+		self.patch(price, 'config_demote_volatile', True)
+		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | {'market_cap': 5e8} for t in ts})
+		price.lambda_handler()
+		self.assertEqual(len(calls), 1)
+		self.assertTrue(calls[0])
+
+	def test_forward_pe_lists_need_as_many_analysts_as_buy(self):
+		"""A forward P/E is analysts' estimate: with few of them it says little, as a perfect 1.0 buy from two brokers does."""
+		analysts = {'T1': 7, 'T2': 6, 'T3': 12, 'T4': 2}
+		self.patch(util, 'get_holdings_and_watchlist', lambda: list(analysts))
+		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | {'price_to_earnings_forward': -5 if t == 'T3' else 20} for t in ts})
+		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'recommend_analysts': analysts[t]}})
+		def listed(action):
+			payload, chart = reports.prepare_value_payload('telegram', action)
+			return payload[0], [line.split('Co ')[1].split(' ')[0] for line in payload[1:]], chart
+		heading, names, chart = listed('forward pe')
+		self.assertEqual((names, [label for label, _ in chart['rows']]), (['T1'], ['T1'])) # T2 and T4: too few; T3: negative
+		self.assertIn('forward pe ratio (7+ analysts)', heading)
+		self.assertEqual(listed('negative forward pe')[1], ['T3'])
+		self.assertIn('Co T2', str(reports.prepare_value_payload('telegram', 'forward pe', 'T2'))) # one ticker asked for: shown
+		self.patch(reports, 'config_min_analysts', 13) # the one setting moves .buy and .forward pe together
+		self.assertEqual(reports.prepare_value_payload('telegram', 'forward pe')[0], ['No tracked stocks with 13+ analysts have a forward pe ratio'])
+		self.assertEqual(reports.prepare_rating_payload('telegram', 'buy')[0], ['No tracked stocks rated buy by 13+ analysts'])
+
+	def test_market_caps_compare_in_us_dollars(self):
+		"""Yahoo gives a market cap in the stock's own currency: unconverted, a won market cap topped every list."""
+		stocks = {'K1.KS': ('KRW', 1.01e15), 'K2.KS': ('KRW', 5e11), 'U1': ('USD', 5.39e12), 'L1.L': ('GBp', 6.25e10), 'Z1': ('XYZ', 1e12)}
+		fx = {'KRW=X': 1380.0, 'GBP=X': 0.79} # units per US dollar; nothing for XYZ
+		def fetch(tickers):
+			return {t: market_entry(t) | {'regularMarketPrice': fx[t]} if t in fx else market_entry(t) | {'currency': stocks[t][0], 'market_cap': stocks[t][1], 'beta': 2.0}
+					for t in tickers if t in fx or t in stocks}
+		self.patch(yahoo, 'fetch', fetch)
+		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: fetch([t]))
+		self.patch(util, 'get_holdings_and_watchlist', lambda: list(stocks))
+		payload = reports.prepare_marketcap_payload('telegram')
+		self.assertIn('by market cap (USD)', payload[0])
+		self.assertEqual([line.split('Co ')[1].split(' ')[0] for line in payload[1:]], ['U1', 'K1.KS', 'L1.L', 'K2.KS']) # Z1: no rate, left out
+		self.assertIn('mkt cap: 731.88 B', payload[2]) # 1.01 Q won
+		self.assertIn('mkt cap: KRW 1.01 Q (USD 731.88 B)', reports.prepare_marketcap_payload('telegram', 'K1.KS')[0])
+		self.assertIn('mkt cap: GBP 62.50 B (USD 79.11 B)', reports.prepare_marketcap_payload('telegram', 'L1.L')[0]) # pence quote, pound cap
+		beta = reports.prepare_beta_payload('telegram')[0]
+		self.assertIn('Co K2.KS', str(beta)) # 500 billion won is US$362M: a small cap, which it never was unconverted
+		self.assertNotIn('Co K1.KS', str(beta))
+
+	def test_prepare_value(self):
+		for action, heading in (('pe', 'Top 5 tracked stocks by pe ratio'), ('bottom forward pe', 'Bottom 5 tracked stocks by forward pe ratio'),
+								('peg', 'Top 5 tracked stocks by peg ratio'), ('negative peg', 'negative peg: no data found')):
+			with self.subTest(action=action):
+				payload, chart = reports.prepare_value_payload('telegram', action, length=5)
+				self.assertIn(heading, payload[0])
+
+	def test_prepare_beta(self):
+		payload, chart = reports.prepare_beta_payload('telegram')
+		self.assertIn('Beta over 1.5', payload[0])
+		self.assertEqual(len(chart['rows']), len(payload) - 1) # a bar per stock listed
+		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'beta': 1.0}})
+		self.assertEqual(reports.prepare_beta_payload('telegram'), (['No tracked stocks have a beta over 1.5 and mkt cap under US$2B'], None))
+
+	def test_prepare_profile(self):
+		self.assertTrue(reports.prepare_profile_payload('telegram', '@u', 'T1'))
+
+	def test_a_profile_leaves_out_what_yahoo_did_not_send(self):
+		"""Each of these fields missing (or a zero equity) used to crash the whole profile."""
+		full = market_entry('T1') | {'insiderBuy': 1, 'insiderSell': 0, 'insiderBuyValue': 5e6, 'insiderSellValue': 0, 'netIncomeToCommon': 1e6,
+									 'sharesOutstanding': 1e6, 'shareholder_equity': 1e9, 'total_debt': 9e8, 'total_cash': 1e8, 'profile_industry': 'Tools'}
+		for missing in ('marketState', 'profile_exchange', 'market_cap', 'currency', 'regularMarketPrice', 'sharesOutstanding', 'percent_change'):
+			with self.subTest(missing=missing):
+				self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: {k: v for k, v in full.items() if k != missing}})
+				self.assertIn('Co T1', reports.prepare_profile_payload('telegram', '@u', 'T1')[0])
+		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: full | {'shareholder_equity': 0}})
+		self.assertNotIn('Net debt/equity', str(reports.prepare_profile_payload('telegram', '@u', 'T1')))
+
+	def test_a_profile_warns_of_a_ps_ratio_only_above_15(self):
+		for ps, warned in ((12.3, False), (15.1, True)): # forward P/E 99 warns (above 30): its warning used to show on the PS line too
+			with self.subTest(ps=ps):
+				self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'price_to_sales': ps, 'price_to_earnings_forward': 99, 'profile_industry': 'Tools'}})
+				line = next(line for line in reports.prepare_profile_payload('telegram', '@u', 'T1') if 'PS ratio' in line)
+				self.assertEqual('⚠️' in line, warned, line)
+
+	def test_prepare_watchlist(self):
+		self.assertTrue(reports.prepare_watchlist('telegram', '@u'))
+
+	def test_prepare_compare_usage(self):
+		caption = reports.prepare_compare('telegram', ['T1'])[0]
+		self.assertIn('Usage', ' '.join(caption))
 
 
 class ScheduledReports(FinbotCase):
-    """The cron entry points (non-interactive: they post to the default channel)."""
-
-    def test_price(self):
-        price.lambda_handler()
-        self.assertTrue(self.sent)
-
-    def test_performance(self):
-        performance.lambda_handler(past_days=7)
-        self.assertTrue(self.sent)
-
-    def test_shorts(self):
-        shorts.lambda_handler()
-        self.assertTrue(self.sent)
-
-    def test_earnings_and_dividends(self):
-        for kwargs in ({'event': 'earnings'}, {'event': 'ex-dividend'}):
-            with self.subTest(kwargs=kwargs):
-                self.sent.clear()
-                cal.lambda_handler(**kwargs)
-                self.assertTrue(self.sent)
-
-    def test_earnings_skip_an_adr_whose_primary_listing_yahoo_does_not_know(self):
-        """T1 has no earnings date, so its primary listing is looked up; Yahoo knowing nothing of it used to crash the report."""
-        self.patch(cal, 'primary_listing', {'T1': 'T1.XX'})
-        self.patch(yahoo, 'fetch', lambda ts: {t: {k: v for k, v in market_entry(t).items() if not (t == 'T1' and k == 'earnings_date')} for t in ts})
-        self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {} if t == 'T1.XX' else {t: market_entry(t)})
-        cal.lambda_handler(event='earnings')
-        self.assertIn('Co T2 ', str(self.sent))
-        self.assertNotIn('Co T1 ', str(self.sent))
-
-    def test_trades(self):
-        trades.lambda_handler(interactive=True, service='telegram', chat_id='5')
-        self.assertTrue(self.sent)
-
-    def test_trades_asks_yahoo_only_for_new_trades(self):
-        fetched = []
-        self.patch(yahoo, 'fetch', lambda tickers: fetched.append(set(tickers)) or {t: market_entry(t) for t in tickers})
-        trades.lambda_handler()
-        self.assertEqual(fetched, [{'T1'}])
-        fetched.clear()
-        trades.lambda_handler() # the trade is known now: nothing to report and nothing to fetch
-        self.assertEqual(fetched, [])
-
-    def test_milestone(self):
-        milestone.lambda_handler()
-        self.assertTrue(self.sent)
-
-    def test_milestone_records_a_ticker_only_with_its_new_values(self):
-        """A first sighting without 52-week values used to be recorded as 0/0, so the next run reported a record high for it."""
-        self.patch(yahoo, 'historic_high', lambda t: (150.0, 50.0)) # beyond the 52-week range (120/80): nothing new
-        self.patch(yahoo, 'fetch', lambda ts: {t: {k: v for k, v in market_entry(t).items() if not k.startswith('fiftyTwoWeek')} for t in ts})
-        milestone.lambda_handler()
-        self.assertFalse(util.json_load('finbot_milestone.json', persist=True))
-        self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts})
-        self.sent.clear()
-        milestone.lambda_handler()
-        self.assertNotIn('Record high', str(self.sent))
-        self.assertEqual(util.json_load('finbot_milestone.json', persist=True)['T1'], {'record_high': 150.0, 'record_low': 50.0}) # all-time
-
-    def test_milestone_records_are_all_time(self):
-        """A new 52-week high below the all-time high is no record. The record used to become the 52-week high after the first
-        run, so a stock far below its peak posted a record high each time it set a new 52-week one."""
-        self.patch(yahoo, 'historic_high', lambda t: (150.0, 50.0))
-        milestone.lambda_handler() # first sighting: the 52-week range (120/80) is inside the history (150/50)
-        for high, expected in ((130.0, False), (160.0, True), (155.0, False)):
-            with self.subTest(high=high):
-                self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | {'fiftyTwoWeekHigh': high} for t in ts})
-                self.sent.clear()
-                milestone.lambda_handler()
-                self.assertEqual('Record high' in str(self.sent), expected)
-
-    def test_a_52_week_record_kept_before_is_replaced_by_the_history(self):
-        util.json_write('finbot_milestone.json', {f'T{i}': {'high': 110.0, 'low': 90.0} for i in range(25)}, persist=True)
-        self.patch(yahoo, 'historic_high', lambda t: (150.0, 50.0))
-        milestone.lambda_handler() # 120 beats the old 52-week record of 110, but not the history's 150
-        self.assertNotIn('Record high', str(self.sent))
-
-    def test_rating(self):
-        rating.lambda_handler(interactive=True, service='telegram', chat_id='5')
-        self.assertTrue(self.sent)
-
-    def test_rating_keeps_the_last_rating_of_a_ticker_missing_from_a_run(self):
-        rating.lambda_handler() # records every ticker's rating
-        self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {} if t == 'T1' else {t: market_entry(t)})
-        rating.lambda_handler() # no T1 this time: its rating used to be dropped from the record
-        self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | ({'recommend': 'sell', 'recommend_index': 4.5} if t == 'T1' else {})})
-        self.sent.clear()
-        rating.lambda_handler()
-        self.assertIn('Co T1 ', str(self.sent)) # so its change from buy to sell went unreported
-
-    def test_an_interactive_rating_records_nothing(self):
-        rating.lambda_handler(interactive=True, service='telegram', chat_id='5')
-        self.assertIsNone(util.json_load('finbot_rating.json', persist=True))
-
-    def test_slack_webhook_without_a_bot_token_gets_scheduled_reports(self):
-        """Scheduled reports reach Slack through its incoming webhook, which takes no channel and no token. They used to send
-        Telegram's chat id along, which made write() need slackBotToken and crash without one, before Telegram got its copy."""
-        self.patch(webhook, 'config_slackBotToken', False)
-        webhook.webhooks['slack'] = 'https://hooks.slack.com/services/T/B/x'
-        self.patch(webhook, 'payload_wrapper', REAL_PAYLOAD_WRAPPER)
-        self.patch(webhook, 'sendPhoto', REAL_SEND_PHOTO)
-        posts = []
-        def post(url, **kwargs):
-            posts.append((url.split('/')[2], 'Authorization' in kwargs.get('headers', {}), (kwargs.get('json') or {}).get('channel')))
-            return mock.Mock(status_code=200, json=lambda: {'ok': True, 'result': {'message_id': 1, 'chat': {'id': 1}}})
-        self.patch(webhook.net, 'post', post)
-        for report in (price.lambda_handler, shorts.lambda_handler, milestone.lambda_handler):
-            with self.subTest(report=report):
-                posts.clear()
-                report()
-                self.assertIn(('hooks.slack.com', False, None), posts)
-                self.assertTrue(any(host == 'api.telegram.org' for host, _, _ in posts))
-
-    def test_nowhere_to_post_is_an_error(self):
-        self.patch(webhook, 'webhooks', {})
-        with self.assertRaisesRegex(RuntimeError, 'nowhere to post'):
-            price.lambda_handler()
+	"""The cron entry points (non-interactive: they post to the default channel)."""
+
+	def test_price(self):
+		price.lambda_handler()
+		self.assertTrue(self.sent)
+
+	def test_performance(self):
+		performance.lambda_handler(past_days=7)
+		self.assertTrue(self.sent)
+
+	def test_shorts(self):
+		shorts.lambda_handler()
+		self.assertTrue(self.sent)
+
+	def test_earnings_and_dividends(self):
+		for kwargs in ({'event': 'earnings'}, {'event': 'ex-dividend'}):
+			with self.subTest(kwargs=kwargs):
+				self.sent.clear()
+				cal.lambda_handler(**kwargs)
+				self.assertTrue(self.sent)
+
+	def test_earnings_skip_an_adr_whose_primary_listing_yahoo_does_not_know(self):
+		"""T1 has no earnings date, so its primary listing is looked up; Yahoo knowing nothing of it used to crash the report."""
+		self.patch(cal, 'primary_listing', {'T1': 'T1.XX'})
+		self.patch(yahoo, 'fetch', lambda ts: {t: {k: v for k, v in market_entry(t).items() if not (t == 'T1' and k == 'earnings_date')} for t in ts})
+		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {} if t == 'T1.XX' else {t: market_entry(t)})
+		cal.lambda_handler(event='earnings')
+		self.assertIn('Co T2 ', str(self.sent))
+		self.assertNotIn('Co T1 ', str(self.sent))
+
+	def test_trades(self):
+		trades.lambda_handler(interactive=True, service='telegram', chat_id='5')
+		self.assertTrue(self.sent)
+
+	def test_trades_asks_yahoo_only_for_new_trades(self):
+		fetched = []
+		self.patch(yahoo, 'fetch', lambda tickers: fetched.append(set(tickers)) or {t: market_entry(t) for t in tickers})
+		trades.lambda_handler()
+		self.assertEqual(fetched, [{'T1'}])
+		fetched.clear()
+		trades.lambda_handler() # the trade is known now: nothing to report and nothing to fetch
+		self.assertEqual(fetched, [])
+
+	def test_milestone(self):
+		milestone.lambda_handler()
+		self.assertTrue(self.sent)
+
+	def test_milestone_records_a_ticker_only_with_its_new_values(self):
+		"""A first sighting without 52-week values used to be recorded as 0/0, so the next run reported a record high for it."""
+		self.patch(yahoo, 'historic_high', lambda t: (150.0, 50.0)) # beyond the 52-week range (120/80): nothing new
+		self.patch(yahoo, 'fetch', lambda ts: {t: {k: v for k, v in market_entry(t).items() if not k.startswith('fiftyTwoWeek')} for t in ts})
+		milestone.lambda_handler()
+		self.assertFalse(util.json_load('finbot_milestone.json', persist=True))
+		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts})
+		self.sent.clear()
+		milestone.lambda_handler()
+		self.assertNotIn('Record high', str(self.sent))
+		self.assertEqual(util.json_load('finbot_milestone.json', persist=True)['T1'], {'record_high': 150.0, 'record_low': 50.0}) # all-time
+
+	def test_milestone_records_are_all_time(self):
+		"""A new 52-week high below the all-time high is no record. The record used to become the 52-week high after the first
+		run, so a stock far below its peak posted a record high each time it set a new 52-week one."""
+		self.patch(yahoo, 'historic_high', lambda t: (150.0, 50.0))
+		milestone.lambda_handler() # first sighting: the 52-week range (120/80) is inside the history (150/50)
+		for high, expected in ((130.0, False), (160.0, True), (155.0, False)):
+			with self.subTest(high=high):
+				self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | {'fiftyTwoWeekHigh': high} for t in ts})
+				self.sent.clear()
+				milestone.lambda_handler()
+				self.assertEqual('Record high' in str(self.sent), expected)
+
+	def test_a_52_week_record_kept_before_is_replaced_by_the_history(self):
+		util.json_write('finbot_milestone.json', {f'T{i}': {'high': 110.0, 'low': 90.0} for i in range(25)}, persist=True)
+		self.patch(yahoo, 'historic_high', lambda t: (150.0, 50.0))
+		milestone.lambda_handler() # 120 beats the old 52-week record of 110, but not the history's 150
+		self.assertNotIn('Record high', str(self.sent))
+
+	def test_rating(self):
+		rating.lambda_handler(interactive=True, service='telegram', chat_id='5')
+		self.assertTrue(self.sent)
+
+	def test_rating_keeps_the_last_rating_of_a_ticker_missing_from_a_run(self):
+		rating.lambda_handler() # records every ticker's rating
+		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {} if t == 'T1' else {t: market_entry(t)})
+		rating.lambda_handler() # no T1 this time: its rating used to be dropped from the record
+		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | ({'recommend': 'sell', 'recommend_index': 4.5} if t == 'T1' else {})})
+		self.sent.clear()
+		rating.lambda_handler()
+		self.assertIn('Co T1 ', str(self.sent)) # so its change from buy to sell went unreported
+
+	def test_an_interactive_rating_records_nothing(self):
+		rating.lambda_handler(interactive=True, service='telegram', chat_id='5')
+		self.assertIsNone(util.json_load('finbot_rating.json', persist=True))
+
+	def test_slack_webhook_without_a_bot_token_gets_scheduled_reports(self):
+		"""Scheduled reports reach Slack through its incoming webhook, which takes no channel and no token. They used to send
+		Telegram's chat id along, which made write() need slackBotToken and crash without one, before Telegram got its copy."""
+		self.patch(webhook, 'config_slackBotToken', False)
+		webhook.webhooks['slack'] = 'https://hooks.slack.com/services/T/B/x'
+		self.patch(webhook, 'payload_wrapper', REAL_PAYLOAD_WRAPPER)
+		self.patch(webhook, 'sendPhoto', REAL_SEND_PHOTO)
+		posts = []
+		def post(url, **kwargs):
+			posts.append((url.split('/')[2], 'Authorization' in kwargs.get('headers', {}), (kwargs.get('json') or {}).get('channel')))
+			return mock.Mock(status_code=200, json=lambda: {'ok': True, 'result': {'message_id': 1, 'chat': {'id': 1}}})
+		self.patch(webhook.net, 'post', post)
+		for report in (price.lambda_handler, shorts.lambda_handler, milestone.lambda_handler):
+			with self.subTest(report=report):
+				posts.clear()
+				report()
+				self.assertIn(('hooks.slack.com', False, None), posts)
+				self.assertTrue(any(host == 'api.telegram.org' for host, _, _ in posts))
+
+	def test_nowhere_to_post_is_an_error(self):
+		self.patch(webhook, 'webhooks', {})
+		with self.assertRaisesRegex(RuntimeError, 'nowhere to post'):
+			price.lambda_handler()
 
 
 class PeriodButtons(FinbotCase):
-    """Default views line up with the buttons: the highlighted button is the period actually shown."""
-
-    def buttons(self, command):
-        self.command(command)
-        self.assertTrue(self.markups and self.markups[0], f'{command} sent no buttons')
-        return [b['text'] for b in self.markups[0]['inline_keyboard'][0]]
-
-    def test_performance_defaults_to_one_month(self):
-        self.assertEqual(self.buttons('.performance'), ['7D', '● 1M', '3M', '1Y', 'Max'])
-        self.assertIn('month', self.sent[0][2]) # "Performance over the past month", not 4 weeks
-
-    def test_performance_period_argument_moves_the_marker(self):
-        self.assertEqual(self.buttons('.performance 7d'), ['● 7D', '1M', '3M', '1Y', 'Max'])
-        self.assertEqual(self.buttons('.performance 1y'), ['7D', '1M', '3M', '● 1Y', 'Max'])
-
-    def test_price_list_has_a_1d_button_marked_by_default(self):
-        self.assertEqual(self.buttons('.price'), ['● 1D', '7D', '1M', '3M', '1Y', 'Max'])
-        self.assertEqual(self.buttons('.price top'), ['● 1D', '7D', '1M', '3M', '1Y', 'Max'])
-        self.assertEqual(self.buttons('.price 1m'), ['1D', '7D', '● 1M', '3M', '1Y', 'Max'])
-
-    def test_other_charts_do_not_offer_1d(self):
-        for command in ('.performance', '.compare T1 T2'):
-            with self.subTest(command=command):
-                self.assertNotIn('1D', [b.lstrip('● ') for b in self.buttons(command)])
-
-    def press(self, data):
-        edits, toasts = [], []
-        self.patch(webhook, 'editMessageMedia', lambda chat, msg, img, caption, markup=None: edits.append((caption.split('\n')[0], markup)))
-        self.patch(telegram, 'answerCallbackQuery', lambda cid, text=None: toasts.append(text))
-        worker.process_callback('telegram', 'cb', '55', '9', data)
-        return edits, toasts
-
-    def test_pressing_1d_rebuilds_todays_list_and_moves_the_marker(self):
-        edits, toasts = self.press('c|l|d|3,0')
-        self.assertEqual(toasts, ['Loading 1D…'])
-        self.assertEqual(self.errors, [])
-        self.assertEqual(len(edits), 1)
-        self.assertEqual([b['text'] for b in edits[0][1]['inline_keyboard'][0]][0], '● 1D')
-
-    def test_a_button_a_chart_does_not_offer_is_ignored(self):
-        edits, toasts = self.press('c|f|d|') # performance has no 1D
-        self.assertEqual((edits, toasts), ([], [None]))
-        edits, toasts = self.press('c|l|zz|3,0')
-        self.assertEqual((edits, toasts), ([], [None]))
+	"""Default views line up with the buttons: the highlighted button is the period actually shown."""
+
+	def buttons(self, command):
+		self.command(command)
+		self.assertTrue(self.markups and self.markups[0], f'{command} sent no buttons')
+		return [b['text'] for b in self.markups[0]['inline_keyboard'][0]]
+
+	def test_performance_defaults_to_one_month(self):
+		self.assertEqual(self.buttons('.performance'), ['7D', '● 1M', '3M', '1Y', 'Max'])
+		self.assertIn('month', self.sent[0][2]) # "Performance over the past month", not 4 weeks
+
+	def test_performance_period_argument_moves_the_marker(self):
+		self.assertEqual(self.buttons('.performance 7d'), ['● 7D', '1M', '3M', '1Y', 'Max'])
+		self.assertEqual(self.buttons('.performance 1y'), ['7D', '1M', '3M', '● 1Y', 'Max'])
+
+	def test_price_list_has_a_1d_button_marked_by_default(self):
+		self.assertEqual(self.buttons('.price'), ['● 1D', '7D', '1M', '3M', '1Y', 'Max'])
+		self.assertEqual(self.buttons('.price top'), ['● 1D', '7D', '1M', '3M', '1Y', 'Max'])
+		self.assertEqual(self.buttons('.price 1m'), ['1D', '7D', '● 1M', '3M', '1Y', 'Max'])
+
+	def test_other_charts_do_not_offer_1d(self):
+		for command in ('.performance', '.compare T1 T2'):
+			with self.subTest(command=command):
+				self.assertNotIn('1D', [b.lstrip('● ') for b in self.buttons(command)])
+
+	def press(self, data):
+		edits, toasts = [], []
+		self.patch(webhook, 'editMessageMedia', lambda chat, msg, img, caption, markup=None: edits.append((caption.split('\n')[0], markup)))
+		self.patch(telegram, 'answerCallbackQuery', lambda cid, text=None: toasts.append(text))
+		worker.process_callback('telegram', 'cb', '55', '9', data)
+		return edits, toasts
+
+	def test_pressing_1d_rebuilds_todays_list_and_moves_the_marker(self):
+		edits, toasts = self.press('c|l|d|3,0')
+		self.assertEqual(toasts, ['Loading 1D…'])
+		self.assertEqual(self.errors, [])
+		self.assertEqual(len(edits), 1)
+		self.assertEqual([b['text'] for b in edits[0][1]['inline_keyboard'][0]][0], '● 1D')
+
+	def test_a_button_a_chart_does_not_offer_is_ignored(self):
+		edits, toasts = self.press('c|f|d|') # performance has no 1D
+		self.assertEqual((edits, toasts), ([], [None]))
+		edits, toasts = self.press('c|l|zz|3,0')
+		self.assertEqual((edits, toasts), ([], [None]))
 
 
 class Podcasts(FinbotCase):
-    def test_format_helpers(self):
-        import podcasts
-        self.assertEqual(podcasts.format_duration('3725'), '01:02')
-        self.assertEqual(podcasts.format_duration('1:02:03'), '01:02')
-        self.assertEqual(podcasts.format_duration('62:05'), '01:02')
-        self.assertEqual(podcasts.format_duration('nonsense'), '')
-        self.assertTrue(podcasts.how_long_ago(podcasts.published('Mon, 01 Jan 2024 00:00:00 +0000')).endswith('d ago'))
-        self.assertEqual(podcasts.published('Mon, 01 Jan 2024 00:00:00 -0000').tzinfo, datetime.timezone.utc) # no zone given: UTC
-        self.assertIsNone(podcasts.published('yesterday'))
-
-    def test_new_episodes_are_posted_once_newest_first(self):
-        """A feed whose items mix dated and undated ones used to crash the sort (aware and naive times)."""
-        import podcasts
-        def item(n, date):
-            return f'<item><guid>g{n}</guid><title>Show: ep {n}</title><enclosure url="https://x/{n}.mp3"/>{date}<itunes:duration>600</itunes:duration></item>'
-        feed = ('<rss xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd"><channel><title>Show</title>'
-                + item(1, '<pubDate>Mon, 01 Jan 2024 00:00:00 +0000</pubDate>') + item(2, '') + item(3, '<pubDate>Tue, 02 Jan 2024 00:00:00 GMT</pubDate>')
-                + '</channel></rss>').encode()
-        self.patch(podcasts.net, 'get', lambda url, **k: mock.Mock(content=feed, raise_for_status=lambda: None))
-        podcasts.lambda_handler('https://example.com/feed')
-        self.assertEqual([line.split(' (')[0] for line in self.sent[0][1]], ['📣 Show: <a href="https://x/3.mp3">ep 3</a> 00:10',
-                         '📣 Show: <a href="https://x/1.mp3">ep 1</a> 00:10', '📣 Show: <a href="https://x/2.mp3">ep 2</a> 00:10'])
-        self.sent.clear()
-        podcasts.lambda_handler('https://example.com/feed')
-        self.assertEqual(self.sent, []) # all seen now
+	def test_format_helpers(self):
+		import podcasts
+		self.assertEqual(podcasts.format_duration('3725'), '01:02')
+		self.assertEqual(podcasts.format_duration('1:02:03'), '01:02')
+		self.assertEqual(podcasts.format_duration('62:05'), '01:02')
+		self.assertEqual(podcasts.format_duration('nonsense'), '')
+		self.assertTrue(podcasts.how_long_ago(podcasts.published('Mon, 01 Jan 2024 00:00:00 +0000')).endswith('d ago'))
+		self.assertEqual(podcasts.published('Mon, 01 Jan 2024 00:00:00 -0000').tzinfo, datetime.timezone.utc) # no zone given: UTC
+		self.assertIsNone(podcasts.published('yesterday'))
+
+	def test_new_episodes_are_posted_once_newest_first(self):
+		"""A feed whose items mix dated and undated ones used to crash the sort (aware and naive times)."""
+		import podcasts
+		def item(n, date):
+			return f'<item><guid>g{n}</guid><title>Show: ep {n}</title><enclosure url="https://x/{n}.mp3"/>{date}<itunes:duration>600</itunes:duration></item>'
+		feed = ('<rss xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd"><channel><title>Show</title>'
+				+ item(1, '<pubDate>Mon, 01 Jan 2024 00:00:00 +0000</pubDate>') + item(2, '') + item(3, '<pubDate>Tue, 02 Jan 2024 00:00:00 GMT</pubDate>')
+				+ '</channel></rss>').encode()
+		self.patch(podcasts.net, 'get', lambda url, **k: mock.Mock(content=feed, raise_for_status=lambda: None))
+		podcasts.lambda_handler('https://example.com/feed')
+		self.assertEqual([line.split(' (')[0] for line in self.sent[0][1]], ['📣 Show: <a href="https://x/3.mp3">ep 3</a> 00:10',
+						 '📣 Show: <a href="https://x/1.mp3">ep 1</a> 00:10', '📣 Show: <a href="https://x/2.mp3">ep 2</a> 00:10'])
+		self.sent.clear()
+		podcasts.lambda_handler('https://example.com/feed')
+		self.assertEqual(self.sent, []) # all seen now
 
 
 class PieCommand(FinbotCase):
-    def setUp(self):
-        super().setUp()
-        def holdings(name, pid):
-            data = [('A', 'Semiconductors', 'Electronic Technology', 500), ('B', 'Semiconductors', 'Electronic Technology', 300),
-                    ('C', 'Biotechnology', 'Health Technology', 200), ('D', None, None, 100), ('E', 'Gold', 'Minerals', 0)]
-            return {c + ':NASDAQ': {'code': c, 'market_code': 'NASDAQ', 'industry_classification_name': i,
-                                    'sector_classification_name': s, 'value': v} for c, i, s, v in data}
-        self.patch(sharesight, 'get_holdings_new', holdings)
-
-    def test_by_industry_default(self):
-        sent = self.command('.pie Rob')
-        self.assertEqual(sent[0][0], 'photo')
-        self.assertIn('Rob by industry', sent[0][2])
-
-    def test_by_sector_and_payload(self):
-        payload, chart = reports.prepare_pie('rob', 'sector', 'telegram')
-        self.assertEqual([r[0] for r in chart['rows']], ['Electronic Technology', 'Health Technology', 'Unclassified'])
-        self.assertIn('Electronic Technology: 72.7%', payload) # 800 of 1100; zero-value Gold is left out
-        self.assertEqual(self.command('.pie Rob sector')[0][0], 'photo')
-
-    def test_folds_into_other_and_renders(self):
-        self.patch(sharesight, 'get_holdings_new', lambda n, p: {f'T{i}:X': {'code': f'T{i}', 'industry_classification_name': f'Ind {i}', 'value': 100 - i} for i in range(12)})
-        payload, chart = reports.prepare_pie('Rob', 'industry', 'telegram')
-        self.assertEqual(len(chart['rows']), 8)
-        self.assertEqual(chart['rows'][-1][0], 'Other')
-        self.assertTrue(util.pie_chart(chart['rows'], chart['title'], chart['subtitle']).read().startswith(b'\x89PNG'))
-
-    def test_unknown_or_missing_portfolio_lists_known(self):
-        for text, first in (('.pie', '.pie: please try again'), ('.pie Nope', 'Nope portfolio not found')):
-            sent = self.command(text)
-            self.assertTrue(sent[0][1][0].startswith(first), text)
-            self.assertIn('Rob', sent[0][1])
-
-    def test_a_service_that_takes_no_images_gets_text(self):
-        self.patch(webhook, 'config_slackBotToken', False) # Slack replies carry charts only with the bot token
-        worker.Request('slack', 'C1', '@u', 'U', BOT).deliver(['x'], {'pie': True, 'rows': [('a', 1.0)], 'title': 't'})
-        self.assertEqual(self.sent[-1][0], 'text')
+	def setUp(self):
+		super().setUp()
+		def holdings(name, pid):
+			data = [('A', 'Semiconductors', 'Electronic Technology', 500), ('B', 'Semiconductors', 'Electronic Technology', 300),
+					('C', 'Biotechnology', 'Health Technology', 200), ('D', None, None, 100), ('E', 'Gold', 'Minerals', 0)]
+			return {c + ':NASDAQ': {'code': c, 'market_code': 'NASDAQ', 'industry_classification_name': i,
+									'sector_classification_name': s, 'value': v} for c, i, s, v in data}
+		self.patch(sharesight, 'get_holdings_new', holdings)
+
+	def test_by_industry_default(self):
+		sent = self.command('.pie Rob')
+		self.assertEqual(sent[0][0], 'photo')
+		self.assertIn('Rob by industry', sent[0][2])
+
+	def test_by_sector_and_payload(self):
+		payload, chart = reports.prepare_pie('rob', 'sector', 'telegram')
+		self.assertEqual([r[0] for r in chart['rows']], ['Electronic Technology', 'Health Technology', 'Unclassified'])
+		self.assertIn('Electronic Technology: 72.7%', payload) # 800 of 1100; zero-value Gold is left out
+		self.assertEqual(self.command('.pie Rob sector')[0][0], 'photo')
+
+	def test_folds_into_other_and_renders(self):
+		self.patch(sharesight, 'get_holdings_new', lambda n, p: {f'T{i}:X': {'code': f'T{i}', 'industry_classification_name': f'Ind {i}', 'value': 100 - i} for i in range(12)})
+		payload, chart = reports.prepare_pie('Rob', 'industry', 'telegram')
+		self.assertEqual(len(chart['rows']), 8)
+		self.assertEqual(chart['rows'][-1][0], 'Other')
+		self.assertTrue(util.pie_chart(chart['rows'], chart['title'], chart['subtitle']).read().startswith(b'\x89PNG'))
+
+	def test_unknown_or_missing_portfolio_lists_known(self):
+		for text, first in (('.pie', '.pie: please try again'), ('.pie Nope', 'Nope portfolio not found')):
+			sent = self.command(text)
+			self.assertTrue(sent[0][1][0].startswith(first), text)
+			self.assertIn('Rob', sent[0][1])
+
+	def test_a_service_that_takes_no_images_gets_text(self):
+		self.patch(webhook, 'config_slackBotToken', False) # Slack replies carry charts only with the bot token
+		worker.Request('slack', 'C1', '@u', 'U', BOT).deliver(['x'], {'pie': True, 'rows': [('a', 1.0)], 'title': 't'})
+		self.assertEqual(self.sent[-1][0], 'text')
 
 
 if __name__ == '__main__':
-    unittest.main()
+	unittest.main()
diff --git a/tests/test_util.py b/tests/test_util.py
index 4b7e03f7..2b9793a8 100644
--- a/tests/test_util.py
+++ b/tests/test_util.py
@@ -5,144 +5,144 @@
 
 
 class Parsing(unittest.TestCase):
-    def test_days_from_human_days(self):
-        for arg, days in (('7', 7), ('7d', 7), ('7D', 7), ('2w', 14), ('3m', 90), ('1y', 365), ('0', 0)):
-            with self.subTest(arg=arg):
-                self.assertEqual(util.days_from_human_days(arg), days)
-        today = datetime.date.today()
-        self.assertEqual(util.days_from_human_days('ytd'), (today - today.replace(month=1, day=1)).days)
-
-    def test_days_from_human_days_rejects_non_periods(self):
-        for arg in ('AAPL', '', 'x5d', '1.5y', 'week'):
-            with self.subTest(arg=arg), self.assertRaises(ValueError):
-                util.days_from_human_days(arg)
-
-    def test_transform_to_yahoo(self):
-        for args, expected in ((('AAPL',), 'AAPL'), (('FANG', 'ASX'), 'FANG.AX'), (('ARM', 'NASDAQ'), 'ARM'),
-                               (('ARM:NASDAQ',), 'ARM'), (('BHP.AX',), 'BHP.AX'), (('0700', 'HKG'), '0700.HK'),
-                               (('BRK', 'B'), 'BRK-B'), (('2330', 'TAI'), '2330.TW'), (('VOD', 'LSE'), 'VOD.L'),
-                               (('7203', 'TSE'), '7203.T'), (('SHOP', 'TSX'), 'SHOP.TO'), (('AIR', 'NZX'), 'AIR.NZ'), (('D05', 'SGX'), 'D05.SI')):
-            with self.subTest(args=args):
-                self.assertEqual(util.transform_to_yahoo(*args), expected)
-
-    def test_each_market_has_its_own_flag_and_currency(self): # Sharesight's market codes: TSE is Tokyo, TSX Toronto, SGX Singapore
-        for market, ticker, flag, currency in (('TSE', '7203.T', '🇯🇵', 'JPY'), ('TSX', 'SHOP.TO', '🇨🇦', 'CAD'), ('NZX', 'AIR.NZ', '🇳🇿', 'NZD'),
-                                               ('SGX', 'D05.SI', '🇸🇬', 'SGD'), ('CPSE', None, '🇩🇰', 'DKK')):
-            with self.subTest(market=market):
-                self.assertEqual((util.flag_from_market(market), util.currency_from_market(market)), (flag, currency))
-                if ticker:
-                    self.assertEqual(util.flag_from_ticker(ticker), flag)
-
-    def test_make_paragraphs_keeps_every_word(self):
-        self.assertEqual(util.make_paragraphs('Acme makes widgets. It sells them.'), 'Acme makes widgets. It sells them.')
-        long = ' '.join(['word'] * 22) + ' end. Tail sentence here.'
-        self.assertEqual(util.make_paragraphs(long), ' '.join(['word'] * 22) + ' end.\n\nTail sentence here.')
-
-    def test_transform_title(self):
-        self.assertEqual(util.transform_title('APPLE INC'), 'Apple')
-        self.assertEqual(util.transform_title('Acme FPO'), 'Acme')
-        self.assertEqual(util.transform_title('Mixed Case Name'), 'Mixed Case Name')
-        for title, expected in (('Allianz Se', 'Allianz'), ('Foo SE AG', 'Foo'), ('Johnson & Johnson &', 'Johnson & Johnson'),
-                                ('Black and White and', 'Black and White'), ('Big Holdings Things Ltd', 'Big Things')):
-            with self.subTest(title=title): # a suffix comes off the end only; no run of spaces is left behind
-                self.assertEqual(util.transform_title(title), expected)
+	def test_days_from_human_days(self):
+		for arg, days in (('7', 7), ('7d', 7), ('7D', 7), ('2w', 14), ('3m', 90), ('1y', 365), ('0', 0)):
+			with self.subTest(arg=arg):
+				self.assertEqual(util.days_from_human_days(arg), days)
+		today = datetime.date.today()
+		self.assertEqual(util.days_from_human_days('ytd'), (today - today.replace(month=1, day=1)).days)
+
+	def test_days_from_human_days_rejects_non_periods(self):
+		for arg in ('AAPL', '', 'x5d', '1.5y', 'week'):
+			with self.subTest(arg=arg), self.assertRaises(ValueError):
+				util.days_from_human_days(arg)
+
+	def test_transform_to_yahoo(self):
+		for args, expected in ((('AAPL',), 'AAPL'), (('FANG', 'ASX'), 'FANG.AX'), (('ARM', 'NASDAQ'), 'ARM'),
+							   (('ARM:NASDAQ',), 'ARM'), (('BHP.AX',), 'BHP.AX'), (('0700', 'HKG'), '0700.HK'),
+							   (('BRK', 'B'), 'BRK-B'), (('2330', 'TAI'), '2330.TW'), (('VOD', 'LSE'), 'VOD.L'),
+							   (('7203', 'TSE'), '7203.T'), (('SHOP', 'TSX'), 'SHOP.TO'), (('AIR', 'NZX'), 'AIR.NZ'), (('D05', 'SGX'), 'D05.SI')):
+			with self.subTest(args=args):
+				self.assertEqual(util.transform_to_yahoo(*args), expected)
+
+	def test_each_market_has_its_own_flag_and_currency(self): # Sharesight's market codes: TSE is Tokyo, TSX Toronto, SGX Singapore
+		for market, ticker, flag, currency in (('TSE', '7203.T', '🇯🇵', 'JPY'), ('TSX', 'SHOP.TO', '🇨🇦', 'CAD'), ('NZX', 'AIR.NZ', '🇳🇿', 'NZD'),
+											   ('SGX', 'D05.SI', '🇸🇬', 'SGD'), ('CPSE', None, '🇩🇰', 'DKK')):
+			with self.subTest(market=market):
+				self.assertEqual((util.flag_from_market(market), util.currency_from_market(market)), (flag, currency))
+				if ticker:
+					self.assertEqual(util.flag_from_ticker(ticker), flag)
+
+	def test_make_paragraphs_keeps_every_word(self):
+		self.assertEqual(util.make_paragraphs('Acme makes widgets. It sells them.'), 'Acme makes widgets. It sells them.')
+		long = ' '.join(['word'] * 22) + ' end. Tail sentence here.'
+		self.assertEqual(util.make_paragraphs(long), ' '.join(['word'] * 22) + ' end.\n\nTail sentence here.')
+
+	def test_transform_title(self):
+		self.assertEqual(util.transform_title('APPLE INC'), 'Apple')
+		self.assertEqual(util.transform_title('Acme FPO'), 'Acme')
+		self.assertEqual(util.transform_title('Mixed Case Name'), 'Mixed Case Name')
+		for title, expected in (('Allianz Se', 'Allianz'), ('Foo SE AG', 'Foo'), ('Johnson & Johnson &', 'Johnson & Johnson'),
+								('Black and White and', 'Black and White'), ('Big Holdings Things Ltd', 'Big Things')):
+			with self.subTest(title=title): # a suffix comes off the end only; no run of spaces is left behind
+				self.assertEqual(util.transform_title(title), expected)
 
 
 class Formatting(unittest.TestCase):
-    def test_signed_percent(self):
-        for value, decimals, expected in ((12, None, '+12%'), (-3.5, None, '-3.5%'), (0, None, '0%'), (1234, None, '+1,234%'),
-                                          (2.345, 1, '+2.3%'), (-0.04, 1, '-0.0%')):
-            with self.subTest(value=value):
-                self.assertEqual(util.signed_percent(value, decimals), expected)
-
-    def test_days_english(self):
-        for days, expected in ((None, 'today'), (0, 'today'), (1, 'the past day'), (7, 'the past week')):
-            with self.subTest(days=days):
-                self.assertEqual(util.days_english(days), expected)
-
-    def test_date_range_english(self):
-        end = datetime.date(2026, 10, 5)
-        self.assertEqual(util.date_range_english(7, end), '28 Sep - 5 Oct 2026')
-        self.assertEqual(util.date_range_english(30, end), '5 Sep - 5 Oct 2026')
-        self.assertEqual(util.date_range_english(10, datetime.date(2026, 1, 5)), '26 Dec 2025 - 5 Jan 2026') # crosses a year
-        with self.assertRaises(TypeError):
-            util.date_range_english(None, end) # callers must pass a period
-
-    def test_ordinal(self):
-        for number, suffix in ((1, 'st'), (2, 'nd'), (3, 'rd'), (4, 'th'), (11, 'th'), (12, 'th'), (13, 'th'), (21, 'st'), (112, 'th')):
-            with self.subTest(number=number):
-                self.assertEqual(util.ordinal(number), suffix)
-
-    def test_get_emoji(self):
-        self.assertEqual([util.get_emoji(n) for n in (1, -1, 0)], ['🔼', '🔻', '▪️'])
-
-    def test_links_per_service(self):
-        telegram_link = util.link('https://example.com/x', 'text', 'telegram')
-        slack_link = util.link('https://example.com/x', 'text', 'slack')
-        self.assertIn('<a href=', telegram_link)
-        self.assertIn('|text', slack_link)
-
-    def test_flags_for_known_and_unknown_markets(self):
-        self.assertTrue(util.flag_from_ticker('BHP.AX'))
-        self.assertTrue(util.flag_from_ticker('AAPL'))
-        self.assertIsInstance(util.flag_from_ticker('XYZ.ZZ'), str) # unknown suffix must not crash
+	def test_signed_percent(self):
+		for value, decimals, expected in ((12, None, '+12%'), (-3.5, None, '-3.5%'), (0, None, '0%'), (1234, None, '+1,234%'),
+										  (2.345, 1, '+2.3%'), (-0.04, 1, '-0.0%')):
+			with self.subTest(value=value):
+				self.assertEqual(util.signed_percent(value, decimals), expected)
+
+	def test_days_english(self):
+		for days, expected in ((None, 'today'), (0, 'today'), (1, 'the past day'), (7, 'the past week')):
+			with self.subTest(days=days):
+				self.assertEqual(util.days_english(days), expected)
+
+	def test_date_range_english(self):
+		end = datetime.date(2026, 10, 5)
+		self.assertEqual(util.date_range_english(7, end), '28 Sep - 5 Oct 2026')
+		self.assertEqual(util.date_range_english(30, end), '5 Sep - 5 Oct 2026')
+		self.assertEqual(util.date_range_english(10, datetime.date(2026, 1, 5)), '26 Dec 2025 - 5 Jan 2026') # crosses a year
+		with self.assertRaises(TypeError):
+			util.date_range_english(None, end) # callers must pass a period
+
+	def test_ordinal(self):
+		for number, suffix in ((1, 'st'), (2, 'nd'), (3, 'rd'), (4, 'th'), (11, 'th'), (12, 'th'), (13, 'th'), (21, 'st'), (112, 'th')):
+			with self.subTest(number=number):
+				self.assertEqual(util.ordinal(number), suffix)
+
+	def test_get_emoji(self):
+		self.assertEqual([util.get_emoji(n) for n in (1, -1, 0)], ['🔼', '🔻', '▪️'])
+
+	def test_links_per_service(self):
+		telegram_link = util.link('https://example.com/x', 'text', 'telegram')
+		slack_link = util.link('https://example.com/x', 'text', 'slack')
+		self.assertIn('<a href=', telegram_link)
+		self.assertIn('|text', slack_link)
+
+	def test_flags_for_known_and_unknown_markets(self):
+		self.assertTrue(util.flag_from_ticker('BHP.AX'))
+		self.assertTrue(util.flag_from_ticker('AAPL'))
+		self.assertIsInstance(util.flag_from_ticker('XYZ.ZZ'), str) # unknown suffix must not crash
 
 
 class StateFiles(FinbotCase):
-    def test_json_roundtrip_and_missing_file(self):
-        self.assertIsNone(util.json_load('absent.json', persist=True))
-        util.json_write('state.json', {'a': [1, 2]}, persist=True)
-        self.assertEqual(util.json_load('state.json', persist=True), {'a': [1, 2]})
-
-    def test_a_failed_write_leaves_the_old_file_and_the_umask_alone(self):
-        util.json_write('state.json', {'a': 1}, persist=True)
-        path = Path(util.config_var_dir, 'state.json')
-        previous = os.umask(0o027)
-        self.addCleanup(os.umask, previous)
-        def disk_full(*a, **k):
-            raise OSError(28, 'No space left on device')
-        for broken in ('replace', 'fsync'):
-            with self.subTest(broken=broken), mock.patch.object(util.os, broken, disk_full):
-                with self.assertRaises(OSError):
-                    util.json_write('state.json', {'a': 2}, persist=True)
-            self.assertEqual(json.loads(path.read_text()), {'a': 1}) # whole or not at all
-            self.assertEqual(os.listdir(util.config_var_dir), ['state.json']) # no temp file left behind
-        with mock.patch.object(util.os, 'replace', disk_full), self.assertRaises(OSError):
-            util.write_binary_cache('img.png', io.BytesIO(b'png'))
-        self.assertEqual(os.listdir(util.config_var_dir), ['state.json'])
-        self.assertEqual(os.umask(0o027), 0o027) # the umask was not touched
-        util.json_write('state.json', {'a': 3}, persist=True)
-        self.assertEqual((json.loads(path.read_text()), oct(path.stat().st_mode & 0o777)), ({'a': 3}, '0o640'))
-
-    def test_read_cache_missing_or_expired_is_falsy(self):
-        self.assertFalse(util.read_cache('absent.json', 60))
-        util.json_write('old.json', {'a': 1})
-        self.assertFalse(util.read_cache('old.json', -1)) # already expired
-
-    def test_a_cache_file_that_goes_while_it_is_read_is_a_miss(self):
-        util.json_write('gone.json', {'a': 1})
-        real = Path.read_text
-        def vanish(path, *a, **k): # cleanup or a flush deleted it between the age check and the read
-            if path.name == 'gone.json':
-                path.unlink()
-                raise FileNotFoundError(path)
-            return real(path, *a, **k)
-        with mock.patch.object(Path, 'read_text', vanish):
-            self.assertIsNone(util.read_cache('gone.json', 60))
-
-    def test_binary_cache_roundtrip(self):
-        self.assertFalse(util.read_binary_cache('absent.png'))
-        util.write_binary_cache('img.png', io.BytesIO(b'png-bytes'))
-        self.assertEqual(util.read_binary_cache('img.png').read(), b'png-bytes')
-
-    def test_holdings_and_watchlist_without_a_watchlist_file(self):
-        # first run: no watchlist exists yet
-        self.patch(util.sharesight, 'get_holdings_wrapper', lambda: ['AAPL', 'GOOG', 'GOOGL'])
-        self.patch(util, 'get_holdings_and_watchlist', REAL_GET_HOLDINGS)
-        self.assertEqual(util.get_holdings_and_watchlist(), ['AAPL', 'GOOG']) # GOOGL dropped when GOOG is present
-        util.json_write('finbot_watchlist.json', ['MSFT'], persist=True)
-        self.assertEqual(util.get_holdings_and_watchlist(), ['AAPL', 'GOOG', 'MSFT'])
+	def test_json_roundtrip_and_missing_file(self):
+		self.assertIsNone(util.json_load('absent.json', persist=True))
+		util.json_write('state.json', {'a': [1, 2]}, persist=True)
+		self.assertEqual(util.json_load('state.json', persist=True), {'a': [1, 2]})
+
+	def test_a_failed_write_leaves_the_old_file_and_the_umask_alone(self):
+		util.json_write('state.json', {'a': 1}, persist=True)
+		path = Path(util.config_var_dir, 'state.json')
+		previous = os.umask(0o027)
+		self.addCleanup(os.umask, previous)
+		def disk_full(*a, **k):
+			raise OSError(28, 'No space left on device')
+		for broken in ('replace', 'fsync'):
+			with self.subTest(broken=broken), mock.patch.object(util.os, broken, disk_full):
+				with self.assertRaises(OSError):
+					util.json_write('state.json', {'a': 2}, persist=True)
+			self.assertEqual(json.loads(path.read_text()), {'a': 1}) # whole or not at all
+			self.assertEqual(os.listdir(util.config_var_dir), ['state.json']) # no temp file left behind
+		with mock.patch.object(util.os, 'replace', disk_full), self.assertRaises(OSError):
+			util.write_binary_cache('img.png', io.BytesIO(b'png'))
+		self.assertEqual(os.listdir(util.config_var_dir), ['state.json'])
+		self.assertEqual(os.umask(0o027), 0o027) # the umask was not touched
+		util.json_write('state.json', {'a': 3}, persist=True)
+		self.assertEqual((json.loads(path.read_text()), oct(path.stat().st_mode & 0o777)), ({'a': 3}, '0o640'))
+
+	def test_read_cache_missing_or_expired_is_falsy(self):
+		self.assertFalse(util.read_cache('absent.json', 60))
+		util.json_write('old.json', {'a': 1})
+		self.assertFalse(util.read_cache('old.json', -1)) # already expired
+
+	def test_a_cache_file_that_goes_while_it_is_read_is_a_miss(self):
+		util.json_write('gone.json', {'a': 1})
+		real = Path.read_text
+		def vanish(path, *a, **k): # cleanup or a flush deleted it between the age check and the read
+			if path.name == 'gone.json':
+				path.unlink()
+				raise FileNotFoundError(path)
+			return real(path, *a, **k)
+		with mock.patch.object(Path, 'read_text', vanish):
+			self.assertIsNone(util.read_cache('gone.json', 60))
+
+	def test_binary_cache_roundtrip(self):
+		self.assertFalse(util.read_binary_cache('absent.png'))
+		util.write_binary_cache('img.png', io.BytesIO(b'png-bytes'))
+		self.assertEqual(util.read_binary_cache('img.png').read(), b'png-bytes')
+
+	def test_holdings_and_watchlist_without_a_watchlist_file(self):
+		# first run: no watchlist exists yet
+		self.patch(util.sharesight, 'get_holdings_wrapper', lambda: ['AAPL', 'GOOG', 'GOOGL'])
+		self.patch(util, 'get_holdings_and_watchlist', REAL_GET_HOLDINGS)
+		self.assertEqual(util.get_holdings_and_watchlist(), ['AAPL', 'GOOG']) # GOOGL dropped when GOOG is present
+		util.json_write('finbot_watchlist.json', ['MSFT'], persist=True)
+		self.assertEqual(util.get_holdings_and_watchlist(), ['AAPL', 'GOOG', 'MSFT'])
 
 
 if __name__ == '__main__':
-    unittest.main()
+	unittest.main()
diff --git a/util/sws_gen.py b/util/sws_gen.py
index 0ec02ebd..518b588d 100644
--- a/util/sws_gen.py
+++ b/util/sws_gen.py
@@ -6,57 +6,57 @@
 import os
 
 def get_symbols(file):
-    symbols = []
-    with open(file) as file:
-        while line := file.readline():
-            symbol = line.rstrip()
-            symbols.append(symbol)
-    return symbols
+	symbols = []
+	with open(file) as file:
+		while line := file.readline():
+			symbol = line.rstrip()
+			symbols.append(symbol)
+	return symbols
 
 def get_sws(symbol, market):
-    query = "site:simplywall.st " + market + " " + symbol + " Stock Price, News & Analysis"
-    url = googlesearch.lucky(query)
-    print(symbol, url)
-    return url
+	query = "site:simplywall.st " + market + " " + symbol + " Stock Price, News & Analysis"
+	url = googlesearch.lucky(query)
+	print(symbol, url)
+	return url
 
 def build_dict(symbols, market):
-    count=1
-    mydict = {}
-    for symbol in symbols:
-        if count % 40 == 0:
-            print("Sleeping for 30 minutes")
-            time.sleep(30 * 60)
-        count = count + 1
-        url = get_sws(symbol, market)
-        if market == 'ASX':
-            symbol = symbol + '.' + 'AX'
-        mydict[symbol] = url
-    return mydict
+	count=1
+	mydict = {}
+	for symbol in symbols:
+		if count % 40 == 0:
+			print("Sleeping for 30 minutes")
+			time.sleep(30 * 60)
+		count = count + 1
+		url = get_sws(symbol, market)
+		if market == 'ASX':
+			symbol = symbol + '.' + 'AX'
+		mydict[symbol] = url
+	return mydict
 
 def do_sws_bulk(file):
-    market = ''
-    if 'asx' in file or 'ASX' in file:
-        market = 'ASX'
-    symbols = get_symbols(file)
-    mydict = build_dict(symbols, market)
-    cache_file = 'finbot_sws_' + file.split('.')[0] + '.json'
-    write_cache(mydict, (os.getenv('DATA_DIR') or 'var') + '/' + cache_file)
-    return mydict
+	market = ''
+	if 'asx' in file or 'ASX' in file:
+		market = 'ASX'
+	symbols = get_symbols(file)
+	mydict = build_dict(symbols, market)
+	cache_file = 'finbot_sws_' + file.split('.')[0] + '.json'
+	write_cache(mydict, (os.getenv('DATA_DIR') or 'var') + '/' + cache_file)
+	return mydict
 
 def write_cache(mydict, cache_file):
-    os.umask(0)
-    def opener(path, flags):
-        return os.open(path, flags, 0o640)
-    with open(cache_file, "w", opener=opener, encoding="utf-8") as f:
-        f.write(json.dumps(mydict))
-    os.umask(0o022)
+	os.umask(0)
+	def opener(path, flags):
+		return os.open(path, flags, 0o640)
+	with open(cache_file, "w", opener=opener, encoding="utf-8") as f:
+		f.write(json.dumps(mydict))
+	os.umask(0o022)
 
 files = ['nasdaq100', 'sp500', 'asx200'] 
 for idx, file in enumerate(files):
-    print("Starting", file)
-    mydict = do_sws_bulk(file)
-    print("Completed", file, "as finbot_sws_" + file.split('.')[0] + ".json")
-    if idx + 1 < len(files):
-        print("Sleeping for 15 minutes")
-        time.sleep(15 * 60)
+	print("Starting", file)
+	mydict = do_sws_bulk(file)
+	print("Completed", file, "as finbot_sws_" + file.split('.')[0] + ".json")
+	if idx + 1 < len(files):
+		print("Sleeping for 15 minutes")
+		time.sleep(15 * 60)
 

From 34b11b7322c5d438c8318fae01c3f0eb4add5215 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Wed, 14 Oct 2026 23:24:27 +0000
Subject: [PATCH 07/81] Config as two objects: config.env (finbot.env) and
 config.prefs (preferences.conf), read where used; no star import

- lib/config.py builds frozen Env and Prefs dataclasses, one place that lists every setting. A reload replaces config.prefs in one
  assignment, so a request never sees half a reload; it used to rewrite each changed name in every module that had copied it with
  `from lib.config import *`. The job processes' snapshot of the preferences lives in the same object.
- Every module reads config.env.<name> / config.prefs.<name> when it runs: no star import, no UNSET sentinel (defaults are None,
  resolved at call time), no names copied into 19 modules. preferences.conf's keys come from the Prefs fields.
- The greeting words moved to lib/words.py, the country suffixes to util.YAHOO_COUNTRY, and cal reads finbot_adr.json when it
  uses it (it was read once at import, so an edit needed a restart).
- Tests change settings through self.env(...) / self.prefs(...); the tests that policed the star import and the sentinel became
  one guard that nothing star-imports config or freezes a preference into a default.
- Checked by recording every request the bot sends for 70 chat commands on Telegram and Slack, 8 buttons and 13 scheduled reports
  to all three networks: byte for byte the same before and after, charts included.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md               |   8 +-
 bot.py                  |  35 ++--
 cal.py                  |  20 +-
 cleanup.py              |   6 +-
 lib/config.py           | 421 +++++++++++++---------------------------
 lib/jobs.py             |   7 +-
 lib/reports.py          |  20 +-
 lib/sharesight.py       |  64 +++---
 lib/shortman.py         |   4 +-
 lib/simplywallst.py     |   4 +-
 lib/telegram.py         |  10 +-
 lib/util.py             |  74 +++++--
 lib/webhook.py          |  38 ++--
 lib/words.py            | 142 ++++++++++++++
 lib/worker.py           |  21 +-
 lib/yahoo.py            |  54 +++---
 milestone.py            |   6 +-
 performance.py          |  14 +-
 price.py                |  36 ++--
 rating.py               |   1 -
 reminder.py             |  10 +-
 shorts.py               |   8 +-
 tests/common.py         |  19 +-
 tests/test_config.py    | 108 +++++------
 tests/test_datadir.py   |  28 +--
 tests/test_edges.py     |  19 +-
 tests/test_finbot.py    |   2 +-
 tests/test_jobs.py      |  15 +-
 tests/test_providers.py |  78 +++++---
 tests/test_reports.py   |  21 +-
 tests/test_util.py      |   6 +-
 trades.py               |  15 +-
 util/sync_defaults.py   |   7 +-
 33 files changed, 691 insertions(+), 630 deletions(-)
 create mode 100644 lib/words.py

diff --git a/CLAUDE.md b/CLAUDE.md
index b0ce1f7e..9e7a1474 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -20,7 +20,7 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 ## Facts
 - Python 3.13.5; tests are offline and need no network (`tests/`). CI runs them on the latest 3.x and 3.13.5, except for pull requests that change only README.md, CLAUDE.md or `img/*.png` (install, compile and tests are skipped; the allowlist in `.github/workflows/ci.yml` must never include a path a test reads).
 - Production runs the Debian system packages (including `python3-croniter`, for the scheduler in `lib/jobs.py`); `requirements.txt` is for pip/uv users and CI.
-- A new preference goes in five places: `PREFERENCE_KEYS` and `preference_values` (`lib/config.py`), `defaults/preferences.conf`, the README's preferences table, and `DEFAULTS` plus the key count in `tests/test_config.py`. The tests catch all but the README (no test may read README.md: CI skips the tests for a README-only change).
+- A new preference goes in five places: a `Prefs` field and its line in `preference_values` (`lib/config.py`), `defaults/preferences.conf`, the README's preferences table, and `DEFAULTS` plus the key count in `tests/test_config.py`. The tests catch all but the README (no test may read README.md: CI skips the tests for a README-only change).
 - If Yahoo and Sharesight are unreachable (the cloud environment's network policy can block them), check behaviour against the offline fixtures in `tests/common.py`. For a refactor, compare old and new output: load `git show HEAD:<file>` as a module beside the new one and diff what both produce for the same inputs.
 
 ## Code map (open the one file you need)
@@ -34,8 +34,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - `lib/charts.py`: Telegram period buttons and the rebuild behind them. `lib/util.py`: formatting, links, market tables, cache and
   state files, all matplotlib rendering.
 - `lib/yahoo.py lib/sharesight.py lib/shortman.py lib/simplywallst.py`: data sources, cached in `DATA_DIR/cache`.
-- `lib/config.py`: finbot.env + preferences.conf → `config_*` names (star-imported), reload. `lib/jobs.py` + `lib/jobrun.py`: the
-  scheduler; each run is a child process.
+- `lib/config.py`: `config.env` (finbot.env, fixed at start) and `config.prefs` (preferences.conf, a new object on each reload).
+  `lib/jobs.py` + `lib/jobrun.py`: the scheduler; each run is a child process.
 - `tests/common.py`: `FinbotCase` (offline fixtures). `self.command('.price')` returns what was sent: `('text', lines)` or
   `('photo', size, caption)`.
 - Which test file: commands and scheduled reports `test_reports`; failures, bad input, WSGI and Slack events `test_edges`;
@@ -44,6 +44,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   `test_finbot`.
 
 ## Conventions
+- Read settings as `config.env.<name>` / `config.prefs.<name>` where they are used, never into a default or module constant (a
+  reload must be seen); in tests, `self.env(...)` / `self.prefs(...)` change them.
 - A report is a list of lines. Markup through `webhook.bold/italic(text, service)`, user-written text through
   `webhook.escape(text, service)`, links through `util.link`/`util.finance_link`.
 - Tabs everywhere, tests included (`.editorconfig`). New and changed lines stay under 120 characters (an edit quotes whole lines).
diff --git a/bot.py b/bot.py
index da1c69ee..1a127277 100755
--- a/bot.py
+++ b/bot.py
@@ -12,7 +12,6 @@
 import threading
 from urllib.parse import urlparse
 
-from lib.config import *
 from lib import config
 from lib import worker
 from lib import jobs
@@ -73,19 +72,19 @@ def print_headers():
 	def reply(status, body=b''):
 		start_response(status, [('Content-type', 'application/json')])
 		return [body]
-	print_headers() if config_print_headers else None
+	print_headers() if config.prefs.print_headers else None
 	# route on the path first, and turn strangers away before reading the body
 	uri = environ.get('PATH_INFO', '')
-	if config_telegramOutgoingWebhook and uri == urlparse(config_telegramOutgoingWebhook).path:
+	if config.env.telegramOutgoingWebhook and uri == urlparse(config.env.telegramOutgoingWebhook).path:
 		service = 'telegram'
-	elif config_slackOutgoingWebhook and uri == urlparse(config_slackOutgoingWebhook).path:
+	elif config.env.slackOutgoingWebhook and uri == urlparse(config.env.slackOutgoingWebhook).path:
 		service = 'slack'
 	else:
 		print("Unknown URI", repr(uri), file=sys.stderr)
 		return reply('404 Not Found', b'<h1>404</h1>')
 	if service == 'telegram': # Telegram proves itself with a header
 		secret = environ.get('HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN')
-		if secret != config_telegramOutgoingToken:
+		if secret != config.env.telegramOutgoingToken:
 			print_headers()
 			print("Fatal: Telegram authorisation header", "not present" if secret is None else "is present but incorrect", file=sys.stderr)
 			return reply('403 Forbidden', b'<h1>Unauthorized</h1>')
@@ -125,7 +124,7 @@ def reply(status, body=b''):
 			user = userRealName = '@' + first_name
 		if "text" in msg:
 			message = msg["text"]
-			print("[Telegram]:", user, message, file=sys.stderr) if debug else None
+			print("[Telegram]:", user, message, file=sys.stderr) if config.prefs.debug else None
 		elif "photo" in msg:
 			message = msg.get("caption", '')
 			file_id = msg["photo"][-1]["file_id"]
@@ -140,7 +139,7 @@ def reply(status, body=b''):
 			print("warning: Slack authorisation field not present", file=sys.stderr)
 			print_body()
 			return reply('403 Forbidden', b'<h1>Unauthorized</h1>')
-		if inbound['token'] != config_slackOutgoingToken:
+		if inbound['token'] != config.env.slackOutgoingToken:
 			print("Slack auth incorrect", file=sys.stderr)
 			print_body()
 			return reply('403 Forbidden', b'<h1>Unauthorized</h1>')
@@ -170,7 +169,7 @@ def reply(status, body=b''):
 			user = userRealName = '<@' + event['user'] + '>' # ZXCVBN becomes <@ZXCVBN>
 			bot_name = '@' + bot_user # QWERTY becomes @QWERTY
 			chat_id = event['channel']
-			print(f"[{service}]:", user, message, file=sys.stderr) if debug else None
+			print(f"[{service}]:", user, message, file=sys.stderr) if config.prefs.debug else None
 			# this condition spawns a worker before returning
 		else:
 			print(f"[{service}]: unhandled 'type'", file=sys.stderr)
@@ -188,7 +187,7 @@ def git_version(checkout=None):
 	"""Short hash of the checked-out commit, read straight from the .git directory: no git binary to run, and no "dubious
 	ownership" refusal when the service user does not own the checkout. 'unknown' when it cannot be read."""
 	try:
-		git = Path(checkout or REPO_DIR) / '.git'
+		git = Path(checkout or config.REPO_DIR) / '.git'
 		if git.is_file(): # a linked worktree or submodule: .git is a file saying "gitdir: <path>"
 			git = (git.parent / git.read_text().split(':', 1)[1].strip()).resolve()
 		common = (git / (git / 'commondir').read_text().strip()).resolve() if (git / 'commondir').is_file() else git # refs of a worktree
@@ -226,7 +225,7 @@ def flush_cache(*args):
 		with yahoo._auth:
 			removed = cleanup.lambda_handler(days=0)
 		sharesight.wake()
-		print(f"cache: flushed {removed} file{'' if removed == 1 else 's'} from {config_cache_dir}", file=sys.stderr)
+		print(f"cache: flushed {removed} file{'' if removed == 1 else 's'} from {config.env.cache_dir}", file=sys.stderr)
 	except Exception as e:
 		print('ERROR', webhook.error_line(e, 'cache flush'), file=sys.stderr)
 
@@ -243,8 +242,8 @@ def flush_cache(*args):
 		print('ERROR config:', line, file=sys.stderr)
 	if os.getuid() == 0:
 		print("Running as superuser. This is not recommended.", file=sys.stderr)
-	if config_scheduler: # read before anything else happens: an impossible time in it stops the start
-		scheduler = jobs.Scheduler(lambda: data_file('schedule.cron'))
+	if config.env.scheduler: # read before anything else happens: an impossible time in it stops the start
+		scheduler = jobs.Scheduler(lambda: config.data_file('schedule.cron'))
 		try:
 			if not scheduler.load():
 				scheduler = None
@@ -256,18 +255,18 @@ def flush_cache(*args):
 	else:
 		print('Scheduled reports are off (scheduler=False)', file=sys.stderr)
 	util.render_off_hub() # charts drawn in a thread of their own: no chat waits while another's chart is drawn
-	httpd = pywsgi.WSGIServer((config_ip, config_port), main)
-	httpd.secure_repr = False if debug else None
-	print(f'Opening socket on http://{config_ip}:{config_port}', file=sys.stderr)
-	if config_telegramBotToken and config_telegramOutgoingWebhook:
+	httpd = pywsgi.WSGIServer((config.env.ip, config.env.port), main)
+	httpd.secure_repr = False if config.prefs.debug else None
+	print(f'Opening socket on http://{config.env.ip}:{config.env.port}', file=sys.stderr)
+	if config.env.telegramBotToken and config.env.telegramOutgoingWebhook:
 		try:
 			botName = telegram.register() # here and only here: the job processes import lib.telegram too
 		except Exception as e:
 			print('ERROR', webhook.error_line(e, 'telegram webhook'), file=sys.stderr)
 			sys.exit(1)
-	if config_telegramBotToken:
+	if config.env.telegramBotToken:
 		threading.Thread(target=charts.refresh_keyboards, daemon=True).start() # update stale DM keyboards in the background
-	if config_sharesight_auth.get('client_id'):
+	if config.env.sharesight_auth.get('client_id'):
 		threading.Thread(target=sharesight.keep_warm, daemon=True).start() # keep the portfolio list and holdings report cached, while cache is on
 	if scheduler:
 		jobs.current = scheduler # for the .jobs chat command
diff --git a/cal.py b/cal.py
index 10b9dd0f..42da8465 100755
--- a/cal.py
+++ b/cal.py
@@ -3,19 +3,23 @@
 import datetime
 import sys
 
-from lib.config import *
+from lib import config
 from lib import telegram
 from lib import util
 from lib import webhook
 from lib import yahoo
 
+def primary_listings():
+	"""{ADR or second listing: its primary listing}, from finbot_adr.json: read when used, so an edit is seen without a restart."""
+	return config.data_json('finbot_adr.json', {})
+
 EVENTS = {'earnings': ('earnings_date', 'Upcoming earnings'), 'ex-dividend': ('ex_dividend_date', 'Upcoming ex-dividends')} # -> (field, heading)
 
-def lambda_handler(chat_id=None, days=UNSET, service=None, specific_stock=None, interactive=False, event='earnings'):
+def lambda_handler(chat_id=None, days=None, service=None, specific_stock=None, interactive=False, event='earnings'):
 	"""The tracked stocks' (or one stock's) upcoming earnings or ex-dividend dates, as `event` says. A scheduled earnings report
 	is pinned on Telegram."""
-	if days is UNSET:
-		days = config_future_days
+	if days is None:
+		days = config.prefs.future_days
 	field, heading = EVENTS[event]
 	def prepare_payload(service, market_data):
 		now = datetime.datetime.now()
@@ -27,14 +31,14 @@ def prepare_payload(service, market_data):
 			ticker_primary = ticker
 			if field not in market_data[ticker]: # an ADR or a second listing: its primary listing may have the date
 				title_orig = market_data[ticker]['profile_title']
-				if ticker in primary_listing: # dict from lib.config
-					ticker_primary = primary_listing[ticker]
+				if ticker in primary_listings():
+					ticker_primary = primary_listings()[ticker]
 				elif ticker.endswith('.AX') and market_data[ticker].get('financialCurrency') == 'NZD':
 					ticker_primary = ticker.replace('.AX', '.NZ') # fixes GTK.AX and SKO.AX
 				market_data = market_data | yahoo.fetch_detail(ticker_primary)
 				country = market_data.get(ticker_primary, {}).get('profile_country') # only in the details
-				if country in yahoo_country and '.' not in ticker_primary: # dict from lib.config
-					ticker_primary = ticker_primary + '.' + yahoo_country[country] # fixes ASML and INFY
+				if country in util.YAHOO_COUNTRY and '.' not in ticker_primary:
+					ticker_primary = ticker_primary + '.' + util.YAHOO_COUNTRY[country] # fixes ASML and INFY
 					market_data = market_data | yahoo.fetch_detail(ticker_primary)
 				primary = market_data.get(ticker_primary, {})
 				if primary.get('profile_title') != title_orig or field not in primary: # not the same company, or no date there either
diff --git a/cleanup.py b/cleanup.py
index 40bafe80..d51bf316 100755
--- a/cleanup.py
+++ b/cleanup.py
@@ -1,7 +1,7 @@
 #!/usr/bin/env python3
 
 import os, sys, time
-from lib.config import *
+from lib import config
 from lib import webhook
 
 def lambda_handler(days=30):
@@ -11,7 +11,7 @@ def lambda_handler(days=30):
 	cutoff = now - days * 86400
 	removed = 0
 	try:
-		entries = list(os.scandir(config_cache_dir))
+		entries = list(os.scandir(config.env.cache_dir))
 	except FileNotFoundError:
 		return 0
 	for entry in entries:
@@ -23,7 +23,7 @@ def lambda_handler(days=30):
 				removed += 1
 			except OSError as e:
 				print("cleanup: can't remove", entry.path, e, file=sys.stderr)
-	print(f"cleanup: removed {removed} cache files older than {days} days from {config_cache_dir}")
+	print(f"cleanup: removed {removed} cache files older than {days} days from {config.env.cache_dir}")
 	return removed
 
 def job(argv):
diff --git a/lib/config.py b/lib/config.py
index e7e3259c..15284210 100644
--- a/lib/config.py
+++ b/lib/config.py
@@ -1,3 +1,4 @@
+import dataclasses
 import difflib
 import json
 import os
@@ -13,15 +14,9 @@
 REPO_DIR = Path(__file__).resolve().parent.parent
 os.chdir(REPO_DIR)
 
-UNSET = object() # "argument not given": a default that follows a preference is resolved when the function runs, so a reload is seen
-
 STARTUP_KEYS = ('DATA_DIR', 'ip', 'port', 'telegramChatID', 'scheduler', 'scheduler_catchup',
 	'sharesight_code', 'sharesight_client_id', 'sharesight_client_secret', 'slack_webhook', 'discord_webhook', 'slackBotToken',
 	'slackOutgoingToken', 'slackOutgoingWebhook', 'telegramBotToken', 'telegramOutgoingToken', 'telegramOutgoingWebhook')
-PREFERENCE_KEYS = ('alliterate', 'cache', 'cache_seconds', 'chunk_maxlines', 'country_code', 'debug', 'demote_leveraged', 'demote_volatile',
-	'exclude_portfolios', 'future_days', 'graph', 'http_timeout', 'hyperlink', 'hyperlinkFooter', 'hyperlinkProvider', 'include_portfolios',
-	'min_analysts', 'past_days', 'performance_cache_seconds', 'performance_use_sharesight', 'price_percent', 'print_headers', 'shorts_percent', 'timezone',
-	'trades_use_yahoo', 'volatility_multiplier')
 SECRET_KEYS = ('sharesight_code', 'sharesight_client_id', 'sharesight_client_secret', 'slack_webhook', 'discord_webhook', 'slackBotToken',
 	'slackOutgoingToken', 'slackOutgoingWebhook', 'telegramBotToken', 'telegramOutgoingToken', 'telegramOutgoingWebhook') # never printed
 PREFERENCES_FILE = 'preferences.conf'
@@ -155,31 +150,34 @@ def check(self):
 			raise ConfigError(self.errors)
 
 
+@dataclasses.dataclass(frozen=True)
+class Env:
+	"""The startup settings: finbot.env, which the launcher loads into the environment. Fixed for the life of the process."""
+	data_dir: str # state: watchlist, seen trades, ratings, the schedule's record, preferences.conf
+	cache_dir: str # data_dir/cache
+	ip: str
+	port: int
+	scheduler: bool # run the scheduled reports (schedule.cron) inside bot.py
+	scheduler_catchup: int # minutes: a run missed while the bot was down still fires if it was due this recently
+	telegramChatID: str # '' when unset, like every token below
+	telegramBotToken: str
+	telegramOutgoingWebhook: str
+	telegramOutgoingToken: str
+	slackBotToken: str
+	slackOutgoingWebhook: str
+	slackOutgoingToken: str
+	sharesight_auth: dict
+	webhooks: dict # service -> outbound URL: Telegram's Bot API base, the Slack and Discord incoming webhooks
+
+	@property
+	def preferences_path(self):
+		return self.data_dir + '/' + PREFERENCES_FILE
+
+
 def startup_values(getenv):
-	"""The settings fixed when the bot starts (finbot.env), computed from getenv(name, default)."""
+	"""The settings fixed when the bot starts (finbot.env), computed from getenv(name, default). Raises ConfigError."""
 	r = _Reader(getenv)
 	data_dir = (getenv('DATA_DIR') or (getenv('STATE_DIRECTORY') or '').split(':')[0] or 'var').rstrip('/') or '/'
-	values = {
-		'config_data_dir': data_dir,
-		'config_sharesight_auth': {
-			"grant_type": 'client_credentials',
-			"code": getenv('sharesight_code'),
-			"client_id": getenv('sharesight_client_id'),
-			"client_secret": getenv('sharesight_client_secret'),
-			"redirect_uri": 'urn:ietf:wg:oauth:2.0:oob'
-		},
-		'config_scheduler': r.bool('scheduler', True), # run the scheduled reports (schedule.cron) inside bot.py
-		'config_scheduler_catchup': r.int('scheduler_catchup', 5), # minutes: a run missed while the bot was down still fires if it was due this recently
-		'config_ip': r.str('ip', '127.0.0.1'),
-		'config_port': r.int('port', 5000),
-		'config_slackBotToken': r.str('slackBotToken', False),
-		'config_slackOutgoingToken': r.str('slackOutgoingToken', False),
-		'config_slackOutgoingWebhook': r.str('slackOutgoingWebhook', False),
-		'config_telegramBotToken': r.str('telegramBotToken', False),
-		'config_telegramOutgoingWebhook': r.str('telegramOutgoingWebhook', False),
-		'config_telegramChatID': r.str('telegramChatID', False),
-		'config_telegramOutgoingToken': r.str('telegramOutgoingToken', False),
-	}
 	hooks = {}
 	if getenv('slack_webhook'):
 		hooks['slack'] = getenv('slack_webhook').rstrip('/')
@@ -187,47 +185,102 @@ def startup_values(getenv):
 		hooks['discord'] = getenv('discord_webhook').rstrip('/').removesuffix('/slack') # the native endpoint, not the Slack-compatible one
 	if getenv('telegramBotToken'):
 		hooks['telegram'] = 'https://api.telegram.org/bot' + getenv('telegramBotToken').rstrip('/') + '/'
-	values['webhooks'] = hooks
+	env = Env(
+		data_dir=data_dir,
+		cache_dir=data_dir + '/cache',
+		ip=r.str('ip', '127.0.0.1'),
+		port=r.int('port', 5000),
+		scheduler=r.bool('scheduler', True),
+		scheduler_catchup=r.int('scheduler_catchup', 5),
+		telegramChatID=r.str('telegramChatID', ''),
+		telegramBotToken=r.str('telegramBotToken', ''),
+		telegramOutgoingWebhook=r.str('telegramOutgoingWebhook', ''),
+		telegramOutgoingToken=r.str('telegramOutgoingToken', ''),
+		slackBotToken=r.str('slackBotToken', ''),
+		slackOutgoingWebhook=r.str('slackOutgoingWebhook', ''),
+		slackOutgoingToken=r.str('slackOutgoingToken', ''),
+		sharesight_auth={
+			"grant_type": 'client_credentials',
+			"code": getenv('sharesight_code'),
+			"client_id": getenv('sharesight_client_id'),
+			"client_secret": getenv('sharesight_client_secret'),
+			"redirect_uri": 'urn:ietf:wg:oauth:2.0:oob'
+		},
+		webhooks=hooks)
 	r.check()
-	return values
-
-
-def preference_values(getenv):
-	"""The preferences (preferences.conf), computed from getenv(name, default). Raises ConfigError listing every bad value."""
+	return env
+
+
+@dataclasses.dataclass(frozen=True)
+class Prefs:
+	"""The preferences: preferences.conf, which the bot reads at start and on every reload. Read them as config.prefs.<name> when
+	they are used, never into a default or a module constant, so that a reload is seen."""
+	alliterate: bool
+	cache: bool
+	cache_seconds: int
+	chunk_maxlines: int
+	country_code: str
+	debug: bool
+	demote_leveraged: bool
+	demote_volatile: bool
+	exclude_portfolios: tuple
+	future_days: int
+	graph: bool
+	http_timeout: int
+	hyperlink: bool
+	hyperlinkFooter: bool
+	hyperlinkProvider: str
+	include_portfolios: tuple
+	min_analysts: int # .buy, .sell and .forward pe list only stocks this many analysts cover
+	past_days: int
+	performance_cache_seconds: int # Sharesight performance reports (30 minutes)
+	performance_use_sharesight: bool
+	price_percent: float
+	print_headers: bool
+	shorts_percent: int
+	timezone: str
+	trades_use_yahoo: bool # set False if Yahoo breaks
+	volatility_multiplier: float
+	source: dict = dataclasses.field(default_factory=dict, compare=False, repr=False) # the file's own lines, for job processes
+
+
+def preference_values(getenv, source=None):
+	"""The preferences (preferences.conf), computed from getenv(name, default); `source` is the file's own {key: value}. Raises
+	ConfigError listing every bad value."""
 	r = _Reader(getenv)
-	values = {
-		'config_alliterate': r.bool('alliterate', False),
-		'config_cache': r.bool('cache', True),
-		'config_cache_seconds': r.int('cache_seconds', 43200),
-		'config_chunk_maxlines': r.int('chunk_maxlines', 20),
-		'config_country_code': r.str('country_code', 'AU'),
-		'debug': r.bool('debug', False),
-		'config_demote_leveraged': r.bool('demote_leveraged', True),
-		'config_demote_volatile': r.bool('demote_volatile', True),
-		'config_exclude_portfolios': r.str('exclude_portfolios', '').split(),
-		'config_future_days': r.int('future_days', 30),
-		'config_graph': r.bool('graph', True),
-		'config_http_timeout': r.int('http_timeout', 18),
-		'config_hyperlink': r.bool('hyperlink', True),
-		'config_hyperlinkFooter': r.bool('hyperlinkFooter', True),
-		'config_hyperlinkProvider': r.str('hyperlinkProvider', 'google').lower(),
-		'config_include_portfolios': r.str('include_portfolios', '').split(),
-		'config_min_analysts': r.int('min_analysts', 7), # .buy, .sell and .forward pe list only stocks this many analysts cover
-		'config_past_days': r.int('past_days', 28),
-		'config_performance_cache_seconds': r.int('performance_cache_seconds', 1800), # Sharesight performance reports (30 minutes)
-		'config_performance_use_sharesight': r.bool('performance_use_sharesight', False),
-		'config_price_percent': r.float('price_percent', 5),
-		'config_print_headers': r.bool('print_headers', False),
-		'config_shorts_percent': r.int('shorts_percent', 15),
-		'config_timezone': r.str('timezone', 'Australia/Melbourne'),
-		'config_trades_use_yahoo': r.bool('trades_use_yahoo', True), # set False if Yahoo breaks
-		'config_volatility_multiplier': r.float('volatility_multiplier', 2),
-	}
+	prefs = Prefs(
+		alliterate=r.bool('alliterate', False),
+		cache=r.bool('cache', True),
+		cache_seconds=r.int('cache_seconds', 43200),
+		chunk_maxlines=r.int('chunk_maxlines', 20),
+		country_code=r.str('country_code', 'AU'),
+		debug=r.bool('debug', False),
+		demote_leveraged=r.bool('demote_leveraged', True),
+		demote_volatile=r.bool('demote_volatile', True),
+		exclude_portfolios=tuple(r.str('exclude_portfolios', '').split()),
+		future_days=r.int('future_days', 30),
+		graph=r.bool('graph', True),
+		http_timeout=r.int('http_timeout', 18),
+		hyperlink=r.bool('hyperlink', True),
+		hyperlinkFooter=r.bool('hyperlinkFooter', True),
+		hyperlinkProvider=r.str('hyperlinkProvider', 'google').lower(),
+		include_portfolios=tuple(r.str('include_portfolios', '').split()),
+		min_analysts=r.int('min_analysts', 7),
+		past_days=r.int('past_days', 28),
+		performance_cache_seconds=r.int('performance_cache_seconds', 1800),
+		performance_use_sharesight=r.bool('performance_use_sharesight', False),
+		price_percent=r.float('price_percent', 5),
+		print_headers=r.bool('print_headers', False),
+		shorts_percent=r.int('shorts_percent', 15),
+		timezone=r.str('timezone', 'Australia/Melbourne'),
+		trades_use_yahoo=r.bool('trades_use_yahoo', True),
+		volatility_multiplier=r.float('volatility_multiplier', 2),
+		source=dict(source or {}))
 	r.check()
-	return values
+	return prefs
 
 
-PREFERENCE_NAMES = frozenset(preference_values(lambda name, default=None: default)) # the config_* names a reload can change
+PREFERENCE_KEYS = tuple(field.name for field in dataclasses.fields(Prefs) if field.compare) # preferences.conf's keys: what a reload can change
 
 
 def _getenv_for(preferences):
@@ -250,16 +303,14 @@ def _fail(e):
 
 
 try:
-	globals().update(startup_values(os.environ.get))
-	config_var_dir = config_data_dir
-	config_cache_dir = config_data_dir + '/cache'
-	preferences_path = config_data_dir + '/' + PREFERENCES_FILE
+	env = startup_values(os.environ.get)
+	preferences_path = env.preferences_path
 	_preferences = {} if os.environ.get(PASSED) else (read_preferences(preferences_path) or {})
-	globals().update(preference_values(_getenv_for(_preferences)))
+	prefs = preference_values(_getenv_for(_preferences), _preferences)
 except ConfigError as e:
 	_fail(e)
 
-for _dir in (config_var_dir, config_cache_dir):
+for _dir in (env.data_dir, env.cache_dir):
 	if os.environ.get('FINBOT_CHECK_ONLY'): # util/check_config.py, possibly run as root: must not create directories the service user then cannot write
 		break
 	try:
@@ -268,11 +319,11 @@ def _fail(e):
 		print("can't create", _dir, e)
 
 # matplotlib needs a writable config/cache dir; without a home directory (systemd DynamicUser, containers) it fails
-os.environ.setdefault('MPLCONFIGDIR', os.path.abspath(config_cache_dir + '/matplotlib'))
+os.environ.setdefault('MPLCONFIGDIR', os.path.abspath(env.cache_dir + '/matplotlib'))
 
 def data_file(name):
 	# user copy in the data dir wins over the shipped default in defaults/
-	user = config_var_dir + '/' + name
+	user = env.data_dir + '/' + name
 	return user if os.path.isfile(user) else str(REPO_DIR / 'defaults' / name)
 
 _data_files = {} # path -> (modification time, parsed JSON)
@@ -294,33 +345,22 @@ def data_json(name, default=None):
 
 
 def reload():
-	"""Re-read preferences.conf and apply it everywhere: this module and every module that copied a name with `from lib.config import *`.
+	"""Re-read preferences.conf and apply it: config.prefs is replaced in one assignment, so a request never sees half of a reload.
 	All or nothing. Returns one line per change; raises ConfigError (nothing applied) if the file is missing, unreadable or invalid."""
-	global _preferences
+	global prefs
 	new = read_preferences(preferences_path)
 	if new is None:
 		raise ConfigError(f'reload requested but {preferences_path} does not exist; nothing reloaded')
-	values = preference_values(_getenv_for(new))
-	changes = []
-	for name, value in values.items():
-		old = globals()[name]
-		if old == value:
-			continue
-		changes.append(f"{name.removeprefix('config_')} {old!r} -> {value!r}")
-		for module in list(sys.modules.values()):
-			namespace = getattr(module, '__dict__', None)
-			try:
-				if namespace is not None and name in namespace and namespace[name] == old:
-					namespace[name] = value
-			except Exception:
-				pass
-	_preferences = new
+	values = preference_values(_getenv_for(new), new)
+	changes = [f"{name} {getattr(prefs, name)!r} -> {getattr(values, name)!r}" for name in PREFERENCE_KEYS
+		if getattr(prefs, name) != getattr(values, name)]
+	prefs = values
 	return changes
 
 
 def child_environment():
 	"""For a job process: the preferences this process runs on, so it never sees a file edited since the last reload."""
-	return dict(_preferences, **{PASSED: '1'})
+	return dict(prefs.source, **{PASSED: '1'})
 
 
 def check_startup_environment():
@@ -334,7 +374,7 @@ def check_startup_environment():
 	if OLD_ENV.exists():
 		problems.append(f'{OLD_ENV} is the old single config file, which finbot no longer reads; it will not start while the file is there')
 		problems.append(f'split it with {REPO_DIR}/util/split_env.py, which writes finbot.env.new and preferences.conf.new beside it; put finbot.env '
-			f'where your launcher loads it and {PREFERENCES_FILE} in {config_data_dir}, then delete {OLD_ENV}')
+			f'where your launcher loads it and {PREFERENCES_FILE} in {env.data_dir}, then delete {OLD_ENV}')
 	if problems:
 		raise ConfigError(problems)
 
@@ -344,208 +384,15 @@ def startup_report():
 	provided = [key for key in STARTUP_KEYS if key in os.environ]
 	lines = [f'startup environment provides {len(provided)} of {len(STARTUP_KEYS)} startup keys (values hidden)']
 	if os.path.isfile(preferences_path):
-		lines.append(f'preferences from {preferences_path} ({len(_preferences)} set; reload with SIGHUP)')
+		lines.append(f'preferences from {preferences_path} ({len(prefs.source)} set; reload with SIGHUP)')
 	else:
-		lines.append(f'no {PREFERENCES_FILE} in {config_data_dir}, using the defaults; reload will have nothing to re-read')
+		lines.append(f'no {PREFERENCES_FILE} in {env.data_dir}, using the defaults; reload will have nothing to re-read')
 	return lines
 
 
 def startup_warnings():
 	"""Problems in finbot.env that do not stop the start, a line each: a setting that leaves part of the bot unable to work."""
 	warnings = []
-	if config_slackOutgoingWebhook and not config_slackBotToken:
+	if env.slackOutgoingWebhook and not env.slackBotToken:
 		warnings.append('slackOutgoingWebhook is set but slackBotToken is not: Slack commands will get no reply')
 	return warnings
-
-
-adjectives = [
-	'absolutely',
-	'amazingly',
-	'awfully',
-	'bitchingly',
-	'bloodly',
-	'comfortably',
-	'darned',
-	'distinctly',
-	'especially',
-	'ever so',
-	'enjoyably',
-	'exceedingly',
-	'exceptionally',
-	'extraordinarily',
-	'extremely',
-	'flatteringly',
-	'fortunately',
-	'fortuitously',
-	'greatly',
-	'honourably',
-	'hugely',
-	'humblingly',
-	'inordinately',
-	'jolly',
-	'joyfully',
-	'keenly',
-	'kindly',
-	'knowingly',
-	'luckily',
-	'memorably',
-	'mostly',
-	'needlessly',
-	'overwhelmingly',
-	'pleasurably',
-	'questionably',
-	'rewardingly',
-	'serendipitously',
-	'seriously',
-	'significantly',
-	'so',
-	'super',
-	'supremely',
-	'surprisingly',
-	'terribly',
-	'terrifically',
-	'thoroughly',
-	'totally',
-	'tremendously',
-	'uber',
-	'uncomprehensibly',
-	'unexceptionably',
-	'unquestionably',
-	'unmentionably',
-	'unusually',
-	'discomfortingly',
-	'incomprehensibly',
-	'very',
-	'wholesomely',
-	'xtra',
-	'yawningly',
-	'zealously'
-]
-
-adjectives_two = [
-	'amazing',
-	'bitching',
-	'comforting',
-	'comprehensible',
-	'dapper',
-	'enjoyable',
-	'exceptional',
-	'extraordinary',
-	'flattering',
-	'fortuitous',
-	'fortunate',
-	'great',
-	'grateful',
-	'honouring',
-	'huge',
-	'humbling',
-	'inordinate',
-	'joyous',
-	'keen',
-	'knowing',
-	'kind',
-	'lucky',
-	'memorable',
-	'mentionable',
-	'nice',
-	'overwhelming',
-	'pleasurable',
-	'questionable',
-	'rewarding',
-	'serendipitous',
-	'significant',
-	'special',
-	'super',
-	'terrific',
-	'tremendous',
-	'useful',
-	'virtuous',
-	'wholesome',
-	'xenial',
-	'yawning',
-	'zany'
-]
-
-verb = [
-	'pretend to greet',
-	'apparently share this moment with',
-	'coincide in temporal reality with',
-	'cross digital paths with',
-	'simulate becoming acquainted with',
-	'fire photons at',
-	'traverse cyberspace with',
-	'co-exist in spacetime with',
-	'fire electrons at',
-	'encode character sets with',
-	'convert utf-8 to binary and then back to utf-8 with',
-	'update this pixel matrix with',
-	'lose money with',
-	'maintain character with',
-	'act like I comprehend'
-]
-
-noTradesVerb = [
-	"The money is probably resting in another account",
-	"Oh well. The market would have just gone down anyway",
-	"Heating bills don't pay themselves you know",
-	"You were fearful when others were also fearful",
-	"I bet you're regretting that extra 1¢ on your limit order now",
-	"This is why you can't have nice things",
-	"But I'm sure that's just a rounding error",
-	"Nothing + nothing = more nothing. Well played",
-	"Or maybe there was. They don't pay me for this you know",
-	"I guess that's right. I'm kinda busy with ChatGPT in another window",
-	"I like this contrarian play you're having on 'being in it to win it'",
-	"With your stock picking skills, this is probably for the best"
-]
-
-primary_listing = data_json('finbot_adr.json', {}) # ADR -> its primary listing
-
-yahoo_country = {
-	'Argentina': 'BA',
-	'Austria': 'VI',
-	'Australia': 'AX',
-	'Belgium': 'BR',
-	'Brazil': 'SA',
-	'Canada': 'TO',
-	'Chile': 'SN',
-	'Colombia': 'CL',
-	'Czech Republic': 'PR',
-	'Denmark': 'CO',
-	'Estonia': 'TL',
-	'Finland': 'HE',
-	'France': 'PA',
-	'Germany': 'DE',
-	'Hungary': 'BD',
-	'Iceland': 'IC',
-	'India': 'NS',
-	'Indonesia': 'JK',
-	'Ireland': 'IR',
-	'Israel': 'TA',
-	'Italy': 'MI',
-	'Latvia': 'RG',
-	'Lithuania': 'VS',
-	'Mexico': 'MX',
-	'Netherlands': 'AS',
-	'New Zealand': 'NZ',
-	'Norway': 'OL',
-	'Portugal': 'LS',
-	'Qatar': 'QA',
-	'Saudi Arabia': 'SR',
-	'South Africa': 'JO',
-	'Spain': 'MC',
-	'Sweden': 'ST',
-	'Switzerland': 'SW',
-	'Thailand': 'BK',
-	'Turkey': 'IS',
-	'United Kingdom': 'L',
-	'Venezuela': 'CR'
-	# markets with numeric tickers purposefully omitted
-}
-
-# `from lib.config import *` is how the other modules get their config_* names; keep the loader machinery out of their namespaces
-_INTERNAL = {'difflib', 're', 'sys', 'ConfigError', 'key_problem', 'parse_assignments', 'read_text', 'read_preferences', 'read_startup_file',
-	'startup_values', 'preference_values', 'reload', 'child_environment', 'print_errors', 'check_startup_environment', 'startup_report', 'preferences_path',
-	'startup_warnings', 'STARTUP_KEYS', 'PREFERENCE_KEYS', 'SECRET_KEYS', 'PREFERENCES_FILE', 'PASSED', 'PREFERENCE_NAMES', 'OLD_ENV',
-	'TRUE_WORDS', 'FALSE_WORDS'}
-__all__ = [name for name in dir() if not name.startswith('_') and name not in _INTERNAL]
diff --git a/lib/jobs.py b/lib/jobs.py
index 7b67dbf7..602a9ed2 100644
--- a/lib/jobs.py
+++ b/lib/jobs.py
@@ -20,7 +20,6 @@
 from gevent import subprocess
 from croniter import croniter, CroniterBadDateError
 
-from lib.config import *
 from lib import config
 from lib import jobrun
 from lib import util
@@ -47,7 +46,7 @@ def run_child(label, module, kwargs):
 	"""Run a command in its own process (python -m lib.jobrun) and return its exit status. The child inherits the environment,
 	user and output of the bot, so it behaves like the old cron job. A timeout or any other interruption stops the child."""
 	env = dict(os.environ, **config.child_environment()) # the preferences this process runs on, not whatever the file says now
-	proc = subprocess.Popen([sys.executable, '-m', 'lib.jobrun', module.__name__, json.dumps(kwargs)], cwd=REPO_DIR, env=env)
+	proc = subprocess.Popen([sys.executable, '-m', 'lib.jobrun', module.__name__, json.dumps(kwargs)], cwd=config.REPO_DIR, env=env)
 	try:
 		return proc.wait()
 	except BaseException:
@@ -240,9 +239,9 @@ class Scheduler:
 	"""Fires the jobs of a schedule file. `path` is a callable returning the file (the data dir copy wins over the default, so a
 	new copy is found on reload). `clock`, `sleep` and `spawn` can be replaced, which the tests do."""
 
-	def __init__(self, path, catchup=config_scheduler_catchup, clock=None, sleep=None, spawn=None, zone=None, runner=None):
+	def __init__(self, path, catchup=None, clock=None, sleep=None, spawn=None, zone=None, runner=None):
 		self.path = path
-		self.catchup = datetime.timedelta(minutes=catchup)
+		self.catchup = datetime.timedelta(minutes=config.env.scheduler_catchup if catchup is None else catchup)
 		self.clock = clock or (lambda: datetime.datetime.now(datetime.timezone.utc))
 		self.sleep = sleep or gevent.sleep
 		self.spawn = spawn or gevent.spawn
diff --git a/lib/reports.py b/lib/reports.py
index 01f8dc45..62a4a40a 100755
--- a/lib/reports.py
+++ b/lib/reports.py
@@ -2,7 +2,7 @@
 import hashlib, re
 import datetime
 import sys
-from lib.config import *
+from lib import config
 from lib import sharesight
 from lib import util
 from lib import webhook
@@ -165,7 +165,7 @@ def prepare_compare(service, args):
 	spec = period_days or ('max' if full_history else 'default')
 	cache_key = hashlib.sha1(f"{service}|{','.join(tickers)}|{spec}".encode()).hexdigest()[:16]
 	image_cache_file, caption_cache_file = f"finbot_compare_{cache_key}.png", f"finbot_compare_{cache_key}.json"
-	if config_cache:
+	if config.prefs.cache:
 		cached_image = util.read_binary_cache(image_cache_file)
 		cached_caption = util.read_cache(caption_cache_file)
 		if cached_image and cached_caption:
@@ -209,7 +209,7 @@ def prepare_compare(service, args):
 	for pct, t in reversed(results):
 		link = util.finance_link(t, exchanges[t], service, days=span_days, brief=False)
 		caption.append(f"{util.get_emoji(pct)} {names[t]} ({link}): {pct:+,.1f}%")
-	if config_cache:
+	if config.prefs.cache:
 		util.write_binary_cache(image_cache_file, image)
 		util.json_write(caption_cache_file, caption)
 		image.seek(0)
@@ -386,7 +386,7 @@ def prepare_rating_payload(service, action, length=15):
 	rating first (1 is a strong buy, 5 a strong sell); on equal ratings, the one more analysts give first."""
 	found = [] # (rating, analysts, line, ticker)
 	for ticker, data in yahoo.fetch_details(util.get_holdings_and_watchlist()).items():
-		if 'recommend' not in data or data['recommend_analysts'] < config_min_analysts:
+		if 'recommend' not in data or data['recommend_analysts'] < config.prefs.min_analysts:
 			continue
 		recommend = data['recommend'].replace('_', ' ')
 		if (action == 'buy' and 'buy' in recommend) or (action == 'sell' and recommend in ('sell', 'underperform')):
@@ -396,8 +396,8 @@ def prepare_rating_payload(service, action, length=15):
 	found = sorted(found, key=lambda row: (row[0] if action == 'buy' else -row[0], -row[1]))[:length]
 	payload = [line for _, _, line, _ in found]
 	if not payload:
-		return [f"No tracked stocks rated {action} by {config_min_analysts}+ analysts"], None
-	message = f"Top {length} analyst {action} ratings for tracked stocks ({config_min_analysts}+ analysts)"
+		return [f"No tracked stocks rated {action} by {config.prefs.min_analysts}+ analysts"], None
+	message = f"Top {length} analyst {action} ratings for tracked stocks ({config.prefs.min_analysts}+ analysts)"
 	shown = [(ticker.split('.')[0], rating) for rating, _, _, ticker in found]
 	low, high = min(r for _, r in shown), max(r for _, r in shown)
 	spread = (high - low) or 0.2
@@ -417,7 +417,7 @@ def prepare_value_payload(service, action='pe', ticker_select=None, length=15):
 		market_data |= yahoo.fetch_details(market_data, strict=bool(ticker_select))
 	covered = None # for a forward P/E list: the tickers enough analysts cover
 	if 'forward' in action and not ticker_select:
-		covered = {ticker for ticker, detail in yahoo.fetch_details(market_data).items() if detail.get('recommend_analysts', 0) >= config_min_analysts}
+		covered = {ticker for ticker, detail in yahoo.fetch_details(market_data).items() if detail.get('recommend_analysts', 0) >= config.prefs.min_analysts}
 	for ticker in market_data:
 		if covered is not None and ticker not in covered:
 			continue
@@ -476,9 +476,9 @@ def prepare_value_payload(service, action='pe', ticker_select=None, length=15):
 			payload.reverse()
 			chart_rows.reverse()
 		payload = payload[:length]
-		coverage = f" ({config_min_analysts}+ analysts)" if covered is not None else ''
+		coverage = f" ({config.prefs.min_analysts}+ analysts)" if covered is not None else ''
 		if not payload:
-			payload = [f"No tracked stocks with {config_min_analysts}+ analysts have a {heading_trail} ratio" if covered is not None else f"{action}: no data found"]
+			payload = [f"No tracked stocks with {config.prefs.min_analysts}+ analysts have a {heading_trail} ratio" if covered is not None else f"{action}: no data found"]
 		else:
 			heading = f'{heading_type} {length} tracked stocks by {heading_trail} ratio{coverage}'
 			payload.insert(0, f"{webhook.bold(heading, service)}")
@@ -668,7 +668,7 @@ def prepare_profile_payload(service, user, ticker):
 				payload.append(f"{util.get_emoji(data[key])} {webhook.bold(label, service)} {data[key]:,}%")
 				break
 
-	if 'profile_website' in data and config_hyperlinkFooter and config_hyperlink:
+	if 'profile_website' in data and config.prefs.hyperlinkFooter and config.prefs.hyperlink:
 		if payload:
 			payload.append("")
 		payload.append(f"{gfinance_link} | {util.yahoo_link(ticker, service, brief=True, text='yahoo')} | {sws_link}")
diff --git a/lib/sharesight.py b/lib/sharesight.py
index a39caf34..8dd9c159 100644
--- a/lib/sharesight.py
+++ b/lib/sharesight.py
@@ -3,7 +3,7 @@
 import datetime
 import threading
 import requests
-from lib.config import *
+from lib import config
 from lib import net
 import lib.util as util
 
@@ -21,7 +21,7 @@ def _post(url, **kwargs):
 def _request(send, url, what, **kwargs):
 	"""Send a request to Sharesight and return the JSON answer. A request that fails raises RuntimeError naming `what`."""
 	try:
-		r = send(url, timeout=config_http_timeout, **kwargs)
+		r = send(url, timeout=config.prefs.http_timeout, **kwargs)
 	except Exception as e:
 		print("Error", net.redact(e), url, file=sys.stderr)
 		raise RuntimeError(f"Sharesight {what} request failed: {e}")
@@ -43,27 +43,27 @@ def __call__(self, r):
 		return r
 
 def get_token():
-	if config_cache:
+	if config.prefs.cache:
 		cacheFile = "finbot_sharesight_token.json"
 		cache = util.json_load(cacheFile)
 		if cache:
 			cache_expiry = datetime.datetime.fromtimestamp(cache['created_at'] + cache['expires_in'])
 			ttl = cache_expiry - datetime.datetime.now()
 			if ttl > datetime.timedelta(seconds=20):
-				print("cache hit:", cacheFile, "TTL:", util.td_to_human(ttl), file=sys.stderr) if debug else None
+				print("cache hit:", cacheFile, "TTL:", util.td_to_human(ttl), file=sys.stderr) if config.prefs.debug else None
 				return cache['access_token']
 			else:
-				print("cache miss:", cacheFile, file=sys.stderr) if debug else None
+				print("cache miss:", cacheFile, file=sys.stderr) if config.prefs.debug else None
 
 	print("Fetching Sharesight auth token")
-	data = _request(_post, "https://api.sharesight.com/oauth2/token", 'auth', data=config_sharesight_auth)
+	data = _request(_post, "https://api.sharesight.com/oauth2/token", 'auth', data=config.env.sharesight_auth)
 	_check(data, 'auth')
-	if config_cache and 'access_token' in data:
+	if config.prefs.cache and 'access_token' in data:
 		util.json_write(cacheFile, data)
-	print("Got a Sharesight token", file=sys.stderr) if debug else None
+	print("Got a Sharesight token", file=sys.stderr) if config.prefs.debug else None
 	return data['access_token']
 
-PORTFOLIOS_CACHE = "finbot_sharesight_portfolios.json" # ttl: config_cache_seconds
+PORTFOLIOS_CACHE = "finbot_sharesight_portfolios.json" # ttl: the cache_seconds preference
 
 def performance_cache_file(portfolio_id, days):
 	return "finbot_sharesight_performance_" + str(portfolio_id) + "_" + str(days) + '.json'
@@ -71,18 +71,18 @@ def performance_cache_file(portfolio_id, days):
 def get_portfolios(refresh=False):
 	"""refresh=True skips the cache read and fetches again (the cache is rewritten)."""
 	portfolio_dict = {}
-	data = util.read_cache(PORTFOLIOS_CACHE, config_cache_seconds) if config_cache and not refresh else None
+	data = util.read_cache(PORTFOLIOS_CACHE, config.prefs.cache_seconds) if config.prefs.cache and not refresh else None
 	if not data:
 		token = get_token()
 		print("Fetching Sharesight portfolios")
 		data = _request(_get, "https://api.sharesight.com/api/v3/portfolios", 'portfolios', headers={'Content-type': 'application/json'}, auth=BearerAuth(token))
 		_check(data, 'portfolios')
-		if config_cache and 'portfolios' in data:
+		if config.prefs.cache and 'portfolios' in data:
 			util.json_write(PORTFOLIOS_CACHE, data)
 	for portfolio in data['portfolios']:
-		if str(portfolio['id']) in config_exclude_portfolios:
+		if str(portfolio['id']) in config.prefs.exclude_portfolios:
 			print(portfolio['id'], "(" + portfolio['name'] + ") in exclusion list. Skipping.")
-		elif str(portfolio['id']) not in config_include_portfolios and config_include_portfolios:
+		elif str(portfolio['id']) not in config.prefs.include_portfolios and config.prefs.include_portfolios:
 			print("Exclusion list is defined and does not contain", portfolio['id'], "(" + portfolio['name'] + "). Skipping.")
 		else:
 			portfolio_dict[portfolio['name']] = portfolio['id']
@@ -109,10 +109,10 @@ def pick_portfolios(name=None, portfolios=None):
 		raise PortfolioNotFound(name, portfolios)
 	return dict([found])
 
-def get_trades(portfolio_name, portfolio_id, days=UNSET):
-	if days is UNSET:
-		days = config_past_days
-	if config_cache:
+def get_trades(portfolio_name, portfolio_id, days=None):
+	if days is None:
+		days = config.prefs.past_days
+	if config.prefs.cache:
 		cache_file = "finbot_sharesight_trades_" + str(portfolio_id) + "_" + str(days) + ".json"
 		cache = util.read_cache(cache_file, 299) # max freq 5 min
 		if cache:
@@ -128,7 +128,7 @@ def get_trades(portfolio_name, portfolio_id, days=UNSET):
 	_check(data, 'trades')
 	for trade in data['trades']:
 		trade['portfolio'] = portfolio_name # inject custom field
-	if config_cache and 'trades' in data:
+	if config.prefs.cache and 'trades' in data:
 		util.json_write(cache_file, data)
 	return data['trades']
 
@@ -165,14 +165,14 @@ def get_holdings_wrapper():
 		tickers.update(get_holdings(portfolio_name, portfolio_id))
 	return sorted(tickers)
 
-def get_performance(portfolio_id, days, config_cache_seconds=UNSET, refresh=False):
-	if config_cache_seconds is UNSET:
-		config_cache_seconds = config_performance_cache_seconds
+def get_performance(portfolio_id, days, seconds=None, refresh=False):
+	if seconds is None:
+		seconds = config.prefs.performance_cache_seconds
 	start_date = datetime.datetime.now() - datetime.timedelta(days=days)
 	start_date = start_date.strftime('%Y-%m-%d') # 2023-04-25
 	cache_file = performance_cache_file(portfolio_id, days)
-	if config_cache and not refresh:
-		cache = util.read_cache(cache_file, config_cache_seconds)
+	if config.prefs.cache and not refresh:
+		cache = util.read_cache(cache_file, seconds)
 		if cache:
 			return cache
 	token = get_token()
@@ -186,13 +186,13 @@ def get_performance(portfolio_id, days, config_cache_seconds=UNSET, refresh=Fals
 			continue
 		break
 	_check(data, 'performance')
-	if config_cache and 'report' in data:
+	if config.prefs.cache and 'report' in data:
 		util.json_write(cache_file, data)
 	return data
 
-def get_performance_wrapper(days=UNSET):
-	if days is UNSET:
-		days = config_past_days
+def get_performance_wrapper(days=None):
+	if days is None:
+		days = config.prefs.past_days
 	performance = {}
 	portfolios = get_portfolios()
 	for portfolio_name, portfolio_id in portfolios.items():
@@ -208,14 +208,14 @@ def warm(margin=0.9):
 	the portfolios on instead of reading the cached list again."""
 	fetched = 0
 	age = util.cache_age(PORTFOLIOS_CACHE)
-	if age is None or age >= config_cache_seconds * margin:
+	if age is None or age >= config.prefs.cache_seconds * margin:
 		portfolios = get_portfolios(refresh=True)
 		fetched += 1
 	else:
 		portfolios = get_portfolios()
 	for portfolio_name, portfolio_id in portfolios.items():
 		age = util.cache_age(performance_cache_file(portfolio_id, 0))
-		if age is None or age >= margin * config_performance_cache_seconds:
+		if age is None or age >= margin * config.prefs.performance_cache_seconds:
 			get_performance(portfolio_id, 0, refresh=True)
 			fetched += 1
 	return fetched, portfolios
@@ -225,10 +225,10 @@ def next_due(margin=0.9, portfolios=None):
 	used; without it the cached list is read."""
 	waits = []
 	age = util.cache_age(PORTFOLIOS_CACHE)
-	waits.append(0 if age is None else margin * config_cache_seconds - age)
+	waits.append(0 if age is None else margin * config.prefs.cache_seconds - age)
 	for portfolio_name, portfolio_id in (get_portfolios() if portfolios is None else portfolios).items():
 		age = util.cache_age(performance_cache_file(portfolio_id, 0))
-		waits.append(0 if age is None else margin * config_performance_cache_seconds - age)
+		waits.append(0 if age is None else margin * config.prefs.performance_cache_seconds - age)
 	return max(0, min(waits))
 
 _wake = threading.Event() # set by wake(): the caches were flushed, refill them now
@@ -242,7 +242,7 @@ def keep_warm(min_wait=5, retry=300, idle=60, stop=None):
 	While the cache preference is off (a reload can switch it either way) it fetches nothing, since nothing would be written and
 	everything would always be due, and looks again every `idle` seconds. Never raises; after a failure it waits `retry` seconds."""
 	while not (stop and stop.is_set()):
-		if not config_cache:
+		if not config.prefs.cache:
 			wait = idle
 		else:
 			try:
diff --git a/lib/shortman.py b/lib/shortman.py
index e1f1a4c4..489fb0e8 100644
--- a/lib/shortman.py
+++ b/lib/shortman.py
@@ -1,5 +1,5 @@
 import sys
-from lib.config import *
+from lib import config
 from lib import net
 
 def fetch(market_data):
@@ -7,7 +7,7 @@ def fetch(market_data):
 	print("Fetching ASX shorts from Shortman")
 	url = 'https://www.shortman.com.au/downloadeddata/latest.csv'
 	try:
-		r = net.get(url, timeout=config_http_timeout)
+		r = net.get(url, timeout=config.prefs.http_timeout)
 	except Exception as e:
 		print("Failure fetching", url, net.redact(e), file=sys.stderr)
 		return market_data # the ASX short interest is missing, the rest of the report is not
diff --git a/lib/simplywallst.py b/lib/simplywallst.py
index c49388d1..206ea1ed 100644
--- a/lib/simplywallst.py
+++ b/lib/simplywallst.py
@@ -1,11 +1,11 @@
-from lib.config import data_json
+from lib import config
 
 FILES = ['finbot_sws_custom.json', 'finbot_sws_asx200.json', 'finbot_sws_sp500.json', 'finbot_sws_nasdaq100.json'] # the first to list a ticker wins
 
 def get_url(symbol, name, exchange):
 	"""The ticker's page on simplywall.st, from the lists in FILES; without one, a search that lands on it."""
 	for file in FILES:
-		url = data_json(file, {}).get(symbol)
+		url = config.data_json(file, {}).get(symbol)
 		if url:
 			return url
 	return 'https://www.google.com/search?q=site:simplywall.st+(' + name + '+' + exchange + ':' + symbol.split('.')[0] + ')+Stock&btnI'
diff --git a/lib/telegram.py b/lib/telegram.py
index c918aca5..908396ca 100644
--- a/lib/telegram.py
+++ b/lib/telegram.py
@@ -1,11 +1,11 @@
 import json
 import sys
-from lib.config import *
+from lib import config
 from lib import net
 
 def call(method, **kwargs):
 	"""POST a Bot API method and return its result. Raises RuntimeError with Telegram's description when it answers not ok."""
-	response = net.post(webhooks['telegram'] + method, timeout=config_http_timeout, **kwargs)
+	response = net.post(config.env.webhooks['telegram'] + method, timeout=config.prefs.http_timeout, **kwargs)
 	try:
 		output = response.json()
 	except ValueError:
@@ -22,12 +22,12 @@ def try_call(method, **kwargs):
 		print(f"Telegram {method} failed:", net.redact(e), file=sys.stderr)
 
 def setWebhook():
-	print("registering", config_telegramOutgoingWebhook, file=sys.stderr)
-	params = {'url': config_telegramOutgoingWebhook, "allowed_updates": json.dumps(['message', 'callback_query']), 'secret_token': config_telegramOutgoingToken}
+	print("registering", config.env.telegramOutgoingWebhook, file=sys.stderr)
+	params = {'url': config.env.telegramOutgoingWebhook, "allowed_updates": json.dumps(['message', 'callback_query']), 'secret_token': config.env.telegramOutgoingToken}
 	return call('setWebhook', params=params)
 
 def delWebhook():
-	print("deregistering", config_telegramOutgoingWebhook, file=sys.stderr)
+	print("deregistering", config.env.telegramOutgoingWebhook, file=sys.stderr)
 	return call('setWebhook', params={'url': ''}) # unsubscribe
 
 def getMe():
diff --git a/lib/util.py b/lib/util.py
index 8a65736a..9e6a9c6b 100644
--- a/lib/util.py
+++ b/lib/util.py
@@ -8,7 +8,7 @@
 import sys
 import tempfile
 import time
-from lib.config import * # first: sets MPLCONFIGDIR before matplotlib is imported (lazily, by the chart functions)
+from lib import config # first: sets MPLCONFIGDIR before matplotlib is imported (lazily, by the chart functions)
 from lib import sharesight
 
 
@@ -88,6 +88,48 @@ def transform_title(title):
 	'WAR': ('🇵🇱', 'PLN'),
 }
 
+YAHOO_COUNTRY = { # country -> its Yahoo ticker suffix, for a listing that has none (cal.py)
+	'Argentina': 'BA',
+	'Austria': 'VI',
+	'Australia': 'AX',
+	'Belgium': 'BR',
+	'Brazil': 'SA',
+	'Canada': 'TO',
+	'Chile': 'SN',
+	'Colombia': 'CL',
+	'Czech Republic': 'PR',
+	'Denmark': 'CO',
+	'Estonia': 'TL',
+	'Finland': 'HE',
+	'France': 'PA',
+	'Germany': 'DE',
+	'Hungary': 'BD',
+	'Iceland': 'IC',
+	'India': 'NS',
+	'Indonesia': 'JK',
+	'Ireland': 'IR',
+	'Israel': 'TA',
+	'Italy': 'MI',
+	'Latvia': 'RG',
+	'Lithuania': 'VS',
+	'Mexico': 'MX',
+	'Netherlands': 'AS',
+	'New Zealand': 'NZ',
+	'Norway': 'OL',
+	'Portugal': 'LS',
+	'Qatar': 'QA',
+	'Saudi Arabia': 'SR',
+	'South Africa': 'JO',
+	'Spain': 'MC',
+	'Sweden': 'ST',
+	'Switzerland': 'SW',
+	'Thailand': 'BK',
+	'Turkey': 'IS',
+	'United Kingdom': 'L',
+	'Venezuela': 'CR'
+	# markets with numeric tickers purposefully omitted
+}
+
 def flag_from_market(market):
 	if market in MARKETS:
 		return MARKETS[market][0]
@@ -104,20 +146,20 @@ def currency_from_market(market):
 
 def cache_path(name, persist=False):
 	"""Where a cache file lives, or with persist=True a state file (in the data directory)."""
-	return (config_var_dir if persist else config_cache_dir) + "/" + name
+	return (config.env.data_dir if persist else config.env.cache_dir) + "/" + name
 
 def _fresh_cache(cacheFile, maxSeconds):
-	"""The cache file's path if it was written less than maxSeconds ago (config_cache_seconds when UNSET), else None."""
-	if maxSeconds is UNSET:
-		maxSeconds = config_cache_seconds
+	"""The cache file's path if it was written less than maxSeconds ago (the cache_seconds preference when None), else None."""
+	if maxSeconds is None:
+		maxSeconds = config.prefs.cache_seconds
 	path, age = cache_path(cacheFile), cache_age(cacheFile)
 	if age is None:
-		print("cache miss:", path, file=sys.stderr) if debug else None
+		print("cache miss:", path, file=sys.stderr) if config.prefs.debug else None
 		return None
 	if age >= maxSeconds:
-		print("cache expired:", path, file=sys.stderr) if debug else None
+		print("cache expired:", path, file=sys.stderr) if config.prefs.debug else None
 		return None
-	print("cache hit:", path, "TTL:", td_to_human(datetime.timedelta(seconds=maxSeconds - age)), file=sys.stderr) if debug else None
+	print("cache hit:", path, "TTL:", td_to_human(datetime.timedelta(seconds=maxSeconds - age)), file=sys.stderr) if config.prefs.debug else None
 	return path
 
 def _read_fresh(cacheFile, maxSeconds, read):
@@ -130,7 +172,7 @@ def _read_fresh(cacheFile, maxSeconds, read):
 	except FileNotFoundError:
 		return None
 
-def read_cache(cacheFile, maxSeconds=UNSET):
+def read_cache(cacheFile, maxSeconds=None):
 	"""A cache file's JSON, or None when it is missing or older than maxSeconds."""
 	return _read_fresh(cacheFile, maxSeconds, lambda path: json.loads(path.read_text(encoding="utf-8")))
 
@@ -165,7 +207,7 @@ def write_file(path, data, sync=False):
 def json_write(filename, data, persist=False):
 	"""A cache file, or with persist=True a state file in the data directory (also flushed to disk). See write_file."""
 	path = cache_path(filename, persist)
-	if debug:
+	if config.prefs.debug:
 		print("cache write:", path, file=sys.stderr)
 	write_file(path, json.dumps(data, indent=4).encode('utf-8'), sync=persist)
 
@@ -173,7 +215,7 @@ def json_load(filename, persist=False):
 	path = Path(cache_path(filename, persist))
 	return json.loads(path.read_text(encoding="utf-8")) if path.is_file() else None
 
-def read_binary_cache(cacheFile, maxSeconds=UNSET):
+def read_binary_cache(cacheFile, maxSeconds=None):
 	"""A cache file's bytes (as a file object), or None when it is missing or older than maxSeconds."""
 	return _read_fresh(cacheFile, maxSeconds, lambda path: io.BytesIO(path.read_bytes()))
 
@@ -201,7 +243,7 @@ def security_line(flag, title, link):
 def link(url, text, service='telegram'):
 	"""text linking to url in the service's markup: HTML on Telegram, <url|text> on Slack, a masked link on Discord (the angle
 	brackets keep its preview away; brackets in the text and parentheses in the url would end it early)."""
-	if not config_hyperlink:
+	if not config.prefs.hyperlink:
 		return text
 	if service == 'telegram':
 		return '<a href="' + url + '">' + text + '</a>'
@@ -212,7 +254,7 @@ def link(url, text, service='telegram'):
 	return text
 
 def finance_link(symbol, exchange, service='telegram', days=1, brief=True, text=None):
-	if config_hyperlinkProvider == 'google':
+	if config.prefs.hyperlinkProvider == 'google':
 		return gfinance_link(symbol, exchange, service, days, brief, text)
 	return yahoo_link(symbol, service, brief, text)
 
@@ -252,7 +294,7 @@ def gfinance_link(symbol, exchange, service='telegram', days=1, brief=True, text
 			text = ticker
 		else:
 			text = symbol # US
-	if service in {'telegram', 'discord', 'slack'} and config_hyperlink:
+	if service in {'telegram', 'discord', 'slack'} and config.prefs.hyperlink:
 		return link(url + ticker + '?window=' + window, text, service)
 	return symbol
 
@@ -380,8 +422,10 @@ def render_off_hub():
 	deadlock matplotlib under gevent's patched locks. matplotlib is imported here, at start, instead of by the first chart. Job
 	processes and the tests draw in place."""
 	global _render_pool
+	import importlib
 	from gevent.threadpool import ThreadPool
-	import matplotlib.backends.backend_agg, matplotlib.dates, matplotlib.figure, matplotlib.patches, matplotlib.path, matplotlib.ticker # noqa: F401
+	for module in ('matplotlib.backends.backend_agg', 'matplotlib.dates', 'matplotlib.figure', 'matplotlib.patches', 'matplotlib.path', 'matplotlib.ticker'):
+		importlib.import_module(module)
 	_render_pool = ThreadPool(1)
 
 def _drawn(draw):
diff --git a/lib/webhook.py b/lib/webhook.py
index ba5ce1bc..b420d517 100644
--- a/lib/webhook.py
+++ b/lib/webhook.py
@@ -6,7 +6,7 @@
 import sys
 import traceback
 
-from lib.config import *
+from lib import config
 from lib import net
 from lib import telegram
 
@@ -28,10 +28,10 @@ def write(service, url, payload_string, chat_id=None, reply_markup=None):
 		payload['unfurl_links'] = False # no link previews: Slack reads these from the body, not the headers
 		payload['unfurl_media'] = False
 		if chat_id:
-			if not config_slackBotToken:
+			if not config.env.slackBotToken:
 				print(f"ERROR slack: a reply to channel {chat_id} needs slackBotToken in finbot.env; not sent", file=sys.stderr)
 				return None
-			headers['Authorization'] = 'Bearer ' + config_slackBotToken
+			headers['Authorization'] = 'Bearer ' + config.env.slackBotToken
 			payload['channel'] = chat_id
 	elif 'api.telegram.org' in url:
 		payload['parse_mode'] = 'HTML'
@@ -40,7 +40,7 @@ def write(service, url, payload_string, chat_id=None, reply_markup=None):
 		if reply_markup:
 			payload['reply_markup'] = reply_markup # keyboard (Telegram only)
 	try:
-		r = net.post(url, headers=headers, json=payload, timeout=config_http_timeout)
+		r = net.post(url, headers=headers, json=payload, timeout=config.prefs.http_timeout)
 	except Exception as e:
 		print("Failure sending to", service + ":", net.redact(e), file=sys.stderr)
 		return None
@@ -59,14 +59,14 @@ def chat_url(service, chat_id):
 	if service == 'slack':
 		return 'https://slack.com/api/chat.postMessage'
 	if service == 'telegram':
-		return webhooks['telegram'] + 'sendMessage?chat_id=' + str(chat_id)
-	return webhooks.get(service)
+		return config.env.webhooks['telegram'] + 'sendMessage?chat_id=' + str(chat_id)
+	return config.env.webhooks.get(service)
 
 def default_chats():
 	"""Where a scheduled report goes: [(service, url, chat id)]. Telegram posts into telegramChatID (left out without one); Slack and
 	Discord post through their incoming webhooks, which take no chat id. Raises RuntimeError when there is nowhere to post."""
-	chats = [(service, chat_url(service, config_telegramChatID), config_telegramChatID) if service == 'telegram' else (service, url, None)
-		for service, url in webhooks.items() if service != 'telegram' or config_telegramChatID]
+	chats = [(service, chat_url(service, config.env.telegramChatID), config.env.telegramChatID) if service == 'telegram' else (service, url, None)
+		for service, url in config.env.webhooks.items() if service != 'telegram' or config.env.telegramChatID]
 	if not chats:
 		raise RuntimeError('nowhere to post: finbot.env sets no telegramChatID, slack_webhook or discord_webhook')
 	return chats
@@ -139,12 +139,12 @@ def payload_wrapper(service, url, payload, chat_id=None, reply_markup=None):
 	payload = [str(line) for line in payload]
 	payload_string = '\n'.join(payload)
 	print("Preparing outbound to", service, str(len(payload_string)), "bytes")
-	print("Payload: " + payload_string) if debug else None
+	print("Payload: " + payload_string) if config.prefs.debug else None
 	limit = DISCORD_LIMIT if service == 'discord' else 4096
 	if len(payload_string) <= limit:
 		return [write(service, url, payload_string, chat_id, reply_markup)]
 	print(service, f"payload is over {limit:,} bytes. Splitting.")
-	chunks = chunked(payload, limit, config_chunk_maxlines)
+	chunks = chunked(payload, limit, config.prefs.chunk_maxlines)
 	results = []
 	for idx, chunk in enumerate(chunks):
 		if idx:
@@ -222,11 +222,11 @@ def takes_charts(service, reply=True):
 	"""Whether a report to `service` can carry a chart image: never with the graph preference off; on Telegram and Discord always;
 	on Slack in a reply to a chat (`reply`), uploaded into its channel with the bot token, but not in a scheduled report, which goes
 	through the incoming webhook and takes text only. Elsewhere a report goes out as text."""
-	return config_graph and (service in ('telegram', 'discord') or (service == 'slack' and reply and bool(config_slackBotToken)))
+	return config.prefs.graph and (service in ('telegram', 'discord') or (service == 'slack' and reply and bool(config.env.slackBotToken)))
 
 def telegram_photo(chat_id, image_data, caption, reply_markup=None):
 	"""sendPhoto, the report as its caption (trimmed to fit). Telegram's answer, MAYBE_SENT, or None if it was refused."""
-	url = webhooks['telegram'] + "sendPhoto?chat_id=" + str(chat_id)
+	url = config.env.webhooks['telegram'] + "sendPhoto?chat_id=" + str(chat_id)
 	data = {
 		'disable_notification': True,
 		'chat_id': chat_id,
@@ -238,7 +238,7 @@ def telegram_photo(chat_id, image_data, caption, reply_markup=None):
 	if hasattr(image_data, 'seek'):
 		image_data.seek(0)
 	try:
-		r = net.post(url, data=data, files={"photo": ('image.png', image_data)}, timeout=config_http_timeout)
+		r = net.post(url, data=data, files={"photo": ('image.png', image_data)}, timeout=config.prefs.http_timeout)
 	except requests.exceptions.ReadTimeout as e:
 		print("Telegram sendPhoto timed out, the chart may still arrive:", net.redact(e), file=sys.stderr)
 		return MAYBE_SENT
@@ -263,8 +263,8 @@ def discord_photo(image_data, caption):
 	payload = {'content': fit_caption(caption, DISCORD_LIMIT, len), 'flags': SUPPRESS_EMBEDS, 'allowed_mentions': NO_MENTIONS,
 		'attachments': [{'id': 0, 'filename': 'chart.png'}]}
 	try:
-		r = net.post(discord_url(webhooks['discord']), data={'payload_json': json.dumps(payload)},
-			files={'files[0]': ('chart.png', image_data, 'image/png')}, timeout=config_http_timeout)
+		r = net.post(discord_url(config.env.webhooks['discord']), data={'payload_json': json.dumps(payload)},
+			files={'files[0]': ('chart.png', image_data, 'image/png')}, timeout=config.prefs.http_timeout)
 	except requests.exceptions.ReadTimeout as e:
 		print("Discord chart timed out, it may still arrive:", net.redact(e), file=sys.stderr)
 		return MAYBE_SENT
@@ -282,9 +282,9 @@ def discord_photo(image_data, caption):
 
 def slack_api(method, **data):
 	"""POST a Slack Web API method as the bot, form-encoded, and return its answer. Raises RuntimeError with Slack's error if not ok."""
-	if not config_slackBotToken:
+	if not config.env.slackBotToken:
 		raise RuntimeError(f"Slack {method}: needs slackBotToken in finbot.env")
-	r = net.post('https://slack.com/api/' + method, headers={'Authorization': 'Bearer ' + config_slackBotToken}, data=data, timeout=config_http_timeout)
+	r = net.post('https://slack.com/api/' + method, headers={'Authorization': 'Bearer ' + config.env.slackBotToken}, data=data, timeout=config.prefs.http_timeout)
 	try:
 		output = r.json()
 	except ValueError:
@@ -302,7 +302,7 @@ def slack_photo(chat_id, image_data, caption):
 	data = image_data.read() if hasattr(image_data, 'read') else bytes(image_data)
 	try:
 		ticket = slack_api('files.getUploadURLExternal', filename='chart.png', length=len(data))
-		r = net.post(ticket['upload_url'], data=data, timeout=config_http_timeout)
+		r = net.post(ticket['upload_url'], data=data, timeout=config.prefs.http_timeout)
 		if r.status_code != 200:
 			raise RuntimeError(f"Slack upload: {r.status_code} {r.text[:200]}")
 		share = {'files': json.dumps([{'id': ticket['file_id']}]), 'channel_id': chat_id, 'initial_comment': caption}
@@ -335,7 +335,7 @@ def sendPhoto(chat_id, image_data, caption, service, reply_markup=None):
 		return output
 	if chartable:
 		print("chart not sent, sending the report as text", file=sys.stderr)
-	return payload_wrapper(service, chat_url(service, chat_id) if chat_id else webhooks.get(service), caption.split('\n'), chat_id)
+	return payload_wrapper(service, chat_url(service, chat_id) if chat_id else config.env.webhooks.get(service), caption.split('\n'), chat_id)
 
 def editMessageText(chat_id, message_id, text, reply_markup=None):
 	"""Telegram: replace a text message's text and buttons in place. A failure is only logged."""
diff --git a/lib/words.py b/lib/words.py
new file mode 100644
index 00000000..265b993f
--- /dev/null
+++ b/lib/words.py
@@ -0,0 +1,142 @@
+"""The words of the easter eggs: .hello, .thanks, and the quip when there were no trades."""
+
+adjectives = [
+	'absolutely',
+	'amazingly',
+	'awfully',
+	'bitchingly',
+	'bloodly',
+	'comfortably',
+	'darned',
+	'distinctly',
+	'especially',
+	'ever so',
+	'enjoyably',
+	'exceedingly',
+	'exceptionally',
+	'extraordinarily',
+	'extremely',
+	'flatteringly',
+	'fortunately',
+	'fortuitously',
+	'greatly',
+	'honourably',
+	'hugely',
+	'humblingly',
+	'inordinately',
+	'jolly',
+	'joyfully',
+	'keenly',
+	'kindly',
+	'knowingly',
+	'luckily',
+	'memorably',
+	'mostly',
+	'needlessly',
+	'overwhelmingly',
+	'pleasurably',
+	'questionably',
+	'rewardingly',
+	'serendipitously',
+	'seriously',
+	'significantly',
+	'so',
+	'super',
+	'supremely',
+	'surprisingly',
+	'terribly',
+	'terrifically',
+	'thoroughly',
+	'totally',
+	'tremendously',
+	'uber',
+	'uncomprehensibly',
+	'unexceptionably',
+	'unquestionably',
+	'unmentionably',
+	'unusually',
+	'discomfortingly',
+	'incomprehensibly',
+	'very',
+	'wholesomely',
+	'xtra',
+	'yawningly',
+	'zealously'
+]
+
+adjectives_two = [
+	'amazing',
+	'bitching',
+	'comforting',
+	'comprehensible',
+	'dapper',
+	'enjoyable',
+	'exceptional',
+	'extraordinary',
+	'flattering',
+	'fortuitous',
+	'fortunate',
+	'great',
+	'grateful',
+	'honouring',
+	'huge',
+	'humbling',
+	'inordinate',
+	'joyous',
+	'keen',
+	'knowing',
+	'kind',
+	'lucky',
+	'memorable',
+	'mentionable',
+	'nice',
+	'overwhelming',
+	'pleasurable',
+	'questionable',
+	'rewarding',
+	'serendipitous',
+	'significant',
+	'special',
+	'super',
+	'terrific',
+	'tremendous',
+	'useful',
+	'virtuous',
+	'wholesome',
+	'xenial',
+	'yawning',
+	'zany'
+]
+
+verb = [
+	'pretend to greet',
+	'apparently share this moment with',
+	'coincide in temporal reality with',
+	'cross digital paths with',
+	'simulate becoming acquainted with',
+	'fire photons at',
+	'traverse cyberspace with',
+	'co-exist in spacetime with',
+	'fire electrons at',
+	'encode character sets with',
+	'convert utf-8 to binary and then back to utf-8 with',
+	'update this pixel matrix with',
+	'lose money with',
+	'maintain character with',
+	'act like I comprehend'
+]
+
+noTradesVerb = [
+	"The money is probably resting in another account",
+	"Oh well. The market would have just gone down anyway",
+	"Heating bills don't pay themselves you know",
+	"You were fearful when others were also fearful",
+	"I bet you're regretting that extra 1¢ on your limit order now",
+	"This is why you can't have nice things",
+	"But I'm sure that's just a rounding error",
+	"Nothing + nothing = more nothing. Well played",
+	"Or maybe there was. They don't pay me for this you know",
+	"I guess that's right. I'm kinda busy with ChatGPT in another window",
+	"I like this contrarian play you're having on 'being in it to win it'",
+	"With your stock picking skills, this is probably for the best"
+]
diff --git a/lib/worker.py b/lib/worker.py
index 193cf716..300c8d42 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -6,7 +6,8 @@
 import re
 import sys
 import threading
-from lib.config import *
+from lib import config
+from lib import words
 from lib import util
 from lib import webhook
 from lib import reports
@@ -215,11 +216,11 @@ def on_help(request, m):
 		charts.mark_keyboard(request.chat_id) # this chat now has the current keyboard
 
 def on_hello(request, m):
-	"""Easter egg: a greeting, alliterated when config_alliterate is on."""
-	verbs = verb + [f"{webhook.strike(said, request.service)} I mean {meant}" for said, meant in (('study', 'meet'), ('observe', 'see'), ('profile', 'know'))]
+	"""Easter egg: a greeting, alliterated when the alliterate preference is on."""
+	verbs = words.verb + [f"{webhook.strike(said, request.service)} I mean {meant}" for said, meant in (('study', 'meet'), ('observe', 'see'), ('profile', 'know'))]
 	while True:
-		first, second = random.choice(adjectives), random.choice(adjectives_two)
-		if not config_alliterate or first[0].lower() == second[0].lower():
+		first, second = random.choice(words.adjectives), random.choice(words.adjectives_two)
+		if not config.prefs.alliterate or first[0].lower() == second[0].lower():
 			break
 	request.say([f"{first.capitalize()} {second} to {random.choice(verbs)} you, {request.user_name}! 😇"])
 
@@ -229,11 +230,11 @@ def on_thanks(request, m):
 	"""Easter egg: you're welcome, glitching one time in a thousand and on April Fools' Day."""
 	today = datetime.date.today()
 	glitch = webhook.strike(GLITCH, request.service) if random.randrange(1, 1000) == 1 or (today.month, today.day) == (4, 1) else ''
-	request.say([f"{glitch}You're {random.choice(adjectives)} welcome, {request.user_name}! 😇"])
+	request.say([f"{glitch}You're {random.choice(words.adjectives)} welcome, {request.user_name}! 😇"])
 
 def on_calendar(request, m):
 	"""Upcoming earnings or ex-dividend dates."""
-	days, ticker = period_or_ticker(m.group('arg'), config_future_days)
+	days, ticker = period_or_ticker(m.group('arg'), config.prefs.future_days)
 	event = 'ex-dividend' if m.group('kind').lower().startswith('div') else 'earnings'
 	with request.typing():
 		cal.lambda_handler(request.chat_id, days, request.service, ticker, interactive=True, event=event)
@@ -246,14 +247,14 @@ def on_performance(request, m):
 
 def on_session(request, m):
 	"""Mid-session moves (.session), or pre-market ones (.premarket, also post-market)."""
-	threshold, ticker = percent_or_ticker(m.group('arg'), config_price_percent)
+	threshold, ticker = percent_or_ticker(m.group('arg'), config.prefs.price_percent)
 	premarket = m.group('kind').lower() != 'session'
 	with request.typing():
 		price.lambda_handler(request.chat_id, threshold, request.service, request.user, ticker, interactive=True, midsession=not premarket, premarket=premarket)
 
 def on_price(request, m):
 	""".price [percent|SYMBOL|top|bottom] [period]: a bare number is a percent, as on the other commands; a period has its unit."""
-	threshold, ticker, days, top = config_price_percent, None, None, None
+	threshold, ticker, days, top = config.prefs.price_percent, None, None, None
 	for i, arg in enumerate(m.groups()):
 		if not arg:
 			continue
@@ -271,7 +272,7 @@ def on_price(request, m):
 		price.lambda_handler(request.chat_id, threshold, request.service, request.user, ticker, interactive=True, interday=days is None, days=days, top=top)
 
 def on_shorts(request, m):
-	threshold, ticker = percent_or_ticker(m.group(1), config_shorts_percent)
+	threshold, ticker = percent_or_ticker(m.group(1), config.prefs.shorts_percent)
 	with request.typing():
 		shorts.lambda_handler(request.chat_id, threshold, ticker, request.service, interactive=True)
 
diff --git a/lib/yahoo.py b/lib/yahoo.py
index 743a2978..75f13d71 100644
--- a/lib/yahoo.py
+++ b/lib/yahoo.py
@@ -6,7 +6,7 @@
 import threading
 import zoneinfo
 import sys
-from lib.config import *
+from lib import config
 from lib import net
 from lib import util
 from http.cookies import SimpleCookie
@@ -24,7 +24,7 @@ def getCookie(maxAge=1209600): # 14 days
 			return cache[0][0] + '=' + cache[0][1]
 		url = 'https://fc.yahoo.com/' # 404, but that is ok because cookie is in response header
 		try:
-			r = net.get(url, headers=HEADERS, timeout=config_http_timeout)
+			r = net.get(url, headers=HEADERS, timeout=config.prefs.http_timeout)
 		except Exception as e:
 			print(net.redact(e), file=sys.stderr)
 			r = requests.Response() # empty: falls through to the fallback cookie below
@@ -45,13 +45,13 @@ def getCrumb(seconds=1209600): # 14 days
 	with _auth:
 		cookie = getCookie()
 		cacheFile = "finbot_yahoo_crumb.json"
-		if config_cache and (cache := util.read_cache(cacheFile, seconds)):
+		if config.prefs.cache and (cache := util.read_cache(cacheFile, seconds)):
 			return cache
 		r = _get_first(_hosts('/v1/test/getcrumb'), {**HEADERS, 'Cookie': cookie})
 		if r is None:
 			print("Exhausted Yahoo API attempts. Returning fallback crumb", file=sys.stderr)
 			return 'jkQEU8yLqxs'
-		if config_cache:
+		if config.prefs.cache:
 			util.json_write(cacheFile, r.text)
 		print("Got a new Yahoo crumb", file=sys.stderr)
 		return r.text
@@ -77,7 +77,7 @@ def _get_first(urls, headers):
 	"""The answer of the first URL that answers 200, or None when none does. Each failure is logged."""
 	for url in urls:
 		try:
-			r = net.get(url, headers=headers, timeout=config_http_timeout)
+			r = net.get(url, headers=headers, timeout=config.prefs.http_timeout)
 		except Exception as e:
 			print(net.redact(e), file=sys.stderr)
 			continue
@@ -146,7 +146,7 @@ def fetch(tickers):
 		return {}
 	tickers = sorted(set(tickers)) # de-dupe
 	cacheFile = "finbot_yahoo_" + hashlib.sha256("_".join(tickers).encode()).hexdigest() + '.json'
-	if config_cache and (cache := util.read_cache(cacheFile, 300)):
+	if config.prefs.cache and (cache := util.read_cache(cacheFile, 300)):
 		return cache
 	cookie, crumb = credentials()
 	r = _get_first(_hosts('/v7/finance/quote?crumb=' + crumb + '&symbols=' + ','.join(tickers)), {**HEADERS, 'Content-type': 'application/json', 'Cookie': cookie})
@@ -172,7 +172,7 @@ def fetch(tickers):
 		if stamps: # the first still to come, else the announced date
 			quote['earnings_date'] = next((stamp for stamp in stamps if stamp > now), stamps[0])
 		yahoo_output[item['symbol']] = quote
-	if config_cache:
+	if config.prefs.cache:
 		util.json_write(cacheFile, yahoo_output)
 	return yahoo_output
 
@@ -227,13 +227,13 @@ def fetch(tickers):
 )
 RECOMMENDATION = ('recommend', 'recommend_index', 'recommend_analysts') # all three, or none
 
-def fetch_detail(ticker, seconds=UNSET):
+def fetch_detail(ticker, seconds=None):
 	"""quoteSummary for one ticker: {ticker: {...}}, or {} when Yahoo has nothing usable (no name or no price)."""
-	if seconds is UNSET:
-		seconds = config_cache_seconds
+	if seconds is None:
+		seconds = config.prefs.cache_seconds
 	now = datetime.datetime.now()
 	cacheFile = "finbot_yahoo_detail_" + ticker + '.json'
-	data = util.read_cache(cacheFile, seconds) if config_cache else None
+	data = util.read_cache(cacheFile, seconds) if config.prefs.cache else None
 	if data:
 		print('.', sep=' ', end='', flush=True)
 	else:
@@ -245,7 +245,7 @@ def fetch_detail(ticker, seconds=UNSET):
 			print(ticker + '†', sep=' ', end='', flush=True)
 			return {} # catches some delisted stocks like "DRNA"
 		data = r.json()
-		if config_cache:
+		if config.prefs.cache:
 			util.json_write(cacheFile, data)
 	result = dig(data, 'quoteSummary', 'result', 0)
 	if result is None: # Yahoo answers result=null for unknown symbols
@@ -303,13 +303,13 @@ def fetch_detail(ticker, seconds=UNSET):
 		detail['insiderSell'], detail['insiderSellValue'] = totals['Sale']
 	return {ticker: dict(sorted(detail.items()))}
 
-def price_history(ticker, days=None, seconds=UNSET, graph=UNSET, graphCache=True):
+def price_history(ticker, days=None, seconds=None, graph=None, graphCache=True):
 	"""({period: percent change}, chart image or None): the change over `days`, else over each standard period. Raises
 	RuntimeError when the history cannot be fetched or has no prices."""
-	if graph is UNSET:
-		graph = config_graph
-	if seconds is UNSET:
-		seconds = config_cache_seconds
+	if graph is None:
+		graph = config.prefs.graph
+	if seconds is None:
+		seconds = config.prefs.cache_seconds
 	import pandas as pd
 	from pandas.tseries.offsets import BDay as businessday
 	percent_dict = {}
@@ -364,10 +364,10 @@ def price_history(ticker, days=None, seconds=UNSET, graph=UNSET, graphCache=True
 		label = "Max" if "Max" in percent_dict else "10Y"
 		title = f"{company_name} ({ticker}) {label} {util.signed_percent(percent_dict[label])} Last: {price}"
 	image_cache_file = "finbot_graph_" + ticker + "_" + str(days) + ".png"
-	if config_cache and graphCache and (image := util.read_binary_cache(image_cache_file, seconds)):
+	if config.prefs.cache and graphCache and (image := util.read_binary_cache(image_cache_file, seconds)):
 		return percent_dict, image
 	image = util.graph(df, title, stock.get('currency'))
-	if config_cache:
+	if config.prefs.cache:
 		util.write_binary_cache(image_cache_file, image)
 	return percent_dict, image
 
@@ -436,28 +436,28 @@ def prefetch_history(tickers, workers=8):
 		if isinstance(result, Exception):
 			print("prefetch failed for", ticker, result, file=sys.stderr)
 
-def fetch_chart_json(ticker, days=3665, seconds=UNSET, full=False):
+def fetch_chart_json(ticker, days=3665, seconds=None, full=False):
 	"""Yahoo's daily chart for the last `days`, or with full=True its entire history (cached separately). Raises RuntimeError when
 	Yahoo cannot be reached or refuses."""
-	if seconds is UNSET:
-		seconds = config_cache_seconds
+	if seconds is None:
+		seconds = config.prefs.cache_seconds
 	now = datetime.datetime.now()
 	cacheFile = "finbot_yahoo_history_" + ticker + ("_max" if full else "") + ".json"
-	if config_cache and (cache := util.read_cache(cacheFile, seconds)):
+	if config.prefs.cache and (cache := util.read_cache(cacheFile, seconds)):
 		return cache
 	cookie, crumb = credentials()
 	start = '0' if full else str(int((now - datetime.timedelta(days=days)).timestamp()))
 	url = f'https://query1.finance.yahoo.com/v8/finance/chart/{ticker}?period1={start}&period2={int(now.timestamp())}&interval=1d&crumb={crumb}'
-	print("Fetching", net.redact(url), file=sys.stderr) if debug else None
+	print("Fetching", net.redact(url), file=sys.stderr) if config.prefs.debug else None
 	try:
-		r = net.get(url, headers={**HEADERS, 'Content-type': 'application/json', 'Cookie': cookie}, timeout=config_http_timeout)
+		r = net.get(url, headers={**HEADERS, 'Content-type': 'application/json', 'Cookie': cookie}, timeout=config.prefs.http_timeout)
 	except Exception as e:
 		raise RuntimeError(f"Yahoo unreachable for {ticker}: {net.redact(e)}") from e
 	if r.status_code != 200:
 		raise RuntimeError(f"{r.status_code} error for {ticker} from Yahoo's chart API")
 	print('↓', sep=' ', end='', flush=True)
 	data = r.json()
-	if config_cache:
+	if config.prefs.cache:
 		util.json_write(cacheFile, data)
 	return data
 
@@ -484,7 +484,7 @@ def historic_high(ticker):
 		raise RuntimeError(f"no high/low history for {ticker}")
 	highrow = df.iloc[df['High'].argmax()]
 	lowrow = df.iloc[df['Low'].argmin()]
-	if debug:
+	if config.prefs.debug:
 		print(ticker, "High", highrow['Date'], round(highrow['High'], 2), file=sys.stderr)
 		print(ticker, "Low", lowrow['Date'], round(lowrow['Low'], 2), file=sys.stderr)
 	return round(highrow['High'], 2), round(lowrow['Low'], 2)
diff --git a/milestone.py b/milestone.py
index 5ccbd3d6..cc782bf2 100755
--- a/milestone.py
+++ b/milestone.py
@@ -2,7 +2,7 @@
 
 import sys
 from itertools import groupby
-from lib.config import *
+from lib import config
 from lib import webhook
 from lib import util
 from lib import yahoo
@@ -35,10 +35,10 @@ def prepare_milestone_payload(service, market_data):
 			emoji = util.flag_from_ticker(ticker)
 			records[ticker] = {'record_high': max(oldhigh, newhigh), 'record_low': min(oldlow, newlow)} # all-time: a 52-week high below it is no record
 			if newhigh > oldhigh:
-				print("DEBUG", ticker, newhigh, "is higher than", oldhigh) if debug else None
+				print("DEBUG", ticker, newhigh, "is higher than", oldhigh) if config.prefs.debug else None
 				payloadhigh.append(f"{emoji} {title} ({ticker_link})")
 			if newlow < oldlow:
-				print("DEBUG", ticker, newlow, "is lower than", oldlow) if debug else None
+				print("DEBUG", ticker, newlow, "is lower than", oldlow) if config.prefs.debug else None
 				payloadlow.append(f"{emoji} {title} ({ticker_link})")
 		payloadlow.sort()
 		payloadhigh.sort()
diff --git a/performance.py b/performance.py
index dc684740..577e6cfc 100755
--- a/performance.py
+++ b/performance.py
@@ -2,7 +2,7 @@
 
 import sys
 
-from lib.config import *
+from lib import config
 from lib import sharesight
 from lib import webhook
 from lib import util
@@ -31,11 +31,11 @@ def benchmark(ticker, market, text, past_days, service):
 	link = util.finance_link(ticker, market, service=service, days=past_days, brief=True, text=text)
 	return f"{util.get_emoji(percent)} {link} {util.signed_percent(percent, 2)}", percent
 
-def report(service, past_days=UNSET, portfolio_select=None, interactive=False, performance=None):
+def report(service, past_days=None, portfolio_select=None, interactive=False, performance=None):
 	"""(payload, chart) of the performance report for one service: each portfolio and the benchmarks, best first. performance:
 	gather()'s, fetched when not given."""
-	if past_days is UNSET:
-		past_days = config_past_days
+	if past_days is None:
+		past_days = config.prefs.past_days
 	if performance is None:
 		performance = gather(past_days, portfolio_select)
 	lines = [] # (percent, text), sorted into the payload below
@@ -61,9 +61,9 @@ def report(service, past_days=UNSET, portfolio_select=None, interactive=False, p
 		graph = util.rows_chart(chart_rows, "Performance", f"% change {'' if period == 'today' else 'over '}{period}", value_fmt=lambda v: util.signed_percent(v, 2))
 	return payload, graph
 
-def lambda_handler(chat_id=None, past_days=UNSET, service=None, user='', portfolio_select=None, interactive=False):
-	if past_days is UNSET:
-		past_days = config_past_days
+def lambda_handler(chat_id=None, past_days=None, service=None, user='', portfolio_select=None, interactive=False):
+	if past_days is None:
+		past_days = config.prefs.past_days
 	performance = gather(past_days, portfolio_select)
 	for service, url, chat in webhook.destinations(interactive, service, chat_id):
 		payload, graph = report(service, past_days, portfolio_select, interactive, performance)
diff --git a/price.py b/price.py
index 37be6dae..a4b95de2 100755
--- a/price.py
+++ b/price.py
@@ -3,7 +3,7 @@
 import sys
 import datetime
 
-from lib.config import *
+from lib import config
 from lib import sharesight
 from lib import webhook
 from lib import util
@@ -15,10 +15,10 @@ def gather(specific_stock=None, days=None):
 	Sharesight has none for (the watchlist) their Yahoo history, fetched together up front."""
 	tickers = [util.transform_to_yahoo(specific_stock)] if specific_stock else util.get_holdings_and_watchlist()
 	market_data = yahoo.fetch(tickers)
-	if config_demote_volatile: # trigger_bar compares market caps, in US dollars
+	if config.prefs.demote_volatile: # trigger_bar compares market caps, in US dollars
 		yahoo.add_usd_market_caps(market_data)
 	# Sharesight reports a holding's performance only for the time it was held: held for 1Y and asked for 5Y, it gives 1Y
-	if days and (not specific_stock or config_performance_use_sharesight):
+	if days and (not specific_stock or config.prefs.performance_use_sharesight):
 		for data in sharesight.get_performance_wrapper(days).values():
 			for holding in data['report']['holdings']:
 				ticker = util.transform_to_yahoo(holding['instrument']['code'], holding['instrument']['market_code'])
@@ -26,40 +26,40 @@ def gather(specific_stock=None, days=None):
 					market_data[ticker]['percent_change_period'] = float(holding['capital_gain_percent'])
 				elif ticker in tickers:
 					print("Notice: price.py:", ticker, "has no data", file=sys.stderr)
-		if not specific_stock and not config_performance_use_sharesight:
+		if not specific_stock and not config.prefs.performance_use_sharesight:
 			yahoo.prefetch_history([t for t in market_data if 'percent_change_period' not in market_data[t]])
 	return tickers, market_data
 
 def needs_beta(quote):
 	"""Whether trigger_bar must know this ticker's beta: a small cap between US$150M and US$1B, which counts as volatile when its
 	beta is over 1.5. The beta is only in Yahoo's details."""
-	return config_demote_volatile and 1.5e8 <= quote.get('market_cap_usd', float('inf')) < 1e9
+	return config.prefs.demote_volatile and 1.5e8 <= quote.get('market_cap_usd', float('inf')) < 1e9
 
 def trigger_bar(ticker, quote, threshold, details=None):
 	"""The move this ticker needs before it alone sends a scheduled report: more for leveraged ETFs (demote_leveraged) and volatile
 	small caps (demote_volatile; their market caps in US dollars), else the threshold itself. details: Yahoo's details of the
 	tickers that need_beta, fetched together; without it this ticker's are fetched here."""
 	title = quote['profile_title'].lower()
-	if config_demote_leveraged and ('2x' in title or '3x' in title):
+	if config.prefs.demote_leveraged and ('2x' in title or '3x' in title):
 		return threshold * 1.3
 	cap = quote.get('market_cap_usd')
-	if config_demote_volatile and cap is not None and cap < 1e9:
+	if config.prefs.demote_volatile and cap is not None and cap < 1e9:
 		if cap < 1e7:
-			return threshold * config_volatility_multiplier * 1.3
+			return threshold * config.prefs.volatility_multiplier * 1.3
 		if cap < 1.5e8:
-			return threshold * config_volatility_multiplier
+			return threshold * config.prefs.volatility_multiplier
 		beta = (yahoo.fetch_detail(ticker) if details is None else details).get(ticker, {}).get('beta')
 		if beta is None or beta > 1.5:
-			return threshold * config_volatility_multiplier
+			return threshold * config.prefs.volatility_multiplier
 	return threshold
 
-def report(service, threshold=UNSET, user='', specific_stock=None, interactive=False, midsession=False, premarket=False, interday=False,
+def report(service, threshold=None, user='', specific_stock=None, interactive=False, midsession=False, premarket=False, interday=False,
 		days=None, close=False, top=None, market=None):
 	"""(payload, chart) of the price report for one service: the moves of `threshold` percent or more (or the top and bottom `top`),
 	today's, over `days`, at the close, mid-session or pre-market. A scheduled report goes out only when one ticker passes its
 	trigger_bar; it then lists every ticker past the threshold. market: gather()'s (tickers, market data), fetched when not given."""
-	if threshold is UNSET:
-		threshold = config_price_percent
+	if threshold is None:
+		threshold = config.prefs.price_percent
 	tickers, market_data = market or gather(specific_stock, days)
 	now = datetime.datetime.now()
 	rows = [] # (ticker, percent, report line without its percent)
@@ -88,10 +88,10 @@ def report(service, threshold=UNSET, user='', specific_stock=None, interactive=F
 		elif days:
 			percent = quote.get('percent_change_period') # Sharesight's figure, for a holding
 			if percent is None:
-				if config_performance_use_sharesight:
+				if config.prefs.performance_use_sharesight:
 					continue
 				if specific_stock: # the one ticker asked for: no history is reported in one line
-					history, graph = yahoo.price_history(ticker, days, graph=config_graph and webhook.takes_charts(service, interactive), graphCache=False)
+					history, graph = yahoo.price_history(ticker, days, graph=config.prefs.graph and webhook.takes_charts(service, interactive), graphCache=False)
 				else:
 					try:
 						history, _ = yahoo.price_history(ticker, days, graph=False)
@@ -161,10 +161,10 @@ def report(service, threshold=UNSET, user='', specific_stock=None, interactive=F
 			payload = [f"{user}, no price movements meet threshold {threshold:g}%"]
 	return payload, graph
 
-def lambda_handler(chat_id=None, threshold=UNSET, service=None, user='', specific_stock=None, interactive=False, midsession=False,
+def lambda_handler(chat_id=None, threshold=None, service=None, user='', specific_stock=None, interactive=False, midsession=False,
 		premarket=False, interday=False, days=None, close=False, top=None):
-	if threshold is UNSET:
-		threshold = config_price_percent
+	if threshold is None:
+		threshold = config.prefs.price_percent
 	market = gather(specific_stock, days)
 	def buttons(service): # a chart's period buttons: one ticker's chart, and the daily and N-day lists (not the intraday ones)
 		if service != 'telegram':
diff --git a/rating.py b/rating.py
index 34c0877d..c9c47d9e 100755
--- a/rating.py
+++ b/rating.py
@@ -1,7 +1,6 @@
 #!/usr/bin/env python3
 
 import sys
-from lib.config import *
 from lib import webhook
 from lib import util
 from lib import yahoo
diff --git a/reminder.py b/reminder.py
index 7205c783..7de0f988 100755
--- a/reminder.py
+++ b/reminder.py
@@ -4,7 +4,7 @@
 import datetime
 import zoneinfo
 from dateutil.relativedelta import relativedelta
-from lib.config import *
+from lib import config
 from lib import util
 from lib import webhook
 
@@ -12,20 +12,20 @@
 def lambda_handler():
 	def prepare_finance_calendar_payload(service):
 		payload = []
-		tz = zoneinfo.ZoneInfo(config_timezone)
+		tz = zoneinfo.ZoneInfo(config.prefs.timezone)
 		localtime = datetime.datetime.now(tz)
 
 		month_and_day = str(localtime.strftime('%m-%d')) # 09-20
 		year = str(localtime.strftime('%Y')) # 2023
 
 		# --- Simple, static one-line reminders loaded from reminders.json ---
-		for reminder in data_json("reminder.json", []):
+		for reminder in config.data_json("reminder.json", []):
 			if reminder.get("month_day") != month_and_day:
 				continue
 			if reminder.get("year") and reminder["year"] != year:
 				continue
 			gate_country = reminder.get("gate_country")
-			if gate_country and config_country_code != gate_country:
+			if gate_country and config.prefs.country_code != gate_country:
 				continue
 			message = reminder["message"]
 			if "{flag}" in message:
@@ -36,7 +36,7 @@ def prepare_finance_calendar_payload(service):
 			payload.append(message)
 
 		# --- Complex, dynamically-computed reminders (kept in code) ---
-		if config_country_code == 'AU':
+		if config.prefs.country_code == 'AU':
 			flag = "🇦🇺"
 			if month_and_day == '06-23':
 				payload.append("💸 Realise capital gains/losses by EOFY June 30" + flag)
diff --git a/shorts.py b/shorts.py
index 2e44772b..af829caa 100755
--- a/shorts.py
+++ b/shorts.py
@@ -1,15 +1,15 @@
 #!/usr/bin/env python3
 
 import sys
-from lib.config import *
+from lib import config
 from lib import webhook
 from lib import util
 from lib import yahoo
 from lib import shortman
 
-def lambda_handler(chat_id=None, threshold=UNSET, specific_stock=None, service=None, interactive=False):
-	if threshold is UNSET:
-		threshold = config_shorts_percent
+def lambda_handler(chat_id=None, threshold=None, specific_stock=None, service=None, interactive=False):
+	if threshold is None:
+		threshold = config.prefs.shorts_percent
 	def prepare_shorts_payload(service, market_data):
 		found = [] # (percent, line, chart label)
 		for ticker in tickers:
diff --git a/tests/common.py b/tests/common.py
index 915e647b..e92773b6 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -1,6 +1,6 @@
 """Shared fixture: points finbot at a temp cache dir and replaces every external service
 (Telegram, Yahoo, Sharesight, Short Man) with canned data. Nothing here touches the network."""
-import collections, io, os, sys, tempfile, time, unittest
+import collections, dataclasses, io, os, sys, tempfile, time, unittest
 import pandas as pd
 from unittest import mock
 
@@ -10,7 +10,7 @@
 os.environ.update(telegramOutgoingWebhook='https://example.com/telegram', telegramOutgoingToken='tok', telegramChatID='1')
 
 import bot # first, and once: it monkey-patches with gevent, which must happen before requests imports ssl and never mid-test
-from lib import config, webhook, util, yahoo, sharesight, shortman, worker, charts, telegram, reports
+from lib import config, webhook, util, yahoo, sharesight, shortman, worker, charts, telegram, reports, words
 import performance, shorts, price, reminder, cal, trades, milestone, rating
 
 REAL = {(m.__name__, n): getattr(m, n) for m, n in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (yahoo, 'price_history'),
@@ -51,19 +51,22 @@ def patch(self, target, attr, value):
 		p.start()
 		self.addCleanup(p.stop)
 
+	def env(self, **changes):
+		"""Startup settings for this test: config.env with these fields changed."""
+		self.patch(config, 'env', dataclasses.replace(config.env, **changes))
+
+	def prefs(self, **changes):
+		"""Preferences for this test: config.prefs with these fields changed (what a reload would do)."""
+		self.patch(config, 'prefs', dataclasses.replace(config.prefs, **changes))
+
 	def setUp(self):
 		fake = mock.Mock(status_code=200, text='{}')
 		fake.json.return_value = {'ok': True, 'result': {'message_id': 1, 'chat': {'id': 5}}}
 		self.patch(__import__('requests').Session, 'request', mock.Mock(return_value=fake)) # no real network traffic
 		tmp = tempfile.TemporaryDirectory()
 		self.addCleanup(tmp.cleanup)
-		self.patch(util, 'config_var_dir', tmp.name)
-		self.patch(util, 'config_cache_dir', tmp.name)
-		self.patch(config, 'config_var_dir', tmp.name)
-		self.patch(webhook, 'webhooks', {'telegram': 'https://api.telegram.org/botX/'})
+		self.env(data_dir=tmp.name, cache_dir=tmp.name, webhooks={'telegram': 'https://api.telegram.org/botX/'})
 		self.patch(bot, '_updates_seen', collections.OrderedDict()) # each test starts with no update taken
-		for m in (performance, shorts, price, cal, trades, milestone, rating, worker, reminder, telegram):
-			self.patch(m, 'webhooks', webhook.webhooks)
 		self.sent = []
 		self.markups = [] # reply_markup of each photo sent
 		self.errors = []
diff --git a/tests/test_config.py b/tests/test_config.py
index 431ab57a..b1b1cde2 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -1,5 +1,5 @@
 """The two config files: finbot.env (the startup file, loaded by the launcher) and preferences.conf (read by the bot, reloadable)."""
-import ast, contextlib, inspect, io, os, re, subprocess, sys, tempfile, unittest
+import ast, contextlib, dataclasses, io, os, re, subprocess, sys, tempfile, unittest
 from pathlib import Path
 from unittest import mock
 from lib import config
@@ -9,13 +9,17 @@
 import performance, price, shorts, trades
 
 REPO = Path(config.__file__).resolve().parent.parent
-DEFAULTS = {'config_past_days': 28, 'config_future_days': 30, 'config_price_percent': 5.0, 'config_shorts_percent': 15,
-	'config_cache_seconds': 43200, 'config_performance_cache_seconds': 1800, 'config_graph': True, 'config_http_timeout': 18,
-	'config_chunk_maxlines': 20, 'config_country_code': 'AU', 'config_timezone': 'Australia/Melbourne', 'config_cache': True,
-	'config_alliterate': False, 'config_hyperlink': True, 'config_hyperlinkFooter': True, 'config_hyperlinkProvider': 'google',
-	'config_demote_volatile': True, 'config_demote_leveraged': True, 'config_volatility_multiplier': 2.0, 'config_exclude_portfolios': [],
-	'config_include_portfolios': [], 'config_trades_use_yahoo': True, 'config_performance_use_sharesight': False,
-	'config_print_headers': False, 'config_min_analysts': 7, 'debug': False}
+DEFAULTS = {'past_days': 28, 'future_days': 30, 'price_percent': 5.0, 'shorts_percent': 15, 'cache_seconds': 43200,
+	'performance_cache_seconds': 1800, 'graph': True, 'http_timeout': 18, 'chunk_maxlines': 20, 'country_code': 'AU',
+	'timezone': 'Australia/Melbourne', 'cache': True, 'alliterate': False, 'hyperlink': True, 'hyperlinkFooter': True,
+	'hyperlinkProvider': 'google', 'demote_volatile': True, 'demote_leveraged': True, 'volatility_multiplier': 2.0, 'exclude_portfolios': (),
+	'include_portfolios': (), 'trades_use_yahoo': True, 'performance_use_sharesight': False, 'print_headers': False, 'min_analysts': 7,
+	'debug': False}
+
+
+def fields(prefs):
+	"""A Prefs as {name: value}, without the file's own lines."""
+	return {name: getattr(prefs, name) for name in config.PREFERENCE_KEYS}
 
 
 def keys_in(path):
@@ -38,13 +42,13 @@ def test_secrets_are_startup_keys(self):
 
 	def test_the_preference_defaults_are_the_old_defaults(self):
 		values = config.preference_values(lambda name, default=None: default)
-		self.assertEqual(values, DEFAULTS)
+		self.assertEqual(fields(values), DEFAULTS)
 
 	def test_templates_have_every_key_and_the_real_defaults(self):
 		self.assertEqual(sorted(keys_in(REPO / 'defaults' / 'preferences.conf')), sorted(config.PREFERENCE_KEYS))
 		self.assertEqual(sorted(keys_in(REPO / 'defaults' / 'finbot.env')), sorted(config.STARTUP_KEYS))
 		prefs = config.read_preferences(REPO / 'defaults' / 'preferences.conf')
-		self.assertEqual(config.preference_values(config._getenv_for(prefs)), DEFAULTS) # the template changes nothing
+		self.assertEqual(fields(config.preference_values(config._getenv_for(prefs))), DEFAULTS) # the template changes nothing
 		startup = config.read_startup_file(REPO / 'defaults' / 'finbot.env')
 		config.startup_values(lambda name, default=None: startup.get(name, default)) # and is valid as it stands
 		self.assertFalse((REPO / 'defaults' / 'env.example').exists())
@@ -101,14 +105,14 @@ def test_invalid_values_are_named(self):
 	def test_booleans_take_the_usual_words_and_refuse_a_typo(self):
 		for word, value in (('True', True), ('yes', True), ('ON', True), ('1', True), ('false', False), ('No', False), ('off', False), ('0', False)):
 			with self.subTest(word=word):
-				self.assertIs(config.preference_values(config._getenv_for({'graph': word}))['config_graph'], value)
+				self.assertIs(config.preference_values(config._getenv_for({'graph': word})).graph, value)
 		with self.assertRaises(config.ConfigError) as caught: # it used to mean False without a word
 			config.preference_values(config._getenv_for({'graph': 'Ture'}))
 		self.assertIn("graph='Ture' is not true or false", caught.exception.errors[0])
 
 	def test_a_blank_value_means_the_default(self):
 		values = config.preference_values(config._getenv_for({'cache': '', 'future_days': '', 'country_code': ''}))
-		self.assertEqual((values['config_cache'], values['config_future_days'], values['config_country_code']), (True, 30, 'AU'))
+		self.assertEqual((values.cache, values.future_days, values.country_code), (True, 30, 'AU'))
 
 	def test_unreadable_file_is_an_error_naming_it(self):
 		path = Path(tempfile.mkdtemp(), 'preferences.conf')
@@ -120,15 +124,15 @@ def test_unreadable_file_is_an_error_naming_it(self):
 	def test_precedence_file_then_environment_then_default(self):
 		with mock.patch.dict(os.environ, {'future_days': '5', 'past_days': '4'}):
 			values = config.preference_values(config._getenv_for({'future_days': '30'}))
-		self.assertEqual((values['config_future_days'], values['config_past_days'], values['config_shorts_percent']), (30, 4, 15))
+		self.assertEqual((values.future_days, values.past_days, values.shorts_percent), (30, 4, 15))
 
 	def test_the_old_multiplier_name_is_gone_everywhere(self):
 		with mock.patch.dict(os.environ, {'volatile_multiplier': '3'}):
-			self.assertEqual(config.preference_values(config._getenv_for({}))['config_volatility_multiplier'], 2.0) # not read any more
+			self.assertEqual(config.preference_values(config._getenv_for({})).volatility_multiplier, 2.0) # not read any more
 			with self.assertRaises(config.ConfigError) as caught:
 				config.check_startup_environment()
 		self.assertIn('volatile_multiplier was renamed to volatility_multiplier', caught.exception.errors[0])
-		self.assertEqual(config.preference_values(config._getenv_for({'volatility_multiplier': '4'}))['config_volatility_multiplier'], 4.0)
+		self.assertEqual(config.preference_values(config._getenv_for({'volatility_multiplier': '4'})).volatility_multiplier, 4.0)
 
 	def test_startup_file_linting(self):
 		path = Path(tempfile.mkdtemp(), 'finbot.env')
@@ -159,22 +163,18 @@ def back_to_defaults(self):
 	def write(self, text):
 		self.file.write_text(text)
 
-	def test_a_change_reaches_config_and_every_module_that_copied_it(self):
+	def test_a_change_is_seen_at_once(self):
 		self.write('future_days=45\nprice_percent=7\nshorts_percent=12\nexclude_portfolios=1156263\nalliterate=1\nhyperlinkProvider=yahoo\n'
 			'hyperlinkFooter=false\ndebug=1\ngraph=0\ndemote_volatile=0\nvolatility_multiplier=3\npast_days=14\nmin_analysts=3\n')
+		before = config.prefs
 		changes = config.reload()
 		self.assertEqual(len(changes), 13)
-		self.assertEqual(reports.config_min_analysts, 3) # .buy, .sell and .forward pe read it
 		self.assertIn('future_days 30 -> 45', changes)
-		want = {'config_future_days': 45, 'config_past_days': 14, 'config_price_percent': 7.0}
-		for module in (config, cal, worker, price, trades, performance, shorts, sharesight, yahoo, util):
-			for name, value in want.items():
-				if hasattr(module, name):
-					self.assertEqual(getattr(module, name), value, f'{module.__name__}.{name}')
-		self.assertEqual(worker.config_future_days, 45)
-		self.assertEqual((config.config_exclude_portfolios, config.config_alliterate, config.config_hyperlinkProvider, config.config_hyperlinkFooter),
-			(['1156263'], True, 'yahoo', False))
-		self.assertEqual((worker.debug, config.debug, yahoo.config_graph, price.config_demote_volatile, price.config_volatility_multiplier), (True, True, False, False, 3.0))
+		self.assertIsNot(config.prefs, before) # one new object: no request sees half a reload
+		self.assertEqual((config.prefs.future_days, config.prefs.past_days, config.prefs.price_percent, config.prefs.min_analysts), (45, 14, 7.0, 3))
+		self.assertEqual((config.prefs.exclude_portfolios, config.prefs.alliterate, config.prefs.hyperlinkProvider, config.prefs.hyperlinkFooter),
+			(('1156263',), True, 'yahoo', False))
+		self.assertEqual((config.prefs.debug, config.prefs.graph, config.prefs.demote_volatile, config.prefs.volatility_multiplier), (True, False, False, 3.0))
 		self.assertEqual(config.reload(), []) # nothing more to change
 
 	def test_a_value_removed_from_the_file_goes_back_to_its_default(self):
@@ -182,10 +182,10 @@ def test_a_value_removed_from_the_file_goes_back_to_its_default(self):
 		config.reload()
 		self.write('# empty\n')
 		self.assertEqual(config.reload(), ['future_days 45 -> 30'])
-		self.assertEqual(cal.config_future_days, 30)
+		self.assertEqual(config.prefs.future_days, 30)
 
 	def test_defaults_that_follow_a_preference_are_read_when_called(self):
-		cache_file = Path(config.config_cache_dir, 'finbot_test_reload.json')
+		cache_file = Path(config.env.cache_dir, 'finbot_test_reload.json')
 		cache_file.write_text('{"a": 1}')
 		self.addCleanup(cache_file.unlink)
 		past = cache_file.stat().st_mtime - 100
@@ -205,7 +205,7 @@ def test_an_error_applies_nothing(self):
 			self.write(text)
 			with self.assertRaises(config.ConfigError):
 				config.reload()
-			self.assertEqual((config.config_future_days, cal.config_future_days), (45, 45))
+			self.assertEqual(config.prefs.future_days, 45)
 
 	def test_a_missing_file_is_an_error_and_keeps_the_running_values(self):
 		self.write('future_days=45\n')
@@ -214,33 +214,27 @@ def test_a_missing_file_is_an_error_and_keeps_the_running_values(self):
 		with self.assertRaises(config.ConfigError) as caught:
 			config.reload()
 		self.assertIn('does not exist; nothing reloaded', caught.exception.errors[0])
-		self.assertEqual(cal.config_future_days, 45)
+		self.assertEqual(config.prefs.future_days, 45)
 		self.file.write_text('')
 
-	def test_signature_defaults_are_the_sentinel_not_a_frozen_value(self):
-		sites = [(cal.lambda_handler, 'days'), (trades.lambda_handler, 'days'), (performance.lambda_handler, 'past_days'),
-			(price.lambda_handler, 'threshold'), (shorts.lambda_handler, 'threshold'), (sharesight.get_trades, 'days'),
-			(sharesight.get_performance, 'config_cache_seconds'), (sharesight.get_performance_wrapper, 'days'), (util.read_cache, 'maxSeconds'),
-			(util.read_binary_cache, 'maxSeconds'), (yahoo.fetch_detail, 'seconds'), (yahoo.price_history, 'seconds'),
-			(yahoo.price_history, 'graph'), (yahoo.fetch_chart_json, 'seconds')]
-		for func, param in sites:
-			self.assertIs(inspect.signature(func).parameters[param].default, config.UNSET, f'{func.__module__}.{func.__name__}({param})')
-
-	def test_no_function_default_or_module_constant_freezes_a_preference(self):
-		names = set(config.PREFERENCE_NAMES)
+	def test_nothing_copies_the_config_or_freezes_a_preference(self):
+		"""Modules read config.prefs.<name> when they run, so that a reload is seen: no star import (which copied every name into
+		each module), and no function default or module constant computed from a preference."""
+		def reads_prefs(node):
+			return any(isinstance(n, ast.Attribute) and n.attr == 'prefs' and isinstance(n.value, ast.Name) and n.value.id == 'config' for n in ast.walk(node))
 		offenders = []
-		for path in sorted([*(REPO / 'lib').glob('*.py'), *REPO.glob('*.py')]):
-			if path.name == 'config.py':
-				continue
+		for path in sorted([*(REPO / 'lib').glob('*.py'), *REPO.glob('*.py'), *(REPO / 'util').glob('*.py')]):
 			tree = ast.parse(path.read_text())
 			for node in ast.walk(tree):
+				if isinstance(node, ast.ImportFrom) and node.module == 'lib.config':
+					offenders.append(f'{path.name}:{node.lineno} from lib.config import {", ".join(a.name for a in node.names)}')
 				if isinstance(node, (ast.FunctionDef, ast.AsyncFunctionDef)):
 					for default in [*node.args.defaults, *[d for d in node.args.kw_defaults if d]]:
-						if {n.id for n in ast.walk(default) if isinstance(n, ast.Name)} & names:
+						if reads_prefs(default):
 							offenders.append(f'{path.name}:{node.lineno} default of {node.name}')
-			for node in tree.body: # module-level constants copied from a preference
-				if isinstance(node, ast.Assign) and {n.id for n in ast.walk(node.value) if isinstance(n, ast.Name)} & names:
-					offenders.append(f'{path.name}:{node.lineno} module constant {[t.id for t in node.targets if isinstance(t, ast.Name)]}')
+			for node in tree.body:
+				if isinstance(node, ast.Assign) and reads_prefs(node.value):
+					offenders.append(f'{path.name}:{node.lineno} module constant')
 		self.assertEqual(offenders, [])
 
 	def test_a_job_process_runs_on_the_snapshot_not_the_edited_file(self):
@@ -251,7 +245,7 @@ def test_a_job_process_runs_on_the_snapshot_not_the_edited_file(self):
 		self.write('future_days=99\n') # edited, no reload sent
 		data = tempfile.mkdtemp()
 		Path(data, 'preferences.conf').write_text('future_days=99\n')
-		show = "from lib import config as c; print(c.config_future_days)"
+		show = "from lib import config as c; print(c.prefs.future_days)"
 		run = lambda **extra: subprocess.run([sys.executable, '-c', show], cwd=REPO, capture_output=True, text=True, check=True,
 			env={**{k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k != config.PASSED}, 'DATA_DIR': data, **extra}).stdout.split()[-1]
 		self.assertEqual(run(**env), '45') # the parent's values, and the file is not read
@@ -273,7 +267,7 @@ def popen(args, **kw):
 
 
 class Startup(unittest.TestCase):
-	def run_config(self, code='from lib import config as c; print(c.config_future_days)', prefs=None, **env):
+	def run_config(self, code='from lib import config as c; print(c.prefs.future_days)', prefs=None, **env):
 		data = tempfile.mkdtemp()
 		if prefs is not None:
 			Path(data, 'preferences.conf').write_text(prefs)
@@ -338,14 +332,12 @@ def test_a_leftover_env_in_the_checkout_refuses_to_start(self):
 	def test_a_discord_webhook_uses_the_native_endpoint(self):
 		for url in ('https://discord.com/api/webhooks/1/abc', 'https://discord.com/api/webhooks/1/abc/slack/'): # the old setup's suffix
 			values = config.startup_values(lambda name, default=None: url if name == 'discord_webhook' else default)
-			self.assertEqual(values['webhooks']['discord'], 'https://discord.com/api/webhooks/1/abc')
+			self.assertEqual(values.webhooks['discord'], 'https://discord.com/api/webhooks/1/abc')
 
 	def test_slack_commands_without_a_bot_token_are_warned_about(self):
-		with mock.patch.object(config, 'config_slackOutgoingWebhook', 'https://example.com/slack'):
-			with mock.patch.object(config, 'config_slackBotToken', False):
-				self.assertEqual(config.startup_warnings(), ['slackOutgoingWebhook is set but slackBotToken is not: Slack commands will get no reply'])
-			with mock.patch.object(config, 'config_slackBotToken', 'xoxb-1'):
-				self.assertEqual(config.startup_warnings(), [])
+		for token, warnings in (('', ['slackOutgoingWebhook is set but slackBotToken is not: Slack commands will get no reply']), ('xoxb-1', [])):
+			with mock.patch.object(config, 'env', dataclasses.replace(config.env, slackOutgoingWebhook='https://example.com/slack', slackBotToken=token)):
+				self.assertEqual(config.startup_warnings(), warnings)
 
 	def test_startup_report_is_a_short_count_never_names_or_values(self):
 		with mock.patch.dict(os.environ, {'slack_webhook': 'https://secret.example/abc', 'port': '5000'}, clear=True):
@@ -379,7 +371,7 @@ def test_reload_all_reloads_preferences_and_the_schedule_independently(self):
 				bot.reload_all()
 			self.assertIn('ERROR config:', err.getvalue())
 			self.assertEqual(scheduler.reload.call_count, 2)
-			self.assertEqual(cal.config_future_days, 45)
+			self.assertEqual(config.prefs.future_days, 45)
 
 	def test_reload_with_the_scheduler_off_still_reloads_preferences(self):
 		file = Path(tempfile.mkdtemp(), 'preferences.conf')
@@ -388,7 +380,7 @@ def test_reload_with_the_scheduler_off_still_reloads_preferences(self):
 		with mock.patch.object(config, 'preferences_path', str(file)), mock.patch.object(bot, 'scheduler', None):
 			with contextlib.redirect_stderr(io.StringIO()):
 				bot.reload_all()
-			self.assertEqual(cal.config_future_days, 11)
+			self.assertEqual(config.prefs.future_days, 11)
 
 
 class Tools(unittest.TestCase):
diff --git a/tests/test_datadir.py b/tests/test_datadir.py
index f59b65af..ba8ce810 100644
--- a/tests/test_datadir.py
+++ b/tests/test_datadir.py
@@ -16,21 +16,21 @@ def run_config(code, **env):
 
 class DataDirTests(FinbotCase):
 	def test_default_unchanged(self):
-		out = run_config("from lib import config as c; print(c.config_var_dir, c.config_cache_dir)")
+		out = run_config("from lib import config as c; print(c.env.data_dir, c.env.cache_dir)")
 		self.assertEqual(out.split()[-2:], ['var', 'var/cache'])
 
 	def test_data_dir_moves_var_and_cache(self):
 		with tempfile.TemporaryDirectory() as d:
-			out = run_config("from lib import config as c; print(c.config_var_dir, c.config_cache_dir)", DATA_DIR=d)
+			out = run_config("from lib import config as c; print(c.env.data_dir, c.env.cache_dir)", DATA_DIR=d)
 			self.assertEqual(out.split()[-2:], [d, d + '/cache'])
 			self.assertTrue(os.path.isdir(d + '/cache'))
 
 	def test_config_reads_environment_only(self):
-		out = run_config("import sys; from lib import config as c; print(c.config_price_percent, 'dotenv' in sys.modules)", price_percent='3.3')
+		out = run_config("import sys; from lib import config as c; print(c.prefs.price_percent, 'dotenv' in sys.modules)", price_percent='3.3')
 		self.assertEqual(out.split()[-2:], ['3.3', 'False'])
 
 	def test_state_directory_fallback(self):
-		show = "from lib import config as c; print(c.config_var_dir, c.config_cache_dir, c.os.environ['MPLCONFIGDIR'])"
+		show = "from lib import config as c; print(c.env.data_dir, c.env.cache_dir, c.os.environ['MPLCONFIGDIR'])"
 		with tempfile.TemporaryDirectory() as d:
 			e = {k: v for k, v in os.environ.items() if k not in ('DATA_DIR', 'MPLCONFIGDIR')}
 			def run(**env):
@@ -49,11 +49,11 @@ def test_util_import_sets_mplconfigdir_without_home(self):
 	def test_data_file_override_and_fallback(self):
 		name = 'finbot_adr.json'
 		self.assertEqual(config.data_file(name), str(REPO / 'defaults' / name))
-		Path(config.config_var_dir, name).write_text('{}')
-		self.assertEqual(config.data_file(name), config.config_var_dir + '/' + name)
+		Path(config.env.data_dir, name).write_text('{}')
+		self.assertEqual(config.data_file(name), config.env.data_dir + '/' + name)
 
 	def test_a_data_file_is_parsed_again_only_when_it_changes(self):
-		path = Path(config.config_var_dir, 'finbot_adr.json')
+		path = Path(config.env.data_dir, 'finbot_adr.json')
 		path.write_text('{"A": "B"}')
 		self.assertEqual(config.data_json('finbot_adr.json'), {'A': 'B'})
 		with mock.patch('json.load', side_effect=AssertionError('parsed again')):
@@ -64,12 +64,12 @@ def test_a_data_file_is_parsed_again_only_when_it_changes(self):
 
 	def test_sync_defaults(self):
 		self.assertIn(('finbot_adr.json', 'would copy'), sync_defaults.sync())
-		self.assertFalse(Path(config.config_var_dir, 'finbot_adr.json').exists()) # dry run
+		self.assertFalse(Path(config.env.data_dir, 'finbot_adr.json').exists()) # dry run
 		sync_defaults.sync(apply=True)
-		self.assertTrue(Path(config.config_var_dir, 'finbot_adr.json').exists())
-		self.assertTrue(Path(config.config_var_dir, 'preferences.conf').exists())
-		self.assertFalse(Path(config.config_var_dir, 'finbot.env').exists()) # the startup file is never copied into the data dir
-		mine = Path(config.config_var_dir, 'finbot_adr.json')
+		self.assertTrue(Path(config.env.data_dir, 'finbot_adr.json').exists())
+		self.assertTrue(Path(config.env.data_dir, 'preferences.conf').exists())
+		self.assertFalse(Path(config.env.data_dir, 'finbot.env').exists()) # the startup file is never copied into the data dir
+		mine = Path(config.env.data_dir, 'finbot_adr.json')
 		mine.write_text('{"mine": 1}')
 		sync_defaults.sync(apply=True)
 		self.assertEqual(mine.read_text(), '{"mine": 1}') # never overwritten without --force
@@ -79,9 +79,9 @@ def test_sync_defaults(self):
 		self.assertEqual(Path(str(mine) + '.bak').read_text(), '{"mine": 1}')
 
 	def test_sync_defaults_to_another_destination(self):
-		other = Path(config.config_var_dir, 'elsewhere')
+		other = Path(config.env.data_dir, 'elsewhere')
 		self.assertIn(('finbot_adr.json', 'would copy'), sync_defaults.sync(dest=other))
 		self.assertFalse(other.exists()) # dry run
 		sync_defaults.sync(apply=True, dest=other)
 		self.assertTrue((other / 'finbot_adr.json').exists())
-		self.assertFalse(Path(config.config_var_dir, 'finbot_adr.json').exists()) # the data dir is untouched
+		self.assertFalse(Path(config.env.data_dir, 'finbot_adr.json').exists()) # the data dir is untouched
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 46992e06..dc212421 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -141,12 +141,12 @@ def test_a_ratio_command_reads_its_ticker_as_the_other_commands_do(self):
 		self.assertIn('Top 15 tracked stocks by pe ratio', str(self.command('.pe top '))) # trailing space and all
 
 	def test_hello_leaves_the_word_list_as_it_was(self):
-		before = list(worker.verb)
+		before = list(words.verb)
 		for _ in range(30):
 			line = self.command('.hello')[0][1][0]
 			self.assertNotIn("('", line) # a verb, not a tuple of them
 			self.assertTrue(line.endswith(' you, U! 😇'), line)
-		self.assertEqual(worker.verb, before)
+		self.assertEqual(words.verb, before)
 		self.assertEqual(self.command('hi there'), []) # hello by name only: "hi @bot"
 
 
@@ -177,8 +177,10 @@ def test_malformed_body_is_refused_not_a_crash(self):
 		self.assertEqual(self.errors, [])
 
 	def test_strangers_cannot_post_into_the_chats(self):
-		self.patch(bot, 'config_slackOutgoingWebhook', 'https://example.com/slack')
-		self.patch(bot, 'config_slackOutgoingToken', 'stok')
+		self.env(slackOutgoingWebhook='https://example.com/slack')
+
+		self.env(slackOutgoingToken='stok')
+
 		for path, body, headers, status in (('/', b'', {}, '404 Not Found'), ('/wp-login.php', b'log=x&pwd=y', {}, '404 Not Found'),
 				('/telegram', b'log=x', {}, '403 Forbidden'), ('/telegram', {'message': {}}, {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'wrong'}, '403 Forbidden'),
 				('/slack', b'', {}, '400 Bad Request'), ('/slack', b'token=stok', {}, '400 Bad Request'), ('/slack', {'type': 'event_callback'}, {}, '403 Forbidden'),
@@ -238,8 +240,10 @@ class SlackEvents(FinbotCase):
 
 	def setUp(self):
 		super().setUp()
-		self.patch(bot, 'config_slackOutgoingWebhook', 'https://example.com/slack')
-		self.patch(bot, 'config_slackOutgoingToken', 'stok')
+		self.env(slackOutgoingWebhook='https://example.com/slack')
+
+		self.env(slackOutgoingToken='stok')
+
 		self.requests = []
 		self.patch(worker, 'process_request', lambda *a: self.requests.append(a))
 
@@ -319,7 +323,8 @@ def test_a_failed_send_logs_no_token(self):
 		def refuse(url, **kwargs):
 			raise ConnectionError(f"Max retries exceeded with url: {url} (headers {kwargs.get('headers')})")
 		self.patch(webhook.net, 'post', refuse)
-		self.patch(webhook, 'config_slackBotToken', 'xoxb-SECRET')
+		self.env(slackBotToken='xoxb-SECRET')
+
 		log = self.stderr_of(lambda: webhook.write('telegram', self.SECRET_URL, 'hi', '1'))
 		log += self.stderr_of(lambda: webhook.write('slack', 'https://hooks.slack.com/services/T0/B0/SECRETpath', 'hi', 'C1'))
 		self.assertIn('Failure sending', log)
diff --git a/tests/test_finbot.py b/tests/test_finbot.py
index fa8c9b2f..76ae706d 100644
--- a/tests/test_finbot.py
+++ b/tests/test_finbot.py
@@ -37,7 +37,7 @@ def test_zoneinfo_offset_is_not_lmt(self):
 		self.assertEqual(datetime.datetime(2022, 8, 18, tzinfo=tz).utcoffset(), datetime.timedelta(hours=10))
 
 	def test_reminder_runs(self):
-		with mock.patch.object(reminder, 'data_json', lambda name, default=None: []):
+		with mock.patch.object(config, 'data_json', lambda name, default=None: []):
 			reminder.lambda_handler()
 
 	def test_bot_handler(self):
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
index c65df849..5d9e6dc3 100644
--- a/tests/test_jobs.py
+++ b/tests/test_jobs.py
@@ -150,7 +150,8 @@ class CleanupJob(FinbotCase):
 	def test_removes_only_old_finbot_cache_files(self):
 		folder = tempfile.mkdtemp()
 		self.addCleanup(lambda: __import__('shutil').rmtree(folder, ignore_errors=True))
-		self.patch(cleanup, 'config_cache_dir', folder)
+		self.env(cache_dir=folder)
+
 		old, new = time.time() - 40 * 86400, time.time() - 5 * 86400
 		for name, stamp in (('finbot_old.json', old), ('finbot_new.json', new), ('other_old.json', old), ('finbot_old.png', old)):
 			path = os.path.join(folder, name)
@@ -164,7 +165,8 @@ def test_removes_only_old_finbot_cache_files(self):
 	def test_a_temp_file_being_written_is_left_alone(self):
 		folder = tempfile.mkdtemp()
 		self.addCleanup(lambda: __import__('shutil').rmtree(folder, ignore_errors=True))
-		self.patch(cleanup, 'config_cache_dir', folder)
+		self.env(cache_dir=folder)
+
 		for name, stamp in (('finbot_a.json.x1.tmp', time.time() - 5), ('finbot_b.json.x2.tmp', time.time() - 40 * 86400)):
 			Path(folder, name).write_text('x')
 			os.utime(Path(folder, name), (stamp, stamp))
@@ -172,13 +174,14 @@ def test_a_temp_file_being_written_is_left_alone(self):
 		self.assertEqual(os.listdir(folder), ['finbot_a.json.x1.tmp'])
 
 	def test_missing_cache_dir_is_fine(self):
-		self.patch(cleanup, 'config_cache_dir', '/nonexistent/finbot-cache')
+		self.env(cache_dir='/nonexistent/finbot-cache')
+
 		self.assertEqual(cleanup.lambda_handler(30), 0)
 
 
 class SchedulerFlag(unittest.TestCase):
 	def flag(self, **env):
-		code = 'from lib import config; print(config.config_scheduler)'
+		code = 'from lib import config; print(config.env.scheduler)'
 		e = {k: v for k, v in os.environ.items() if k != 'scheduler'}
 		e.update(env)
 		repo = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
@@ -244,7 +247,7 @@ def test_a_run_missed_within_the_catch_up_window_fires_once(self):
 		self.assertEqual([t for _, t in s.due(self.clock.now)], [at(13, 21, 40)]) # 21:20 is older than 5 minutes: skipped
 
 	def test_a_corrupt_state_file_is_reported_and_the_schedule_starts_afresh(self):
-		Path(util.config_var_dir, jobs.STATE_FILE).write_text('{"*/20 * * * Mon-Fri|UTC|trades.py": "2026-10-13T21:') # cut off mid-write
+		Path(config.env.data_dir, jobs.STATE_FILE).write_text('{"*/20 * * * Mon-Fri|UTC|trades.py": "2026-10-13T21:') # cut off mid-write
 		s = self.scheduler()
 		self.assertEqual((s.last, len(self.errors)), ({}, 1))
 		self.assertIn('cannot read finbot_jobs.json', self.errors[0])
@@ -389,7 +392,7 @@ def test_the_scheduler_runs_a_daily_job_once_across_the_change(self):
 		file.write_text('CRON_TZ=Australia/Melbourne\n0 9 * * * trades.py\n')
 		clock = Clock(self.mel(2026, 10, 3, 12).astimezone(UTC))
 		s = jobs.Scheduler(lambda: file, clock=clock, zone=UTC, runner=lambda *a: 0)
-		with mock.patch.object(util, 'config_var_dir', tempfile.mkdtemp()), contextlib.redirect_stderr(io.StringIO()):
+		with mock.patch.object(config, 'env', dataclasses.replace(config.env, data_dir=tempfile.mkdtemp())), contextlib.redirect_stderr(io.StringIO()):
 			self.assertTrue(s.load())
 			fired = []
 			while clock.now < self.mel(2026, 10, 5, 12):
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 43736735..04f7da63 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -218,11 +218,12 @@ def test_the_chart_request_carries_the_cookie_of_its_crumb(self):
 
 	def test_with_the_cache_off_no_chart_is_written(self):
 		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
-		self.patch(yahoo, 'config_cache', False)
+		self.prefs(cache=False)
+
 		self.http(yahoo, Resp(self.chart([100.0] * 30 + [110.0])))
 		table, image = yahoo.price_history('T1', 7, graph=True)
 		self.assertIsNotNone(image)
-		self.assertEqual([name for name in os.listdir(util.config_cache_dir) if name.startswith('finbot_graph_')], [])
+		self.assertEqual([name for name in os.listdir(config.env.cache_dir) if name.startswith('finbot_graph_')], [])
 
 	def test_price_series_raises_on_failure_and_on_empty_history(self):
 		self.http(yahoo, Resp(status=500))
@@ -242,10 +243,13 @@ def test_portfolios_from_real_response(self):
 
 	def test_portfolio_exclusion_and_inclusion(self):
 		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_portfolios.json')))
-		self.patch(sharesight, 'config_exclude_portfolios', ['445825'])
+		self.prefs(exclude_portfolios=['445825'])
+
 		self.assertEqual(sharesight.get_portfolios(), {'RobSMSF': 684141})
-		self.patch(sharesight, 'config_exclude_portfolios', [])
-		self.patch(sharesight, 'config_include_portfolios', ['445825'])
+		self.prefs(exclude_portfolios=[])
+
+		self.prefs(include_portfolios=['445825'])
+
 		self.assertEqual(sharesight.get_portfolios(), {'Rob': 445825})
 
 	def test_portfolio_failures_raise(self):
@@ -306,7 +310,7 @@ def get(url, **kwargs):
 		return calls
 
 	def age(self, cache_file, seconds):
-		path = os.path.join(util.config_cache_dir, cache_file)
+		path = os.path.join(config.env.cache_dir, cache_file)
 		old = os.path.getmtime(path) - seconds
 		os.utime(path, (old, old))
 
@@ -322,12 +326,12 @@ def test_refreshes_before_the_ttl_runs_out(self):
 		calls = self.serve_sharesight()
 		sharesight.warm()[0]
 		calls.clear()
-		self.age(sharesight.performance_cache_file(684141, 0), 0.95 * sharesight.config_performance_cache_seconds) # 95% of its lifetime
+		self.age(sharesight.performance_cache_file(684141, 0), 0.95 * config.prefs.performance_cache_seconds) # 95% of its lifetime
 		self.assertEqual(sharesight.warm()[0], 1)
 		self.assertEqual(len(calls), 1)
 		self.assertIn('684141/performance', calls[0])
 		calls.clear()
-		self.age(sharesight.PORTFOLIOS_CACHE, 0.95 * sharesight.config_cache_seconds)
+		self.age(sharesight.PORTFOLIOS_CACHE, 0.95 * config.prefs.cache_seconds)
 		self.assertEqual(sharesight.warm()[0], 1)
 		self.assertTrue(calls[0].endswith('/portfolios'))
 
@@ -350,7 +354,7 @@ def test_holdings_wrapper_is_served_entirely_from_the_warmed_cache(self):
 	def test_failed_refresh_keeps_the_old_cache(self):
 		self.serve_sharesight()
 		sharesight.warm()[0]
-		self.age(sharesight.PORTFOLIOS_CACHE, 0.95 * sharesight.config_cache_seconds)
+		self.age(sharesight.PORTFOLIOS_CACHE, 0.95 * config.prefs.cache_seconds)
 		self.http(sharesight, ConnectionError('down'))
 		with self.assertRaises(RuntimeError):
 			sharesight.warm()[0]
@@ -380,11 +384,12 @@ def is_set(self): return len(self.waits) >= 4
 			def wait(self, seconds):
 				self.waits.append(seconds)
 				if len(self.waits) == 2:
-					sharesight.config_cache = True # what a reload that switches it on does
+					config.prefs = dataclasses.replace(config.prefs, cache=True) # what a reload that switches it on does
 		passes = []
 		self.patch(sharesight, 'warm', lambda: (passes.append(1), (0, {}))[1])
 		self.patch(sharesight, 'next_due', lambda portfolios=None: 3000)
-		self.patch(sharesight, 'config_cache', False)
+		self.prefs(cache=False)
+
 		stop = Stop()
 		sharesight.keep_warm(idle=60, stop=stop)
 		self.assertEqual((stop.waits, len(passes)), ([60, 60, 3000, 3000], 2)) # idle while off: no Sharesight request every 5 seconds
@@ -418,7 +423,7 @@ def test_next_due_is_90_percent_of_the_ttl(self):
 		self.assertEqual(sharesight.next_due(), 0) # cold cache
 		sharesight.warm()[0]
 		due = sharesight.next_due()
-		self.assertAlmostEqual(due, 0.9 * min(sharesight.config_cache_seconds, sharesight.config_performance_cache_seconds), delta=5)
+		self.assertAlmostEqual(due, 0.9 * min(config.prefs.cache_seconds, config.prefs.performance_cache_seconds), delta=5)
 
 	def test_refresh_flag_bypasses_a_fresh_cache(self):
 		calls = self.serve_sharesight()
@@ -460,8 +465,9 @@ def test_error_line_redacts_secrets_and_truncates(self):
 	def test_error_reports_are_escaped_for_telegram_and_slack(self):
 		sent = []
 		self.patch(webhook, 'write', lambda service, url, text, *a, **k: sent.append((service, text)))
-		self.patch(webhook, 'webhooks', {'telegram': 'https://api.telegram.org/botX/', 'slack': 'https://hooks.slack.com/x', 'discord': 'https://discord.com/x/slack'})
-		self.patch(webhook, 'config_telegramChatID', '5')
+		self.env(webhooks={'telegram': 'https://api.telegram.org/botX/', 'slack': 'https://hooks.slack.com/x', 'discord': 'https://discord.com/x/slack'})
+		self.env(telegramChatID='5')
+
 		error = RuntimeError("404 for ZZ at https://x/?a=1&crumb=abc; '<' not supported between <urllib3.connection>")
 		with contextlib.redirect_stderr(io.StringIO()):
 			REAL_REPORT_ERROR(error, context='.compare <b>')
@@ -513,14 +519,17 @@ def upload_steps(self):
 		return [Resp({'ok': True, 'upload_url': self.UPLOAD_URL, 'file_id': 'F1'}), Resp(None, text='OK - 3'), Resp({'ok': True, 'files': [{'id': 'F1'}]})]
 
 	def test_where_charts_can_go(self):
-		self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
+		self.env(slackBotToken='xoxb-test')
+
 		cases = (('telegram',), ('telegram', False), ('slack',), ('slack', False), ('discord', False))
 		self.assertEqual([webhook.takes_charts(*case) for case in cases], [True, True, True, False, True]) # Slack: replies, not scheduled reports
-		self.patch(webhook, 'config_slackBotToken', False)
+		self.env(slackBotToken=False)
+
 		self.assertFalse(webhook.takes_charts('slack'))
 
 	def test_slack_replies_upload_their_charts(self):
-		self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
+		self.env(slackBotToken='xoxb-test')
+
 		for cmd, heading in (('.beta', '*Beta over 1.5'), ('.price', '*5% at close'), ('.compare T1 T2', '*')):
 			with self.subTest(cmd=cmd):
 				calls = self.slack_answers(*self.upload_steps())
@@ -538,7 +547,8 @@ def test_slack_replies_upload_their_charts(self):
 				self.assertNotIn(share['data']['initial_comment'].split('\n'), [payload for _, payload in self.sent]) # not sent again as text
 
 	def test_a_slack_upload_that_fails_goes_out_as_text(self):
-		self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
+		self.env(slackBotToken='xoxb-test')
+
 		good = self.upload_steps()
 		text = [('text', ['*Heading*', 'line 2'])]
 		for outcomes, expect in (([Resp({'ok': False, 'error': 'missing_scope'})], text), ([ConnectionError('down')], text),
@@ -554,8 +564,9 @@ def test_a_slack_upload_that_fails_goes_out_as_text(self):
 				self.assertEqual(self.sent, expect)
 
 	def test_scheduled_reports_to_slack_stay_text(self):
-		self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
-		webhook.webhooks['slack'] = 'https://hooks.slack.com/services/x' # an incoming webhook cannot upload
+		self.env(slackBotToken='xoxb-test')
+
+		self.env(webhooks={**config.env.webhooks, 'slack': 'https://hooks.slack.com/services/x'}) # an incoming webhook cannot upload
 		sent = []
 		self.patch(webhook, 'sendPhoto', lambda chat, image, caption, service, **k: sent.append((service, 'chart')))
 		self.patch(webhook, 'payload_wrapper', lambda service, url, payload, *a, **k: sent.append((service, 'text')) or [])
@@ -566,12 +577,13 @@ def test_scheduled_reports_to_slack_stay_text(self):
 				self.assertEqual(sorted(set(sent)), [('slack', 'text'), ('telegram', 'chart')])
 
 	def test_slack_without_a_bot_token_gets_text_and_no_chart_is_drawn(self):
-		self.patch(webhook, 'config_slackBotToken', False)
+		self.env(slackBotToken=False)
+
 		def refuse(*a, **k):
 			raise AssertionError('a chart was drawn for Slack')
 		for name in ('rows_chart', 'pie_chart', 'bar_graph', 'shorts_chart'):
 			self.patch(util, name, refuse)
-		webhook.webhooks['slack'] = 'https://hooks.slack.com/services/x'
+		self.env(webhooks={**config.env.webhooks, 'slack': 'https://hooks.slack.com/services/x'})
 		for cmd in ('.beta', '.pie Rob', '.buy', '.sell', '.peg', '.price', '.price top', '.price T1 7d', '.performance 7d', '.shorts', '.history T1', '.compare T1 T2'):
 			with self.subTest(cmd=cmd):
 				self.sent.clear(); self.errors.clear()
@@ -582,8 +594,10 @@ def refuse(*a, **k):
 
 	def test_the_graph_preference_off_sends_no_chart_anywhere(self):
 		"""graph=False means no charts, as the README says; it used to hold back only the price-line charts."""
-		self.patch(webhook, 'config_graph', False)
-		self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
+		self.prefs(graph=False)
+
+		self.env(slackBotToken='xoxb-test')
+
 		posts = []
 		self.patch(webhook.net, 'post', lambda url, **k: posts.append(url) or Resp({'ok': True, 'result': True}))
 		for service, chat in (('telegram', '55'), ('slack', 'C1')):
@@ -621,7 +635,7 @@ class DiscordWebhook(FinbotCase):
 
 	def setUp(self):
 		super().setUp()
-		self.patch(webhook, 'webhooks', {'discord': self.URL})
+		self.env(webhooks={'discord': self.URL})
 		self.patch(webhook, 'payload_wrapper', REAL_PAYLOAD_WRAPPER)
 		self.patch(webhook, 'sendPhoto', REAL_SEND_PHOTO)
 		self.posts = []
@@ -667,7 +681,8 @@ def setUp(self):
 		self.patch(webhook.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or Resp({'ok': True}))
 
 	def test_link_previews_are_turned_off_in_the_body(self):
-		self.patch(webhook, 'config_slackBotToken', 'xoxb-test')
+		self.env(slackBotToken='xoxb-test')
+
 		webhook.write('slack', 'https://slack.com/api/chat.postMessage', 'hi', 'C1')
 		webhook.write('slack', 'https://hooks.slack.com/services/T/B/x', 'hi')
 		self.assertEqual(len(self.posts), 2)
@@ -677,7 +692,8 @@ def test_link_previews_are_turned_off_in_the_body(self):
 				self.assertNotIn('unfurl_links', kwargs['headers'])
 
 	def test_a_reply_without_a_bot_token_is_logged_not_sent(self):
-		self.patch(webhook, 'config_slackBotToken', False)
+		self.env(slackBotToken=False)
+
 		err = io.StringIO()
 		with contextlib.redirect_stderr(err), contextlib.redirect_stdout(io.StringIO()):
 			for cmd in ('.help', '.price', '.compare T1 T2'):
@@ -686,7 +702,8 @@ def test_a_reply_without_a_bot_token_is_logged_not_sent(self):
 		self.assertEqual(err.getvalue().count('ERROR slack: a reply to channel C1 needs slackBotToken in finbot.env; not sent'), 3)
 
 	def test_the_slack_api_says_what_is_missing(self):
-		self.patch(webhook, 'config_slackBotToken', False)
+		self.env(slackBotToken=False)
+
 		with self.assertRaisesRegex(RuntimeError, 'Slack files.getUploadURLExternal: needs slackBotToken in finbot.env'):
 			webhook.slack_api('files.getUploadURLExternal', filename='chart.png', length=1) # it used to be a TypeError
 		self.assertEqual(self.posts, [])
@@ -697,8 +714,9 @@ class TelegramWebhook(FinbotCase):
 
 	def setUp(self):
 		super().setUp()
-		self.patch(telegram, 'webhooks', {'telegram': 'https://api.telegram.org/botX/'})
-		self.patch(telegram, 'config_telegramOutgoingWebhook', 'https://example.com/telegram')
+		self.env(webhooks={'telegram': 'https://api.telegram.org/botX/'})
+		self.env(telegramOutgoingWebhook='https://example.com/telegram')
+
 
 	def test_register_unregisters_registers_and_names_the_bot(self):
 		calls = []
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 084ea115..b5745ab4 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -95,17 +95,19 @@ def order(action):
 	def test_a_small_cap_quoted_in_won_needs_the_bigger_move_too(self):
 		"""demote_volatile compares market caps in US dollars: unconverted, a small cap quoted in won was never small."""
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | ({'regularMarketPrice': 1380.0} if t == 'KRW=X' else {'currency': 'KRW', 'market_cap': 1e11}) for t in ts})
-		self.patch(price, 'config_demote_volatile', True)
+		self.prefs(demote_volatile=True)
+
 		quote = price.gather()[1]['T1']
 		self.assertAlmostEqual(quote['market_cap_usd'], 1e11 / 1380.0) # US$72M
-		self.assertEqual(price.trigger_bar('T1', quote, 5), 5 * price.config_volatility_multiplier)
+		self.assertEqual(price.trigger_bar('T1', quote, 5), 5 * config.prefs.volatility_multiplier)
 
 	def test_the_small_caps_betas_are_fetched_together(self):
 		"""A scheduled price report asks for the beta of each small cap that moved (US$150M to US$1B): together, not one by one."""
 		calls = []
 		self.patch(yahoo, 'fetch_details', lambda tickers, strict=False: calls.append(sorted(tickers)) or {t: market_entry(t) for t in tickers})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: self.fail('a beta fetched on its own'))
-		self.patch(price, 'config_demote_volatile', True)
+		self.prefs(demote_volatile=True)
+
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | {'market_cap': 5e8} for t in ts})
 		price.lambda_handler()
 		self.assertEqual(len(calls), 1)
@@ -125,7 +127,7 @@ def listed(action):
 		self.assertIn('forward pe ratio (7+ analysts)', heading)
 		self.assertEqual(listed('negative forward pe')[1], ['T3'])
 		self.assertIn('Co T2', str(reports.prepare_value_payload('telegram', 'forward pe', 'T2'))) # one ticker asked for: shown
-		self.patch(reports, 'config_min_analysts', 13) # the one setting moves .buy and .forward pe together
+		self.prefs(min_analysts=13) # the one setting moves .buy and .forward pe together
 		self.assertEqual(reports.prepare_value_payload('telegram', 'forward pe')[0], ['No tracked stocks with 13+ analysts have a forward pe ratio'])
 		self.assertEqual(reports.prepare_rating_payload('telegram', 'buy')[0], ['No tracked stocks rated buy by 13+ analysts'])
 
@@ -216,7 +218,7 @@ def test_earnings_and_dividends(self):
 
 	def test_earnings_skip_an_adr_whose_primary_listing_yahoo_does_not_know(self):
 		"""T1 has no earnings date, so its primary listing is looked up; Yahoo knowing nothing of it used to crash the report."""
-		self.patch(cal, 'primary_listing', {'T1': 'T1.XX'})
+		self.patch(cal, 'primary_listings', lambda: {'T1': 'T1.XX'})
 		self.patch(yahoo, 'fetch', lambda ts: {t: {k: v for k, v in market_entry(t).items() if not (t == 'T1' and k == 'earnings_date')} for t in ts})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {} if t == 'T1.XX' else {t: market_entry(t)})
 		cal.lambda_handler(event='earnings')
@@ -290,8 +292,9 @@ def test_an_interactive_rating_records_nothing(self):
 	def test_slack_webhook_without_a_bot_token_gets_scheduled_reports(self):
 		"""Scheduled reports reach Slack through its incoming webhook, which takes no channel and no token. They used to send
 		Telegram's chat id along, which made write() need slackBotToken and crash without one, before Telegram got its copy."""
-		self.patch(webhook, 'config_slackBotToken', False)
-		webhook.webhooks['slack'] = 'https://hooks.slack.com/services/T/B/x'
+		self.env(slackBotToken=False)
+
+		self.env(webhooks={**config.env.webhooks, 'slack': 'https://hooks.slack.com/services/T/B/x'})
 		self.patch(webhook, 'payload_wrapper', REAL_PAYLOAD_WRAPPER)
 		self.patch(webhook, 'sendPhoto', REAL_SEND_PHOTO)
 		posts = []
@@ -307,7 +310,7 @@ def post(url, **kwargs):
 				self.assertTrue(any(host == 'api.telegram.org' for host, _, _ in posts))
 
 	def test_nowhere_to_post_is_an_error(self):
-		self.patch(webhook, 'webhooks', {})
+		self.env(webhooks={})
 		with self.assertRaisesRegex(RuntimeError, 'nowhere to post'):
 			price.lambda_handler()
 
@@ -422,7 +425,7 @@ def test_unknown_or_missing_portfolio_lists_known(self):
 			self.assertIn('Rob', sent[0][1])
 
 	def test_a_service_that_takes_no_images_gets_text(self):
-		self.patch(webhook, 'config_slackBotToken', False) # Slack replies carry charts only with the bot token
+		self.env(slackBotToken=False) # Slack replies carry charts only with the bot token
 		worker.Request('slack', 'C1', '@u', 'U', BOT).deliver(['x'], {'pie': True, 'rows': [('a', 1.0)], 'title': 't'})
 		self.assertEqual(self.sent[-1][0], 'text')
 
diff --git a/tests/test_util.py b/tests/test_util.py
index 2b9793a8..2dac74ad 100644
--- a/tests/test_util.py
+++ b/tests/test_util.py
@@ -96,7 +96,7 @@ def test_json_roundtrip_and_missing_file(self):
 
 	def test_a_failed_write_leaves_the_old_file_and_the_umask_alone(self):
 		util.json_write('state.json', {'a': 1}, persist=True)
-		path = Path(util.config_var_dir, 'state.json')
+		path = Path(config.env.data_dir, 'state.json')
 		previous = os.umask(0o027)
 		self.addCleanup(os.umask, previous)
 		def disk_full(*a, **k):
@@ -106,10 +106,10 @@ def disk_full(*a, **k):
 				with self.assertRaises(OSError):
 					util.json_write('state.json', {'a': 2}, persist=True)
 			self.assertEqual(json.loads(path.read_text()), {'a': 1}) # whole or not at all
-			self.assertEqual(os.listdir(util.config_var_dir), ['state.json']) # no temp file left behind
+			self.assertEqual(os.listdir(config.env.data_dir), ['state.json']) # no temp file left behind
 		with mock.patch.object(util.os, 'replace', disk_full), self.assertRaises(OSError):
 			util.write_binary_cache('img.png', io.BytesIO(b'png'))
-		self.assertEqual(os.listdir(util.config_var_dir), ['state.json'])
+		self.assertEqual(os.listdir(config.env.data_dir), ['state.json'])
 		self.assertEqual(os.umask(0o027), 0o027) # the umask was not touched
 		util.json_write('state.json', {'a': 3}, persist=True)
 		self.assertEqual((json.loads(path.read_text()), oct(path.stat().st_mode & 0o777)), ({'a': 3}, '0o640'))
diff --git a/trades.py b/trades.py
index 7e0fdcd1..90600758 100755
--- a/trades.py
+++ b/trades.py
@@ -3,17 +3,18 @@
 import random, sys
 import datetime
 
-from lib.config import *
+from lib import config
+from lib import words
 from lib import sharesight
 from lib import webhook
 from lib import util
 from lib import yahoo
 
-def lambda_handler(chat_id=None, days=UNSET, service=None, user='', portfolio_select=None, interactive=False):
+def lambda_handler(chat_id=None, days=None, service=None, user='', portfolio_select=None, interactive=False):
 	"""The buys and sells of the past `days`, in every portfolio or the one named: on a scheduled run only those not reported
 	before. Raises sharesight.PortfolioNotFound for a name that is none of them."""
-	if days is UNSET:
-		days = config_past_days
+	if days is None:
+		days = config.prefs.past_days
 	state_file = "finbot_sharesight_trades.json"
 
 	def prepare_trade_payload(service, trades):
@@ -30,7 +31,7 @@ def prepare_trade_payload(service, trades):
 		if not wanted: # the usual case on a scheduled run
 			return []
 		market_data = {}
-		if config_trades_use_yahoo:
+		if config.prefs.trades_use_yahoo:
 			tickers = {util.transform_to_yahoo(trade['symbol'], trade['market']) for trade in wanted} # one yahoo.fetch for all
 			try:
 				print("Fetching tickers from Yahoo: ", tickers, file=sys.stderr)
@@ -48,7 +49,7 @@ def prepare_trade_payload(service, trades):
 			units = float(trade['quantity'])
 			price = float(trade['price'])
 			currency = trade['brokerage_currency_code']
-			if config_trades_use_yahoo:
+			if config.prefs.trades_use_yahoo:
 				ticker = util.transform_to_yahoo(symbol, market)
 				try:
 					market = market_data[ticker]['profile_exchange']
@@ -94,7 +95,7 @@ def prepare_trade_payload(service, trades):
 	for service, url, chat in webhook.destinations(interactive, service, chat_id):
 		payload = prepare_trade_payload(service, trades)
 		if interactive and not payload: # no trades, or none of them a buy or a sell
-			payload = [f"{user} No trades {util.days_english(days, 'in the past ')}. {random.choice(noTradesVerb)}"]
+			payload = [f"{user} No trades {util.days_english(days, 'in the past ')}. {random.choice(words.noTradesVerb)}"]
 		webhook.send(service, url, chat, payload)
 
 	# write state file
diff --git a/util/sync_defaults.py b/util/sync_defaults.py
index d5e74657..3716e6cb 100755
--- a/util/sync_defaults.py
+++ b/util/sync_defaults.py
@@ -11,15 +11,14 @@
 
 sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
 from lib import config
-from lib.config import REPO_DIR
 
 
 def sync(apply=False, force=False, dest=None):
 	"""One (file, outcome) per default: copied / would copy, replaced (old copy kept as .bak), or skipped."""
 	apply = apply or force
-	dest_dir = Path(dest or config.config_var_dir)
+	dest_dir = Path(dest or config.env.data_dir)
 	results = []
-	for src in sorted((REPO_DIR / 'defaults').iterdir()):
+	for src in sorted((config.REPO_DIR / 'defaults').iterdir()):
 		if not src.is_file():
 			continue
 		if src.name == 'finbot.env':
@@ -50,7 +49,7 @@ def sync(apply=False, force=False, dest=None):
 	ap.add_argument('--force', action='store_true', help='also replace differing files (keeps .bak); implies --apply')
 	ap.add_argument('--dest', metavar='DIR', help='destination directory (default: the data dir, DATA_DIR or var/)')
 	args = ap.parse_args()
-	print(f"{REPO_DIR / 'defaults'} -> {Path(args.dest or config.config_var_dir).resolve()}")
+	print(f"{config.REPO_DIR / 'defaults'} -> {Path(args.dest or config.env.data_dir).resolve()}")
 	for name, outcome in sync(args.apply or args.force, args.force, args.dest):
 		label = {'copied': '', 'skipped - destination identical': '', 'skipped - destination differs': '[differs]'}.get(outcome, f'[{outcome}]')
 		print(f"{name} {label}".rstrip())

From af891b4123065c740e4c19c51c5580b31b54a0bb Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Wed, 14 Oct 2026 23:32:28 +0000
Subject: [PATCH 08/81] Cache and state files in their own modules:
 lib/cache.py and lib/state.py

util.py's cache_path/read_cache/json_write/json_load family splits by what the file is for. lib/cache.py holds the copies of
Yahoo and Sharesight answers and drawn charts, any of which may vanish at any time; lib/state.py holds what must last in
DATA_DIR (the watchlist, plans, seen trades and so on), written whole and flushed. state.edit() is the locked
read-modify-write for the next step. Output is unchanged: every command, button and scheduled report posts what it did.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md               |   6 ++-
 cleanup.py              |   2 +-
 lib/cache.py            | 104 ++++++++++++++++++++++++++++++++++++++++
 lib/charts.py           |  19 ++++----
 lib/jobs.py             |   6 +--
 lib/reports.py          |  14 +++---
 lib/sharesight.py       |  39 +++++++--------
 lib/state.py            |  47 ++++++++++++++++++
 lib/util.py             |  88 +---------------------------------
 lib/worker.py           |   5 +-
 lib/yahoo.py            |  37 +++++++-------
 milestone.py            |   5 +-
 podcasts.py             |   5 +-
 rating.py               |   5 +-
 tests/common.py         |   2 +-
 tests/test_config.py    |   8 ++--
 tests/test_edges.py     |   6 +--
 tests/test_jobs.py      |   4 +-
 tests/test_providers.py |   4 +-
 tests/test_reports.py   |   8 ++--
 tests/test_util.py      |  36 +++++++-------
 trades.py               |   5 +-
 22 files changed, 267 insertions(+), 188 deletions(-)
 create mode 100644 lib/cache.py
 create mode 100644 lib/state.py

diff --git a/CLAUDE.md b/CLAUDE.md
index 9e7a1474..da1efbed 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -31,9 +31,11 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - `price.py performance.py trades.py shorts.py cal.py rating.py milestone.py reminder.py podcasts.py cleanup.py`: reports that also
   run on a schedule; each has `lambda_handler()` and `job(argv)` (its schedule-line argument parser).
 - `lib/webhook.py`: sending (chunking, photos, Slack upload), markup, escaping, error reporting. `lib/telegram.py`: Bot API calls.
-- `lib/charts.py`: Telegram period buttons and the rebuild behind them. `lib/util.py`: formatting, links, market tables, cache and
-  state files, all matplotlib rendering.
+- `lib/charts.py`: Telegram period buttons and the rebuild behind them. `lib/util.py`: formatting, links, market tables, all
+  matplotlib rendering.
 - `lib/yahoo.py lib/sharesight.py lib/shortman.py lib/simplywallst.py`: data sources, cached in `DATA_DIR/cache`.
+- `lib/cache.py`: cache files (any may vanish at any time). `lib/state.py`: state files in `DATA_DIR`, which must last;
+  change one with `with state.edit(name, default) as data:`, never load-then-save.
 - `lib/config.py`: `config.env` (finbot.env, fixed at start) and `config.prefs` (preferences.conf, a new object on each reload).
   `lib/jobs.py` + `lib/jobrun.py`: the scheduler; each run is a child process.
 - `tests/common.py`: `FinbotCase` (offline fixtures). `self.command('.price')` returns what was sent: `('text', lines)` or
diff --git a/cleanup.py b/cleanup.py
index d51bf316..78ee9fd6 100755
--- a/cleanup.py
+++ b/cleanup.py
@@ -6,7 +6,7 @@
 
 def lambda_handler(days=30):
 	"""Delete the cache files (finbot_*) that were last written more than `days` days ago (days=0: all of them, a flush), but not
-	a temp file written in the last minute: that is an atomic write in flight (util.write_file)."""
+	a temp file written in the last minute: that is an atomic write in flight (cache.write_file)."""
 	now = time.time()
 	cutoff = now - days * 86400
 	removed = 0
diff --git a/lib/cache.py b/lib/cache.py
new file mode 100644
index 00000000..98a15895
--- /dev/null
+++ b/lib/cache.py
@@ -0,0 +1,104 @@
+"""Cache files in DATA_DIR/cache: copies of Yahoo and Sharesight answers and of drawn charts, each good for so many seconds.
+Anything here may be deleted at any time (cleanup.py, a flush); what must last lives in lib/state.py."""
+import io
+import json
+import os
+import sys
+import tempfile
+import time
+from pathlib import Path
+
+from lib import config
+
+
+def path(name):
+	"""Where a cache file lives."""
+	return config.env.cache_dir + '/' + name
+
+
+def age(name):
+	"""Seconds since the cache file was written, or None when it does not exist."""
+	try:
+		return time.time() - os.path.getmtime(path(name))
+	except OSError:
+		return None
+
+
+def _fresh(name, seconds):
+	"""The cache file's path if it was written less than `seconds` ago (the cache_seconds preference when None), else None."""
+	if seconds is None:
+		seconds = config.prefs.cache_seconds
+	where, written = path(name), age(name)
+	if written is None:
+		print("cache miss:", where, file=sys.stderr) if config.prefs.debug else None
+		return None
+	if written >= seconds:
+		print("cache expired:", where, file=sys.stderr) if config.prefs.debug else None
+		return None
+	print("cache hit:", where, "TTL:", f"{seconds - written:.0f} s", file=sys.stderr) if config.prefs.debug else None
+	return where
+
+
+def _read_fresh(name, seconds, read):
+	"""read(path) of the cache file if it is fresh, else None; also None if the file goes (cleanup, a flush) after the check."""
+	where = _fresh(name, seconds)
+	if not where:
+		return None
+	try:
+		return read(Path(where))
+	except FileNotFoundError:
+		return None
+
+
+def read(name, seconds=None):
+	"""A cache file's JSON, or None when it is missing or older than `seconds` (the cache_seconds preference when None)."""
+	return _read_fresh(name, seconds, lambda where: json.loads(where.read_text(encoding="utf-8")))
+
+
+def read_bytes(name, seconds=None):
+	"""A cache file's bytes (as a file object), or None when it is missing or older than `seconds`."""
+	return _read_fresh(name, seconds, lambda where: io.BytesIO(where.read_bytes()))
+
+
+def load(name):
+	"""A cache file's JSON whatever its age, or None: for an answer that carries its own expiry (the Sharesight token)."""
+	where = Path(path(name))
+	try:
+		return json.loads(where.read_text(encoding="utf-8"))
+	except FileNotFoundError:
+		return None
+
+
+def write(name, data):
+	"""Write a cache file's JSON. See write_file."""
+	where = path(name)
+	print("cache write:", where, file=sys.stderr) if config.prefs.debug else None
+	write_file(where, json.dumps(data, indent=4).encode('utf-8'))
+
+
+def write_bytes(name, data):
+	"""Write a cache file's bytes from a file object (a drawn chart)."""
+	data.seek(0)
+	write_file(path(name), data.getvalue())
+
+
+def write_file(where, data, sync=False):
+	"""Write `data` (bytes) to `where` whole or not at all, mode 0640: it goes to a temp file beside it, which then replaces the
+	file, so a failure part way (a full disk, a kill) leaves the old file as it was. sync=True also flushes it to disk, for state
+	that must survive a power cut. The process umask is left alone."""
+	where = os.path.realpath(where) # through a symlink, as an in-place write would
+	fd, tmp = tempfile.mkstemp(dir=os.path.dirname(where), prefix=os.path.basename(where) + '.', suffix='.tmp')
+	try:
+		with os.fdopen(fd, 'wb') as f:
+			f.write(data)
+			if sync:
+				f.flush()
+				os.fsync(f.fileno())
+		os.chmod(tmp, 0o640)
+		os.replace(tmp, where)
+	except BaseException:
+		try:
+			os.unlink(tmp)
+		except OSError:
+			pass
+		raise
diff --git a/lib/charts.py b/lib/charts.py
index 124b7276..615206f8 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -5,6 +5,7 @@
 import threading
 
 from lib import util
+from lib import state
 from lib import webhook
 from lib import yahoo
 from lib import reports
@@ -28,15 +29,15 @@ def _ref(tickers):
 	if len(('c|h|w|' + inline).encode()) <= MAX_CALLBACK_BYTES:
 		return inline
 	key = hashlib.sha1(inline.encode()).hexdigest()[:10]
-	refs = util.json_load(REF_FILE, persist=True) or {}
+	refs = state.load(REF_FILE) or {}
 	if refs.get(key) != tickers:
 		refs[key] = tickers
-		util.json_write(REF_FILE, refs, persist=True)
+		state.save(REF_FILE, refs)
 	return '#' + key
 
 def resolve(ref):
 	if ref.startswith('#'):
-		tickers = (util.json_load(REF_FILE, persist=True) or {}).get(ref[1:])
+		tickers = (state.load(REF_FILE) or {}).get(ref[1:])
 		if not tickers:
 			raise KeyError("these buttons have expired, please re-run the command")
 		return tickers
@@ -59,9 +60,9 @@ def keyboard_version():
 def mark_keyboard(chat_id):
 	"""Record that this chat now has the current command keyboard."""
 	with _keyboard_lock:
-		state = util.json_load(KEYBOARD_FILE, persist=True) or {}
-		state[str(chat_id)] = keyboard_version()
-		util.json_write(KEYBOARD_FILE, state, persist=True)
+		versions = state.load(KEYBOARD_FILE) or {}
+		versions[str(chat_id)] = keyboard_version()
+		state.save(KEYBOARD_FILE, versions)
 
 def push_keyboard(chat_id):
 	"""Telegram can only change a reply keyboard by sending a message, so send a short one carrying it."""
@@ -75,15 +76,15 @@ def ensure_keyboard(chat_id):
 	if int(chat_id) <= 0:
 		return
 	with _keyboard_lock:
-		current = (util.json_load(KEYBOARD_FILE, persist=True) or {}).get(str(chat_id)) == keyboard_version()
+		current = (state.load(KEYBOARD_FILE) or {}).get(str(chat_id)) == keyboard_version()
 	if not current:
 		push_keyboard(chat_id)
 
 def refresh_keyboards():
 	"""At startup: update every chat we know of whose keyboard is out of date."""
 	with _keyboard_lock:
-		state = util.json_load(KEYBOARD_FILE, persist=True) or {}
-	for chat_id, version in state.items():
+		versions = state.load(KEYBOARD_FILE) or {}
+	for chat_id, version in versions.items():
 		if version != keyboard_version():
 			try:
 				push_keyboard(chat_id)
diff --git a/lib/jobs.py b/lib/jobs.py
index 602a9ed2..86c6bb0f 100644
--- a/lib/jobs.py
+++ b/lib/jobs.py
@@ -21,8 +21,8 @@
 from croniter import croniter, CroniterBadDateError
 
 from lib import config
+from lib import state
 from lib import jobrun
-from lib import util
 from lib import webhook
 import cal, cleanup, milestone, performance, podcasts, price, rating, reminder, shorts, trades
 
@@ -270,7 +270,7 @@ def load(self):
 			print('schedule: no usable line in', path, '- keeping the current schedule', file=sys.stderr)
 			return False
 		try:
-			last = util.json_load(STATE_FILE, persist=True)
+			last = state.load(STATE_FILE)
 		except (OSError, ValueError) as e: # unreadable or not JSON: carry on as if there were no record, the next run rewrites it
 			webhook.report_error(ScheduleError(f'cannot read {STATE_FILE}, starting without a record of past runs: {e}'), context='schedule')
 			last = None
@@ -324,7 +324,7 @@ def due(self, now):
 				fired.append((job, latest))
 		if fired:
 			try:
-				util.json_write(STATE_FILE, self.last, persist=True)
+				state.save(STATE_FILE, self.last)
 			except OSError as e:
 				print('schedule: cannot save', STATE_FILE, e, file=sys.stderr)
 		return fired
diff --git a/lib/reports.py b/lib/reports.py
index 62a4a40a..7fdd7552 100755
--- a/lib/reports.py
+++ b/lib/reports.py
@@ -3,6 +3,8 @@
 import datetime
 import sys
 from lib import config
+from lib import cache
+from lib import state
 from lib import sharesight
 from lib import util
 from lib import webhook
@@ -60,7 +62,7 @@ def prepare_watchlist(service, user, action=None, ticker=None):
 		ticker_link = util.finance_link(ticker, ticker, service)
 	duplicate = False
 	transformed = False
-	watchlist = util.json_load('finbot_watchlist.json', persist=True) or [] # absent until the first add
+	watchlist = state.load('finbot_watchlist.json') or [] # absent until the first add
 	if action == 'add':
 		if ticker in watchlist:
 			duplicate = True
@@ -136,7 +138,7 @@ def profile_title_sort(e): # disregards markup in sort command
 		payload.insert(0, f"Hi {user}, I'm currently tracking:")
 	else:
 		payload.append('Watchlist is empty. Try ".watchlist add SYMBOL" to create it')
-	util.json_write('finbot_watchlist.json', watchlist, persist=True)
+	state.save('finbot_watchlist.json', watchlist)
 	if action in ('add', 'delete'):
 		payload = payload[:1] # confirmation only; don't list the whole watchlist
 	return payload
@@ -166,8 +168,8 @@ def prepare_compare(service, args):
 	cache_key = hashlib.sha1(f"{service}|{','.join(tickers)}|{spec}".encode()).hexdigest()[:16]
 	image_cache_file, caption_cache_file = f"finbot_compare_{cache_key}.png", f"finbot_compare_{cache_key}.json"
 	if config.prefs.cache:
-		cached_image = util.read_binary_cache(image_cache_file)
-		cached_caption = util.read_cache(caption_cache_file)
+		cached_image = cache.read_bytes(image_cache_file)
+		cached_caption = cache.read(caption_cache_file)
 		if cached_image and cached_caption:
 			return cached_caption, cached_image, tickers, period_days
 	names, exchanges, closes = {}, {}, {}
@@ -210,8 +212,8 @@ def prepare_compare(service, args):
 		link = util.finance_link(t, exchanges[t], service, days=span_days, brief=False)
 		caption.append(f"{util.get_emoji(pct)} {names[t]} ({link}): {pct:+,.1f}%")
 	if config.prefs.cache:
-		util.write_binary_cache(image_cache_file, image)
-		util.json_write(caption_cache_file, caption)
+		cache.write_bytes(image_cache_file, image)
+		cache.write(caption_cache_file, caption)
 		image.seek(0)
 	return caption, image, tickers, period_days
 
diff --git a/lib/sharesight.py b/lib/sharesight.py
index 8dd9c159..466ec446 100644
--- a/lib/sharesight.py
+++ b/lib/sharesight.py
@@ -4,6 +4,7 @@
 import threading
 import requests
 from lib import config
+from lib import cache
 from lib import net
 import lib.util as util
 
@@ -45,13 +46,13 @@ def __call__(self, r):
 def get_token():
 	if config.prefs.cache:
 		cacheFile = "finbot_sharesight_token.json"
-		cache = util.json_load(cacheFile)
-		if cache:
-			cache_expiry = datetime.datetime.fromtimestamp(cache['created_at'] + cache['expires_in'])
+		cached = cache.load(cacheFile)
+		if cached:
+			cache_expiry = datetime.datetime.fromtimestamp(cached['created_at'] + cached['expires_in'])
 			ttl = cache_expiry - datetime.datetime.now()
 			if ttl > datetime.timedelta(seconds=20):
 				print("cache hit:", cacheFile, "TTL:", util.td_to_human(ttl), file=sys.stderr) if config.prefs.debug else None
-				return cache['access_token']
+				return cached['access_token']
 			else:
 				print("cache miss:", cacheFile, file=sys.stderr) if config.prefs.debug else None
 
@@ -59,7 +60,7 @@ def get_token():
 	data = _request(_post, "https://api.sharesight.com/oauth2/token", 'auth', data=config.env.sharesight_auth)
 	_check(data, 'auth')
 	if config.prefs.cache and 'access_token' in data:
-		util.json_write(cacheFile, data)
+		cache.write(cacheFile, data)
 	print("Got a Sharesight token", file=sys.stderr) if config.prefs.debug else None
 	return data['access_token']
 
@@ -71,14 +72,14 @@ def performance_cache_file(portfolio_id, days):
 def get_portfolios(refresh=False):
 	"""refresh=True skips the cache read and fetches again (the cache is rewritten)."""
 	portfolio_dict = {}
-	data = util.read_cache(PORTFOLIOS_CACHE, config.prefs.cache_seconds) if config.prefs.cache and not refresh else None
+	data = cache.read(PORTFOLIOS_CACHE, config.prefs.cache_seconds) if config.prefs.cache and not refresh else None
 	if not data:
 		token = get_token()
 		print("Fetching Sharesight portfolios")
 		data = _request(_get, "https://api.sharesight.com/api/v3/portfolios", 'portfolios', headers={'Content-type': 'application/json'}, auth=BearerAuth(token))
 		_check(data, 'portfolios')
 		if config.prefs.cache and 'portfolios' in data:
-			util.json_write(PORTFOLIOS_CACHE, data)
+			cache.write(PORTFOLIOS_CACHE, data)
 	for portfolio in data['portfolios']:
 		if str(portfolio['id']) in config.prefs.exclude_portfolios:
 			print(portfolio['id'], "(" + portfolio['name'] + ") in exclusion list. Skipping.")
@@ -114,9 +115,9 @@ def get_trades(portfolio_name, portfolio_id, days=None):
 		days = config.prefs.past_days
 	if config.prefs.cache:
 		cache_file = "finbot_sharesight_trades_" + str(portfolio_id) + "_" + str(days) + ".json"
-		cache = util.read_cache(cache_file, 299) # max freq 5 min
-		if cache:
-			return cache['trades']
+		cached = cache.read(cache_file, 299) # max freq 5 min
+		if cached:
+			return cached['trades']
 	start_date = datetime.datetime.now() - datetime.timedelta(days=days)
 	start_date = start_date.strftime('%Y-%m-%d') # 2022-08-20
 	token = get_token()
@@ -129,7 +130,7 @@ def get_trades(portfolio_name, portfolio_id, days=None):
 	for trade in data['trades']:
 		trade['portfolio'] = portfolio_name # inject custom field
 	if config.prefs.cache and 'trades' in data:
-		util.json_write(cache_file, data)
+		cache.write(cache_file, data)
 	return data['trades']
 
 def get_holdings(portfolio_name, portfolio_id):
@@ -172,9 +173,9 @@ def get_performance(portfolio_id, days, seconds=None, refresh=False):
 	start_date = start_date.strftime('%Y-%m-%d') # 2023-04-25
 	cache_file = performance_cache_file(portfolio_id, days)
 	if config.prefs.cache and not refresh:
-		cache = util.read_cache(cache_file, seconds)
-		if cache:
-			return cache
+		cached = cache.read(cache_file, seconds)
+		if cached:
+			return cached
 	token = get_token()
 	endpoint = 'https://api.sharesight.com/api/v3/portfolios/'
 	url = endpoint + str(portfolio_id) + '/performance?grouping=ungrouped&start_date=' + start_date
@@ -187,7 +188,7 @@ def get_performance(portfolio_id, days, seconds=None, refresh=False):
 		break
 	_check(data, 'performance')
 	if config.prefs.cache and 'report' in data:
-		util.json_write(cache_file, data)
+		cache.write(cache_file, data)
 	return data
 
 def get_performance_wrapper(days=None):
@@ -207,14 +208,14 @@ def warm(margin=0.9):
 	so a request never has to wait on Sharesight. Returns (the number of fetches made, the portfolios): the caller can hand
 	the portfolios on instead of reading the cached list again."""
 	fetched = 0
-	age = util.cache_age(PORTFOLIOS_CACHE)
+	age = cache.age(PORTFOLIOS_CACHE)
 	if age is None or age >= config.prefs.cache_seconds * margin:
 		portfolios = get_portfolios(refresh=True)
 		fetched += 1
 	else:
 		portfolios = get_portfolios()
 	for portfolio_name, portfolio_id in portfolios.items():
-		age = util.cache_age(performance_cache_file(portfolio_id, 0))
+		age = cache.age(performance_cache_file(portfolio_id, 0))
 		if age is None or age >= margin * config.prefs.performance_cache_seconds:
 			get_performance(portfolio_id, 0, refresh=True)
 			fetched += 1
@@ -224,10 +225,10 @@ def next_due(margin=0.9, portfolios=None):
 	"""Seconds until the soonest warmed cache needs refreshing (0 if one already does). `portfolios` is the list warm() just
 	used; without it the cached list is read."""
 	waits = []
-	age = util.cache_age(PORTFOLIOS_CACHE)
+	age = cache.age(PORTFOLIOS_CACHE)
 	waits.append(0 if age is None else margin * config.prefs.cache_seconds - age)
 	for portfolio_name, portfolio_id in (get_portfolios() if portfolios is None else portfolios).items():
-		age = util.cache_age(performance_cache_file(portfolio_id, 0))
+		age = cache.age(performance_cache_file(portfolio_id, 0))
 		waits.append(0 if age is None else margin * config.prefs.performance_cache_seconds - age)
 	return max(0, min(waits))
 
diff --git a/lib/state.py b/lib/state.py
new file mode 100644
index 00000000..0edb81af
--- /dev/null
+++ b/lib/state.py
@@ -0,0 +1,47 @@
+"""State files in DATA_DIR: what must last from one run to the next (the watchlist, plans, seen trades, ratings, record highs,
+the schedule's last runs, chart button refs, keyboard versions, seen podcast episodes). Written whole or not at all, and flushed."""
+import contextlib
+import copy
+import json
+import threading
+from pathlib import Path
+
+from lib import cache
+from lib import config
+
+_locks = {} # file name -> the lock edit() holds while it changes the file
+_locks_lock = threading.Lock()
+
+
+def path(name):
+	"""Where a state file lives."""
+	return config.env.data_dir + '/' + name
+
+
+def load(name, default=None):
+	"""A state file's JSON, or `default` when there is none yet."""
+	where = Path(path(name))
+	return json.loads(where.read_text(encoding="utf-8")) if where.is_file() else default
+
+
+def save(name, data):
+	"""Write a state file's JSON, whole or not at all, flushed to disk (cache.write_file)."""
+	cache.write_file(path(name), json.dumps(data, indent=4).encode('utf-8'), sync=True)
+
+
+def _lock(name):
+	with _locks_lock:
+		return _locks.setdefault(name, threading.Lock())
+
+
+@contextlib.contextmanager
+def edit(name, default):
+	"""Change a state file in place: `with state.edit(name, {}) as data: data[key] = value`. The file is read as the block starts
+	and written when it ends without an error; meanwhile no other edit of the same file runs in this process, so two chats
+	changing it at once both count. Keep slow work, a network call say, outside the block."""
+	with _lock(name):
+		data = load(name)
+		if data is None:
+			data = copy.deepcopy(default)
+		yield data
+		save(name, data)
diff --git a/lib/util.py b/lib/util.py
index 9e6a9c6b..a727495a 100644
--- a/lib/util.py
+++ b/lib/util.py
@@ -1,15 +1,10 @@
 import functools
-import os
 import io
 import datetime
-import json
-from pathlib import Path
 import re
-import sys
-import tempfile
-import time
 from lib import config # first: sets MPLCONFIGDIR before matplotlib is imported (lazily, by the chart functions)
 from lib import sharesight
+from lib import state
 
 
 TITLE_TRIMS = [ # transform_title: (text, replacement), in order, as each sees what the ones before it made
@@ -144,85 +139,6 @@ def flag_from_ticker(ticker):
 def currency_from_market(market):
 	return MARKETS.get(market, (None, None))[1]
 
-def cache_path(name, persist=False):
-	"""Where a cache file lives, or with persist=True a state file (in the data directory)."""
-	return (config.env.data_dir if persist else config.env.cache_dir) + "/" + name
-
-def _fresh_cache(cacheFile, maxSeconds):
-	"""The cache file's path if it was written less than maxSeconds ago (the cache_seconds preference when None), else None."""
-	if maxSeconds is None:
-		maxSeconds = config.prefs.cache_seconds
-	path, age = cache_path(cacheFile), cache_age(cacheFile)
-	if age is None:
-		print("cache miss:", path, file=sys.stderr) if config.prefs.debug else None
-		return None
-	if age >= maxSeconds:
-		print("cache expired:", path, file=sys.stderr) if config.prefs.debug else None
-		return None
-	print("cache hit:", path, "TTL:", td_to_human(datetime.timedelta(seconds=maxSeconds - age)), file=sys.stderr) if config.prefs.debug else None
-	return path
-
-def _read_fresh(cacheFile, maxSeconds, read):
-	"""read(path) of the cache file if it is fresh, else None; also None if the file goes (cleanup, a flush) after the check."""
-	path = _fresh_cache(cacheFile, maxSeconds)
-	if not path:
-		return None
-	try:
-		return read(Path(path))
-	except FileNotFoundError:
-		return None
-
-def read_cache(cacheFile, maxSeconds=None):
-	"""A cache file's JSON, or None when it is missing or older than maxSeconds."""
-	return _read_fresh(cacheFile, maxSeconds, lambda path: json.loads(path.read_text(encoding="utf-8")))
-
-def cache_age(cacheFile):
-	"""Seconds since the cache file was written, or None when it does not exist."""
-	try:
-		return time.time() - os.path.getmtime(cache_path(cacheFile))
-	except OSError:
-		return None
-
-def write_file(path, data, sync=False):
-	"""Write `data` (bytes) to `path` whole or not at all, mode 0640: it goes to a temp file beside it, which then replaces the file,
-	so a failure part way (a full disk, a kill) leaves the old file as it was. sync=True also flushes it to disk, for state that
-	must survive a power cut. The process umask is left alone."""
-	path = os.path.realpath(path) # through a symlink, as an in-place write would
-	fd, tmp = tempfile.mkstemp(dir=os.path.dirname(path), prefix=os.path.basename(path) + '.', suffix='.tmp')
-	try:
-		with os.fdopen(fd, 'wb') as f:
-			f.write(data)
-			if sync:
-				f.flush()
-				os.fsync(f.fileno())
-		os.chmod(tmp, 0o640)
-		os.replace(tmp, path)
-	except BaseException:
-		try:
-			os.unlink(tmp)
-		except OSError:
-			pass
-		raise
-
-def json_write(filename, data, persist=False):
-	"""A cache file, or with persist=True a state file in the data directory (also flushed to disk). See write_file."""
-	path = cache_path(filename, persist)
-	if config.prefs.debug:
-		print("cache write:", path, file=sys.stderr)
-	write_file(path, json.dumps(data, indent=4).encode('utf-8'), sync=persist)
-
-def json_load(filename, persist=False):
-	path = Path(cache_path(filename, persist))
-	return json.loads(path.read_text(encoding="utf-8")) if path.is_file() else None
-
-def read_binary_cache(cacheFile, maxSeconds=None):
-	"""A cache file's bytes (as a file object), or None when it is missing or older than maxSeconds."""
-	return _read_fresh(cacheFile, maxSeconds, lambda path: io.BytesIO(path.read_bytes()))
-
-def write_binary_cache(cacheFile, data):
-	data.seek(0)
-	write_file(cache_path(cacheFile), data.getvalue())
-
 def humanUnits(value, decimal_places=2):
 	"""1234567 -> '1.23 M'."""
 	for unit in ['', 'K', 'M', 'B', 'T', 'Q']:
@@ -852,7 +768,7 @@ def days_from_human_days(arg):
 
 def get_holdings_and_watchlist():
 	tickers = set(sharesight.get_holdings_wrapper())
-	tickers.update(json_load('finbot_watchlist.json', persist=True) or []) # absent until the first add
+	tickers.update(state.load('finbot_watchlist.json', [])) # absent until the first add
 	if 'GOOG' in tickers and 'GOOGL' in tickers:
 		tickers.remove("GOOGL")
 	return sorted(tickers)
diff --git a/lib/worker.py b/lib/worker.py
index 300c8d42..74966386 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -7,6 +7,7 @@
 import sys
 import threading
 from lib import config
+from lib import state
 from lib import words
 from lib import util
 from lib import webhook
@@ -347,10 +348,10 @@ def on_history(request, m):
 	request.send(caption.split('\n'), image, charts.keyboard('h', [ticker], 'x') if image and request.service == 'telegram' else None)
 
 def on_plan(request, m):
-	plan = util.json_load('finbot_plan.json', persist=True) or {}
+	plan = state.load('finbot_plan.json') or {}
 	if m.group(1) and not m.group(1).startswith('@'):
 		plan[request.user] = m.group(1)
-		util.json_write('finbot_plan.json', plan, persist=True)
+		state.save('finbot_plan.json', plan)
 	def who(user): # a Slack user is a mention (<@U123>), which is markup to keep; elsewhere a handle is text
 		user = user.removeprefix('@')
 		return user if request.service == 'slack' else webhook.escape(user, request.service)
diff --git a/lib/yahoo.py b/lib/yahoo.py
index 75f13d71..bd9b77fd 100644
--- a/lib/yahoo.py
+++ b/lib/yahoo.py
@@ -7,6 +7,7 @@
 import zoneinfo
 import sys
 from lib import config
+from lib import cache
 from lib import net
 from lib import util
 from http.cookies import SimpleCookie
@@ -19,9 +20,9 @@ def getCookie(maxAge=1209600): # 14 days
 	# we must always cache this, as crumb must match cookie
 	with _auth:
 		cacheFile = "finbot_yahoo_cookie.json"
-		cache = util.read_cache(cacheFile, maxAge - 30) # renewed 30 seconds before it expires
-		if cache:
-			return cache[0][0] + '=' + cache[0][1]
+		cached = cache.read(cacheFile, maxAge - 30) # renewed 30 seconds before it expires
+		if cached:
+			return cached[0][0] + '=' + cached[0][1]
 		url = 'https://fc.yahoo.com/' # 404, but that is ok because cookie is in response header
 		try:
 			r = net.get(url, headers=HEADERS, timeout=config.prefs.http_timeout)
@@ -36,7 +37,7 @@ def getCookie(maxAge=1209600): # 14 days
 		cookie = SimpleCookie()
 		cookie.load(r.headers['Set-Cookie'])
 		cookielist = [[name, morsel.value, int(morsel.get('max-age') or 0)] for name, morsel in cookie.items()]
-		util.json_write(cacheFile, cookielist)
+		cache.write(cacheFile, cookielist)
 		print("Got a new Yahoo cookie", file=sys.stderr)
 		rmCrumb() # we must invalidate the non-matching crumb
 		return cookielist[0][0] + '=' + cookielist[0][1]
@@ -45,14 +46,14 @@ def getCrumb(seconds=1209600): # 14 days
 	with _auth:
 		cookie = getCookie()
 		cacheFile = "finbot_yahoo_crumb.json"
-		if config.prefs.cache and (cache := util.read_cache(cacheFile, seconds)):
-			return cache
+		if config.prefs.cache and (cached := cache.read(cacheFile, seconds)):
+			return cached
 		r = _get_first(_hosts('/v1/test/getcrumb'), {**HEADERS, 'Cookie': cookie})
 		if r is None:
 			print("Exhausted Yahoo API attempts. Returning fallback crumb", file=sys.stderr)
 			return 'jkQEU8yLqxs'
 		if config.prefs.cache:
-			util.json_write(cacheFile, r.text)
+			cache.write(cacheFile, r.text)
 		print("Got a new Yahoo crumb", file=sys.stderr)
 		return r.text
 
@@ -65,7 +66,7 @@ def credentials():
 
 def rmCrumb():
 	"""Set the cached crumb aside (as .old): it belongs to the cookie being replaced."""
-	crumb = Path(util.cache_path("finbot_yahoo_crumb.json"))
+	crumb = Path(cache.path("finbot_yahoo_crumb.json"))
 	if crumb.is_file():
 		crumb.replace(crumb.with_name(crumb.name + '.old'))
 
@@ -146,8 +147,8 @@ def fetch(tickers):
 		return {}
 	tickers = sorted(set(tickers)) # de-dupe
 	cacheFile = "finbot_yahoo_" + hashlib.sha256("_".join(tickers).encode()).hexdigest() + '.json'
-	if config.prefs.cache and (cache := util.read_cache(cacheFile, 300)):
-		return cache
+	if config.prefs.cache and (cached := cache.read(cacheFile, 300)):
+		return cached
 	cookie, crumb = credentials()
 	r = _get_first(_hosts('/v7/finance/quote?crumb=' + crumb + '&symbols=' + ','.join(tickers)), {**HEADERS, 'Content-type': 'application/json', 'Cookie': cookie})
 	if r is None:
@@ -173,7 +174,7 @@ def fetch(tickers):
 			quote['earnings_date'] = next((stamp for stamp in stamps if stamp > now), stamps[0])
 		yahoo_output[item['symbol']] = quote
 	if config.prefs.cache:
-		util.json_write(cacheFile, yahoo_output)
+		cache.write(cacheFile, yahoo_output)
 	return yahoo_output
 
 DETAIL_FIELDS = ( # (key, path in the quoteSummary result[, conversion]); a field Yahoo leaves out, sends as null or garbles is left out
@@ -233,7 +234,7 @@ def fetch_detail(ticker, seconds=None):
 		seconds = config.prefs.cache_seconds
 	now = datetime.datetime.now()
 	cacheFile = "finbot_yahoo_detail_" + ticker + '.json'
-	data = util.read_cache(cacheFile, seconds) if config.prefs.cache else None
+	data = cache.read(cacheFile, seconds) if config.prefs.cache else None
 	if data:
 		print('.', sep=' ', end='', flush=True)
 	else:
@@ -246,7 +247,7 @@ def fetch_detail(ticker, seconds=None):
 			return {} # catches some delisted stocks like "DRNA"
 		data = r.json()
 		if config.prefs.cache:
-			util.json_write(cacheFile, data)
+			cache.write(cacheFile, data)
 	result = dig(data, 'quoteSummary', 'result', 0)
 	if result is None: # Yahoo answers result=null for unknown symbols
 		print(ticker + 'x', sep=' ', end='', flush=True, file=sys.stderr)
@@ -364,11 +365,11 @@ def price_history(ticker, days=None, seconds=None, graph=None, graphCache=True):
 		label = "Max" if "Max" in percent_dict else "10Y"
 		title = f"{company_name} ({ticker}) {label} {util.signed_percent(percent_dict[label])} Last: {price}"
 	image_cache_file = "finbot_graph_" + ticker + "_" + str(days) + ".png"
-	if config.prefs.cache and graphCache and (image := util.read_binary_cache(image_cache_file, seconds)):
+	if config.prefs.cache and graphCache and (image := cache.read_bytes(image_cache_file, seconds)):
 		return percent_dict, image
 	image = util.graph(df, title, stock.get('currency'))
 	if config.prefs.cache:
-		util.write_binary_cache(image_cache_file, image)
+		cache.write_bytes(image_cache_file, image)
 	return percent_dict, image
 
 def price_series(ticker):
@@ -443,8 +444,8 @@ def fetch_chart_json(ticker, days=3665, seconds=None, full=False):
 		seconds = config.prefs.cache_seconds
 	now = datetime.datetime.now()
 	cacheFile = "finbot_yahoo_history_" + ticker + ("_max" if full else "") + ".json"
-	if config.prefs.cache and (cache := util.read_cache(cacheFile, seconds)):
-		return cache
+	if config.prefs.cache and (cached := cache.read(cacheFile, seconds)):
+		return cached
 	cookie, crumb = credentials()
 	start = '0' if full else str(int((now - datetime.timedelta(days=days)).timestamp()))
 	url = f'https://query1.finance.yahoo.com/v8/finance/chart/{ticker}?period1={start}&period2={int(now.timestamp())}&interval=1d&crumb={crumb}'
@@ -458,7 +459,7 @@ def fetch_chart_json(ticker, days=3665, seconds=None, full=False):
 	print('↓', sep=' ', end='', flush=True)
 	data = r.json()
 	if config.prefs.cache:
-		util.json_write(cacheFile, data)
+		cache.write(cacheFile, data)
 	return data
 
 def chart_json_to_df(chart_json):
diff --git a/milestone.py b/milestone.py
index cc782bf2..e2748705 100755
--- a/milestone.py
+++ b/milestone.py
@@ -3,6 +3,7 @@
 import sys
 from itertools import groupby
 from lib import config
+from lib import state
 from lib import webhook
 from lib import util
 from lib import yahoo
@@ -16,7 +17,7 @@ def prepare_milestone_payload(service, market_data):
 		payload = []
 		payloadhigh = []
 		payloadlow = []
-		records = util.json_load('finbot_milestone.json', persist=True) or {}
+		records = state.load('finbot_milestone.json') or {}
 		for ticker in tickers:
 			try:
 				oldhigh, oldlow = records[ticker]['record_high'], records[ticker]['record_low']
@@ -65,7 +66,7 @@ def prepare_milestone_payload(service, market_data):
 		webhook.send(service, url, chat, payload)
 
 	if records:
-		util.json_write('finbot_milestone.json', records, persist=True)
+		state.save('finbot_milestone.json', records)
 
 def job(argv):
 	return 'milestone.py', {}
diff --git a/podcasts.py b/podcasts.py
index 1d088a55..a2d62f04 100755
--- a/podcasts.py
+++ b/podcasts.py
@@ -7,6 +7,7 @@
 from email.utils import parsedate_to_datetime
 from datetime import datetime, timezone
 from lib import net
+from lib import state
 from lib import webhook
 from lib import util
 
@@ -87,7 +88,7 @@ def fetch_new_episodes(feed_url, seen):
 def lambda_handler(feed_url):
 	"""Post the feed's newest episodes not posted before, then record every new one as seen."""
 	seen_file = get_seen_filename(feed_url)
-	seen = set(util.json_load(seen_file, persist=True) or [])
+	seen = set(state.load(seen_file) or [])
 	try:
 		episodes, guids = fetch_new_episodes(feed_url, seen)
 	except Exception as e:
@@ -103,7 +104,7 @@ def lambda_handler(feed_url):
 			duration = f" {episode['duration']}" if episode['duration'] else ""
 			payload.append(f"📣 {podcast_name}: {util.link(episode['url'], title, service)}{duration} ({episode['ago']})")
 		webhook.send(service, url, chat, payload)
-	util.json_write(seen_file, sorted(seen | guids), persist=True) # last, so that a failure above leaves these episodes for next run
+	state.save(seen_file, sorted(seen | guids)) # last, so that a failure above leaves these episodes for next run
 
 def job(argv):
 	"""Arguments of a scheduled run: (label, lambda_handler kwargs). Raises ValueError with the usage on a bad argument."""
diff --git a/rating.py b/rating.py
index c9c47d9e..a35662eb 100755
--- a/rating.py
+++ b/rating.py
@@ -2,6 +2,7 @@
 
 import sys
 from lib import webhook
+from lib import state
 from lib import util
 from lib import yahoo
 
@@ -54,7 +55,7 @@ def get_emoji(old, new):
 	# MAIN #
 
 	tickers = [util.transform_to_yahoo(specific_stock)] if specific_stock else util.get_holdings_and_watchlist()
-	old = util.json_load('finbot_rating.json', persist=True)
+	old = state.load('finbot_rating.json')
 	old = old if isinstance(old, dict) else {} # absent until the first run
 	market_data = yahoo.fetch_details(tickers, strict=bool(specific_stock))
 
@@ -63,7 +64,7 @@ def get_emoji(old, new):
 		webhook.send(service, url, chat, payload)
 
 	if new and not interactive: # added to what is recorded: a ticker missing from this run keeps its last rating
-		util.json_write('finbot_rating.json', old | new, persist=True)
+		state.save('finbot_rating.json', old | new)
 
 def job(argv):
 	return 'rating.py', {}
diff --git a/tests/common.py b/tests/common.py
index e92773b6..0febcfcb 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -10,7 +10,7 @@
 os.environ.update(telegramOutgoingWebhook='https://example.com/telegram', telegramOutgoingToken='tok', telegramChatID='1')
 
 import bot # first, and once: it monkey-patches with gevent, which must happen before requests imports ssl and never mid-test
-from lib import config, webhook, util, yahoo, sharesight, shortman, worker, charts, telegram, reports, words
+from lib import cache, config, state, webhook, util, yahoo, sharesight, shortman, worker, charts, telegram, reports, words
 import performance, shorts, price, reminder, cal, trades, milestone, rating
 
 REAL = {(m.__name__, n): getattr(m, n) for m, n in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (yahoo, 'price_history'),
diff --git a/tests/test_config.py b/tests/test_config.py
index b1b1cde2..40f0202b 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -5,7 +5,7 @@
 from lib import config
 from util import check_config, split_env
 import bot, cal
-from lib import reports, sharesight, util, worker, yahoo
+from lib import cache, reports, sharesight, util, worker, yahoo
 import performance, price, shorts, trades
 
 REPO = Path(config.__file__).resolve().parent.parent
@@ -192,11 +192,11 @@ def test_defaults_that_follow_a_preference_are_read_when_called(self):
 		os.utime(cache_file, (past, past)) # 100 seconds old
 		self.write('cache_seconds=50\n')
 		config.reload()
-		self.assertIsNone(util.read_cache('finbot_test_reload.json')) # too old for 50 s
+		self.assertIsNone(cache.read('finbot_test_reload.json')) # too old for 50 s
 		self.write('cache_seconds=200\n')
 		config.reload()
-		self.assertEqual(util.read_cache('finbot_test_reload.json'), {'a': 1})
-		self.assertIsNone(util.read_cache('finbot_test_reload.json', 50)) # an explicit value always wins
+		self.assertEqual(cache.read('finbot_test_reload.json'), {'a': 1})
+		self.assertIsNone(cache.read('finbot_test_reload.json', 50)) # an explicit value always wins
 
 	def test_an_error_applies_nothing(self):
 		self.write('future_days=45\n')
diff --git a/tests/test_edges.py b/tests/test_edges.py
index dc212421..366a0cdb 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -71,12 +71,12 @@ class WatchlistActions(FinbotCase):
 	def test_add_duplicate_delete_and_unknown_action(self):
 		add = self.command('.watchlist add MSFT')
 		self.assertTrue(add)
-		self.assertEqual(util.json_load('finbot_watchlist.json', persist=True), ['MSFT'])
+		self.assertEqual(state.load('finbot_watchlist.json'), ['MSFT'])
 		again = self.command('.watchlist add MSFT')
 		self.assertNotEqual(add, again) # a duplicate is reported differently from a first add
-		self.assertEqual(util.json_load('finbot_watchlist.json', persist=True), ['MSFT'])
+		self.assertEqual(state.load('finbot_watchlist.json'), ['MSFT'])
 		self.command('.watchlist rm MSFT')
-		self.assertEqual(util.json_load('finbot_watchlist.json', persist=True), [])
+		self.assertEqual(state.load('finbot_watchlist.json'), [])
 		self.assertIn('not a valid watchlist action', str(self.command('.watchlist frobnicate MSFT')))
 
 
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
index 5d9e6dc3..55b19333 100644
--- a/tests/test_jobs.py
+++ b/tests/test_jobs.py
@@ -241,7 +241,7 @@ def test_tick_runs_the_job_and_returns_the_wait(self):
 	def test_a_run_missed_within_the_catch_up_window_fires_once(self):
 		first = self.scheduler()
 		first.last = {first.jobs[0].key: at(13, 21, 0).isoformat()}
-		util.json_write(jobs.STATE_FILE, first.last, persist=True)
+		state.save(jobs.STATE_FILE, first.last)
 		self.clock.now = at(13, 21, 43)
 		s = self.scheduler()
 		self.assertEqual([t for _, t in s.due(self.clock.now)], [at(13, 21, 40)]) # 21:20 is older than 5 minutes: skipped
@@ -253,7 +253,7 @@ def test_a_corrupt_state_file_is_reported_and_the_schedule_starts_afresh(self):
 		self.assertIn('cannot read finbot_jobs.json', self.errors[0])
 		self.clock.now = at(13, 21, 20)
 		self.assertEqual(len(s.due(self.clock.now)), 1) # runs as usual, and the record is written afresh
-		self.assertEqual(util.json_load(jobs.STATE_FILE, persist=True), {s.jobs[0].key: at(13, 21, 20).isoformat()})
+		self.assertEqual(state.load(jobs.STATE_FILE), {s.jobs[0].key: at(13, 21, 20).isoformat()})
 		s.reload() # and a reload is no trouble either
 		self.assertEqual(len(self.errors), 1)
 
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 04f7da63..d23ac476 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -410,8 +410,8 @@ def test_a_warming_pass_reads_the_cached_portfolio_list_once(self):
 		self.serve_sharesight()
 		sharesight.warm()[0] # fill the caches
 		reads = []
-		real = util.read_cache
-		self.patch(util, 'read_cache', lambda name, *a, **k: (reads.append(name), real(name, *a, **k))[1])
+		real = cache.read
+		self.patch(cache, 'read', lambda name, *a, **k: (reads.append(name), real(name, *a, **k))[1])
 		fetched, portfolios = sharesight.warm() # what one pass of keep_warm does
 		due = sharesight.next_due(portfolios=portfolios)
 		self.assertEqual((fetched, reads), (0, [sharesight.PORTFOLIOS_CACHE]))
diff --git a/tests/test_reports.py b/tests/test_reports.py
index b5745ab4..8bdb222d 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -247,12 +247,12 @@ def test_milestone_records_a_ticker_only_with_its_new_values(self):
 		self.patch(yahoo, 'historic_high', lambda t: (150.0, 50.0)) # beyond the 52-week range (120/80): nothing new
 		self.patch(yahoo, 'fetch', lambda ts: {t: {k: v for k, v in market_entry(t).items() if not k.startswith('fiftyTwoWeek')} for t in ts})
 		milestone.lambda_handler()
-		self.assertFalse(util.json_load('finbot_milestone.json', persist=True))
+		self.assertFalse(state.load('finbot_milestone.json'))
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts})
 		self.sent.clear()
 		milestone.lambda_handler()
 		self.assertNotIn('Record high', str(self.sent))
-		self.assertEqual(util.json_load('finbot_milestone.json', persist=True)['T1'], {'record_high': 150.0, 'record_low': 50.0}) # all-time
+		self.assertEqual(state.load('finbot_milestone.json')['T1'], {'record_high': 150.0, 'record_low': 50.0}) # all-time
 
 	def test_milestone_records_are_all_time(self):
 		"""A new 52-week high below the all-time high is no record. The record used to become the 52-week high after the first
@@ -267,7 +267,7 @@ def test_milestone_records_are_all_time(self):
 				self.assertEqual('Record high' in str(self.sent), expected)
 
 	def test_a_52_week_record_kept_before_is_replaced_by_the_history(self):
-		util.json_write('finbot_milestone.json', {f'T{i}': {'high': 110.0, 'low': 90.0} for i in range(25)}, persist=True)
+		state.save('finbot_milestone.json', {f'T{i}': {'high': 110.0, 'low': 90.0} for i in range(25)})
 		self.patch(yahoo, 'historic_high', lambda t: (150.0, 50.0))
 		milestone.lambda_handler() # 120 beats the old 52-week record of 110, but not the history's 150
 		self.assertNotIn('Record high', str(self.sent))
@@ -287,7 +287,7 @@ def test_rating_keeps_the_last_rating_of_a_ticker_missing_from_a_run(self):
 
 	def test_an_interactive_rating_records_nothing(self):
 		rating.lambda_handler(interactive=True, service='telegram', chat_id='5')
-		self.assertIsNone(util.json_load('finbot_rating.json', persist=True))
+		self.assertIsNone(state.load('finbot_rating.json'))
 
 	def test_slack_webhook_without_a_bot_token_gets_scheduled_reports(self):
 		"""Scheduled reports reach Slack through its incoming webhook, which takes no channel and no token. They used to send
diff --git a/tests/test_util.py b/tests/test_util.py
index 2dac74ad..92a46c59 100644
--- a/tests/test_util.py
+++ b/tests/test_util.py
@@ -90,37 +90,37 @@ def test_flags_for_known_and_unknown_markets(self):
 
 class StateFiles(FinbotCase):
 	def test_json_roundtrip_and_missing_file(self):
-		self.assertIsNone(util.json_load('absent.json', persist=True))
-		util.json_write('state.json', {'a': [1, 2]}, persist=True)
-		self.assertEqual(util.json_load('state.json', persist=True), {'a': [1, 2]})
+		self.assertIsNone(state.load('absent.json'))
+		state.save('state.json', {'a': [1, 2]})
+		self.assertEqual(state.load('state.json'), {'a': [1, 2]})
 
 	def test_a_failed_write_leaves_the_old_file_and_the_umask_alone(self):
-		util.json_write('state.json', {'a': 1}, persist=True)
+		state.save('state.json', {'a': 1})
 		path = Path(config.env.data_dir, 'state.json')
 		previous = os.umask(0o027)
 		self.addCleanup(os.umask, previous)
 		def disk_full(*a, **k):
 			raise OSError(28, 'No space left on device')
 		for broken in ('replace', 'fsync'):
-			with self.subTest(broken=broken), mock.patch.object(util.os, broken, disk_full):
+			with self.subTest(broken=broken), mock.patch.object(cache.os, broken, disk_full):
 				with self.assertRaises(OSError):
-					util.json_write('state.json', {'a': 2}, persist=True)
+					state.save('state.json', {'a': 2})
 			self.assertEqual(json.loads(path.read_text()), {'a': 1}) # whole or not at all
 			self.assertEqual(os.listdir(config.env.data_dir), ['state.json']) # no temp file left behind
-		with mock.patch.object(util.os, 'replace', disk_full), self.assertRaises(OSError):
-			util.write_binary_cache('img.png', io.BytesIO(b'png'))
+		with mock.patch.object(cache.os, 'replace', disk_full), self.assertRaises(OSError):
+			cache.write_bytes('img.png', io.BytesIO(b'png'))
 		self.assertEqual(os.listdir(config.env.data_dir), ['state.json'])
 		self.assertEqual(os.umask(0o027), 0o027) # the umask was not touched
-		util.json_write('state.json', {'a': 3}, persist=True)
+		state.save('state.json', {'a': 3})
 		self.assertEqual((json.loads(path.read_text()), oct(path.stat().st_mode & 0o777)), ({'a': 3}, '0o640'))
 
 	def test_read_cache_missing_or_expired_is_falsy(self):
-		self.assertFalse(util.read_cache('absent.json', 60))
-		util.json_write('old.json', {'a': 1})
-		self.assertFalse(util.read_cache('old.json', -1)) # already expired
+		self.assertFalse(cache.read('absent.json', 60))
+		cache.write('old.json', {'a': 1})
+		self.assertFalse(cache.read('old.json', -1)) # already expired
 
 	def test_a_cache_file_that_goes_while_it_is_read_is_a_miss(self):
-		util.json_write('gone.json', {'a': 1})
+		cache.write('gone.json', {'a': 1})
 		real = Path.read_text
 		def vanish(path, *a, **k): # cleanup or a flush deleted it between the age check and the read
 			if path.name == 'gone.json':
@@ -128,19 +128,19 @@ def vanish(path, *a, **k): # cleanup or a flush deleted it between the age check
 				raise FileNotFoundError(path)
 			return real(path, *a, **k)
 		with mock.patch.object(Path, 'read_text', vanish):
-			self.assertIsNone(util.read_cache('gone.json', 60))
+			self.assertIsNone(cache.read('gone.json', 60))
 
 	def test_binary_cache_roundtrip(self):
-		self.assertFalse(util.read_binary_cache('absent.png'))
-		util.write_binary_cache('img.png', io.BytesIO(b'png-bytes'))
-		self.assertEqual(util.read_binary_cache('img.png').read(), b'png-bytes')
+		self.assertFalse(cache.read_bytes('absent.png'))
+		cache.write_bytes('img.png', io.BytesIO(b'png-bytes'))
+		self.assertEqual(cache.read_bytes('img.png').read(), b'png-bytes')
 
 	def test_holdings_and_watchlist_without_a_watchlist_file(self):
 		# first run: no watchlist exists yet
 		self.patch(util.sharesight, 'get_holdings_wrapper', lambda: ['AAPL', 'GOOG', 'GOOGL'])
 		self.patch(util, 'get_holdings_and_watchlist', REAL_GET_HOLDINGS)
 		self.assertEqual(util.get_holdings_and_watchlist(), ['AAPL', 'GOOG']) # GOOGL dropped when GOOG is present
-		util.json_write('finbot_watchlist.json', ['MSFT'], persist=True)
+		state.save('finbot_watchlist.json', ['MSFT'])
 		self.assertEqual(util.get_holdings_and_watchlist(), ['AAPL', 'GOOG', 'MSFT'])
 
 
diff --git a/trades.py b/trades.py
index 90600758..b04f489e 100755
--- a/trades.py
+++ b/trades.py
@@ -4,6 +4,7 @@
 import datetime
 
 from lib import config
+from lib import state
 from lib import words
 from lib import sharesight
 from lib import webhook
@@ -89,7 +90,7 @@ def prepare_trade_payload(service, trades):
 	# MAIN #
 	newtrades = set()
 	portfolios = sharesight.pick_portfolios(portfolio_select)
-	known_trades = set(util.json_load(state_file, persist=True) or ())
+	known_trades = set(state.load(state_file) or ())
 	trades = [trade for name, portfolio_id in portfolios.items() for trade in sharesight.get_trades(name, portfolio_id, days)]
 
 	for service, url, chat in webhook.destinations(interactive, service, chat_id):
@@ -101,7 +102,7 @@ def prepare_trade_payload(service, trades):
 	# write state file
 	if newtrades and not interactive:
 		known_trades.update(newtrades)
-		util.json_write(state_file, list(known_trades), persist=True)
+		state.save(state_file, list(known_trades))
 
 def job(argv):
 	return 'trades.py', {}

From 1348e87194cd57f2c343abe55e154cff099787b0 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Wed, 14 Oct 2026 23:41:25 +0000
Subject: [PATCH 09/81] State files change under a lock; the watchlist is an
 object (lib/watchlist.py)

state.edit() reads a state file, lets the block change it and writes it back only if it changed, while no other edit of
that file runs in the process. Every read-modify-write uses it: the watchlist, plans, chart button refs, keyboard
versions, and the seen trades, ratings, records and podcast episodes, which now add to the file as it is when the run
ends instead of writing back what was read when it began. charts' keyboard lock goes.

The watchlist: tickers(), find() (Yahoo, with the .AX retry), add() and remove(). A ticker is checked with Yahoo before
the file is opened, so two chats adding at once both count; before, both read the list, waited on Yahoo, and the later
write dropped the other's add. prepare_watchlist shrinks to phrasing the outcome: it fetches only the ticker it is
changing, not the whole list it then threw away. yahoo.resolve() is the one .AX retry, shared with the company profile.

Visible changes: the add/remove confirmation links to the stock's exchange (MSFT:NASDAQ, not MSFT:MSFT), and removing a
ticker Yahoo no longer knows says it was removed (it was, but the reply said it could not be found).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md             |   2 +-
 lib/charts.py         |  21 ++-----
 lib/reports.py        | 126 +++++++++++-------------------------------
 lib/state.py          |  13 +++--
 lib/util.py           |   4 +-
 lib/watchlist.py      |  36 ++++++++++++
 lib/worker.py         |   6 +-
 lib/yahoo.py          |   9 +++
 milestone.py          |  14 +++--
 podcasts.py           |   5 +-
 rating.py             |   6 +-
 tests/common.py       |   2 +-
 tests/test_edges.py   |  29 +++++++++-
 tests/test_reports.py |   7 +++
 tests/test_util.py    |  28 +++++++++-
 trades.py             |   7 +--
 16 files changed, 176 insertions(+), 139 deletions(-)
 create mode 100644 lib/watchlist.py

diff --git a/CLAUDE.md b/CLAUDE.md
index da1efbed..4b9fc585 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -35,7 +35,7 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   matplotlib rendering.
 - `lib/yahoo.py lib/sharesight.py lib/shortman.py lib/simplywallst.py`: data sources, cached in `DATA_DIR/cache`.
 - `lib/cache.py`: cache files (any may vanish at any time). `lib/state.py`: state files in `DATA_DIR`, which must last;
-  change one with `with state.edit(name, default) as data:`, never load-then-save.
+  change one with `with state.edit(name, default) as data:`, never load-then-save. `lib/watchlist.py`: the watchlist.
 - `lib/config.py`: `config.env` (finbot.env, fixed at start) and `config.prefs` (preferences.conf, a new object on each reload).
   `lib/jobs.py` + `lib/jobrun.py`: the scheduler; each run is a child process.
 - `tests/common.py`: `FinbotCase` (offline fixtures). `self.command('.price')` returns what was sent: `('text', lines)` or
diff --git a/lib/charts.py b/lib/charts.py
index 615206f8..667741da 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -2,7 +2,6 @@
 import hashlib
 import json
 import sys
-import threading
 
 from lib import util
 from lib import state
@@ -29,15 +28,13 @@ def _ref(tickers):
 	if len(('c|h|w|' + inline).encode()) <= MAX_CALLBACK_BYTES:
 		return inline
 	key = hashlib.sha1(inline.encode()).hexdigest()[:10]
-	refs = state.load(REF_FILE) or {}
-	if refs.get(key) != tickers:
+	with state.edit(REF_FILE, {}) as refs:
 		refs[key] = tickers
-		state.save(REF_FILE, refs)
 	return '#' + key
 
 def resolve(ref):
 	if ref.startswith('#'):
-		tickers = (state.load(REF_FILE) or {}).get(ref[1:])
+		tickers = state.load(REF_FILE, {}).get(ref[1:])
 		if not tickers:
 			raise KeyError("these buttons have expired, please re-run the command")
 		return tickers
@@ -52,17 +49,13 @@ def command_keyboard(per_row=3):
 	return {'keyboard': rows, 'resize_keyboard': True, 'is_persistent': True}
 
 KEYBOARD_FILE = 'finbot_keyboards.json' # chat id -> version of the command keyboard that chat last received
-_keyboard_lock = threading.Lock()
-
 def keyboard_version():
 	return hashlib.sha1(json.dumps(command_keyboard(), sort_keys=True).encode()).hexdigest()[:8]
 
 def mark_keyboard(chat_id):
 	"""Record that this chat now has the current command keyboard."""
-	with _keyboard_lock:
-		versions = state.load(KEYBOARD_FILE) or {}
+	with state.edit(KEYBOARD_FILE, {}) as versions:
 		versions[str(chat_id)] = keyboard_version()
-		state.save(KEYBOARD_FILE, versions)
 
 def push_keyboard(chat_id):
 	"""Telegram can only change a reply keyboard by sending a message, so send a short one carrying it."""
@@ -75,16 +68,12 @@ def ensure_keyboard(chat_id):
 	"""DMs only (positive chat ids): bring a chat's keyboard up to date if its buttons have changed or it never had them."""
 	if int(chat_id) <= 0:
 		return
-	with _keyboard_lock:
-		current = (state.load(KEYBOARD_FILE) or {}).get(str(chat_id)) == keyboard_version()
-	if not current:
+	if state.load(KEYBOARD_FILE, {}).get(str(chat_id)) != keyboard_version():
 		push_keyboard(chat_id)
 
 def refresh_keyboards():
 	"""At startup: update every chat we know of whose keyboard is out of date."""
-	with _keyboard_lock:
-		versions = state.load(KEYBOARD_FILE) or {}
-	for chat_id, version in versions.items():
+	for chat_id, version in state.load(KEYBOARD_FILE, {}).items():
 		if version != keyboard_version():
 			try:
 				push_keyboard(chat_id)
diff --git a/lib/reports.py b/lib/reports.py
index 7fdd7552..c7ca86d1 100755
--- a/lib/reports.py
+++ b/lib/reports.py
@@ -4,7 +4,7 @@
 import sys
 from lib import config
 from lib import cache
-from lib import state
+from lib import watchlist
 from lib import sharesight
 from lib import util
 from lib import webhook
@@ -57,91 +57,37 @@ def doDelta(inputList):
 	return deltaString
 
 def prepare_watchlist(service, user, action=None, ticker=None):
-	if ticker:
-		ticker = ticker_orig = util.transform_to_yahoo(ticker)
-		ticker_link = util.finance_link(ticker, ticker, service)
-	duplicate = False
-	transformed = False
-	watchlist = state.load('finbot_watchlist.json') or [] # absent until the first add
-	if action == 'add':
-		if ticker in watchlist:
-			duplicate = True
-		else:
-			watchlist.append(ticker)
-	if len(watchlist):
-		market_data = yahoo.fetch(watchlist)
-	else:
-		market_data = None
-	print("")
-	if action == 'delete':
-		if ticker in watchlist:
-			watchlist.remove(ticker)
-		else:
-			print(ticker, "not in watchlist")
-	if action == 'add':
-		if '.' not in ticker and ticker not in market_data:
-			watchlist.remove(ticker)
-			ticker = ticker + '.AX'
-			transformed = True
-			ticker_link = util.finance_link(ticker, ticker, service)
-			print(ticker_orig, "not found. Trying", ticker)
-			if ticker in watchlist:
-				print(ticker, "already in watchlist")
-				duplicate = True
-			else:
-				watchlist.append(ticker)
-				market_data = yahoo.fetch(watchlist)
-				print("")
-				if ticker in market_data:
-					print("found", ticker)
-				else:
-					watchlist.remove(ticker)
-					print(ticker, "not found")
-		elif ticker not in market_data:
-			watchlist.remove(ticker)
-	def described(): # "Full Name (TICKER)" for the ticker being changed
-		title = ((market_data or {}).get(ticker) or {}).get('profile_title')
-		return webhook.bold(f"{title} ({ticker_link})" if title else ticker_link, service)
-	payload = []
-	if market_data:
-		for item in market_data:
-			flag = util.flag_from_ticker(item)
-			item_link = util.finance_link(item, market_data.get(item).get('profile_exchange'), service)
-			profile_title = market_data[item]['profile_title']
-			if item == ticker and action == 'delete':
-				pass
-			elif item == ticker and action == 'add': # highlight requested item
-				text = webhook.bold(webhook.italic(f"{flag} {profile_title} ({item_link})", service), service)
-				payload.append(text)
-			else:
-				payload.append(f"{flag} {profile_title} ({item_link})")
-	def profile_title_sort(e): # disregards markup in sort command
-		return re.findall('[A-Z].*', e)
-	payload.sort(key=profile_title_sort)
+	"""The .watchlist reply: the companies watched, or one line on adding ('add') or removing ('delete') a ticker."""
+	if action:
+		return [watchlist_change(service, user, action, util.transform_to_yahoo(ticker))]
+	watched = watchlist.tickers()
+	market_data = yahoo.fetch(watched) if watched else {}
+	lines = sorted((f"{util.flag_from_ticker(t)} {quote['profile_title']} ({util.finance_link(t, quote.get('profile_exchange'), service)})"
+		for t, quote in market_data.items()), key=lambda line: re.findall('[A-Z].*', line)) # by name, the flag aside
+	if not lines:
+		return ['Watchlist is empty. Try ".watchlist add SYMBOL" to create it']
+	return [f"Hi {user}, I'm currently tracking:"] + lines
+
+def watchlist_change(service, user, action, asked):
+	"""The reply to adding or removing `asked` (a Yahoo ticker)."""
+	def described(ticker, quote): # "Full Name (TICKER)", bold
+		quote = quote or {}
+		link = util.finance_link(ticker, quote.get('profile_exchange') or ticker, service)
+		return webhook.bold(f"{quote['profile_title']} ({link})" if quote.get('profile_title') else link, service)
 	if action == 'delete':
-		if not market_data or ticker not in market_data:
-			payload.insert(0, "Beep Boop. I could not find " + webhook.bold(ticker, service) + " to remove it")
-		else:
-			payload.insert(0, f"Ok {user}, I deleted " + described())
-	elif action == 'add':
-		if ticker not in market_data:
-			payload = ["Beep Boop. I could not find " + webhook.bold(ticker_orig, service) + " to add it"]
-		elif transformed and duplicate:
-			payload.insert(0, "Beep Boop. I could not find " + webhook.bold(ticker_orig, service) + " and I'm already tracking " + described())
-		elif transformed:
-			payload.insert(0, "Beep Boop. I could not find " + webhook.bold(ticker_orig, service) + " so I added " + described())
-		elif duplicate:
-			payload.insert(0, f"{user}, I'm already tracking " + described())
-		else:
-			payload.insert(0, f"Ok {user}, I added " + described())
-	elif not action and payload:
-		payload.insert(0, f"Hi {user}, I'm currently tracking:")
-	else:
-		payload.append('Watchlist is empty. Try ".watchlist add SYMBOL" to create it')
-	state.save('finbot_watchlist.json', watchlist)
-	if action in ('add', 'delete'):
-		payload = payload[:1] # confirmation only; don't list the whole watchlist
-	return payload
+		if asked not in watchlist.tickers():
+			return f"Beep Boop. I could not find {webhook.bold(asked, service)} to remove it"
+		quote = yahoo.fetch([asked]).get(asked)
+		watchlist.remove(asked)
+		return f"Ok {user}, I deleted {described(asked, quote)}"
+	found, quote = watchlist.find(asked)
+	if not quote:
+		return f"Beep Boop. I could not find {webhook.bold(asked, service)} to add it"
+	added = watchlist.add(found)
+	if found != asked:
+		done = "so I added" if added else "and I'm already tracking"
+		return f"Beep Boop. I could not find {webhook.bold(asked, service)} {done} {described(found, quote)}"
+	return f"Ok {user}, I added {described(found, quote)}" if added else f"{user}, I'm already tracking {described(found, quote)}"
 
 def prepare_compare(service, args):
 	"""Compare 2-6 tickers as rebased % change. A trailing time period (3m/1y/90d/ytd) is optional;
@@ -508,16 +454,10 @@ def prepare_beta_payload(service):
 
 def prepare_profile_payload(service, user, ticker):
 	"""The .SYMBOL report: what Yahoo knows about one company. A field Yahoo did not send leaves its line out."""
-	ticker = ticker_orig = util.transform_to_yahoo(ticker)
+	ticker_orig = util.transform_to_yahoo(ticker)
 	now = datetime.datetime.now()
 	payload = []
-	market_data = yahoo.fetch_detail(ticker, 600)
-	print("")
-	if ticker not in market_data and '.' not in ticker:
-		ticker = ticker + '.AX'
-		print("trying again with", ticker)
-		market_data = yahoo.fetch_detail(ticker, 600)
-		print("")
+	ticker, market_data = yahoo.resolve(ticker_orig, lambda symbol: yahoo.fetch_detail(symbol, 600))
 	if ticker not in market_data:
 		return [f"{user} 🛑 Beep Boop. I could not find {ticker_orig}"]
 	data = market_data[ticker]
diff --git a/lib/state.py b/lib/state.py
index 0edb81af..a709dec1 100644
--- a/lib/state.py
+++ b/lib/state.py
@@ -37,11 +37,12 @@ def _lock(name):
 @contextlib.contextmanager
 def edit(name, default):
 	"""Change a state file in place: `with state.edit(name, {}) as data: data[key] = value`. The file is read as the block starts
-	and written when it ends without an error; meanwhile no other edit of the same file runs in this process, so two chats
-	changing it at once both count. Keep slow work, a network call say, outside the block."""
+	(a copy of `default` when there is none yet) and written as the block ends without an error, if the block changed it.
+	Meanwhile no other edit of the same file runs in this process, so two chats changing it at once both count. Keep slow work,
+	a network call say, outside the block."""
 	with _lock(name):
-		data = load(name)
-		if data is None:
-			data = copy.deepcopy(default)
+		data = load(name, copy.deepcopy(default))
+		before = copy.deepcopy(data)
 		yield data
-		save(name, data)
+		if data != before:
+			save(name, data)
diff --git a/lib/util.py b/lib/util.py
index a727495a..3268ee7d 100644
--- a/lib/util.py
+++ b/lib/util.py
@@ -4,7 +4,7 @@
 import re
 from lib import config # first: sets MPLCONFIGDIR before matplotlib is imported (lazily, by the chart functions)
 from lib import sharesight
-from lib import state
+from lib import watchlist
 
 
 TITLE_TRIMS = [ # transform_title: (text, replacement), in order, as each sees what the ones before it made
@@ -768,7 +768,7 @@ def days_from_human_days(arg):
 
 def get_holdings_and_watchlist():
 	tickers = set(sharesight.get_holdings_wrapper())
-	tickers.update(state.load('finbot_watchlist.json', [])) # absent until the first add
+	tickers.update(watchlist.tickers())
 	if 'GOOG' in tickers and 'GOOGL' in tickers:
 		tickers.remove("GOOGL")
 	return sorted(tickers)
diff --git a/lib/watchlist.py b/lib/watchlist.py
new file mode 100644
index 00000000..b1ba936f
--- /dev/null
+++ b/lib/watchlist.py
@@ -0,0 +1,36 @@
+"""The watchlist: tickers followed beside the Sharesight holdings, in DATA_DIR/finbot_watchlist.json. Every call reads the file,
+so the bot and its scheduled reports (other processes) always see the same list."""
+from lib import state
+from lib import yahoo
+
+FILE = 'finbot_watchlist.json'
+
+
+def tickers():
+	"""The watched tickers, in the order they were added."""
+	return state.load(FILE, [])
+
+
+def find(ticker):
+	"""(ticker, its Yahoo quote) for a ticker Yahoo knows, tried on the ASX when a bare symbol is unknown (BHP: BHP.AX); the
+	quote is None when Yahoo knows neither."""
+	ticker, found = yahoo.resolve(ticker, lambda symbol: yahoo.fetch([symbol]))
+	return ticker, found.get(ticker)
+
+
+def add(ticker):
+	"""Watch a ticker (check it with find first). False when it was already watched."""
+	with state.edit(FILE, []) as watched:
+		if ticker in watched:
+			return False
+		watched.append(ticker)
+	return True
+
+
+def remove(ticker):
+	"""Stop watching a ticker. False when it was not watched."""
+	with state.edit(FILE, []) as watched:
+		if ticker not in watched:
+			return False
+		watched.remove(ticker)
+	return True
diff --git a/lib/worker.py b/lib/worker.py
index 74966386..e48522eb 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -348,10 +348,10 @@ def on_history(request, m):
 	request.send(caption.split('\n'), image, charts.keyboard('h', [ticker], 'x') if image and request.service == 'telegram' else None)
 
 def on_plan(request, m):
-	plan = state.load('finbot_plan.json') or {}
 	if m.group(1) and not m.group(1).startswith('@'):
-		plan[request.user] = m.group(1)
-		state.save('finbot_plan.json', plan)
+		with state.edit('finbot_plan.json', {}) as plans:
+			plans[request.user] = m.group(1)
+	plan = state.load('finbot_plan.json', {})
 	def who(user): # a Slack user is a mention (<@U123>), which is markup to keep; elsewhere a handle is text
 		user = user.removeprefix('@')
 		return user if request.service == 'slack' else webhook.escape(user, request.service)
diff --git a/lib/yahoo.py b/lib/yahoo.py
index bd9b77fd..d9f7875a 100644
--- a/lib/yahoo.py
+++ b/lib/yahoo.py
@@ -430,6 +430,15 @@ def fetch_details(tickers, strict=False):
 			details |= detail
 	return details
 
+def resolve(ticker, lookup):
+	"""(ticker, lookup(ticker)) for a ticker Yahoo knows: a bare symbol it does not know is tried on the ASX (BHP: BHP.AX).
+	lookup is fetch-shaped ({ticker: data}, without the tickers Yahoo does not know); its {} means Yahoo knows neither."""
+	found = lookup(ticker)
+	if ticker not in found and '.' not in ticker:
+		ticker += '.AX'
+		found = lookup(ticker)
+	return ticker, found
+
 def prefetch_history(tickers, workers=8):
 	"""Warm the price-history cache for many tickers at once, so the per-ticker lookups that follow are cache hits."""
 	tickers = list(tickers)
diff --git a/milestone.py b/milestone.py
index e2748705..2e38fcf7 100755
--- a/milestone.py
+++ b/milestone.py
@@ -17,7 +17,7 @@ def prepare_milestone_payload(service, market_data):
 		payload = []
 		payloadhigh = []
 		payloadlow = []
-		records = state.load('finbot_milestone.json') or {}
+		changed = {} # ticker -> its records after this run
 		for ticker in tickers:
 			try:
 				oldhigh, oldlow = records[ticker]['record_high'], records[ticker]['record_low']
@@ -34,7 +34,7 @@ def prepare_milestone_payload(service, market_data):
 			title = market_data[ticker]['profile_title']
 			ticker_link = util.finance_link(ticker, market_data[ticker]['profile_exchange'], service)
 			emoji = util.flag_from_ticker(ticker)
-			records[ticker] = {'record_high': max(oldhigh, newhigh), 'record_low': min(oldlow, newlow)} # all-time: a 52-week high below it is no record
+			changed[ticker] = {'record_high': max(oldhigh, newhigh), 'record_low': min(oldlow, newlow)} # all-time: a 52-week high below it is no record
 			if newhigh > oldhigh:
 				print("DEBUG", ticker, newhigh, "is higher than", oldhigh) if config.prefs.debug else None
 				payloadhigh.append(f"{emoji} {title} ({ticker_link})")
@@ -55,18 +55,20 @@ def prepare_milestone_payload(service, market_data):
 			payload.append(message)
 			payload = payload + payloadlow
 		payload = [i[0] for i in groupby(payload)] # de-dupe white space
-		return payload, records
+		return payload, changed
 
 	# MAIN #
 
 	tickers = util.get_holdings_and_watchlist()
 	market_data = yahoo.fetch(tickers)
+	records, changed = state.load('finbot_milestone.json', {}), {}
 	for service, url, chat in webhook.default_chats():
-		payload, records = prepare_milestone_payload(service, market_data)
+		payload, changed = prepare_milestone_payload(service, market_data)
 		webhook.send(service, url, chat, payload)
 
-	if records:
-		state.save('finbot_milestone.json', records)
+	if changed:
+		with state.edit('finbot_milestone.json', {}) as recorded:
+			recorded.update(changed)
 
 def job(argv):
 	return 'milestone.py', {}
diff --git a/podcasts.py b/podcasts.py
index a2d62f04..1eb93b2f 100755
--- a/podcasts.py
+++ b/podcasts.py
@@ -88,7 +88,7 @@ def fetch_new_episodes(feed_url, seen):
 def lambda_handler(feed_url):
 	"""Post the feed's newest episodes not posted before, then record every new one as seen."""
 	seen_file = get_seen_filename(feed_url)
-	seen = set(state.load(seen_file) or [])
+	seen = set(state.load(seen_file, []))
 	try:
 		episodes, guids = fetch_new_episodes(feed_url, seen)
 	except Exception as e:
@@ -104,7 +104,8 @@ def lambda_handler(feed_url):
 			duration = f" {episode['duration']}" if episode['duration'] else ""
 			payload.append(f"📣 {podcast_name}: {util.link(episode['url'], title, service)}{duration} ({episode['ago']})")
 		webhook.send(service, url, chat, payload)
-	state.save(seen_file, sorted(seen | guids)) # last, so that a failure above leaves these episodes for next run
+	with state.edit(seen_file, []) as recorded: # last, so that a failure above leaves these episodes for next run
+		recorded[:] = sorted(guids.union(recorded))
 
 def job(argv):
 	"""Arguments of a scheduled run: (label, lambda_handler kwargs). Raises ValueError with the usage on a bad argument."""
diff --git a/rating.py b/rating.py
index a35662eb..d1807646 100755
--- a/rating.py
+++ b/rating.py
@@ -55,8 +55,7 @@ def get_emoji(old, new):
 	# MAIN #
 
 	tickers = [util.transform_to_yahoo(specific_stock)] if specific_stock else util.get_holdings_and_watchlist()
-	old = state.load('finbot_rating.json')
-	old = old if isinstance(old, dict) else {} # absent until the first run
+	old = state.load('finbot_rating.json', {})
 	market_data = yahoo.fetch_details(tickers, strict=bool(specific_stock))
 
 	for service, url, chat in webhook.destinations(interactive, service, chat_id):
@@ -64,7 +63,8 @@ def get_emoji(old, new):
 		webhook.send(service, url, chat, payload)
 
 	if new and not interactive: # added to what is recorded: a ticker missing from this run keeps its last rating
-		state.save('finbot_rating.json', old | new)
+		with state.edit('finbot_rating.json', {}) as ratings:
+			ratings.update(new)
 
 def job(argv):
 	return 'rating.py', {}
diff --git a/tests/common.py b/tests/common.py
index 0febcfcb..b46cb934 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -10,7 +10,7 @@
 os.environ.update(telegramOutgoingWebhook='https://example.com/telegram', telegramOutgoingToken='tok', telegramChatID='1')
 
 import bot # first, and once: it monkey-patches with gevent, which must happen before requests imports ssl and never mid-test
-from lib import cache, config, state, webhook, util, yahoo, sharesight, shortman, worker, charts, telegram, reports, words
+from lib import cache, config, state, webhook, util, yahoo, sharesight, shortman, worker, charts, telegram, reports, watchlist, words
 import performance, shorts, price, reminder, cal, trades, milestone, rating
 
 REAL = {(m.__name__, n): getattr(m, n) for m, n in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (yahoo, 'price_history'),
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 366a0cdb..4a1c61f0 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -1,5 +1,5 @@
 """Failure and no-data scenarios: empty or broken upstream data, bad requests, rejected callers."""
-import contextlib, io, json, os, signal, subprocess, sys, tempfile, time, unittest
+import contextlib, io, json, os, signal, subprocess, sys, tempfile, threading, time, unittest
 from pathlib import Path
 import gevent
 from tests.common import *
@@ -79,6 +79,33 @@ def test_add_duplicate_delete_and_unknown_action(self):
 		self.assertEqual(state.load('finbot_watchlist.json'), [])
 		self.assertIn('not a valid watchlist action', str(self.command('.watchlist frobnicate MSFT')))
 
+	def test_a_bare_symbol_yahoo_does_not_know_is_tried_on_the_asx(self):
+		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts if t.endswith('.AX')})
+		self.assertIn('so I added', str(self.command('.watchlist add BHP')))
+		self.assertIn('already tracking', str(self.command('.watchlist add BHP')))
+		self.assertIn('could not find', str(self.command('.watchlist add BHP.L'))) # a ticker with a market is not retried
+		self.assertEqual(watchlist.tickers(), ['BHP.AX'])
+
+	def test_two_chats_adding_at_once_both_count(self):
+		# Yahoo is asked before the file is opened: both adds used to read the list, wait on Yahoo, and the later write won
+		def slow_fetch(tickers):
+			time.sleep(0.01)
+			return {t: market_entry(t) for t in tickers}
+		self.patch(yahoo, 'fetch', slow_fetch)
+		threads = [threading.Thread(target=reports.prepare_watchlist, args=('telegram', '@u', 'add', t)) for t in ('T1', 'T2')]
+		for thread in threads:
+			thread.start()
+		for thread in threads:
+			thread.join()
+		self.assertEqual(sorted(watchlist.tickers()), ['T1', 'T2'])
+
+	def test_removing_a_ticker_yahoo_no_longer_knows(self):
+		state.save('finbot_watchlist.json', ['GONE', 'T1'])
+		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts if t != 'GONE'})
+		self.assertIn('I deleted', str(self.command('.watchlist rm GONE')))
+		self.assertIn('could not find', str(self.command('.watchlist rm GONE')))
+		self.assertEqual(watchlist.tickers(), ['T1'])
+
 
 class BadArguments(FinbotCase):
 	def test_compare_argument_errors_give_usage(self):
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 8bdb222d..27127c26 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -238,6 +238,13 @@ def test_trades_asks_yahoo_only_for_new_trades(self):
 		trades.lambda_handler() # the trade is known now: nothing to report and nothing to fetch
 		self.assertEqual(fetched, [])
 
+	def test_a_scheduled_run_keeps_the_trades_another_run_recorded_meanwhile(self):
+		meanwhile = lambda tickers: state.save('finbot_sharesight_trades.json', [999]) or {t: market_entry(t) for t in tickers}
+		self.patch(yahoo, 'fetch', meanwhile) # called after the seen trades were read, before the new ones are recorded
+		trades.lambda_handler()
+		known = state.load('finbot_sharesight_trades.json')
+		self.assertEqual((known[0], len(known)), (999, 2))
+
 	def test_milestone(self):
 		milestone.lambda_handler()
 		self.assertTrue(self.sent)
diff --git a/tests/test_util.py b/tests/test_util.py
index 92a46c59..6f21f283 100644
--- a/tests/test_util.py
+++ b/tests/test_util.py
@@ -1,5 +1,5 @@
 """Pure helpers in lib/util.py: table-driven pass and fail cases."""
-import datetime, json, unittest
+import datetime, json, threading, unittest
 from pathlib import Path
 from tests.common import *
 
@@ -114,6 +114,32 @@ def disk_full(*a, **k):
 		state.save('state.json', {'a': 3})
 		self.assertEqual((json.loads(path.read_text()), oct(path.stat().st_mode & 0o777)), ({'a': 3}, '0o640'))
 
+	def test_edit_writes_only_a_change_and_never_changes_the_default(self):
+		default = {'seen': []}
+		with state.edit('edit.json', default):
+			pass
+		self.assertIsNone(state.load('edit.json')) # nothing changed, nothing written
+		with state.edit('edit.json', default) as data:
+			data['seen'].append(1)
+		self.assertEqual((state.load('edit.json'), default), ({'seen': [1]}, {'seen': []}))
+		with self.assertRaises(ValueError), state.edit('edit.json', default) as data:
+			data['seen'].append(2)
+			raise ValueError
+		self.assertEqual(state.load('edit.json'), {'seen': [1]}) # a block that fails writes nothing
+
+	def test_edits_at_once_all_count(self):
+		# each edit reads the file after the one before it wrote, so none overwrites another with what it read earlier
+		def edit(key):
+			with state.edit('edit.json', {}) as data:
+				time.sleep(0.01) # the other edits start meanwhile
+				data[key] = True
+		threads = [threading.Thread(target=edit, args=(key,)) for key in 'abc']
+		for thread in threads:
+			thread.start()
+		for thread in threads:
+			thread.join()
+		self.assertEqual(state.load('edit.json'), dict.fromkeys('abc', True))
+
 	def test_read_cache_missing_or_expired_is_falsy(self):
 		self.assertFalse(cache.read('absent.json', 60))
 		cache.write('old.json', {'a': 1})
diff --git a/trades.py b/trades.py
index b04f489e..2829604e 100755
--- a/trades.py
+++ b/trades.py
@@ -90,7 +90,7 @@ def prepare_trade_payload(service, trades):
 	# MAIN #
 	newtrades = set()
 	portfolios = sharesight.pick_portfolios(portfolio_select)
-	known_trades = set(state.load(state_file) or ())
+	known_trades = set(state.load(state_file, []))
 	trades = [trade for name, portfolio_id in portfolios.items() for trade in sharesight.get_trades(name, portfolio_id, days)]
 
 	for service, url, chat in webhook.destinations(interactive, service, chat_id):
@@ -99,10 +99,9 @@ def prepare_trade_payload(service, trades):
 			payload = [f"{user} No trades {util.days_english(days, 'in the past ')}. {random.choice(words.noTradesVerb)}"]
 		webhook.send(service, url, chat, payload)
 
-	# write state file
 	if newtrades and not interactive:
-		known_trades.update(newtrades)
-		state.save(state_file, list(known_trades))
+		with state.edit(state_file, []) as known:
+			known.extend(sorted(newtrades.difference(known)))
 
 def job(argv):
 	return 'trades.py', {}

From 7d9dcf244a08320c78ed7110922bd3567813e98f Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Wed, 14 Oct 2026 23:46:53 +0000
Subject: [PATCH 10/81] Sharesight portfolios are objects:
 sharesight.portfolios() gives Portfolio(name, id)

A Portfolio has holdings(), tickers(), performance(days) and trades(days); sharesight.portfolios(name=None) picks every
portfolio or the one named (in any case, else PortfolioNotFound), and sharesight.tickers() is everything held. They
replace pick_portfolios, get_holdings, get_holdings_new, get_holdings_wrapper and get_performance_wrapper, so callers
pass one object instead of a (name, id) pair. The cached, one-at-a-time requests stay in the module functions.

A performance answer without a report is now an error naming the portfolio (it was a KeyError: 'report' further on),
and is not cached. Output is otherwise unchanged.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md               |   2 +
 lib/reports.py          |  27 +++++-----
 lib/sharesight.py       | 109 ++++++++++++++++++----------------------
 lib/util.py             |   2 +-
 performance.py          |   4 +-
 price.py                |   4 +-
 tests/common.py         |   3 +-
 tests/test_edges.py     |   4 +-
 tests/test_providers.py |  16 ++++--
 tests/test_reports.py   |   8 +--
 tests/test_util.py      |   2 +-
 trades.py               |   3 +-
 12 files changed, 91 insertions(+), 93 deletions(-)

diff --git a/CLAUDE.md b/CLAUDE.md
index 4b9fc585..9dce94db 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -34,6 +34,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - `lib/charts.py`: Telegram period buttons and the rebuild behind them. `lib/util.py`: formatting, links, market tables, all
   matplotlib rendering.
 - `lib/yahoo.py lib/sharesight.py lib/shortman.py lib/simplywallst.py`: data sources, cached in `DATA_DIR/cache`.
+  `sharesight.portfolios(name=None)` gives `Portfolio` objects: `holdings()`, `tickers()`, `performance(days)`,
+  `trades(days)`.
 - `lib/cache.py`: cache files (any may vanish at any time). `lib/state.py`: state files in `DATA_DIR`, which must last;
   change one with `with state.edit(name, default) as data:`, never load-then-save. `lib/watchlist.py`: the watchlist.
 - `lib/config.py`: `config.env` (finbot.env, fixed at start) and `config.prefs` (preferences.conf, a new object on each reload).
diff --git a/lib/reports.py b/lib/reports.py
index c7ca86d1..6b4def1c 100755
--- a/lib/reports.py
+++ b/lib/reports.py
@@ -250,36 +250,37 @@ def prepare_help(service, botName):
 
 def prepare_holdings_payload(portfolio_name, service):
 	"""The holdings of the portfolio called `portfolio_name`. Raises sharesight.PortfolioNotFound."""
-	portfolios = sharesight.get_portfolios()
 	if not portfolio_name:
-		return [".holdings: please try again specifying a portfolio:"] + list(portfolios)
-	(name, portfolio_id), = sharesight.pick_portfolios(portfolio_name, portfolios).items()
-	holdings = sharesight.get_holdings_new(name, portfolio_id)
+		names = [portfolio.name for portfolio in sharesight.portfolios()]
+		return [".holdings: please try again specifying a portfolio:"] + names
+	portfolio, = sharesight.portfolios(portfolio_name)
+	holdings = portfolio.holdings()
 	payload = sorted(holding_line(holding, service) for holding in holdings.values())
 	if payload:
-		payload.insert(0, webhook.bold(f"Holdings for {name} ({len(holdings)})", service))
+		payload.insert(0, webhook.bold(f"Holdings for {portfolio.name} ({len(holdings)})", service))
 	return payload
 
 def prepare_who(service, code):
 	"""Which portfolios hold `code` (a ticker without its exchange), each holding linked to Sharesight."""
 	payload = []
-	for portfolio_name, portfolio_id in sharesight.get_portfolios().items():
-		holdings = sharesight.get_holdings_new(portfolio_name, portfolio_id)
-		found = [holding_line(holding, service, code) for holding in holdings.values() if holding['code'].upper() == code]
+	for portfolio in sharesight.portfolios():
+		found = [holding_line(holding, service, code)
+			for holding in portfolio.holdings().values() if holding['code'].upper() == code]
 		if found:
-			payload += [webhook.bold(f"{portfolio_name}:", service), "\n".join(found), ""]
+			payload += [webhook.bold(f"{portfolio.name}:", service), "\n".join(found), ""]
 	return payload or [f'{code} not found in any portfolio']
 
 def prepare_pie(portfolio_name, by, service, max_slices=8):
 	"""Portfolio value by industry (or by='sector'). Returns (payload, chart); chart is None unless there is something to draw.
 	Raises sharesight.PortfolioNotFound."""
-	portfolios = sharesight.get_portfolios()
 	if not portfolio_name:
-		return [".pie: please try again specifying a portfolio:"] + list(portfolios), None
-	(name, portfolio_id), = sharesight.pick_portfolios(portfolio_name, portfolios).items()
+		names = [portfolio.name for portfolio in sharesight.portfolios()]
+		return [".pie: please try again specifying a portfolio:"] + names, None
+	portfolio, = sharesight.portfolios(portfolio_name)
+	name = portfolio.name
 	key = 'sector_classification_name' if by == 'sector' else 'industry_classification_name'
 	totals = {}
-	for holding in sharesight.get_holdings_new(name, portfolio_id).values():
+	for holding in portfolio.holdings().values():
 		try:
 			value = float(holding.get('value'))
 		except (TypeError, ValueError):
diff --git a/lib/sharesight.py b/lib/sharesight.py
index 466ec446..a9fa941b 100644
--- a/lib/sharesight.py
+++ b/lib/sharesight.py
@@ -1,7 +1,8 @@
-import sys
-import time
+import dataclasses
 import datetime
+import sys
 import threading
+import time
 import requests
 from lib import config
 from lib import cache
@@ -95,20 +96,51 @@ def get_portfolios(refresh=False):
 
 class PortfolioNotFound(LookupError):
 	"""No portfolio is called what was asked for. `lines` is the answer to give: the name, and the portfolios there are."""
-	def __init__(self, name, portfolios):
-		self.lines = [f"{name} portfolio not found. I only know about:"] + list(portfolios)
+	def __init__(self, name, names):
+		self.lines = [f"{name} portfolio not found. I only know about:"] + list(names)
 		super().__init__(' '.join(self.lines))
 
-def pick_portfolios(name=None, portfolios=None):
-	"""{name as Sharesight spells it: id} of the portfolio called `name` (in any case), or of every portfolio when no name is given.
-	`portfolios` is get_portfolios() when not given. Raises PortfolioNotFound."""
-	portfolios = get_portfolios() if portfolios is None else portfolios
-	if not name:
-		return portfolios
-	found = {k.lower(): (k, v) for k, v in portfolios.items()}.get(name.lower())
-	if not found:
-		raise PortfolioNotFound(name, portfolios)
-	return dict([found])
+@dataclasses.dataclass(frozen=True)
+class Portfolio:
+	"""A Sharesight portfolio. Its methods ask Sharesight through the module functions (cached, one request at once)."""
+	name: str # as Sharesight spells it
+	id: int
+
+	def performance(self, days=None):
+		"""Sharesight's performance report over the past `days` (past_days when None); days=0 gives today's holdings."""
+		return get_performance(self.id, config.prefs.past_days if days is None else days)
+
+	def holdings(self):
+		"""{'CODE:MARKET': Sharesight's instrument, with the holding's id and value added} of what it holds."""
+		holdings = {}
+		for item in self.performance(0)['report']['holdings']:
+			instrument = item['instrument']
+			key = instrument['code'] + ':' + instrument['market_code']
+			holdings[key] = instrument | {'holding_id': item['id'], 'value': item.get('value')}
+		return holdings
+
+	def tickers(self):
+		"""What it holds, as Yahoo tickers, sorted."""
+		return sorted({util.transform_to_yahoo(each['code'], each['market_code']) for each in self.holdings().values()})
+
+	def trades(self, days=None):
+		"""Its trades of the past `days` (the past_days preference when None)."""
+		return get_trades(self.name, self.id, days)
+
+def portfolios(select=None):
+	"""The portfolios to report on (after include_portfolios and exclude_portfolios), or the one called `select`, in any
+	case. Raises PortfolioNotFound for a name that is none of them."""
+	found = [Portfolio(name, portfolio_id) for name, portfolio_id in get_portfolios().items()]
+	if not select:
+		return found
+	chosen = [portfolio for portfolio in found if portfolio.name.lower() == select.lower()]
+	if not chosen:
+		raise PortfolioNotFound(select, [portfolio.name for portfolio in found])
+	return chosen
+
+def tickers():
+	"""Every portfolio's holdings as Yahoo tickers, sorted."""
+	return sorted({ticker for portfolio in portfolios() for ticker in portfolio.tickers()})
 
 def get_trades(portfolio_name, portfolio_id, days=None):
 	if days is None:
@@ -133,39 +165,6 @@ def get_trades(portfolio_name, portfolio_id, days=None):
 		cache.write(cache_file, data)
 	return data['trades']
 
-def get_holdings(portfolio_name, portfolio_id):
-	tickers = set()
-	holdings = get_holdings_new(portfolio_name, portfolio_id)
-	for holding in holdings:
-		symbol = holdings[holding]['code']
-		market = holdings[holding]['market_code']
-		ticker = util.transform_to_yahoo(symbol, market)
-		tickers.add(ticker)
-	return sorted(tickers)
-
-def get_holdings_new(portfolio_name, portfolio_id):
-	print("Fetching Sharesight holdings", portfolio_name, end=": ")
-	data = get_performance(portfolio_id, 0)
-	print(len(data['report']['holdings']))
-	holdings = {}
-	for item in data['report']['holdings']:
-		code = item['instrument']['code']
-		market_code = item['instrument']['market_code']
-		symbol = code + ":" + market_code
-		holdings[symbol] = item['instrument']
-		holdings[symbol]['holding_id'] = item['id']
-		holdings[symbol]['value'] = item.get('value')
-	return holdings
-
-def get_holdings_wrapper():
-	tickers = set()
-	portfolios = get_portfolios()
-	if not portfolios:
-		return None
-	for portfolio_name, portfolio_id in portfolios.items():
-		tickers.update(get_holdings(portfolio_name, portfolio_id))
-	return sorted(tickers)
-
 def get_performance(portfolio_id, days, seconds=None, refresh=False):
 	if seconds is None:
 		seconds = config.prefs.performance_cache_seconds
@@ -187,22 +186,12 @@ def get_performance(portfolio_id, days, seconds=None, refresh=False):
 			continue
 		break
 	_check(data, 'performance')
-	if config.prefs.cache and 'report' in data:
+	if 'report' not in data:
+		raise RuntimeError(f"Sharesight sent no performance report for portfolio {portfolio_id}")
+	if config.prefs.cache:
 		cache.write(cache_file, data)
 	return data
 
-def get_performance_wrapper(days=None):
-	if days is None:
-		days = config.prefs.past_days
-	performance = {}
-	portfolios = get_portfolios()
-	for portfolio_name, portfolio_id in portfolios.items():
-		performance[portfolio_id] = get_performance(portfolio_id, days)
-		if not performance[portfolio_id]:
-			print("Could not get performance for portfolio:", portfolio_id, file=sys.stderr)
-			raise RuntimeError(f"Sharesight returned no performance for portfolio {portfolio_id}")
-	return performance
-
 def warm(margin=0.9):
 	"""Refresh the portfolio list and each portfolio's holdings report (days=0) once they reach `margin` of their cache lifetime,
 	so a request never has to wait on Sharesight. Returns (the number of fetches made, the portfolios): the caller can hand
diff --git a/lib/util.py b/lib/util.py
index 3268ee7d..a76fa71f 100644
--- a/lib/util.py
+++ b/lib/util.py
@@ -767,7 +767,7 @@ def days_from_human_days(arg):
 	return int(arg.removesuffix('Y')) * 365
 
 def get_holdings_and_watchlist():
-	tickers = set(sharesight.get_holdings_wrapper())
+	tickers = set(sharesight.tickers())
 	tickers.update(watchlist.tickers())
 	if 'GOOG' in tickers and 'GOOGL' in tickers:
 		tickers.remove("GOOGL")
diff --git a/performance.py b/performance.py
index 577e6cfc..314b9f5f 100755
--- a/performance.py
+++ b/performance.py
@@ -14,8 +14,8 @@
 def gather(past_days, portfolio_select=None):
 	"""{portfolio id: (its name, Sharesight's performance report over past_days)} of every portfolio, or of the one named (in any
 	case). Raises sharesight.PortfolioNotFound for a name that is none of them, and RuntimeError when there is no report."""
-	portfolios = sharesight.pick_portfolios(portfolio_select)
-	performance = {portfolio_id: (name, sharesight.get_performance(portfolio_id, past_days)) for name, portfolio_id in portfolios.items()}
+	performance = {portfolio.id: (portfolio.name, portfolio.performance(past_days))
+		for portfolio in sharesight.portfolios(portfolio_select)}
 	if not performance:
 		print("Error: no Sharesight data found", file=sys.stderr)
 		raise RuntimeError("no Sharesight data found")
diff --git a/price.py b/price.py
index a4b95de2..9e6bdb92 100755
--- a/price.py
+++ b/price.py
@@ -19,8 +19,8 @@ def gather(specific_stock=None, days=None):
 		yahoo.add_usd_market_caps(market_data)
 	# Sharesight reports a holding's performance only for the time it was held: held for 1Y and asked for 5Y, it gives 1Y
 	if days and (not specific_stock or config.prefs.performance_use_sharesight):
-		for data in sharesight.get_performance_wrapper(days).values():
-			for holding in data['report']['holdings']:
+		for portfolio in sharesight.portfolios():
+			for holding in portfolio.performance(days)['report']['holdings']:
 				ticker = util.transform_to_yahoo(holding['instrument']['code'], holding['instrument']['market_code'])
 				if ticker in market_data:
 					market_data[ticker]['percent_change_period'] = float(holding['capital_gain_percent'])
diff --git a/tests/common.py b/tests/common.py
index b46cb934..3f278f4d 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -14,7 +14,7 @@
 import performance, shorts, price, reminder, cal, trades, milestone, rating
 
 REAL = {(m.__name__, n): getattr(m, n) for m, n in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (yahoo, 'price_history'),
-	(sharesight, 'get_portfolios'), (sharesight, 'get_trades'), (sharesight, 'get_performance'), (sharesight, 'get_performance_wrapper'))}
+	(sharesight, 'get_portfolios'), (sharesight, 'get_trades'), (sharesight, 'get_performance'))}
 REAL_GET_HOLDINGS = util.get_holdings_and_watchlist
 REAL_PAYLOAD_WRAPPER = webhook.payload_wrapper # FinbotCase replaces it; tests of the real thing use this
 REAL_REPORT_ERROR = webhook.report_error # likewise
@@ -84,7 +84,6 @@ def setUp(self):
 		self.patch(shortman, 'fetch', lambda md: md)
 		holdings = [{'id': i, 'instrument': {'code': t, 'market_code': 'NASDAQ'}, 'portfolio': {'name': 'Rob'},
 					 'capital_gain_percent': str(i * 4 - 40), 'value': '1000'} for i, t in enumerate(TICKERS)]
-		self.patch(sharesight, 'get_performance_wrapper', lambda d: {1: {'report': {'holdings': holdings}}})
 		self.patch(sharesight, 'get_portfolios', lambda: {'Rob': 1})
 		self.patch(sharesight, 'get_performance', lambda pid, d, *a: {'report': {
 			'holdings': holdings, 'currency_gain_percent': '0', 'capital_gain_percent': '2.5', 'total_gain_percent': '0'}})
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 4a1c61f0..c2d99bfa 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -44,7 +44,6 @@ def down(*a, **k):
 	def test_sharesight_down_is_reported_once(self):
 		def down(*a, **k):
 			raise RuntimeError('Sharesight performance error: boom')
-		self.patch(sharesight, 'get_performance_wrapper', down)
 		self.patch(sharesight, 'get_performance', down)
 		self.patch(sharesight, 'get_portfolios', down)
 		self.sent.clear(); self.errors.clear()
@@ -133,7 +132,8 @@ def test_an_unknown_portfolio_gets_the_same_answer_everywhere(self):
 				self.assertEqual(self.command(cmd), [('text', ['Nope portfolio not found. I only know about:', 'Rob'])])
 
 	def test_a_portfolio_name_in_any_case(self):
-		self.patch(sharesight, 'get_holdings_new', lambda name, pid: {'T1:NASDAQ': {'code': 'T1', 'market_code': 'NASDAQ', 'name': 'Co T1', 'holding_id': 1}})
+		holding = {'code': 'T1', 'market_code': 'NASDAQ', 'name': 'Co T1', 'holding_id': 1}
+		self.patch(sharesight.Portfolio, 'holdings', lambda portfolio: {'T1:NASDAQ': holding})
 		self.assertIn('Holdings for Rob (1)', str(self.command('.holdings rob')))
 
 	def test_trades_with_no_buy_or_sell_still_get_an_answer(self):
diff --git a/tests/test_providers.py b/tests/test_providers.py
index d23ac476..75f2d05a 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -41,7 +41,7 @@ def setUp(self):
 		super().setUp()
 		# these tests exercise the real parsers, so undo the canned data from the base fixture
 		for module, name in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (sharesight, 'get_portfolios'),
-							 (sharesight, 'get_trades'), (sharesight, 'get_performance'), (sharesight, 'get_performance_wrapper')):
+							 (sharesight, 'get_trades'), (sharesight, 'get_performance')):
 			self.patch(module, name, REAL[(module.__name__, name)])
 		self.patch(yahoo, 'getCookie', lambda *a, **k: 'A=B')
 		self.patch(yahoo, 'getCrumb', lambda *a, **k: 'crumb')
@@ -185,7 +185,7 @@ def test_one_failed_ticker_does_not_take_the_report_down(self):
 		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts}) # quotes as canned in the base fixture, charts over HTTP
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t)})
-		self.patch(sharesight, 'get_performance_wrapper', lambda d: {}) # no Sharesight figures: every ticker needs its Yahoo history
+		self.patch(sharesight, 'get_portfolios', lambda: {}) # no Sharesight figures: every ticker needs its history
 		good = Resp(self.chart([100.0] * 20 + [110.0]))
 		def get(url, **kwargs):
 			return Resp(status=404) if '/chart/T3?' in url or '/chart/SPY?' in url else good
@@ -277,7 +277,7 @@ def test_trade_failures_raise(self):
 	def test_performance_and_holdings_from_real_response(self):
 		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_performance_684141_0.json')))
 		self.assertEqual(len(sharesight.get_performance(684141, 0)['report']['holdings']), 18)
-		tickers = sharesight.get_holdings('RobSMSF', 684141)
+		tickers = sharesight.Portfolio('RobSMSF', 684141).tickers()
 		self.assertIn('ARM', tickers) # US venues carry no Yahoo suffix
 		self.assertEqual(tickers, sorted(tickers))
 
@@ -295,6 +295,12 @@ def test_performance_gives_up_after_repeated_busy_or_hard_errors(self):
 				with self.assertRaises(RuntimeError):
 					sharesight.get_performance(684141, 7)
 
+	def test_an_answer_without_a_report_says_so(self):
+		self.http(sharesight, Resp({}))
+		with self.assertRaisesRegex(RuntimeError, 'no performance report for portfolio 684141'):
+			sharesight.get_performance(684141, 7)
+		self.assertIsNone(cache.read(sharesight.performance_cache_file(684141, 7))) # and is not cached
+
 
 class SharesightKeepWarm(ProviderCase):
 	"""The background refresher keeps the portfolio list and the holdings report (days=0) cached ahead of their TTLs."""
@@ -343,11 +349,11 @@ def test_a_request_never_waits_on_a_cache_the_job_has_warmed(self):
 		self.assertEqual(len(sharesight.get_performance(684141, 0)['report']['holdings']), 18)
 		self.assertEqual(calls, [])
 
-	def test_holdings_wrapper_is_served_entirely_from_the_warmed_cache(self):
+	def test_every_holding_is_served_from_the_warmed_cache(self):
 		calls = self.serve_sharesight()
 		sharesight.warm()[0]
 		calls.clear()
-		tickers = sharesight.get_holdings_wrapper() # get_portfolios() + get_holdings() per portfolio, no cache of its own
+		tickers = sharesight.tickers() # the portfolio list and each portfolio's holdings report, no cache of its own
 		self.assertIn('ARM', tickers)
 		self.assertEqual(calls, [])
 
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 27127c26..bece5552 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -400,12 +400,12 @@ def item(n, date):
 class PieCommand(FinbotCase):
 	def setUp(self):
 		super().setUp()
-		def holdings(name, pid):
+		def holdings(portfolio):
 			data = [('A', 'Semiconductors', 'Electronic Technology', 500), ('B', 'Semiconductors', 'Electronic Technology', 300),
 					('C', 'Biotechnology', 'Health Technology', 200), ('D', None, None, 100), ('E', 'Gold', 'Minerals', 0)]
 			return {c + ':NASDAQ': {'code': c, 'market_code': 'NASDAQ', 'industry_classification_name': i,
 									'sector_classification_name': s, 'value': v} for c, i, s, v in data}
-		self.patch(sharesight, 'get_holdings_new', holdings)
+		self.patch(sharesight.Portfolio, 'holdings', holdings)
 
 	def test_by_industry_default(self):
 		sent = self.command('.pie Rob')
@@ -419,7 +419,9 @@ def test_by_sector_and_payload(self):
 		self.assertEqual(self.command('.pie Rob sector')[0][0], 'photo')
 
 	def test_folds_into_other_and_renders(self):
-		self.patch(sharesight, 'get_holdings_new', lambda n, p: {f'T{i}:X': {'code': f'T{i}', 'industry_classification_name': f'Ind {i}', 'value': 100 - i} for i in range(12)})
+		holdings = {f'T{i}:X': {'code': f'T{i}', 'industry_classification_name': f'Ind {i}', 'value': 100 - i}
+			for i in range(12)}
+		self.patch(sharesight.Portfolio, 'holdings', lambda portfolio: holdings)
 		payload, chart = reports.prepare_pie('Rob', 'industry', 'telegram')
 		self.assertEqual(len(chart['rows']), 8)
 		self.assertEqual(chart['rows'][-1][0], 'Other')
diff --git a/tests/test_util.py b/tests/test_util.py
index 6f21f283..1ffa8310 100644
--- a/tests/test_util.py
+++ b/tests/test_util.py
@@ -163,7 +163,7 @@ def test_binary_cache_roundtrip(self):
 
 	def test_holdings_and_watchlist_without_a_watchlist_file(self):
 		# first run: no watchlist exists yet
-		self.patch(util.sharesight, 'get_holdings_wrapper', lambda: ['AAPL', 'GOOG', 'GOOGL'])
+		self.patch(util.sharesight, 'tickers', lambda: ['AAPL', 'GOOG', 'GOOGL'])
 		self.patch(util, 'get_holdings_and_watchlist', REAL_GET_HOLDINGS)
 		self.assertEqual(util.get_holdings_and_watchlist(), ['AAPL', 'GOOG']) # GOOGL dropped when GOOG is present
 		state.save('finbot_watchlist.json', ['MSFT'])
diff --git a/trades.py b/trades.py
index 2829604e..ef5798df 100755
--- a/trades.py
+++ b/trades.py
@@ -89,9 +89,8 @@ def prepare_trade_payload(service, trades):
 
 	# MAIN #
 	newtrades = set()
-	portfolios = sharesight.pick_portfolios(portfolio_select)
 	known_trades = set(state.load(state_file, []))
-	trades = [trade for name, portfolio_id in portfolios.items() for trade in sharesight.get_trades(name, portfolio_id, days)]
+	trades = [trade for portfolio in sharesight.portfolios(portfolio_select) for trade in portfolio.trades(days)]
 
 	for service, url, chat in webhook.destinations(interactive, service, chat_id):
 		payload = prepare_trade_payload(service, trades)

From bd85fb1f3efd155e9a3045a327b354de156aec06 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Wed, 14 Oct 2026 23:51:00 +0000
Subject: [PATCH 11/81] Stdout prints removed: systemd discarded every one

finbot.service sends stdout to null, and job processes inherit it, so the 30 prints there (progress dots, "Preparing
outbound", "Fetching ...", skip notices, a debug payload dump) were never read. They go, and a test keeps the bot and its
reports printing only to stderr; the util/ tools, run by hand, still print to stdout. The one error among them, a data
or cache directory that cannot be created, moves to stderr.

On the way: get_trades printed len(data['trades']) before checking the answer, so a Sharesight error surfaced as
KeyError: 'trades'; get_portfolios' include/exclude filter is one expression.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md            |  3 ++-
 bot.py               |  3 ---
 cal.py               |  1 -
 cleanup.py           |  1 -
 lib/config.py        |  2 +-
 lib/reports.py       |  4 ----
 lib/sharesight.py    | 18 ++++--------------
 lib/shortman.py      |  5 +----
 lib/webhook.py       |  5 -----
 lib/yahoo.py         |  7 +------
 milestone.py         |  3 ---
 podcasts.py          |  1 -
 price.py             |  6 ++----
 shorts.py            |  1 -
 tests/test_finbot.py | 15 ++++++++++++++-
 trades.py            |  1 -
 16 files changed, 25 insertions(+), 51 deletions(-)

diff --git a/CLAUDE.md b/CLAUDE.md
index 9dce94db..7c175d07 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -53,7 +53,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - A report is a list of lines. Markup through `webhook.bold/italic(text, service)`, user-written text through
   `webhook.escape(text, service)`, links through `util.link`/`util.finance_link`.
 - Tabs everywhere, tests included (`.editorconfig`). New and changed lines stay under 120 characters (an edit quotes whole lines).
-- stdout is discarded under systemd (`StandardOutput=null`): anything worth logging goes to stderr.
+- Log with `print(..., file=sys.stderr)`. systemd discards stdout (`StandardOutput=null`), so the bot and its reports never
+  print there (a test checks); only the util/ tools, run by hand, do.
 - A chart renderer in `lib/util.py` gets `@_drawn` (the bot draws in one native thread); it must not call another `@_drawn`
   function, or that thread waits on itself.
 - A new command: `COMMANDS`, `prepare_help`, the README command list, and `command_keyboard` if it gets a button.
diff --git a/bot.py b/bot.py
index 1a127277..dceb2f2a 100755
--- a/bot.py
+++ b/bot.py
@@ -127,8 +127,6 @@ def reply(status, body=b''):
 			print("[Telegram]:", user, message, file=sys.stderr) if config.prefs.debug else None
 		elif "photo" in msg:
 			message = msg.get("caption", '')
-			file_id = msg["photo"][-1]["file_id"]
-			print("[Telegram photo]:", user, file_id, message)
 		elif IGNORED_TELEGRAM_CONTENT & msg.keys():
 			return reply('200 OK') # stickers, GIFs, voice notes and the like: nothing to answer, nothing worth logging
 		else:
@@ -144,7 +142,6 @@ def reply(status, body=b''):
 			print_body()
 			return reply('403 Forbidden', b'<h1>Unauthorized</h1>')
 		environ[AUTHENTICATED] = True
-		print("Incoming Slack request authenticated")
 		if 'type' not in inbound:
 			print(f"[{service}]: unhandled: no 'type'", file=sys.stderr)
 			return reply('200 OK', b'Unhandled')
diff --git a/cal.py b/cal.py
index 42da8465..d737e996 100755
--- a/cal.py
+++ b/cal.py
@@ -79,7 +79,6 @@ def pin_if_earnings(service, results):
 		market_data = yahoo.fetch(tickers)
 	else: # ex-dividend dates are in Yahoo's details
 		market_data = yahoo.fetch_details(tickers, strict=bool(specific_stock))
-		print("")
 
 	for service, url, chat in webhook.destinations(interactive, service, chat_id):
 		pin_if_earnings(service, webhook.send(service, url, chat, prepare_payload(service, market_data)))
diff --git a/cleanup.py b/cleanup.py
index 78ee9fd6..cb1eb62f 100755
--- a/cleanup.py
+++ b/cleanup.py
@@ -23,7 +23,6 @@ def lambda_handler(days=30):
 				removed += 1
 			except OSError as e:
 				print("cleanup: can't remove", entry.path, e, file=sys.stderr)
-	print(f"cleanup: removed {removed} cache files older than {days} days from {config.env.cache_dir}")
 	return removed
 
 def job(argv):
diff --git a/lib/config.py b/lib/config.py
index 15284210..49070c90 100644
--- a/lib/config.py
+++ b/lib/config.py
@@ -316,7 +316,7 @@ def _fail(e):
 	try:
 		os.makedirs(_dir, exist_ok=True)
 	except OSError as e: # e.g. read-only filesystem (Lambda)
-		print("can't create", _dir, e)
+		print("config: can't create", _dir, e, file=sys.stderr)
 
 # matplotlib needs a writable config/cache dir; without a home directory (systemd DynamicUser, containers) it fails
 os.environ.setdefault('MPLCONFIGDIR', os.path.abspath(env.cache_dir + '/matplotlib'))
diff --git a/lib/reports.py b/lib/reports.py
index 6b4def1c..a24004c2 100755
--- a/lib/reports.py
+++ b/lib/reports.py
@@ -523,16 +523,12 @@ def prepare_profile_payload(service, user, ticker):
 		earnings_date = datetime.datetime.fromtimestamp(data['earnings_date'])
 		if earnings_date > now:
 			payload.append(webhook.bold("Earnings date:", service) + f" {earnings_date:%b %d}")
-		else:
-			print("Skipping past earnings:", ticker, f"{earnings_date:%b %d}")
 	if data.get('dividend', 0) > 0:
 		payload.append(webhook.bold("Dividend:", service) + f" {data['dividend']}%")
 		if 'ex_dividend_date' in data:
 			ex_dividend_date = datetime.datetime.fromtimestamp(data['ex_dividend_date'])
 			if ex_dividend_date > now:
 				payload.append(webhook.bold("Ex-dividend date:", service) + f" {ex_dividend_date:%b %d}")
-			else:
-				print("Skipping past ex-dividend:", ticker, f"{ex_dividend_date:%b %d}")
 	cashflow = data.get('free_cashflow', data.get('operating_cashflow'))
 	if cashflow:
 		payload.append(webhook.bold("Cashflow positive:", service) + (" no ⚠️ " if cashflow < 0 else " yes"))
diff --git a/lib/sharesight.py b/lib/sharesight.py
index a9fa941b..53223f5d 100644
--- a/lib/sharesight.py
+++ b/lib/sharesight.py
@@ -57,7 +57,6 @@ def get_token():
 			else:
 				print("cache miss:", cacheFile, file=sys.stderr) if config.prefs.debug else None
 
-	print("Fetching Sharesight auth token")
 	data = _request(_post, "https://api.sharesight.com/oauth2/token", 'auth', data=config.env.sharesight_auth)
 	_check(data, 'auth')
 	if config.prefs.cache and 'access_token' in data:
@@ -72,26 +71,19 @@ def performance_cache_file(portfolio_id, days):
 
 def get_portfolios(refresh=False):
 	"""refresh=True skips the cache read and fetches again (the cache is rewritten)."""
-	portfolio_dict = {}
 	data = cache.read(PORTFOLIOS_CACHE, config.prefs.cache_seconds) if config.prefs.cache and not refresh else None
 	if not data:
 		token = get_token()
-		print("Fetching Sharesight portfolios")
 		data = _request(_get, "https://api.sharesight.com/api/v3/portfolios", 'portfolios', headers={'Content-type': 'application/json'}, auth=BearerAuth(token))
 		_check(data, 'portfolios')
 		if config.prefs.cache and 'portfolios' in data:
 			cache.write(PORTFOLIOS_CACHE, data)
-	for portfolio in data['portfolios']:
-		if str(portfolio['id']) in config.prefs.exclude_portfolios:
-			print(portfolio['id'], "(" + portfolio['name'] + ") in exclusion list. Skipping.")
-		elif str(portfolio['id']) not in config.prefs.include_portfolios and config.prefs.include_portfolios:
-			print("Exclusion list is defined and does not contain", portfolio['id'], "(" + portfolio['name'] + "). Skipping.")
-		else:
-			portfolio_dict[portfolio['name']] = portfolio['id']
-	if not len(portfolio_dict):
+	include, exclude = config.prefs.include_portfolios, config.prefs.exclude_portfolios
+	portfolio_dict = {portfolio['name']: portfolio['id'] for portfolio in data['portfolios']
+		if str(portfolio['id']) not in exclude and (not include or str(portfolio['id']) in include)}
+	if not portfolio_dict:
 		print("No portfolios found. Exiting.", file=sys.stderr)
 		raise RuntimeError("No Sharesight portfolios found")
-	print(portfolio_dict)
 	return portfolio_dict
 
 class PortfolioNotFound(LookupError):
@@ -153,11 +145,9 @@ def get_trades(portfolio_name, portfolio_id, days=None):
 	start_date = datetime.datetime.now() - datetime.timedelta(days=days)
 	start_date = start_date.strftime('%Y-%m-%d') # 2022-08-20
 	token = get_token()
-	print("Fetching Sharesight trades for", portfolio_name, end=": ")
 	url = 'https://api.sharesight.com/api/v2/portfolios/'
 	url = url + str(portfolio_id) + '/trades.json' + '?start_date=' + start_date
 	data = _request(_get, url, 'trades', auth=BearerAuth(token))
-	print(len(data['trades']))
 	_check(data, 'trades')
 	for trade in data['trades']:
 		trade['portfolio'] = portfolio_name # inject custom field
diff --git a/lib/shortman.py b/lib/shortman.py
index 489fb0e8..023a87a0 100644
--- a/lib/shortman.py
+++ b/lib/shortman.py
@@ -4,16 +4,13 @@
 
 def fetch(market_data):
 	local_market_data = market_data.copy()
-	print("Fetching ASX shorts from Shortman")
 	url = 'https://www.shortman.com.au/downloadeddata/latest.csv'
 	try:
 		r = net.get(url, timeout=config.prefs.http_timeout)
 	except Exception as e:
 		print("Failure fetching", url, net.redact(e), file=sys.stderr)
 		return market_data # the ASX short interest is missing, the rest of the report is not
-	if r.status_code == 200:
-		print(r.status_code, "success shortman")
-	else:
+	if r.status_code != 200:
 		print(r.status_code, "error communicating with", url, file=sys.stderr)
 		return market_data
 	csv = r.content.decode('utf-8')
diff --git a/lib/webhook.py b/lib/webhook.py
index b420d517..0aad651f 100644
--- a/lib/webhook.py
+++ b/lib/webhook.py
@@ -134,16 +134,12 @@ def cli(script, job, handler, argv=None):
 def payload_wrapper(service, url, payload, chat_id=None, reply_markup=None):
 	"""Returns a list with the response of each message sent (several if the payload was chunked)."""
 	if not payload:
-		print(service + ": Nothing to send") # informational
 		return None
 	payload = [str(line) for line in payload]
 	payload_string = '\n'.join(payload)
-	print("Preparing outbound to", service, str(len(payload_string)), "bytes")
-	print("Payload: " + payload_string) if config.prefs.debug else None
 	limit = DISCORD_LIMIT if service == 'discord' else 4096
 	if len(payload_string) <= limit:
 		return [write(service, url, payload_string, chat_id, reply_markup)]
-	print(service, f"payload is over {limit:,} bytes. Splitting.")
 	chunks = chunked(payload, limit, config.prefs.chunk_maxlines)
 	results = []
 	for idx, chunk in enumerate(chunks):
@@ -250,7 +246,6 @@ def telegram_photo(chat_id, image_data, caption, reply_markup=None):
 	except ValueError:
 		output = {}
 	if r.status_code == 200 and output.get('ok'):
-		print(r.status_code, "OK telegram sendPhoto", data['caption'])
 		return output
 	print(r.status_code, "error telegram sendPhoto", output.get('error_code'), output.get('description') or r.text[:300], file=sys.stderr)
 	return None
diff --git a/lib/yahoo.py b/lib/yahoo.py
index d9f7875a..515e6af0 100644
--- a/lib/yahoo.py
+++ b/lib/yahoo.py
@@ -235,15 +235,11 @@ def fetch_detail(ticker, seconds=None):
 	now = datetime.datetime.now()
 	cacheFile = "finbot_yahoo_detail_" + ticker + '.json'
 	data = cache.read(cacheFile, seconds) if config.prefs.cache else None
-	if data:
-		print('.', sep=' ', end='', flush=True)
-	else:
-		print('↓', sep=' ', end='', flush=True)
+	if not data:
 		cookie, crumb = credentials() # only on a miss: a cached detail needs neither, nor the lock
 		modules = 'calendarEvents,defaultKeyStatistics,balanceSheetHistoryQuarterly,financialData,summaryProfile,summaryDetail,price,earnings,earningsTrend,insiderTransactions'
 		r = _get_first(_hosts('/v10/finance/quoteSummary/' + ticker + '?modules=' + modules + '&crumb=' + crumb), {**HEADERS, 'Cookie': cookie})
 		if r is None:
-			print(ticker + '†', sep=' ', end='', flush=True)
 			return {} # catches some delisted stocks like "DRNA"
 		data = r.json()
 		if config.prefs.cache:
@@ -465,7 +461,6 @@ def fetch_chart_json(ticker, days=3665, seconds=None, full=False):
 		raise RuntimeError(f"Yahoo unreachable for {ticker}: {net.redact(e)}") from e
 	if r.status_code != 200:
 		raise RuntimeError(f"{r.status_code} error for {ticker} from Yahoo's chart API")
-	print('↓', sep=' ', end='', flush=True)
 	data = r.json()
 	if config.prefs.cache:
 		cache.write(cacheFile, data)
diff --git a/milestone.py b/milestone.py
index 2e38fcf7..bf496a57 100755
--- a/milestone.py
+++ b/milestone.py
@@ -2,7 +2,6 @@
 
 import sys
 from itertools import groupby
-from lib import config
 from lib import state
 from lib import webhook
 from lib import util
@@ -36,10 +35,8 @@ def prepare_milestone_payload(service, market_data):
 			emoji = util.flag_from_ticker(ticker)
 			changed[ticker] = {'record_high': max(oldhigh, newhigh), 'record_low': min(oldlow, newlow)} # all-time: a 52-week high below it is no record
 			if newhigh > oldhigh:
-				print("DEBUG", ticker, newhigh, "is higher than", oldhigh) if config.prefs.debug else None
 				payloadhigh.append(f"{emoji} {title} ({ticker_link})")
 			if newlow < oldlow:
-				print("DEBUG", ticker, newlow, "is lower than", oldlow) if config.prefs.debug else None
 				payloadlow.append(f"{emoji} {title} ({ticker_link})")
 		payloadlow.sort()
 		payloadhigh.sort()
diff --git a/podcasts.py b/podcasts.py
index 1eb93b2f..0239ef4c 100755
--- a/podcasts.py
+++ b/podcasts.py
@@ -94,7 +94,6 @@ def lambda_handler(feed_url):
 	except Exception as e:
 		raise RuntimeError(f"fetching feed {feed_url}: {e}")
 	if not guids:
-		print("no new episodes found")
 		return
 	for service, url, chat in webhook.default_chats():
 		payload = []
diff --git a/price.py b/price.py
index 9e6bdb92..90d29cf2 100755
--- a/price.py
+++ b/price.py
@@ -76,11 +76,9 @@ def report(service, threshold=None, user='', specific_stock=None, interactive=Fa
 		if midsession and state != 'REGULAR':
 			continue # not in session (states: PREPRE PRE REGULAR POST POSTPOST CLOSED)
 		if interday and not interactive and now - quoted > datetime.timedelta(days=1):
-			print(ticker + '⏭', sep='', end=' ', flush=True) # a market holiday: no repeat of the last session's moves
-			continue
+			continue # a market holiday: no repeat of the last session's moves
 		if close and not interactive and (now - quoted > datetime.timedelta(hours=1.5) or state in ('PREPRE', 'PRE', 'REGULAR')):
-			print(ticker + '⏭', sep='', end=' ', flush=True) # its market did not just close
-			continue
+			continue # its market did not just close
 		if premarket:
 			percent = quote.get('percent_change_premarket', quote.get('percent_change_postmarket'))
 			if percent is None:
diff --git a/shorts.py b/shorts.py
index af829caa..258a9d69 100755
--- a/shorts.py
+++ b/shorts.py
@@ -41,7 +41,6 @@ def prepare_shorts_payload(service, market_data):
 
 	market_data = yahoo.fetch(tickers)
 	market_data |= yahoo.fetch_details((ticker for ticker in tickers if '.' not in ticker), strict=bool(specific_stock)) # US short interest; Short Man has the ASX's
-	print("")
 	market_data = shortman.fetch(market_data)
 
 	for service, url, chat in webhook.destinations(interactive, service, chat_id):
diff --git a/tests/test_finbot.py b/tests/test_finbot.py
index 76ae706d..1914a4d9 100644
--- a/tests/test_finbot.py
+++ b/tests/test_finbot.py
@@ -1,5 +1,5 @@
 """Plumbing: message chunking, timezones, the WSGI handler."""
-import contextlib, datetime, io, json, subprocess, tempfile, unittest, zoneinfo
+import ast, contextlib, datetime, io, json, subprocess, tempfile, unittest, zoneinfo
 from pathlib import Path
 from unittest import mock
 from tests.common import *
@@ -55,6 +55,19 @@ def post(message):
 		self.assertEqual(bot.git_version(), head[:7]) # matches what git says
 
 
+class Logging(unittest.TestCase):
+	def test_the_bot_and_its_reports_print_only_to_stderr(self):
+		"""systemd discards finbot's stdout (StandardOutput=null), so a print there is never read. The util/ tools, run by
+		hand, may use it."""
+		offenders = []
+		for path in sorted([*Path(ROOT, 'lib').glob('*.py'), *Path(ROOT).glob('*.py')]):
+			for node in ast.walk(ast.parse(path.read_text())):
+				if isinstance(node, ast.Call) and getattr(node.func, 'id', None) == 'print' and \
+						not any(keyword.arg == 'file' for keyword in node.keywords):
+					offenders.append(f'{path.name}:{node.lineno}')
+		self.assertEqual(offenders, [])
+
+
 class GitVersion(unittest.TestCase):
 	SHA = '0123456789abcdef0123456789abcdef01234567'
 
diff --git a/trades.py b/trades.py
index ef5798df..a57ff93d 100755
--- a/trades.py
+++ b/trades.py
@@ -26,7 +26,6 @@ def prepare_trade_payload(service, trades):
 			if trade['transaction_type'] not in {'BUY', 'SELL'}:
 				continue
 			if not interactive and int(trade['id']) in known_trades:
-				print("Skipping known trade_id:", trade['id'], trade['transaction_date'], trade['portfolio'], trade['transaction_type'], trade['symbol'])
 				continue
 			wanted.append(trade)
 		if not wanted: # the usual case on a scheduled run

From 5c13613523c290708c26615e8de82c4d65339563 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Wed, 14 Oct 2026 23:57:58 +0000
Subject: [PATCH 12/81] util.py split by subject: markets, formats, links and
 render

lib/util.py (807 lines) held four unrelated things; each moves unchanged into its own module:
- lib/markets.py: market tables, flags, currencies, exchange names, ticker translation;
- lib/formats.py: company names trimmed, units, dates and spans in words, signed percents, ordinals;
- lib/links.py: chat-markup links and Yahoo/Google Finance links;
- lib/render.py: every matplotlib chart and the render pool.
get_holdings_and_watchlist becomes watchlist.with_holdings(). util imported sharesight, which imported util back; that
cycle is gone. Callers and test patch targets follow; the profile's local `links` list is now `sites`. Output is unchanged.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md                  |   9 +-
 bot.py                     |   4 +-
 cal.py                     |  17 +-
 lib/charts.py              |  11 +-
 lib/formats.py             | 171 +++++++++++++++++
 lib/links.py               |  66 +++++++
 lib/markets.py             | 132 +++++++++++++
 lib/{util.py => render.py} | 378 +------------------------------------
 lib/reports.py             |  99 +++++-----
 lib/sharesight.py          |   7 +-
 lib/watchlist.py           |   9 +
 lib/worker.py              |  16 +-
 lib/yahoo.py               |  17 +-
 milestone.py               |  10 +-
 performance.py             |  16 +-
 podcasts.py                |   4 +-
 price.py                   |  26 +--
 rating.py                  |   8 +-
 reminder.py                |  11 +-
 shorts.py                  |  13 +-
 tests/common.py            |   6 +-
 tests/test_charts.py       |  66 +++----
 tests/test_config.py       |   2 +-
 tests/test_datadir.py      |   4 +-
 tests/test_edges.py        |   2 +-
 tests/test_providers.py    |   4 +-
 tests/test_reports.py      |  10 +-
 tests/test_util.py         |  60 +++---
 trades.py                  |  18 +-
 29 files changed, 615 insertions(+), 581 deletions(-)
 create mode 100644 lib/formats.py
 create mode 100644 lib/links.py
 create mode 100644 lib/markets.py
 rename lib/{util.py => render.py} (61%)

diff --git a/CLAUDE.md b/CLAUDE.md
index 7c175d07..c59183c3 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -31,8 +31,9 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - `price.py performance.py trades.py shorts.py cal.py rating.py milestone.py reminder.py podcasts.py cleanup.py`: reports that also
   run on a schedule; each has `lambda_handler()` and `job(argv)` (its schedule-line argument parser).
 - `lib/webhook.py`: sending (chunking, photos, Slack upload), markup, escaping, error reporting. `lib/telegram.py`: Bot API calls.
-- `lib/charts.py`: Telegram period buttons and the rebuild behind them. `lib/util.py`: formatting, links, market tables, all
-  matplotlib rendering.
+- `lib/charts.py`: Telegram period buttons and the rebuild behind them. `lib/markets.py`: market tables, flags, ticker
+  translation. `lib/formats.py`: names, units, dates, percents in words. `lib/links.py`: links. `lib/render.py`: every chart
+  (matplotlib).
 - `lib/yahoo.py lib/sharesight.py lib/shortman.py lib/simplywallst.py`: data sources, cached in `DATA_DIR/cache`.
   `sharesight.portfolios(name=None)` gives `Portfolio` objects: `holdings()`, `tickers()`, `performance(days)`,
   `trades(days)`.
@@ -51,11 +52,11 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - Read settings as `config.env.<name>` / `config.prefs.<name>` where they are used, never into a default or module constant (a
   reload must be seen); in tests, `self.env(...)` / `self.prefs(...)` change them.
 - A report is a list of lines. Markup through `webhook.bold/italic(text, service)`, user-written text through
-  `webhook.escape(text, service)`, links through `util.link`/`util.finance_link`.
+  `webhook.escape(text, service)`, links through `links.link`/`links.finance_link`.
 - Tabs everywhere, tests included (`.editorconfig`). New and changed lines stay under 120 characters (an edit quotes whole lines).
 - Log with `print(..., file=sys.stderr)`. systemd discards stdout (`StandardOutput=null`), so the bot and its reports never
   print there (a test checks); only the util/ tools, run by hand, do.
-- A chart renderer in `lib/util.py` gets `@_drawn` (the bot draws in one native thread); it must not call another `@_drawn`
+- A chart renderer in `lib/render.py` gets `@_drawn` (the bot draws in one native thread); it must not call another `@_drawn`
   function, or that thread waits on itself.
 - A new command: `COMMANDS`, `prepare_help`, the README command list, and `command_keyboard` if it gets a button.
 
diff --git a/bot.py b/bot.py
index dceb2f2a..82ec25c6 100755
--- a/bot.py
+++ b/bot.py
@@ -21,7 +21,7 @@
 from lib import telegram
 from lib import webhook
 from lib import yahoo
-from lib import util
+from lib import render
 import cleanup
 
 botName = None # the Telegram bot's @username, from telegram.register() at start
@@ -251,7 +251,7 @@ def flush_cache(*args):
 			sys.exit(1)
 	else:
 		print('Scheduled reports are off (scheduler=False)', file=sys.stderr)
-	util.render_off_hub() # charts drawn in a thread of their own: no chat waits while another's chart is drawn
+	render.render_off_hub() # charts drawn in a thread of their own: no chat waits while another's chart is drawn
 	httpd = pywsgi.WSGIServer((config.env.ip, config.env.port), main)
 	httpd.secure_repr = False if config.prefs.debug else None
 	print(f'Opening socket on http://{config.env.ip}:{config.env.port}', file=sys.stderr)
diff --git a/cal.py b/cal.py
index d737e996..256d7444 100755
--- a/cal.py
+++ b/cal.py
@@ -5,7 +5,10 @@
 
 from lib import config
 from lib import telegram
-from lib import util
+from lib import formats
+from lib import links
+from lib import markets
+from lib import watchlist
 from lib import webhook
 from lib import yahoo
 
@@ -37,15 +40,15 @@ def prepare_payload(service, market_data):
 					ticker_primary = ticker.replace('.AX', '.NZ') # fixes GTK.AX and SKO.AX
 				market_data = market_data | yahoo.fetch_detail(ticker_primary)
 				country = market_data.get(ticker_primary, {}).get('profile_country') # only in the details
-				if country in util.YAHOO_COUNTRY and '.' not in ticker_primary:
-					ticker_primary = ticker_primary + '.' + util.YAHOO_COUNTRY[country] # fixes ASML and INFY
+				if country in markets.YAHOO_COUNTRY and '.' not in ticker_primary:
+					ticker_primary = ticker_primary + '.' + markets.YAHOO_COUNTRY[country] # fixes ASML and INFY
 					market_data = market_data | yahoo.fetch_detail(ticker_primary)
 				primary = market_data.get(ticker_primary, {})
 				if primary.get('profile_title') != title_orig or field not in primary: # not the same company, or no date there either
 					continue
 			when = datetime.datetime.fromtimestamp(market_data[ticker_primary][field])
 			if now < when < soon or specific_stock:
-				found.append((when, util.flag_from_ticker(ticker_primary), market_data[ticker]['profile_title'], util.yahoo_link(ticker, service)))
+				found.append((when, markets.flag_from_ticker(ticker_primary), market_data[ticker]['profile_title'], links.yahoo_link(ticker, service)))
 		found.sort(key=lambda item: item[0])
 		payload = []
 		for date in sorted({when.date() for when, *_ in found}):
@@ -55,9 +58,9 @@ def prepare_payload(service, market_data):
 				if when.date() == date:
 					payload.append(f"{flag} {date:%b %d} {title} ({link})" if specific_stock else f"{flag} {title} ({link})")
 		if payload and not specific_stock:
-			payload.insert(0, webhook.bold(f"{heading} {util.days_english(days, 'for the next ')}:", service))
+			payload.insert(0, webhook.bold(f"{heading} {formats.days_english(days, 'for the next ')}:", service))
 		elif not payload and interactive:
-			payload = [f"No events found for {tickers[0]}" if specific_stock else f"No events found {util.days_english(days, 'for the next ')}"]
+			payload = [f"No events found for {tickers[0]}" if specific_stock else f"No events found {formats.days_english(days, 'for the next ')}"]
 		return payload
 
 	def pin_if_earnings(service, results):
@@ -74,7 +77,7 @@ def pin_if_earnings(service, results):
 			telegram.pinChatMessage(chat, message_id)
 
 	# MAIN #
-	tickers = [util.transform_to_yahoo(specific_stock)] if specific_stock else util.get_holdings_and_watchlist()
+	tickers = [markets.transform_to_yahoo(specific_stock)] if specific_stock else watchlist.with_holdings()
 	if event == 'earnings':
 		market_data = yahoo.fetch(tickers)
 	else: # ex-dividend dates are in Yahoo's details
diff --git a/lib/charts.py b/lib/charts.py
index 667741da..d6d782be 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -3,7 +3,8 @@
 import json
 import sys
 
-from lib import util
+from lib import formats
+from lib import links
 from lib import state
 from lib import webhook
 from lib import yahoo
@@ -94,7 +95,7 @@ def history_payload(ticker, service, highlight=None, graph=False):
 	the full history's chart with graph=True (if the graph preference is on), else None."""
 	market_data = yahoo.fetch_detail(ticker, 600)
 	title = market_data.get(ticker, {}).get('profile_title', '')
-	ticker_link = util.finance_link(ticker, market_data.get(ticker, {}).get('profile_exchange', ''), service, days=1825, brief=False)
+	ticker_link = links.finance_link(ticker, market_data.get(ticker, {}).get('profile_exchange', ''), service, days=1825, brief=False)
 	if ticker not in market_data or 'percent_change' not in market_data[ticker]:
 		return [f".history: no data found for ticker {ticker}"], None, None
 	table, image = yahoo.price_history(ticker) if graph else yahoo.price_history(ticker, graph=False)
@@ -105,7 +106,7 @@ def history_payload(ticker, service, highlight=None, graph=False):
 			label = webhook.bold(interval + ':', service)
 			if interval == highlight:
 				label = '● ' + label
-			lines.append(f"{util.get_emoji(percent)} {label} {util.signed_percent(percent)}")
+			lines.append(f"{formats.get_emoji(percent)} {label} {formats.signed_percent(percent)}")
 	return lines, table, image
 
 def build(kind, tickers, period, service='telegram'):
@@ -126,8 +127,8 @@ def build(kind, tickers, period, service='telegram'):
 			raise RuntimeError(f"no data found for {ticker}")
 		percent, image = yahoo.price_history(ticker, days, graphCache=False)
 		percent = float(percent.get(days, percent.get('Max', next(iter(percent.values()), 0))))
-		link = util.quote_link(ticker, market_data[ticker]['profile_exchange'], service, days=days)
-		return f"{util.get_emoji(percent)} {market_data[ticker]['profile_title']} ({link}) {util.signed_percent(round(percent))}", image
+		link = links.quote_link(ticker, market_data[ticker]['profile_exchange'], service, days=days)
+		return f"{formats.get_emoji(percent)} {market_data[ticker]['profile_title']} ({link}) {formats.signed_percent(round(percent))}", image
 	if kind == 'c':
 		args = list(tickers) + ([f'{days}d'] if days else ['max'])
 		caption, image, *_ = reports.prepare_compare(service, args)
diff --git a/lib/formats.py b/lib/formats.py
new file mode 100644
index 00000000..fa42a6ea
--- /dev/null
+++ b/lib/formats.py
@@ -0,0 +1,171 @@
+"""Text for people: company names trimmed, numbers in units, dates and spans in words, signed percents, ordinals."""
+import datetime
+import re
+
+TITLE_TRIMS = [ # transform_title: (text, replacement), in order, as each sees what the ones before it made
+	(' - ', ' '), ('Roundhill ETF Trust ', ''), ('Class A Common Stock', ''),
+	('BlackRock Institutional Trust Company N.A.', ''), ('Listed Funds Trust', ''), ('GraniteShares ETF Trust', ''),
+	('ProShares Trust', ''), ('VanEck ETF Trust', ''), ('Direxion Shares ETF Trust', ''), (' of California', ''),
+	('Magnificent', 'Mag'), ('First Trust NASDAQ Clean Edge Green Energy Index Fund', 'First Trust Clean Energy'),
+	('First Trust Exchange-Traded Fund III First Trust Nasdaq Clean Edge Clean Energy Index Fund', 'First Trust Green Energy'),
+	('Mirae Asset Global Investments (Hong Kong) Limited', ''), ('Space Exploration Technologies', 'SpaceX'),
+	('Exchange Traded Funds Series OFC', ''), ('Global X Global X', 'Global X'), ('Tema ETF Trust', ''), (' AUD', ' '),
+	(' USD', ' '), ('New York Shares', ''), ('China Electric Vehicle and Battery ETF', 'China EV ETF'),
+	('China Electric Vehicle ETF', 'China EV ETF'), ('Atlantica Sustainable Infrastructure', 'Atlantica Sustainable'),
+	('Advanced Micro Devices', 'AMD'), ('Taiwan Semiconductor Manufacturing', 'TSM'),
+	('Flight Centre Travel', 'Flight Centre'), ('Global X Funds ', ''), (' of Australia', ''),
+	('National Australia Bank', 'NAB'), (' PLC', ''), (' p.l.c.', ''), (' P.l.c.', ''), (' P.L.C.', ''), (' Units', '')
+]
+
+TITLE_SHORTENINGS = [ # and these after the suffixes come off
+	('Microbalifesciences', 'Microba Life Sciences'), ('Walt Disney Co (The)', 'Disney'), ('Lisenergylimited', 'LI-S Energy'),
+	('Invesco Capital Management LLC ', ''), ('QUALCOMM', 'Qualcomm'), ('Ordinary Shares Class A', ''),
+	('Ordinary Shares Class C', ''), ('Lbt Innovations', 'LBT Innovations'),
+	('Vanguard Information Technology', 'Vanguard Infotech'), ('The ', ''), (' Shares', ''), ('Rea ', 'REA '),
+	('Csl ', 'CSL '), ('Battery ', 'Batt '), ('.com', ''), ('Etf', 'ETF'), ('N.V.', ''), (' NV ', ' '), ('New York Re', ''),
+	(' Australian', ' Aus'), (' Australia', ' Aus'), (' Infrastructure', ' Infra'), (' Manufacturing Company', ' '),
+	(' Limited', ' '), (' Ltd', ' '), (' Holdings', ' '), (' Holding', ' '), (' (Holdings)', ''), (' Corporation', ' '),
+	(' Incorporated', ' '), (' incorporated', ' '), (' Technologies', ' '), (' Technology', ' '), (' Enterprises', ' '),
+	(' Enterprise', ' '), (' Enterpr', ' '), (' Ventures', ' '), (' Resources', ' '), (' (Intrada', ''), (' Structured', ''),
+	(' Co.', ' '), (' Corp.', ' '), (' Corp', ' '), (' Tech ', ' '), (' Company', ' '), (' Group', ' '), (', Inc', ' '),
+	(' Inc', ' '), (' Plc', ' '), (' plc', ' '), (' Index', ' '), (' ADR', ' '), (' Daily', ' '), (' Long', ' '),
+	(' Bull', ' '), (' .', ' '), (' ,', ' ')
+]
+
+def transform_title(title):
+	"""A company or fund name as the reports show it: shorter, without fund-trust boilerplate and legal suffixes."""
+	title = title.replace(' FPO', '')
+	if title.isupper() or title.islower():
+		title = title.title()
+	for text, replacement in TITLE_TRIMS:
+		title = title.replace(text, replacement)
+	for suffix in (' AG', ' SE', ' Se'):
+		title = title.removesuffix(suffix)
+	title = re.sub(r'\[\w+\]$', '', title)
+	for text, replacement in TITLE_SHORTENINGS:
+		title = title.replace(text, replacement)
+	if title.islower():
+		title = title.title()
+	for suffix in (' &', ' and'):
+		title = title.strip().removesuffix(suffix)
+	return ' '.join(title.replace(' Usd', ' USD').replace('Ishares', 'iShares').split())
+
+def humanUnits(value, decimal_places=2):
+	"""1234567 -> '1.23 M'."""
+	for unit in ['', 'K', 'M', 'B', 'T', 'Q']:
+		if value < 1000.0 or unit == 'Q':
+			break
+		value /= 1000.0
+	return f"{value:.{decimal_places}f} {unit}".rstrip()
+
+def security_line(flag, title, link):
+	"""'🇦🇺 Title (link)': how a report line names a security (without a leading space where its market has no flag)."""
+	return f"{flag} {title} ({link})" if flag else f"{title} ({link})"
+
+def strip_url(url):
+	url = url.removeprefix('https://www.')
+	url = url.removeprefix('http://www.')
+	url = url.removeprefix('https://')
+	url = url.removeprefix('http://')
+	return url
+
+def make_paragraphs(walloftext):
+	"""The text in paragraphs, each ending with the first sentence that takes it past 22 words."""
+	paragraphs, buffer = [], []
+	for word in walloftext.split():
+		buffer.append(word)
+		if word.endswith(('!', '.')) and len(buffer) > 22:
+			paragraphs.append(' '.join(buffer))
+			buffer = []
+	if buffer: # the rest: it used to be dropped, and a short text with it
+		paragraphs.append(' '.join(buffer))
+	return '\n\n'.join(paragraphs)
+
+def date_short(d, year=False):
+	"""5 Oct (or 5 Oct 2026)."""
+	return f"{d.day} {d:%b}" + (f" {d.year}" if year else "")
+
+def date_range_english(days, end):
+	"""'28 Sep - 5 Oct 2026' for the `days` ending on `end`; both ends carry their year when the range crosses a year boundary."""
+	start = end - datetime.timedelta(days=days)
+	return f"{date_short(start, start.year != end.year)} - {date_short(end, True)}"
+
+def days_english(days, prefix='the past ', article=''):
+	if days is None or days == 0:
+		return 'today'
+	elif days == 1:
+		return prefix + article + 'day'
+	elif days == 7:
+		return prefix + article + 'week'
+	elif days == 30:
+		return prefix + article + 'month'
+	elif days == 365:
+		return prefix + article + 'year'
+	elif days % 7 == 0:
+		return prefix + str(int(days/7)) + ' weeks'
+	elif days % 30 == 0:
+		return prefix + str(int(days/30)) + ' months'
+	elif days % 365 == 0:
+		return prefix + str(int(days/365)) + ' years'
+	else:
+		return prefix + str(days) + ' days'
+
+def signed_percent(value, decimals=None):
+	"""12 -> '+12%', -3.5 -> '-3.5%', 0 -> '0%'; thousands separated. decimals=None keeps the value's own digits."""
+	if value == 0:
+		return '0%'
+	return f"{value:+,}%" if decimals is None else f"{value:+,.{decimals}f}%"
+
+def get_emoji(number):
+	if number > 0:
+		return '🔼'
+	elif number < 0:
+		return '🔻'
+	else:
+		return '▪️'
+
+def days_from_human_days(arg):
+	"""7, 7d, 2w, 3m, 1y or ytd -> days. Raises ValueError for anything else."""
+	arg = arg.upper()
+	if arg == 'YTD':
+		today = datetime.datetime.now().date()
+		return (today - today.replace(day=1, month=1)).days
+	for suffix, days in (('D', 1), ('W', 7), ('M', 30)):
+		try:
+			return int(arg.removesuffix(suffix)) * days
+		except ValueError:
+			pass
+	return int(arg.removesuffix('Y')) * 365
+
+def ordinal(num):
+	value = str(num)
+	if len(value) > 1:
+		secondToLastDigit = value[-2]
+		if secondToLastDigit == '1':
+			return 'th'
+	lastDigit = value[-1]
+	if (lastDigit == '1'):
+		return 'st'
+	elif (lastDigit == '2'):
+		return 'nd'
+	elif (lastDigit == '3'):
+		return 'rd'
+	else:
+		return 'th'
+
+def td_to_human(timedelta):
+	days, remainder = divmod(timedelta.total_seconds(), 86400)
+	hours, remainder = divmod(remainder, 3600)
+	minutes, seconds = divmod(remainder, 60)
+	parts = []
+	if days:
+		parts.append(f"{int(days)} day{'s' if days != 1 else ''}")
+	if hours:
+		parts.append(f"{int(hours)} hour{'s' if hours != 1 else ''}")
+	if minutes:
+		parts.append(f"{int(minutes)} minute{'s' if minutes != 1 else ''}")
+	if seconds:
+		parts.append(f"{int(seconds)} second{'s' if seconds != 1 else ''}")
+	if not parts:
+		return "0 seconds"
+	return ", ".join(parts)
diff --git a/lib/links.py b/lib/links.py
new file mode 100644
index 00000000..66881bac
--- /dev/null
+++ b/lib/links.py
@@ -0,0 +1,66 @@
+"""Links: chat markup for a link, and a stock's page on Yahoo or Google Finance."""
+from lib import config
+from lib import markets
+
+def yahoo_link(ticker, service='telegram', brief=True, text=None):
+	if not text:
+		text = ticker.split('.')[0] if brief else ticker
+	return link("https://au.finance.yahoo.com/quote/" + ticker, text, service)
+
+def link(url, text, service='telegram'):
+	"""text linking to url in the service's markup: HTML on Telegram, <url|text> on Slack, a masked link on Discord (the angle
+	brackets keep its preview away; brackets in the text and parentheses in the url would end it early)."""
+	if not config.prefs.hyperlink:
+		return text
+	if service == 'telegram':
+		return '<a href="' + url + '">' + text + '</a>'
+	if service == 'slack':
+		return '<' + url + '|' + text + '>'
+	if service == 'discord':
+		return '[' + text.replace('[', '\\[').replace(']', '\\]') + '](<' + url.replace('(', '%28').replace(')', '%29') + '>)'
+	return text
+
+def finance_link(symbol, exchange, service='telegram', days=1, brief=True, text=None):
+	if config.prefs.hyperlinkProvider == 'google':
+		return gfinance_link(symbol, exchange, service, days, brief, text)
+	return yahoo_link(symbol, service, brief, text)
+
+def quote_link(ticker, exchange, service='telegram', days=1):
+	"""finance_link, but to Yahoo for what Google Finance has no page for: Taipei Exchange, crypto (CCC) and indices (^)."""
+	if exchange in ('Taipei Exchange', 'CCC') or ticker.startswith('^'):
+		return yahoo_link(ticker, service)
+	return finance_link(ticker, exchange, service, days=days)
+
+def gfinance_link(symbol, exchange, service='telegram', days=1, brief=True, text=None):
+	window = '1D'
+	if not days:
+		days = 1
+	if days > 1:
+		window = '5D'
+	if days > 7:
+		window = '1M'
+	if days > 31:
+		window = '6M'
+	if days > 183:
+		window = '1Y'
+	if days > 365:
+		window = '5Y'
+	if days > 1825:
+		window = 'Max'
+	url = "https://www.google.com/finance/quote/"
+	if '.' in exchange:
+		exchange = exchange.split('.')[1]
+	exchange = markets.transform_to_google(exchange)
+	symbol_short = symbol.replace(':', '.').split('.')[0]
+	symbol_short = symbol_short.replace('-', '.') # class shares e.g. BRK.A
+	ticker = symbol_short + ':' + exchange
+	if brief and not text:
+		text = symbol_short
+	elif not text:
+		if '.' in symbol:
+			text = ticker
+		else:
+			text = symbol # US
+	if service in {'telegram', 'discord', 'slack'} and config.prefs.hyperlink:
+		return link(url + ticker + '?window=' + window, text, service)
+	return symbol
diff --git a/lib/markets.py b/lib/markets.py
new file mode 100644
index 00000000..7b5855ff
--- /dev/null
+++ b/lib/markets.py
@@ -0,0 +1,132 @@
+"""Markets: flags and currencies by market, exchange names, and ticker translation between Sharesight, Google and Yahoo."""
+
+MARKETS = { # Sharesight market code -> (flag, currency); no currency where a market quotes in several (LSE and HKEX allow non-home currencies)
+	'ASX': ('🇦🇺', 'AUD'),
+	'BOM': ('🇮🇳', 'INR'), 'NSE': ('🇮🇳', 'INR'),
+	'BMV': ('🇲🇽', 'MXN'),
+	'BKK': ('🇹🇭', 'THB'),
+	'BVMF': ('🇧🇷', 'BRL'),
+	'SHE': ('🇨🇳', 'CNY'), 'SHA': ('🇨🇳', 'CNY'),
+	'CPSE': ('🇩🇰', 'DKK'),
+	**dict.fromkeys(('EURONEXT', 'AMS', 'ATH', 'BIT', 'BME', 'DUB', 'EBR', 'EPA', 'ETR', 'FWB', 'FRA', 'VIE'), ('🇪🇺', 'EUR')),
+	'HKG': ('🇭🇰', None),
+	'ICSE': ('🇮🇸', 'ISK'),
+	'JSE': ('🇿🇦', 'ZAR'),
+	'KRX': ('🇰🇷', 'KRW'), 'KOSDAQ': ('🇰🇷', 'KRW'),
+	'LSE': ('🇬🇧', 'GBP'), # better than trades.py using brokerage_currency_code, which is AUD on CMC
+	'MISX': ('🇷🇺', 'RUB'),
+	'OM': ('🇸🇪', 'SEK'), 'STO': ('🇸🇪', 'SEK'),
+	'SWX': ('🇨🇭', 'CHF'), 'VTX': ('🇨🇭', 'CHF'),
+	'TAI': ('🇹🇼', 'TWD'), 'TPE': ('🇹🇼', 'TWD'),
+	'TASE': ('🇮🇱', 'ILS'),
+	'NZX': ('🇳🇿', 'NZD'),
+	'OB': ('🇳🇴', 'NOK'),
+	'SGX': ('🇸🇬', 'SGD'),
+	'TSE': ('🇯🇵', 'JPY'),
+	'TSX': ('🇨🇦', 'CAD'),
+	'NASDAQ': ('🇺🇸', 'USD'), 'NYSE': ('🇺🇸', 'USD'), 'BATS': ('🇺🇸', 'USD'), 'AMEX': ('🇺🇸', None),
+	'WAR': ('🇵🇱', 'PLN'),
+}
+
+YAHOO_COUNTRY = { # country -> its Yahoo ticker suffix, for a listing that has none (cal.py)
+	'Argentina': 'BA',
+	'Austria': 'VI',
+	'Australia': 'AX',
+	'Belgium': 'BR',
+	'Brazil': 'SA',
+	'Canada': 'TO',
+	'Chile': 'SN',
+	'Colombia': 'CL',
+	'Czech Republic': 'PR',
+	'Denmark': 'CO',
+	'Estonia': 'TL',
+	'Finland': 'HE',
+	'France': 'PA',
+	'Germany': 'DE',
+	'Hungary': 'BD',
+	'Iceland': 'IC',
+	'India': 'NS',
+	'Indonesia': 'JK',
+	'Ireland': 'IR',
+	'Israel': 'TA',
+	'Italy': 'MI',
+	'Latvia': 'RG',
+	'Lithuania': 'VS',
+	'Mexico': 'MX',
+	'Netherlands': 'AS',
+	'New Zealand': 'NZ',
+	'Norway': 'OL',
+	'Portugal': 'LS',
+	'Qatar': 'QA',
+	'Saudi Arabia': 'SR',
+	'South Africa': 'JO',
+	'Spain': 'MC',
+	'Sweden': 'ST',
+	'Switzerland': 'SW',
+	'Thailand': 'BK',
+	'Turkey': 'IS',
+	'United Kingdom': 'L',
+	'Venezuela': 'CR'
+	# markets with numeric tickers purposefully omitted
+}
+
+def flag_from_market(market):
+	if market in MARKETS:
+		return MARKETS[market][0]
+	return '🇺🇸' if market.lower().startswith(('nasdaq', 'nyse')) else ''
+
+TICKER_FLAGS = {'AS': '🇳🇱', 'AX': '🇦🇺', 'HK': '🇭🇰', 'KS': '🇰🇷', 'KQ': '🇰🇷', 'L': '🇬🇧', 'NS': '🇮🇳', 'NZ': '🇳🇿', 'SI': '🇸🇬', 'T': '🇯🇵', 'TA': '🇮🇱',
+	'TW': '🇹🇼', 'TWO': '🇹🇼', 'TO': '🇨🇦'} # Yahoo suffix -> flag
+
+def flag_from_ticker(ticker):
+	return TICKER_FLAGS.get(ticker.split('.')[1], '') if '.' in ticker else '🇺🇸'
+
+def currency_from_market(market):
+	return MARKETS.get(market, (None, None))[1]
+
+VENUES = { # the names Yahoo, Sharesight and Google give an exchange -> (Google Finance code, country)
+	**dict.fromkeys(('OTC', 'PNK', 'Other OTC', 'OTCPK'), ('OTCMKTS', None)),
+	**dict.fromkeys(('TO', 'TOR', 'Toronto'), ('TSE', 'CA')),
+	**dict.fromkeys(('TW', 'TWO', 'TAI', 'Taiwan', 'Taipei Exchange', 'Taipei'), ('TPE', 'TW')),
+	**dict.fromkeys(('HK', 'HKG', 'HKSE'), ('HKG', 'HK')),
+	**dict.fromkeys(('KQ', 'KOE', 'KOSDAQ'), ('KOSDAQ', 'KR')),
+	**dict.fromkeys(('KS', 'KRX', 'KSE', 'KSC'), ('KRX', 'KR')),
+	**dict.fromkeys(('L', 'LSE', 'London'), ('LON', 'UK')),
+	**dict.fromkeys(('T', 'TYO', 'JPX', 'Tokyo'), ('TYO', 'JP')),
+	**dict.fromkeys(('TA', 'TLV', 'Tel Aviv'), ('TLV', 'IL')),
+	**dict.fromkeys(('NZSE', 'NZE', 'New Zealand', 'Auckland'), (None, 'NZ')),
+}
+
+def transform_to_google(exchange):
+	if 'Nasdaq' in exchange:
+		return 'NASDAQ'
+	if exchange == 'Cboe US':
+		return 'BATS'
+	return VENUES.get(exchange, (None, None))[0] or exchange
+
+def exchange_human(exchange):
+	if 'Nasdaq' in exchange or 'NYSE' in exchange or 'Other OTC' in exchange:
+		return 'US'
+	return VENUES.get(exchange, (None, None))[1] or exchange
+
+YAHOO_SUFFIX = { # exchange code (Sharesight, Google) -> Yahoo's ticker suffix
+	'ASX': 'AX', 'HKG': 'HK', 'KRX': 'KS', 'KOSDAQ': 'KQ', 'LSE': 'L', 'NZX': 'NZ', 'SGX': 'SI',
+	'TAI': 'TW', # Taiwan in YF and Sharesight; TPE in GF
+	'TPE': 'TWO', # Taipei in YF, Taiwan in GF (GF format will not work as input); missing from GF
+	'TSE': 'T', 'TYO': 'T', # Tokyo: TSE in Sharesight (Google's TSE is Toronto, which Sharesight calls TSX), TYO in GF
+	'TSX': 'TO', 'TLV': 'TA',
+}
+
+def transform_to_yahoo(ticker, market=None):
+	split = ticker.replace(':', '.').split('.')
+	if len(split) > 1:
+		ticker = split[0]
+		market = split[1]
+	if not market:
+		return ticker
+	market = YAHOO_SUFFIX.get(market, market)
+	if market in {'NASDAQ', 'NYSE', 'BATS', 'OTCMKTS', 'AMEX', 'NYSEARCA'}: # US venues: Yahoo symbols carry no suffix
+		return ticker
+	if market in {'A', 'B', 'C'}: # class shares
+		return ticker + '-' + market
+	return ticker + '.' + market
diff --git a/lib/util.py b/lib/render.py
similarity index 61%
rename from lib/util.py
rename to lib/render.py
index a76fa71f..ce03aaf7 100644
--- a/lib/util.py
+++ b/lib/render.py
@@ -1,313 +1,8 @@
+"""Charts drawn as PNGs with matplotlib: the bot draws them in one native thread, off the gevent hub (render_off_hub)."""
 import functools
 import io
-import datetime
-import re
 from lib import config # first: sets MPLCONFIGDIR before matplotlib is imported (lazily, by the chart functions)
-from lib import sharesight
-from lib import watchlist
-
-
-TITLE_TRIMS = [ # transform_title: (text, replacement), in order, as each sees what the ones before it made
-	(' - ', ' '), ('Roundhill ETF Trust ', ''), ('Class A Common Stock', ''),
-	('BlackRock Institutional Trust Company N.A.', ''), ('Listed Funds Trust', ''), ('GraniteShares ETF Trust', ''),
-	('ProShares Trust', ''), ('VanEck ETF Trust', ''), ('Direxion Shares ETF Trust', ''), (' of California', ''),
-	('Magnificent', 'Mag'), ('First Trust NASDAQ Clean Edge Green Energy Index Fund', 'First Trust Clean Energy'),
-	('First Trust Exchange-Traded Fund III First Trust Nasdaq Clean Edge Clean Energy Index Fund', 'First Trust Green Energy'),
-	('Mirae Asset Global Investments (Hong Kong) Limited', ''), ('Space Exploration Technologies', 'SpaceX'),
-	('Exchange Traded Funds Series OFC', ''), ('Global X Global X', 'Global X'), ('Tema ETF Trust', ''), (' AUD', ' '),
-	(' USD', ' '), ('New York Shares', ''), ('China Electric Vehicle and Battery ETF', 'China EV ETF'),
-	('China Electric Vehicle ETF', 'China EV ETF'), ('Atlantica Sustainable Infrastructure', 'Atlantica Sustainable'),
-	('Advanced Micro Devices', 'AMD'), ('Taiwan Semiconductor Manufacturing', 'TSM'),
-	('Flight Centre Travel', 'Flight Centre'), ('Global X Funds ', ''), (' of Australia', ''),
-	('National Australia Bank', 'NAB'), (' PLC', ''), (' p.l.c.', ''), (' P.l.c.', ''), (' P.L.C.', ''), (' Units', '')
-]
-
-TITLE_SHORTENINGS = [ # and these after the suffixes come off
-	('Microbalifesciences', 'Microba Life Sciences'), ('Walt Disney Co (The)', 'Disney'), ('Lisenergylimited', 'LI-S Energy'),
-	('Invesco Capital Management LLC ', ''), ('QUALCOMM', 'Qualcomm'), ('Ordinary Shares Class A', ''),
-	('Ordinary Shares Class C', ''), ('Lbt Innovations', 'LBT Innovations'),
-	('Vanguard Information Technology', 'Vanguard Infotech'), ('The ', ''), (' Shares', ''), ('Rea ', 'REA '),
-	('Csl ', 'CSL '), ('Battery ', 'Batt '), ('.com', ''), ('Etf', 'ETF'), ('N.V.', ''), (' NV ', ' '), ('New York Re', ''),
-	(' Australian', ' Aus'), (' Australia', ' Aus'), (' Infrastructure', ' Infra'), (' Manufacturing Company', ' '),
-	(' Limited', ' '), (' Ltd', ' '), (' Holdings', ' '), (' Holding', ' '), (' (Holdings)', ''), (' Corporation', ' '),
-	(' Incorporated', ' '), (' incorporated', ' '), (' Technologies', ' '), (' Technology', ' '), (' Enterprises', ' '),
-	(' Enterprise', ' '), (' Enterpr', ' '), (' Ventures', ' '), (' Resources', ' '), (' (Intrada', ''), (' Structured', ''),
-	(' Co.', ' '), (' Corp.', ' '), (' Corp', ' '), (' Tech ', ' '), (' Company', ' '), (' Group', ' '), (', Inc', ' '),
-	(' Inc', ' '), (' Plc', ' '), (' plc', ' '), (' Index', ' '), (' ADR', ' '), (' Daily', ' '), (' Long', ' '),
-	(' Bull', ' '), (' .', ' '), (' ,', ' ')
-]
-
-def transform_title(title):
-	"""A company or fund name as the reports show it: shorter, without fund-trust boilerplate and legal suffixes."""
-	title = title.replace(' FPO', '')
-	if title.isupper() or title.islower():
-		title = title.title()
-	for text, replacement in TITLE_TRIMS:
-		title = title.replace(text, replacement)
-	for suffix in (' AG', ' SE', ' Se'):
-		title = title.removesuffix(suffix)
-	title = re.sub(r'\[\w+\]$', '', title)
-	for text, replacement in TITLE_SHORTENINGS:
-		title = title.replace(text, replacement)
-	if title.islower():
-		title = title.title()
-	for suffix in (' &', ' and'):
-		title = title.strip().removesuffix(suffix)
-	return ' '.join(title.replace(' Usd', ' USD').replace('Ishares', 'iShares').split())
-
-MARKETS = { # Sharesight market code -> (flag, currency); no currency where a market quotes in several (LSE and HKEX allow non-home currencies)
-	'ASX': ('🇦🇺', 'AUD'),
-	'BOM': ('🇮🇳', 'INR'), 'NSE': ('🇮🇳', 'INR'),
-	'BMV': ('🇲🇽', 'MXN'),
-	'BKK': ('🇹🇭', 'THB'),
-	'BVMF': ('🇧🇷', 'BRL'),
-	'SHE': ('🇨🇳', 'CNY'), 'SHA': ('🇨🇳', 'CNY'),
-	'CPSE': ('🇩🇰', 'DKK'),
-	**dict.fromkeys(('EURONEXT', 'AMS', 'ATH', 'BIT', 'BME', 'DUB', 'EBR', 'EPA', 'ETR', 'FWB', 'FRA', 'VIE'), ('🇪🇺', 'EUR')),
-	'HKG': ('🇭🇰', None),
-	'ICSE': ('🇮🇸', 'ISK'),
-	'JSE': ('🇿🇦', 'ZAR'),
-	'KRX': ('🇰🇷', 'KRW'), 'KOSDAQ': ('🇰🇷', 'KRW'),
-	'LSE': ('🇬🇧', 'GBP'), # better than trades.py using brokerage_currency_code, which is AUD on CMC
-	'MISX': ('🇷🇺', 'RUB'),
-	'OM': ('🇸🇪', 'SEK'), 'STO': ('🇸🇪', 'SEK'),
-	'SWX': ('🇨🇭', 'CHF'), 'VTX': ('🇨🇭', 'CHF'),
-	'TAI': ('🇹🇼', 'TWD'), 'TPE': ('🇹🇼', 'TWD'),
-	'TASE': ('🇮🇱', 'ILS'),
-	'NZX': ('🇳🇿', 'NZD'),
-	'OB': ('🇳🇴', 'NOK'),
-	'SGX': ('🇸🇬', 'SGD'),
-	'TSE': ('🇯🇵', 'JPY'),
-	'TSX': ('🇨🇦', 'CAD'),
-	'NASDAQ': ('🇺🇸', 'USD'), 'NYSE': ('🇺🇸', 'USD'), 'BATS': ('🇺🇸', 'USD'), 'AMEX': ('🇺🇸', None),
-	'WAR': ('🇵🇱', 'PLN'),
-}
-
-YAHOO_COUNTRY = { # country -> its Yahoo ticker suffix, for a listing that has none (cal.py)
-	'Argentina': 'BA',
-	'Austria': 'VI',
-	'Australia': 'AX',
-	'Belgium': 'BR',
-	'Brazil': 'SA',
-	'Canada': 'TO',
-	'Chile': 'SN',
-	'Colombia': 'CL',
-	'Czech Republic': 'PR',
-	'Denmark': 'CO',
-	'Estonia': 'TL',
-	'Finland': 'HE',
-	'France': 'PA',
-	'Germany': 'DE',
-	'Hungary': 'BD',
-	'Iceland': 'IC',
-	'India': 'NS',
-	'Indonesia': 'JK',
-	'Ireland': 'IR',
-	'Israel': 'TA',
-	'Italy': 'MI',
-	'Latvia': 'RG',
-	'Lithuania': 'VS',
-	'Mexico': 'MX',
-	'Netherlands': 'AS',
-	'New Zealand': 'NZ',
-	'Norway': 'OL',
-	'Portugal': 'LS',
-	'Qatar': 'QA',
-	'Saudi Arabia': 'SR',
-	'South Africa': 'JO',
-	'Spain': 'MC',
-	'Sweden': 'ST',
-	'Switzerland': 'SW',
-	'Thailand': 'BK',
-	'Turkey': 'IS',
-	'United Kingdom': 'L',
-	'Venezuela': 'CR'
-	# markets with numeric tickers purposefully omitted
-}
-
-def flag_from_market(market):
-	if market in MARKETS:
-		return MARKETS[market][0]
-	return '🇺🇸' if market.lower().startswith(('nasdaq', 'nyse')) else ''
-
-TICKER_FLAGS = {'AS': '🇳🇱', 'AX': '🇦🇺', 'HK': '🇭🇰', 'KS': '🇰🇷', 'KQ': '🇰🇷', 'L': '🇬🇧', 'NS': '🇮🇳', 'NZ': '🇳🇿', 'SI': '🇸🇬', 'T': '🇯🇵', 'TA': '🇮🇱',
-	'TW': '🇹🇼', 'TWO': '🇹🇼', 'TO': '🇨🇦'} # Yahoo suffix -> flag
-
-def flag_from_ticker(ticker):
-	return TICKER_FLAGS.get(ticker.split('.')[1], '') if '.' in ticker else '🇺🇸'
-
-def currency_from_market(market):
-	return MARKETS.get(market, (None, None))[1]
-
-def humanUnits(value, decimal_places=2):
-	"""1234567 -> '1.23 M'."""
-	for unit in ['', 'K', 'M', 'B', 'T', 'Q']:
-		if value < 1000.0 or unit == 'Q':
-			break
-		value /= 1000.0
-	return f"{value:.{decimal_places}f} {unit}".rstrip()
-
-def yahoo_link(ticker, service='telegram', brief=True, text=None):
-	if not text:
-		text = ticker.split('.')[0] if brief else ticker
-	return link("https://au.finance.yahoo.com/quote/" + ticker, text, service)
-
-def security_line(flag, title, link):
-	"""'🇦🇺 Title (link)': how a report line names a security (without a leading space where its market has no flag)."""
-	return f"{flag} {title} ({link})" if flag else f"{title} ({link})"
-
-def link(url, text, service='telegram'):
-	"""text linking to url in the service's markup: HTML on Telegram, <url|text> on Slack, a masked link on Discord (the angle
-	brackets keep its preview away; brackets in the text and parentheses in the url would end it early)."""
-	if not config.prefs.hyperlink:
-		return text
-	if service == 'telegram':
-		return '<a href="' + url + '">' + text + '</a>'
-	if service == 'slack':
-		return '<' + url + '|' + text + '>'
-	if service == 'discord':
-		return '[' + text.replace('[', '\\[').replace(']', '\\]') + '](<' + url.replace('(', '%28').replace(')', '%29') + '>)'
-	return text
-
-def finance_link(symbol, exchange, service='telegram', days=1, brief=True, text=None):
-	if config.prefs.hyperlinkProvider == 'google':
-		return gfinance_link(symbol, exchange, service, days, brief, text)
-	return yahoo_link(symbol, service, brief, text)
-
-def quote_link(ticker, exchange, service='telegram', days=1):
-	"""finance_link, but to Yahoo for what Google Finance has no page for: Taipei Exchange, crypto (CCC) and indices (^)."""
-	if exchange in ('Taipei Exchange', 'CCC') or ticker.startswith('^'):
-		return yahoo_link(ticker, service)
-	return finance_link(ticker, exchange, service, days=days)
-
-def gfinance_link(symbol, exchange, service='telegram', days=1, brief=True, text=None):
-	window = '1D'
-	if not days:
-		days = 1
-	if days > 1:
-		window = '5D'
-	if days > 7:
-		window = '1M'
-	if days > 31:
-		window = '6M'
-	if days > 183:
-		window = '1Y'
-	if days > 365:
-		window = '5Y'
-	if days > 1825:
-		window = 'Max'
-	url = "https://www.google.com/finance/quote/"
-	if '.' in exchange:
-		exchange = exchange.split('.')[1]
-	exchange = transform_to_google(exchange)
-	symbol_short = symbol.replace(':', '.').split('.')[0]
-	symbol_short = symbol_short.replace('-', '.') # class shares e.g. BRK.A
-	ticker = symbol_short + ':' + exchange
-	if brief and not text:
-		text = symbol_short
-	elif not text:
-		if '.' in symbol:
-			text = ticker
-		else:
-			text = symbol # US
-	if service in {'telegram', 'discord', 'slack'} and config.prefs.hyperlink:
-		return link(url + ticker + '?window=' + window, text, service)
-	return symbol
-
-VENUES = { # the names Yahoo, Sharesight and Google give an exchange -> (Google Finance code, country)
-	**dict.fromkeys(('OTC', 'PNK', 'Other OTC', 'OTCPK'), ('OTCMKTS', None)),
-	**dict.fromkeys(('TO', 'TOR', 'Toronto'), ('TSE', 'CA')),
-	**dict.fromkeys(('TW', 'TWO', 'TAI', 'Taiwan', 'Taipei Exchange', 'Taipei'), ('TPE', 'TW')),
-	**dict.fromkeys(('HK', 'HKG', 'HKSE'), ('HKG', 'HK')),
-	**dict.fromkeys(('KQ', 'KOE', 'KOSDAQ'), ('KOSDAQ', 'KR')),
-	**dict.fromkeys(('KS', 'KRX', 'KSE', 'KSC'), ('KRX', 'KR')),
-	**dict.fromkeys(('L', 'LSE', 'London'), ('LON', 'UK')),
-	**dict.fromkeys(('T', 'TYO', 'JPX', 'Tokyo'), ('TYO', 'JP')),
-	**dict.fromkeys(('TA', 'TLV', 'Tel Aviv'), ('TLV', 'IL')),
-	**dict.fromkeys(('NZSE', 'NZE', 'New Zealand', 'Auckland'), (None, 'NZ')),
-}
-
-def transform_to_google(exchange):
-	if 'Nasdaq' in exchange:
-		return 'NASDAQ'
-	if exchange == 'Cboe US':
-		return 'BATS'
-	return VENUES.get(exchange, (None, None))[0] or exchange
-
-def exchange_human(exchange):
-	if 'Nasdaq' in exchange or 'NYSE' in exchange or 'Other OTC' in exchange:
-		return 'US'
-	return VENUES.get(exchange, (None, None))[1] or exchange
-
-YAHOO_SUFFIX = { # exchange code (Sharesight, Google) -> Yahoo's ticker suffix
-	'ASX': 'AX', 'HKG': 'HK', 'KRX': 'KS', 'KOSDAQ': 'KQ', 'LSE': 'L', 'NZX': 'NZ', 'SGX': 'SI',
-	'TAI': 'TW', # Taiwan in YF and Sharesight; TPE in GF
-	'TPE': 'TWO', # Taipei in YF, Taiwan in GF (GF format will not work as input); missing from GF
-	'TSE': 'T', 'TYO': 'T', # Tokyo: TSE in Sharesight (Google's TSE is Toronto, which Sharesight calls TSX), TYO in GF
-	'TSX': 'TO', 'TLV': 'TA',
-}
-
-def transform_to_yahoo(ticker, market=None):
-	split = ticker.replace(':', '.').split('.')
-	if len(split) > 1:
-		ticker = split[0]
-		market = split[1]
-	if not market:
-		return ticker
-	market = YAHOO_SUFFIX.get(market, market)
-	if market in {'NASDAQ', 'NYSE', 'BATS', 'OTCMKTS', 'AMEX', 'NYSEARCA'}: # US venues: Yahoo symbols carry no suffix
-		return ticker
-	if market in {'A', 'B', 'C'}: # class shares
-		return ticker + '-' + market
-	return ticker + '.' + market
-
-def strip_url(url):
-	url = url.removeprefix('https://www.')
-	url = url.removeprefix('http://www.')
-	url = url.removeprefix('https://')
-	url = url.removeprefix('http://')
-	return url
-
-def make_paragraphs(walloftext):
-	"""The text in paragraphs, each ending with the first sentence that takes it past 22 words."""
-	paragraphs, buffer = [], []
-	for word in walloftext.split():
-		buffer.append(word)
-		if word.endswith(('!', '.')) and len(buffer) > 22:
-			paragraphs.append(' '.join(buffer))
-			buffer = []
-	if buffer: # the rest: it used to be dropped, and a short text with it
-		paragraphs.append(' '.join(buffer))
-	return '\n\n'.join(paragraphs)
-
-def date_short(d, year=False):
-	"""5 Oct (or 5 Oct 2026)."""
-	return f"{d.day} {d:%b}" + (f" {d.year}" if year else "")
-
-def date_range_english(days, end):
-	"""'28 Sep - 5 Oct 2026' for the `days` ending on `end`; both ends carry their year when the range crosses a year boundary."""
-	start = end - datetime.timedelta(days=days)
-	return f"{date_short(start, start.year != end.year)} - {date_short(end, True)}"
-
-def days_english(days, prefix='the past ', article=''):
-	if days is None or days == 0:
-		return 'today'
-	elif days == 1:
-		return prefix + article + 'day'
-	elif days == 7:
-		return prefix + article + 'week'
-	elif days == 30:
-		return prefix + article + 'month'
-	elif days == 365:
-		return prefix + article + 'year'
-	elif days % 7 == 0:
-		return prefix + str(int(days/7)) + ' weeks'
-	elif days % 30 == 0:
-		return prefix + str(int(days/30)) + ' months'
-	elif days % 365 == 0:
-		return prefix + str(int(days/365)) + ' years'
-	else:
-		return prefix + str(days) + ' days'
+from lib import formats
 
 def fit_left_margin(fig, ax, pad_px=18):
 	"""Widen the left margin so the longest Y tick label is never clipped; the right edge stays put."""
@@ -497,7 +192,7 @@ def hbar_chart(rows, title, subtitle='', value_fmt=None, xlim=None, session=None
 	Blue for >= 0, red below. Each row's label sits at the bar's base and its value at the tip.
 	rows: [(label, value)]; a (label, None) row is a blank spacer. xlim zooms the value axis (bars then start at its left edge).
 	session: 'midsession' or 'premarket' washes the background green or yellow."""
-	value_fmt = value_fmt or (lambda v: signed_percent(round(v)))
+	value_fmt = value_fmt or (lambda v: formats.signed_percent(round(v)))
 	bg, ink, ink2, _ = PALETTE
 	up, down = '#3987e5', '#fb7185'
 	if session in SESSION_TINTS:
@@ -738,70 +433,3 @@ def pie_chart(rows, title, subtitle=''):
 def bar_graph(top_rows, bottom_rows, title, subtitle=''):
 	"""Top performers then, after a gap, the worst performers: the same order as the caption's two lists."""
 	return hbar_chart(list(top_rows) + [('', None)] + list(bottom_rows), title, subtitle)
-
-def signed_percent(value, decimals=None):
-	"""12 -> '+12%', -3.5 -> '-3.5%', 0 -> '0%'; thousands separated. decimals=None keeps the value's own digits."""
-	if value == 0:
-		return '0%'
-	return f"{value:+,}%" if decimals is None else f"{value:+,.{decimals}f}%"
-
-def get_emoji(number):
-	if number > 0:
-		return '🔼'
-	elif number < 0:
-		return '🔻'
-	else:
-		return '▪️'
-
-def days_from_human_days(arg):
-	"""7, 7d, 2w, 3m, 1y or ytd -> days. Raises ValueError for anything else."""
-	arg = arg.upper()
-	if arg == 'YTD':
-		today = datetime.datetime.now().date()
-		return (today - today.replace(day=1, month=1)).days
-	for suffix, days in (('D', 1), ('W', 7), ('M', 30)):
-		try:
-			return int(arg.removesuffix(suffix)) * days
-		except ValueError:
-			pass
-	return int(arg.removesuffix('Y')) * 365
-
-def get_holdings_and_watchlist():
-	tickers = set(sharesight.tickers())
-	tickers.update(watchlist.tickers())
-	if 'GOOG' in tickers and 'GOOGL' in tickers:
-		tickers.remove("GOOGL")
-	return sorted(tickers)
-
-def ordinal(num):
-	value = str(num)
-	if len(value) > 1:
-		secondToLastDigit = value[-2]
-		if secondToLastDigit == '1':
-			return 'th'
-	lastDigit = value[-1]
-	if (lastDigit == '1'):
-		return 'st'
-	elif (lastDigit == '2'):
-		return 'nd'
-	elif (lastDigit == '3'):
-		return 'rd'
-	else:
-		return 'th'
-
-def td_to_human(timedelta):
-	days, remainder = divmod(timedelta.total_seconds(), 86400)
-	hours, remainder = divmod(remainder, 3600)
-	minutes, seconds = divmod(remainder, 60)
-	parts = []
-	if days:
-		parts.append(f"{int(days)} day{'s' if days != 1 else ''}")
-	if hours:
-		parts.append(f"{int(hours)} hour{'s' if hours != 1 else ''}")
-	if minutes:
-		parts.append(f"{int(minutes)} minute{'s' if minutes != 1 else ''}")
-	if seconds:
-		parts.append(f"{int(seconds)} second{'s' if seconds != 1 else ''}")
-	if not parts:
-		return "0 seconds"
-	return ", ".join(parts)
diff --git a/lib/reports.py b/lib/reports.py
index a24004c2..5d4d3b71 100755
--- a/lib/reports.py
+++ b/lib/reports.py
@@ -6,7 +6,10 @@
 from lib import cache
 from lib import watchlist
 from lib import sharesight
-from lib import util
+from lib import formats
+from lib import links
+from lib import markets
+from lib import render
 from lib import webhook
 from lib import yahoo
 from lib import simplywallst
@@ -21,8 +24,8 @@ def last_col(line):
 
 def holding_line(holding, service, text=None):
 	"""'🇦🇺 Title (CODE)' for a Sharesight holding, CODE (or `text`) linking to the holding on Sharesight."""
-	link = util.link(f"https://portfolio.sharesight.com/holdings/{holding['holding_id']}/dashboard", holding['code'] if text is None else text, service)
-	return f"{util.flag_from_market(holding['market_code'])} {util.transform_title(holding['name'])} ({link})"
+	link = links.link(f"https://portfolio.sharesight.com/holdings/{holding['holding_id']}/dashboard", holding['code'] if text is None else text, service)
+	return f"{markets.flag_from_market(holding['market_code'])} {formats.transform_title(holding['name'])} ({link})"
 
 def doDelta(inputList):
 	deltaString = ''
@@ -59,10 +62,10 @@ def doDelta(inputList):
 def prepare_watchlist(service, user, action=None, ticker=None):
 	"""The .watchlist reply: the companies watched, or one line on adding ('add') or removing ('delete') a ticker."""
 	if action:
-		return [watchlist_change(service, user, action, util.transform_to_yahoo(ticker))]
+		return [watchlist_change(service, user, action, markets.transform_to_yahoo(ticker))]
 	watched = watchlist.tickers()
 	market_data = yahoo.fetch(watched) if watched else {}
-	lines = sorted((f"{util.flag_from_ticker(t)} {quote['profile_title']} ({util.finance_link(t, quote.get('profile_exchange'), service)})"
+	lines = sorted((f"{markets.flag_from_ticker(t)} {quote['profile_title']} ({links.finance_link(t, quote.get('profile_exchange'), service)})"
 		for t, quote in market_data.items()), key=lambda line: re.findall('[A-Z].*', line)) # by name, the flag aside
 	if not lines:
 		return ['Watchlist is empty. Try ".watchlist add SYMBOL" to create it']
@@ -72,7 +75,7 @@ def watchlist_change(service, user, action, asked):
 	"""The reply to adding or removing `asked` (a Yahoo ticker)."""
 	def described(ticker, quote): # "Full Name (TICKER)", bold
 		quote = quote or {}
-		link = util.finance_link(ticker, quote.get('profile_exchange') or ticker, service)
+		link = links.finance_link(ticker, quote.get('profile_exchange') or ticker, service)
 		return webhook.bold(f"{quote['profile_title']} ({link})" if quote.get('profile_title') else link, service)
 	if action == 'delete':
 		if asked not in watchlist.tickers():
@@ -103,11 +106,11 @@ def prepare_compare(service, args):
 			args.pop()
 		else:
 			try:
-				period_days = util.days_from_human_days(args[-1])
+				period_days = formats.days_from_human_days(args[-1])
 				args.pop()
 			except ValueError:
 				pass
-	tickers = list(dict.fromkeys(util.transform_to_yahoo(a.upper()) for a in args)) # de-duplicated, order kept
+	tickers = list(dict.fromkeys(markets.transform_to_yahoo(a.upper()) for a in args)) # de-duplicated, order kept
 	if not 2 <= len(tickers) <= 6:
 		return usage, None, tickers, period_days
 	spec = period_days or ('max' if full_history else 'default')
@@ -146,17 +149,17 @@ def prepare_compare(service, args):
 	if any(len(series) < 2 for _, series in cropped):
 		raise RuntimeError("not enough price history in that period to compare")
 	span_days = (datetime.datetime.now().date() - start.date()).days
-	period_label = util.days_english(period_days) if period_days and not note else f"{span_days} days"
+	period_label = formats.days_english(period_days) if period_days and not note else f"{span_days} days"
 	by_performance = sorted(cropped, key=lambda item: item[1].iloc[-1] / item[1].iloc[0], reverse=True)
 	title = ' vs '.join(t.split('.')[0] for t, _ in by_performance) # best first; line colours stay tied to input order
 	subtitle = f"% change over {period_label}, since {start:%d %b %Y}" + note
-	image = util.compare_graph([(t.split('.')[0], series) for t, series in cropped], title, subtitle)
+	image = render.compare_graph([(t.split('.')[0], series) for t, series in cropped], title, subtitle)
 	image.seek(0)
 	caption = [webhook.bold(f"{period_label}, since {start:%d %b %Y}", service) + note]
 	results = sorted(((series.iloc[-1] / series.iloc[0] - 1) * 100, t) for t, series in cropped)
 	for pct, t in reversed(results):
-		link = util.finance_link(t, exchanges[t], service, days=span_days, brief=False)
-		caption.append(f"{util.get_emoji(pct)} {names[t]} ({link}): {pct:+,.1f}%")
+		link = links.finance_link(t, exchanges[t], service, days=span_days, brief=False)
+		caption.append(f"{formats.get_emoji(pct)} {names[t]} ({link}): {pct:+,.1f}%")
 	if config.prefs.cache:
 		cache.write_bytes(image_cache_file, image)
 		cache.write(caption_cache_file, caption)
@@ -245,7 +248,7 @@ def prepare_help(service, botName):
 	else:
 		payload.append(botName + ' price')
 	payload.append("")
-	payload.append(f"Full help on {util.link('https://github.com/robdevops/finbot', 'GitHub', service)}")
+	payload.append(f"Full help on {links.link('https://github.com/robdevops/finbot', 'GitHub', service)}")
 	return payload
 
 def prepare_holdings_payload(portfolio_name, service):
@@ -307,23 +310,23 @@ def prepare_marketcap_payload(service, action='top', length=15):
 	"""The `length` tracked stocks of largest (action 'top') or smallest ('bottom') market cap, compared in US dollars at Yahoo's
 	exchange rates: a market cap in won or yen is a far bigger number. Any other action is one ticker's, in its own currency too."""
 	if action not in ('top', 'bottom'):
-		ticker = util.transform_to_yahoo(action.upper())
+		ticker = markets.transform_to_yahoo(action.upper())
 		data = yahoo.add_usd_market_caps(yahoo.fetch_detail(ticker, 600)).get(ticker, {})
 		if 'market_cap' not in data:
 			return [f"Mkt cap not found for {ticker}"]
 		currency = yahoo.market_cap_currency(data) or ''
-		cap = f"{currency} {util.humanUnits(data['market_cap'])}".strip()
+		cap = f"{currency} {formats.humanUnits(data['market_cap'])}".strip()
 		if currency != 'USD' and 'market_cap_usd' in data:
-			cap += f" (USD {util.humanUnits(data['market_cap_usd'])})"
-		link = util.finance_link(ticker, data.get('profile_exchange', ''), service)
-		return [util.security_line(util.flag_from_ticker(ticker), data.get('profile_title', ticker), link) + f" mkt cap: {cap}"]
+			cap += f" (USD {formats.humanUnits(data['market_cap_usd'])})"
+		link = links.finance_link(ticker, data.get('profile_exchange', ''), service)
+		return [formats.security_line(markets.flag_from_ticker(ticker), data.get('profile_title', ticker), link) + f" mkt cap: {cap}"]
 	caps = [] # (market cap in US dollars, line)
-	for ticker, data in yahoo.add_usd_market_caps(yahoo.fetch(util.get_holdings_and_watchlist())).items():
+	for ticker, data in yahoo.add_usd_market_caps(yahoo.fetch(watchlist.with_holdings())).items():
 		if 'market_cap_usd' not in data:
 			print(ticker, "no market cap" if 'market_cap' not in data else f"no exchange rate for {data.get('currency')}", file=sys.stderr)
 			continue
-		link = util.finance_link(ticker, data['profile_exchange'], service)
-		caps.append((data['market_cap_usd'], util.security_line(util.flag_from_ticker(ticker), data['profile_title'], link) + f" mkt cap: {util.humanUnits(data['market_cap_usd'])}"))
+		link = links.finance_link(ticker, data['profile_exchange'], service)
+		caps.append((data['market_cap_usd'], formats.security_line(markets.flag_from_ticker(ticker), data['profile_title'], link) + f" mkt cap: {formats.humanUnits(data['market_cap_usd'])}"))
 	if not caps:
 		return ['No market cap data found']
 	caps.sort(key=lambda cap: cap[0], reverse=action == 'top')
@@ -334,14 +337,14 @@ def prepare_rating_payload(service, action, length=15):
 	and the consensus says little: ASX small caps covered by two to six brokers are nearly all a perfect 1.0 buy), the strongest
 	rating first (1 is a strong buy, 5 a strong sell); on equal ratings, the one more analysts give first."""
 	found = [] # (rating, analysts, line, ticker)
-	for ticker, data in yahoo.fetch_details(util.get_holdings_and_watchlist()).items():
+	for ticker, data in yahoo.fetch_details(watchlist.with_holdings()).items():
 		if 'recommend' not in data or data['recommend_analysts'] < config.prefs.min_analysts:
 			continue
 		recommend = data['recommend'].replace('_', ' ')
 		if (action == 'buy' and 'buy' in recommend) or (action == 'sell' and recommend in ('sell', 'underperform')):
 			rating, analysts = round(data['recommend_index'], 2), data['recommend_analysts']
-			link = util.finance_link(ticker, data['profile_exchange'], service)
-			found.append((rating, analysts, f"{util.flag_from_ticker(ticker)} {data['profile_title']} ({link}) {rating} ({analysts} analysts)", ticker))
+			link = links.finance_link(ticker, data['profile_exchange'], service)
+			found.append((rating, analysts, f"{markets.flag_from_ticker(ticker)} {data['profile_title']} ({link}) {rating} ({analysts} analysts)", ticker))
 	found = sorted(found, key=lambda row: (row[0] if action == 'buy' else -row[0], -row[1]))[:length]
 	payload = [line for _, _, line, _ in found]
 	if not payload:
@@ -360,7 +363,7 @@ def prepare_value_payload(service, action='pe', ticker_select=None, length=15):
 	take only stocks that min_analysts or more analysts cover, as .buy does."""
 	payload = []
 	chart_rows = [] # (ticker, ratio)
-	tickers = [ticker_select] if ticker_select else util.get_holdings_and_watchlist()
+	tickers = [ticker_select] if ticker_select else watchlist.with_holdings()
 	market_data = yahoo.fetch(tickers)
 	if action in ('peg', 'bottom peg', 'negative peg'): # the PEG is in Yahoo's details
 		market_data |= yahoo.fetch_details(market_data, strict=bool(ticker_select))
@@ -403,8 +406,8 @@ def prepare_value_payload(service, action='pe', ticker_select=None, length=15):
 		if ratio is None:
 			continue
 		profile_title = market_data.get(ticker).get('profile_title')
-		ticker_link = util.finance_link(ticker, market_data.get(ticker).get('profile_exchange'), service)
-		flag = util.flag_from_ticker(ticker)
+		ticker_link = links.finance_link(ticker, market_data.get(ticker).get('profile_exchange'), service)
+		flag = markets.flag_from_ticker(ticker)
 		if ticker_select and action in {'pe', 'forward pe'}:
 			parts = []
 			if trailing_ratio is not None:
@@ -437,13 +440,13 @@ def prepare_value_payload(service, action='pe', ticker_select=None, length=15):
 def prepare_beta_payload(service):
 	"""(payload, chart) of the tracked stocks with a beta over 1.5 and a market cap under US$2B, the highest beta first."""
 	found = [] # (beta, line, chart row)
-	for ticker, data in yahoo.add_usd_market_caps(yahoo.fetch_details(util.get_holdings_and_watchlist())).items():
+	for ticker, data in yahoo.add_usd_market_caps(yahoo.fetch_details(watchlist.with_holdings())).items():
 		if 'beta' not in data:
 			continue
 		beta = round(data['beta'], 1)
 		if beta > 1.5 and data.get('market_cap_usd', float('inf')) < 2e9:
-			link = util.finance_link(ticker, data['profile_exchange'], service)
-			found.append((beta, util.security_line(util.flag_from_ticker(ticker), data['profile_title'], link) + f" {beta} β", (ticker.split('.')[0], float(beta))))
+			link = links.finance_link(ticker, data['profile_exchange'], service)
+			found.append((beta, formats.security_line(markets.flag_from_ticker(ticker), data['profile_title'], link) + f" {beta} β", (ticker.split('.')[0], float(beta))))
 	if not found:
 		return ['No tracked stocks have a beta over 1.5 and mkt cap under US$2B'], None
 	found.sort(key=lambda row: row[0], reverse=True)
@@ -455,7 +458,7 @@ def prepare_beta_payload(service):
 
 def prepare_profile_payload(service, user, ticker):
 	"""The .SYMBOL report: what Yahoo knows about one company. A field Yahoo did not send leaves its line out."""
-	ticker_orig = util.transform_to_yahoo(ticker)
+	ticker_orig = markets.transform_to_yahoo(ticker)
 	now = datetime.datetime.now()
 	payload = []
 	ticker, market_data = yahoo.resolve(ticker_orig, lambda symbol: yahoo.fetch_detail(symbol, 600))
@@ -467,12 +470,12 @@ def prepare_profile_payload(service, user, ticker):
 	state = data.get('marketState', '').rstrip()
 	currency = data.get('currency', '')
 	exchange = data.get('profile_exchange', '').replace('NasdaqCM', 'Nasdaq').replace('NasdaqGS', 'Nasdaq').replace('NYSEArca', 'NYSE')
-	ticker_link = util.finance_link(ticker, exchange, service, brief=False)
-	gfinance_link = util.gfinance_link(ticker, exchange, service, brief=True, text='google') # all three before exchange is renamed below
-	sws_link = util.link(simplywallst.get_url(ticker, title, exchange), 'simplywall.st', service)
-	macrotrends_link = util.link('https://www.google.com/search?q=site:macrotrends.net+' + title + '+PE Ratio+' + code + '&btnI', 'macrotrends', service)
+	ticker_link = links.finance_link(ticker, exchange, service, brief=False)
+	gfinance_link = links.gfinance_link(ticker, exchange, service, brief=True, text='google') # all three before exchange is renamed below
+	sws_link = links.link(simplywallst.get_url(ticker, title, exchange), 'simplywall.st', service)
+	macrotrends_link = links.link('https://www.google.com/search?q=site:macrotrends.net+' + title + '+PE Ratio+' + code + '&btnI', 'macrotrends', service)
 	if 'profile_bio' in data:
-		payload.append(util.make_paragraphs(data['profile_bio']))
+		payload.append(formats.make_paragraphs(data['profile_bio']))
 
 	if exchange == 'ASX':
 		exchange_url = 'https://www2.asx.com.au/markets/company/' + code
@@ -490,7 +493,7 @@ def prepare_profile_payload(service, user, ticker):
 		exchange_url = 'https://quote.jpx.co.jp/jpx/template/quote.cgi?F=tmp/e_stock_detail&MKTN=T&QCODE=' + code
 	else:
 		exchange_url = 'https://www.google.com/search?q=stock+exchange+' + exchange + '+' + code + '&btnI'
-	exchange_link = util.link(exchange_url, exchange, service)
+	exchange_link = links.link(exchange_url, exchange, service)
 	location = [data[key] for key in ('profile_city', 'profile_state', 'profile_country') if key in data]
 	if payload:
 		payload.append("")
@@ -502,19 +505,19 @@ def prepare_profile_payload(service, user, ticker):
 	if 'profile_employees' in data:
 		payload.append(webhook.bold("Employees: ", service) + f"{data['profile_employees']:,}")
 	if 'profile_website' in data:
-		links = [util.link(data['profile_website'], util.strip_url(data['profile_website']), service), exchange_link]
+		sites = [links.link(data['profile_website'], formats.strip_url(data['profile_website']), service), exchange_link]
 		if 'NYSE' in exchange or 'Nasdaq' in exchange:
-			links += [util.link('https://finviz.com/quote.ashx?t=' + ticker, 'finviz', service), macrotrends_link,
-				util.link('https://seekingalpha.com/symbol/' + ticker, 'seekingalpha', service)]
+			sites += [links.link('https://finviz.com/quote.ashx?t=' + ticker, 'finviz', service), macrotrends_link,
+				links.link('https://seekingalpha.com/symbol/' + ticker, 'seekingalpha', service)]
 		elif exchange == 'ASX':
-			links.append(util.link('https://www.shortman.com.au/stock?q=' + code.lower(), 'shortman', service))
-		payload.append(webhook.bold("Links: ", service) + ' | '.join(links))
+			sites.append(links.link('https://www.shortman.com.au/stock?q=' + code.lower(), 'shortman', service))
+		payload.append(webhook.bold("Links: ", service) + ' | '.join(sites))
 
 	if payload:
 		payload.append("")
 
 	if 'currency' in data and 'market_cap' in data:
-		payload.append(webhook.bold("Mkt cap:", service) + f" {yahoo.market_cap_currency(data)} {util.humanUnits(data['market_cap'])}") # GBP, not pence
+		payload.append(webhook.bold("Mkt cap:", service) + f" {yahoo.market_cap_currency(data)} {formats.humanUnits(data['market_cap'])}") # GBP, not pence
 	if all(key in data for key in ('shareholder_equity', 'total_debt', 'total_cash', 'profile_industry')) and data['shareholder_equity'] and 'Bank' not in data['profile_industry']:
 		net_debt_equity_ratio = round((data['total_debt'] - data['total_cash']) / data['shareholder_equity'] * 100)
 		if net_debt_equity_ratio > 0:
@@ -554,9 +557,9 @@ def prepare_profile_payload(service, user, ticker):
 			payload.append(webhook.bold(label, service) + (f" +{growth}% " if growth > 0 else f" {growth}% ⚠️ "))
 	if 'insiderBuy' in data:
 		if data['insiderBuy'] > data['insiderSell']:
-			payload.append(webhook.bold("Net insider action (3M):", service) + f" Buy {currency} {util.humanUnits(data['insiderBuyValue'])} ")
+			payload.append(webhook.bold("Net insider action (3M):", service) + f" Buy {currency} {formats.humanUnits(data['insiderBuyValue'])} ")
 		elif data['insiderBuy'] < data['insiderSell']:
-			payload.append(webhook.bold("Net insider action (3M):", service) + f" Sell {currency} {util.humanUnits(data['insiderSellValue'])}⚠️ ")
+			payload.append(webhook.bold("Net insider action (3M):", service) + f" Sell {currency} {formats.humanUnits(data['insiderSellValue'])}⚠️ ")
 	if 'short_percent' in data:
 		payload.append(webhook.bold("Shorted stock:", service) + f" {data['short_percent']}%" + ('⚠️ ' if data['short_percent'] > 10 else ''))
 	if 'recommend' in data:
@@ -600,17 +603,17 @@ def prepare_profile_payload(service, user, ticker):
 	for interval in ('5Y', '1Y', '1M', '1D'):
 		if interval in price_history:
 			percent = price_history[interval]
-			payload.append(f"{util.get_emoji(percent)} {webhook.bold(interval + ':', service)} {percent}%")
+			payload.append(f"{formats.get_emoji(percent)} {webhook.bold(interval + ':', service)} {percent}%")
 	if state != 'REGULAR':
 		for key, label in (('percent_change_premarket', 'Pre-market:'), ('percent_change_postmarket', 'Post-market:')):
 			if key in data:
-				payload.append(f"{util.get_emoji(data[key])} {webhook.bold(label, service)} {data[key]:,}%")
+				payload.append(f"{formats.get_emoji(data[key])} {webhook.bold(label, service)} {data[key]:,}%")
 				break
 
 	if 'profile_website' in data and config.prefs.hyperlinkFooter and config.prefs.hyperlink:
 		if payload:
 			payload.append("")
-		payload.append(f"{gfinance_link} | {util.yahoo_link(ticker, service, brief=True, text='yahoo')} | {sws_link}")
+		payload.append(f"{gfinance_link} | {links.yahoo_link(ticker, service, brief=True, text='yahoo')} | {sws_link}")
 	heading = f"{webhook.bold(title, service)} ({ticker_link}) {MARKET_STATE_EMOJI.get(state, '🔴') if state else ''}"
 	if ticker_orig == ticker:
 		payload.insert(0, heading)
diff --git a/lib/sharesight.py b/lib/sharesight.py
index 53223f5d..347841f2 100644
--- a/lib/sharesight.py
+++ b/lib/sharesight.py
@@ -7,7 +7,8 @@
 from lib import config
 from lib import cache
 from lib import net
-import lib.util as util
+from lib import formats
+from lib import markets
 
 MAX_CONCURRENT = 1 # Sharesight rejects parallel requests beyond a few per account: allow one at a time across all chats and threads
 _request_slots = threading.BoundedSemaphore(MAX_CONCURRENT)
@@ -52,7 +53,7 @@ def get_token():
 			cache_expiry = datetime.datetime.fromtimestamp(cached['created_at'] + cached['expires_in'])
 			ttl = cache_expiry - datetime.datetime.now()
 			if ttl > datetime.timedelta(seconds=20):
-				print("cache hit:", cacheFile, "TTL:", util.td_to_human(ttl), file=sys.stderr) if config.prefs.debug else None
+				print("cache hit:", cacheFile, "TTL:", formats.td_to_human(ttl), file=sys.stderr) if config.prefs.debug else None
 				return cached['access_token']
 			else:
 				print("cache miss:", cacheFile, file=sys.stderr) if config.prefs.debug else None
@@ -113,7 +114,7 @@ def holdings(self):
 
 	def tickers(self):
 		"""What it holds, as Yahoo tickers, sorted."""
-		return sorted({util.transform_to_yahoo(each['code'], each['market_code']) for each in self.holdings().values()})
+		return sorted({markets.transform_to_yahoo(each['code'], each['market_code']) for each in self.holdings().values()})
 
 	def trades(self, days=None):
 		"""Its trades of the past `days` (the past_days preference when None)."""
diff --git a/lib/watchlist.py b/lib/watchlist.py
index b1ba936f..d0edac4d 100644
--- a/lib/watchlist.py
+++ b/lib/watchlist.py
@@ -1,5 +1,6 @@
 """The watchlist: tickers followed beside the Sharesight holdings, in DATA_DIR/finbot_watchlist.json. Every call reads the file,
 so the bot and its scheduled reports (other processes) always see the same list."""
+from lib import sharesight
 from lib import state
 from lib import yahoo
 
@@ -34,3 +35,11 @@ def remove(ticker):
 			return False
 		watched.remove(ticker)
 	return True
+
+
+def with_holdings():
+	"""The watchlist and every Sharesight holding: the tickers the scheduled reports cover, sorted."""
+	followed = set(sharesight.tickers()) | set(tickers())
+	if 'GOOG' in followed and 'GOOGL' in followed:
+		followed.remove("GOOGL")
+	return sorted(followed)
diff --git a/lib/worker.py b/lib/worker.py
index e48522eb..27b2732e 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -9,7 +9,9 @@
 from lib import config
 from lib import state
 from lib import words
-from lib import util
+from lib import formats
+from lib import markets
+from lib import render
 from lib import webhook
 from lib import reports
 from lib import charts
@@ -94,7 +96,7 @@ def deliver(self, payload, chart=None):
 			return
 		spec = dict(chart)
 		rows, title, subtitle = spec.pop('rows'), spec.pop('title'), spec.pop('subtitle', '')
-		image = util.pie_chart(rows, title, subtitle) if spec.pop('pie', False) else util.rows_chart(rows, title, subtitle, **spec)
+		image = render.pie_chart(rows, title, subtitle) if spec.pop('pie', False) else render.rows_chart(rows, title, subtitle, **spec)
 		webhook.sendPhoto(self.chat_id, image, '\n'.join(payload) if payload else title, self.service)
 
 def process_job_callback(service, callback_id, chat_id, message_id, data):
@@ -168,7 +170,7 @@ def period_or_ticker(arg, default):
 	if not arg:
 		return default, None
 	try:
-		return util.days_from_human_days(arg), None
+		return formats.days_from_human_days(arg), None
 	except ValueError:
 		return default, arg.upper()
 
@@ -188,7 +190,7 @@ def period_and_portfolio(text, days):
 	for i in (0, -1):
 		if words:
 			try:
-				days = util.days_from_human_days(words[i])
+				days = formats.days_from_human_days(words[i])
 			except ValueError:
 				continue
 			del words[i]
@@ -263,7 +265,7 @@ def on_price(request, m):
 			threshold = parse_percent(arg)
 		except ValueError:
 			try:
-				days = util.days_from_human_days(arg)
+				days = formats.days_from_human_days(arg)
 			except ValueError:
 				if i == 0 and arg in ('top', 'bottom'): # one chart shows both ends
 					threshold, top = 0, 10
@@ -303,7 +305,7 @@ def handler(request, m):
 		elif negative and 'neg' in arg:
 			action = f'negative {kind}'
 		elif arg and arg != 'top':
-			ticker = util.transform_to_yahoo(arg.upper())
+			ticker = markets.transform_to_yahoo(arg.upper())
 		with request.typing():
 			payload, spec = reports.prepare_value_payload(request.service, action, ticker, length=15)
 		request.deliver(payload, spec if chart else None)
@@ -342,7 +344,7 @@ def on_history(request, m):
 	if not m.group('ticker') or m.group('extra'):
 		request.say(["Usage: .history SYMBOL"])
 		return
-	ticker = util.transform_to_yahoo(m.group('ticker').upper())
+	ticker = markets.transform_to_yahoo(m.group('ticker').upper())
 	with request.typing():
 		caption, image = charts.build('h', [ticker], None, request.service)
 	request.send(caption.split('\n'), image, charts.keyboard('h', [ticker], 'x') if image and request.service == 'telegram' else None)
diff --git a/lib/yahoo.py b/lib/yahoo.py
index 515e6af0..aa064c2e 100644
--- a/lib/yahoo.py
+++ b/lib/yahoo.py
@@ -9,7 +9,8 @@
 from lib import config
 from lib import cache
 from lib import net
-from lib import util
+from lib import formats
+from lib import render
 from http.cookies import SimpleCookie
 from pathlib import Path
 
@@ -164,7 +165,7 @@ def fetch(tickers):
 		title = next((name for name in (item.get('longName'), item.get('shortName')) if isinstance(name, str) and name), None)
 		if not title or item.get('regularMarketPrice') is None:
 			continue
-		quote = {'profile_title': util.transform_title(title), 'ticker': item['symbol'], 'regularMarketPrice': item['regularMarketPrice'],
+		quote = {'profile_title': formats.transform_title(title), 'ticker': item['symbol'], 'regularMarketPrice': item['regularMarketPrice'],
 			'currency': item.get('currency'), 'percent_change': 0.0, 'dividend': 0.0}
 		for key, field, convert in QUOTE_FIELDS:
 			if (value := _convert(convert, item.get(field))) is not None:
@@ -252,7 +253,7 @@ def fetch_detail(ticker, seconds=None):
 	if not title: # catches some delisted stocks like "DUB"
 		print(f"{ticker}†", sep=' ', end='', flush=True, file=sys.stderr)
 		return {}
-	detail = {'profile_title': util.transform_title(title)}
+	detail = {'profile_title': formats.transform_title(title)}
 	for key, path, *convert in DETAIL_FIELDS:
 		if (value := _convert(convert[0] if convert else None, dig(result, *path))) is not None:
 			detail[key] = value
@@ -354,16 +355,16 @@ def price_history(ticker, days=None, seconds=None, graph=None, graphCache=True):
 			percent_dict[period] = round((price - past_price) / past_price * 100)
 	if not graph or (days and days < 2):
 		return percent_dict, None
-	company_name = util.transform_title(stock.get('longName', ''))
+	company_name = formats.transform_title(stock.get('longName', ''))
 	if days:
-		title = f"{company_name} ({ticker}) {min(days, max_days)} days {util.signed_percent(percent_dict[days])} Last: {price}"
+		title = f"{company_name} ({ticker}) {min(days, max_days)} days {formats.signed_percent(percent_dict[days])} Last: {price}"
 	else:
 		label = "Max" if "Max" in percent_dict else "10Y"
-		title = f"{company_name} ({ticker}) {label} {util.signed_percent(percent_dict[label])} Last: {price}"
+		title = f"{company_name} ({ticker}) {label} {formats.signed_percent(percent_dict[label])} Last: {price}"
 	image_cache_file = "finbot_graph_" + ticker + "_" + str(days) + ".png"
 	if config.prefs.cache and graphCache and (image := cache.read_bytes(image_cache_file, seconds)):
 		return percent_dict, image
-	image = util.graph(df, title, stock.get('currency'))
+	image = render.graph(df, title, stock.get('currency'))
 	if config.prefs.cache:
 		cache.write_bytes(image_cache_file, image)
 	return percent_dict, image
@@ -379,7 +380,7 @@ def price_series(ticker):
 	if df.empty:
 		raise RuntimeError(f"no price history for {ticker}")
 	series = pd.Series(df['Close'].astype(float).values, index=pd.to_datetime(df['Date'].astype(str)), name=ticker)
-	name = util.transform_title(stock.get('longName') or stock.get('shortName') or ticker)
+	name = formats.transform_title(stock.get('longName') or stock.get('shortName') or ticker)
 	return name, stock.get('fullExchangeName') or stock.get('exchangeName') or '', series
 
 def fetch_each(get, tickers, workers=8):
diff --git a/milestone.py b/milestone.py
index bf496a57..dc5303b6 100755
--- a/milestone.py
+++ b/milestone.py
@@ -4,7 +4,9 @@
 from itertools import groupby
 from lib import state
 from lib import webhook
-from lib import util
+from lib import links
+from lib import markets
+from lib import watchlist
 from lib import yahoo
 
 # BUGS:
@@ -31,8 +33,8 @@ def prepare_milestone_payload(service, market_data):
 			except (KeyError, ValueError):
 				continue # likewise: a ticker is recorded only with both of its new values
 			title = market_data[ticker]['profile_title']
-			ticker_link = util.finance_link(ticker, market_data[ticker]['profile_exchange'], service)
-			emoji = util.flag_from_ticker(ticker)
+			ticker_link = links.finance_link(ticker, market_data[ticker]['profile_exchange'], service)
+			emoji = markets.flag_from_ticker(ticker)
 			changed[ticker] = {'record_high': max(oldhigh, newhigh), 'record_low': min(oldlow, newlow)} # all-time: a 52-week high below it is no record
 			if newhigh > oldhigh:
 				payloadhigh.append(f"{emoji} {title} ({ticker_link})")
@@ -56,7 +58,7 @@ def prepare_milestone_payload(service, market_data):
 
 	# MAIN #
 
-	tickers = util.get_holdings_and_watchlist()
+	tickers = watchlist.with_holdings()
 	market_data = yahoo.fetch(tickers)
 	records, changed = state.load('finbot_milestone.json', {}), {}
 	for service, url, chat in webhook.default_chats():
diff --git a/performance.py b/performance.py
index 314b9f5f..dc200aa6 100755
--- a/performance.py
+++ b/performance.py
@@ -5,7 +5,9 @@
 from lib import config
 from lib import sharesight
 from lib import webhook
-from lib import util
+from lib import formats
+from lib import links
+from lib import render
 from lib import yahoo
 from lib import charts
 
@@ -28,8 +30,8 @@ def benchmark(ticker, market, text, past_days, service):
 	except Exception as e: # the rest of the report still goes out
 		print(ticker, e, file=sys.stderr)
 		return f"⚠️ {text}: no price history", None
-	link = util.finance_link(ticker, market, service=service, days=past_days, brief=True, text=text)
-	return f"{util.get_emoji(percent)} {link} {util.signed_percent(percent, 2)}", percent
+	link = links.finance_link(ticker, market, service=service, days=past_days, brief=True, text=text)
+	return f"{formats.get_emoji(percent)} {link} {formats.signed_percent(percent, 2)}", percent
 
 def report(service, past_days=None, portfolio_select=None, interactive=False, performance=None):
 	"""(payload, chart) of the performance report for one service: each portfolio and the benchmarks, best first. performance:
@@ -41,9 +43,9 @@ def report(service, past_days=None, portfolio_select=None, interactive=False, pe
 	lines = [] # (percent, text), sorted into the payload below
 	chart_rows = []
 	for portfolio_id, (name, data) in performance.items(): # the name from the portfolio list: one with no holdings names none
-		link = util.link("https://portfolio.sharesight.com/portfolios/" + str(portfolio_id), name, service)
+		link = links.link("https://portfolio.sharesight.com/portfolios/" + str(portfolio_id), name, service)
 		percent = float(data['report']['capital_gain_percent'])
-		lines.append((percent, f"{util.get_emoji(percent)} {link} {util.signed_percent(percent, 2)}"))
+		lines.append((percent, f"{formats.get_emoji(percent)} {link} {formats.signed_percent(percent, 2)}"))
 		chart_rows.append((name, percent))
 	if not lines:
 		return [], None
@@ -53,12 +55,12 @@ def report(service, past_days=None, portfolio_select=None, interactive=False, pe
 		if percent is not None:
 			chart_rows.append((text, float(percent)))
 	lines.sort(key=lambda item: item[0], reverse=True)
-	period = util.days_english(past_days)
+	period = formats.days_english(past_days)
 	payload = [webhook.bold(f"Performance over {period}", service)] + [line for _, line in lines]
 	graph = None
 	if webhook.takes_charts(service, interactive):
 		chart_rows.sort(key=lambda row: row[1], reverse=True) # best first
-		graph = util.rows_chart(chart_rows, "Performance", f"% change {'' if period == 'today' else 'over '}{period}", value_fmt=lambda v: util.signed_percent(v, 2))
+		graph = render.rows_chart(chart_rows, "Performance", f"% change {'' if period == 'today' else 'over '}{period}", value_fmt=lambda v: formats.signed_percent(v, 2))
 	return payload, graph
 
 def lambda_handler(chat_id=None, past_days=None, service=None, user='', portfolio_select=None, interactive=False):
diff --git a/podcasts.py b/podcasts.py
index 0239ef4c..650f2651 100755
--- a/podcasts.py
+++ b/podcasts.py
@@ -9,7 +9,7 @@
 from lib import net
 from lib import state
 from lib import webhook
-from lib import util
+from lib import links
 
 NAMESPACES = {
 	"media": "http://search.yahoo.com/mrss/",
@@ -101,7 +101,7 @@ def lambda_handler(feed_url):
 			podcast_name = episode['podcast']
 			title = episode['title'].removeprefix(podcast_name).lstrip(string.punctuation + " ")
 			duration = f" {episode['duration']}" if episode['duration'] else ""
-			payload.append(f"📣 {podcast_name}: {util.link(episode['url'], title, service)}{duration} ({episode['ago']})")
+			payload.append(f"📣 {podcast_name}: {links.link(episode['url'], title, service)}{duration} ({episode['ago']})")
 		webhook.send(service, url, chat, payload)
 	with state.edit(seen_file, []) as recorded: # last, so that a failure above leaves these episodes for next run
 		recorded[:] = sorted(guids.union(recorded))
diff --git a/price.py b/price.py
index 90d29cf2..8bf4c0b0 100755
--- a/price.py
+++ b/price.py
@@ -6,14 +6,18 @@
 from lib import config
 from lib import sharesight
 from lib import webhook
-from lib import util
+from lib import formats
+from lib import links
+from lib import markets
+from lib import render
+from lib import watchlist
 from lib import yahoo
 from lib import charts
 
 def gather(specific_stock=None, days=None):
 	"""(tickers, market data) for a price report: Yahoo's quotes; over `days` each holding's Sharesight figure, and for the tickers
 	Sharesight has none for (the watchlist) their Yahoo history, fetched together up front."""
-	tickers = [util.transform_to_yahoo(specific_stock)] if specific_stock else util.get_holdings_and_watchlist()
+	tickers = [markets.transform_to_yahoo(specific_stock)] if specific_stock else watchlist.with_holdings()
 	market_data = yahoo.fetch(tickers)
 	if config.prefs.demote_volatile: # trigger_bar compares market caps, in US dollars
 		yahoo.add_usd_market_caps(market_data)
@@ -21,7 +25,7 @@ def gather(specific_stock=None, days=None):
 	if days and (not specific_stock or config.prefs.performance_use_sharesight):
 		for portfolio in sharesight.portfolios():
 			for holding in portfolio.performance(days)['report']['holdings']:
-				ticker = util.transform_to_yahoo(holding['instrument']['code'], holding['instrument']['market_code'])
+				ticker = markets.transform_to_yahoo(holding['instrument']['code'], holding['instrument']['market_code'])
 				if ticker in market_data:
 					market_data[ticker]['percent_change_period'] = float(holding['capital_gain_percent'])
 				elif ticker in tickers:
@@ -102,19 +106,19 @@ def report(service, threshold=None, user='', specific_stock=None, interactive=Fa
 		percent = float(percent)
 		if abs(percent) >= threshold or (specific_stock and interactive):
 			exchange = quote['profile_exchange']
-			exchanges.add(util.exchange_human(exchange))
-			rows.append((ticker, percent, f"{util.get_emoji(percent)} {quote['profile_title']} ({util.quote_link(ticker, exchange, service, days=days)})"))
+			exchanges.add(markets.exchange_human(exchange))
+			rows.append((ticker, percent, f"{formats.get_emoji(percent)} {quote['profile_title']} ({links.quote_link(ticker, exchange, service, days=days)})"))
 	if not triggered: # a scheduled report: is there a move big enough for its ticker? Small caps' betas are fetched together
 		details = yahoo.fetch_details([ticker for ticker, _, _ in rows if needs_beta(market_data[ticker])])
 		if not any(abs(percent) >= trigger_bar(ticker, market_data[ticker], threshold, details) for ticker, percent, _ in rows):
 			rows = []
 	rows.sort(key=lambda row: row[1], reverse=True)
-	payload = [f"{line} {util.signed_percent(round(percent))}" for _, percent, line in rows]
+	payload = [f"{line} {formats.signed_percent(round(percent))}" for _, percent, line in rows]
 	chart_rows = [(ticker, percent) for ticker, percent, _ in rows]
 	label = lambda chart_rows: [(ticker.split('.')[0].removesuffix('-USD'), percent) for ticker, percent in chart_rows]
 	if payload and not specific_stock:
 		as_of = latest or datetime.date.today()
-		on = util.date_short(as_of, True)
+		on = formats.date_short(as_of, True)
 		where = ", ".join(exchanges)
 		if midsession:
 			heading = f'≥ {threshold:g}% mid-session, {on} ({where})'
@@ -123,9 +127,9 @@ def report(service, threshold=None, user='', specific_stock=None, interactive=Fa
 		elif close:
 			heading = f'≥ {threshold:g}% at close, {on} ({where})'
 		elif top:
-			heading = f'Top {top} performers {util.date_range_english(days, as_of) if days else on}'
+			heading = f'Top {top} performers {formats.date_range_english(days, as_of) if days else on}'
 		elif days:
-			heading = f'Price movement {threshold:g}%, {util.date_range_english(days, as_of)}'
+			heading = f'Price movement {threshold:g}%, {formats.date_range_english(days, as_of)}'
 		else:
 			heading = f'{threshold:g}% at close, {on}'
 		chartable = webhook.takes_charts(service, interactive)
@@ -135,11 +139,11 @@ def report(service, threshold=None, user='', specific_stock=None, interactive=Fa
 			chart_top = chart_rows[:top]
 			chart_bottom = [row for row in reversed(chart_rows[-top:]) if row not in chart_top] # worst first
 			if chartable and chart_top:
-				graph = util.bar_graph(label(chart_top), label(chart_bottom), f"Top & bottom {top}", f"% change {util.days_english(days, 'in ', 'the past ')}")
+				graph = render.bar_graph(label(chart_top), label(chart_bottom), f"Top & bottom {top}", f"% change {formats.days_english(days, 'in ', 'the past ')}")
 		else:
 			payload.insert(0, webhook.bold(heading, service))
 			if chartable:
-				graph = util.rows_chart(label(chart_rows), heading, session='midsession' if midsession else 'premarket' if premarket else None)
+				graph = render.rows_chart(label(chart_rows), heading, session='midsession' if midsession else 'premarket' if premarket else None)
 	elif not payload and interactive:
 		if specific_stock:
 			if midsession:
diff --git a/rating.py b/rating.py
index d1807646..816f31ee 100755
--- a/rating.py
+++ b/rating.py
@@ -3,7 +3,9 @@
 import sys
 from lib import webhook
 from lib import state
-from lib import util
+from lib import links
+from lib import markets
+from lib import watchlist
 from lib import yahoo
 
 def lambda_handler(chat_id=None, specific_stock=None, service=None, interactive=False):
@@ -28,7 +30,7 @@ def get_emoji(old, new):
 			except (KeyError, ValueError):
 				continue
 			title = market_data[ticker]['profile_title']
-			ticker_link = util.finance_link(ticker, market_data[ticker]['profile_exchange'], service, brief=False)
+			ticker_link = links.finance_link(ticker, market_data[ticker]['profile_exchange'], service, brief=False)
 			new[ticker] = [action, index]
 			try:
 				old_action, old_index = old[ticker]
@@ -54,7 +56,7 @@ def get_emoji(old, new):
 
 	# MAIN #
 
-	tickers = [util.transform_to_yahoo(specific_stock)] if specific_stock else util.get_holdings_and_watchlist()
+	tickers = [markets.transform_to_yahoo(specific_stock)] if specific_stock else watchlist.with_holdings()
 	old = state.load('finbot_rating.json', {})
 	market_data = yahoo.fetch_details(tickers, strict=bool(specific_stock))
 
diff --git a/reminder.py b/reminder.py
index 7de0f988..7aaaf741 100755
--- a/reminder.py
+++ b/reminder.py
@@ -5,7 +5,8 @@
 import zoneinfo
 from dateutil.relativedelta import relativedelta
 from lib import config
-from lib import util
+from lib import formats
+from lib import links
 from lib import webhook
 
 
@@ -47,9 +48,9 @@ def prepare_finance_calendar_payload(service):
 				payload.append("• work expenses‡")
 				payload.append("• investment subscriptions§")
 				payload.append ("")
-				abr_link = util.link('https://abr.business.gov.au/Tools/DgrListing', 'DGR-registered', service)
-				stake_note = util.link('https://hellostake.com/au/support/stake-super/employer-and-personal-contributions/33442327233305#h_01JQAPN9A6ZM41NZKP6XZWN1KP', 'note about XX PCC in transaction description', service)
-				work_link = util.link('https://www.ato.gov.au/individuals-and-families/income-deductions-offsets-and-records/deductions-you-can-claim', 'eligible work expenses', service)
+				abr_link = links.link('https://abr.business.gov.au/Tools/DgrListing', 'DGR-registered', service)
+				stake_note = links.link('https://hellostake.com/au/support/stake-super/employer-and-personal-contributions/33442327233305#h_01JQAPN9A6ZM41NZKP6XZWN1KP', 'note about XX PCC in transaction description', service)
+				work_link = links.link('https://www.ato.gov.au/individuals-and-families/income-deductions-offsets-and-records/deductions-you-can-claim', 'eligible work expenses', service)
 				payload.append(webhook.italic("* Organisation must be " + abr_link, service))
 				payload.append(webhook.italic("† Allow transfer time for Super, as contributions can only be deducted for the year they're received. Stake users: see " + stake_note, service))
 				payload.append(webhook.italic("‡ See " + work_link, service))
@@ -58,7 +59,7 @@ def prepare_finance_calendar_payload(service):
 		if month_and_day == '08-18':
 			myBirthday = datetime.datetime(2022,8,18,0,0,0,0, tzinfo=tz)
 			difference = relativedelta(localtime, myBirthday)
-			payload.append("It's my " + str(difference.years) + util.ordinal(difference.years) + " birthday! 🥳")
+			payload.append("It's my " + str(difference.years) + formats.ordinal(difference.years) + " birthday! 🥳")
 
 		if payload:
 			heading = webhook.bold("Finance event reminders:", service)
diff --git a/shorts.py b/shorts.py
index 258a9d69..6819b4d1 100755
--- a/shorts.py
+++ b/shorts.py
@@ -3,7 +3,10 @@
 import sys
 from lib import config
 from lib import webhook
-from lib import util
+from lib import links
+from lib import markets
+from lib import render
+from lib import watchlist
 from lib import yahoo
 from lib import shortman
 
@@ -20,13 +23,13 @@ def prepare_shorts_payload(service, market_data):
 			if short_percent > threshold or specific_stock:
 				code = ticker.split('.')[0]
 				url = 'https://www.shortman.com.au/stock?q=' + code if ticker.endswith('.AX') else 'https://finance.yahoo.com/quote/' + ticker + '/key-statistics?p=' + ticker
-				found.append((short_percent, f"🩳 {market_data[ticker]['profile_title']} ({util.link(url, code, service)}) {round(short_percent)}%", code))
+				found.append((short_percent, f"🩳 {market_data[ticker]['profile_title']} ({links.link(url, code, service)}) {round(short_percent)}%", code))
 		found.sort(key=lambda row: row[0], reverse=True) # largest first, as the chart has them
 		payload, graph = [line for _, line, _ in found], None
 		if payload and not specific_stock:
 			payload.insert(0, webhook.bold(f'Tracked stocks shorted over {threshold:g}%', service))
 			if webhook.takes_charts(service, interactive):
-				graph = util.shorts_chart([(code, percent) for percent, _, code in found], threshold)
+				graph = render.shorts_chart([(code, percent) for percent, _, code in found], threshold)
 		elif not payload and interactive:
 			payload = [f"🩳 No short interest found for {tickers[0]}" if specific_stock else f"🩳 No stocks shorted over {threshold:g}%. Try specifying a number."]
 		return payload, graph
@@ -34,10 +37,10 @@ def prepare_shorts_payload(service, market_data):
 	# MAIN #
 
 	if specific_stock:
-		ticker = util.transform_to_yahoo(specific_stock)
+		ticker = markets.transform_to_yahoo(specific_stock)
 		tickers = [ticker]
 	else:
-		tickers = util.get_holdings_and_watchlist()
+		tickers = watchlist.with_holdings()
 
 	market_data = yahoo.fetch(tickers)
 	market_data |= yahoo.fetch_details((ticker for ticker in tickers if '.' not in ticker), strict=bool(specific_stock)) # US short interest; Short Man has the ASX's
diff --git a/tests/common.py b/tests/common.py
index 3f278f4d..225052fd 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -10,12 +10,12 @@
 os.environ.update(telegramOutgoingWebhook='https://example.com/telegram', telegramOutgoingToken='tok', telegramChatID='1')
 
 import bot # first, and once: it monkey-patches with gevent, which must happen before requests imports ssl and never mid-test
-from lib import cache, config, state, webhook, util, yahoo, sharesight, shortman, worker, charts, telegram, reports, watchlist, words
+from lib import cache, config, state, webhook, formats, links, markets, render, yahoo, sharesight, shortman, worker, charts, telegram, reports, watchlist, words
 import performance, shorts, price, reminder, cal, trades, milestone, rating
 
 REAL = {(m.__name__, n): getattr(m, n) for m, n in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (yahoo, 'price_history'),
 	(sharesight, 'get_portfolios'), (sharesight, 'get_trades'), (sharesight, 'get_performance'))}
-REAL_GET_HOLDINGS = util.get_holdings_and_watchlist
+REAL_GET_HOLDINGS = watchlist.with_holdings
 REAL_PAYLOAD_WRAPPER = webhook.payload_wrapper # FinbotCase replaces it; tests of the real thing use this
 REAL_REPORT_ERROR = webhook.report_error # likewise
 REAL_SEND_PHOTO = webhook.sendPhoto # likewise
@@ -73,7 +73,7 @@ def setUp(self):
 		self.patch(webhook, 'report_error', lambda e, *a, **k: self.errors.append(webhook.error_line(e, k.get('context'))))
 		self.patch(webhook, 'sendPhoto', lambda chat, img, cap, svc, **k: (self.sent.append(('photo', len(img.read()), cap.split('\n')[0])), self.markups.append(k.get('reply_markup'))))
 		self.patch(webhook, 'payload_wrapper', lambda svc, url, payload, *a, **k: self.sent.append(('text', payload)) or [])
-		self.patch(util, 'get_holdings_and_watchlist', lambda: TICKERS)
+		self.patch(watchlist, 'with_holdings', lambda: TICKERS)
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t)})
 		self.patch(yahoo, 'price_history', lambda t, days=None, **k: ({days: 1.5} if days else {'Max': 5, '1Y': 3}, None))
diff --git a/tests/test_charts.py b/tests/test_charts.py
index 0af42bd1..76c30a96 100644
--- a/tests/test_charts.py
+++ b/tests/test_charts.py
@@ -41,10 +41,10 @@ def compare_series():
 
 
 CHARTS = {
-	'hbar_mixed': lambda: util.hbar_chart(MIXED, 'Price movement 5.0%, 5 Sep - 5 Oct 2026', 'subtitle here'),
-	'price_line': lambda: util.graph(price_frame(), 'Apple Inc (AAPL)', 'USD'),
-	'compare_lines': lambda: util.compare_graph(compare_series(), 'Compare 120 days', 'subtitle here'),
-	'shorts': lambda: util.shorts_chart([('GME', 47.3), ('AMC', 32.1), ('TSLA', 21.6), ('AAPL', 15.2)], 15),
+	'hbar_mixed': lambda: render.hbar_chart(MIXED, 'Price movement 5.0%, 5 Sep - 5 Oct 2026', 'subtitle here'),
+	'price_line': lambda: render.graph(price_frame(), 'Apple Inc (AAPL)', 'USD'),
+	'compare_lines': lambda: render.compare_graph(compare_series(), 'Compare 120 days', 'subtitle here'),
+	'shorts': lambda: render.shorts_chart([('GME', 47.3), ('AMC', 32.1), ('TSLA', 21.6), ('AAPL', 15.2)], 15),
 }
 # inputs that have broken chart code before or are likely to: sizes, signs, long names, zeros, a single row
 EDGE_ROWS = {
@@ -85,44 +85,44 @@ def test_each_chart_type(self):
 	def test_edge_case_rows(self):
 		for name, rows in EDGE_ROWS.items():
 			with self.subTest(rows=name):
-				self.check_image(util.hbar_chart(rows, 'Title', 'sub'), name)
-				self.check_image(util.rows_chart(rows, 'Title'), name + '/rows_chart')
+				self.check_image(render.hbar_chart(rows, 'Title', 'sub'), name)
+				self.check_image(render.rows_chart(rows, 'Title'), name + '/rows_chart')
 
 	def test_zoomed_axis(self):
-		self.check_image(util.hbar_chart([('AAA', 1.2), ('BBB', 3.4)], 'Ratings', xlim=(0.9, 4.0)), 'zoomed')
-		self.check_image(util.hbar_chart([('WWWWWWWW', 1.2), ('MMM', 3.4)], 'Ratings', xlim=(0.9, 4.0)), 'zoomed wide labels')
-		self.check_image(util.hbar_chart([(f'T{i}', 1 + i / 10) for i in range(15)], 'Ratings', xlim=(0.9, 3.0)), 'zoomed 15 rows')
+		self.check_image(render.hbar_chart([('AAA', 1.2), ('BBB', 3.4)], 'Ratings', xlim=(0.9, 4.0)), 'zoomed')
+		self.check_image(render.hbar_chart([('WWWWWWWW', 1.2), ('MMM', 3.4)], 'Ratings', xlim=(0.9, 4.0)), 'zoomed wide labels')
+		self.check_image(render.hbar_chart([(f'T{i}', 1 + i / 10) for i in range(15)], 'Ratings', xlim=(0.9, 3.0)), 'zoomed 15 rows')
 
 	def test_shorts_chart_sizes_and_verdicts(self):
 		for n in (1, 5, 45):
 			rows = [(f'T{i}', 60.0 - i) for i in range(n)]
-			self.check_image(util.shorts_chart(rows, 15), f'shorts {n} rows', bg=SHORTS_BG, edge_clean=False)
-		self.check_image(util.shorts_chart([('A VERY LONG LABEL', 15.0), ('B', 0.4)], 15), 'shorts long label and tiny value', bg=SHORTS_BG, edge_clean=False)
+			self.check_image(render.shorts_chart(rows, 15), f'shorts {n} rows', bg=SHORTS_BG, edge_clean=False)
+		self.check_image(render.shorts_chart([('A VERY LONG LABEL', 15.0), ('B', 0.4)], 15), 'shorts long label and tiny value', bg=SHORTS_BG, edge_clean=False)
 		tiers = [(0, 'board shorts'), (4.9, 'board shorts'), (5, 'cargo shorts'), (10, 'boxers'), (15, 'gym shorts'), (20, 'cutoffs'),
 				 (30, 'short shorts'), (40, 'daisy dukes'), (50, 'hot pants'), (60, 'speedos'), (70, 'budgie smugglers'),
 				 (80, 'just a belt'), (90, 'never nudes'), (200, 'never nudes')]
-		self.assertEqual([util.shorts_verdict(p) for p, _ in tiers], [name for _, name in tiers])
+		self.assertEqual([render.shorts_verdict(p) for p, _ in tiers], [name for _, name in tiers])
 
 	def test_bar_graph_has_top_and_bottom(self):
-		self.check_image(util.bar_graph([('AAA', 9)], [('ZZZ', -9)], 'Top and bottom'), 'bar_graph')
+		self.check_image(render.bar_graph([('AAA', 9)], [('ZZZ', -9)], 'Top and bottom'), 'bar_graph')
 
 	def test_flat_and_tiny_price_series(self):
 		flat = pd.DataFrame({'Date': [datetime.date(2026, 1, 1) + datetime.timedelta(days=i) for i in range(10)], 'Close': [5.0] * 10})
-		self.check_image(util.graph(flat, 'Flat', 'USD'), 'flat price')
+		self.check_image(render.graph(flat, 'Flat', 'USD'), 'flat price')
 		two = price_frame(2)
-		self.check_image(util.graph(two, 'Two points', 'USD'), 'two points')
+		self.check_image(render.graph(two, 'Two points', 'USD'), 'two points')
 
 
 class RenderPool(unittest.TestCase):
-	"""The bot draws charts in one native thread (util.render_off_hub), so the gevent hub keeps serving while a chart is drawn."""
+	"""The bot draws charts in one native thread (render.render_off_hub), so the gevent hub keeps serving while a chart is drawn."""
 
 	def setUp(self):
-		util.render_off_hub()
-		self.addCleanup(setattr, util, '_render_pool', None)
+		render.render_off_hub()
+		self.addCleanup(setattr, render, '_render_pool', None)
 
 	def test_charts_asked_for_at_once_all_finish(self):
-		jobs = [gevent.spawn(util.pie_chart, [('a', 1.0), ('b', 2.0)], f't{i}') for i in range(4)]
-		jobs += [gevent.spawn(util.hbar_chart, MIXED, 'T'), gevent.spawn(util.rows_chart, MIXED, 'T')]
+		jobs = [gevent.spawn(render.pie_chart, [('a', 1.0), ('b', 2.0)], f't{i}') for i in range(4)]
+		jobs += [gevent.spawn(render.hbar_chart, MIXED, 'T'), gevent.spawn(render.rows_chart, MIXED, 'T')]
 		gevent.joinall(jobs, timeout=120)
 		self.assertTrue(all(job.successful() for job in jobs), 'a chart did not finish (several native threads deadlock matplotlib)')
 		self.assertTrue(all(decode(job.value).size == (1280, 800) for job in jobs))
@@ -137,7 +137,7 @@ def ticker():
 				gaps.append(now - last)
 				last = now
 		tick = gevent.spawn(ticker)
-		util.shorts_chart([(f'T{i}', 60.0 - 2 * i) for i in range(24)], 15) # about a second of drawing
+		render.shorts_chart([(f'T{i}', 60.0 - 2 * i) for i in range(24)], 15) # about a second of drawing
 		done.append(True)
 		tick.join()
 		self.assertLess(max(gaps), 0.5) # drawn in the hub, the ticker stopped for the whole chart
@@ -151,7 +151,7 @@ def spy(ax, y, width, *a, **k):
 			calls.append({'y': list(y), 'width': list(width), 'color': list(k.get('color', []))})
 			return real(ax, y, width, *a, **k)
 		with mock.patch.object(Axes, 'barh', spy):
-			util.hbar_chart(rows, 'T', **kwargs)
+			render.hbar_chart(rows, 'T', **kwargs)
 		return calls[0]
 
 	def capture_text(self, make):
@@ -175,12 +175,12 @@ def test_first_row_is_on_top_and_spacers_are_skipped(self):
 		calls = []
 		real = Axes.set_ylim
 		with mock.patch.object(Axes, 'set_ylim', lambda ax, *a, **k: (calls.append(a), real(ax, *a, **k))[1]):
-			util.hbar_chart(MIXED, 'T')
+			render.hbar_chart(MIXED, 'T')
 		self.assertTrue(any(len(a) == 2 and a[0] > a[1] for a in calls), 'y axis must be inverted so row 0 is drawn first (top)')
 
 	def test_pixels_follow_the_sign_of_the_data(self):
 		def count(rows):
-			pixels = np.asarray(decode(util.hbar_chart(rows, 'T'))).reshape(-1, 3)
+			pixels = np.asarray(decode(render.hbar_chart(rows, 'T'))).reshape(-1, 3)
 			return (pixels == BLUE).all(axis=1).sum(), (pixels == RED).all(axis=1).sum()
 		blue, red = count([('A', 5), ('B', 3)])
 		self.assertGreater(blue, 1000); self.assertEqual(red, 0)
@@ -190,7 +190,7 @@ def count(rows):
 		self.assertGreater(blue, 1000); self.assertGreater(red, 1000)
 
 	def test_session_tints_the_background(self):
-		corner = lambda session: tuple(np.asarray(decode(util.hbar_chart(MIXED, 'T', session=session)))[2, 2][:3])
+		corner = lambda session: tuple(np.asarray(decode(render.hbar_chart(MIXED, 'T', session=session)))[2, 2][:3])
 		plain, close, mid, pre = corner(None), corner('close'), corner('midsession'), corner('premarket')
 		self.assertEqual(plain, (0x14, 0x18, 0x1f))
 		self.assertEqual(close, plain) # at close stays untinted
@@ -199,16 +199,16 @@ def test_session_tints_the_background(self):
 		self.assertLess(max(abs(int(a) - int(b)) for a, b in zip(pre, plain)), 20) # subtle
 
 	def test_longer_bar_is_drawn_longer(self):
-		pixels = np.asarray(decode(util.hbar_chart([('A', 40), ('B', 10)], 'T')))
+		pixels = np.asarray(decode(render.hbar_chart([('A', 40), ('B', 10)], 'T')))
 		blue_rows = np.where((pixels == BLUE).all(axis=2).any(axis=1))[0]
 		top_band, bottom_band = pixels[blue_rows[0]], pixels[blue_rows[-1]]
 		self.assertGreater((top_band == BLUE).all(axis=1).sum(), 2 * (bottom_band == BLUE).all(axis=1).sum())
 
 	def test_titles_are_drawn(self):
-		texts = self.capture_text(lambda: util.hbar_chart(MIXED, 'My Title', 'My subtitle'))
+		texts = self.capture_text(lambda: render.hbar_chart(MIXED, 'My Title', 'My subtitle'))
 		self.assertIn('My Title', texts)
 		self.assertIn('My subtitle', texts)
-		texts = self.capture_text(lambda: util.graph(price_frame(), 'Apple Inc (AAPL)', 'USD'))
+		texts = self.capture_text(lambda: render.graph(price_frame(), 'Apple Inc (AAPL)', 'USD'))
 		self.assertIn('Apple Inc (AAPL)', texts)
 		self.assertTrue(any('over period' in t for t in texts))
 
@@ -218,7 +218,7 @@ def line_colour(closes):
 			texts = []
 			real = Figure.text
 			with mock.patch.object(Figure, 'text', lambda fig, x, y, s, *a, **k: (texts.append((s, k.get('color'))), real(fig, x, y, s, *a, **k))[1]):
-				util.graph(frame, 'T', 'USD')
+				render.graph(frame, 'T', 'USD')
 			return next(color for s, color in texts if 'over period' in s)
 		self.assertEqual(line_colour([1, 2, 3]), '#34d399') # up: green
 		self.assertEqual(line_colour([3, 2, 1]), '#fb7185') # down: red
@@ -230,11 +230,11 @@ class ReportChartInputs(FinbotCase):
 
 	def run_report(self, call):
 		captured = []
-		real = util.hbar_chart
+		real = render.hbar_chart
 		def spy(rows, title, subtitle='', **kw):
 			captured.append({'rows': list(rows), 'title': title, 'subtitle': subtitle})
 			return real(rows, title, subtitle, **kw)
-		self.patch(util, 'hbar_chart', spy)
+		self.patch(render, 'hbar_chart', spy)
 		call()
 		self.assertTrue(captured, 'no chart was drawn')
 		return captured[0]
@@ -265,11 +265,11 @@ def test_performance_is_sorted_by_percent(self):
 
 	def test_shorts_is_sorted_and_carries_the_threshold(self):
 		captured = []
-		real = util.shorts_chart
+		real = render.shorts_chart
 		def spy(rows, threshold, *a, **k):
 			captured.append((list(rows), threshold))
 			return real(rows, threshold, *a, **k)
-		self.patch(util, 'shorts_chart', spy)
+		self.patch(render, 'shorts_chart', spy)
 		self.command('.shorts')
 		self.assertTrue(captured, 'no chart was drawn')
 		rows, threshold = captured[0]
diff --git a/tests/test_config.py b/tests/test_config.py
index 40f0202b..6bba073e 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -5,7 +5,7 @@
 from lib import config
 from util import check_config, split_env
 import bot, cal
-from lib import cache, reports, sharesight, util, worker, yahoo
+from lib import cache, reports, sharesight, worker, yahoo
 import performance, price, shorts, trades
 
 REPO = Path(config.__file__).resolve().parent.parent
diff --git a/tests/test_datadir.py b/tests/test_datadir.py
index ba8ce810..de6d55fa 100644
--- a/tests/test_datadir.py
+++ b/tests/test_datadir.py
@@ -39,9 +39,9 @@ def run(**env):
 			self.assertEqual(run(STATE_DIRECTORY='/x/y', DATA_DIR=d)[0], d) # DATA_DIR wins
 			self.assertEqual(run(DATA_DIR=d, MPLCONFIGDIR='/custom')[2], '/custom') # explicit value kept
 
-	def test_util_import_sets_mplconfigdir_without_home(self):
+	def test_render_import_sets_mplconfigdir_without_home(self):
 		e = {k: v for k, v in os.environ.items() if k not in ('MPLCONFIGDIR', 'XDG_CONFIG_HOME', 'XDG_CACHE_HOME', 'HOME')}
-		code = "import os; from lib import util; import matplotlib.dates; print(os.environ['MPLCONFIGDIR'])" # util loads matplotlib lazily
+		code = "import os; from lib import render; import matplotlib.dates; print(os.environ['MPLCONFIGDIR'])" # render loads matplotlib lazily
 		with tempfile.TemporaryDirectory() as d:
 			out = subprocess.run([sys.executable, '-c', code], cwd=REPO, env=dict(e, DATA_DIR=d), capture_output=True, text=True, check=True).stdout.strip()
 			self.assertEqual(out.split()[-1], d + '/cache/matplotlib')
diff --git a/tests/test_edges.py b/tests/test_edges.py
index c2d99bfa..5a174530 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -60,7 +60,7 @@ def down(*a, **k):
 		self.assertEqual(typing_when_reported, [[False]]) # its one indicator had stopped when the error line went out
 
 	def test_no_holdings_and_empty_watchlist(self):
-		self.patch(util, 'get_holdings_and_watchlist', lambda: [])
+		self.patch(watchlist, 'with_holdings', lambda: [])
 		for cmd in ('.watchlist', '.price', '.earnings'):
 			with self.subTest(cmd=cmd):
 				self.command(cmd)
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 75f2d05a..d3e368c7 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -588,7 +588,7 @@ def test_slack_without_a_bot_token_gets_text_and_no_chart_is_drawn(self):
 		def refuse(*a, **k):
 			raise AssertionError('a chart was drawn for Slack')
 		for name in ('rows_chart', 'pie_chart', 'bar_graph', 'shorts_chart'):
-			self.patch(util, name, refuse)
+			self.patch(render, name, refuse)
 		self.env(webhooks={**config.env.webhooks, 'slack': 'https://hooks.slack.com/services/x'})
 		for cmd in ('.beta', '.pie Rob', '.buy', '.sell', '.peg', '.price', '.price top', '.price T1 7d', '.performance 7d', '.shorts', '.history T1', '.compare T1 T2'):
 			with self.subTest(cmd=cmd):
@@ -673,7 +673,7 @@ def test_a_long_report_is_split_by_size(self):
 		self.assertTrue(all(size <= 2000 for size in sizes), sizes) # 20 lines a chunk used to make 3,019 characters
 
 	def test_links_are_masked_links(self):
-		self.assertEqual(util.link('https://x.example/a(b)', 'T[1]', 'discord'), '[T\\[1\\]](<https://x.example/a%28b%29>)')
+		self.assertEqual(links.link('https://x.example/a(b)', 'T[1]', 'discord'), '[T\\[1\\]](<https://x.example/a%28b%29>)')
 
 
 class SlackSending(FinbotCase):
diff --git a/tests/test_reports.py b/tests/test_reports.py
index bece5552..b22500d4 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -81,7 +81,7 @@ def test_ratings_list_the_strongest_first(self):
 		fewer than 7 analysts and the stock is left out (T7, T8: the ASX small caps that are always a perfect 1.0)."""
 		ratings = {'T1': ('buy', 1.5, 7), 'T2': ('strong_buy', 1.2, 12), 'T3': ('buy', 1.5, 9), 'T7': ('strong_buy', 1.0, 6),
 				   'T4': ('sell', 4.2, 7), 'T5': ('underperform', 4.6, 10), 'T6': ('sell', 4.2, 8), 'T8': ('strong_buy', 1.0, 2)}
-		self.patch(util, 'get_holdings_and_watchlist', lambda: list(ratings))
+		self.patch(watchlist, 'with_holdings', lambda: list(ratings))
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | dict(zip(('recommend', 'recommend_index', 'recommend_analysts'), ratings[t]))})
 		def order(action):
 			payload, chart = reports.prepare_rating_payload('telegram', action)
@@ -89,7 +89,7 @@ def order(action):
 		self.assertEqual(order('buy'), (['T2', 'T3', 'T1'], ['T2', 'T3', 'T1']))
 		self.assertEqual(order('sell'), (['T5', 'T6', 'T4'], ['T5', 'T6', 'T4']))
 		self.assertIn('(7+ analysts)', reports.prepare_rating_payload('telegram', 'buy')[0][0])
-		self.patch(util, 'get_holdings_and_watchlist', lambda: ['T7', 'T8'])
+		self.patch(watchlist, 'with_holdings', lambda: ['T7', 'T8'])
 		self.assertEqual(reports.prepare_rating_payload('telegram', 'buy'), (['No tracked stocks rated buy by 7+ analysts'], None))
 
 	def test_a_small_cap_quoted_in_won_needs_the_bigger_move_too(self):
@@ -116,7 +116,7 @@ def test_the_small_caps_betas_are_fetched_together(self):
 	def test_forward_pe_lists_need_as_many_analysts_as_buy(self):
 		"""A forward P/E is analysts' estimate: with few of them it says little, as a perfect 1.0 buy from two brokers does."""
 		analysts = {'T1': 7, 'T2': 6, 'T3': 12, 'T4': 2}
-		self.patch(util, 'get_holdings_and_watchlist', lambda: list(analysts))
+		self.patch(watchlist, 'with_holdings', lambda: list(analysts))
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | {'price_to_earnings_forward': -5 if t == 'T3' else 20} for t in ts})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'recommend_analysts': analysts[t]}})
 		def listed(action):
@@ -140,7 +140,7 @@ def fetch(tickers):
 					for t in tickers if t in fx or t in stocks}
 		self.patch(yahoo, 'fetch', fetch)
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: fetch([t]))
-		self.patch(util, 'get_holdings_and_watchlist', lambda: list(stocks))
+		self.patch(watchlist, 'with_holdings', lambda: list(stocks))
 		payload = reports.prepare_marketcap_payload('telegram')
 		self.assertIn('by market cap (USD)', payload[0])
 		self.assertEqual([line.split('Co ')[1].split(' ')[0] for line in payload[1:]], ['U1', 'K1.KS', 'L1.L', 'K2.KS']) # Z1: no rate, left out
@@ -425,7 +425,7 @@ def test_folds_into_other_and_renders(self):
 		payload, chart = reports.prepare_pie('Rob', 'industry', 'telegram')
 		self.assertEqual(len(chart['rows']), 8)
 		self.assertEqual(chart['rows'][-1][0], 'Other')
-		self.assertTrue(util.pie_chart(chart['rows'], chart['title'], chart['subtitle']).read().startswith(b'\x89PNG'))
+		self.assertTrue(render.pie_chart(chart['rows'], chart['title'], chart['subtitle']).read().startswith(b'\x89PNG'))
 
 	def test_unknown_or_missing_portfolio_lists_known(self):
 		for text, first in (('.pie', '.pie: please try again'), ('.pie Nope', 'Nope portfolio not found')):
diff --git a/tests/test_util.py b/tests/test_util.py
index 1ffa8310..75671a68 100644
--- a/tests/test_util.py
+++ b/tests/test_util.py
@@ -1,4 +1,4 @@
-"""Pure helpers in lib/util.py: table-driven pass and fail cases."""
+"""Pure helpers (lib/markets.py, formats.py, links.py) and the cache and state files: table-driven pass and fail cases."""
 import datetime, json, threading, unittest
 from pathlib import Path
 from tests.common import *
@@ -8,14 +8,14 @@ class Parsing(unittest.TestCase):
 	def test_days_from_human_days(self):
 		for arg, days in (('7', 7), ('7d', 7), ('7D', 7), ('2w', 14), ('3m', 90), ('1y', 365), ('0', 0)):
 			with self.subTest(arg=arg):
-				self.assertEqual(util.days_from_human_days(arg), days)
+				self.assertEqual(formats.days_from_human_days(arg), days)
 		today = datetime.date.today()
-		self.assertEqual(util.days_from_human_days('ytd'), (today - today.replace(month=1, day=1)).days)
+		self.assertEqual(formats.days_from_human_days('ytd'), (today - today.replace(month=1, day=1)).days)
 
 	def test_days_from_human_days_rejects_non_periods(self):
 		for arg in ('AAPL', '', 'x5d', '1.5y', 'week'):
 			with self.subTest(arg=arg), self.assertRaises(ValueError):
-				util.days_from_human_days(arg)
+				formats.days_from_human_days(arg)
 
 	def test_transform_to_yahoo(self):
 		for args, expected in ((('AAPL',), 'AAPL'), (('FANG', 'ASX'), 'FANG.AX'), (('ARM', 'NASDAQ'), 'ARM'),
@@ -23,29 +23,29 @@ def test_transform_to_yahoo(self):
 							   (('BRK', 'B'), 'BRK-B'), (('2330', 'TAI'), '2330.TW'), (('VOD', 'LSE'), 'VOD.L'),
 							   (('7203', 'TSE'), '7203.T'), (('SHOP', 'TSX'), 'SHOP.TO'), (('AIR', 'NZX'), 'AIR.NZ'), (('D05', 'SGX'), 'D05.SI')):
 			with self.subTest(args=args):
-				self.assertEqual(util.transform_to_yahoo(*args), expected)
+				self.assertEqual(markets.transform_to_yahoo(*args), expected)
 
 	def test_each_market_has_its_own_flag_and_currency(self): # Sharesight's market codes: TSE is Tokyo, TSX Toronto, SGX Singapore
 		for market, ticker, flag, currency in (('TSE', '7203.T', '🇯🇵', 'JPY'), ('TSX', 'SHOP.TO', '🇨🇦', 'CAD'), ('NZX', 'AIR.NZ', '🇳🇿', 'NZD'),
 											   ('SGX', 'D05.SI', '🇸🇬', 'SGD'), ('CPSE', None, '🇩🇰', 'DKK')):
 			with self.subTest(market=market):
-				self.assertEqual((util.flag_from_market(market), util.currency_from_market(market)), (flag, currency))
+				self.assertEqual((markets.flag_from_market(market), markets.currency_from_market(market)), (flag, currency))
 				if ticker:
-					self.assertEqual(util.flag_from_ticker(ticker), flag)
+					self.assertEqual(markets.flag_from_ticker(ticker), flag)
 
 	def test_make_paragraphs_keeps_every_word(self):
-		self.assertEqual(util.make_paragraphs('Acme makes widgets. It sells them.'), 'Acme makes widgets. It sells them.')
+		self.assertEqual(formats.make_paragraphs('Acme makes widgets. It sells them.'), 'Acme makes widgets. It sells them.')
 		long = ' '.join(['word'] * 22) + ' end. Tail sentence here.'
-		self.assertEqual(util.make_paragraphs(long), ' '.join(['word'] * 22) + ' end.\n\nTail sentence here.')
+		self.assertEqual(formats.make_paragraphs(long), ' '.join(['word'] * 22) + ' end.\n\nTail sentence here.')
 
 	def test_transform_title(self):
-		self.assertEqual(util.transform_title('APPLE INC'), 'Apple')
-		self.assertEqual(util.transform_title('Acme FPO'), 'Acme')
-		self.assertEqual(util.transform_title('Mixed Case Name'), 'Mixed Case Name')
+		self.assertEqual(formats.transform_title('APPLE INC'), 'Apple')
+		self.assertEqual(formats.transform_title('Acme FPO'), 'Acme')
+		self.assertEqual(formats.transform_title('Mixed Case Name'), 'Mixed Case Name')
 		for title, expected in (('Allianz Se', 'Allianz'), ('Foo SE AG', 'Foo'), ('Johnson & Johnson &', 'Johnson & Johnson'),
 								('Black and White and', 'Black and White'), ('Big Holdings Things Ltd', 'Big Things')):
 			with self.subTest(title=title): # a suffix comes off the end only; no run of spaces is left behind
-				self.assertEqual(util.transform_title(title), expected)
+				self.assertEqual(formats.transform_title(title), expected)
 
 
 class Formatting(unittest.TestCase):
@@ -53,39 +53,39 @@ def test_signed_percent(self):
 		for value, decimals, expected in ((12, None, '+12%'), (-3.5, None, '-3.5%'), (0, None, '0%'), (1234, None, '+1,234%'),
 										  (2.345, 1, '+2.3%'), (-0.04, 1, '-0.0%')):
 			with self.subTest(value=value):
-				self.assertEqual(util.signed_percent(value, decimals), expected)
+				self.assertEqual(formats.signed_percent(value, decimals), expected)
 
 	def test_days_english(self):
 		for days, expected in ((None, 'today'), (0, 'today'), (1, 'the past day'), (7, 'the past week')):
 			with self.subTest(days=days):
-				self.assertEqual(util.days_english(days), expected)
+				self.assertEqual(formats.days_english(days), expected)
 
 	def test_date_range_english(self):
 		end = datetime.date(2026, 10, 5)
-		self.assertEqual(util.date_range_english(7, end), '28 Sep - 5 Oct 2026')
-		self.assertEqual(util.date_range_english(30, end), '5 Sep - 5 Oct 2026')
-		self.assertEqual(util.date_range_english(10, datetime.date(2026, 1, 5)), '26 Dec 2025 - 5 Jan 2026') # crosses a year
+		self.assertEqual(formats.date_range_english(7, end), '28 Sep - 5 Oct 2026')
+		self.assertEqual(formats.date_range_english(30, end), '5 Sep - 5 Oct 2026')
+		self.assertEqual(formats.date_range_english(10, datetime.date(2026, 1, 5)), '26 Dec 2025 - 5 Jan 2026') # crosses a year
 		with self.assertRaises(TypeError):
-			util.date_range_english(None, end) # callers must pass a period
+			formats.date_range_english(None, end) # callers must pass a period
 
 	def test_ordinal(self):
 		for number, suffix in ((1, 'st'), (2, 'nd'), (3, 'rd'), (4, 'th'), (11, 'th'), (12, 'th'), (13, 'th'), (21, 'st'), (112, 'th')):
 			with self.subTest(number=number):
-				self.assertEqual(util.ordinal(number), suffix)
+				self.assertEqual(formats.ordinal(number), suffix)
 
 	def test_get_emoji(self):
-		self.assertEqual([util.get_emoji(n) for n in (1, -1, 0)], ['🔼', '🔻', '▪️'])
+		self.assertEqual([formats.get_emoji(n) for n in (1, -1, 0)], ['🔼', '🔻', '▪️'])
 
 	def test_links_per_service(self):
-		telegram_link = util.link('https://example.com/x', 'text', 'telegram')
-		slack_link = util.link('https://example.com/x', 'text', 'slack')
+		telegram_link = links.link('https://example.com/x', 'text', 'telegram')
+		slack_link = links.link('https://example.com/x', 'text', 'slack')
 		self.assertIn('<a href=', telegram_link)
 		self.assertIn('|text', slack_link)
 
 	def test_flags_for_known_and_unknown_markets(self):
-		self.assertTrue(util.flag_from_ticker('BHP.AX'))
-		self.assertTrue(util.flag_from_ticker('AAPL'))
-		self.assertIsInstance(util.flag_from_ticker('XYZ.ZZ'), str) # unknown suffix must not crash
+		self.assertTrue(markets.flag_from_ticker('BHP.AX'))
+		self.assertTrue(markets.flag_from_ticker('AAPL'))
+		self.assertIsInstance(markets.flag_from_ticker('XYZ.ZZ'), str) # unknown suffix must not crash
 
 
 class StateFiles(FinbotCase):
@@ -163,11 +163,11 @@ def test_binary_cache_roundtrip(self):
 
 	def test_holdings_and_watchlist_without_a_watchlist_file(self):
 		# first run: no watchlist exists yet
-		self.patch(util.sharesight, 'tickers', lambda: ['AAPL', 'GOOG', 'GOOGL'])
-		self.patch(util, 'get_holdings_and_watchlist', REAL_GET_HOLDINGS)
-		self.assertEqual(util.get_holdings_and_watchlist(), ['AAPL', 'GOOG']) # GOOGL dropped when GOOG is present
+		self.patch(sharesight, 'tickers', lambda: ['AAPL', 'GOOG', 'GOOGL'])
+		self.patch(watchlist, 'with_holdings', REAL_GET_HOLDINGS)
+		self.assertEqual(watchlist.with_holdings(), ['AAPL', 'GOOG']) # GOOGL dropped when GOOG is present
 		state.save('finbot_watchlist.json', ['MSFT'])
-		self.assertEqual(util.get_holdings_and_watchlist(), ['AAPL', 'GOOG', 'MSFT'])
+		self.assertEqual(watchlist.with_holdings(), ['AAPL', 'GOOG', 'MSFT'])
 
 
 if __name__ == '__main__':
diff --git a/trades.py b/trades.py
index a57ff93d..04cd4390 100755
--- a/trades.py
+++ b/trades.py
@@ -8,7 +8,9 @@
 from lib import words
 from lib import sharesight
 from lib import webhook
-from lib import util
+from lib import formats
+from lib import links
+from lib import markets
 from lib import yahoo
 
 def lambda_handler(chat_id=None, days=None, service=None, user='', portfolio_select=None, interactive=False):
@@ -32,7 +34,7 @@ def prepare_trade_payload(service, trades):
 			return []
 		market_data = {}
 		if config.prefs.trades_use_yahoo:
-			tickers = {util.transform_to_yahoo(trade['symbol'], trade['market']) for trade in wanted} # one yahoo.fetch for all
+			tickers = {markets.transform_to_yahoo(trade['symbol'], trade['market']) for trade in wanted} # one yahoo.fetch for all
 			try:
 				print("Fetching tickers from Yahoo: ", tickers, file=sys.stderr)
 				market_data = yahoo.fetch(tickers)
@@ -50,7 +52,7 @@ def prepare_trade_payload(service, trades):
 			price = float(trade['price'])
 			currency = trade['brokerage_currency_code']
 			if config.prefs.trades_use_yahoo:
-				ticker = util.transform_to_yahoo(symbol, market)
+				ticker = markets.transform_to_yahoo(symbol, market)
 				try:
 					market = market_data[ticker]['profile_exchange']
 				except Exception as e:
@@ -62,15 +64,15 @@ def prepare_trade_payload(service, trades):
 			newtrades.add(trade_id) # recorded by the main body once the report is out
 			dates.add(dt_date)
 
-			flag = util.flag_from_market(market)
+			flag = markets.flag_from_market(market)
 			# lookup currency or fall back to less reliable sharesight currency
-			currency_temp = util.currency_from_market(market)
+			currency_temp = markets.currency_from_market(market)
 			if currency_temp:
 				currency = currency_temp
 			sharesight_url = "https://portfolio.sharesight.com/holdings/"
 			trade_url = sharesight_url + holding_id + '/dashboard/transactions'
-			trade_link = util.link(trade_url, action, service)
-			holding_link = util.finance_link(symbol, market, service)
+			trade_link = links.link(trade_url, action, service)
+			holding_link = links.finance_link(symbol, market, service)
 			payload_staging.append([dt_date, trade_id, emoji, portfolio_name, trade_link, currency, f'{value}', holding_link, flag])
 
 		payload = []
@@ -94,7 +96,7 @@ def prepare_trade_payload(service, trades):
 	for service, url, chat in webhook.destinations(interactive, service, chat_id):
 		payload = prepare_trade_payload(service, trades)
 		if interactive and not payload: # no trades, or none of them a buy or a sell
-			payload = [f"{user} No trades {util.days_english(days, 'in the past ')}. {random.choice(words.noTradesVerb)}"]
+			payload = [f"{user} No trades {formats.days_english(days, 'in the past ')}. {random.choice(words.noTradesVerb)}"]
 		webhook.send(service, url, chat, payload)
 
 	if newtrades and not interactive:

From 3da57c0b1e083680d9c754dd9cc79d72689d30f6 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 00:09:09 +0000
Subject: [PATCH 13/81] Reports move into a lib/reports/ package; the button
 helpers into lib/buttons.py

The ten scheduled scripts move from the top level into lib/reports/ (renamed, so their history follows), and
lib/reports.py (622 lines) splits by subject beside them: about (.help, .jobs), history (.history, .compare), holdings
(.holdings, .who, .pie, .watchlist), profile (.SYMBOL) and rankings (.marketcap, .pe, .fpe, .peg, .beta, .buy, .sell).

The period-button helpers (PERIODS, keyboard, the callback refs) leave lib/charts.py for lib/buttons.py, which the
reports use; charts.py keeps the rebuild behind a button press and imports the reports at the top. That ends the second
import cycle: charts imported reports, which imported jobs, which imported every script, two of which imported charts.

lib/jobrun.py takes a module path (lib.reports.price) and names the run by its last part. The next commit puts a
script back at the top level for each report, for schedule lines and runs by hand.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md                                    |  18 +-
 bot.py                                       |   2 +-
 lib/buttons.py                               |  44 ++
 lib/charts.py                                |  77 +--
 lib/jobrun.py                                |  14 +-
 lib/jobs.py                                  |   2 +-
 lib/reports.py                               | 625 -------------------
 lib/reports/__init__.py                      |  14 +
 lib/reports/about.py                         |  87 +++
 cal.py => lib/reports/cal.py                 |   6 +-
 cleanup.py => lib/reports/cleanup.py         |   7 +-
 lib/reports/history.py                       | 104 +++
 lib/reports/holdings.py                      | 102 +++
 milestone.py => lib/reports/milestone.py     |   7 +-
 performance.py => lib/reports/performance.py |  12 +-
 podcasts.py => lib/reports/podcasts.py       |   7 +-
 price.py => lib/reports/price.py             |  17 +-
 lib/reports/profile.py                       | 214 +++++++
 lib/reports/rankings.py                      | 164 +++++
 rating.py => lib/reports/rating.py           |   7 +-
 reminder.py => lib/reports/reminder.py       |   7 +-
 shorts.py => lib/reports/shorts.py           |   7 +-
 trades.py => lib/reports/trades.py           |   6 +-
 lib/worker.py                                |  50 +-
 tests/common.py                              |   8 +-
 tests/test_config.py                         |   8 +-
 tests/test_edges.py                          |   2 +-
 tests/test_finbot.py                         |   2 +-
 tests/test_jobs.py                           |  12 +-
 tests/test_reports.py                        |  61 +-
 30 files changed, 845 insertions(+), 848 deletions(-)
 create mode 100644 lib/buttons.py
 delete mode 100755 lib/reports.py
 create mode 100644 lib/reports/__init__.py
 create mode 100644 lib/reports/about.py
 rename cal.py => lib/reports/cal.py (97%)
 mode change 100755 => 100644
 rename cleanup.py => lib/reports/cleanup.py (90%)
 mode change 100755 => 100644
 create mode 100644 lib/reports/history.py
 create mode 100644 lib/reports/holdings.py
 rename milestone.py => lib/reports/milestone.py (96%)
 mode change 100755 => 100644
 rename performance.py => lib/reports/performance.py (93%)
 mode change 100755 => 100644
 rename podcasts.py => lib/reports/podcasts.py (97%)
 mode change 100755 => 100644
 rename price.py => lib/reports/price.py (94%)
 mode change 100755 => 100644
 create mode 100644 lib/reports/profile.py
 create mode 100644 lib/reports/rankings.py
 rename rating.py => lib/reports/rating.py (95%)
 mode change 100755 => 100644
 rename reminder.py => lib/reports/reminder.py (96%)
 mode change 100755 => 100644
 rename shorts.py => lib/reports/shorts.py (95%)
 mode change 100755 => 100644
 rename trades.py => lib/reports/trades.py (97%)
 mode change 100755 => 100644

diff --git a/CLAUDE.md b/CLAUDE.md
index c59183c3..b6fddb0d 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -27,13 +27,15 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - `bot.py`: WSGI entry; routes /telegram and /slack, authenticates, hands each message to `lib/worker.py` in a thread; SIGHUP reload, SIGUSR1
   cache flush.
 - `lib/worker.py`: `COMMANDS` (regex → `on_*` handler, first match wins, `.SYMBOL` last); `Request`; the typing indicator.
-- `lib/reports.py`: `prepare_*` builders for the chat-only reports (lines, plus a chart spec).
-- `price.py performance.py trades.py shorts.py cal.py rating.py milestone.py reminder.py podcasts.py cleanup.py`: reports that also
-  run on a schedule; each has `lambda_handler()` and `job(argv)` (its schedule-line argument parser).
+- `lib/reports/`: what finbot says. Scheduled (each has `lambda_handler()` and `job(argv)`, its schedule-line parser): `price
+  performance trades shorts cal rating milestone reminder podcasts cleanup`; the top-level `price.py` and the rest are shims
+  that run them. Chat-only (`prepare_*`): `about` (.help .jobs), `history` (.history .compare), `holdings` (.holdings .who .pie
+  .watchlist), `profile` (.SYMBOL), `rankings` (.marketcap .pe .fpe .peg .beta .buy .sell).
 - `lib/webhook.py`: sending (chunking, photos, Slack upload), markup, escaping, error reporting. `lib/telegram.py`: Bot API calls.
-- `lib/charts.py`: Telegram period buttons and the rebuild behind them. `lib/markets.py`: market tables, flags, ticker
-  translation. `lib/formats.py`: names, units, dates, percents in words. `lib/links.py`: links. `lib/render.py`: every chart
-  (matplotlib).
+- `lib/buttons.py`: period buttons under a chart. `lib/charts.py`: a button press rebuilds its chart; the Telegram command
+  keyboard.
+- `lib/markets.py`: market tables, flags, ticker translation. `lib/formats.py`: names, units, dates, percents in words.
+  `lib/links.py`: links. `lib/render.py`: every chart (matplotlib).
 - `lib/yahoo.py lib/sharesight.py lib/shortman.py lib/simplywallst.py`: data sources, cached in `DATA_DIR/cache`.
   `sharesight.portfolios(name=None)` gives `Portfolio` objects: `holdings()`, `tickers()`, `performance(days)`,
   `trades(days)`.
@@ -45,8 +47,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   `('photo', size, caption)`.
 - Which test file: commands and scheduled reports `test_reports`; failures, bad input, WSGI and Slack events `test_edges`;
   Yahoo/Sharesight parsing, sending, Telegram API `test_providers`; scheduler `test_jobs`; config `test_config`; charts
-  `test_charts` (refresh goldens with `UPDATE_GOLDEN=1`); pure helpers `test_util`; data dir `test_datadir`; chunking, git_version
-  `test_finbot`.
+  `test_charts` (refresh goldens with `UPDATE_GOLDEN=1`); markets, formats, links, cache and state files `test_util`; data dir
+  `test_datadir`; chunking, git_version `test_finbot`.
 
 ## Conventions
 - Read settings as `config.env.<name>` / `config.prefs.<name>` where they are used, never into a default or module constant (a
diff --git a/bot.py b/bot.py
index 82ec25c6..8d1f851e 100755
--- a/bot.py
+++ b/bot.py
@@ -22,7 +22,7 @@
 from lib import webhook
 from lib import yahoo
 from lib import render
-import cleanup
+from lib.reports import cleanup
 
 botName = None # the Telegram bot's @username, from telegram.register() at start
 AUTHENTICATED = 'finbot.authenticated' # set in a request's WSGI environ once it has passed its service's check
diff --git a/lib/buttons.py b/lib/buttons.py
new file mode 100644
index 00000000..752a0081
--- /dev/null
+++ b/lib/buttons.py
@@ -0,0 +1,44 @@
+"""Period buttons (Week | Month | Quarter | Max) under a chart: their markup, and the callback data that names the tickers."""
+import hashlib
+
+from lib import state
+
+PERIODS = {'d': ('1D', 1), 'w': ('7D', 7), 'm': ('1M', 30), 'q': ('3M', 90), 'y': ('1Y', 365), 'x': ('Max', None)}
+BUTTONS = {'h': 'wmqyx', 'p': 'wmqyx', 'c': 'wmqyx', 'f': 'wmqyx', 'l': 'dwmqyx'} # chart kind -> buttons offered (h=history, p=price chart, c=compare, f=performance, l=price list)
+MAX_DAYS = 3650 # what Max means for the Sharesight-based reports
+MAX_CALLBACK_BYTES = 64 # Telegram limit on callback_data
+REF_FILE = 'finbot_chart_buttons.json'
+
+def period_for_days(days, default=None):
+	"""Button key for a day count, if it is one of the button periods; `default` when it is not (or days is None)."""
+	for key, (_, d) in PERIODS.items():
+		if d and d == days:
+			return key
+	return default
+
+def _ref(tickers):
+	"""Tickers as callback data: inline if it fits, else a short id persisted to disk."""
+	inline = ','.join(tickers)
+	if len(('c|h|w|' + inline).encode()) <= MAX_CALLBACK_BYTES:
+		return inline
+	key = hashlib.sha1(inline.encode()).hexdigest()[:10]
+	with state.edit(REF_FILE, {}) as refs:
+		refs[key] = tickers
+	return '#' + key
+
+def resolve(ref):
+	if ref.startswith('#'):
+		tickers = state.load(REF_FILE, {}).get(ref[1:])
+		if not tickers:
+			raise KeyError("these buttons have expired, please re-run the command")
+		return tickers
+	return ref.split(',')
+
+def keyboard(kind, tickers, active=None):
+	"""Telegram inline keyboard; the active period is marked."""
+	ref = _ref(tickers)
+	row = []
+	for key in BUTTONS[kind]:
+		label = PERIODS[key][0]
+		row.append({'text': ('● ' + label) if key == active else label, 'callback_data': f"c|{kind}|{key}|{ref}"})
+	return {'inline_keyboard': [row]}
diff --git a/lib/charts.py b/lib/charts.py
index d6d782be..c195b604 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -1,45 +1,15 @@
-"""Period buttons (Week | Month | Quarter | Max) for Telegram chart messages, and the rebuild behind them."""
+"""A period button pressed: the chart rebuilt for that period. And the Telegram command keyboard, kept current in every DM."""
 import hashlib
 import json
 import sys
 
+from lib import buttons
 from lib import formats
 from lib import links
 from lib import state
 from lib import webhook
 from lib import yahoo
-from lib import reports
-
-PERIODS = {'d': ('1D', 1), 'w': ('7D', 7), 'm': ('1M', 30), 'q': ('3M', 90), 'y': ('1Y', 365), 'x': ('Max', None)}
-BUTTONS = {'h': 'wmqyx', 'p': 'wmqyx', 'c': 'wmqyx', 'f': 'wmqyx', 'l': 'dwmqyx'} # chart kind -> buttons offered (h=history, p=price chart, c=compare, f=performance, l=price list)
-MAX_DAYS = 3650 # what Max means for the Sharesight-based reports
-MAX_CALLBACK_BYTES = 64 # Telegram limit on callback_data
-REF_FILE = 'finbot_chart_buttons.json'
-
-def period_for_days(days, default=None):
-	"""Button key for a day count, if it is one of the button periods; `default` when it is not (or days is None)."""
-	for key, (_, d) in PERIODS.items():
-		if d and d == days:
-			return key
-	return default
-
-def _ref(tickers):
-	"""Tickers as callback data: inline if it fits, else a short id persisted to disk."""
-	inline = ','.join(tickers)
-	if len(('c|h|w|' + inline).encode()) <= MAX_CALLBACK_BYTES:
-		return inline
-	key = hashlib.sha1(inline.encode()).hexdigest()[:10]
-	with state.edit(REF_FILE, {}) as refs:
-		refs[key] = tickers
-	return '#' + key
-
-def resolve(ref):
-	if ref.startswith('#'):
-		tickers = state.load(REF_FILE, {}).get(ref[1:])
-		if not tickers:
-			raise KeyError("these buttons have expired, please re-run the command")
-		return tickers
-	return ref.split(',')
+from lib.reports import history, performance, price
 
 COMMAND_BUTTONS = sorted(['.beta', '.buy & sell', '.dividend', '.earnings', '.forward pe', '.marketcap', '.pe', '.peg', '.performance', '.premarket',
 	'.price', '.session', '.shorts', '.trades', '.watchlist']) # alphabetical, row by row; '.buy & sell' sends both rating reports
@@ -50,6 +20,7 @@ def command_keyboard(per_row=3):
 	return {'keyboard': rows, 'resize_keyboard': True, 'is_persistent': True}
 
 KEYBOARD_FILE = 'finbot_keyboards.json' # chat id -> version of the command keyboard that chat last received
+
 def keyboard_version():
 	return hashlib.sha1(json.dumps(command_keyboard(), sort_keys=True).encode()).hexdigest()[:8]
 
@@ -81,40 +52,12 @@ def refresh_keyboards():
 			except Exception as e:
 				print("keyboard refresh failed for", chat_id, e, file=sys.stderr)
 
-def keyboard(kind, tickers, active=None):
-	"""Telegram inline keyboard; the active period is marked."""
-	ref = _ref(tickers)
-	row = []
-	for key in BUTTONS[kind]:
-		label = PERIODS[key][0]
-		row.append({'text': ('● ' + label) if key == active else label, 'callback_data': f"c|{kind}|{key}|{ref}"})
-	return {'inline_keyboard': [row]}
-
-def history_payload(ticker, service, highlight=None, graph=False):
-	"""Caption lines for .history: performance table, with the selected interval in bold. Returns (lines, percent table, image):
-	the full history's chart with graph=True (if the graph preference is on), else None."""
-	market_data = yahoo.fetch_detail(ticker, 600)
-	title = market_data.get(ticker, {}).get('profile_title', '')
-	ticker_link = links.finance_link(ticker, market_data.get(ticker, {}).get('profile_exchange', ''), service, days=1825, brief=False)
-	if ticker not in market_data or 'percent_change' not in market_data[ticker]:
-		return [f".history: no data found for ticker {ticker}"], None, None
-	table, image = yahoo.price_history(ticker) if graph else yahoo.price_history(ticker, graph=False)
-	lines = [webhook.bold(f"{title} ({ticker_link}) performance history", service)]
-	for interval in ('Max', '10Y', '5Y', '3Y', '1Y', 'YTD', '6M', '3M', '1M', '7D', '1D'):
-		if interval in table:
-			percent = table[interval]
-			label = webhook.bold(interval + ':', service)
-			if interval == highlight:
-				label = '● ' + label
-			lines.append(f"{formats.get_emoji(percent)} {label} {formats.signed_percent(percent)}")
-	return lines, table, image
-
 def build(kind, tickers, period, service='telegram'):
 	"""(caption, image) for a chart kind at a button period; period None = the command's default view."""
-	days = PERIODS[period][1] if period else None
+	days = buttons.PERIODS[period][1] if period else None
 	if kind == 'h':
 		ticker = tickers[0]
-		lines, table, image = history_payload(ticker, service, PERIODS[period or 'x'][0], graph=not days) # default view is the full history
+		lines, table, image = history.history_payload(ticker, service, buttons.PERIODS[period or 'x'][0], graph=not days) # default view is the full history
 		if table is None:
 			raise RuntimeError(lines[0])
 		if days:
@@ -131,19 +74,17 @@ def build(kind, tickers, period, service='telegram'):
 		return f"{formats.get_emoji(percent)} {market_data[ticker]['profile_title']} ({link}) {formats.signed_percent(round(percent))}", image
 	if kind == 'c':
 		args = list(tickers) + ([f'{days}d'] if days else ['max'])
-		caption, image, *_ = reports.prepare_compare(service, args)
+		caption, image, *_ = history.prepare_compare(service, args)
 		if not image:
 			raise RuntimeError(caption[0])
 		return '\n'.join(caption), image
 	if kind == 'f': # performance: tickers = [portfolio name or '']
-		import performance
-		payload, image = performance.report(service, days or MAX_DAYS, tickers[0] or None, interactive=True)
+		payload, image = performance.report(service, days or buttons.MAX_DAYS, tickers[0] or None, interactive=True)
 		if not image:
 			raise RuntimeError(payload[0] if payload else 'no performance data')
 		return '\n'.join(payload), image
 	if kind == 'l': # price list: tickers = [threshold, top]
-		import price
-		span = {'interday': True} if period == 'd' else {'days': days or MAX_DAYS} # 'd' is today's moves: the command's default view, not a 1-day lookback
+		span = {'interday': True} if period == 'd' else {'days': days or buttons.MAX_DAYS} # 'd' is today's moves: the command's default view, not a 1-day lookback
 		payload, image = price.report(service, threshold=float(tickers[0]), interactive=True, top=int(tickers[1]) or None, **span)
 		if not image:
 			raise RuntimeError(payload[0] if payload else 'no price data')
diff --git a/lib/jobrun.py b/lib/jobrun.py
index c3bc0d29..1f6b7c7d 100644
--- a/lib/jobrun.py
+++ b/lib/jobrun.py
@@ -1,4 +1,4 @@
-"""Runs one scheduled command in its own process: `python -m lib.jobrun <script> '<json kwargs>'`.
+"""Runs one scheduled command in its own process: `python -m lib.jobrun lib.reports.<name> '<json kwargs>'`.
 
 The bot starts this for every scheduled or manual run, so a job's CPU-heavy work (charts, pandas) happens off the bot's event
 loop and cannot hold up incoming webhooks. Failures are reported to the channel and give the exit status REPORTED."""
@@ -14,14 +14,14 @@
 
 
 def main(argv):
-	if len(argv) != 2 or not re.fullmatch(r'[a-z]+', argv[0]):
-		print('usage: python -m lib.jobrun <script> <json kwargs>', file=sys.stderr)
+	if len(argv) != 2 or not re.fullmatch(r'[a-z]+(\.[a-z]+)*', argv[0]):
+		print('usage: python -m lib.jobrun <module> <json kwargs>', file=sys.stderr)
 		return 2
-	name = argv[0]
-	def run(): # the import too: a script that cannot even be imported is reported like any other failure
-		importlib.import_module(name).lambda_handler(**json.loads(argv[1]))
+	module = argv[0]
+	def run(): # the import too: a report that cannot even be imported is reported like any other failure
+		importlib.import_module(module).lambda_handler(**json.loads(argv[1]))
 		return True
-	return 0 if webhook.guarded(f'{name}.py', run, exit=False) else REPORTED
+	return 0 if webhook.guarded(module.rsplit('.', 1)[-1] + '.py', run, exit=False) else REPORTED
 
 
 if __name__ == '__main__':
diff --git a/lib/jobs.py b/lib/jobs.py
index 86c6bb0f..80cb7a3c 100644
--- a/lib/jobs.py
+++ b/lib/jobs.py
@@ -24,7 +24,7 @@
 from lib import state
 from lib import jobrun
 from lib import webhook
-import cal, cleanup, milestone, performance, podcasts, price, rating, reminder, shorts, trades
+from lib.reports import cal, cleanup, milestone, performance, podcasts, price, rating, reminder, shorts, trades
 
 SCRIPTS = {'cal': cal, 'cleanup': cleanup, 'milestone': milestone, 'performance': performance, 'podcasts': podcasts, 'price': price,
 	'rating': rating, 'reminder': reminder, 'shorts': shorts, 'trades': trades}
diff --git a/lib/reports.py b/lib/reports.py
deleted file mode 100755
index 5d4d3b71..00000000
--- a/lib/reports.py
+++ /dev/null
@@ -1,625 +0,0 @@
-from itertools import pairwise
-import hashlib, re
-import datetime
-import sys
-from lib import config
-from lib import cache
-from lib import watchlist
-from lib import sharesight
-from lib import formats
-from lib import links
-from lib import markets
-from lib import render
-from lib import webhook
-from lib import yahoo
-from lib import simplywallst
-from itertools import groupby
-
-def last_col(line):
-	"""Sort key: the number a report line ends with; a line without one sorts first."""
-	try:
-		return float(line.split()[-1])
-	except ValueError:
-		return float('-inf')
-
-def holding_line(holding, service, text=None):
-	"""'🇦🇺 Title (CODE)' for a Sharesight holding, CODE (or `text`) linking to the holding on Sharesight."""
-	link = links.link(f"https://portfolio.sharesight.com/holdings/{holding['holding_id']}/dashboard", holding['code'] if text is None else text, service)
-	return f"{markets.flag_from_market(holding['market_code'])} {formats.transform_title(holding['name'])} ({link})"
-
-def doDelta(inputList):
-	deltaString = ''
-	inputListFixed = inputList.copy()
-	# replace any NoneType to allow delta calculation
-	for idx, absolute in enumerate(inputList):
-		if absolute is None:
-			if idx == 0:
-				# get the next value that's not None
-				inputListFixed[idx] = next((x for x in inputList if x is not None), 0)
-			else:
-				# get the previous value
-				inputListFixed[idx] = inputListFixed[idx-1]
-	deltaList = [y-x for (x,y) in pairwise(inputListFixed)]
-	for idx, delta in enumerate(deltaList):
-		absolute = inputList[idx+1]
-		if absolute is None or (idx == 0 and inputList[0] is None):
-			deltaString = deltaString + '❌'
-		elif delta < 0 and absolute < 0:
-			deltaString = deltaString + '🔻'
-		elif delta < 0 and absolute >= 0:
-			deltaString = deltaString + '🔽'
-		elif delta > 0 and absolute < 0:
-			deltaString = deltaString + '🔺'
-		elif delta > 0 and absolute >= 0:
-			deltaString = deltaString + '🔼'
-		else:
-			deltaString = deltaString + '▪️'
-	# fallback if input has missing elements
-	missingfromstart = 3 - len(deltaString) # desired length hard coded
-	deltaString = ("❌" * missingfromstart) + deltaString
-	return deltaString
-
-def prepare_watchlist(service, user, action=None, ticker=None):
-	"""The .watchlist reply: the companies watched, or one line on adding ('add') or removing ('delete') a ticker."""
-	if action:
-		return [watchlist_change(service, user, action, markets.transform_to_yahoo(ticker))]
-	watched = watchlist.tickers()
-	market_data = yahoo.fetch(watched) if watched else {}
-	lines = sorted((f"{markets.flag_from_ticker(t)} {quote['profile_title']} ({links.finance_link(t, quote.get('profile_exchange'), service)})"
-		for t, quote in market_data.items()), key=lambda line: re.findall('[A-Z].*', line)) # by name, the flag aside
-	if not lines:
-		return ['Watchlist is empty. Try ".watchlist add SYMBOL" to create it']
-	return [f"Hi {user}, I'm currently tracking:"] + lines
-
-def watchlist_change(service, user, action, asked):
-	"""The reply to adding or removing `asked` (a Yahoo ticker)."""
-	def described(ticker, quote): # "Full Name (TICKER)", bold
-		quote = quote or {}
-		link = links.finance_link(ticker, quote.get('profile_exchange') or ticker, service)
-		return webhook.bold(f"{quote['profile_title']} ({link})" if quote.get('profile_title') else link, service)
-	if action == 'delete':
-		if asked not in watchlist.tickers():
-			return f"Beep Boop. I could not find {webhook.bold(asked, service)} to remove it"
-		quote = yahoo.fetch([asked]).get(asked)
-		watchlist.remove(asked)
-		return f"Ok {user}, I deleted {described(asked, quote)}"
-	found, quote = watchlist.find(asked)
-	if not quote:
-		return f"Beep Boop. I could not find {webhook.bold(asked, service)} to add it"
-	added = watchlist.add(found)
-	if found != asked:
-		done = "so I added" if added else "and I'm already tracking"
-		return f"Beep Boop. I could not find {webhook.bold(asked, service)} {done} {described(found, quote)}"
-	return f"Ok {user}, I added {described(found, quote)}" if added else f"{user}, I'm already tracking {described(found, quote)}"
-
-def prepare_compare(service, args):
-	"""Compare 2-6 tickers as rebased % change. A trailing time period (3m/1y/90d/ytd) is optional;
-	without one, the span is the earliest date covered by every ticker. Returns (caption lines, image, tickers, period days)."""
-	import pandas as pd
-	usage = ['Usage: .compare SYMBOL SYMBOL [SYMBOL...up to 6] [period|max]  (default: 10 years)']
-	period_days = None
-	full_history = False
-	args = list(args)
-	if args:
-		if args[-1].lower() in ('max', 'all'):
-			full_history = True
-			args.pop()
-		else:
-			try:
-				period_days = formats.days_from_human_days(args[-1])
-				args.pop()
-			except ValueError:
-				pass
-	tickers = list(dict.fromkeys(markets.transform_to_yahoo(a.upper()) for a in args)) # de-duplicated, order kept
-	if not 2 <= len(tickers) <= 6:
-		return usage, None, tickers, period_days
-	spec = period_days or ('max' if full_history else 'default')
-	cache_key = hashlib.sha1(f"{service}|{','.join(tickers)}|{spec}".encode()).hexdigest()[:16]
-	image_cache_file, caption_cache_file = f"finbot_compare_{cache_key}.png", f"finbot_compare_{cache_key}.json"
-	if config.prefs.cache:
-		cached_image = cache.read_bytes(image_cache_file)
-		cached_caption = cache.read(caption_cache_file)
-		if cached_image and cached_caption:
-			return cached_caption, cached_image, tickers, period_days
-	names, exchanges, closes = {}, {}, {}
-	for ticker, history in zip(tickers, yahoo.fetch_each(yahoo.price_series, tickers)): # downloaded at the same time
-		if isinstance(history, Exception):
-			raise history
-		names[ticker], exchanges[ticker], closes[ticker] = history
-	limiting = max(closes, key=lambda t: closes[t].index[0]) # ticker with the shortest history
-	common_start = closes[limiting].index[0]
-	start = common_start
-	note = ''
-	if not period_days and not full_history: # default window; ask for a longer period (or max) to go further
-		default_start = pd.Timestamp(datetime.datetime.now().date() - datetime.timedelta(days=3655))
-		start = max(common_start, default_start)
-	if period_days:
-		requested = pd.Timestamp(datetime.datetime.now().date() - datetime.timedelta(days=period_days))
-		if requested >= common_start:
-			start = requested
-		else:
-			note = f" (limited by {limiting} history)"
-	cropped = []
-	for ticker in tickers:
-		series = closes[ticker]
-		base = series[series.index <= start]
-		base_value = base.iloc[-1] if len(base) else series.iloc[0] # last close on/before the start date
-		series = pd.concat([pd.Series([base_value], index=[start]), series[series.index > start]])
-		cropped.append((ticker, series))
-	if any(len(series) < 2 for _, series in cropped):
-		raise RuntimeError("not enough price history in that period to compare")
-	span_days = (datetime.datetime.now().date() - start.date()).days
-	period_label = formats.days_english(period_days) if period_days and not note else f"{span_days} days"
-	by_performance = sorted(cropped, key=lambda item: item[1].iloc[-1] / item[1].iloc[0], reverse=True)
-	title = ' vs '.join(t.split('.')[0] for t, _ in by_performance) # best first; line colours stay tied to input order
-	subtitle = f"% change over {period_label}, since {start:%d %b %Y}" + note
-	image = render.compare_graph([(t.split('.')[0], series) for t, series in cropped], title, subtitle)
-	image.seek(0)
-	caption = [webhook.bold(f"{period_label}, since {start:%d %b %Y}", service) + note]
-	results = sorted(((series.iloc[-1] / series.iloc[0] - 1) * 100, t) for t, series in cropped)
-	for pct, t in reversed(results):
-		link = links.finance_link(t, exchanges[t], service, days=span_days, brief=False)
-		caption.append(f"{formats.get_emoji(pct)} {names[t]} ({link}): {pct:+,.1f}%")
-	if config.prefs.cache:
-		cache.write_bytes(image_cache_file, image)
-		cache.write(caption_cache_file, caption)
-		image.seek(0)
-	return caption, image, tickers, period_days
-
-def prepare_jobs_payload(service, args):
-	"""`.jobs` lists the scheduled commands, each once; `.jobs run <number|command>` runs one now, as cron would.
-	Returns (payload, reply_markup): Telegram gets a button per command."""
-	from lib import jobs # here: jobs imports every script, and some scripts import this module
-	scheduler = jobs.current
-	if scheduler is None:
-		return ['Scheduled jobs are off (scheduler=False)'], None
-	args = (args or '').strip()
-	if not args:
-		commands = scheduler.commands()
-		if not commands:
-			return ['No scheduled jobs'], None
-		payload = [webhook.bold('Scheduled commands:', service)]
-		for number, (label, _, _) in enumerate(commands, 1):
-			payload.append(f"{number}. {jobs.display(label)}" + (' (running)' if label in scheduler.busy else ''))
-		payload.append("\nRun one with .jobs run NUMBER" + (' or tap it' if service == 'telegram' else ''))
-		markup = None
-		if service == 'telegram':
-			buttons = [{'text': f"{number}. {jobs.display(label)}"[:30], 'callback_data': f"j|{number}|{jobs.checksum(label)}"} for number, (label, _, _) in enumerate(commands, 1)]
-			markup = {'inline_keyboard': [buttons[i:i + 2] for i in range(0, len(buttons), 2)]}
-		return payload, markup
-	m = re.fullmatch(r'run(?:\s+(.+))?', args, re.IGNORECASE)
-	if not m or not m.group(1):
-		return ['Usage: .jobs [run NUMBER|COMMAND]'], None
-	return [run_job_line(scheduler, m.group(1))], None
-
-def run_job_line(scheduler, text):
-	"""Run a scheduled command now and say how it went in one line."""
-	from lib import jobs # here: jobs imports every script, and some scripts import this module
-	try:
-		return job_line(scheduler, *scheduler.run_now(text))
-	except jobs.ScheduleError as e:
-		return str(e)
-
-def job_line(scheduler, label, outcome):
-	from lib import jobs # here: jobs imports every script, and some scripts import this module
-	number = [command[0] for command in scheduler.commands()].index(label) + 1
-	return f"{number}. {jobs.display(label)}: {outcome}"
-
-def prepare_help(service, botName):
-	payload = []
-	payload.append(webhook.bold("Tracked securities:", service))
-	payload.append(".holdings")
-	payload.append(".pie PORTFOLIO [sector]")
-	payload.append(".watchlist [add|del SYMBOL]")
-	payload.append(".who SYMBOL")
-
-	payload.append(webhook.bold("\nCompany info:", service))
-	payload.append('.SYMBOL')
-	payload.append(".dividend [period|SYMBOL]")
-	payload.append(".earnings [period|SYMBOL]")
-	payload.append(".marketcap [SYMBOL|bottom|top]")
-
-	payload.append(webhook.bold("\nPrice:", service))
-	payload.append(".beta")
-	payload.append(".compare SYMBOL SYMBOL [...] [period|max]")
-	payload.append(".history SYMBOL")
-	payload.append(".performance [period] [portfolio]")
-	payload.append(".price [percent|SYMBOL] [period]")
-	payload.append(".session [percent|SYMBOL]")
-	payload.append(".premarket [percent|SYMBOL]")
-
-	payload.append(webhook.bold("\nRisk & Value:", service))
-	payload.append(".buy")
-	payload.append(".sell")
-	payload.append(".pe [SYMBOL|top|bottom]")
-	payload.append(".forwardpe [SYMBOL|top|bottom|negative]")
-	payload.append(".peg [SYMBOL|top|bottom|negative]")
-	payload.append(".shorts [percent|SYMBOL]")
-
-	payload.append(webhook.bold("\nTrades:", service))
-	payload.append(".trades [period] [portfolio]")
-
-	payload.append(webhook.bold("\nScheduled jobs:", service))
-	payload.append(".jobs [run NUMBER|COMMAND]")
-
-	payload.append(webhook.bold("\nAlternative syntax:", service))
-	if service == 'slack':
-		payload.append('<' + botName + '> trades')
-	else:
-		payload.append(botName + ' price')
-	payload.append("")
-	payload.append(f"Full help on {links.link('https://github.com/robdevops/finbot', 'GitHub', service)}")
-	return payload
-
-def prepare_holdings_payload(portfolio_name, service):
-	"""The holdings of the portfolio called `portfolio_name`. Raises sharesight.PortfolioNotFound."""
-	if not portfolio_name:
-		names = [portfolio.name for portfolio in sharesight.portfolios()]
-		return [".holdings: please try again specifying a portfolio:"] + names
-	portfolio, = sharesight.portfolios(portfolio_name)
-	holdings = portfolio.holdings()
-	payload = sorted(holding_line(holding, service) for holding in holdings.values())
-	if payload:
-		payload.insert(0, webhook.bold(f"Holdings for {portfolio.name} ({len(holdings)})", service))
-	return payload
-
-def prepare_who(service, code):
-	"""Which portfolios hold `code` (a ticker without its exchange), each holding linked to Sharesight."""
-	payload = []
-	for portfolio in sharesight.portfolios():
-		found = [holding_line(holding, service, code)
-			for holding in portfolio.holdings().values() if holding['code'].upper() == code]
-		if found:
-			payload += [webhook.bold(f"{portfolio.name}:", service), "\n".join(found), ""]
-	return payload or [f'{code} not found in any portfolio']
-
-def prepare_pie(portfolio_name, by, service, max_slices=8):
-	"""Portfolio value by industry (or by='sector'). Returns (payload, chart); chart is None unless there is something to draw.
-	Raises sharesight.PortfolioNotFound."""
-	if not portfolio_name:
-		names = [portfolio.name for portfolio in sharesight.portfolios()]
-		return [".pie: please try again specifying a portfolio:"] + names, None
-	portfolio, = sharesight.portfolios(portfolio_name)
-	name = portfolio.name
-	key = 'sector_classification_name' if by == 'sector' else 'industry_classification_name'
-	totals = {}
-	for holding in portfolio.holdings().values():
-		try:
-			value = float(holding.get('value'))
-		except (TypeError, ValueError):
-			continue
-		if value > 0:
-			label = holding.get(key) or 'Unclassified'
-			totals[label] = totals.get(label, 0.0) + value
-	if not totals:
-		return [f"{name} has no holdings with a value"], None
-	rows = sorted(totals.items(), key=lambda kv: kv[1], reverse=True)
-	if len(rows) > max_slices:
-		rest = sum(v for _, v in rows[max_slices - 1:])
-		rows = rows[:max_slices - 1]
-		rows.append(('Other', rest))
-	total = sum(v for _, v in rows)
-	title = f"{name} by {by}"
-	payload = [webhook.bold(title, service)]
-	payload += [f"{label}: {value / total * 100:.1f}%" for label, value in rows]
-	noun = {'sector': ('sector', 'sectors'), 'industry': ('industry', 'industries')}[by]
-	subtitle = f"{len(totals)} {noun[0] if len(totals) == 1 else noun[1]}"
-	return payload, {'pie': True, 'rows': rows, 'title': title, 'subtitle': subtitle}
-
-def prepare_marketcap_payload(service, action='top', length=15):
-	"""The `length` tracked stocks of largest (action 'top') or smallest ('bottom') market cap, compared in US dollars at Yahoo's
-	exchange rates: a market cap in won or yen is a far bigger number. Any other action is one ticker's, in its own currency too."""
-	if action not in ('top', 'bottom'):
-		ticker = markets.transform_to_yahoo(action.upper())
-		data = yahoo.add_usd_market_caps(yahoo.fetch_detail(ticker, 600)).get(ticker, {})
-		if 'market_cap' not in data:
-			return [f"Mkt cap not found for {ticker}"]
-		currency = yahoo.market_cap_currency(data) or ''
-		cap = f"{currency} {formats.humanUnits(data['market_cap'])}".strip()
-		if currency != 'USD' and 'market_cap_usd' in data:
-			cap += f" (USD {formats.humanUnits(data['market_cap_usd'])})"
-		link = links.finance_link(ticker, data.get('profile_exchange', ''), service)
-		return [formats.security_line(markets.flag_from_ticker(ticker), data.get('profile_title', ticker), link) + f" mkt cap: {cap}"]
-	caps = [] # (market cap in US dollars, line)
-	for ticker, data in yahoo.add_usd_market_caps(yahoo.fetch(watchlist.with_holdings())).items():
-		if 'market_cap_usd' not in data:
-			print(ticker, "no market cap" if 'market_cap' not in data else f"no exchange rate for {data.get('currency')}", file=sys.stderr)
-			continue
-		link = links.finance_link(ticker, data['profile_exchange'], service)
-		caps.append((data['market_cap_usd'], formats.security_line(markets.flag_from_ticker(ticker), data['profile_title'], link) + f" mkt cap: {formats.humanUnits(data['market_cap_usd'])}"))
-	if not caps:
-		return ['No market cap data found']
-	caps.sort(key=lambda cap: cap[0], reverse=action == 'top')
-	return [webhook.bold(f'{action.title()} {length} tracked stocks by market cap (USD)', service)] + [line for _, line in caps[:length]]
-
-def prepare_rating_payload(service, action, length=15):
-	"""(payload, chart) of the `length` tracked stocks rated `action` ('buy' or 'sell') by min_analysts or more analysts (fewer,
-	and the consensus says little: ASX small caps covered by two to six brokers are nearly all a perfect 1.0 buy), the strongest
-	rating first (1 is a strong buy, 5 a strong sell); on equal ratings, the one more analysts give first."""
-	found = [] # (rating, analysts, line, ticker)
-	for ticker, data in yahoo.fetch_details(watchlist.with_holdings()).items():
-		if 'recommend' not in data or data['recommend_analysts'] < config.prefs.min_analysts:
-			continue
-		recommend = data['recommend'].replace('_', ' ')
-		if (action == 'buy' and 'buy' in recommend) or (action == 'sell' and recommend in ('sell', 'underperform')):
-			rating, analysts = round(data['recommend_index'], 2), data['recommend_analysts']
-			link = links.finance_link(ticker, data['profile_exchange'], service)
-			found.append((rating, analysts, f"{markets.flag_from_ticker(ticker)} {data['profile_title']} ({link}) {rating} ({analysts} analysts)", ticker))
-	found = sorted(found, key=lambda row: (row[0] if action == 'buy' else -row[0], -row[1]))[:length]
-	payload = [line for _, _, line, _ in found]
-	if not payload:
-		return [f"No tracked stocks rated {action} by {config.prefs.min_analysts}+ analysts"], None
-	message = f"Top {length} analyst {action} ratings for tracked stocks ({config.prefs.min_analysts}+ analysts)"
-	shown = [(ticker.split('.')[0], rating) for rating, _, _, ticker in found]
-	low, high = min(r for _, r in shown), max(r for _, r in shown)
-	spread = (high - low) or 0.2
-	# zoom the axis onto the ratings so the bars fill the panel (ratings only run 1-5)
-	chart = {'rows': shown, 'title': message, 'subtitle': 'Mean analyst rating: 1 = strong buy, 5 = strong sell (axis zoomed)',
-		'value_fmt': lambda v: f"{v:g}", 'xlim': (max(0.9, low - spread * 0.9), min(5.1, high + spread * 0.25))}
-	return [webhook.bold(message, service)] + payload, chart
-
-def prepare_value_payload(service, action='pe', ticker_select=None, length=15):
-	"""(payload, chart): the tracked stocks by a ratio (`action`), or one ticker's. A forward P/E is analysts' estimate, so its lists
-	take only stocks that min_analysts or more analysts cover, as .buy does."""
-	payload = []
-	chart_rows = [] # (ticker, ratio)
-	tickers = [ticker_select] if ticker_select else watchlist.with_holdings()
-	market_data = yahoo.fetch(tickers)
-	if action in ('peg', 'bottom peg', 'negative peg'): # the PEG is in Yahoo's details
-		market_data |= yahoo.fetch_details(market_data, strict=bool(ticker_select))
-	covered = None # for a forward P/E list: the tickers enough analysts cover
-	if 'forward' in action and not ticker_select:
-		covered = {ticker for ticker, detail in yahoo.fetch_details(market_data).items() if detail.get('recommend_analysts', 0) >= config.prefs.min_analysts}
-	for ticker in market_data:
-		if covered is not None and ticker not in covered:
-			continue
-		trailing_ratio = None
-		forward_ratio = None
-		peg_ratio = None
-		ratio = None
-		if action in ('pe', 'bottom pe', 'forward pe', 'bottom forward pe'):
-			trailing_ratio = market_data.get(ticker).get('price_to_earnings_trailing')
-			forward_ratio = market_data.get(ticker).get('price_to_earnings_forward')
-			if action in ('forward pe', 'bottom forward pe'):
-				if not ticker_select and forward_ratio is not None and forward_ratio < 0:
-					continue
-				ratio = forward_ratio
-			else:
-				if not ticker_select and trailing_ratio is not None and trailing_ratio < 0:
-					continue
-				ratio = trailing_ratio
-		elif action == 'negative forward pe':
-			forward_ratio = market_data.get(ticker).get('price_to_earnings_forward')
-			if not ticker_select and (forward_ratio is None or forward_ratio >= 0):
-				continue
-			ratio = forward_ratio
-		elif action in ('peg', 'bottom peg'):
-			peg_ratio = market_data.get(ticker).get('price_to_earnings_peg')
-			if not ticker_select and peg_ratio is not None and peg_ratio < 0:
-				continue
-			ratio = peg_ratio
-		elif action == 'negative peg':
-			peg_ratio = market_data.get(ticker).get('price_to_earnings_peg')
-			if not ticker_select and (peg_ratio is None or peg_ratio >= 0):
-				continue
-			ratio = peg_ratio
-		if ratio is None:
-			continue
-		profile_title = market_data.get(ticker).get('profile_title')
-		ticker_link = links.finance_link(ticker, market_data.get(ticker).get('profile_exchange'), service)
-		flag = markets.flag_from_ticker(ticker)
-		if ticker_select and action in {'pe', 'forward pe'}:
-			parts = []
-			if trailing_ratio is not None:
-				parts.append(f"trailing {trailing_ratio}")
-			if forward_ratio is not None:
-				parts.append(f"forward {forward_ratio}")
-			payload.append(f"{flag} {profile_title} ({ticker_link}) PE: {', '.join(parts)}")
-		else:
-			payload.append(f"{flag} {profile_title} ({ticker_link}) {ratio}")
-			chart_rows.append((ticker.split('.')[0], float(ratio)))
-	payload.sort(key=last_col)
-	chart_rows.sort(key=lambda row: row[1])
-	chart = None
-	if not ticker_select:
-		heading_type = "Bottom" if 'bottom' in action else "Top"
-		heading_trail = action.replace('bottom ', '')
-		if 'bottom' in action:
-			payload.reverse()
-			chart_rows.reverse()
-		payload = payload[:length]
-		coverage = f" ({config.prefs.min_analysts}+ analysts)" if covered is not None else ''
-		if not payload:
-			payload = [f"No tracked stocks with {config.prefs.min_analysts}+ analysts have a {heading_trail} ratio" if covered is not None else f"{action}: no data found"]
-		else:
-			heading = f'{heading_type} {length} tracked stocks by {heading_trail} ratio{coverage}'
-			payload.insert(0, f"{webhook.bold(heading, service)}")
-			chart = {'rows': chart_rows[:length], 'title': heading, 'value_fmt': lambda v: f"{v:g}"}
-	return payload, chart
-
-def prepare_beta_payload(service):
-	"""(payload, chart) of the tracked stocks with a beta over 1.5 and a market cap under US$2B, the highest beta first."""
-	found = [] # (beta, line, chart row)
-	for ticker, data in yahoo.add_usd_market_caps(yahoo.fetch_details(watchlist.with_holdings())).items():
-		if 'beta' not in data:
-			continue
-		beta = round(data['beta'], 1)
-		if beta > 1.5 and data.get('market_cap_usd', float('inf')) < 2e9:
-			link = links.finance_link(ticker, data['profile_exchange'], service)
-			found.append((beta, formats.security_line(markets.flag_from_ticker(ticker), data['profile_title'], link) + f" {beta} β", (ticker.split('.')[0], float(beta))))
-	if not found:
-		return ['No tracked stocks have a beta over 1.5 and mkt cap under US$2B'], None
-	found.sort(key=lambda row: row[0], reverse=True)
-	heading = 'Beta over 1.5 and mkt cap under US$2B'
-	chart = {'rows': [row for _, _, row in found], 'title': heading, 'value_fmt': lambda v: f"{v:g}"}
-	return [webhook.bold(heading, service)] + [line for _, line, _ in found], chart
-
-MARKET_STATE_EMOJI = {'REGULAR': '🟢', 'PRE': '🟠', 'POST': '🟠'} # any other state is closed: 🔴
-
-def prepare_profile_payload(service, user, ticker):
-	"""The .SYMBOL report: what Yahoo knows about one company. A field Yahoo did not send leaves its line out."""
-	ticker_orig = markets.transform_to_yahoo(ticker)
-	now = datetime.datetime.now()
-	payload = []
-	ticker, market_data = yahoo.resolve(ticker_orig, lambda symbol: yahoo.fetch_detail(symbol, 600))
-	if ticker not in market_data:
-		return [f"{user} 🛑 Beep Boop. I could not find {ticker_orig}"]
-	data = market_data[ticker]
-	code = ticker.split('.')[0]
-	title = data.get('profile_title', '')
-	state = data.get('marketState', '').rstrip()
-	currency = data.get('currency', '')
-	exchange = data.get('profile_exchange', '').replace('NasdaqCM', 'Nasdaq').replace('NasdaqGS', 'Nasdaq').replace('NYSEArca', 'NYSE')
-	ticker_link = links.finance_link(ticker, exchange, service, brief=False)
-	gfinance_link = links.gfinance_link(ticker, exchange, service, brief=True, text='google') # all three before exchange is renamed below
-	sws_link = links.link(simplywallst.get_url(ticker, title, exchange), 'simplywall.st', service)
-	macrotrends_link = links.link('https://www.google.com/search?q=site:macrotrends.net+' + title + '+PE Ratio+' + code + '&btnI', 'macrotrends', service)
-	if 'profile_bio' in data:
-		payload.append(formats.make_paragraphs(data['profile_bio']))
-
-	if exchange == 'ASX':
-		exchange_url = 'https://www2.asx.com.au/markets/company/' + code
-	elif exchange == 'HKSE':
-		exchange_url = 'https://www.hkex.com.hk/Market-Data/Securities-Prices/Equities/Equities-Quote?sym=' + code + '&sc_lang=en'
-	elif 'Nasdaq' in exchange:
-		exchange_url = 'https://www.nasdaq.com/market-activity/stocks/' + ticker.lower()
-	elif exchange == 'NYSE':
-		exchange_url = 'https://www.nyse.com/quote/XNYS:' + ticker
-	elif exchange == 'Taiwan':
-		exchange = 'TWSE'
-		exchange_url = 'https://mis.twse.com.tw/stock/fibest.jsp?stock=' + code + '&lang=en_us'
-	elif exchange == 'Tokyo':
-		exchange = 'JPX'
-		exchange_url = 'https://quote.jpx.co.jp/jpx/template/quote.cgi?F=tmp/e_stock_detail&MKTN=T&QCODE=' + code
-	else:
-		exchange_url = 'https://www.google.com/search?q=stock+exchange+' + exchange + '+' + code + '&btnI'
-	exchange_link = links.link(exchange_url, exchange, service)
-	location = [data[key] for key in ('profile_city', 'profile_state', 'profile_country') if key in data]
-	if payload:
-		payload.append("")
-
-	if location:
-		payload.append(webhook.bold("Location:", service) + " " + ', '.join(location))
-	if 'profile_industry' in data and 'profile_sector' in data:
-		payload.append(webhook.bold("Classification: ", service) + f"{data['profile_industry']}, {data['profile_sector']}")
-	if 'profile_employees' in data:
-		payload.append(webhook.bold("Employees: ", service) + f"{data['profile_employees']:,}")
-	if 'profile_website' in data:
-		sites = [links.link(data['profile_website'], formats.strip_url(data['profile_website']), service), exchange_link]
-		if 'NYSE' in exchange or 'Nasdaq' in exchange:
-			sites += [links.link('https://finviz.com/quote.ashx?t=' + ticker, 'finviz', service), macrotrends_link,
-				links.link('https://seekingalpha.com/symbol/' + ticker, 'seekingalpha', service)]
-		elif exchange == 'ASX':
-			sites.append(links.link('https://www.shortman.com.au/stock?q=' + code.lower(), 'shortman', service))
-		payload.append(webhook.bold("Links: ", service) + ' | '.join(sites))
-
-	if payload:
-		payload.append("")
-
-	if 'currency' in data and 'market_cap' in data:
-		payload.append(webhook.bold("Mkt cap:", service) + f" {yahoo.market_cap_currency(data)} {formats.humanUnits(data['market_cap'])}") # GBP, not pence
-	if all(key in data for key in ('shareholder_equity', 'total_debt', 'total_cash', 'profile_industry')) and data['shareholder_equity'] and 'Bank' not in data['profile_industry']:
-		net_debt_equity_ratio = round((data['total_debt'] - data['total_cash']) / data['shareholder_equity'] * 100)
-		if net_debt_equity_ratio > 0:
-			payload.append(webhook.bold("Net debt/equity ratio:", service) + f" {net_debt_equity_ratio}%" + ('⚠️ ' if net_debt_equity_ratio > 40 else ''))
-	if 'earnings_date' in data:
-		earnings_date = datetime.datetime.fromtimestamp(data['earnings_date'])
-		if earnings_date > now:
-			payload.append(webhook.bold("Earnings date:", service) + f" {earnings_date:%b %d}")
-	if data.get('dividend', 0) > 0:
-		payload.append(webhook.bold("Dividend:", service) + f" {data['dividend']}%")
-		if 'ex_dividend_date' in data:
-			ex_dividend_date = datetime.datetime.fromtimestamp(data['ex_dividend_date'])
-			if ex_dividend_date > now:
-				payload.append(webhook.bold("Ex-dividend date:", service) + f" {ex_dividend_date:%b %d}")
-	cashflow = data.get('free_cashflow', data.get('operating_cashflow'))
-	if cashflow:
-		payload.append(webhook.bold("Cashflow positive:", service) + (" no ⚠️ " if cashflow < 0 else " yes"))
-	if 'net_income' in data:
-		payload.append(webhook.bold("Profitable:", service) + (" no ⚠️ " if data['net_income'] <= 0 else " yes"))
-	else:
-		payload.append(webhook.bold("Profitable:", service) + " unknown ⚠️")
-
-	if payload:
-		payload.append("")
-
-	if 'earningsQ' in data:
-		deltas = [(doDelta(data[key]), label) for key, label in (('revenueQ', 'quarterly revenue'), ('earningsQ', 'quarterly earnings'),
-			('revenueY', 'annual revenue'), ('earningsY', 'annual earnings'))]
-		payload += [f"{delta}\t{label} delta" for delta, label in deltas] # doDelta always gives at least three symbols
-
-	if payload:
-		payload.append("")
-
-	for key, label in (('revenueEstimateY', 'Revenue forecast (1Y):'), ('earningsEstimateY', 'Earnings forecast (1Y):')):
-		if key in data:
-			growth = int(round(data[key]))
-			payload.append(webhook.bold(label, service) + (f" +{growth}% " if growth > 0 else f" {growth}% ⚠️ "))
-	if 'insiderBuy' in data:
-		if data['insiderBuy'] > data['insiderSell']:
-			payload.append(webhook.bold("Net insider action (3M):", service) + f" Buy {currency} {formats.humanUnits(data['insiderBuyValue'])} ")
-		elif data['insiderBuy'] < data['insiderSell']:
-			payload.append(webhook.bold("Net insider action (3M):", service) + f" Sell {currency} {formats.humanUnits(data['insiderSellValue'])}⚠️ ")
-	if 'short_percent' in data:
-		payload.append(webhook.bold("Shorted stock:", service) + f" {data['short_percent']}%" + ('⚠️ ' if data['short_percent'] > 10 else ''))
-	if 'recommend' in data:
-		payload.append(webhook.bold("Score:", service) + f" {data['recommend_index']} {data['recommend'].replace('_', ' ')} ({data['recommend_analysts']} analysts)")
-
-	if payload:
-		payload.append("")
-
-	if 'regularMarketPrice' in data:
-		price_line = webhook.bold("Price:", service) + f" {currency} {data['regularMarketPrice']:,.2f}"
-		if state != 'REGULAR' and 'prePostMarketPrice' in data:
-			price_line += f" ({data['prePostMarketPrice']:,.2f} after hrs)"
-		payload.append(price_line)
-	trailing_pe = str(int(round(data['price_to_earnings_trailing']))) if 'price_to_earnings_trailing' in data else 'N/A ⚠️ '
-	if data.get('net_income', 0) > 0:
-		payload.append(webhook.bold("Trailing P/E:", service) + f" {trailing_pe}")
-	if data.get('netIncomeToCommon', 0) > 0 and data.get('sharesOutstanding') and 'regularMarketPrice' in data:
-		current_pe = round(data['regularMarketPrice'] / (data['netIncomeToCommon'] / data['sharesOutstanding']))
-		payload.append(webhook.bold("Current P/E:", service) + f" {current_pe}")
-	if 'price_to_earnings_forward' in data:
-		forward_pe = int(round(data['price_to_earnings_forward']))
-		warn = forward_pe < 0
-		if 'profile_industry' in data: # software is dear: it warns above 100, the rest above 30 or above their trailing P/E
-			warn = warn or forward_pe > (100 if 'Software' in data['profile_industry'] else 30)
-			warn = warn or ('price_to_earnings_trailing' in data and forward_pe > int(trailing_pe))
-		payload.append(webhook.bold("Forward P/E:", service) + f" {forward_pe} " + ('⚠️ ' if warn else ''))
-	if 'price_to_earnings_peg' in data:
-		payload.append(webhook.bold("PEG ratio:", service) + f" {round(data['price_to_earnings_peg'], 1)}")
-	if 'price_to_sales' in data:
-		price_to_sales = round(data['price_to_sales'], 1)
-		payload.append(webhook.bold("PS ratio:", service) + f" {price_to_sales} " + ('⚠️ ' if price_to_sales > 15 else ''))
-
-	if payload:
-		payload.append("")
-
-	try:
-		price_history, _ = yahoo.price_history(ticker, graph=False) # the percentages only: the profile has no chart
-	except RuntimeError as e: # no chart data: the profile goes out without these lines
-		print(ticker, e, file=sys.stderr)
-		price_history = {}
-	for interval in ('5Y', '1Y', '1M', '1D'):
-		if interval in price_history:
-			percent = price_history[interval]
-			payload.append(f"{formats.get_emoji(percent)} {webhook.bold(interval + ':', service)} {percent}%")
-	if state != 'REGULAR':
-		for key, label in (('percent_change_premarket', 'Pre-market:'), ('percent_change_postmarket', 'Post-market:')):
-			if key in data:
-				payload.append(f"{formats.get_emoji(data[key])} {webhook.bold(label, service)} {data[key]:,}%")
-				break
-
-	if 'profile_website' in data and config.prefs.hyperlinkFooter and config.prefs.hyperlink:
-		if payload:
-			payload.append("")
-		payload.append(f"{gfinance_link} | {links.yahoo_link(ticker, service, brief=True, text='yahoo')} | {sws_link}")
-	heading = f"{webhook.bold(title, service)} ({ticker_link}) {MARKET_STATE_EMOJI.get(state, '🔴') if state else ''}"
-	if ticker_orig == ticker:
-		payload.insert(0, heading)
-	else:
-		payload[:0] = [f"I could not find {ticker_orig} but I found {ticker_link}:", "", heading]
-	payload = [i[0] for i in groupby(payload)] # de-dupe white space
-	if len(payload) < 2:
-		payload.append("no data found")
-	return payload
diff --git a/lib/reports/__init__.py b/lib/reports/__init__.py
new file mode 100644
index 00000000..eb2923ea
--- /dev/null
+++ b/lib/reports/__init__.py
@@ -0,0 +1,14 @@
+"""What finbot says. A module per scheduled report: cal, cleanup (of the cache), milestone, performance, podcasts, price, rating,
+reminder, shorts and trades, each with job(argv), which parses its schedule line, and lambda_handler(**kwargs), which runs it; the
+top-level script of the same name runs it from the command line. And the chat-only reports: about (.help, .jobs), history
+(.history, .compare), holdings (.holdings, .who, .pie, .watchlist), profile (.SYMBOL) and rankings (.marketcap, .pe and the rest)."""
+import os
+import sys
+
+
+def cli(path):
+	"""Run the report a top-level script is named for (price.py: lib/reports/price.py) with the script's arguments."""
+	import importlib
+	from lib import webhook
+	report = importlib.import_module('lib.reports.' + os.path.basename(path).removesuffix('.py'))
+	webhook.cli(sys.argv[0], report.job, report.lambda_handler)
diff --git a/lib/reports/about.py b/lib/reports/about.py
new file mode 100644
index 00000000..e6a2e7d6
--- /dev/null
+++ b/lib/reports/about.py
@@ -0,0 +1,87 @@
+"""What finbot can do: the .help text, and the .jobs listing of the schedule (with running a job from it)."""
+import re
+from lib import jobs
+from lib import links
+from lib import webhook
+
+def prepare_jobs_payload(service, args):
+	"""`.jobs` lists the scheduled commands, each once; `.jobs run <number|command>` runs one now, as cron would.
+	Returns (payload, reply_markup): Telegram gets a button per command."""
+	scheduler = jobs.current
+	if scheduler is None:
+		return ['Scheduled jobs are off (scheduler=False)'], None
+	args = (args or '').strip()
+	if not args:
+		commands = scheduler.commands()
+		if not commands:
+			return ['No scheduled jobs'], None
+		payload = [webhook.bold('Scheduled commands:', service)]
+		for number, (label, _, _) in enumerate(commands, 1):
+			payload.append(f"{number}. {jobs.display(label)}" + (' (running)' if label in scheduler.busy else ''))
+		payload.append("\nRun one with .jobs run NUMBER" + (' or tap it' if service == 'telegram' else ''))
+		markup = None
+		if service == 'telegram':
+			buttons = [{'text': f"{number}. {jobs.display(label)}"[:30], 'callback_data': f"j|{number}|{jobs.checksum(label)}"} for number, (label, _, _) in enumerate(commands, 1)]
+			markup = {'inline_keyboard': [buttons[i:i + 2] for i in range(0, len(buttons), 2)]}
+		return payload, markup
+	m = re.fullmatch(r'run(?:\s+(.+))?', args, re.IGNORECASE)
+	if not m or not m.group(1):
+		return ['Usage: .jobs [run NUMBER|COMMAND]'], None
+	return [run_job_line(scheduler, m.group(1))], None
+
+def run_job_line(scheduler, text):
+	"""Run a scheduled command now and say how it went in one line."""
+	try:
+		return job_line(scheduler, *scheduler.run_now(text))
+	except jobs.ScheduleError as e:
+		return str(e)
+
+def job_line(scheduler, label, outcome):
+	number = [command[0] for command in scheduler.commands()].index(label) + 1
+	return f"{number}. {jobs.display(label)}: {outcome}"
+
+def prepare_help(service, botName):
+	payload = []
+	payload.append(webhook.bold("Tracked securities:", service))
+	payload.append(".holdings")
+	payload.append(".pie PORTFOLIO [sector]")
+	payload.append(".watchlist [add|del SYMBOL]")
+	payload.append(".who SYMBOL")
+
+	payload.append(webhook.bold("\nCompany info:", service))
+	payload.append('.SYMBOL')
+	payload.append(".dividend [period|SYMBOL]")
+	payload.append(".earnings [period|SYMBOL]")
+	payload.append(".marketcap [SYMBOL|bottom|top]")
+
+	payload.append(webhook.bold("\nPrice:", service))
+	payload.append(".beta")
+	payload.append(".compare SYMBOL SYMBOL [...] [period|max]")
+	payload.append(".history SYMBOL")
+	payload.append(".performance [period] [portfolio]")
+	payload.append(".price [percent|SYMBOL] [period]")
+	payload.append(".session [percent|SYMBOL]")
+	payload.append(".premarket [percent|SYMBOL]")
+
+	payload.append(webhook.bold("\nRisk & Value:", service))
+	payload.append(".buy")
+	payload.append(".sell")
+	payload.append(".pe [SYMBOL|top|bottom]")
+	payload.append(".forwardpe [SYMBOL|top|bottom|negative]")
+	payload.append(".peg [SYMBOL|top|bottom|negative]")
+	payload.append(".shorts [percent|SYMBOL]")
+
+	payload.append(webhook.bold("\nTrades:", service))
+	payload.append(".trades [period] [portfolio]")
+
+	payload.append(webhook.bold("\nScheduled jobs:", service))
+	payload.append(".jobs [run NUMBER|COMMAND]")
+
+	payload.append(webhook.bold("\nAlternative syntax:", service))
+	if service == 'slack':
+		payload.append('<' + botName + '> trades')
+	else:
+		payload.append(botName + ' price')
+	payload.append("")
+	payload.append(f"Full help on {links.link('https://github.com/robdevops/finbot', 'GitHub', service)}")
+	return payload
diff --git a/cal.py b/lib/reports/cal.py
old mode 100755
new mode 100644
similarity index 97%
rename from cal.py
rename to lib/reports/cal.py
index 256d7444..07984730
--- a/cal.py
+++ b/lib/reports/cal.py
@@ -1,5 +1,4 @@
-#!/usr/bin/env python3
-
+""".earnings and .dividend: upcoming earnings and ex-dividend dates (scheduled: cal.py earnings | ex-dividend)."""
 import datetime
 import sys
 
@@ -93,6 +92,3 @@ def job(argv):
 			return f'cal.py {event}', {'event': event}
 		case _:
 			raise ValueError('[earnings|ex-dividend]')
-
-if __name__ == "__main__":
-	webhook.cli(sys.argv[0], job, lambda_handler)
diff --git a/cleanup.py b/lib/reports/cleanup.py
old mode 100755
new mode 100644
similarity index 90%
rename from cleanup.py
rename to lib/reports/cleanup.py
index cb1eb62f..23b3d99b
--- a/cleanup.py
+++ b/lib/reports/cleanup.py
@@ -1,8 +1,6 @@
-#!/usr/bin/env python3
-
+"""Deletes old cache files (scheduled: cleanup.py [days]); with days=0 it is the cache flush."""
 import os, sys, time
 from lib import config
-from lib import webhook
 
 def lambda_handler(days=30):
 	"""Delete the cache files (finbot_*) that were last written more than `days` days ago (days=0: all of them, a flush), but not
@@ -34,6 +32,3 @@ def job(argv):
 			return f'cleanup.py {days}', {'days': int(days)}
 		case _:
 			raise ValueError('[days]')
-
-if __name__ == "__main__":
-	webhook.cli(sys.argv[0], job, lambda_handler)
diff --git a/lib/reports/history.py b/lib/reports/history.py
new file mode 100644
index 00000000..1ed17e08
--- /dev/null
+++ b/lib/reports/history.py
@@ -0,0 +1,104 @@
+"""Price history: .history (one stock over every period) and .compare (two to six stocks, rebased)."""
+import hashlib
+import datetime
+from lib import config
+from lib import cache
+from lib import formats
+from lib import links
+from lib import markets
+from lib import render
+from lib import webhook
+from lib import yahoo
+
+def prepare_compare(service, args):
+	"""Compare 2-6 tickers as rebased % change. A trailing time period (3m/1y/90d/ytd) is optional;
+	without one, the span is the earliest date covered by every ticker. Returns (caption lines, image, tickers, period days)."""
+	import pandas as pd
+	usage = ['Usage: .compare SYMBOL SYMBOL [SYMBOL...up to 6] [period|max]  (default: 10 years)']
+	period_days = None
+	full_history = False
+	args = list(args)
+	if args:
+		if args[-1].lower() in ('max', 'all'):
+			full_history = True
+			args.pop()
+		else:
+			try:
+				period_days = formats.days_from_human_days(args[-1])
+				args.pop()
+			except ValueError:
+				pass
+	tickers = list(dict.fromkeys(markets.transform_to_yahoo(a.upper()) for a in args)) # de-duplicated, order kept
+	if not 2 <= len(tickers) <= 6:
+		return usage, None, tickers, period_days
+	spec = period_days or ('max' if full_history else 'default')
+	cache_key = hashlib.sha1(f"{service}|{','.join(tickers)}|{spec}".encode()).hexdigest()[:16]
+	image_cache_file, caption_cache_file = f"finbot_compare_{cache_key}.png", f"finbot_compare_{cache_key}.json"
+	if config.prefs.cache:
+		cached_image = cache.read_bytes(image_cache_file)
+		cached_caption = cache.read(caption_cache_file)
+		if cached_image and cached_caption:
+			return cached_caption, cached_image, tickers, period_days
+	names, exchanges, closes = {}, {}, {}
+	for ticker, history in zip(tickers, yahoo.fetch_each(yahoo.price_series, tickers)): # downloaded at the same time
+		if isinstance(history, Exception):
+			raise history
+		names[ticker], exchanges[ticker], closes[ticker] = history
+	limiting = max(closes, key=lambda t: closes[t].index[0]) # ticker with the shortest history
+	common_start = closes[limiting].index[0]
+	start = common_start
+	note = ''
+	if not period_days and not full_history: # default window; ask for a longer period (or max) to go further
+		default_start = pd.Timestamp(datetime.datetime.now().date() - datetime.timedelta(days=3655))
+		start = max(common_start, default_start)
+	if period_days:
+		requested = pd.Timestamp(datetime.datetime.now().date() - datetime.timedelta(days=period_days))
+		if requested >= common_start:
+			start = requested
+		else:
+			note = f" (limited by {limiting} history)"
+	cropped = []
+	for ticker in tickers:
+		series = closes[ticker]
+		base = series[series.index <= start]
+		base_value = base.iloc[-1] if len(base) else series.iloc[0] # last close on/before the start date
+		series = pd.concat([pd.Series([base_value], index=[start]), series[series.index > start]])
+		cropped.append((ticker, series))
+	if any(len(series) < 2 for _, series in cropped):
+		raise RuntimeError("not enough price history in that period to compare")
+	span_days = (datetime.datetime.now().date() - start.date()).days
+	period_label = formats.days_english(period_days) if period_days and not note else f"{span_days} days"
+	by_performance = sorted(cropped, key=lambda item: item[1].iloc[-1] / item[1].iloc[0], reverse=True)
+	title = ' vs '.join(t.split('.')[0] for t, _ in by_performance) # best first; line colours stay tied to input order
+	subtitle = f"% change over {period_label}, since {start:%d %b %Y}" + note
+	image = render.compare_graph([(t.split('.')[0], series) for t, series in cropped], title, subtitle)
+	image.seek(0)
+	caption = [webhook.bold(f"{period_label}, since {start:%d %b %Y}", service) + note]
+	results = sorted(((series.iloc[-1] / series.iloc[0] - 1) * 100, t) for t, series in cropped)
+	for pct, t in reversed(results):
+		link = links.finance_link(t, exchanges[t], service, days=span_days, brief=False)
+		caption.append(f"{formats.get_emoji(pct)} {names[t]} ({link}): {pct:+,.1f}%")
+	if config.prefs.cache:
+		cache.write_bytes(image_cache_file, image)
+		cache.write(caption_cache_file, caption)
+		image.seek(0)
+	return caption, image, tickers, period_days
+
+def history_payload(ticker, service, highlight=None, graph=False):
+	"""Caption lines for .history: performance table, with the selected interval in bold. Returns (lines, percent table, image):
+	the full history's chart with graph=True (if the graph preference is on), else None."""
+	market_data = yahoo.fetch_detail(ticker, 600)
+	title = market_data.get(ticker, {}).get('profile_title', '')
+	ticker_link = links.finance_link(ticker, market_data.get(ticker, {}).get('profile_exchange', ''), service, days=1825, brief=False)
+	if ticker not in market_data or 'percent_change' not in market_data[ticker]:
+		return [f".history: no data found for ticker {ticker}"], None, None
+	table, image = yahoo.price_history(ticker) if graph else yahoo.price_history(ticker, graph=False)
+	lines = [webhook.bold(f"{title} ({ticker_link}) performance history", service)]
+	for interval in ('Max', '10Y', '5Y', '3Y', '1Y', 'YTD', '6M', '3M', '1M', '7D', '1D'):
+		if interval in table:
+			percent = table[interval]
+			label = webhook.bold(interval + ':', service)
+			if interval == highlight:
+				label = '● ' + label
+			lines.append(f"{formats.get_emoji(percent)} {label} {formats.signed_percent(percent)}")
+	return lines, table, image
diff --git a/lib/reports/holdings.py b/lib/reports/holdings.py
new file mode 100644
index 00000000..c644927c
--- /dev/null
+++ b/lib/reports/holdings.py
@@ -0,0 +1,102 @@
+"""What is held and followed: .holdings, .who, .pie and .watchlist."""
+import re
+from lib import watchlist
+from lib import sharesight
+from lib import formats
+from lib import links
+from lib import markets
+from lib import webhook
+from lib import yahoo
+
+def holding_line(holding, service, text=None):
+	"""'🇦🇺 Title (CODE)' for a Sharesight holding, CODE (or `text`) linking to the holding on Sharesight."""
+	link = links.link(f"https://portfolio.sharesight.com/holdings/{holding['holding_id']}/dashboard", holding['code'] if text is None else text, service)
+	return f"{markets.flag_from_market(holding['market_code'])} {formats.transform_title(holding['name'])} ({link})"
+
+def prepare_holdings_payload(portfolio_name, service):
+	"""The holdings of the portfolio called `portfolio_name`. Raises sharesight.PortfolioNotFound."""
+	if not portfolio_name:
+		names = [portfolio.name for portfolio in sharesight.portfolios()]
+		return [".holdings: please try again specifying a portfolio:"] + names
+	portfolio, = sharesight.portfolios(portfolio_name)
+	holdings = portfolio.holdings()
+	payload = sorted(holding_line(holding, service) for holding in holdings.values())
+	if payload:
+		payload.insert(0, webhook.bold(f"Holdings for {portfolio.name} ({len(holdings)})", service))
+	return payload
+
+def prepare_who(service, code):
+	"""Which portfolios hold `code` (a ticker without its exchange), each holding linked to Sharesight."""
+	payload = []
+	for portfolio in sharesight.portfolios():
+		found = [holding_line(holding, service, code)
+			for holding in portfolio.holdings().values() if holding['code'].upper() == code]
+		if found:
+			payload += [webhook.bold(f"{portfolio.name}:", service), "\n".join(found), ""]
+	return payload or [f'{code} not found in any portfolio']
+
+def prepare_pie(portfolio_name, by, service, max_slices=8):
+	"""Portfolio value by industry (or by='sector'). Returns (payload, chart); chart is None unless there is something to draw.
+	Raises sharesight.PortfolioNotFound."""
+	if not portfolio_name:
+		names = [portfolio.name for portfolio in sharesight.portfolios()]
+		return [".pie: please try again specifying a portfolio:"] + names, None
+	portfolio, = sharesight.portfolios(portfolio_name)
+	name = portfolio.name
+	key = 'sector_classification_name' if by == 'sector' else 'industry_classification_name'
+	totals = {}
+	for holding in portfolio.holdings().values():
+		try:
+			value = float(holding.get('value'))
+		except (TypeError, ValueError):
+			continue
+		if value > 0:
+			label = holding.get(key) or 'Unclassified'
+			totals[label] = totals.get(label, 0.0) + value
+	if not totals:
+		return [f"{name} has no holdings with a value"], None
+	rows = sorted(totals.items(), key=lambda kv: kv[1], reverse=True)
+	if len(rows) > max_slices:
+		rest = sum(v for _, v in rows[max_slices - 1:])
+		rows = rows[:max_slices - 1]
+		rows.append(('Other', rest))
+	total = sum(v for _, v in rows)
+	title = f"{name} by {by}"
+	payload = [webhook.bold(title, service)]
+	payload += [f"{label}: {value / total * 100:.1f}%" for label, value in rows]
+	noun = {'sector': ('sector', 'sectors'), 'industry': ('industry', 'industries')}[by]
+	subtitle = f"{len(totals)} {noun[0] if len(totals) == 1 else noun[1]}"
+	return payload, {'pie': True, 'rows': rows, 'title': title, 'subtitle': subtitle}
+
+def prepare_watchlist(service, user, action=None, ticker=None):
+	"""The .watchlist reply: the companies watched, or one line on adding ('add') or removing ('delete') a ticker."""
+	if action:
+		return [watchlist_change(service, user, action, markets.transform_to_yahoo(ticker))]
+	watched = watchlist.tickers()
+	market_data = yahoo.fetch(watched) if watched else {}
+	lines = sorted((f"{markets.flag_from_ticker(t)} {quote['profile_title']} ({links.finance_link(t, quote.get('profile_exchange'), service)})"
+		for t, quote in market_data.items()), key=lambda line: re.findall('[A-Z].*', line)) # by name, the flag aside
+	if not lines:
+		return ['Watchlist is empty. Try ".watchlist add SYMBOL" to create it']
+	return [f"Hi {user}, I'm currently tracking:"] + lines
+
+def watchlist_change(service, user, action, asked):
+	"""The reply to adding or removing `asked` (a Yahoo ticker)."""
+	def described(ticker, quote): # "Full Name (TICKER)", bold
+		quote = quote or {}
+		link = links.finance_link(ticker, quote.get('profile_exchange') or ticker, service)
+		return webhook.bold(f"{quote['profile_title']} ({link})" if quote.get('profile_title') else link, service)
+	if action == 'delete':
+		if asked not in watchlist.tickers():
+			return f"Beep Boop. I could not find {webhook.bold(asked, service)} to remove it"
+		quote = yahoo.fetch([asked]).get(asked)
+		watchlist.remove(asked)
+		return f"Ok {user}, I deleted {described(asked, quote)}"
+	found, quote = watchlist.find(asked)
+	if not quote:
+		return f"Beep Boop. I could not find {webhook.bold(asked, service)} to add it"
+	added = watchlist.add(found)
+	if found != asked:
+		done = "so I added" if added else "and I'm already tracking"
+		return f"Beep Boop. I could not find {webhook.bold(asked, service)} {done} {described(found, quote)}"
+	return f"Ok {user}, I added {described(found, quote)}" if added else f"{user}, I'm already tracking {described(found, quote)}"
diff --git a/milestone.py b/lib/reports/milestone.py
old mode 100755
new mode 100644
similarity index 96%
rename from milestone.py
rename to lib/reports/milestone.py
index dc5303b6..8dbd19e2
--- a/milestone.py
+++ b/lib/reports/milestone.py
@@ -1,6 +1,4 @@
-#!/usr/bin/env python3
-
-import sys
+"""Record highs and lows (all-time) among the stocks followed (scheduled: milestone.py)."""
 from itertools import groupby
 from lib import state
 from lib import webhook
@@ -71,6 +69,3 @@ def prepare_milestone_payload(service, market_data):
 
 def job(argv):
 	return 'milestone.py', {}
-
-if __name__ == "__main__":
-	webhook.cli(sys.argv[0], job, lambda_handler)
diff --git a/performance.py b/lib/reports/performance.py
old mode 100755
new mode 100644
similarity index 93%
rename from performance.py
rename to lib/reports/performance.py
index dc200aa6..cecef39f
--- a/performance.py
+++ b/lib/reports/performance.py
@@ -1,5 +1,4 @@
-#!/usr/bin/env python3
-
+""".performance: each portfolio's return over a period, beside two benchmarks (scheduled: performance.py [days])."""
 import sys
 
 from lib import config
@@ -9,7 +8,7 @@
 from lib import links
 from lib import render
 from lib import yahoo
-from lib import charts
+from lib import buttons
 
 BENCHMARKS = (('SPY', 'NYSEARCA', 'S&P 500'), ('QQQ', 'NasdaqGM', 'NASDAQ 100'))
 
@@ -69,8 +68,8 @@ def lambda_handler(chat_id=None, past_days=None, service=None, user='', portfoli
 	performance = gather(past_days, portfolio_select)
 	for service, url, chat in webhook.destinations(interactive, service, chat_id):
 		payload, graph = report(service, past_days, portfolio_select, interactive, performance)
-		buttons = charts.keyboard('f', [portfolio_select or ''], charts.period_for_days(past_days)) if graph and service == 'telegram' else None
-		webhook.send(service, url, chat, payload, graph, reply_markup=buttons)
+		markup = buttons.keyboard('f', [portfolio_select or ''], buttons.period_for_days(past_days)) if graph and service == 'telegram' else None
+		webhook.send(service, url, chat, payload, graph, reply_markup=markup)
 
 def job(argv):
 	"""Arguments of a scheduled run: (label, lambda_handler kwargs). Raises ValueError with the usage on a bad argument."""
@@ -81,6 +80,3 @@ def job(argv):
 			return f'performance.py {days}', {'past_days': int(days)}
 		case _:
 			raise ValueError('[integer]')
-
-if __name__ == "__main__":
-	webhook.cli(sys.argv[0], job, lambda_handler)
diff --git a/podcasts.py b/lib/reports/podcasts.py
old mode 100755
new mode 100644
similarity index 97%
rename from podcasts.py
rename to lib/reports/podcasts.py
index 650f2651..7ba6d72a
--- a/podcasts.py
+++ b/lib/reports/podcasts.py
@@ -1,6 +1,4 @@
-#!/usr/bin/env python3
-
-import sys
+"""New episodes of a podcast feed (scheduled: podcasts.py "<feed url>")."""
 import hashlib
 import string
 import xml.etree.ElementTree as ET
@@ -113,6 +111,3 @@ def job(argv):
 			return 'podcasts.py ' + hashlib.md5(feed_url.encode()).hexdigest()[:8], {'feed_url': feed_url} # the label matches the seen file's name
 		case _:
 			raise ValueError('"<podcast_feed_url>"')
-
-if __name__ == "__main__":
-	webhook.cli(sys.argv[0], job, lambda_handler)
diff --git a/price.py b/lib/reports/price.py
old mode 100755
new mode 100644
similarity index 94%
rename from price.py
rename to lib/reports/price.py
index 8bf4c0b0..88710ed3
--- a/price.py
+++ b/lib/reports/price.py
@@ -1,5 +1,5 @@
-#!/usr/bin/env python3
-
+""".price, .session, .premarket, .postmarket: price moves past a threshold (scheduled: price.py midsession | interday |
+premarket | close | <days>)."""
 import sys
 import datetime
 
@@ -12,7 +12,7 @@
 from lib import render
 from lib import watchlist
 from lib import yahoo
-from lib import charts
+from lib import buttons
 
 def gather(specific_stock=None, days=None):
 	"""(tickers, market data) for a price report: Yahoo's quotes; over `days` each holding's Sharesight figure, and for the tickers
@@ -168,17 +168,17 @@ def lambda_handler(chat_id=None, threshold=None, service=None, user='', specific
 	if threshold is None:
 		threshold = config.prefs.price_percent
 	market = gather(specific_stock, days)
-	def buttons(service): # a chart's period buttons: one ticker's chart, and the daily and N-day lists (not the intraday ones)
+	def markup(service): # a chart's period buttons: one ticker's chart, and the daily and N-day lists (not the intraday ones)
 		if service != 'telegram':
 			return None
 		if specific_stock:
-			return charts.keyboard('p', market[0], charts.period_for_days(days))
+			return buttons.keyboard('p', market[0], buttons.period_for_days(days))
 		if not (midsession or premarket or close):
-			return charts.keyboard('l', [f"{threshold:g}", str(top or 0)], charts.period_for_days(days, 'd' if not days else None))
+			return buttons.keyboard('l', [f"{threshold:g}", str(top or 0)], buttons.period_for_days(days, 'd' if not days else None))
 		return None
 	for service, url, chat in webhook.destinations(interactive, service, chat_id):
 		payload, graph = report(service, threshold, user, specific_stock, interactive, midsession, premarket, interday, days, close, top, market)
-		webhook.send(service, url, chat, payload, graph, reply_markup=buttons(service) if graph else None)
+		webhook.send(service, url, chat, payload, graph, reply_markup=markup(service) if graph else None)
 
 def job(argv):
 	"""Arguments of a scheduled run: (label, lambda_handler kwargs). Raises ValueError with the usage on a bad argument."""
@@ -189,6 +189,3 @@ def job(argv):
 			return f'price.py {days}', {'threshold': 0, 'days': int(days), 'top': 10}
 		case _:
 			raise ValueError('[midsession|interday|premarket|close|days (int)]')
-
-if __name__ == "__main__":
-	webhook.cli(sys.argv[0], job, lambda_handler)
diff --git a/lib/reports/profile.py b/lib/reports/profile.py
new file mode 100644
index 00000000..04291a71
--- /dev/null
+++ b/lib/reports/profile.py
@@ -0,0 +1,214 @@
+"""The .SYMBOL report: what Yahoo and Simply Wall St know about one company."""
+from itertools import pairwise
+import datetime
+import sys
+from lib import config
+from lib import formats
+from lib import links
+from lib import markets
+from lib import webhook
+from lib import yahoo
+from lib import simplywallst
+from itertools import groupby
+
+def doDelta(inputList):
+	deltaString = ''
+	inputListFixed = inputList.copy()
+	# replace any NoneType to allow delta calculation
+	for idx, absolute in enumerate(inputList):
+		if absolute is None:
+			if idx == 0:
+				# get the next value that's not None
+				inputListFixed[idx] = next((x for x in inputList if x is not None), 0)
+			else:
+				# get the previous value
+				inputListFixed[idx] = inputListFixed[idx-1]
+	deltaList = [y-x for (x,y) in pairwise(inputListFixed)]
+	for idx, delta in enumerate(deltaList):
+		absolute = inputList[idx+1]
+		if absolute is None or (idx == 0 and inputList[0] is None):
+			deltaString = deltaString + '❌'
+		elif delta < 0 and absolute < 0:
+			deltaString = deltaString + '🔻'
+		elif delta < 0 and absolute >= 0:
+			deltaString = deltaString + '🔽'
+		elif delta > 0 and absolute < 0:
+			deltaString = deltaString + '🔺'
+		elif delta > 0 and absolute >= 0:
+			deltaString = deltaString + '🔼'
+		else:
+			deltaString = deltaString + '▪️'
+	# fallback if input has missing elements
+	missingfromstart = 3 - len(deltaString) # desired length hard coded
+	deltaString = ("❌" * missingfromstart) + deltaString
+	return deltaString
+
+MARKET_STATE_EMOJI = {'REGULAR': '🟢', 'PRE': '🟠', 'POST': '🟠'} # any other state is closed: 🔴
+
+def prepare_profile_payload(service, user, ticker):
+	"""The .SYMBOL report: what Yahoo knows about one company. A field Yahoo did not send leaves its line out."""
+	ticker_orig = markets.transform_to_yahoo(ticker)
+	now = datetime.datetime.now()
+	payload = []
+	ticker, market_data = yahoo.resolve(ticker_orig, lambda symbol: yahoo.fetch_detail(symbol, 600))
+	if ticker not in market_data:
+		return [f"{user} 🛑 Beep Boop. I could not find {ticker_orig}"]
+	data = market_data[ticker]
+	code = ticker.split('.')[0]
+	title = data.get('profile_title', '')
+	state = data.get('marketState', '').rstrip()
+	currency = data.get('currency', '')
+	exchange = data.get('profile_exchange', '').replace('NasdaqCM', 'Nasdaq').replace('NasdaqGS', 'Nasdaq').replace('NYSEArca', 'NYSE')
+	ticker_link = links.finance_link(ticker, exchange, service, brief=False)
+	gfinance_link = links.gfinance_link(ticker, exchange, service, brief=True, text='google') # all three before exchange is renamed below
+	sws_link = links.link(simplywallst.get_url(ticker, title, exchange), 'simplywall.st', service)
+	macrotrends_link = links.link('https://www.google.com/search?q=site:macrotrends.net+' + title + '+PE Ratio+' + code + '&btnI', 'macrotrends', service)
+	if 'profile_bio' in data:
+		payload.append(formats.make_paragraphs(data['profile_bio']))
+
+	if exchange == 'ASX':
+		exchange_url = 'https://www2.asx.com.au/markets/company/' + code
+	elif exchange == 'HKSE':
+		exchange_url = 'https://www.hkex.com.hk/Market-Data/Securities-Prices/Equities/Equities-Quote?sym=' + code + '&sc_lang=en'
+	elif 'Nasdaq' in exchange:
+		exchange_url = 'https://www.nasdaq.com/market-activity/stocks/' + ticker.lower()
+	elif exchange == 'NYSE':
+		exchange_url = 'https://www.nyse.com/quote/XNYS:' + ticker
+	elif exchange == 'Taiwan':
+		exchange = 'TWSE'
+		exchange_url = 'https://mis.twse.com.tw/stock/fibest.jsp?stock=' + code + '&lang=en_us'
+	elif exchange == 'Tokyo':
+		exchange = 'JPX'
+		exchange_url = 'https://quote.jpx.co.jp/jpx/template/quote.cgi?F=tmp/e_stock_detail&MKTN=T&QCODE=' + code
+	else:
+		exchange_url = 'https://www.google.com/search?q=stock+exchange+' + exchange + '+' + code + '&btnI'
+	exchange_link = links.link(exchange_url, exchange, service)
+	location = [data[key] for key in ('profile_city', 'profile_state', 'profile_country') if key in data]
+	if payload:
+		payload.append("")
+
+	if location:
+		payload.append(webhook.bold("Location:", service) + " " + ', '.join(location))
+	if 'profile_industry' in data and 'profile_sector' in data:
+		payload.append(webhook.bold("Classification: ", service) + f"{data['profile_industry']}, {data['profile_sector']}")
+	if 'profile_employees' in data:
+		payload.append(webhook.bold("Employees: ", service) + f"{data['profile_employees']:,}")
+	if 'profile_website' in data:
+		sites = [links.link(data['profile_website'], formats.strip_url(data['profile_website']), service), exchange_link]
+		if 'NYSE' in exchange or 'Nasdaq' in exchange:
+			sites += [links.link('https://finviz.com/quote.ashx?t=' + ticker, 'finviz', service), macrotrends_link,
+				links.link('https://seekingalpha.com/symbol/' + ticker, 'seekingalpha', service)]
+		elif exchange == 'ASX':
+			sites.append(links.link('https://www.shortman.com.au/stock?q=' + code.lower(), 'shortman', service))
+		payload.append(webhook.bold("Links: ", service) + ' | '.join(sites))
+
+	if payload:
+		payload.append("")
+
+	if 'currency' in data and 'market_cap' in data:
+		payload.append(webhook.bold("Mkt cap:", service) + f" {yahoo.market_cap_currency(data)} {formats.humanUnits(data['market_cap'])}") # GBP, not pence
+	if all(key in data for key in ('shareholder_equity', 'total_debt', 'total_cash', 'profile_industry')) and data['shareholder_equity'] and 'Bank' not in data['profile_industry']:
+		net_debt_equity_ratio = round((data['total_debt'] - data['total_cash']) / data['shareholder_equity'] * 100)
+		if net_debt_equity_ratio > 0:
+			payload.append(webhook.bold("Net debt/equity ratio:", service) + f" {net_debt_equity_ratio}%" + ('⚠️ ' if net_debt_equity_ratio > 40 else ''))
+	if 'earnings_date' in data:
+		earnings_date = datetime.datetime.fromtimestamp(data['earnings_date'])
+		if earnings_date > now:
+			payload.append(webhook.bold("Earnings date:", service) + f" {earnings_date:%b %d}")
+	if data.get('dividend', 0) > 0:
+		payload.append(webhook.bold("Dividend:", service) + f" {data['dividend']}%")
+		if 'ex_dividend_date' in data:
+			ex_dividend_date = datetime.datetime.fromtimestamp(data['ex_dividend_date'])
+			if ex_dividend_date > now:
+				payload.append(webhook.bold("Ex-dividend date:", service) + f" {ex_dividend_date:%b %d}")
+	cashflow = data.get('free_cashflow', data.get('operating_cashflow'))
+	if cashflow:
+		payload.append(webhook.bold("Cashflow positive:", service) + (" no ⚠️ " if cashflow < 0 else " yes"))
+	if 'net_income' in data:
+		payload.append(webhook.bold("Profitable:", service) + (" no ⚠️ " if data['net_income'] <= 0 else " yes"))
+	else:
+		payload.append(webhook.bold("Profitable:", service) + " unknown ⚠️")
+
+	if payload:
+		payload.append("")
+
+	if 'earningsQ' in data:
+		deltas = [(doDelta(data[key]), label) for key, label in (('revenueQ', 'quarterly revenue'), ('earningsQ', 'quarterly earnings'),
+			('revenueY', 'annual revenue'), ('earningsY', 'annual earnings'))]
+		payload += [f"{delta}\t{label} delta" for delta, label in deltas] # doDelta always gives at least three symbols
+
+	if payload:
+		payload.append("")
+
+	for key, label in (('revenueEstimateY', 'Revenue forecast (1Y):'), ('earningsEstimateY', 'Earnings forecast (1Y):')):
+		if key in data:
+			growth = int(round(data[key]))
+			payload.append(webhook.bold(label, service) + (f" +{growth}% " if growth > 0 else f" {growth}% ⚠️ "))
+	if 'insiderBuy' in data:
+		if data['insiderBuy'] > data['insiderSell']:
+			payload.append(webhook.bold("Net insider action (3M):", service) + f" Buy {currency} {formats.humanUnits(data['insiderBuyValue'])} ")
+		elif data['insiderBuy'] < data['insiderSell']:
+			payload.append(webhook.bold("Net insider action (3M):", service) + f" Sell {currency} {formats.humanUnits(data['insiderSellValue'])}⚠️ ")
+	if 'short_percent' in data:
+		payload.append(webhook.bold("Shorted stock:", service) + f" {data['short_percent']}%" + ('⚠️ ' if data['short_percent'] > 10 else ''))
+	if 'recommend' in data:
+		payload.append(webhook.bold("Score:", service) + f" {data['recommend_index']} {data['recommend'].replace('_', ' ')} ({data['recommend_analysts']} analysts)")
+
+	if payload:
+		payload.append("")
+
+	if 'regularMarketPrice' in data:
+		price_line = webhook.bold("Price:", service) + f" {currency} {data['regularMarketPrice']:,.2f}"
+		if state != 'REGULAR' and 'prePostMarketPrice' in data:
+			price_line += f" ({data['prePostMarketPrice']:,.2f} after hrs)"
+		payload.append(price_line)
+	trailing_pe = str(int(round(data['price_to_earnings_trailing']))) if 'price_to_earnings_trailing' in data else 'N/A ⚠️ '
+	if data.get('net_income', 0) > 0:
+		payload.append(webhook.bold("Trailing P/E:", service) + f" {trailing_pe}")
+	if data.get('netIncomeToCommon', 0) > 0 and data.get('sharesOutstanding') and 'regularMarketPrice' in data:
+		current_pe = round(data['regularMarketPrice'] / (data['netIncomeToCommon'] / data['sharesOutstanding']))
+		payload.append(webhook.bold("Current P/E:", service) + f" {current_pe}")
+	if 'price_to_earnings_forward' in data:
+		forward_pe = int(round(data['price_to_earnings_forward']))
+		warn = forward_pe < 0
+		if 'profile_industry' in data: # software is dear: it warns above 100, the rest above 30 or above their trailing P/E
+			warn = warn or forward_pe > (100 if 'Software' in data['profile_industry'] else 30)
+			warn = warn or ('price_to_earnings_trailing' in data and forward_pe > int(trailing_pe))
+		payload.append(webhook.bold("Forward P/E:", service) + f" {forward_pe} " + ('⚠️ ' if warn else ''))
+	if 'price_to_earnings_peg' in data:
+		payload.append(webhook.bold("PEG ratio:", service) + f" {round(data['price_to_earnings_peg'], 1)}")
+	if 'price_to_sales' in data:
+		price_to_sales = round(data['price_to_sales'], 1)
+		payload.append(webhook.bold("PS ratio:", service) + f" {price_to_sales} " + ('⚠️ ' if price_to_sales > 15 else ''))
+
+	if payload:
+		payload.append("")
+
+	try:
+		price_history, _ = yahoo.price_history(ticker, graph=False) # the percentages only: the profile has no chart
+	except RuntimeError as e: # no chart data: the profile goes out without these lines
+		print(ticker, e, file=sys.stderr)
+		price_history = {}
+	for interval in ('5Y', '1Y', '1M', '1D'):
+		if interval in price_history:
+			percent = price_history[interval]
+			payload.append(f"{formats.get_emoji(percent)} {webhook.bold(interval + ':', service)} {percent}%")
+	if state != 'REGULAR':
+		for key, label in (('percent_change_premarket', 'Pre-market:'), ('percent_change_postmarket', 'Post-market:')):
+			if key in data:
+				payload.append(f"{formats.get_emoji(data[key])} {webhook.bold(label, service)} {data[key]:,}%")
+				break
+
+	if 'profile_website' in data and config.prefs.hyperlinkFooter and config.prefs.hyperlink:
+		if payload:
+			payload.append("")
+		payload.append(f"{gfinance_link} | {links.yahoo_link(ticker, service, brief=True, text='yahoo')} | {sws_link}")
+	heading = f"{webhook.bold(title, service)} ({ticker_link}) {MARKET_STATE_EMOJI.get(state, '🔴') if state else ''}"
+	if ticker_orig == ticker:
+		payload.insert(0, heading)
+	else:
+		payload[:0] = [f"I could not find {ticker_orig} but I found {ticker_link}:", "", heading]
+	payload = [i[0] for i in groupby(payload)] # de-dupe white space
+	if len(payload) < 2:
+		payload.append("no data found")
+	return payload
diff --git a/lib/reports/rankings.py b/lib/reports/rankings.py
new file mode 100644
index 00000000..68911256
--- /dev/null
+++ b/lib/reports/rankings.py
@@ -0,0 +1,164 @@
+"""Ranked lists of the stocks followed: .marketcap, .pe, .fpe and .peg (top, bottom, negative), .beta, .buy and .sell."""
+import sys
+from lib import config
+from lib import watchlist
+from lib import formats
+from lib import links
+from lib import markets
+from lib import webhook
+from lib import yahoo
+
+def last_col(line):
+	"""Sort key: the number a report line ends with; a line without one sorts first."""
+	try:
+		return float(line.split()[-1])
+	except ValueError:
+		return float('-inf')
+
+def prepare_marketcap_payload(service, action='top', length=15):
+	"""The `length` tracked stocks of largest (action 'top') or smallest ('bottom') market cap, compared in US dollars at Yahoo's
+	exchange rates: a market cap in won or yen is a far bigger number. Any other action is one ticker's, in its own currency too."""
+	if action not in ('top', 'bottom'):
+		ticker = markets.transform_to_yahoo(action.upper())
+		data = yahoo.add_usd_market_caps(yahoo.fetch_detail(ticker, 600)).get(ticker, {})
+		if 'market_cap' not in data:
+			return [f"Mkt cap not found for {ticker}"]
+		currency = yahoo.market_cap_currency(data) or ''
+		cap = f"{currency} {formats.humanUnits(data['market_cap'])}".strip()
+		if currency != 'USD' and 'market_cap_usd' in data:
+			cap += f" (USD {formats.humanUnits(data['market_cap_usd'])})"
+		link = links.finance_link(ticker, data.get('profile_exchange', ''), service)
+		return [formats.security_line(markets.flag_from_ticker(ticker), data.get('profile_title', ticker), link) + f" mkt cap: {cap}"]
+	caps = [] # (market cap in US dollars, line)
+	for ticker, data in yahoo.add_usd_market_caps(yahoo.fetch(watchlist.with_holdings())).items():
+		if 'market_cap_usd' not in data:
+			print(ticker, "no market cap" if 'market_cap' not in data else f"no exchange rate for {data.get('currency')}", file=sys.stderr)
+			continue
+		link = links.finance_link(ticker, data['profile_exchange'], service)
+		caps.append((data['market_cap_usd'], formats.security_line(markets.flag_from_ticker(ticker), data['profile_title'], link) + f" mkt cap: {formats.humanUnits(data['market_cap_usd'])}"))
+	if not caps:
+		return ['No market cap data found']
+	caps.sort(key=lambda cap: cap[0], reverse=action == 'top')
+	return [webhook.bold(f'{action.title()} {length} tracked stocks by market cap (USD)', service)] + [line for _, line in caps[:length]]
+
+def prepare_rating_payload(service, action, length=15):
+	"""(payload, chart) of the `length` tracked stocks rated `action` ('buy' or 'sell') by min_analysts or more analysts (fewer,
+	and the consensus says little: ASX small caps covered by two to six brokers are nearly all a perfect 1.0 buy), the strongest
+	rating first (1 is a strong buy, 5 a strong sell); on equal ratings, the one more analysts give first."""
+	found = [] # (rating, analysts, line, ticker)
+	for ticker, data in yahoo.fetch_details(watchlist.with_holdings()).items():
+		if 'recommend' not in data or data['recommend_analysts'] < config.prefs.min_analysts:
+			continue
+		recommend = data['recommend'].replace('_', ' ')
+		if (action == 'buy' and 'buy' in recommend) or (action == 'sell' and recommend in ('sell', 'underperform')):
+			rating, analysts = round(data['recommend_index'], 2), data['recommend_analysts']
+			link = links.finance_link(ticker, data['profile_exchange'], service)
+			found.append((rating, analysts, f"{markets.flag_from_ticker(ticker)} {data['profile_title']} ({link}) {rating} ({analysts} analysts)", ticker))
+	found = sorted(found, key=lambda row: (row[0] if action == 'buy' else -row[0], -row[1]))[:length]
+	payload = [line for _, _, line, _ in found]
+	if not payload:
+		return [f"No tracked stocks rated {action} by {config.prefs.min_analysts}+ analysts"], None
+	message = f"Top {length} analyst {action} ratings for tracked stocks ({config.prefs.min_analysts}+ analysts)"
+	shown = [(ticker.split('.')[0], rating) for rating, _, _, ticker in found]
+	low, high = min(r for _, r in shown), max(r for _, r in shown)
+	spread = (high - low) or 0.2
+	# zoom the axis onto the ratings so the bars fill the panel (ratings only run 1-5)
+	chart = {'rows': shown, 'title': message, 'subtitle': 'Mean analyst rating: 1 = strong buy, 5 = strong sell (axis zoomed)',
+		'value_fmt': lambda v: f"{v:g}", 'xlim': (max(0.9, low - spread * 0.9), min(5.1, high + spread * 0.25))}
+	return [webhook.bold(message, service)] + payload, chart
+
+def prepare_value_payload(service, action='pe', ticker_select=None, length=15):
+	"""(payload, chart): the tracked stocks by a ratio (`action`), or one ticker's. A forward P/E is analysts' estimate, so its lists
+	take only stocks that min_analysts or more analysts cover, as .buy does."""
+	payload = []
+	chart_rows = [] # (ticker, ratio)
+	tickers = [ticker_select] if ticker_select else watchlist.with_holdings()
+	market_data = yahoo.fetch(tickers)
+	if action in ('peg', 'bottom peg', 'negative peg'): # the PEG is in Yahoo's details
+		market_data |= yahoo.fetch_details(market_data, strict=bool(ticker_select))
+	covered = None # for a forward P/E list: the tickers enough analysts cover
+	if 'forward' in action and not ticker_select:
+		covered = {ticker for ticker, detail in yahoo.fetch_details(market_data).items() if detail.get('recommend_analysts', 0) >= config.prefs.min_analysts}
+	for ticker in market_data:
+		if covered is not None and ticker not in covered:
+			continue
+		trailing_ratio = None
+		forward_ratio = None
+		peg_ratio = None
+		ratio = None
+		if action in ('pe', 'bottom pe', 'forward pe', 'bottom forward pe'):
+			trailing_ratio = market_data.get(ticker).get('price_to_earnings_trailing')
+			forward_ratio = market_data.get(ticker).get('price_to_earnings_forward')
+			if action in ('forward pe', 'bottom forward pe'):
+				if not ticker_select and forward_ratio is not None and forward_ratio < 0:
+					continue
+				ratio = forward_ratio
+			else:
+				if not ticker_select and trailing_ratio is not None and trailing_ratio < 0:
+					continue
+				ratio = trailing_ratio
+		elif action == 'negative forward pe':
+			forward_ratio = market_data.get(ticker).get('price_to_earnings_forward')
+			if not ticker_select and (forward_ratio is None or forward_ratio >= 0):
+				continue
+			ratio = forward_ratio
+		elif action in ('peg', 'bottom peg'):
+			peg_ratio = market_data.get(ticker).get('price_to_earnings_peg')
+			if not ticker_select and peg_ratio is not None and peg_ratio < 0:
+				continue
+			ratio = peg_ratio
+		elif action == 'negative peg':
+			peg_ratio = market_data.get(ticker).get('price_to_earnings_peg')
+			if not ticker_select and (peg_ratio is None or peg_ratio >= 0):
+				continue
+			ratio = peg_ratio
+		if ratio is None:
+			continue
+		profile_title = market_data.get(ticker).get('profile_title')
+		ticker_link = links.finance_link(ticker, market_data.get(ticker).get('profile_exchange'), service)
+		flag = markets.flag_from_ticker(ticker)
+		if ticker_select and action in {'pe', 'forward pe'}:
+			parts = []
+			if trailing_ratio is not None:
+				parts.append(f"trailing {trailing_ratio}")
+			if forward_ratio is not None:
+				parts.append(f"forward {forward_ratio}")
+			payload.append(f"{flag} {profile_title} ({ticker_link}) PE: {', '.join(parts)}")
+		else:
+			payload.append(f"{flag} {profile_title} ({ticker_link}) {ratio}")
+			chart_rows.append((ticker.split('.')[0], float(ratio)))
+	payload.sort(key=last_col)
+	chart_rows.sort(key=lambda row: row[1])
+	chart = None
+	if not ticker_select:
+		heading_type = "Bottom" if 'bottom' in action else "Top"
+		heading_trail = action.replace('bottom ', '')
+		if 'bottom' in action:
+			payload.reverse()
+			chart_rows.reverse()
+		payload = payload[:length]
+		coverage = f" ({config.prefs.min_analysts}+ analysts)" if covered is not None else ''
+		if not payload:
+			payload = [f"No tracked stocks with {config.prefs.min_analysts}+ analysts have a {heading_trail} ratio" if covered is not None else f"{action}: no data found"]
+		else:
+			heading = f'{heading_type} {length} tracked stocks by {heading_trail} ratio{coverage}'
+			payload.insert(0, f"{webhook.bold(heading, service)}")
+			chart = {'rows': chart_rows[:length], 'title': heading, 'value_fmt': lambda v: f"{v:g}"}
+	return payload, chart
+
+def prepare_beta_payload(service):
+	"""(payload, chart) of the tracked stocks with a beta over 1.5 and a market cap under US$2B, the highest beta first."""
+	found = [] # (beta, line, chart row)
+	for ticker, data in yahoo.add_usd_market_caps(yahoo.fetch_details(watchlist.with_holdings())).items():
+		if 'beta' not in data:
+			continue
+		beta = round(data['beta'], 1)
+		if beta > 1.5 and data.get('market_cap_usd', float('inf')) < 2e9:
+			link = links.finance_link(ticker, data['profile_exchange'], service)
+			found.append((beta, formats.security_line(markets.flag_from_ticker(ticker), data['profile_title'], link) + f" {beta} β", (ticker.split('.')[0], float(beta))))
+	if not found:
+		return ['No tracked stocks have a beta over 1.5 and mkt cap under US$2B'], None
+	found.sort(key=lambda row: row[0], reverse=True)
+	heading = 'Beta over 1.5 and mkt cap under US$2B'
+	chart = {'rows': [row for _, _, row in found], 'title': heading, 'value_fmt': lambda v: f"{v:g}"}
+	return [webhook.bold(heading, service)] + [line for _, line, _ in found], chart
diff --git a/rating.py b/lib/reports/rating.py
old mode 100755
new mode 100644
similarity index 95%
rename from rating.py
rename to lib/reports/rating.py
index 816f31ee..9ac394d9
--- a/rating.py
+++ b/lib/reports/rating.py
@@ -1,6 +1,4 @@
-#!/usr/bin/env python3
-
-import sys
+"""Analysts' consensus rating changes since the last scheduled run (scheduled: rating.py). .buy and .sell are in rankings."""
 from lib import webhook
 from lib import state
 from lib import links
@@ -70,6 +68,3 @@ def get_emoji(old, new):
 
 def job(argv):
 	return 'rating.py', {}
-
-if __name__ == "__main__":
-	webhook.cli(sys.argv[0], job, lambda_handler)
diff --git a/reminder.py b/lib/reports/reminder.py
old mode 100755
new mode 100644
similarity index 96%
rename from reminder.py
rename to lib/reports/reminder.py
index 7aaaf741..4d4f87bf
--- a/reminder.py
+++ b/lib/reports/reminder.py
@@ -1,6 +1,4 @@
-#!/usr/bin/env python3
-
-import sys
+"""Reminders for the day: those in reminder.json, and the Australian end of financial year (scheduled: reminder.py)."""
 import datetime
 import zoneinfo
 from dateutil.relativedelta import relativedelta
@@ -74,6 +72,3 @@ def prepare_finance_calendar_payload(service):
 
 def job(argv):
 	return 'reminder.py', {}
-
-if __name__ == "__main__":
-	webhook.cli(sys.argv[0], job, lambda_handler)
diff --git a/shorts.py b/lib/reports/shorts.py
old mode 100755
new mode 100644
similarity index 95%
rename from shorts.py
rename to lib/reports/shorts.py
index 6819b4d1..2230ef27
--- a/shorts.py
+++ b/lib/reports/shorts.py
@@ -1,6 +1,4 @@
-#!/usr/bin/env python3
-
-import sys
+""".shorts: the most shorted stocks followed, from Short Man (ASX) and Yahoo (scheduled: shorts.py)."""
 from lib import config
 from lib import webhook
 from lib import links
@@ -52,6 +50,3 @@ def prepare_shorts_payload(service, market_data):
 
 def job(argv):
 	return 'shorts.py', {}
-
-if __name__ == "__main__":
-	webhook.cli(sys.argv[0], job, lambda_handler)
diff --git a/trades.py b/lib/reports/trades.py
old mode 100755
new mode 100644
similarity index 97%
rename from trades.py
rename to lib/reports/trades.py
index 04cd4390..16cf07e5
--- a/trades.py
+++ b/lib/reports/trades.py
@@ -1,5 +1,4 @@
-#!/usr/bin/env python3
-
+""".trades: Sharesight buys and sells; a scheduled run reports only new ones (scheduled: trades.py)."""
 import random, sys
 import datetime
 
@@ -105,6 +104,3 @@ def prepare_trade_payload(service, trades):
 
 def job(argv):
 	return 'trades.py', {}
-
-if __name__ == "__main__":
-	webhook.cli(sys.argv[0], job, lambda_handler)
diff --git a/lib/worker.py b/lib/worker.py
index 27b2732e..43456348 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -13,16 +13,12 @@
 from lib import markets
 from lib import render
 from lib import webhook
-from lib import reports
+from lib import buttons
 from lib import charts
+from lib.reports import about, cal, history, holdings, performance, price, profile, rankings, shorts, trades
 from lib import telegram
 from lib import sharesight
 from lib import jobs
-import cal
-import performance
-import price
-import shorts
-import trades
 
 class TypingIndicator:
 	"""Telegram's "typing…" while a with block runs, sent again every 7 seconds, at most max_loops times. It stops however the block
@@ -106,12 +102,12 @@ def process_job_callback(service, callback_id, chat_id, message_id, data):
 		command = jobs.current.pick(number, crc)
 	except (ValueError, jobs.ScheduleError, AttributeError):
 		telegram.answerCallbackQuery(callback_id, 'Buttons updated, try again')
-		payload, markup = reports.prepare_jobs_payload(service, '') # the list as it is now, in place of the stale one
+		payload, markup = about.prepare_jobs_payload(service, '') # the list as it is now, in place of the stale one
 		webhook.editMessageText(chat_id, message_id, '\n'.join(payload), markup or {'inline_keyboard': []})
 		return
 	telegram.answerCallbackQuery(callback_id, f'Running {jobs.display(command[0])}…'[:100])
 	try:
-		webhook.payload_wrapper(service, webhook.chat_url(service, chat_id), [reports.job_line(jobs.current, *jobs.current.run_command(command))], chat_id)
+		webhook.payload_wrapper(service, webhook.chat_url(service, chat_id), [about.job_line(jobs.current, *jobs.current.run_command(command))], chat_id)
 	except Exception as e:
 		webhook.report_error(e, service, chat_id, context='jobs')
 
@@ -120,16 +116,16 @@ def process_callback(service, callback_id, chat_id, message_id, data):
 	if data.startswith('j|'):
 		return process_job_callback(service, callback_id, chat_id, message_id, data)
 	parts = data.split('|', 3)
-	valid = len(parts) == 4 and parts[0] == 'c' and parts[1] in charts.BUTTONS and parts[2] in charts.BUTTONS[parts[1]]
+	valid = len(parts) == 4 and parts[0] == 'c' and parts[1] in buttons.BUTTONS and parts[2] in buttons.BUTTONS[parts[1]]
 	# answering now shows Telegram's toast bubble ("Loading 3M…") while the new chart is built
-	telegram.answerCallbackQuery(callback_id, f"Loading {charts.PERIODS[parts[2]][0]}…" if valid else None)
+	telegram.answerCallbackQuery(callback_id, f"Loading {buttons.PERIODS[parts[2]][0]}…" if valid else None)
 	if not valid:
 		return
 	try:
 		_, kind, period, ref = parts
-		tickers = charts.resolve(ref)
+		tickers = buttons.resolve(ref)
 		caption, image = charts.build(kind, tickers, period, service)
-		webhook.editMessageMedia(chat_id, message_id, image, caption, charts.keyboard(kind, tickers, period))
+		webhook.editMessageMedia(chat_id, message_id, image, caption, buttons.keyboard(kind, tickers, period))
 	except Exception as e:
 		webhook.report_error(e, service, chat_id, context='chart')
 
@@ -207,14 +203,14 @@ def on_watchlist(request, m):
 		request.say([f'"{action}" is not a valid watchlist action'])
 		return
 	with request.typing():
-		payload = reports.prepare_watchlist(request.service, request.user, action, ticker)
+		payload = holdings.prepare_watchlist(request.service, request.user, action, ticker)
 	request.say(payload)
 
 def on_help(request, m):
 	markup = None
 	if request.service == 'telegram' and int(request.chat_id) > 0: # positive chat ids are DMs; groups are negative
 		markup = charts.command_keyboard()
-	request.say(reports.prepare_help(request.service, request.bot_name), reply_markup=markup)
+	request.say(about.prepare_help(request.service, request.bot_name), reply_markup=markup)
 	if markup:
 		charts.mark_keyboard(request.chat_id) # this chat now has the current keyboard
 
@@ -243,7 +239,7 @@ def on_calendar(request, m):
 		cal.lambda_handler(request.chat_id, days, request.service, ticker, interactive=True, event=event)
 
 def on_performance(request, m):
-	days, portfolio = period_and_portfolio(m.group('args'), charts.PERIODS['m'][1]) # default view is the 1M button
+	days, portfolio = period_and_portfolio(m.group('args'), buttons.PERIODS['m'][1]) # default view is the 1M button
 	if days > 0:
 		with request.typing():
 			performance.lambda_handler(request.chat_id, days, request.service, request.user, portfolio, interactive=True)
@@ -286,12 +282,12 @@ def on_trades(request, m):
 
 def on_holdings(request, m):
 	with request.typing():
-		payload = reports.prepare_holdings_payload((m.group(1) or '').strip() or None, request.service)
+		payload = holdings.prepare_holdings_payload((m.group(1) or '').strip() or None, request.service)
 	request.say(payload)
 
 def on_marketcap(request, m):
 	with request.typing():
-		payload = reports.prepare_marketcap_payload(request.service, m.group(1) or 'top', length=15)
+		payload = rankings.prepare_marketcap_payload(request.service, m.group(1) or 'top', length=15)
 	request.say(payload)
 
 def on_ratio(kind, negative=False, chart=False):
@@ -307,13 +303,13 @@ def handler(request, m):
 		elif arg and arg != 'top':
 			ticker = markets.transform_to_yahoo(arg.upper())
 		with request.typing():
-			payload, spec = reports.prepare_value_payload(request.service, action, ticker, length=15)
+			payload, spec = rankings.prepare_value_payload(request.service, action, ticker, length=15)
 		request.deliver(payload, spec if chart else None)
 	return handler
 
 def on_beta(request, m):
 	with request.typing():
-		payload, chart = reports.prepare_beta_payload(request.service)
+		payload, chart = rankings.prepare_beta_payload(request.service)
 	request.deliver(payload, chart)
 
 def on_pie(request, m):
@@ -322,22 +318,22 @@ def on_pie(request, m):
 	if words and words[-1].lower() in ('sector', 'sectors', 'industry', 'industries'):
 		by = 'sector' if words.pop().lower().startswith('sector') else 'industry'
 	with request.typing():
-		payload, chart = reports.prepare_pie(' '.join(words), by, request.service)
+		payload, chart = holdings.prepare_pie(' '.join(words), by, request.service)
 	request.deliver(payload, chart)
 
 def on_rating(request, m):
 	"""The analysts' strongest buy ratings (.buy), sell ratings (.sell), or both (.buy & sell, a keyboard button), as asked."""
 	for action in dict.fromkeys(word.lower() for word in m.group('first', 'second') if word): # each once, in the order asked
 		with request.typing():
-			payload, chart = reports.prepare_rating_payload(request.service, action, length=15)
+			payload, chart = rankings.prepare_rating_payload(request.service, action, length=15)
 		request.deliver(payload, chart)
 
 def on_compare(request, m):
 	with request.typing():
-		payload, image, tickers, days = reports.prepare_compare(request.service, (m.group('args') or '').split())
+		payload, image, tickers, days = history.prepare_compare(request.service, (m.group('args') or '').split())
 	markup = None
 	if image and request.service == 'telegram':
-		markup = charts.keyboard('c', tickers, charts.period_for_days(days) if days else 'x')
+		markup = buttons.keyboard('c', tickers, buttons.period_for_days(days) if days else 'x')
 	request.send(payload, image, markup)
 
 def on_history(request, m):
@@ -347,7 +343,7 @@ def on_history(request, m):
 	ticker = markets.transform_to_yahoo(m.group('ticker').upper())
 	with request.typing():
 		caption, image = charts.build('h', [ticker], None, request.service)
-	request.send(caption.split('\n'), image, charts.keyboard('h', [ticker], 'x') if image and request.service == 'telegram' else None)
+	request.send(caption.split('\n'), image, buttons.keyboard('h', [ticker], 'x') if image and request.service == 'telegram' else None)
 
 def on_plan(request, m):
 	if m.group(1) and not m.group(1).startswith('@'):
@@ -365,17 +361,17 @@ def on_who(request, m):
 		request.say([".who: please try again specifying a ticker"])
 		return
 	with request.typing():
-		payload = reports.prepare_who(request.service, re.split(r'[.:]', m.group('ticker').upper())[0])
+		payload = holdings.prepare_who(request.service, re.split(r'[.:]', m.group('ticker').upper())[0])
 	request.say(payload)
 
 def on_jobs(request, m):
 	with request.typing(max_loops=10):
-		payload, markup = reports.prepare_jobs_payload(request.service, m.group('args'))
+		payload, markup = about.prepare_jobs_payload(request.service, m.group('args'))
 	request.say(payload, reply_markup=markup)
 
 def on_profile(request, m):
 	with request.typing():
-		payload = reports.prepare_profile_payload(request.service, request.user, m.group('ticker').upper())
+		payload = profile.prepare_profile_payload(request.service, request.user, m.group('ticker').upper())
 	request.say(payload)
 
 COMMANDS = [ # (pattern, function): tried in this order, the first match answers. .SYMBOL takes anything, so it stays last
diff --git a/tests/common.py b/tests/common.py
index 225052fd..834490e3 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -10,8 +10,8 @@
 os.environ.update(telegramOutgoingWebhook='https://example.com/telegram', telegramOutgoingToken='tok', telegramChatID='1')
 
 import bot # first, and once: it monkey-patches with gevent, which must happen before requests imports ssl and never mid-test
-from lib import cache, config, state, webhook, formats, links, markets, render, yahoo, sharesight, shortman, worker, charts, telegram, reports, watchlist, words
-import performance, shorts, price, reminder, cal, trades, milestone, rating
+from lib import buttons, cache, config, state, webhook, formats, links, markets, render, yahoo, sharesight, shortman, worker, charts, telegram, watchlist, words
+from lib.reports import about, cal, history, holdings, milestone, performance, price, profile, rankings, rating, reminder, shorts, trades
 
 REAL = {(m.__name__, n): getattr(m, n) for m, n in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (yahoo, 'price_history'),
 	(sharesight, 'get_portfolios'), (sharesight, 'get_trades'), (sharesight, 'get_performance'))}
@@ -82,11 +82,11 @@ def setUp(self):
 		dates = pd.date_range(end=pd.Timestamp.today().normalize(), periods=400)
 		self.patch(yahoo, 'price_series', lambda t: ('Co ' + t, 'NASDAQ', pd.Series(range(100, 500), index=dates, name=t, dtype=float)))
 		self.patch(shortman, 'fetch', lambda md: md)
-		holdings = [{'id': i, 'instrument': {'code': t, 'market_code': 'NASDAQ'}, 'portfolio': {'name': 'Rob'},
+		held = [{'id': i, 'instrument': {'code': t, 'market_code': 'NASDAQ'}, 'portfolio': {'name': 'Rob'},
 					 'capital_gain_percent': str(i * 4 - 40), 'value': '1000'} for i, t in enumerate(TICKERS)]
 		self.patch(sharesight, 'get_portfolios', lambda: {'Rob': 1})
 		self.patch(sharesight, 'get_performance', lambda pid, d, *a: {'report': {
-			'holdings': holdings, 'currency_gain_percent': '0', 'capital_gain_percent': '2.5', 'total_gain_percent': '0'}})
+			'holdings': held, 'currency_gain_percent': '0', 'capital_gain_percent': '2.5', 'total_gain_percent': '0'}})
 		self.patch(sharesight, 'get_trades', lambda name, pid, days=7: [{
 			'id': 1, 'transaction_type': 'BUY', 'symbol': 'T1', 'market': 'NASDAQ', 'portfolio': name,
 			'transaction_date': time.strftime('%Y-%m-%d'), 'quantity': '10', 'price': '5.0', 'value': '50', 'brokerage_currency_code': 'USD', 'holding_id': 7}])
diff --git a/tests/test_config.py b/tests/test_config.py
index 6bba073e..fa6a0dea 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -5,8 +5,8 @@
 from lib import config
 from util import check_config, split_env
 import bot, cal
-from lib import cache, reports, sharesight, worker, yahoo
-import performance, price, shorts, trades
+from lib import cache, sharesight, worker, yahoo
+from lib.reports import performance, price, shorts, trades
 
 REPO = Path(config.__file__).resolve().parent.parent
 DEFAULTS = {'past_days': 28, 'future_days': 30, 'price_percent': 5.0, 'shorts_percent': 15, 'cache_seconds': 43200,
@@ -223,7 +223,7 @@ def test_nothing_copies_the_config_or_freezes_a_preference(self):
 		def reads_prefs(node):
 			return any(isinstance(n, ast.Attribute) and n.attr == 'prefs' and isinstance(n.value, ast.Name) and n.value.id == 'config' for n in ast.walk(node))
 		offenders = []
-		for path in sorted([*(REPO / 'lib').glob('*.py'), *REPO.glob('*.py'), *(REPO / 'util').glob('*.py')]):
+		for path in sorted([*(REPO / 'lib').rglob('*.py'), *REPO.glob('*.py'), *(REPO / 'util').glob('*.py')]):
 			tree = ast.parse(path.read_text())
 			for node in ast.walk(tree):
 				if isinstance(node, ast.ImportFrom) and node.module == 'lib.config':
@@ -261,7 +261,7 @@ def popen(args, **kw):
 		self.write('future_days=45\n')
 		config.reload()
 		with mock.patch.object(jobs.subprocess, 'Popen', popen):
-			import cleanup
+			from lib.reports import cleanup
 			self.assertEqual(jobs.run_child('cleanup.py', cleanup, {}), 0)
 		self.assertEqual((seen['future_days'], seen[config.PASSED]), ('45', '1'))
 
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 5a174530..52d9152c 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -91,7 +91,7 @@ def slow_fetch(tickers):
 			time.sleep(0.01)
 			return {t: market_entry(t) for t in tickers}
 		self.patch(yahoo, 'fetch', slow_fetch)
-		threads = [threading.Thread(target=reports.prepare_watchlist, args=('telegram', '@u', 'add', t)) for t in ('T1', 'T2')]
+		threads = [threading.Thread(target=holdings.prepare_watchlist, args=('telegram', '@u', 'add', t)) for t in ('T1', 'T2')]
 		for thread in threads:
 			thread.start()
 		for thread in threads:
diff --git a/tests/test_finbot.py b/tests/test_finbot.py
index 1914a4d9..692bf311 100644
--- a/tests/test_finbot.py
+++ b/tests/test_finbot.py
@@ -60,7 +60,7 @@ def test_the_bot_and_its_reports_print_only_to_stderr(self):
 		"""systemd discards finbot's stdout (StandardOutput=null), so a print there is never read. The util/ tools, run by
 		hand, may use it."""
 		offenders = []
-		for path in sorted([*Path(ROOT, 'lib').glob('*.py'), *Path(ROOT).glob('*.py')]):
+		for path in sorted([*Path(ROOT, 'lib').rglob('*.py'), *Path(ROOT).glob('*.py')]):
 			for node in ast.walk(ast.parse(path.read_text())):
 				if isinstance(node, ast.Call) and getattr(node.func, 'id', None) == 'print' and \
 						not any(keyword.arg == 'file' for keyword in node.keywords):
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
index 55b19333..79793b45 100644
--- a/tests/test_jobs.py
+++ b/tests/test_jobs.py
@@ -5,7 +5,7 @@
 import gevent
 from tests.common import *
 from lib import jobs
-import cal, cleanup, performance, podcasts, price
+from lib.reports import cal, cleanup, performance, podcasts, price
 
 UTC = zoneinfo.ZoneInfo('UTC')
 DEFAULT_SCHEDULE = Path(os.path.dirname(os.path.dirname(os.path.abspath(__file__)))) / 'defaults' / 'schedule.cron'
@@ -484,13 +484,13 @@ def test_list_in_chat_is_just_the_commands(self):
 		self.assertIn('6. price 7 (running)', '\n'.join(map(str, self.list_in_chat())))
 
 	def test_telegram_gets_a_button_per_command_and_slack_does_not(self):
-		payload, markup = reports.prepare_jobs_payload('telegram', '')
+		payload, markup = about.prepare_jobs_payload('telegram', '')
 		buttons = [b for row in markup['inline_keyboard'] for b in row]
 		self.assertEqual([b['text'] for b in buttons], ['1. performance 7', '2. performance 30', '3. performance 90', '4. performance 180',
 													  '5. performance 365', '6. price 7', '7. price close', '8. trades'])
 		for b in buttons:
 			self.assertLessEqual(len(b['callback_data'].encode()), 64)
-		payload, markup = reports.prepare_jobs_payload('slack', '')
+		payload, markup = about.prepare_jobs_payload('slack', '')
 		self.assertIsNone(markup)
 
 	def test_the_chat_command_passes_the_keyboard_on(self):
@@ -502,7 +502,7 @@ def test_the_chat_command_passes_the_keyboard_on(self):
 	def test_button_press_runs_that_command(self):
 		sent = []
 		self.patch(telegram, 'answerCallbackQuery', lambda cid, text=None: sent.append(text))
-		markup = reports.prepare_jobs_payload('telegram', '')[1]
+		markup = about.prepare_jobs_payload('telegram', '')[1]
 		data = next(b['callback_data'] for row in markup['inline_keyboard'] for b in row if b['text'].startswith('7.'))
 		self.sent.clear()
 		worker.process_callback('telegram', 'cb1', '55', '9', data)
@@ -547,7 +547,7 @@ def test_button_from_an_old_schedule_updates_the_message(self):
 		chat, mid, text, markup = self.edits[0]
 		self.assertEqual((chat, mid), ('55', '9')) # the message with the old buttons is edited in place
 		self.assertIn('7. price close', text)
-		self.assertEqual(markup, reports.prepare_jobs_payload('telegram', '')[1]) # with the current buttons
+		self.assertEqual(markup, about.prepare_jobs_payload('telegram', '')[1]) # with the current buttons
 		self.patch(jobs, 'current', None) # scheduler off: the list is replaced by the notice and the buttons removed
 		worker.process_callback('telegram', 'cb1', '55', '9', 'j|7|0000')
 		self.assertEqual(self.edits[-1][2:], ('Scheduled jobs are off (scheduler=False)', {'inline_keyboard': []}))
@@ -636,7 +636,7 @@ def test_scheduler_reports_a_child_that_ended_without_saying_why(self):
 	def test_a_report_script_loads_neither_pandas_nor_the_scheduler(self):
 		"""A job process imports only its own script: price used to pull in pandas and, through charts, reports and jobs, every
 		script there is."""
-		code = 'import sys, price, performance, trades, shorts, cal; print(sorted(m for m in ("pandas", "lib.jobs", "matplotlib") if m in sys.modules))'
+		code = 'import sys; from lib.reports import price, performance, trades, shorts, cal; print(sorted(m for m in ("pandas", "lib.jobs", "matplotlib") if m in sys.modules))'
 		with tempfile.TemporaryDirectory() as data:
 			out = subprocess.run([sys.executable, '-c', code], cwd=ROOT, capture_output=True, text=True, env={**os.environ, 'DATA_DIR': data})
 		self.assertEqual(out.stdout.strip().splitlines()[-1:], ['[]'], out.stderr)
diff --git a/tests/test_reports.py b/tests/test_reports.py
index b22500d4..271679e6 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -58,7 +58,7 @@ def test_unknown_stickers_and_noise_are_ignored(self):
 
 class ReportFunctions(FinbotCase):
 	def test_prepare_help(self):
-		self.assertTrue(reports.prepare_help('telegram', BOT))
+		self.assertTrue(about.prepare_help('telegram', BOT))
 
 	def test_a_portfolio_with_no_holdings_is_reported_by_name(self):
 		"""Its name used to be read from its first holding: an empty portfolio crashed .performance."""
@@ -67,14 +67,14 @@ def test_a_portfolio_with_no_holdings_is_reported_by_name(self):
 		self.assertTrue(any('>Rob</a>' in line for line in payload), payload)
 
 	def test_prepare_holdings(self):
-		self.assertTrue(reports.prepare_holdings_payload(None, 'telegram'))
+		self.assertTrue(holdings.prepare_holdings_payload(None, 'telegram'))
 
 	def test_prepare_marketcap(self):
-		self.assertTrue(reports.prepare_marketcap_payload('telegram', 'top', length=5))
+		self.assertTrue(rankings.prepare_marketcap_payload('telegram', 'top', length=5))
 
 	def test_prepare_rating(self):
 		for action in ('buy', 'sell'):
-			self.assertTrue(reports.prepare_rating_payload('telegram', action, length=5)[0])
+			self.assertTrue(rankings.prepare_rating_payload('telegram', action, length=5)[0])
 
 	def test_ratings_list_the_strongest_first(self):
 		"""1 is a strong buy and 5 a strong sell: .sell used to list the mildest sells first. On equal ratings, more analysts first;
@@ -84,13 +84,13 @@ def test_ratings_list_the_strongest_first(self):
 		self.patch(watchlist, 'with_holdings', lambda: list(ratings))
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | dict(zip(('recommend', 'recommend_index', 'recommend_analysts'), ratings[t]))})
 		def order(action):
-			payload, chart = reports.prepare_rating_payload('telegram', action)
+			payload, chart = rankings.prepare_rating_payload('telegram', action)
 			return [line.split('Co ')[1].split(' ')[0] for line in payload[1:]], [label for label, _ in chart['rows']]
 		self.assertEqual(order('buy'), (['T2', 'T3', 'T1'], ['T2', 'T3', 'T1']))
 		self.assertEqual(order('sell'), (['T5', 'T6', 'T4'], ['T5', 'T6', 'T4']))
-		self.assertIn('(7+ analysts)', reports.prepare_rating_payload('telegram', 'buy')[0][0])
+		self.assertIn('(7+ analysts)', rankings.prepare_rating_payload('telegram', 'buy')[0][0])
 		self.patch(watchlist, 'with_holdings', lambda: ['T7', 'T8'])
-		self.assertEqual(reports.prepare_rating_payload('telegram', 'buy'), (['No tracked stocks rated buy by 7+ analysts'], None))
+		self.assertEqual(rankings.prepare_rating_payload('telegram', 'buy'), (['No tracked stocks rated buy by 7+ analysts'], None))
 
 	def test_a_small_cap_quoted_in_won_needs_the_bigger_move_too(self):
 		"""demote_volatile compares market caps in US dollars: unconverted, a small cap quoted in won was never small."""
@@ -120,16 +120,16 @@ def test_forward_pe_lists_need_as_many_analysts_as_buy(self):
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | {'price_to_earnings_forward': -5 if t == 'T3' else 20} for t in ts})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'recommend_analysts': analysts[t]}})
 		def listed(action):
-			payload, chart = reports.prepare_value_payload('telegram', action)
+			payload, chart = rankings.prepare_value_payload('telegram', action)
 			return payload[0], [line.split('Co ')[1].split(' ')[0] for line in payload[1:]], chart
 		heading, names, chart = listed('forward pe')
 		self.assertEqual((names, [label for label, _ in chart['rows']]), (['T1'], ['T1'])) # T2 and T4: too few; T3: negative
 		self.assertIn('forward pe ratio (7+ analysts)', heading)
 		self.assertEqual(listed('negative forward pe')[1], ['T3'])
-		self.assertIn('Co T2', str(reports.prepare_value_payload('telegram', 'forward pe', 'T2'))) # one ticker asked for: shown
+		self.assertIn('Co T2', str(rankings.prepare_value_payload('telegram', 'forward pe', 'T2'))) # one ticker asked for: shown
 		self.prefs(min_analysts=13) # the one setting moves .buy and .forward pe together
-		self.assertEqual(reports.prepare_value_payload('telegram', 'forward pe')[0], ['No tracked stocks with 13+ analysts have a forward pe ratio'])
-		self.assertEqual(reports.prepare_rating_payload('telegram', 'buy')[0], ['No tracked stocks rated buy by 13+ analysts'])
+		self.assertEqual(rankings.prepare_value_payload('telegram', 'forward pe')[0], ['No tracked stocks with 13+ analysts have a forward pe ratio'])
+		self.assertEqual(rankings.prepare_rating_payload('telegram', 'buy')[0], ['No tracked stocks rated buy by 13+ analysts'])
 
 	def test_market_caps_compare_in_us_dollars(self):
 		"""Yahoo gives a market cap in the stock's own currency: unconverted, a won market cap topped every list."""
@@ -141,13 +141,13 @@ def fetch(tickers):
 		self.patch(yahoo, 'fetch', fetch)
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: fetch([t]))
 		self.patch(watchlist, 'with_holdings', lambda: list(stocks))
-		payload = reports.prepare_marketcap_payload('telegram')
+		payload = rankings.prepare_marketcap_payload('telegram')
 		self.assertIn('by market cap (USD)', payload[0])
 		self.assertEqual([line.split('Co ')[1].split(' ')[0] for line in payload[1:]], ['U1', 'K1.KS', 'L1.L', 'K2.KS']) # Z1: no rate, left out
 		self.assertIn('mkt cap: 731.88 B', payload[2]) # 1.01 Q won
-		self.assertIn('mkt cap: KRW 1.01 Q (USD 731.88 B)', reports.prepare_marketcap_payload('telegram', 'K1.KS')[0])
-		self.assertIn('mkt cap: GBP 62.50 B (USD 79.11 B)', reports.prepare_marketcap_payload('telegram', 'L1.L')[0]) # pence quote, pound cap
-		beta = reports.prepare_beta_payload('telegram')[0]
+		self.assertIn('mkt cap: KRW 1.01 Q (USD 731.88 B)', rankings.prepare_marketcap_payload('telegram', 'K1.KS')[0])
+		self.assertIn('mkt cap: GBP 62.50 B (USD 79.11 B)', rankings.prepare_marketcap_payload('telegram', 'L1.L')[0]) # pence quote, pound cap
+		beta = rankings.prepare_beta_payload('telegram')[0]
 		self.assertIn('Co K2.KS', str(beta)) # 500 billion won is US$362M: a small cap, which it never was unconverted
 		self.assertNotIn('Co K1.KS', str(beta))
 
@@ -155,18 +155,18 @@ def test_prepare_value(self):
 		for action, heading in (('pe', 'Top 5 tracked stocks by pe ratio'), ('bottom forward pe', 'Bottom 5 tracked stocks by forward pe ratio'),
 								('peg', 'Top 5 tracked stocks by peg ratio'), ('negative peg', 'negative peg: no data found')):
 			with self.subTest(action=action):
-				payload, chart = reports.prepare_value_payload('telegram', action, length=5)
+				payload, chart = rankings.prepare_value_payload('telegram', action, length=5)
 				self.assertIn(heading, payload[0])
 
 	def test_prepare_beta(self):
-		payload, chart = reports.prepare_beta_payload('telegram')
+		payload, chart = rankings.prepare_beta_payload('telegram')
 		self.assertIn('Beta over 1.5', payload[0])
 		self.assertEqual(len(chart['rows']), len(payload) - 1) # a bar per stock listed
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'beta': 1.0}})
-		self.assertEqual(reports.prepare_beta_payload('telegram'), (['No tracked stocks have a beta over 1.5 and mkt cap under US$2B'], None))
+		self.assertEqual(rankings.prepare_beta_payload('telegram'), (['No tracked stocks have a beta over 1.5 and mkt cap under US$2B'], None))
 
 	def test_prepare_profile(self):
-		self.assertTrue(reports.prepare_profile_payload('telegram', '@u', 'T1'))
+		self.assertTrue(profile.prepare_profile_payload('telegram', '@u', 'T1'))
 
 	def test_a_profile_leaves_out_what_yahoo_did_not_send(self):
 		"""Each of these fields missing (or a zero equity) used to crash the whole profile."""
@@ -175,22 +175,22 @@ def test_a_profile_leaves_out_what_yahoo_did_not_send(self):
 		for missing in ('marketState', 'profile_exchange', 'market_cap', 'currency', 'regularMarketPrice', 'sharesOutstanding', 'percent_change'):
 			with self.subTest(missing=missing):
 				self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: {k: v for k, v in full.items() if k != missing}})
-				self.assertIn('Co T1', reports.prepare_profile_payload('telegram', '@u', 'T1')[0])
+				self.assertIn('Co T1', profile.prepare_profile_payload('telegram', '@u', 'T1')[0])
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: full | {'shareholder_equity': 0}})
-		self.assertNotIn('Net debt/equity', str(reports.prepare_profile_payload('telegram', '@u', 'T1')))
+		self.assertNotIn('Net debt/equity', str(profile.prepare_profile_payload('telegram', '@u', 'T1')))
 
 	def test_a_profile_warns_of_a_ps_ratio_only_above_15(self):
 		for ps, warned in ((12.3, False), (15.1, True)): # forward P/E 99 warns (above 30): its warning used to show on the PS line too
 			with self.subTest(ps=ps):
 				self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'price_to_sales': ps, 'price_to_earnings_forward': 99, 'profile_industry': 'Tools'}})
-				line = next(line for line in reports.prepare_profile_payload('telegram', '@u', 'T1') if 'PS ratio' in line)
+				line = next(line for line in profile.prepare_profile_payload('telegram', '@u', 'T1') if 'PS ratio' in line)
 				self.assertEqual('⚠️' in line, warned, line)
 
 	def test_prepare_watchlist(self):
-		self.assertTrue(reports.prepare_watchlist('telegram', '@u'))
+		self.assertTrue(holdings.prepare_watchlist('telegram', '@u'))
 
 	def test_prepare_compare_usage(self):
-		caption = reports.prepare_compare('telegram', ['T1'])[0]
+		caption = history.prepare_compare('telegram', ['T1'])[0]
 		self.assertIn('Usage', ' '.join(caption))
 
 
@@ -371,7 +371,7 @@ def test_a_button_a_chart_does_not_offer_is_ignored(self):
 
 class Podcasts(FinbotCase):
 	def test_format_helpers(self):
-		import podcasts
+		from lib.reports import podcasts
 		self.assertEqual(podcasts.format_duration('3725'), '01:02')
 		self.assertEqual(podcasts.format_duration('1:02:03'), '01:02')
 		self.assertEqual(podcasts.format_duration('62:05'), '01:02')
@@ -382,7 +382,7 @@ def test_format_helpers(self):
 
 	def test_new_episodes_are_posted_once_newest_first(self):
 		"""A feed whose items mix dated and undated ones used to crash the sort (aware and naive times)."""
-		import podcasts
+		from lib.reports import podcasts
 		def item(n, date):
 			return f'<item><guid>g{n}</guid><title>Show: ep {n}</title><enclosure url="https://x/{n}.mp3"/>{date}<itunes:duration>600</itunes:duration></item>'
 		feed = ('<rss xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd"><channel><title>Show</title>'
@@ -413,16 +413,15 @@ def test_by_industry_default(self):
 		self.assertIn('Rob by industry', sent[0][2])
 
 	def test_by_sector_and_payload(self):
-		payload, chart = reports.prepare_pie('rob', 'sector', 'telegram')
+		payload, chart = holdings.prepare_pie('rob', 'sector', 'telegram')
 		self.assertEqual([r[0] for r in chart['rows']], ['Electronic Technology', 'Health Technology', 'Unclassified'])
 		self.assertIn('Electronic Technology: 72.7%', payload) # 800 of 1100; zero-value Gold is left out
 		self.assertEqual(self.command('.pie Rob sector')[0][0], 'photo')
 
 	def test_folds_into_other_and_renders(self):
-		holdings = {f'T{i}:X': {'code': f'T{i}', 'industry_classification_name': f'Ind {i}', 'value': 100 - i}
-			for i in range(12)}
-		self.patch(sharesight.Portfolio, 'holdings', lambda portfolio: holdings)
-		payload, chart = reports.prepare_pie('Rob', 'industry', 'telegram')
+		held = {f'T{i}:X': {'code': f'T{i}', 'industry_classification_name': f'Ind {i}', 'value': 100 - i} for i in range(12)}
+		self.patch(sharesight.Portfolio, 'holdings', lambda portfolio: held)
+		payload, chart = holdings.prepare_pie('Rob', 'industry', 'telegram')
 		self.assertEqual(len(chart['rows']), 8)
 		self.assertEqual(chart['rows'][-1][0], 'Other')
 		self.assertTrue(render.pie_chart(chart['rows'], chart['title'], chart['subtitle']).read().startswith(b'\x89PNG'))

From 03d8147647e034ae74ff9cfb7d72bdf79fd63e17 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 00:10:05 +0000
Subject: [PATCH 14/81] A top-level script runs each report, as before

price.py, trades.py and the other eight are back at the top level as three-line shims that run lib/reports/<name>.py
with their arguments (reports.cli), so schedule lines, cron entries and runs by hand work unchanged. A test runs two of
them and checks there is one per scheduled report.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 cal.py             |  5 +++++
 cleanup.py         |  5 +++++
 milestone.py       |  5 +++++
 performance.py     |  5 +++++
 podcasts.py        |  5 +++++
 price.py           |  5 +++++
 rating.py          |  5 +++++
 reminder.py        |  5 +++++
 shorts.py          |  5 +++++
 tests/test_jobs.py | 12 ++++++++++++
 trades.py          |  5 +++++
 11 files changed, 62 insertions(+)
 create mode 100755 cal.py
 create mode 100755 cleanup.py
 create mode 100755 milestone.py
 create mode 100755 performance.py
 create mode 100755 podcasts.py
 create mode 100755 price.py
 create mode 100755 rating.py
 create mode 100755 reminder.py
 create mode 100755 shorts.py
 create mode 100755 trades.py

diff --git a/cal.py b/cal.py
new file mode 100755
index 00000000..e631d21d
--- /dev/null
+++ b/cal.py
@@ -0,0 +1,5 @@
+#!/usr/bin/env python3
+"""./cal.py [arguments]: runs lib/reports/cal.py once, as its schedule line would."""
+if __name__ == '__main__':
+	from lib import reports
+	reports.cli(__file__)
diff --git a/cleanup.py b/cleanup.py
new file mode 100755
index 00000000..530ee071
--- /dev/null
+++ b/cleanup.py
@@ -0,0 +1,5 @@
+#!/usr/bin/env python3
+"""./cleanup.py [arguments]: runs lib/reports/cleanup.py once, as its schedule line would."""
+if __name__ == '__main__':
+	from lib import reports
+	reports.cli(__file__)
diff --git a/milestone.py b/milestone.py
new file mode 100755
index 00000000..29829cf3
--- /dev/null
+++ b/milestone.py
@@ -0,0 +1,5 @@
+#!/usr/bin/env python3
+"""./milestone.py [arguments]: runs lib/reports/milestone.py once, as its schedule line would."""
+if __name__ == '__main__':
+	from lib import reports
+	reports.cli(__file__)
diff --git a/performance.py b/performance.py
new file mode 100755
index 00000000..e58e5d5b
--- /dev/null
+++ b/performance.py
@@ -0,0 +1,5 @@
+#!/usr/bin/env python3
+"""./performance.py [arguments]: runs lib/reports/performance.py once, as its schedule line would."""
+if __name__ == '__main__':
+	from lib import reports
+	reports.cli(__file__)
diff --git a/podcasts.py b/podcasts.py
new file mode 100755
index 00000000..c86a624e
--- /dev/null
+++ b/podcasts.py
@@ -0,0 +1,5 @@
+#!/usr/bin/env python3
+"""./podcasts.py [arguments]: runs lib/reports/podcasts.py once, as its schedule line would."""
+if __name__ == '__main__':
+	from lib import reports
+	reports.cli(__file__)
diff --git a/price.py b/price.py
new file mode 100755
index 00000000..4aee8e91
--- /dev/null
+++ b/price.py
@@ -0,0 +1,5 @@
+#!/usr/bin/env python3
+"""./price.py [arguments]: runs lib/reports/price.py once, as its schedule line would."""
+if __name__ == '__main__':
+	from lib import reports
+	reports.cli(__file__)
diff --git a/rating.py b/rating.py
new file mode 100755
index 00000000..4a44d44f
--- /dev/null
+++ b/rating.py
@@ -0,0 +1,5 @@
+#!/usr/bin/env python3
+"""./rating.py [arguments]: runs lib/reports/rating.py once, as its schedule line would."""
+if __name__ == '__main__':
+	from lib import reports
+	reports.cli(__file__)
diff --git a/reminder.py b/reminder.py
new file mode 100755
index 00000000..09adbc26
--- /dev/null
+++ b/reminder.py
@@ -0,0 +1,5 @@
+#!/usr/bin/env python3
+"""./reminder.py [arguments]: runs lib/reports/reminder.py once, as its schedule line would."""
+if __name__ == '__main__':
+	from lib import reports
+	reports.cli(__file__)
diff --git a/shorts.py b/shorts.py
new file mode 100755
index 00000000..14347057
--- /dev/null
+++ b/shorts.py
@@ -0,0 +1,5 @@
+#!/usr/bin/env python3
+"""./shorts.py [arguments]: runs lib/reports/shorts.py once, as its schedule line would."""
+if __name__ == '__main__':
+	from lib import reports
+	reports.cli(__file__)
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
index 79793b45..9f1c6f33 100644
--- a/tests/test_jobs.py
+++ b/tests/test_jobs.py
@@ -145,6 +145,18 @@ def test_cli_runs_the_handler(self):
 		webhook.cli('price.py', price.job, lambda **k: calls.append(k), ['midsession'])
 		self.assertEqual(calls, [{'midsession': True}])
 
+	def test_each_scheduled_report_runs_from_its_top_level_script(self):
+		"""./price.py and the rest run lib/reports/<name>.py, so cron lines and runs by hand keep working."""
+		self.assertEqual(sorted(path.stem for path in Path(ROOT).glob('*.py') if path.stem not in ('bot', 'bot_send_stdin')),
+						 sorted(jobs.SCRIPTS))
+		with tempfile.TemporaryDirectory() as data:
+			run = lambda *args: subprocess.run([sys.executable, *args], cwd=ROOT, capture_output=True, text=True, timeout=60,
+				env={**os.environ, 'DATA_DIR': data})
+			self.assertEqual(run('cleanup.py', '0').returncode, 0)
+			bad = run('price.py', 'bogus')
+			self.assertEqual(bad.returncode, 1)
+			self.assertIn('Usage: price.py [midsession', bad.stderr)
+
 
 class CleanupJob(FinbotCase):
 	def test_removes_only_old_finbot_cache_files(self):
diff --git a/trades.py b/trades.py
new file mode 100755
index 00000000..d9f3cc9f
--- /dev/null
+++ b/trades.py
@@ -0,0 +1,5 @@
+#!/usr/bin/env python3
+"""./trades.py [arguments]: runs lib/reports/trades.py once, as its schedule line would."""
+if __name__ == '__main__':
+	from lib import reports
+	reports.cli(__file__)

From 1940085e0fe5e2631ef9b22f98f0cfab653e52b1 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 00:17:42 +0000
Subject: [PATCH 15/81] Markup by dialect: service.bold(text) in place of
 webhook.bold(text, service)

lib/dialects.py holds each network's markup: bold, italic, strike, links, user text made safe, and its message sizes. A
dialect is a str equal to the network's name (TELEGRAM == 'telegram'), so it goes wherever the name did; the bot turns a
name into its dialect where a message comes in or a report goes out (worker, webhook.destinations). The 73 markup calls
that passed the service name along become service.bold(text), service.link(url, text) and the like; webhook's MARKUP
table and links.link go. Tests pass TELEGRAM, SLACK and DISCORD. Output is unchanged.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md                  |  9 ++--
 lib/charts.py              |  2 +-
 lib/dialects.py            | 88 ++++++++++++++++++++++++++++++++++++++
 lib/links.py               | 29 ++++---------
 lib/reports/about.py       | 20 ++++-----
 lib/reports/cal.py         |  4 +-
 lib/reports/history.py     |  7 ++-
 lib/reports/holdings.py    | 17 ++++----
 lib/reports/milestone.py   |  4 +-
 lib/reports/performance.py |  4 +-
 lib/reports/podcasts.py    |  3 +-
 lib/reports/price.py       |  6 +--
 lib/reports/profile.py     | 67 ++++++++++++++---------------
 lib/reports/rankings.py    |  9 ++--
 lib/reports/rating.py      |  4 +-
 lib/reports/reminder.py    | 17 ++++----
 lib/reports/shorts.py      |  5 +--
 lib/reports/trades.py      |  4 +-
 lib/webhook.py             | 34 ++-------------
 lib/worker.py              | 12 +++---
 tests/common.py            |  3 +-
 tests/test_edges.py        |  2 +-
 tests/test_jobs.py         |  8 ++--
 tests/test_providers.py    |  2 +-
 tests/test_reports.py      | 56 ++++++++++++------------
 tests/test_util.py         |  4 +-
 26 files changed, 231 insertions(+), 189 deletions(-)
 create mode 100644 lib/dialects.py

diff --git a/CLAUDE.md b/CLAUDE.md
index b6fddb0d..d6bb78b9 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -34,8 +34,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - `lib/webhook.py`: sending (chunking, photos, Slack upload), markup, escaping, error reporting. `lib/telegram.py`: Bot API calls.
 - `lib/buttons.py`: period buttons under a chart. `lib/charts.py`: a button press rebuilds its chart; the Telegram command
   keyboard.
-- `lib/markets.py`: market tables, flags, ticker translation. `lib/formats.py`: names, units, dates, percents in words.
-  `lib/links.py`: links. `lib/render.py`: every chart (matplotlib).
+- `lib/dialects.py`: each network's markup and message sizes. `lib/markets.py`: market tables, flags, ticker translation.
+  `lib/formats.py`: names, units, dates, percents in words. `lib/links.py`: stock links. `lib/render.py`: every chart.
 - `lib/yahoo.py lib/sharesight.py lib/shortman.py lib/simplywallst.py`: data sources, cached in `DATA_DIR/cache`.
   `sharesight.portfolios(name=None)` gives `Portfolio` objects: `holdings()`, `tickers()`, `performance(days)`,
   `trades(days)`.
@@ -53,8 +53,9 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 ## Conventions
 - Read settings as `config.env.<name>` / `config.prefs.<name>` where they are used, never into a default or module constant (a
   reload must be seen); in tests, `self.env(...)` / `self.prefs(...)` change them.
-- A report is a list of lines. Markup through `webhook.bold/italic(text, service)`, user-written text through
-  `webhook.escape(text, service)`, links through `links.link`/`links.finance_link`.
+- A report is a list of lines. `service` is a dialect (`lib/dialects.py`), which equals the network's name: markup with
+  `service.bold(text)`, `.italic`, `.strike`, `.link(url, text)`, user-written text through `service.escape(text)`, stock links
+  through `links.finance_link(symbol, exchange, service)`. Tests pass `TELEGRAM`, `SLACK`, `DISCORD`.
 - Tabs everywhere, tests included (`.editorconfig`). New and changed lines stay under 120 characters (an edit quotes whole lines).
 - Log with `print(..., file=sys.stderr)`. systemd discards stdout (`StandardOutput=null`), so the bot and its reports never
   print there (a test checks); only the util/ tools, run by hand, do.
diff --git a/lib/charts.py b/lib/charts.py
index c195b604..5046ddd7 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -52,7 +52,7 @@ def refresh_keyboards():
 			except Exception as e:
 				print("keyboard refresh failed for", chat_id, e, file=sys.stderr)
 
-def build(kind, tickers, period, service='telegram'):
+def build(kind, tickers, period, service):
 	"""(caption, image) for a chart kind at a button period; period None = the command's default view."""
 	days = buttons.PERIODS[period][1] if period else None
 	if kind == 'h':
diff --git a/lib/dialects.py b/lib/dialects.py
new file mode 100644
index 00000000..bd9c18ed
--- /dev/null
+++ b/lib/dialects.py
@@ -0,0 +1,88 @@
+"""How each chat network marks text up: bold, italic, struck-through, links, user text made safe, and message sizes. A dialect is
+also the network's name (TELEGRAM == 'telegram'), so it can stand wherever the name did: reports take one as `service` and write
+`service.bold(text)`. dialects.of('slack') gives the one for a name."""
+import html
+import re
+
+from lib import config
+
+
+class Dialect(str):
+	"""A network's markup. Subclasses set the styles, link markup and limits."""
+	__slots__ = ()
+	styles = {} # style -> (before, after)
+	text_limit = 4096 # characters in one message
+	caption_limit = None # characters under a chart, as visible() counts them; None: no limit of its own
+
+	def bold(self, text):
+		return self._styled('bold', text)
+
+	def italic(self, text):
+		return self._styled('italic', text)
+
+	def strike(self, text):
+		return self._styled('strike', text)
+
+	def _styled(self, style, text):
+		before, after = self.styles[style]
+		return before + text + after
+
+	def link(self, url, text):
+		"""text linking to url; just the text with the hyperlink preference off."""
+		return self._link(url, text) if config.prefs.hyperlink else text
+
+	def escape(self, text):
+		"""Text a user wrote (a .plan entry, a name) made safe to show: Telegram's HTML and Slack's markup both take &amp; &lt; &gt;.
+		It is unescaped first because it may arrive escaped already (Slack escapes what it sends; bot.py escapes Telegram names):
+		escaping twice would show &amp;amp;."""
+		return html.escape(html.unescape(str(text)), quote=False)
+
+	def visible(self, text):
+		"""The length of text as the network counts it against caption_limit."""
+		return len(text)
+
+
+class Telegram(Dialect):
+	"""Telegram's HTML parse mode."""
+	__slots__ = ()
+	styles = {'bold': ('<b>', '</b>'), 'italic': ('<i>', '</i>'), 'strike': ('<s>', '</s>')}
+	caption_limit = 1024
+
+	def _link(self, url, text):
+		return '<a href="' + url + '">' + text + '</a>'
+
+	def visible(self, text):
+		return len(re.sub(r'<[^>]+>', '', text)) # the tags do not count
+
+
+class Slack(Dialect):
+	"""Slack's mrkdwn."""
+	__slots__ = ()
+	styles = {'bold': ('*', '*'), 'italic': ('_', '_'), 'strike': ('~', '~')}
+
+	def _link(self, url, text):
+		return '<' + url + '|' + text + '>'
+
+
+class Discord(Dialect):
+	"""Discord's markdown."""
+	__slots__ = ()
+	styles = {'bold': ('**', '**'), 'italic': ('_', '_'), 'strike': ('~~', '~~')}
+	text_limit = caption_limit = 2000 # markup included
+
+	def _link(self, url, text):
+		"""A masked link: the angle brackets keep its preview away; brackets in the text and parentheses in the url would end it
+		early."""
+		return '[' + text.replace('[', '\\[').replace(']', '\\]') + '](<' + url.replace('(', '%28').replace(')', '%29') + '>)'
+
+	def escape(self, text):
+		return html.unescape(str(text)) # markdown shows entities as they are: the plain text
+
+
+TELEGRAM, SLACK, DISCORD = Telegram('telegram'), Slack('slack'), Discord('discord')
+_BY_NAME = {dialect: dialect for dialect in (TELEGRAM, SLACK, DISCORD)}
+
+
+def of(network):
+	"""The dialect of a network by name ('telegram', 'slack' or 'discord'); a dialect is its own."""
+	return _BY_NAME[network]
diff --git a/lib/links.py b/lib/links.py
index 66881bac..27159b8b 100644
--- a/lib/links.py
+++ b/lib/links.py
@@ -1,37 +1,24 @@
-"""Links: chat markup for a link, and a stock's page on Yahoo or Google Finance."""
+"""Links to a stock's page on Yahoo or Google Finance, in the markup of `service` (a dialect: see lib/dialects.py)."""
 from lib import config
 from lib import markets
 
-def yahoo_link(ticker, service='telegram', brief=True, text=None):
+def yahoo_link(ticker, service, brief=True, text=None):
 	if not text:
 		text = ticker.split('.')[0] if brief else ticker
-	return link("https://au.finance.yahoo.com/quote/" + ticker, text, service)
+	return service.link("https://au.finance.yahoo.com/quote/" + ticker, text)
 
-def link(url, text, service='telegram'):
-	"""text linking to url in the service's markup: HTML on Telegram, <url|text> on Slack, a masked link on Discord (the angle
-	brackets keep its preview away; brackets in the text and parentheses in the url would end it early)."""
-	if not config.prefs.hyperlink:
-		return text
-	if service == 'telegram':
-		return '<a href="' + url + '">' + text + '</a>'
-	if service == 'slack':
-		return '<' + url + '|' + text + '>'
-	if service == 'discord':
-		return '[' + text.replace('[', '\\[').replace(']', '\\]') + '](<' + url.replace('(', '%28').replace(')', '%29') + '>)'
-	return text
-
-def finance_link(symbol, exchange, service='telegram', days=1, brief=True, text=None):
+def finance_link(symbol, exchange, service, days=1, brief=True, text=None):
 	if config.prefs.hyperlinkProvider == 'google':
 		return gfinance_link(symbol, exchange, service, days, brief, text)
 	return yahoo_link(symbol, service, brief, text)
 
-def quote_link(ticker, exchange, service='telegram', days=1):
+def quote_link(ticker, exchange, service, days=1):
 	"""finance_link, but to Yahoo for what Google Finance has no page for: Taipei Exchange, crypto (CCC) and indices (^)."""
 	if exchange in ('Taipei Exchange', 'CCC') or ticker.startswith('^'):
 		return yahoo_link(ticker, service)
 	return finance_link(ticker, exchange, service, days=days)
 
-def gfinance_link(symbol, exchange, service='telegram', days=1, brief=True, text=None):
+def gfinance_link(symbol, exchange, service, days=1, brief=True, text=None):
 	window = '1D'
 	if not days:
 		days = 1
@@ -61,6 +48,6 @@ def gfinance_link(symbol, exchange, service='telegram', days=1, brief=True, text
 			text = ticker
 		else:
 			text = symbol # US
-	if service in {'telegram', 'discord', 'slack'} and config.prefs.hyperlink:
-		return link(url + ticker + '?window=' + window, text, service)
+	if config.prefs.hyperlink:
+		return service.link(url + ticker + '?window=' + window, text)
 	return symbol
diff --git a/lib/reports/about.py b/lib/reports/about.py
index e6a2e7d6..faf0acab 100644
--- a/lib/reports/about.py
+++ b/lib/reports/about.py
@@ -1,8 +1,6 @@
 """What finbot can do: the .help text, and the .jobs listing of the schedule (with running a job from it)."""
 import re
 from lib import jobs
-from lib import links
-from lib import webhook
 
 def prepare_jobs_payload(service, args):
 	"""`.jobs` lists the scheduled commands, each once; `.jobs run <number|command>` runs one now, as cron would.
@@ -15,7 +13,7 @@ def prepare_jobs_payload(service, args):
 		commands = scheduler.commands()
 		if not commands:
 			return ['No scheduled jobs'], None
-		payload = [webhook.bold('Scheduled commands:', service)]
+		payload = [service.bold('Scheduled commands:')]
 		for number, (label, _, _) in enumerate(commands, 1):
 			payload.append(f"{number}. {jobs.display(label)}" + (' (running)' if label in scheduler.busy else ''))
 		payload.append("\nRun one with .jobs run NUMBER" + (' or tap it' if service == 'telegram' else ''))
@@ -42,19 +40,19 @@ def job_line(scheduler, label, outcome):
 
 def prepare_help(service, botName):
 	payload = []
-	payload.append(webhook.bold("Tracked securities:", service))
+	payload.append(service.bold("Tracked securities:"))
 	payload.append(".holdings")
 	payload.append(".pie PORTFOLIO [sector]")
 	payload.append(".watchlist [add|del SYMBOL]")
 	payload.append(".who SYMBOL")
 
-	payload.append(webhook.bold("\nCompany info:", service))
+	payload.append(service.bold("\nCompany info:"))
 	payload.append('.SYMBOL')
 	payload.append(".dividend [period|SYMBOL]")
 	payload.append(".earnings [period|SYMBOL]")
 	payload.append(".marketcap [SYMBOL|bottom|top]")
 
-	payload.append(webhook.bold("\nPrice:", service))
+	payload.append(service.bold("\nPrice:"))
 	payload.append(".beta")
 	payload.append(".compare SYMBOL SYMBOL [...] [period|max]")
 	payload.append(".history SYMBOL")
@@ -63,7 +61,7 @@ def prepare_help(service, botName):
 	payload.append(".session [percent|SYMBOL]")
 	payload.append(".premarket [percent|SYMBOL]")
 
-	payload.append(webhook.bold("\nRisk & Value:", service))
+	payload.append(service.bold("\nRisk & Value:"))
 	payload.append(".buy")
 	payload.append(".sell")
 	payload.append(".pe [SYMBOL|top|bottom]")
@@ -71,17 +69,17 @@ def prepare_help(service, botName):
 	payload.append(".peg [SYMBOL|top|bottom|negative]")
 	payload.append(".shorts [percent|SYMBOL]")
 
-	payload.append(webhook.bold("\nTrades:", service))
+	payload.append(service.bold("\nTrades:"))
 	payload.append(".trades [period] [portfolio]")
 
-	payload.append(webhook.bold("\nScheduled jobs:", service))
+	payload.append(service.bold("\nScheduled jobs:"))
 	payload.append(".jobs [run NUMBER|COMMAND]")
 
-	payload.append(webhook.bold("\nAlternative syntax:", service))
+	payload.append(service.bold("\nAlternative syntax:"))
 	if service == 'slack':
 		payload.append('<' + botName + '> trades')
 	else:
 		payload.append(botName + ' price')
 	payload.append("")
-	payload.append(f"Full help on {links.link('https://github.com/robdevops/finbot', 'GitHub', service)}")
+	payload.append(f"Full help on {service.link('https://github.com/robdevops/finbot', 'GitHub')}")
 	return payload
diff --git a/lib/reports/cal.py b/lib/reports/cal.py
index 07984730..53069955 100644
--- a/lib/reports/cal.py
+++ b/lib/reports/cal.py
@@ -52,12 +52,12 @@ def prepare_payload(service, market_data):
 		payload = []
 		for date in sorted({when.date() for when, *_ in found}):
 			if not specific_stock:
-				payload += ["", webhook.bold(f"{date:%b %d}", service)]
+				payload += ["", service.bold(f"{date:%b %d}")]
 			for when, flag, title, link in found:
 				if when.date() == date:
 					payload.append(f"{flag} {date:%b %d} {title} ({link})" if specific_stock else f"{flag} {title} ({link})")
 		if payload and not specific_stock:
-			payload.insert(0, webhook.bold(f"{heading} {formats.days_english(days, 'for the next ')}:", service))
+			payload.insert(0, service.bold(f"{heading} {formats.days_english(days, 'for the next ')}:"))
 		elif not payload and interactive:
 			payload = [f"No events found for {tickers[0]}" if specific_stock else f"No events found {formats.days_english(days, 'for the next ')}"]
 		return payload
diff --git a/lib/reports/history.py b/lib/reports/history.py
index 1ed17e08..87e4c39d 100644
--- a/lib/reports/history.py
+++ b/lib/reports/history.py
@@ -7,7 +7,6 @@
 from lib import links
 from lib import markets
 from lib import render
-from lib import webhook
 from lib import yahoo
 
 def prepare_compare(service, args):
@@ -73,7 +72,7 @@ def prepare_compare(service, args):
 	subtitle = f"% change over {period_label}, since {start:%d %b %Y}" + note
 	image = render.compare_graph([(t.split('.')[0], series) for t, series in cropped], title, subtitle)
 	image.seek(0)
-	caption = [webhook.bold(f"{period_label}, since {start:%d %b %Y}", service) + note]
+	caption = [service.bold(f"{period_label}, since {start:%d %b %Y}") + note]
 	results = sorted(((series.iloc[-1] / series.iloc[0] - 1) * 100, t) for t, series in cropped)
 	for pct, t in reversed(results):
 		link = links.finance_link(t, exchanges[t], service, days=span_days, brief=False)
@@ -93,11 +92,11 @@ def history_payload(ticker, service, highlight=None, graph=False):
 	if ticker not in market_data or 'percent_change' not in market_data[ticker]:
 		return [f".history: no data found for ticker {ticker}"], None, None
 	table, image = yahoo.price_history(ticker) if graph else yahoo.price_history(ticker, graph=False)
-	lines = [webhook.bold(f"{title} ({ticker_link}) performance history", service)]
+	lines = [service.bold(f"{title} ({ticker_link}) performance history")]
 	for interval in ('Max', '10Y', '5Y', '3Y', '1Y', 'YTD', '6M', '3M', '1M', '7D', '1D'):
 		if interval in table:
 			percent = table[interval]
-			label = webhook.bold(interval + ':', service)
+			label = service.bold(interval + ':')
 			if interval == highlight:
 				label = '● ' + label
 			lines.append(f"{formats.get_emoji(percent)} {label} {formats.signed_percent(percent)}")
diff --git a/lib/reports/holdings.py b/lib/reports/holdings.py
index c644927c..42b5bb9b 100644
--- a/lib/reports/holdings.py
+++ b/lib/reports/holdings.py
@@ -5,12 +5,11 @@
 from lib import formats
 from lib import links
 from lib import markets
-from lib import webhook
 from lib import yahoo
 
 def holding_line(holding, service, text=None):
 	"""'🇦🇺 Title (CODE)' for a Sharesight holding, CODE (or `text`) linking to the holding on Sharesight."""
-	link = links.link(f"https://portfolio.sharesight.com/holdings/{holding['holding_id']}/dashboard", holding['code'] if text is None else text, service)
+	link = service.link(f"https://portfolio.sharesight.com/holdings/{holding['holding_id']}/dashboard", holding['code'] if text is None else text)
 	return f"{markets.flag_from_market(holding['market_code'])} {formats.transform_title(holding['name'])} ({link})"
 
 def prepare_holdings_payload(portfolio_name, service):
@@ -22,7 +21,7 @@ def prepare_holdings_payload(portfolio_name, service):
 	holdings = portfolio.holdings()
 	payload = sorted(holding_line(holding, service) for holding in holdings.values())
 	if payload:
-		payload.insert(0, webhook.bold(f"Holdings for {portfolio.name} ({len(holdings)})", service))
+		payload.insert(0, service.bold(f"Holdings for {portfolio.name} ({len(holdings)})"))
 	return payload
 
 def prepare_who(service, code):
@@ -32,7 +31,7 @@ def prepare_who(service, code):
 		found = [holding_line(holding, service, code)
 			for holding in portfolio.holdings().values() if holding['code'].upper() == code]
 		if found:
-			payload += [webhook.bold(f"{portfolio.name}:", service), "\n".join(found), ""]
+			payload += [service.bold(f"{portfolio.name}:"), "\n".join(found), ""]
 	return payload or [f'{code} not found in any portfolio']
 
 def prepare_pie(portfolio_name, by, service, max_slices=8):
@@ -62,7 +61,7 @@ def prepare_pie(portfolio_name, by, service, max_slices=8):
 		rows.append(('Other', rest))
 	total = sum(v for _, v in rows)
 	title = f"{name} by {by}"
-	payload = [webhook.bold(title, service)]
+	payload = [service.bold(title)]
 	payload += [f"{label}: {value / total * 100:.1f}%" for label, value in rows]
 	noun = {'sector': ('sector', 'sectors'), 'industry': ('industry', 'industries')}[by]
 	subtitle = f"{len(totals)} {noun[0] if len(totals) == 1 else noun[1]}"
@@ -85,18 +84,18 @@ def watchlist_change(service, user, action, asked):
 	def described(ticker, quote): # "Full Name (TICKER)", bold
 		quote = quote or {}
 		link = links.finance_link(ticker, quote.get('profile_exchange') or ticker, service)
-		return webhook.bold(f"{quote['profile_title']} ({link})" if quote.get('profile_title') else link, service)
+		return service.bold(f"{quote['profile_title']} ({link})" if quote.get('profile_title') else link)
 	if action == 'delete':
 		if asked not in watchlist.tickers():
-			return f"Beep Boop. I could not find {webhook.bold(asked, service)} to remove it"
+			return f"Beep Boop. I could not find {service.bold(asked)} to remove it"
 		quote = yahoo.fetch([asked]).get(asked)
 		watchlist.remove(asked)
 		return f"Ok {user}, I deleted {described(asked, quote)}"
 	found, quote = watchlist.find(asked)
 	if not quote:
-		return f"Beep Boop. I could not find {webhook.bold(asked, service)} to add it"
+		return f"Beep Boop. I could not find {service.bold(asked)} to add it"
 	added = watchlist.add(found)
 	if found != asked:
 		done = "so I added" if added else "and I'm already tracking"
-		return f"Beep Boop. I could not find {webhook.bold(asked, service)} {done} {described(found, quote)}"
+		return f"Beep Boop. I could not find {service.bold(asked)} {done} {described(found, quote)}"
 	return f"Ok {user}, I added {described(found, quote)}" if added else f"{user}, I'm already tracking {described(found, quote)}"
diff --git a/lib/reports/milestone.py b/lib/reports/milestone.py
index 8dbd19e2..b6606037 100644
--- a/lib/reports/milestone.py
+++ b/lib/reports/milestone.py
@@ -42,13 +42,13 @@ def prepare_milestone_payload(service, market_data):
 		payloadhigh.sort()
 		if payloadhigh:
 			message = 'Record high:'
-			message = webhook.bold(message, service)
+			message = service.bold(message)
 			payload.append(message)
 			payload = payload + payloadhigh
 		if payloadlow:
 			payload.append("")
 			message = 'Record low:'
-			message = webhook.bold(message, service)
+			message = service.bold(message)
 			payload.append(message)
 			payload = payload + payloadlow
 		payload = [i[0] for i in groupby(payload)] # de-dupe white space
diff --git a/lib/reports/performance.py b/lib/reports/performance.py
index cecef39f..cceaaa5f 100644
--- a/lib/reports/performance.py
+++ b/lib/reports/performance.py
@@ -42,7 +42,7 @@ def report(service, past_days=None, portfolio_select=None, interactive=False, pe
 	lines = [] # (percent, text), sorted into the payload below
 	chart_rows = []
 	for portfolio_id, (name, data) in performance.items(): # the name from the portfolio list: one with no holdings names none
-		link = links.link("https://portfolio.sharesight.com/portfolios/" + str(portfolio_id), name, service)
+		link = service.link("https://portfolio.sharesight.com/portfolios/" + str(portfolio_id), name)
 		percent = float(data['report']['capital_gain_percent'])
 		lines.append((percent, f"{formats.get_emoji(percent)} {link} {formats.signed_percent(percent, 2)}"))
 		chart_rows.append((name, percent))
@@ -55,7 +55,7 @@ def report(service, past_days=None, portfolio_select=None, interactive=False, pe
 			chart_rows.append((text, float(percent)))
 	lines.sort(key=lambda item: item[0], reverse=True)
 	period = formats.days_english(past_days)
-	payload = [webhook.bold(f"Performance over {period}", service)] + [line for _, line in lines]
+	payload = [service.bold(f"Performance over {period}")] + [line for _, line in lines]
 	graph = None
 	if webhook.takes_charts(service, interactive):
 		chart_rows.sort(key=lambda row: row[1], reverse=True) # best first
diff --git a/lib/reports/podcasts.py b/lib/reports/podcasts.py
index 7ba6d72a..1bd9e232 100644
--- a/lib/reports/podcasts.py
+++ b/lib/reports/podcasts.py
@@ -7,7 +7,6 @@
 from lib import net
 from lib import state
 from lib import webhook
-from lib import links
 
 NAMESPACES = {
 	"media": "http://search.yahoo.com/mrss/",
@@ -99,7 +98,7 @@ def lambda_handler(feed_url):
 			podcast_name = episode['podcast']
 			title = episode['title'].removeprefix(podcast_name).lstrip(string.punctuation + " ")
 			duration = f" {episode['duration']}" if episode['duration'] else ""
-			payload.append(f"📣 {podcast_name}: {links.link(episode['url'], title, service)}{duration} ({episode['ago']})")
+			payload.append(f"📣 {podcast_name}: {service.link(episode['url'], title)}{duration} ({episode['ago']})")
 		webhook.send(service, url, chat, payload)
 	with state.edit(seen_file, []) as recorded: # last, so that a failure above leaves these episodes for next run
 		recorded[:] = sorted(guids.union(recorded))
diff --git a/lib/reports/price.py b/lib/reports/price.py
index 88710ed3..72c53822 100644
--- a/lib/reports/price.py
+++ b/lib/reports/price.py
@@ -134,14 +134,14 @@ def report(service, threshold=None, user='', specific_stock=None, interactive=Fa
 			heading = f'{threshold:g}% at close, {on}'
 		chartable = webhook.takes_charts(service, interactive)
 		if top:
-			bottom = [webhook.bold(f'Bottom {top}', service)] + list(reversed(payload[-top:]))
-			payload = [webhook.bold(heading, service)] + payload[:top] + [''] + bottom
+			bottom = [service.bold(f'Bottom {top}')] + list(reversed(payload[-top:]))
+			payload = [service.bold(heading)] + payload[:top] + [''] + bottom
 			chart_top = chart_rows[:top]
 			chart_bottom = [row for row in reversed(chart_rows[-top:]) if row not in chart_top] # worst first
 			if chartable and chart_top:
 				graph = render.bar_graph(label(chart_top), label(chart_bottom), f"Top & bottom {top}", f"% change {formats.days_english(days, 'in ', 'the past ')}")
 		else:
-			payload.insert(0, webhook.bold(heading, service))
+			payload.insert(0, service.bold(heading))
 			if chartable:
 				graph = render.rows_chart(label(chart_rows), heading, session='midsession' if midsession else 'premarket' if premarket else None)
 	elif not payload and interactive:
diff --git a/lib/reports/profile.py b/lib/reports/profile.py
index 04291a71..38bd2eb1 100644
--- a/lib/reports/profile.py
+++ b/lib/reports/profile.py
@@ -6,7 +6,6 @@
 from lib import formats
 from lib import links
 from lib import markets
-from lib import webhook
 from lib import yahoo
 from lib import simplywallst
 from itertools import groupby
@@ -61,8 +60,8 @@ def prepare_profile_payload(service, user, ticker):
 	exchange = data.get('profile_exchange', '').replace('NasdaqCM', 'Nasdaq').replace('NasdaqGS', 'Nasdaq').replace('NYSEArca', 'NYSE')
 	ticker_link = links.finance_link(ticker, exchange, service, brief=False)
 	gfinance_link = links.gfinance_link(ticker, exchange, service, brief=True, text='google') # all three before exchange is renamed below
-	sws_link = links.link(simplywallst.get_url(ticker, title, exchange), 'simplywall.st', service)
-	macrotrends_link = links.link('https://www.google.com/search?q=site:macrotrends.net+' + title + '+PE Ratio+' + code + '&btnI', 'macrotrends', service)
+	sws_link = service.link(simplywallst.get_url(ticker, title, exchange), 'simplywall.st')
+	macrotrends_link = service.link('https://www.google.com/search?q=site:macrotrends.net+' + title + '+PE Ratio+' + code + '&btnI', 'macrotrends')
 	if 'profile_bio' in data:
 		payload.append(formats.make_paragraphs(data['profile_bio']))
 
@@ -82,52 +81,52 @@ def prepare_profile_payload(service, user, ticker):
 		exchange_url = 'https://quote.jpx.co.jp/jpx/template/quote.cgi?F=tmp/e_stock_detail&MKTN=T&QCODE=' + code
 	else:
 		exchange_url = 'https://www.google.com/search?q=stock+exchange+' + exchange + '+' + code + '&btnI'
-	exchange_link = links.link(exchange_url, exchange, service)
+	exchange_link = service.link(exchange_url, exchange)
 	location = [data[key] for key in ('profile_city', 'profile_state', 'profile_country') if key in data]
 	if payload:
 		payload.append("")
 
 	if location:
-		payload.append(webhook.bold("Location:", service) + " " + ', '.join(location))
+		payload.append(service.bold("Location:") + " " + ', '.join(location))
 	if 'profile_industry' in data and 'profile_sector' in data:
-		payload.append(webhook.bold("Classification: ", service) + f"{data['profile_industry']}, {data['profile_sector']}")
+		payload.append(service.bold("Classification: ") + f"{data['profile_industry']}, {data['profile_sector']}")
 	if 'profile_employees' in data:
-		payload.append(webhook.bold("Employees: ", service) + f"{data['profile_employees']:,}")
+		payload.append(service.bold("Employees: ") + f"{data['profile_employees']:,}")
 	if 'profile_website' in data:
-		sites = [links.link(data['profile_website'], formats.strip_url(data['profile_website']), service), exchange_link]
+		sites = [service.link(data['profile_website'], formats.strip_url(data['profile_website'])), exchange_link]
 		if 'NYSE' in exchange or 'Nasdaq' in exchange:
-			sites += [links.link('https://finviz.com/quote.ashx?t=' + ticker, 'finviz', service), macrotrends_link,
-				links.link('https://seekingalpha.com/symbol/' + ticker, 'seekingalpha', service)]
+			sites += [service.link('https://finviz.com/quote.ashx?t=' + ticker, 'finviz'), macrotrends_link,
+				service.link('https://seekingalpha.com/symbol/' + ticker, 'seekingalpha')]
 		elif exchange == 'ASX':
-			sites.append(links.link('https://www.shortman.com.au/stock?q=' + code.lower(), 'shortman', service))
-		payload.append(webhook.bold("Links: ", service) + ' | '.join(sites))
+			sites.append(service.link('https://www.shortman.com.au/stock?q=' + code.lower(), 'shortman'))
+		payload.append(service.bold("Links: ") + ' | '.join(sites))
 
 	if payload:
 		payload.append("")
 
 	if 'currency' in data and 'market_cap' in data:
-		payload.append(webhook.bold("Mkt cap:", service) + f" {yahoo.market_cap_currency(data)} {formats.humanUnits(data['market_cap'])}") # GBP, not pence
+		payload.append(service.bold("Mkt cap:") + f" {yahoo.market_cap_currency(data)} {formats.humanUnits(data['market_cap'])}") # GBP, not pence
 	if all(key in data for key in ('shareholder_equity', 'total_debt', 'total_cash', 'profile_industry')) and data['shareholder_equity'] and 'Bank' not in data['profile_industry']:
 		net_debt_equity_ratio = round((data['total_debt'] - data['total_cash']) / data['shareholder_equity'] * 100)
 		if net_debt_equity_ratio > 0:
-			payload.append(webhook.bold("Net debt/equity ratio:", service) + f" {net_debt_equity_ratio}%" + ('⚠️ ' if net_debt_equity_ratio > 40 else ''))
+			payload.append(service.bold("Net debt/equity ratio:") + f" {net_debt_equity_ratio}%" + ('⚠️ ' if net_debt_equity_ratio > 40 else ''))
 	if 'earnings_date' in data:
 		earnings_date = datetime.datetime.fromtimestamp(data['earnings_date'])
 		if earnings_date > now:
-			payload.append(webhook.bold("Earnings date:", service) + f" {earnings_date:%b %d}")
+			payload.append(service.bold("Earnings date:") + f" {earnings_date:%b %d}")
 	if data.get('dividend', 0) > 0:
-		payload.append(webhook.bold("Dividend:", service) + f" {data['dividend']}%")
+		payload.append(service.bold("Dividend:") + f" {data['dividend']}%")
 		if 'ex_dividend_date' in data:
 			ex_dividend_date = datetime.datetime.fromtimestamp(data['ex_dividend_date'])
 			if ex_dividend_date > now:
-				payload.append(webhook.bold("Ex-dividend date:", service) + f" {ex_dividend_date:%b %d}")
+				payload.append(service.bold("Ex-dividend date:") + f" {ex_dividend_date:%b %d}")
 	cashflow = data.get('free_cashflow', data.get('operating_cashflow'))
 	if cashflow:
-		payload.append(webhook.bold("Cashflow positive:", service) + (" no ⚠️ " if cashflow < 0 else " yes"))
+		payload.append(service.bold("Cashflow positive:") + (" no ⚠️ " if cashflow < 0 else " yes"))
 	if 'net_income' in data:
-		payload.append(webhook.bold("Profitable:", service) + (" no ⚠️ " if data['net_income'] <= 0 else " yes"))
+		payload.append(service.bold("Profitable:") + (" no ⚠️ " if data['net_income'] <= 0 else " yes"))
 	else:
-		payload.append(webhook.bold("Profitable:", service) + " unknown ⚠️")
+		payload.append(service.bold("Profitable:") + " unknown ⚠️")
 
 	if payload:
 		payload.append("")
@@ -143,43 +142,43 @@ def prepare_profile_payload(service, user, ticker):
 	for key, label in (('revenueEstimateY', 'Revenue forecast (1Y):'), ('earningsEstimateY', 'Earnings forecast (1Y):')):
 		if key in data:
 			growth = int(round(data[key]))
-			payload.append(webhook.bold(label, service) + (f" +{growth}% " if growth > 0 else f" {growth}% ⚠️ "))
+			payload.append(service.bold(label) + (f" +{growth}% " if growth > 0 else f" {growth}% ⚠️ "))
 	if 'insiderBuy' in data:
 		if data['insiderBuy'] > data['insiderSell']:
-			payload.append(webhook.bold("Net insider action (3M):", service) + f" Buy {currency} {formats.humanUnits(data['insiderBuyValue'])} ")
+			payload.append(service.bold("Net insider action (3M):") + f" Buy {currency} {formats.humanUnits(data['insiderBuyValue'])} ")
 		elif data['insiderBuy'] < data['insiderSell']:
-			payload.append(webhook.bold("Net insider action (3M):", service) + f" Sell {currency} {formats.humanUnits(data['insiderSellValue'])}⚠️ ")
+			payload.append(service.bold("Net insider action (3M):") + f" Sell {currency} {formats.humanUnits(data['insiderSellValue'])}⚠️ ")
 	if 'short_percent' in data:
-		payload.append(webhook.bold("Shorted stock:", service) + f" {data['short_percent']}%" + ('⚠️ ' if data['short_percent'] > 10 else ''))
+		payload.append(service.bold("Shorted stock:") + f" {data['short_percent']}%" + ('⚠️ ' if data['short_percent'] > 10 else ''))
 	if 'recommend' in data:
-		payload.append(webhook.bold("Score:", service) + f" {data['recommend_index']} {data['recommend'].replace('_', ' ')} ({data['recommend_analysts']} analysts)")
+		payload.append(service.bold("Score:") + f" {data['recommend_index']} {data['recommend'].replace('_', ' ')} ({data['recommend_analysts']} analysts)")
 
 	if payload:
 		payload.append("")
 
 	if 'regularMarketPrice' in data:
-		price_line = webhook.bold("Price:", service) + f" {currency} {data['regularMarketPrice']:,.2f}"
+		price_line = service.bold("Price:") + f" {currency} {data['regularMarketPrice']:,.2f}"
 		if state != 'REGULAR' and 'prePostMarketPrice' in data:
 			price_line += f" ({data['prePostMarketPrice']:,.2f} after hrs)"
 		payload.append(price_line)
 	trailing_pe = str(int(round(data['price_to_earnings_trailing']))) if 'price_to_earnings_trailing' in data else 'N/A ⚠️ '
 	if data.get('net_income', 0) > 0:
-		payload.append(webhook.bold("Trailing P/E:", service) + f" {trailing_pe}")
+		payload.append(service.bold("Trailing P/E:") + f" {trailing_pe}")
 	if data.get('netIncomeToCommon', 0) > 0 and data.get('sharesOutstanding') and 'regularMarketPrice' in data:
 		current_pe = round(data['regularMarketPrice'] / (data['netIncomeToCommon'] / data['sharesOutstanding']))
-		payload.append(webhook.bold("Current P/E:", service) + f" {current_pe}")
+		payload.append(service.bold("Current P/E:") + f" {current_pe}")
 	if 'price_to_earnings_forward' in data:
 		forward_pe = int(round(data['price_to_earnings_forward']))
 		warn = forward_pe < 0
 		if 'profile_industry' in data: # software is dear: it warns above 100, the rest above 30 or above their trailing P/E
 			warn = warn or forward_pe > (100 if 'Software' in data['profile_industry'] else 30)
 			warn = warn or ('price_to_earnings_trailing' in data and forward_pe > int(trailing_pe))
-		payload.append(webhook.bold("Forward P/E:", service) + f" {forward_pe} " + ('⚠️ ' if warn else ''))
+		payload.append(service.bold("Forward P/E:") + f" {forward_pe} " + ('⚠️ ' if warn else ''))
 	if 'price_to_earnings_peg' in data:
-		payload.append(webhook.bold("PEG ratio:", service) + f" {round(data['price_to_earnings_peg'], 1)}")
+		payload.append(service.bold("PEG ratio:") + f" {round(data['price_to_earnings_peg'], 1)}")
 	if 'price_to_sales' in data:
 		price_to_sales = round(data['price_to_sales'], 1)
-		payload.append(webhook.bold("PS ratio:", service) + f" {price_to_sales} " + ('⚠️ ' if price_to_sales > 15 else ''))
+		payload.append(service.bold("PS ratio:") + f" {price_to_sales} " + ('⚠️ ' if price_to_sales > 15 else ''))
 
 	if payload:
 		payload.append("")
@@ -192,18 +191,18 @@ def prepare_profile_payload(service, user, ticker):
 	for interval in ('5Y', '1Y', '1M', '1D'):
 		if interval in price_history:
 			percent = price_history[interval]
-			payload.append(f"{formats.get_emoji(percent)} {webhook.bold(interval + ':', service)} {percent}%")
+			payload.append(f"{formats.get_emoji(percent)} {service.bold(interval + ':')} {percent}%")
 	if state != 'REGULAR':
 		for key, label in (('percent_change_premarket', 'Pre-market:'), ('percent_change_postmarket', 'Post-market:')):
 			if key in data:
-				payload.append(f"{formats.get_emoji(data[key])} {webhook.bold(label, service)} {data[key]:,}%")
+				payload.append(f"{formats.get_emoji(data[key])} {service.bold(label)} {data[key]:,}%")
 				break
 
 	if 'profile_website' in data and config.prefs.hyperlinkFooter and config.prefs.hyperlink:
 		if payload:
 			payload.append("")
 		payload.append(f"{gfinance_link} | {links.yahoo_link(ticker, service, brief=True, text='yahoo')} | {sws_link}")
-	heading = f"{webhook.bold(title, service)} ({ticker_link}) {MARKET_STATE_EMOJI.get(state, '🔴') if state else ''}"
+	heading = f"{service.bold(title)} ({ticker_link}) {MARKET_STATE_EMOJI.get(state, '🔴') if state else ''}"
 	if ticker_orig == ticker:
 		payload.insert(0, heading)
 	else:
diff --git a/lib/reports/rankings.py b/lib/reports/rankings.py
index 68911256..98bcb6c4 100644
--- a/lib/reports/rankings.py
+++ b/lib/reports/rankings.py
@@ -5,7 +5,6 @@
 from lib import formats
 from lib import links
 from lib import markets
-from lib import webhook
 from lib import yahoo
 
 def last_col(line):
@@ -39,7 +38,7 @@ def prepare_marketcap_payload(service, action='top', length=15):
 	if not caps:
 		return ['No market cap data found']
 	caps.sort(key=lambda cap: cap[0], reverse=action == 'top')
-	return [webhook.bold(f'{action.title()} {length} tracked stocks by market cap (USD)', service)] + [line for _, line in caps[:length]]
+	return [service.bold(f'{action.title()} {length} tracked stocks by market cap (USD)')] + [line for _, line in caps[:length]]
 
 def prepare_rating_payload(service, action, length=15):
 	"""(payload, chart) of the `length` tracked stocks rated `action` ('buy' or 'sell') by min_analysts or more analysts (fewer,
@@ -65,7 +64,7 @@ def prepare_rating_payload(service, action, length=15):
 	# zoom the axis onto the ratings so the bars fill the panel (ratings only run 1-5)
 	chart = {'rows': shown, 'title': message, 'subtitle': 'Mean analyst rating: 1 = strong buy, 5 = strong sell (axis zoomed)',
 		'value_fmt': lambda v: f"{v:g}", 'xlim': (max(0.9, low - spread * 0.9), min(5.1, high + spread * 0.25))}
-	return [webhook.bold(message, service)] + payload, chart
+	return [service.bold(message)] + payload, chart
 
 def prepare_value_payload(service, action='pe', ticker_select=None, length=15):
 	"""(payload, chart): the tracked stocks by a ratio (`action`), or one ticker's. A forward P/E is analysts' estimate, so its lists
@@ -142,7 +141,7 @@ def prepare_value_payload(service, action='pe', ticker_select=None, length=15):
 			payload = [f"No tracked stocks with {config.prefs.min_analysts}+ analysts have a {heading_trail} ratio" if covered is not None else f"{action}: no data found"]
 		else:
 			heading = f'{heading_type} {length} tracked stocks by {heading_trail} ratio{coverage}'
-			payload.insert(0, f"{webhook.bold(heading, service)}")
+			payload.insert(0, f"{service.bold(heading)}")
 			chart = {'rows': chart_rows[:length], 'title': heading, 'value_fmt': lambda v: f"{v:g}"}
 	return payload, chart
 
@@ -161,4 +160,4 @@ def prepare_beta_payload(service):
 	found.sort(key=lambda row: row[0], reverse=True)
 	heading = 'Beta over 1.5 and mkt cap under US$2B'
 	chart = {'rows': [row for _, _, row in found], 'title': heading, 'value_fmt': lambda v: f"{v:g}"}
-	return [webhook.bold(heading, service)] + [line for _, line, _ in found], chart
+	return [service.bold(heading)] + [line for _, line, _ in found], chart
diff --git a/lib/reports/rating.py b/lib/reports/rating.py
index 9ac394d9..2b7899cc 100644
--- a/lib/reports/rating.py
+++ b/lib/reports/rating.py
@@ -36,13 +36,13 @@ def get_emoji(old, new):
 				pass
 			if old_action and action != old_action:
 				emoji = get_emoji(old_index, index)
-				message = f"{webhook.bold(f'{old_index} {old_action}', service)} to {webhook.bold(f'{index} {action}', service)} ({analysts} analysts)"
+				message = f"{service.bold(f'{old_index} {old_action}')} to {service.bold(f'{index} {action}')} ({analysts} analysts)"
 				payload.append(f"{emoji} {title} ({ticker_link}) changed from {message}")
 		payload.sort()
 		if payload:
 			if not specific_stock:
 				message = 'Consensus rating changes:'
-				message = webhook.bold(message, service)
+				message = service.bold(message)
 				payload.insert(0, message)
 		else:
 			if interactive:
diff --git a/lib/reports/reminder.py b/lib/reports/reminder.py
index 4d4f87bf..a9e161aa 100644
--- a/lib/reports/reminder.py
+++ b/lib/reports/reminder.py
@@ -4,7 +4,6 @@
 from dateutil.relativedelta import relativedelta
 from lib import config
 from lib import formats
-from lib import links
 from lib import webhook
 
 
@@ -46,13 +45,13 @@ def prepare_finance_calendar_payload(service):
 				payload.append("• work expenses‡")
 				payload.append("• investment subscriptions§")
 				payload.append ("")
-				abr_link = links.link('https://abr.business.gov.au/Tools/DgrListing', 'DGR-registered', service)
-				stake_note = links.link('https://hellostake.com/au/support/stake-super/employer-and-personal-contributions/33442327233305#h_01JQAPN9A6ZM41NZKP6XZWN1KP', 'note about XX PCC in transaction description', service)
-				work_link = links.link('https://www.ato.gov.au/individuals-and-families/income-deductions-offsets-and-records/deductions-you-can-claim', 'eligible work expenses', service)
-				payload.append(webhook.italic("* Organisation must be " + abr_link, service))
-				payload.append(webhook.italic("† Allow transfer time for Super, as contributions can only be deducted for the year they're received. Stake users: see " + stake_note, service))
-				payload.append(webhook.italic("‡ See " + work_link, service))
-				payload.append(webhook.italic("§ Must relate directly to specific trades and gains/income (e.g. not general finance news)", service))
+				abr_link = service.link('https://abr.business.gov.au/Tools/DgrListing', 'DGR-registered')
+				stake_note = service.link('https://hellostake.com/au/support/stake-super/employer-and-personal-contributions/33442327233305#h_01JQAPN9A6ZM41NZKP6XZWN1KP', 'note about XX PCC in transaction description')
+				work_link = service.link('https://www.ato.gov.au/individuals-and-families/income-deductions-offsets-and-records/deductions-you-can-claim', 'eligible work expenses')
+				payload.append(service.italic("* Organisation must be " + abr_link))
+				payload.append(service.italic("† Allow transfer time for Super, as contributions can only be deducted for the year they're received. Stake users: see " + stake_note))
+				payload.append(service.italic("‡ See " + work_link))
+				payload.append(service.italic("§ Must relate directly to specific trades and gains/income (e.g. not general finance news)"))
 
 		if month_and_day == '08-18':
 			myBirthday = datetime.datetime(2022,8,18,0,0,0,0, tzinfo=tz)
@@ -60,7 +59,7 @@ def prepare_finance_calendar_payload(service):
 			payload.append("It's my " + str(difference.years) + formats.ordinal(difference.years) + " birthday! 🥳")
 
 		if payload:
-			heading = webhook.bold("Finance event reminders:", service)
+			heading = service.bold("Finance event reminders:")
 			payload.insert(0, heading)
 		return payload
 
diff --git a/lib/reports/shorts.py b/lib/reports/shorts.py
index 2230ef27..e27401c2 100644
--- a/lib/reports/shorts.py
+++ b/lib/reports/shorts.py
@@ -1,7 +1,6 @@
 """.shorts: the most shorted stocks followed, from Short Man (ASX) and Yahoo (scheduled: shorts.py)."""
 from lib import config
 from lib import webhook
-from lib import links
 from lib import markets
 from lib import render
 from lib import watchlist
@@ -21,11 +20,11 @@ def prepare_shorts_payload(service, market_data):
 			if short_percent > threshold or specific_stock:
 				code = ticker.split('.')[0]
 				url = 'https://www.shortman.com.au/stock?q=' + code if ticker.endswith('.AX') else 'https://finance.yahoo.com/quote/' + ticker + '/key-statistics?p=' + ticker
-				found.append((short_percent, f"🩳 {market_data[ticker]['profile_title']} ({links.link(url, code, service)}) {round(short_percent)}%", code))
+				found.append((short_percent, f"🩳 {market_data[ticker]['profile_title']} ({service.link(url, code)}) {round(short_percent)}%", code))
 		found.sort(key=lambda row: row[0], reverse=True) # largest first, as the chart has them
 		payload, graph = [line for _, line, _ in found], None
 		if payload and not specific_stock:
-			payload.insert(0, webhook.bold(f'Tracked stocks shorted over {threshold:g}%', service))
+			payload.insert(0, service.bold(f'Tracked stocks shorted over {threshold:g}%'))
 			if webhook.takes_charts(service, interactive):
 				graph = render.shorts_chart([(code, percent) for percent, _, code in found], threshold)
 		elif not payload and interactive:
diff --git a/lib/reports/trades.py b/lib/reports/trades.py
index 16cf07e5..e6333851 100644
--- a/lib/reports/trades.py
+++ b/lib/reports/trades.py
@@ -70,7 +70,7 @@ def prepare_trade_payload(service, trades):
 				currency = currency_temp
 			sharesight_url = "https://portfolio.sharesight.com/holdings/"
 			trade_url = sharesight_url + holding_id + '/dashboard/transactions'
-			trade_link = links.link(trade_url, action, service)
+			trade_link = service.link(trade_url, action)
 			holding_link = links.finance_link(symbol, market, service)
 			payload_staging.append([dt_date, trade_id, emoji, portfolio_name, trade_link, currency, f'{value}', holding_link, flag])
 
@@ -80,7 +80,7 @@ def prepare_trade_payload(service, trades):
 			if len(dates) > 1 or (interactive and days > 1):
 				human_date = date.strftime('%b %d') # Dec 30
 				payload.append("")
-				payload.append(webhook.bold(human_date, service))
+				payload.append(service.bold(human_date))
 			for trade in payload_staging:
 				if date == trade[0]:
 					payload.append(' '.join(trade[2:]))
diff --git a/lib/webhook.py b/lib/webhook.py
index 0aad651f..9391b353 100644
--- a/lib/webhook.py
+++ b/lib/webhook.py
@@ -7,6 +7,7 @@
 import traceback
 
 from lib import config
+from lib import dialects
 from lib import net
 from lib import telegram
 
@@ -65,7 +66,7 @@ def chat_url(service, chat_id):
 def default_chats():
 	"""Where a scheduled report goes: [(service, url, chat id)]. Telegram posts into telegramChatID (left out without one); Slack and
 	Discord post through their incoming webhooks, which take no chat id. Raises RuntimeError when there is nowhere to post."""
-	chats = [(service, chat_url(service, config.env.telegramChatID), config.env.telegramChatID) if service == 'telegram' else (service, url, None)
+	chats = [(dialects.TELEGRAM, chat_url(service, config.env.telegramChatID), config.env.telegramChatID) if service == 'telegram' else (dialects.of(service), url, None)
 		for service, url in config.env.webhooks.items() if service != 'telegram' or config.env.telegramChatID]
 	if not chats:
 		raise RuntimeError('nowhere to post: finbot.env sets no telegramChatID, slack_webhook or discord_webhook')
@@ -73,7 +74,7 @@ def default_chats():
 
 def destinations(interactive, service, chat_id):
 	"""Where a report goes: back to the chat that asked for it (interactive), else to the default chats."""
-	return [(service, chat_url(service, chat_id), chat_id)] if interactive else default_chats()
+	return [(dialects.of(service), chat_url(service, chat_id), chat_id)] if interactive else default_chats()
 
 def send(service, url, chat_id, payload, image=None, reply_markup=None):
 	"""A report: its chart with the text as caption when there is one (sendPhoto falls back to the text where a chart cannot go),
@@ -165,35 +166,6 @@ def chunked(lines, limit, max_lines):
 		chunks.append(current)
 	return chunks
 
-MARKUP = { # how each service marks text up: style -> service -> (before, after)
-	'bold': {'telegram': ('<b>', '</b>'), 'slack': ('*', '*'), 'discord': ('**', '**')},
-	'italic': {'telegram': ('<i>', '</i>'), 'slack': ('_', '_'), 'discord': ('_', '_')},
-	'strike': {'telegram': ('<s>', '</s>'), 'slack': ('~', '~'), 'discord': ('~~', '~~')},
-}
-
-def escape(text, service):
-	"""Text a user wrote (a .plan entry, a name) made safe to show: Telegram's HTML and Slack's markup both take &amp; &lt; &gt;.
-	It is unescaped first because it may arrive escaped already (Slack escapes what it sends; bot.py escapes Telegram names):
-	escaping twice would show &amp;amp;. Discord's markdown shows entities as they are, so it gets the plain text."""
-	text = html.unescape(str(text))
-	return text if service == 'discord' else html.escape(text, quote=False)
-
-def _styled(style, message, service):
-	if service not in MARKUP[style]:
-		return message
-	before, after = MARKUP[style][service]
-	return before + message + after
-
-def bold(message, service):
-	return _styled('bold', message, service)
-
-def italic(message, service):
-	return _styled('italic', message, service)
-
-def strike(message, service):
-	return _styled('strike', message, service)
-
-
 def fit_caption(caption, limit=1024, visible=None):
 	"""Telegram rejects photo captions over 1024 visible characters; Discord takes 2,000 characters, markup included
 	(visible=len). Keep whole lines that fit and say how many were left out."""
diff --git a/lib/worker.py b/lib/worker.py
index 43456348..7b7631fc 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -7,6 +7,7 @@
 import sys
 import threading
 from lib import config
+from lib import dialects
 from lib import state
 from lib import words
 from lib import formats
@@ -113,6 +114,7 @@ def process_job_callback(service, callback_id, chat_id, message_id, data):
 
 def process_callback(service, callback_id, chat_id, message_id, data):
 	"""Button press: run a scheduled command from .jobs, or rebuild a chart message for the chosen period and edit it in place."""
+	service = dialects.of(service)
 	if data.startswith('j|'):
 		return process_job_callback(service, callback_id, chat_id, message_id, data)
 	parts = data.split('|', 3)
@@ -132,7 +134,7 @@ def process_callback(service, callback_id, chat_id, message_id, data):
 def process_request(service, chat_id, user, message, botName, userRealName):
 	"""Entry point for inbound chat requests: the first command in COMMANDS that fits the message answers it (a message that fits
 	none gets no answer). Any failure becomes a one-line reply in the originating chat, after the typing stops."""
-	request = Request(service, chat_id, user, userRealName, botName)
+	request = Request(dialects.of(service), chat_id, user, userRealName, botName)
 	try:
 		for pattern, handler in COMMANDS:
 			if m := match(pattern, message, botName):
@@ -216,7 +218,7 @@ def on_help(request, m):
 
 def on_hello(request, m):
 	"""Easter egg: a greeting, alliterated when the alliterate preference is on."""
-	verbs = words.verb + [f"{webhook.strike(said, request.service)} I mean {meant}" for said, meant in (('study', 'meet'), ('observe', 'see'), ('profile', 'know'))]
+	verbs = words.verb + [f"{request.service.strike(said)} I mean {meant}" for said, meant in (('study', 'meet'), ('observe', 'see'), ('profile', 'know'))]
 	while True:
 		first, second = random.choice(words.adjectives), random.choice(words.adjectives_two)
 		if not config.prefs.alliterate or first[0].lower() == second[0].lower():
@@ -228,7 +230,7 @@ def on_hello(request, m):
 def on_thanks(request, m):
 	"""Easter egg: you're welcome, glitching one time in a thousand and on April Fools' Day."""
 	today = datetime.date.today()
-	glitch = webhook.strike(GLITCH, request.service) if random.randrange(1, 1000) == 1 or (today.month, today.day) == (4, 1) else ''
+	glitch = request.service.strike(GLITCH) if random.randrange(1, 1000) == 1 or (today.month, today.day) == (4, 1) else ''
 	request.say([f"{glitch}You're {random.choice(words.adjectives)} welcome, {request.user_name}! 😇"])
 
 def on_calendar(request, m):
@@ -352,8 +354,8 @@ def on_plan(request, m):
 	plan = state.load('finbot_plan.json', {})
 	def who(user): # a Slack user is a mention (<@U123>), which is markup to keep; elsewhere a handle is text
 		user = user.removeprefix('@')
-		return user if request.service == 'slack' else webhook.escape(user, request.service)
-	request.say([f"{webhook.bold(who(k), request.service)}: {webhook.italic(webhook.escape(v, request.service), request.service)}\n"
+		return user if request.service == 'slack' else request.service.escape(user)
+	request.say([f"{request.service.bold(who(k))}: {request.service.italic(request.service.escape(v))}\n"
 		for k, v in plan.items()])
 
 def on_who(request, m):
diff --git a/tests/common.py b/tests/common.py
index 834490e3..74803fa9 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -10,7 +10,7 @@
 os.environ.update(telegramOutgoingWebhook='https://example.com/telegram', telegramOutgoingToken='tok', telegramChatID='1')
 
 import bot # first, and once: it monkey-patches with gevent, which must happen before requests imports ssl and never mid-test
-from lib import buttons, cache, config, state, webhook, formats, links, markets, render, yahoo, sharesight, shortman, worker, charts, telegram, watchlist, words
+from lib import buttons, cache, config, dialects, state, webhook, formats, links, markets, render, yahoo, sharesight, shortman, worker, charts, telegram, watchlist, words
 from lib.reports import about, cal, history, holdings, milestone, performance, price, profile, rankings, rating, reminder, shorts, trades
 
 REAL = {(m.__name__, n): getattr(m, n) for m, n in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (yahoo, 'price_history'),
@@ -20,6 +20,7 @@
 REAL_REPORT_ERROR = webhook.report_error # likewise
 REAL_SEND_PHOTO = webhook.sendPhoto # likewise
 TICKERS = [f'T{i}' for i in range(25)]
+TELEGRAM, SLACK, DISCORD = dialects.TELEGRAM, dialects.SLACK, dialects.DISCORD
 BOT = '@bot'
 
 
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 52d9152c..19550542 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -91,7 +91,7 @@ def slow_fetch(tickers):
 			time.sleep(0.01)
 			return {t: market_entry(t) for t in tickers}
 		self.patch(yahoo, 'fetch', slow_fetch)
-		threads = [threading.Thread(target=holdings.prepare_watchlist, args=('telegram', '@u', 'add', t)) for t in ('T1', 'T2')]
+		threads = [threading.Thread(target=holdings.prepare_watchlist, args=(TELEGRAM, '@u', 'add', t)) for t in ('T1', 'T2')]
 		for thread in threads:
 			thread.start()
 		for thread in threads:
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
index 9f1c6f33..86e13b9b 100644
--- a/tests/test_jobs.py
+++ b/tests/test_jobs.py
@@ -496,13 +496,13 @@ def test_list_in_chat_is_just_the_commands(self):
 		self.assertIn('6. price 7 (running)', '\n'.join(map(str, self.list_in_chat())))
 
 	def test_telegram_gets_a_button_per_command_and_slack_does_not(self):
-		payload, markup = about.prepare_jobs_payload('telegram', '')
+		payload, markup = about.prepare_jobs_payload(TELEGRAM, '')
 		buttons = [b for row in markup['inline_keyboard'] for b in row]
 		self.assertEqual([b['text'] for b in buttons], ['1. performance 7', '2. performance 30', '3. performance 90', '4. performance 180',
 													  '5. performance 365', '6. price 7', '7. price close', '8. trades'])
 		for b in buttons:
 			self.assertLessEqual(len(b['callback_data'].encode()), 64)
-		payload, markup = about.prepare_jobs_payload('slack', '')
+		payload, markup = about.prepare_jobs_payload(SLACK, '')
 		self.assertIsNone(markup)
 
 	def test_the_chat_command_passes_the_keyboard_on(self):
@@ -514,7 +514,7 @@ def test_the_chat_command_passes_the_keyboard_on(self):
 	def test_button_press_runs_that_command(self):
 		sent = []
 		self.patch(telegram, 'answerCallbackQuery', lambda cid, text=None: sent.append(text))
-		markup = about.prepare_jobs_payload('telegram', '')[1]
+		markup = about.prepare_jobs_payload(TELEGRAM, '')[1]
 		data = next(b['callback_data'] for row in markup['inline_keyboard'] for b in row if b['text'].startswith('7.'))
 		self.sent.clear()
 		worker.process_callback('telegram', 'cb1', '55', '9', data)
@@ -559,7 +559,7 @@ def test_button_from_an_old_schedule_updates_the_message(self):
 		chat, mid, text, markup = self.edits[0]
 		self.assertEqual((chat, mid), ('55', '9')) # the message with the old buttons is edited in place
 		self.assertIn('7. price close', text)
-		self.assertEqual(markup, about.prepare_jobs_payload('telegram', '')[1]) # with the current buttons
+		self.assertEqual(markup, about.prepare_jobs_payload(TELEGRAM, '')[1]) # with the current buttons
 		self.patch(jobs, 'current', None) # scheduler off: the list is replaced by the notice and the buttons removed
 		worker.process_callback('telegram', 'cb1', '55', '9', 'j|7|0000')
 		self.assertEqual(self.edits[-1][2:], ('Scheduled jobs are off (scheduler=False)', {'inline_keyboard': []}))
diff --git a/tests/test_providers.py b/tests/test_providers.py
index d3e368c7..a9063c56 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -673,7 +673,7 @@ def test_a_long_report_is_split_by_size(self):
 		self.assertTrue(all(size <= 2000 for size in sizes), sizes) # 20 lines a chunk used to make 3,019 characters
 
 	def test_links_are_masked_links(self):
-		self.assertEqual(links.link('https://x.example/a(b)', 'T[1]', 'discord'), '[T\\[1\\]](<https://x.example/a%28b%29>)')
+		self.assertEqual(DISCORD.link('https://x.example/a(b)', 'T[1]'), '[T\\[1\\]](<https://x.example/a%28b%29>)')
 
 
 class SlackSending(FinbotCase):
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 271679e6..b975117d 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -58,23 +58,23 @@ def test_unknown_stickers_and_noise_are_ignored(self):
 
 class ReportFunctions(FinbotCase):
 	def test_prepare_help(self):
-		self.assertTrue(about.prepare_help('telegram', BOT))
+		self.assertTrue(about.prepare_help(TELEGRAM, BOT))
 
 	def test_a_portfolio_with_no_holdings_is_reported_by_name(self):
 		"""Its name used to be read from its first holding: an empty portfolio crashed .performance."""
 		self.patch(sharesight, 'get_performance', lambda pid, d, *a: {'report': {'holdings': [], 'capital_gain_percent': '0.0'}})
-		payload, _ = performance.report('telegram', 7)
+		payload, _ = performance.report(TELEGRAM, 7)
 		self.assertTrue(any('>Rob</a>' in line for line in payload), payload)
 
 	def test_prepare_holdings(self):
-		self.assertTrue(holdings.prepare_holdings_payload(None, 'telegram'))
+		self.assertTrue(holdings.prepare_holdings_payload(None, TELEGRAM))
 
 	def test_prepare_marketcap(self):
-		self.assertTrue(rankings.prepare_marketcap_payload('telegram', 'top', length=5))
+		self.assertTrue(rankings.prepare_marketcap_payload(TELEGRAM, 'top', length=5))
 
 	def test_prepare_rating(self):
 		for action in ('buy', 'sell'):
-			self.assertTrue(rankings.prepare_rating_payload('telegram', action, length=5)[0])
+			self.assertTrue(rankings.prepare_rating_payload(TELEGRAM, action, length=5)[0])
 
 	def test_ratings_list_the_strongest_first(self):
 		"""1 is a strong buy and 5 a strong sell: .sell used to list the mildest sells first. On equal ratings, more analysts first;
@@ -84,13 +84,13 @@ def test_ratings_list_the_strongest_first(self):
 		self.patch(watchlist, 'with_holdings', lambda: list(ratings))
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | dict(zip(('recommend', 'recommend_index', 'recommend_analysts'), ratings[t]))})
 		def order(action):
-			payload, chart = rankings.prepare_rating_payload('telegram', action)
+			payload, chart = rankings.prepare_rating_payload(TELEGRAM, action)
 			return [line.split('Co ')[1].split(' ')[0] for line in payload[1:]], [label for label, _ in chart['rows']]
 		self.assertEqual(order('buy'), (['T2', 'T3', 'T1'], ['T2', 'T3', 'T1']))
 		self.assertEqual(order('sell'), (['T5', 'T6', 'T4'], ['T5', 'T6', 'T4']))
-		self.assertIn('(7+ analysts)', rankings.prepare_rating_payload('telegram', 'buy')[0][0])
+		self.assertIn('(7+ analysts)', rankings.prepare_rating_payload(TELEGRAM, 'buy')[0][0])
 		self.patch(watchlist, 'with_holdings', lambda: ['T7', 'T8'])
-		self.assertEqual(rankings.prepare_rating_payload('telegram', 'buy'), (['No tracked stocks rated buy by 7+ analysts'], None))
+		self.assertEqual(rankings.prepare_rating_payload(TELEGRAM, 'buy'), (['No tracked stocks rated buy by 7+ analysts'], None))
 
 	def test_a_small_cap_quoted_in_won_needs_the_bigger_move_too(self):
 		"""demote_volatile compares market caps in US dollars: unconverted, a small cap quoted in won was never small."""
@@ -120,16 +120,16 @@ def test_forward_pe_lists_need_as_many_analysts_as_buy(self):
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | {'price_to_earnings_forward': -5 if t == 'T3' else 20} for t in ts})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'recommend_analysts': analysts[t]}})
 		def listed(action):
-			payload, chart = rankings.prepare_value_payload('telegram', action)
+			payload, chart = rankings.prepare_value_payload(TELEGRAM, action)
 			return payload[0], [line.split('Co ')[1].split(' ')[0] for line in payload[1:]], chart
 		heading, names, chart = listed('forward pe')
 		self.assertEqual((names, [label for label, _ in chart['rows']]), (['T1'], ['T1'])) # T2 and T4: too few; T3: negative
 		self.assertIn('forward pe ratio (7+ analysts)', heading)
 		self.assertEqual(listed('negative forward pe')[1], ['T3'])
-		self.assertIn('Co T2', str(rankings.prepare_value_payload('telegram', 'forward pe', 'T2'))) # one ticker asked for: shown
+		self.assertIn('Co T2', str(rankings.prepare_value_payload(TELEGRAM, 'forward pe', 'T2'))) # one ticker asked for: shown
 		self.prefs(min_analysts=13) # the one setting moves .buy and .forward pe together
-		self.assertEqual(rankings.prepare_value_payload('telegram', 'forward pe')[0], ['No tracked stocks with 13+ analysts have a forward pe ratio'])
-		self.assertEqual(rankings.prepare_rating_payload('telegram', 'buy')[0], ['No tracked stocks rated buy by 13+ analysts'])
+		self.assertEqual(rankings.prepare_value_payload(TELEGRAM, 'forward pe')[0], ['No tracked stocks with 13+ analysts have a forward pe ratio'])
+		self.assertEqual(rankings.prepare_rating_payload(TELEGRAM, 'buy')[0], ['No tracked stocks rated buy by 13+ analysts'])
 
 	def test_market_caps_compare_in_us_dollars(self):
 		"""Yahoo gives a market cap in the stock's own currency: unconverted, a won market cap topped every list."""
@@ -141,13 +141,13 @@ def fetch(tickers):
 		self.patch(yahoo, 'fetch', fetch)
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: fetch([t]))
 		self.patch(watchlist, 'with_holdings', lambda: list(stocks))
-		payload = rankings.prepare_marketcap_payload('telegram')
+		payload = rankings.prepare_marketcap_payload(TELEGRAM)
 		self.assertIn('by market cap (USD)', payload[0])
 		self.assertEqual([line.split('Co ')[1].split(' ')[0] for line in payload[1:]], ['U1', 'K1.KS', 'L1.L', 'K2.KS']) # Z1: no rate, left out
 		self.assertIn('mkt cap: 731.88 B', payload[2]) # 1.01 Q won
-		self.assertIn('mkt cap: KRW 1.01 Q (USD 731.88 B)', rankings.prepare_marketcap_payload('telegram', 'K1.KS')[0])
-		self.assertIn('mkt cap: GBP 62.50 B (USD 79.11 B)', rankings.prepare_marketcap_payload('telegram', 'L1.L')[0]) # pence quote, pound cap
-		beta = rankings.prepare_beta_payload('telegram')[0]
+		self.assertIn('mkt cap: KRW 1.01 Q (USD 731.88 B)', rankings.prepare_marketcap_payload(TELEGRAM, 'K1.KS')[0])
+		self.assertIn('mkt cap: GBP 62.50 B (USD 79.11 B)', rankings.prepare_marketcap_payload(TELEGRAM, 'L1.L')[0]) # pence quote, pound cap
+		beta = rankings.prepare_beta_payload(TELEGRAM)[0]
 		self.assertIn('Co K2.KS', str(beta)) # 500 billion won is US$362M: a small cap, which it never was unconverted
 		self.assertNotIn('Co K1.KS', str(beta))
 
@@ -155,18 +155,18 @@ def test_prepare_value(self):
 		for action, heading in (('pe', 'Top 5 tracked stocks by pe ratio'), ('bottom forward pe', 'Bottom 5 tracked stocks by forward pe ratio'),
 								('peg', 'Top 5 tracked stocks by peg ratio'), ('negative peg', 'negative peg: no data found')):
 			with self.subTest(action=action):
-				payload, chart = rankings.prepare_value_payload('telegram', action, length=5)
+				payload, chart = rankings.prepare_value_payload(TELEGRAM, action, length=5)
 				self.assertIn(heading, payload[0])
 
 	def test_prepare_beta(self):
-		payload, chart = rankings.prepare_beta_payload('telegram')
+		payload, chart = rankings.prepare_beta_payload(TELEGRAM)
 		self.assertIn('Beta over 1.5', payload[0])
 		self.assertEqual(len(chart['rows']), len(payload) - 1) # a bar per stock listed
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'beta': 1.0}})
-		self.assertEqual(rankings.prepare_beta_payload('telegram'), (['No tracked stocks have a beta over 1.5 and mkt cap under US$2B'], None))
+		self.assertEqual(rankings.prepare_beta_payload(TELEGRAM), (['No tracked stocks have a beta over 1.5 and mkt cap under US$2B'], None))
 
 	def test_prepare_profile(self):
-		self.assertTrue(profile.prepare_profile_payload('telegram', '@u', 'T1'))
+		self.assertTrue(profile.prepare_profile_payload(TELEGRAM, '@u', 'T1'))
 
 	def test_a_profile_leaves_out_what_yahoo_did_not_send(self):
 		"""Each of these fields missing (or a zero equity) used to crash the whole profile."""
@@ -175,22 +175,22 @@ def test_a_profile_leaves_out_what_yahoo_did_not_send(self):
 		for missing in ('marketState', 'profile_exchange', 'market_cap', 'currency', 'regularMarketPrice', 'sharesOutstanding', 'percent_change'):
 			with self.subTest(missing=missing):
 				self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: {k: v for k, v in full.items() if k != missing}})
-				self.assertIn('Co T1', profile.prepare_profile_payload('telegram', '@u', 'T1')[0])
+				self.assertIn('Co T1', profile.prepare_profile_payload(TELEGRAM, '@u', 'T1')[0])
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: full | {'shareholder_equity': 0}})
-		self.assertNotIn('Net debt/equity', str(profile.prepare_profile_payload('telegram', '@u', 'T1')))
+		self.assertNotIn('Net debt/equity', str(profile.prepare_profile_payload(TELEGRAM, '@u', 'T1')))
 
 	def test_a_profile_warns_of_a_ps_ratio_only_above_15(self):
 		for ps, warned in ((12.3, False), (15.1, True)): # forward P/E 99 warns (above 30): its warning used to show on the PS line too
 			with self.subTest(ps=ps):
 				self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'price_to_sales': ps, 'price_to_earnings_forward': 99, 'profile_industry': 'Tools'}})
-				line = next(line for line in profile.prepare_profile_payload('telegram', '@u', 'T1') if 'PS ratio' in line)
+				line = next(line for line in profile.prepare_profile_payload(TELEGRAM, '@u', 'T1') if 'PS ratio' in line)
 				self.assertEqual('⚠️' in line, warned, line)
 
 	def test_prepare_watchlist(self):
-		self.assertTrue(holdings.prepare_watchlist('telegram', '@u'))
+		self.assertTrue(holdings.prepare_watchlist(TELEGRAM, '@u'))
 
 	def test_prepare_compare_usage(self):
-		caption = history.prepare_compare('telegram', ['T1'])[0]
+		caption = history.prepare_compare(TELEGRAM, ['T1'])[0]
 		self.assertIn('Usage', ' '.join(caption))
 
 
@@ -413,7 +413,7 @@ def test_by_industry_default(self):
 		self.assertIn('Rob by industry', sent[0][2])
 
 	def test_by_sector_and_payload(self):
-		payload, chart = holdings.prepare_pie('rob', 'sector', 'telegram')
+		payload, chart = holdings.prepare_pie('rob', 'sector', TELEGRAM)
 		self.assertEqual([r[0] for r in chart['rows']], ['Electronic Technology', 'Health Technology', 'Unclassified'])
 		self.assertIn('Electronic Technology: 72.7%', payload) # 800 of 1100; zero-value Gold is left out
 		self.assertEqual(self.command('.pie Rob sector')[0][0], 'photo')
@@ -421,7 +421,7 @@ def test_by_sector_and_payload(self):
 	def test_folds_into_other_and_renders(self):
 		held = {f'T{i}:X': {'code': f'T{i}', 'industry_classification_name': f'Ind {i}', 'value': 100 - i} for i in range(12)}
 		self.patch(sharesight.Portfolio, 'holdings', lambda portfolio: held)
-		payload, chart = holdings.prepare_pie('Rob', 'industry', 'telegram')
+		payload, chart = holdings.prepare_pie('Rob', 'industry', TELEGRAM)
 		self.assertEqual(len(chart['rows']), 8)
 		self.assertEqual(chart['rows'][-1][0], 'Other')
 		self.assertTrue(render.pie_chart(chart['rows'], chart['title'], chart['subtitle']).read().startswith(b'\x89PNG'))
@@ -434,7 +434,7 @@ def test_unknown_or_missing_portfolio_lists_known(self):
 
 	def test_a_service_that_takes_no_images_gets_text(self):
 		self.env(slackBotToken=False) # Slack replies carry charts only with the bot token
-		worker.Request('slack', 'C1', '@u', 'U', BOT).deliver(['x'], {'pie': True, 'rows': [('a', 1.0)], 'title': 't'})
+		worker.Request(SLACK, 'C1', '@u', 'U', BOT).deliver(['x'], {'pie': True, 'rows': [('a', 1.0)], 'title': 't'})
 		self.assertEqual(self.sent[-1][0], 'text')
 
 
diff --git a/tests/test_util.py b/tests/test_util.py
index 75671a68..643bc581 100644
--- a/tests/test_util.py
+++ b/tests/test_util.py
@@ -77,8 +77,8 @@ def test_get_emoji(self):
 		self.assertEqual([formats.get_emoji(n) for n in (1, -1, 0)], ['🔼', '🔻', '▪️'])
 
 	def test_links_per_service(self):
-		telegram_link = links.link('https://example.com/x', 'text', 'telegram')
-		slack_link = links.link('https://example.com/x', 'text', 'slack')
+		telegram_link = TELEGRAM.link('https://example.com/x', 'text')
+		slack_link = SLACK.link('https://example.com/x', 'text')
 		self.assertIn('<a href=', telegram_link)
 		self.assertIn('|text', slack_link)
 

From a3cd2aa5733d9a72fde70e267622fdd92ca56d21 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 00:24:41 +0000
Subject: [PATCH 16/81] A docstring for every function in the bot, lib/ and
 util/

The 128 functions and methods that had none get one, nested functions included; a comment that said the same on the def
line goes. A test now fails on any function in the bot, lib/ or util/ without a docstring, and CLAUDE.md says so.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md                  |  3 ++-
 bot.py                     |  6 ++++++
 lib/buttons.py             |  2 ++
 lib/charts.py              |  1 +
 lib/config.py              | 16 +++++++++++++++-
 lib/dialects.py            | 12 ++++++++++--
 lib/formats.py             |  6 ++++++
 lib/jobrun.py              |  5 ++++-
 lib/jobs.py                |  5 +++++
 lib/links.py               |  6 ++++++
 lib/markets.py             |  7 +++++++
 lib/net.py                 |  2 ++
 lib/render.py              |  6 +++++-
 lib/reports/about.py       |  2 ++
 lib/reports/cal.py         |  2 ++
 lib/reports/holdings.py    |  3 ++-
 lib/reports/milestone.py   |  4 ++++
 lib/reports/performance.py |  3 +++
 lib/reports/podcasts.py    |  1 +
 lib/reports/price.py       |  6 +++++-
 lib/reports/profile.py     |  2 ++
 lib/reports/rating.py      |  3 +++
 lib/reports/reminder.py    |  4 ++++
 lib/reports/shorts.py      |  5 +++++
 lib/reports/trades.py      |  3 +++
 lib/sharesight.py          | 13 +++++++++++++
 lib/shortman.py            |  2 ++
 lib/state.py               |  1 +
 lib/telegram.py            |  4 ++++
 lib/webhook.py             |  4 ++++
 lib/worker.py              | 30 +++++++++++++++++++++++++++++-
 lib/yahoo.py               | 12 +++++++++---
 tests/test_finbot.py       | 12 +++++++++++-
 util/sws_gen.py            |  7 +++++++
 34 files changed, 187 insertions(+), 13 deletions(-)

diff --git a/CLAUDE.md b/CLAUDE.md
index d6bb78b9..c30ff02e 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -56,7 +56,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - A report is a list of lines. `service` is a dialect (`lib/dialects.py`), which equals the network's name: markup with
   `service.bold(text)`, `.italic`, `.strike`, `.link(url, text)`, user-written text through `service.escape(text)`, stock links
   through `links.finance_link(symbol, exchange, service)`. Tests pass `TELEGRAM`, `SLACK`, `DISCORD`.
-- Tabs everywhere, tests included (`.editorconfig`). New and changed lines stay under 120 characters (an edit quotes whole lines).
+- Every function and method has a docstring (a test checks the bot, lib/ and util/). Tabs everywhere, tests included
+  (`.editorconfig`). New and changed lines stay under 120 characters (an edit quotes whole lines).
 - Log with `print(..., file=sys.stderr)`. systemd discards stdout (`StandardOutput=null`), so the bot and its reports never
   print there (a test checks); only the util/ tools, run by hand, do.
 - A chart renderer in `lib/render.py` gets `@_drawn` (the bot draws in one native thread); it must not call another `@_drawn`
diff --git a/bot.py b/bot.py
index 8d1f851e..a187320c 100755
--- a/bot.py
+++ b/bot.py
@@ -63,13 +63,19 @@ def main(environ, start_response):
 		return [b'']
 
 def handle(environ, start_response):
+	"""Route a request by its path, check that it comes from Telegram (its secret header) or Slack (its verification
+	token), turn away repeats, and hand a command or a button press to lib/worker in a thread of its own: the service
+	gets its answer at once."""
 	def print_body():
+		"""Log the request body, Slack's verification token redacted."""
 		shown = {k: '<redacted>' if k == 'token' else v for k, v in inbound.items()} # Slack's verification token
 		print(f"inbound {uri} ", json.dumps(shown, indent=4, default=str), file=sys.stderr)
 	def print_headers():
+		"""Log the request headers, Telegram's secret header redacted."""
 		for name, value in sorted(environ.items()):
 			print((name, '<redacted>' if name == 'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN' else value), file=sys.stderr)
 	def reply(status, body=b''):
+		"""Answer the request: a status line and a body."""
 		start_response(status, [('Content-type', 'application/json')])
 		return [body]
 	print_headers() if config.prefs.print_headers else None
diff --git a/lib/buttons.py b/lib/buttons.py
index 752a0081..685cc51c 100644
--- a/lib/buttons.py
+++ b/lib/buttons.py
@@ -27,6 +27,8 @@ def _ref(tickers):
 	return '#' + key
 
 def resolve(ref):
+	"""The tickers a button's callback data names: listed inline ('T1,T2'), or a stored list's '#id' (see _ref). Raises
+	KeyError when the stored list is gone."""
 	if ref.startswith('#'):
 		tickers = state.load(REF_FILE, {}).get(ref[1:])
 		if not tickers:
diff --git a/lib/charts.py b/lib/charts.py
index 5046ddd7..a3e0f5cd 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -22,6 +22,7 @@ def command_keyboard(per_row=3):
 KEYBOARD_FILE = 'finbot_keyboards.json' # chat id -> version of the command keyboard that chat last received
 
 def keyboard_version():
+	"""A short hash of the command keyboard, which changes whenever its buttons do."""
 	return hashlib.sha1(json.dumps(command_keyboard(), sort_keys=True).encode()).hexdigest()[:8]
 
 def mark_keyboard(chat_id):
diff --git a/lib/config.py b/lib/config.py
index 49070c90..1d3ba215 100644
--- a/lib/config.py
+++ b/lib/config.py
@@ -27,6 +27,7 @@
 class ConfigError(Exception):
 	"""One or more problems, each already naming the file and line. `errors` is the list."""
 	def __init__(self, errors):
+		"""errors: one message or a list of them."""
 		self.errors = [errors] if isinstance(errors, str) else list(errors)
 		super().__init__('\n'.join(self.errors))
 
@@ -113,17 +114,22 @@ class _Reader:
 	"""Typed reads from a getenv(name, default); a bad value is remembered, the default used, and all problems raised together.
 	A blank value means unset: the default."""
 	def __init__(self, getenv):
+		"""getenv(name, default) supplies the values; the problems found collect in `errors` until check()."""
 		self.getenv = getenv
 		self.errors = []
 
 	def value(self, name, default):
+		"""The raw value, or default when it is unset or blank."""
 		value = self.getenv(name, default)
 		return default if value == '' else value
 
 	def str(self, name, default=None):
+		"""A text setting."""
 		return self.value(name, default)
 
 	def bool(self, name, default):
+		"""A true/false setting: true/false, yes/no, on/off, 1/0 or t/f, in any case. Anything else is an error, and
+		gives default."""
 		value = self.value(name, default)
 		word = str(value).strip().lower()
 		if word in TRUE_WORDS or word in FALSE_WORDS:
@@ -132,6 +138,7 @@ def bool(self, name, default):
 		return default
 
 	def number(self, kind, name, default):
+		"""A setting of type `kind` (int or float). A value that is not one is an error, and gives default."""
 		value = self.value(name, default)
 		try:
 			return kind(value)
@@ -140,12 +147,15 @@ def number(self, kind, name, default):
 			return kind(default)
 
 	def int(self, name, default):
+		"""A whole-number setting."""
 		return self.number(int, name, default)
 
 	def float(self, name, default):
+		"""A number setting."""
 		return self.number(float, name, default)
 
 	def check(self):
+		"""Raise ConfigError with every problem found so far."""
 		if self.errors:
 			raise ConfigError(self.errors)
 
@@ -171,6 +181,7 @@ class Env:
 
 	@property
 	def preferences_path(self):
+		"""Where preferences.conf is read from: the data dir."""
 		return self.data_dir + '/' + PREFERENCES_FILE
 
 
@@ -286,6 +297,7 @@ def preference_values(getenv, source=None):
 def _getenv_for(preferences):
 	"""getenv(name, default): preferences.conf first (preference keys only), then the process environment."""
 	def getenv(name, default=None):
+		"""A preference from preferences.conf, else the process environment's value, else default."""
 		if name in preferences:
 			return preferences[name]
 		return os.environ.get(name, default)
@@ -298,6 +310,7 @@ def print_errors(e):
 		print('ERROR config:', line, file=sys.stderr)
 
 def _fail(e):
+	"""Log the configuration errors and exit: finbot does not start with a bad setting."""
 	print_errors(e)
 	sys.exit(1)
 
@@ -322,7 +335,8 @@ def _fail(e):
 os.environ.setdefault('MPLCONFIGDIR', os.path.abspath(env.cache_dir + '/matplotlib'))
 
 def data_file(name):
-	# user copy in the data dir wins over the shipped default in defaults/
+	"""Where a data file is read from: the user's copy in the data dir when there is one, else the default shipped in
+	defaults/."""
 	user = env.data_dir + '/' + name
 	return user if os.path.isfile(user) else str(REPO_DIR / 'defaults' / name)
 
diff --git a/lib/dialects.py b/lib/dialects.py
index bd9c18ed..ff4c05fe 100644
--- a/lib/dialects.py
+++ b/lib/dialects.py
@@ -15,15 +15,19 @@ class Dialect(str):
 	caption_limit = None # characters under a chart, as visible() counts them; None: no limit of its own
 
 	def bold(self, text):
+		"""text in bold."""
 		return self._styled('bold', text)
 
 	def italic(self, text):
+		"""text in italics."""
 		return self._styled('italic', text)
 
 	def strike(self, text):
+		"""text struck through."""
 		return self._styled('strike', text)
 
 	def _styled(self, style, text):
+		"""text between the markers of a style: bold, italic or strike."""
 		before, after = self.styles[style]
 		return before + text + after
 
@@ -49,10 +53,12 @@ class Telegram(Dialect):
 	caption_limit = 1024
 
 	def _link(self, url, text):
+		"""An HTML link."""
 		return '<a href="' + url + '">' + text + '</a>'
 
 	def visible(self, text):
-		return len(re.sub(r'<[^>]+>', '', text)) # the tags do not count
+		"""The length without the HTML tags, which Telegram does not count."""
+		return len(re.sub(r'<[^>]+>', '', text))
 
 
 class Slack(Dialect):
@@ -61,6 +67,7 @@ class Slack(Dialect):
 	styles = {'bold': ('*', '*'), 'italic': ('_', '_'), 'strike': ('~', '~')}
 
 	def _link(self, url, text):
+		"""A mrkdwn link: <url|text>."""
 		return '<' + url + '|' + text + '>'
 
 
@@ -76,7 +83,8 @@ def _link(self, url, text):
 		return '[' + text.replace('[', '\\[').replace(']', '\\]') + '](<' + url.replace('(', '%28').replace(')', '%29') + '>)'
 
 	def escape(self, text):
-		return html.unescape(str(text)) # markdown shows entities as they are: the plain text
+		"""The plain text: Discord's markdown shows entities as written."""
+		return html.unescape(str(text))
 
 
 TELEGRAM, SLACK, DISCORD = Telegram('telegram'), Slack('slack'), Discord('discord')
diff --git a/lib/formats.py b/lib/formats.py
index fa42a6ea..d5a6d6c5 100644
--- a/lib/formats.py
+++ b/lib/formats.py
@@ -63,6 +63,7 @@ def security_line(flag, title, link):
 	return f"{flag} {title} ({link})" if flag else f"{title} ({link})"
 
 def strip_url(url):
+	"""url without its scheme and www., to show as a link's text."""
 	url = url.removeprefix('https://www.')
 	url = url.removeprefix('http://www.')
 	url = url.removeprefix('https://')
@@ -91,6 +92,8 @@ def date_range_english(days, end):
 	return f"{date_short(start, start.year != end.year)} - {date_short(end, True)}"
 
 def days_english(days, prefix='the past ', article=''):
+	"""A span of days in words: 'today' for 0 or None; prefix + article + 'day', 'week', 'month' or 'year'; else prefix
+	and a count of weeks, months, years or days ('the past 3 weeks')."""
 	if days is None or days == 0:
 		return 'today'
 	elif days == 1:
@@ -117,6 +120,7 @@ def signed_percent(value, decimals=None):
 	return f"{value:+,}%" if decimals is None else f"{value:+,.{decimals}f}%"
 
 def get_emoji(number):
+	"""🔼 for a rise, 🔻 for a fall, ▪️ for no change."""
 	if number > 0:
 		return '🔼'
 	elif number < 0:
@@ -138,6 +142,7 @@ def days_from_human_days(arg):
 	return int(arg.removesuffix('Y')) * 365
 
 def ordinal(num):
+	"""The ordinal suffix of a number: 'st' for 1, 'nd' for 22, 'th' for 11 and 13."""
 	value = str(num)
 	if len(value) > 1:
 		secondToLastDigit = value[-2]
@@ -154,6 +159,7 @@ def ordinal(num):
 		return 'th'
 
 def td_to_human(timedelta):
+	"""A duration in words: '1 day, 2 hours, 5 seconds', zero parts left out ('0 seconds' for none)."""
 	days, remainder = divmod(timedelta.total_seconds(), 86400)
 	hours, remainder = divmod(remainder, 3600)
 	minutes, seconds = divmod(remainder, 60)
diff --git a/lib/jobrun.py b/lib/jobrun.py
index 1f6b7c7d..ff848c68 100644
--- a/lib/jobrun.py
+++ b/lib/jobrun.py
@@ -14,11 +14,14 @@
 
 
 def main(argv):
+	"""Run one scheduled command: argv is the report's module and its JSON keyword arguments. Returns the exit status: 0
+	when it ran, REPORTED when it failed and said so in the chat, 2 for bad arguments."""
 	if len(argv) != 2 or not re.fullmatch(r'[a-z]+(\.[a-z]+)*', argv[0]):
 		print('usage: python -m lib.jobrun <module> <json kwargs>', file=sys.stderr)
 		return 2
 	module = argv[0]
-	def run(): # the import too: a report that cannot even be imported is reported like any other failure
+	def run():
+		"""Import the report and run it: an import that fails is reported like any other failure."""
 		importlib.import_module(module).lambda_handler(**json.loads(argv[1]))
 		return True
 	return 0 if webhook.guarded(module.rsplit('.', 1)[-1] + '.py', run, exit=False) else REPORTED
diff --git a/lib/jobs.py b/lib/jobs.py
index 80cb7a3c..d1897661 100644
--- a/lib/jobs.py
+++ b/lib/jobs.py
@@ -88,6 +88,7 @@ class ImpossibleTime(ScheduleError):
 	"""A time that cannot be (minute 61) or never comes (30 February). Other bad lines are reported and skipped, but this refuses the
 	whole file: finbot does not start with it, and a reload keeps the schedule that is running. `lines` says where, one per line."""
 	def __init__(self, lines):
+		"""lines: where, one message or a list of them."""
 		self.lines = [lines] if isinstance(lines, str) else list(lines)
 		super().__init__('; '.join(self.lines))
 
@@ -119,6 +120,7 @@ def tokenize(text):
 	"""A command line as ('word', text) and ('op', text) tokens. Quoted text is always a word, whatever it contains."""
 	tokens, word, started, i = [], '', False, 0
 	def flush():
+		"""End the word being read, if there is one, and add it to the tokens."""
 		nonlocal word, started
 		if started:
 			tokens.append(('word', word))
@@ -240,6 +242,9 @@ class Scheduler:
 	new copy is found on reload). `clock`, `sleep` and `spawn` can be replaced, which the tests do."""
 
 	def __init__(self, path, catchup=None, clock=None, sleep=None, spawn=None, zone=None, runner=None):
+		"""path() gives the schedule file's path. catchup is how late, in minutes, a missed run may still start (the
+		scheduler_catchup setting when None); clock, sleep, spawn, zone and runner stand in for the real ones in
+		tests."""
 		self.path = path
 		self.catchup = datetime.timedelta(minutes=config.env.scheduler_catchup if catchup is None else catchup)
 		self.clock = clock or (lambda: datetime.datetime.now(datetime.timezone.utc))
diff --git a/lib/links.py b/lib/links.py
index 27159b8b..8bcd091a 100644
--- a/lib/links.py
+++ b/lib/links.py
@@ -3,11 +3,15 @@
 from lib import markets
 
 def yahoo_link(ticker, service, brief=True, text=None):
+	"""A link to the ticker's Yahoo Finance page, reading `text`, else the symbol without its market (brief) or the
+	whole ticker."""
 	if not text:
 		text = ticker.split('.')[0] if brief else ticker
 	return service.link("https://au.finance.yahoo.com/quote/" + ticker, text)
 
 def finance_link(symbol, exchange, service, days=1, brief=True, text=None):
+	"""A link to the stock's page on the site the hyperlinkProvider preference names: Google Finance (google) or
+	Yahoo."""
 	if config.prefs.hyperlinkProvider == 'google':
 		return gfinance_link(symbol, exchange, service, days, brief, text)
 	return yahoo_link(symbol, service, brief, text)
@@ -19,6 +23,8 @@ def quote_link(ticker, exchange, service, days=1):
 	return finance_link(ticker, exchange, service, days=days)
 
 def gfinance_link(symbol, exchange, service, days=1, brief=True, text=None):
+	"""A link to the stock's Google Finance page, its chart window the one nearest to `days`, reading `text`, else the
+	symbol (brief) or SYMBOL:EXCHANGE (a US listing: the symbol). The plain symbol with the hyperlink preference off."""
 	window = '1D'
 	if not days:
 		days = 1
diff --git a/lib/markets.py b/lib/markets.py
index 7b5855ff..7ec4d557 100644
--- a/lib/markets.py
+++ b/lib/markets.py
@@ -71,6 +71,7 @@
 }
 
 def flag_from_market(market):
+	"""The flag of a market by its Sharesight code: US for the NASDAQ and NYSE variants, '' when unknown."""
 	if market in MARKETS:
 		return MARKETS[market][0]
 	return '🇺🇸' if market.lower().startswith(('nasdaq', 'nyse')) else ''
@@ -79,9 +80,11 @@ def flag_from_market(market):
 	'TW': '🇹🇼', 'TWO': '🇹🇼', 'TO': '🇨🇦'} # Yahoo suffix -> flag
 
 def flag_from_ticker(ticker):
+	"""The flag of a Yahoo ticker's market suffix, '' when unknown; a ticker without one is a US listing."""
 	return TICKER_FLAGS.get(ticker.split('.')[1], '') if '.' in ticker else '🇺🇸'
 
 def currency_from_market(market):
+	"""The currency of a market by its Sharesight code; None where it quotes in several, or is unknown."""
 	return MARKETS.get(market, (None, None))[1]
 
 VENUES = { # the names Yahoo, Sharesight and Google give an exchange -> (Google Finance code, country)
@@ -98,6 +101,7 @@ def currency_from_market(market):
 }
 
 def transform_to_google(exchange):
+	"""Google Finance's code for an exchange, as Yahoo, Sharesight or Google name it; the name itself when unknown."""
 	if 'Nasdaq' in exchange:
 		return 'NASDAQ'
 	if exchange == 'Cboe US':
@@ -105,6 +109,7 @@ def transform_to_google(exchange):
 	return VENUES.get(exchange, (None, None))[0] or exchange
 
 def exchange_human(exchange):
+	"""The country an exchange is in ('US', 'AU' and so on), or its name when unknown."""
 	if 'Nasdaq' in exchange or 'NYSE' in exchange or 'Other OTC' in exchange:
 		return 'US'
 	return VENUES.get(exchange, (None, None))[1] or exchange
@@ -118,6 +123,8 @@ def exchange_human(exchange):
 }
 
 def transform_to_yahoo(ticker, market=None):
+	"""Yahoo's symbol for a ticker: 'BHP:ASX', 'BHP.ASX' or ('BHP', 'ASX') give 'BHP.AX'. US venues take no suffix, and
+	class shares a dash ('BRK.B' gives 'BRK-B')."""
 	split = ticker.replace(':', '.').split('.')
 	if len(split) > 1:
 		ticker = split[0]
diff --git a/lib/net.py b/lib/net.py
index d29a1cc6..2ae24541 100644
--- a/lib/net.py
+++ b/lib/net.py
@@ -11,9 +11,11 @@
 session.cookies.set_policy(http.cookiejar.DefaultCookiePolicy(allowed_domains=[])) # no domain may set or receive a cookie
 
 def get(url, **kwargs):
+	"""requests.get through the shared session, which reuses connections."""
 	return session.get(url, **kwargs)
 
 def post(url, **kwargs):
+	"""requests.post through the shared session, which reuses connections."""
 	return session.post(url, **kwargs)
 
 def redact(text):
diff --git a/lib/render.py b/lib/render.py
index ce03aaf7..ee9be8e1 100644
--- a/lib/render.py
+++ b/lib/render.py
@@ -44,6 +44,7 @@ def _drawn(draw):
 	pool's one thread waiting on itself would never finish."""
 	@functools.wraps(draw)
 	def in_pool(*args, **kwargs):
+		"""Draw in the render thread when the bot has one (render_off_hub), else here."""
 		if _render_pool is None:
 			return draw(*args, **kwargs)
 		return _render_pool.apply(draw, args, kwargs)
@@ -113,6 +114,7 @@ def graph(df, title, ylabel):
 	_style_time_axes(fig, ax, x.iloc[0], x.iloc[-1])
 
 	def annotate(i, name, above):
+		"""Label the point at index i, the high or the low, with its value and date, above or below it."""
 		xd, yv = x.iloc[i], y.iloc[i]
 		frac = (mdates.date2num(xd) - mdates.date2num(x.iloc[0])) / max(mdates.date2num(x.iloc[-1]) - mdates.date2num(x.iloc[0]), 1)
 		# anchor text toward the inside of the plot so it cannot run off either edge
@@ -305,6 +307,7 @@ def shorts_chart(rows, threshold, step=5):
 	def shorts(x, y, w, h, color, frac=1.0, lw=1.6):
 		"""One pair, top-left at (x, y); frac < 1 crops it from the right."""
 		def P(px, py):
+			"""A point of the pair's box, from fractions of its width and height."""
 			return (x + px * w, y + py * h)
 		belt = '#' + ''.join(f"{int(int(color[i:i + 2], 16) * 0.7):02x}" for i in (1, 3, 5))
 		legs = Polygon([P(0.04, 0.2), P(0.96, 0.2), P(1.0, 0.97), P(0.56, 0.97), P(0.5, 0.58), P(0.44, 0.97), P(0.0, 0.97)],
@@ -329,7 +332,8 @@ def P(px, py):
 		a0, a1 = k * 2 * math.pi / rays + tilt, (k + 1) * 2 * math.pi / rays + tilt
 		ax.add_patch(Polygon([(sx, sy), (sx + 2000 * math.cos(a0), sy + 2000 * math.sin(a0)), (sx + 2000 * math.cos(a1), sy + 2000 * math.sin(a1))],
 			closed=True, facecolor=ray, edgecolor='none', zorder=-1))
-	def at(u, v): # face coordinates in sun radii, y down
+	def at(u, v):
+		"""A point on the sun's face, in sun radii from its centre with y down, in the axes' pixels."""
 		return (sx + u * R, sy + v * R)
 	ax.add_patch(Circle((sx, sy), R, facecolor='#ffb000', edgecolor='none'))
 	for dx, dy, r, c in ((-0.05, -0.06, 0.93, '#ffb90a'), (-0.1, -0.12, 0.78, '#ffc21f'), (-0.16, -0.2, 0.55, '#ffcb35')): # fake gradient
diff --git a/lib/reports/about.py b/lib/reports/about.py
index faf0acab..a6a0fe80 100644
--- a/lib/reports/about.py
+++ b/lib/reports/about.py
@@ -35,10 +35,12 @@ def run_job_line(scheduler, text):
 		return str(e)
 
 def job_line(scheduler, label, outcome):
+	"""'3. trades.py: done': a command's number in the .jobs list, its name, and how its run went."""
 	number = [command[0] for command in scheduler.commands()].index(label) + 1
 	return f"{number}. {jobs.display(label)}: {outcome}"
 
 def prepare_help(service, botName):
+	"""The .help text: every command, by section. botName is the bot's @name, which can start a command too."""
 	payload = []
 	payload.append(service.bold("Tracked securities:"))
 	payload.append(".holdings")
diff --git a/lib/reports/cal.py b/lib/reports/cal.py
index 53069955..65d384ea 100644
--- a/lib/reports/cal.py
+++ b/lib/reports/cal.py
@@ -24,6 +24,8 @@ def lambda_handler(chat_id=None, days=None, service=None, specific_stock=None, i
 		days = config.prefs.future_days
 	field, heading = EVENTS[event]
 	def prepare_payload(service, market_data):
+		"""The report in one network's markup: the events within `days`, grouped by date. An ADR or a second listing
+		without a date of its own takes its primary listing's."""
 		now = datetime.datetime.now()
 		soon = now + datetime.timedelta(days=days)
 		found = [] # (when, flag, title, link)
diff --git a/lib/reports/holdings.py b/lib/reports/holdings.py
index 42b5bb9b..131730df 100644
--- a/lib/reports/holdings.py
+++ b/lib/reports/holdings.py
@@ -81,7 +81,8 @@ def prepare_watchlist(service, user, action=None, ticker=None):
 
 def watchlist_change(service, user, action, asked):
 	"""The reply to adding or removing `asked` (a Yahoo ticker)."""
-	def described(ticker, quote): # "Full Name (TICKER)", bold
+	def described(ticker, quote):
+		"""'Full Name (TICKER)' in bold, linked to the stock's page."""
 		quote = quote or {}
 		link = links.finance_link(ticker, quote.get('profile_exchange') or ticker, service)
 		return service.bold(f"{quote['profile_title']} ({link})" if quote.get('profile_title') else link)
diff --git a/lib/reports/milestone.py b/lib/reports/milestone.py
index b6606037..6b8dce09 100644
--- a/lib/reports/milestone.py
+++ b/lib/reports/milestone.py
@@ -12,7 +12,10 @@
 # https://finance.yahoo.com/quote/TSLA/history?period1=0&period2=1707609600&filter=split
 
 def lambda_handler():
+	"""Post the stocks followed that set an all-time high or low (their 52-week range against the record, which a first
+	sighting takes from the stock's history), and record the new records."""
 	def prepare_milestone_payload(service, market_data):
+		"""(the report in one network's markup, the records it changes)."""
 		payload = []
 		payloadhigh = []
 		payloadlow = []
@@ -68,4 +71,5 @@ def prepare_milestone_payload(service, market_data):
 			recorded.update(changed)
 
 def job(argv):
+	"""Arguments of a scheduled run: milestone.py takes none."""
 	return 'milestone.py', {}
diff --git a/lib/reports/performance.py b/lib/reports/performance.py
index cceaaa5f..c72dc3d6 100644
--- a/lib/reports/performance.py
+++ b/lib/reports/performance.py
@@ -63,6 +63,9 @@ def report(service, past_days=None, portfolio_select=None, interactive=False, pe
 	return payload, graph
 
 def lambda_handler(chat_id=None, past_days=None, service=None, user='', portfolio_select=None, interactive=False):
+	"""Post each portfolio's performance over past_days (the past_days preference when None), or the one named, beside
+	the benchmarks: to the chat that asked (interactive), else to the default chats, with its chart and on Telegram
+	period buttons."""
 	if past_days is None:
 		past_days = config.prefs.past_days
 	performance = gather(past_days, portfolio_select)
diff --git a/lib/reports/podcasts.py b/lib/reports/podcasts.py
index 1bd9e232..3e9e771f 100644
--- a/lib/reports/podcasts.py
+++ b/lib/reports/podcasts.py
@@ -19,6 +19,7 @@ def get_seen_filename(url):
 	return f"finbot_podcast_seen_{hashlib.md5(url.encode()).hexdigest()}.json"
 
 def fetch_and_parse_feed(url):
+	"""The feed at url, parsed as XML. Raises on an HTTP error."""
 	resp = net.get(url, timeout=30)
 	resp.raise_for_status()
 	return ET.fromstring(resp.content)
diff --git a/lib/reports/price.py b/lib/reports/price.py
index 72c53822..ac44ff71 100644
--- a/lib/reports/price.py
+++ b/lib/reports/price.py
@@ -165,10 +165,14 @@ def report(service, threshold=None, user='', specific_stock=None, interactive=Fa
 
 def lambda_handler(chat_id=None, threshold=None, service=None, user='', specific_stock=None, interactive=False, midsession=False,
 		premarket=False, interday=False, days=None, close=False, top=None):
+	"""Post the price report (report() has the modes): to the chat that asked (interactive), else to the default chats,
+	with its chart and on Telegram period buttons."""
 	if threshold is None:
 		threshold = config.prefs.price_percent
 	market = gather(specific_stock, days)
-	def markup(service): # a chart's period buttons: one ticker's chart, and the daily and N-day lists (not the intraday ones)
+	def markup(service):
+		"""A chart's period buttons on Telegram: one ticker's chart, and the daily and N-day lists, but not the intraday
+		ones."""
 		if service != 'telegram':
 			return None
 		if specific_stock:
diff --git a/lib/reports/profile.py b/lib/reports/profile.py
index 38bd2eb1..49fdeb77 100644
--- a/lib/reports/profile.py
+++ b/lib/reports/profile.py
@@ -11,6 +11,8 @@
 from itertools import groupby
 
 def doDelta(inputList):
+	"""Trend marks between consecutive figures (four quarters give three): 🔼 up and 🔽 down while the figure is positive,
+	🔺 up and 🔻 down while it is negative, ▪️ unchanged, ❌ missing; padded with ❌ to three."""
 	deltaString = ''
 	inputListFixed = inputList.copy()
 	# replace any NoneType to allow delta calculation
diff --git a/lib/reports/rating.py b/lib/reports/rating.py
index 2b7899cc..7b6c378c 100644
--- a/lib/reports/rating.py
+++ b/lib/reports/rating.py
@@ -9,7 +9,9 @@
 def lambda_handler(chat_id=None, specific_stock=None, service=None, interactive=False):
 	"""Analysts' consensus ratings that changed since the last scheduled run, which records the ratings it saw."""
 	def prepare_rating_payload(service, market_data):
+		"""(the report in one network's markup, the ratings seen this run)."""
 		def get_emoji(old, new):
+			"""📈 when the consensus improved (its index fell: 1 is a strong buy), 📉 when it worsened, ▪️ otherwise."""
 			if old > new:
 				return '📈'
 			if old < new:
@@ -67,4 +69,5 @@ def get_emoji(old, new):
 			ratings.update(new)
 
 def job(argv):
+	"""Arguments of a scheduled run: rating.py takes none."""
 	return 'rating.py', {}
diff --git a/lib/reports/reminder.py b/lib/reports/reminder.py
index a9e161aa..d33ad2fe 100644
--- a/lib/reports/reminder.py
+++ b/lib/reports/reminder.py
@@ -8,7 +8,10 @@
 
 
 def lambda_handler():
+	"""Post today's reminders to the default chats, if there are any."""
 	def prepare_finance_calendar_payload(service):
+		"""Today's reminders in one network's markup: those in reminder.json for this day (and year, and country,
+		when given), the Australian end of financial year on 23 June, and finbot's birthday."""
 		payload = []
 		tz = zoneinfo.ZoneInfo(config.prefs.timezone)
 		localtime = datetime.datetime.now(tz)
@@ -70,4 +73,5 @@ def prepare_finance_calendar_payload(service):
 		webhook.send(service, url, chat, prepare_finance_calendar_payload(service))
 
 def job(argv):
+	"""Arguments of a scheduled run: reminder.py takes none."""
 	return 'reminder.py', {}
diff --git a/lib/reports/shorts.py b/lib/reports/shorts.py
index e27401c2..2b41ea65 100644
--- a/lib/reports/shorts.py
+++ b/lib/reports/shorts.py
@@ -8,9 +8,13 @@
 from lib import shortman
 
 def lambda_handler(chat_id=None, threshold=None, specific_stock=None, service=None, interactive=False):
+	"""Post the stocks followed whose short interest is over `threshold` percent (the shorts_percent preference when
+	None), or one stock's: to the chat that asked (interactive), else to the default chats. Short Man has the ASX's
+	figures, Yahoo the US's."""
 	if threshold is None:
 		threshold = config.prefs.shorts_percent
 	def prepare_shorts_payload(service, market_data):
+		"""(the report in one network's markup, its chart or None), the most shorted first."""
 		found = [] # (percent, line, chart label)
 		for ticker in tickers:
 			try:
@@ -48,4 +52,5 @@ def prepare_shorts_payload(service, market_data):
 		webhook.send(service, url, chat, payload, graph)
 
 def job(argv):
+	"""Arguments of a scheduled run: shorts.py takes none."""
 	return 'shorts.py', {}
diff --git a/lib/reports/trades.py b/lib/reports/trades.py
index e6333851..5e4d1450 100644
--- a/lib/reports/trades.py
+++ b/lib/reports/trades.py
@@ -20,6 +20,8 @@ def lambda_handler(chat_id=None, days=None, service=None, user='', portfolio_sel
 	state_file = "finbot_sharesight_trades.json"
 
 	def prepare_trade_payload(service, trades):
+		"""The report in one network's markup: the buys and sells (on a scheduled run only new ones), grouped by date,
+		each linked to Sharesight and to the stock's page."""
 		payload_staging = []
 		dates = set()
 		wanted = [] # the trades to report: buys and sells that are new (or all of them, when interactive)
@@ -103,4 +105,5 @@ def prepare_trade_payload(service, trades):
 			known.extend(sorted(newtrades.difference(known)))
 
 def job(argv):
+	"""Arguments of a scheduled run: trades.py takes none."""
 	return 'trades.py', {}
diff --git a/lib/sharesight.py b/lib/sharesight.py
index 347841f2..c544a86c 100644
--- a/lib/sharesight.py
+++ b/lib/sharesight.py
@@ -14,10 +14,12 @@
 _request_slots = threading.BoundedSemaphore(MAX_CONCURRENT)
 
 def _get(url, **kwargs):
+	"""net.get, waiting its turn: one Sharesight request at a time (MAX_CONCURRENT)."""
 	with _request_slots:
 		return net.get(url, **kwargs)
 
 def _post(url, **kwargs):
+	"""net.post, waiting its turn like _get."""
 	with _request_slots:
 		return net.post(url, **kwargs)
 
@@ -40,12 +42,16 @@ def _check(data, what):
 
 class BearerAuth(requests.auth.AuthBase):
 	def __init__(self, token):
+		"""token: the access token."""
 		self.token = token
 	def __call__(self, r):
+		"""Add the token to a request's Authorization header."""
 		r.headers["Authorization"] = "Bearer " + self.token
 		return r
 
 def get_token():
+	"""An access token for the Sharesight API: the cached one while it has more than 20 seconds left, else a new one
+	(cached while the cache preference is on)."""
 	if config.prefs.cache:
 		cacheFile = "finbot_sharesight_token.json"
 		cached = cache.load(cacheFile)
@@ -68,6 +74,7 @@ def get_token():
 PORTFOLIOS_CACHE = "finbot_sharesight_portfolios.json" # ttl: the cache_seconds preference
 
 def performance_cache_file(portfolio_id, days):
+	"""The cache file of a portfolio's performance report over `days`."""
 	return "finbot_sharesight_performance_" + str(portfolio_id) + "_" + str(days) + '.json'
 
 def get_portfolios(refresh=False):
@@ -90,6 +97,7 @@ def get_portfolios(refresh=False):
 class PortfolioNotFound(LookupError):
 	"""No portfolio is called what was asked for. `lines` is the answer to give: the name, and the portfolios there are."""
 	def __init__(self, name, names):
+		"""name: what was asked for; names: the portfolios there are."""
 		self.lines = [f"{name} portfolio not found. I only know about:"] + list(names)
 		super().__init__(' '.join(self.lines))
 
@@ -136,6 +144,8 @@ def tickers():
 	return sorted({ticker for portfolio in portfolios() for ticker in portfolio.tickers()})
 
 def get_trades(portfolio_name, portfolio_id, days=None):
+	"""The portfolio's trades of the past `days` (the past_days preference when None), each tagged with its portfolio's
+	name; cached for 5 minutes. Raises RuntimeError when Sharesight cannot be reached or answers with an error."""
 	if days is None:
 		days = config.prefs.past_days
 	if config.prefs.cache:
@@ -157,6 +167,9 @@ def get_trades(portfolio_name, portfolio_id, days=None):
 	return data['trades']
 
 def get_performance(portfolio_id, days, seconds=None, refresh=False):
+	"""Sharesight's performance report of a portfolio since `days` ago (0: today's holdings), cached for `seconds` (the
+	performance_cache_seconds preference when None); refresh=True asks again. While Sharesight says too many requests
+	are in flight it retries, up to five tries. Raises RuntimeError on an error answer, or one without a report."""
 	if seconds is None:
 		seconds = config.prefs.performance_cache_seconds
 	start_date = datetime.datetime.now() - datetime.timedelta(days=days)
diff --git a/lib/shortman.py b/lib/shortman.py
index 023a87a0..aa878e9a 100644
--- a/lib/shortman.py
+++ b/lib/shortman.py
@@ -3,6 +3,8 @@
 from lib import net
 
 def fetch(market_data):
+	"""market_data with each ASX ticker's short interest added from Short Man's daily CSV; as it was when Short Man
+	cannot be reached."""
 	local_market_data = market_data.copy()
 	url = 'https://www.shortman.com.au/downloadeddata/latest.csv'
 	try:
diff --git a/lib/state.py b/lib/state.py
index a709dec1..f299349e 100644
--- a/lib/state.py
+++ b/lib/state.py
@@ -30,6 +30,7 @@ def save(name, data):
 
 
 def _lock(name):
+	"""The lock edit() holds while it changes the file `name`."""
 	with _locks_lock:
 		return _locks.setdefault(name, threading.Lock())
 
diff --git a/lib/telegram.py b/lib/telegram.py
index 908396ca..427564ae 100644
--- a/lib/telegram.py
+++ b/lib/telegram.py
@@ -22,15 +22,19 @@ def try_call(method, **kwargs):
 		print(f"Telegram {method} failed:", net.redact(e), file=sys.stderr)
 
 def setWebhook():
+	"""Have Telegram post updates (messages and button presses) to telegramOutgoingWebhook, signed with
+	telegramOutgoingToken."""
 	print("registering", config.env.telegramOutgoingWebhook, file=sys.stderr)
 	params = {'url': config.env.telegramOutgoingWebhook, "allowed_updates": json.dumps(['message', 'callback_query']), 'secret_token': config.env.telegramOutgoingToken}
 	return call('setWebhook', params=params)
 
 def delWebhook():
+	"""Have Telegram stop posting updates."""
 	print("deregistering", config.env.telegramOutgoingWebhook, file=sys.stderr)
 	return call('setWebhook', params={'url': ''}) # unsubscribe
 
 def getMe():
+	"""The bot's own account: its username is the bot's name in chats."""
 	return call('getMe')
 
 def register():
diff --git a/lib/webhook.py b/lib/webhook.py
index 9391b353..50b1589b 100644
--- a/lib/webhook.py
+++ b/lib/webhook.py
@@ -20,6 +20,9 @@ def discord_url(url):
 	return url + ('&' if '?' in url else '?') + 'wait=true'
 
 def write(service, url, payload_string, chat_id=None, reply_markup=None):
+	"""POST one message to a service and return its answer, or None when it failed (logged). Telegram: HTML, no link
+	previews, silent, reply_markup as its buttons. Slack: into channel chat_id through chat.postMessage with the bot
+	token, or through the incoming webhook. Discord: the native webhook, without previews or mentions."""
 	headers = {'Content-type': 'application/json'}
 	payload = {'text': payload_string}
 	if service == 'discord': # the native webhook endpoint: Discord's Slack-compatible one takes no flags
@@ -171,6 +174,7 @@ def fit_caption(caption, limit=1024, visible=None):
 	(visible=len). Keep whole lines that fit and say how many were left out."""
 	if visible is None:
 		def visible(text):
+			"""The length without the HTML tags."""
 			return len(re.sub(r'<[^>]+>', '', text))
 	if visible(caption) <= limit:
 		return caption
diff --git a/lib/worker.py b/lib/worker.py
index 7b7631fc..4f183c46 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -25,6 +25,7 @@ class TypingIndicator:
 	"""Telegram's "typing…" while a with block runs, sent again every 7 seconds, at most max_loops times. It stops however the block
 	ends, so a command that fails stops it before its error line goes out. Only Telegram has one: elsewhere it does nothing."""
 	def __init__(self, service, chat_id, max_loops=3):
+		"""service, chat_id: where; max_loops: how many notices at most, one every 7 seconds."""
 		self.service = service
 		self.chat_id = chat_id
 		self.max_loops = max_loops
@@ -32,6 +33,7 @@ def __init__(self, service, chat_id, max_loops=3):
 		self._thread = None
 
 	def _worker(self):
+		"""Send the notice every 7 seconds until stopped, or until max_loops are sent."""
 		loop_count = 0
 		while not self._stop_event.is_set() and loop_count < self.max_loops:
 			webhook.pleaseHold('typing', self.service, self.chat_id)
@@ -41,25 +43,30 @@ def _worker(self):
 				break
 
 	def start(self):
+		"""Start the notices, on Telegram only."""
 		if self.service != 'telegram':
 			return
 		self._thread = threading.Thread(target=self._worker, daemon=True)
 		self._thread.start()
 
 	def stop(self):
+		"""Stop the notices and wait for their thread to end."""
 		if not self._thread: # never started: only Telegram has a typing indicator
 			return
 		self._stop_event.set()
 		self._thread.join()
 
 	def is_active(self):
+		"""Whether the notices are still going."""
 		return self._thread is not None and self._thread.is_alive()
 
 	def __enter__(self):
+		"""Start the notices as the with block starts."""
 		self.start()
 		return self
 
 	def __exit__(self, *exc):
+		"""Stop them however the block ends."""
 		self.stop()
 
 @dataclasses.dataclass
@@ -73,12 +80,15 @@ class Request:
 
 	@property
 	def url(self):
+		"""The URL that posts into this chat."""
 		return webhook.chat_url(self.service, self.chat_id)
 
 	def typing(self, max_loops=3):
+		"""A typing indicator for this chat: `with request.typing():` around slow work."""
 		return TypingIndicator(self.service, self.chat_id, max_loops)
 
 	def say(self, payload, reply_markup=None):
+		"""Send lines of text to this chat, in several messages when long; reply_markup goes on the last."""
 		webhook.payload_wrapper(self.service, self.url, payload, self.chat_id, reply_markup=reply_markup)
 
 	def send(self, payload, image=None, reply_markup=None):
@@ -196,6 +206,7 @@ def period_and_portfolio(text, days):
 	return days, ' '.join(words) or None
 
 def on_watchlist(request, m):
+	""".watchlist [add|del SYMBOL]: the watchlist, or a change to it."""
 	action = ticker = None
 	if m.group('action') and m.group('ticker'):
 		action, ticker = m.group('action').lower(), m.group('ticker').upper()
@@ -209,6 +220,7 @@ def on_watchlist(request, m):
 	request.say(payload)
 
 def on_help(request, m):
+	""".help: every command; in a Telegram DM the command keyboard too."""
 	markup = None
 	if request.service == 'telegram' and int(request.chat_id) > 0: # positive chat ids are DMs; groups are negative
 		markup = charts.command_keyboard()
@@ -241,6 +253,7 @@ def on_calendar(request, m):
 		cal.lambda_handler(request.chat_id, days, request.service, ticker, interactive=True, event=event)
 
 def on_performance(request, m):
+	""".performance [period] [portfolio]: each portfolio's return, over a month by default."""
 	days, portfolio = period_and_portfolio(m.group('args'), buttons.PERIODS['m'][1]) # default view is the 1M button
 	if days > 0:
 		with request.typing():
@@ -273,21 +286,26 @@ def on_price(request, m):
 		price.lambda_handler(request.chat_id, threshold, request.service, request.user, ticker, interactive=True, interday=days is None, days=days, top=top)
 
 def on_shorts(request, m):
+	""".shorts [percent|SYMBOL]: the most shorted stocks followed, or one stock's short interest."""
 	threshold, ticker = percent_or_ticker(m.group(1), config.prefs.shorts_percent)
 	with request.typing():
 		shorts.lambda_handler(request.chat_id, threshold, ticker, request.service, interactive=True)
 
 def on_trades(request, m):
+	""".trades [period] [portfolio]: the buys and sells, of the past day by default."""
 	days, portfolio = period_and_portfolio(m.group('args'), 1)
 	with request.typing():
 		trades.lambda_handler(request.chat_id, days, request.service, request.user, portfolio, interactive=True)
 
 def on_holdings(request, m):
+	""".holdings PORTFOLIO: what it holds."""
 	with request.typing():
 		payload = holdings.prepare_holdings_payload((m.group(1) or '').strip() or None, request.service)
 	request.say(payload)
 
 def on_marketcap(request, m):
+	""".marketcap [top|bottom|SYMBOL]: the stocks followed by market cap, largest (or smallest) first, or one
+	stock's."""
 	with request.typing():
 		payload = rankings.prepare_marketcap_payload(request.service, m.group(1) or 'top', length=15)
 	request.say(payload)
@@ -296,6 +314,7 @@ def on_ratio(kind, negative=False, chart=False):
 	"""The function that answers .pe, .fpe or .peg: the top (the default) or bottom 15 by the ratio, the negative ones (where
 	there can be any), or one ticker's ratio. `chart`: with its chart."""
 	def handler(request, m):
+		""".pe, .fpe or .peg [top|bottom|negative|SYMBOL]."""
 		arg = (m.group(1) or '').strip()
 		action, ticker = kind, None
 		if arg == 'bottom':
@@ -310,11 +329,13 @@ def handler(request, m):
 	return handler
 
 def on_beta(request, m):
+	""".beta: the volatile small caps among the stocks followed."""
 	with request.typing():
 		payload, chart = rankings.prepare_beta_payload(request.service)
 	request.deliver(payload, chart)
 
 def on_pie(request, m):
+	""".pie PORTFOLIO [sector]: its value by industry, or by sector."""
 	words = m.group('args').split()
 	by = 'industry'
 	if words and words[-1].lower() in ('sector', 'sectors', 'industry', 'industries'):
@@ -331,6 +352,7 @@ def on_rating(request, m):
 		request.deliver(payload, chart)
 
 def on_compare(request, m):
+	""".compare SYMBOL SYMBOL... [period]: their price changes on one chart, with period buttons on Telegram."""
 	with request.typing():
 		payload, image, tickers, days = history.prepare_compare(request.service, (m.group('args') or '').split())
 	markup = None
@@ -339,6 +361,7 @@ def on_compare(request, m):
 	request.send(payload, image, markup)
 
 def on_history(request, m):
+	""".history SYMBOL: the stock's change over every period, with its price chart and, on Telegram, period buttons."""
 	if not m.group('ticker') or m.group('extra'):
 		request.say(["Usage: .history SYMBOL"])
 		return
@@ -348,17 +371,20 @@ def on_history(request, m):
 	request.send(caption.split('\n'), image, buttons.keyboard('h', [ticker], 'x') if image and request.service == 'telegram' else None)
 
 def on_plan(request, m):
+	""".plan [text]: record the sender's plan, and list everyone's."""
 	if m.group(1) and not m.group(1).startswith('@'):
 		with state.edit('finbot_plan.json', {}) as plans:
 			plans[request.user] = m.group(1)
 	plan = state.load('finbot_plan.json', {})
-	def who(user): # a Slack user is a mention (<@U123>), which is markup to keep; elsewhere a handle is text
+	def who(user):
+		"""A sender as shown: on Slack a mention (<@U123>), which is markup to keep; elsewhere the handle as text."""
 		user = user.removeprefix('@')
 		return user if request.service == 'slack' else request.service.escape(user)
 	request.say([f"{request.service.bold(who(k))}: {request.service.italic(request.service.escape(v))}\n"
 		for k, v in plan.items()])
 
 def on_who(request, m):
+	""".who SYMBOL: the portfolios that hold it."""
 	if not m.group('ticker'):
 		request.say([".who: please try again specifying a ticker"])
 		return
@@ -367,11 +393,13 @@ def on_who(request, m):
 	request.say(payload)
 
 def on_jobs(request, m):
+	""".jobs [run NUMBER|COMMAND]: the schedule, or one of its commands run now."""
 	with request.typing(max_loops=10):
 		payload, markup = about.prepare_jobs_payload(request.service, m.group('args'))
 	request.say(payload, reply_markup=markup)
 
 def on_profile(request, m):
+	""".SYMBOL: the company's profile."""
 	with request.typing():
 		payload = profile.prepare_profile_payload(request.service, request.user, m.group('ticker').upper())
 	request.say(payload)
diff --git a/lib/yahoo.py b/lib/yahoo.py
index aa064c2e..cfef0c2e 100644
--- a/lib/yahoo.py
+++ b/lib/yahoo.py
@@ -17,8 +17,10 @@
 _auth = threading.RLock() # Yahoo's cookie and crumb are renewed by one thread at a time: two at once could cache a crumb for the cookie the other replaced
 HEADERS = {'User-Agent': 'Mozilla/5.0'}
 
-def getCookie(maxAge=1209600): # 14 days
-	# we must always cache this, as crumb must match cookie
+def getCookie(maxAge=1209600):
+	"""Yahoo's session cookie as 'name=value', cached for maxAge seconds (14 days, renewed 30 seconds early) whatever
+	the cache preference, since the crumb must match it: a new cookie drops the crumb. A fallback cookie when Yahoo
+	sends none."""
 	with _auth:
 		cacheFile = "finbot_yahoo_cookie.json"
 		cached = cache.read(cacheFile, maxAge - 30) # renewed 30 seconds before it expires
@@ -43,7 +45,9 @@ def getCookie(maxAge=1209600): # 14 days
 		rmCrumb() # we must invalidate the non-matching crumb
 		return cookielist[0][0] + '=' + cookielist[0][1]
 
-def getCrumb(seconds=1209600): # 14 days
+def getCrumb(seconds=1209600):
+	"""Yahoo's crumb, which goes with the cookie, cached for `seconds` (14 days) while the cache preference is on; a
+	fallback crumb when Yahoo gives none."""
 	with _auth:
 		cookie = getCookie()
 		cacheFile = "finbot_yahoo_crumb.json"
@@ -106,6 +110,7 @@ def _percent(fraction):
 	return round(float(fraction) * 100, 1)
 
 def _round2(value):
+	"""value as a float rounded to two places."""
 	return round(float(value), 2)
 
 def _positive(value):
@@ -386,6 +391,7 @@ def price_series(ticker):
 def fetch_each(get, tickers, workers=8):
 	"""get(ticker) for every ticker, `workers` at a time so their waits overlap: [its result, or the exception it raised], in order."""
 	def one(ticker):
+		"""get(ticker), or the exception it raised."""
 		try:
 			return get(ticker)
 		except Exception as e:
diff --git a/tests/test_finbot.py b/tests/test_finbot.py
index 692bf311..a7cadc60 100644
--- a/tests/test_finbot.py
+++ b/tests/test_finbot.py
@@ -55,7 +55,17 @@ def post(message):
 		self.assertEqual(bot.git_version(), head[:7]) # matches what git says
 
 
-class Logging(unittest.TestCase):
+class Conventions(unittest.TestCase):
+	def test_every_function_has_a_docstring(self):
+		"""Every function and method in the bot, lib/ and util/, nested ones included, says what it does."""
+		missing = []
+		code = [*Path(ROOT, 'lib').rglob('*.py'), *Path(ROOT).glob('*.py'), *Path(ROOT, 'util').glob('*.py')]
+		for path in sorted(code):
+			for node in ast.walk(ast.parse(path.read_text())):
+				if isinstance(node, (ast.FunctionDef, ast.AsyncFunctionDef)) and ast.get_docstring(node) is None:
+					missing.append(f'{path.relative_to(ROOT)}:{node.lineno} {node.name}')
+		self.assertEqual(missing, [])
+
 	def test_the_bot_and_its_reports_print_only_to_stderr(self):
 		"""systemd discards finbot's stdout (StandardOutput=null), so a print there is never read. The util/ tools, run by
 		hand, may use it."""
diff --git a/util/sws_gen.py b/util/sws_gen.py
index 518b588d..46f768db 100644
--- a/util/sws_gen.py
+++ b/util/sws_gen.py
@@ -6,6 +6,7 @@
 import os
 
 def get_symbols(file):
+	"""The symbols in a file, one per line."""
 	symbols = []
 	with open(file) as file:
 		while line := file.readline():
@@ -14,12 +15,14 @@ def get_symbols(file):
 	return symbols
 
 def get_sws(symbol, market):
+	"""The Simply Wall St page of a stock, found with Google's I'm Feeling Lucky."""
 	query = "site:simplywall.st " + market + " " + symbol + " Stock Price, News & Analysis"
 	url = googlesearch.lucky(query)
 	print(symbol, url)
 	return url
 
 def build_dict(symbols, market):
+	"""{symbol: its Simply Wall St page}, pausing 30 minutes after every 40 searches; ASX symbols get Yahoo's .AX."""
 	count=1
 	mydict = {}
 	for symbol in symbols:
@@ -34,6 +37,8 @@ def build_dict(symbols, market):
 	return mydict
 
 def do_sws_bulk(file):
+	"""Look up every symbol in a file (as ASX listings when its name says asx) and write them to finbot_sws_<name>.json
+	in the data dir."""
 	market = ''
 	if 'asx' in file or 'ASX' in file:
 		market = 'ASX'
@@ -44,8 +49,10 @@ def do_sws_bulk(file):
 	return mydict
 
 def write_cache(mydict, cache_file):
+	"""Write the dict as JSON to cache_file, mode 0640."""
 	os.umask(0)
 	def opener(path, flags):
+		"""open() that creates the file with mode 0640."""
 		return os.open(path, flags, 0o640)
 	with open(cache_file, "w", opener=opener, encoding="utf-8") as f:
 		f.write(json.dumps(mydict))

From db95ce0dc4290b5f95d03eb62f5c60d6a0bae2fe Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 00:32:50 +0000
Subject: [PATCH 17/81] A docstring for every test and test helper too

The 421 test functions without one get one: each test method says what it checks, and each helper and stub what it
stands in for; a comment that said the same at the top of a test goes. The guard test now covers tests/ as well, so
every function in the project has a docstring. Four one-line stubs became two-line ones to take theirs; test_config
no longer imports the cal.py shim it never used.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md               |   4 +-
 tests/common.py         |   6 ++
 tests/test_charts.py    |  65 +++++++++++++++++++--
 tests/test_config.py    |  63 +++++++++++++++++++-
 tests/test_datadir.py   |  14 +++++
 tests/test_edges.py     |  59 ++++++++++++++++++-
 tests/test_finbot.py    |  22 ++++++-
 tests/test_jobs.py      |  95 +++++++++++++++++++++++++++++-
 tests/test_providers.py | 124 ++++++++++++++++++++++++++++++++++++++--
 tests/test_reports.py   |  56 +++++++++++++++++-
 tests/test_util.py      |  35 ++++++++++--
 11 files changed, 519 insertions(+), 24 deletions(-)

diff --git a/CLAUDE.md b/CLAUDE.md
index c30ff02e..d461896d 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -56,8 +56,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - A report is a list of lines. `service` is a dialect (`lib/dialects.py`), which equals the network's name: markup with
   `service.bold(text)`, `.italic`, `.strike`, `.link(url, text)`, user-written text through `service.escape(text)`, stock links
   through `links.finance_link(symbol, exchange, service)`. Tests pass `TELEGRAM`, `SLACK`, `DISCORD`.
-- Every function and method has a docstring (a test checks the bot, lib/ and util/). Tabs everywhere, tests included
-  (`.editorconfig`). New and changed lines stay under 120 characters (an edit quotes whole lines).
+- Every function and method has a docstring, tests included (a test checks). Tabs everywhere (`.editorconfig`). New and
+  changed lines stay under 120 characters (an edit quotes whole lines).
 - Log with `print(..., file=sys.stderr)`. systemd discards stdout (`StandardOutput=null`), so the bot and its reports never
   print there (a test checks); only the util/ tools, run by hand, do.
 - A chart renderer in `lib/render.py` gets `@_drawn` (the bot draws in one native thread); it must not call another `@_drawn`
diff --git a/tests/common.py b/tests/common.py
index 74803fa9..bf2cd837 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -34,6 +34,8 @@ def skip_if_old_env(case):
 
 
 def market_entry(ticker):
+	"""A Yahoo quote for a fixture ticker; T0 to T24 vary in market state, move, rating and so on, so that every
+	threshold, sort and branch has rows."""
 	i = int(ticker[1:]) if ticker[1:].isdigit() else 0 # varied data so threshold/sort/sell branches all have rows
 	return {'marketState': STATES[i % 4], 'regularMarketTime': time.time(), 'quoteType': 'EQUITY',
 			'profile_title': 'Co ' + ticker, 'profile_exchange': 'NasdaqGS', 'percent_change': (i - 12) * 1.5, 'percent_change_premarket': (i - 12) * 1.5, 'percent_change_postmarket': (i - 12) * 1.5,
@@ -48,6 +50,7 @@ class FinbotCase(unittest.TestCase):
 	"""Base class: self.sent collects ('photo'|'text', ...) tuples for everything the bot tried to send."""
 
 	def patch(self, target, attr, value):
+		"""Replace target.attr with value for the rest of the test."""
 		p = mock.patch.object(target, attr, value)
 		p.start()
 		self.addCleanup(p.stop)
@@ -61,6 +64,8 @@ def prefs(self, **changes):
 		self.patch(config, 'prefs', dataclasses.replace(config.prefs, **changes))
 
 	def setUp(self):
+		"""A fresh data dir and the offline fixtures: no request leaves the machine, and whatever the bot sends is
+		collected in self.sent (and buttons in self.markups, error lines in self.errors)."""
 		fake = mock.Mock(status_code=200, text='{}')
 		fake.json.return_value = {'ok': True, 'result': {'message_id': 1, 'chat': {'id': 5}}}
 		self.patch(__import__('requests').Session, 'request', mock.Mock(return_value=fake)) # no real network traffic
@@ -97,6 +102,7 @@ def setUp(self):
 		self.addCleanup(self.no_typing_left)
 
 	def no_typing_left(self):
+		"""Fail the test if a typing indicator it started is still running, after stopping it."""
 		left = [indicator for indicator in self.typing if indicator.is_active()]
 		for indicator in left:
 			indicator.stop()
diff --git a/tests/test_charts.py b/tests/test_charts.py
index 76c30a96..220d499e 100644
--- a/tests/test_charts.py
+++ b/tests/test_charts.py
@@ -24,17 +24,20 @@
 
 
 def decode(buf):
+	"""A PNG buffer as an RGB image."""
 	buf.seek(0)
 	return Image.open(buf).convert('RGB')
 
 
 def price_frame(days=120):
+	"""A made-up price history of `days` days, wavy and rising."""
 	dates = [datetime.date(2026, 1, 1) + datetime.timedelta(days=i) for i in range(days)]
 	closes = [100 + 12 * np.sin(i / 9) + i * 0.15 for i in range(days)]
 	return pd.DataFrame({'Date': dates, 'Close': closes})
 
 
 def compare_series():
+	"""Three made-up price series over the same 120 days, for the compare chart."""
 	index = pd.date_range('2026-01-01', periods=120)
 	return [(name, pd.Series(100 + np.cumsum(np.sin(np.arange(120) / k) + drift), index=index))
 			for name, k, drift in (('AAA', 7, 0.2), ('BBB', 11, -0.1), ('CCC', 5, 0.05))]
@@ -62,6 +65,8 @@ def compare_series():
 
 class ImageSanity(unittest.TestCase):
 	def check_image(self, buf, name, bg=BG, edge_clean=True):
+		"""Fail unless buf is a 1280x800 PNG that is not blank, starts with the background, and (edge_clean) draws
+		nothing on its edges, where clipped text or bars would show."""
 		image = decode(buf)
 		self.assertEqual(image.size, (1280, 800), name)
 		pixels = np.asarray(image)
@@ -75,6 +80,7 @@ def check_image(self, buf, name, bg=BG, edge_clean=True):
 		self.assertTrue((edge == bg).all(), f'{name}: something is drawn on the image edge (clipped text or bars?)')
 
 	def test_each_chart_type(self):
+		"""Every chart type draws a sane image."""
 		for name, make in CHARTS.items():
 			with self.subTest(chart=name):
 				if name == 'shorts':
@@ -83,17 +89,23 @@ def test_each_chart_type(self):
 					self.check_image(make(), name)
 
 	def test_edge_case_rows(self):
+		"""Rows that have broken charts before (one row, all zero, forty rows, long labels, huge or tiny values) draw
+		sane images."""
 		for name, rows in EDGE_ROWS.items():
 			with self.subTest(rows=name):
 				self.check_image(render.hbar_chart(rows, 'Title', 'sub'), name)
 				self.check_image(render.rows_chart(rows, 'Title'), name + '/rows_chart')
 
 	def test_zoomed_axis(self):
+		"""A bar chart with a narrow x range (ratings from 1 to 5) draws sane images, wide labels and many rows
+		included."""
 		self.check_image(render.hbar_chart([('AAA', 1.2), ('BBB', 3.4)], 'Ratings', xlim=(0.9, 4.0)), 'zoomed')
 		self.check_image(render.hbar_chart([('WWWWWWWW', 1.2), ('MMM', 3.4)], 'Ratings', xlim=(0.9, 4.0)), 'zoomed wide labels')
 		self.check_image(render.hbar_chart([(f'T{i}', 1 + i / 10) for i in range(15)], 'Ratings', xlim=(0.9, 3.0)), 'zoomed 15 rows')
 
 	def test_shorts_chart_sizes_and_verdicts(self):
+		"""The shorts pictogram draws for 1 to 45 rows and a long label, and each tier of short interest gets its
+		verdict."""
 		for n in (1, 5, 45):
 			rows = [(f'T{i}', 60.0 - i) for i in range(n)]
 			self.check_image(render.shorts_chart(rows, 15), f'shorts {n} rows', bg=SHORTS_BG, edge_clean=False)
@@ -104,9 +116,11 @@ def test_shorts_chart_sizes_and_verdicts(self):
 		self.assertEqual([render.shorts_verdict(p) for p, _ in tiers], [name for _, name in tiers])
 
 	def test_bar_graph_has_top_and_bottom(self):
+		"""The top-and-bottom bar chart draws a sane image."""
 		self.check_image(render.bar_graph([('AAA', 9)], [('ZZZ', -9)], 'Top and bottom'), 'bar_graph')
 
 	def test_flat_and_tiny_price_series(self):
+		"""A flat price line and one of two points draw sane images."""
 		flat = pd.DataFrame({'Date': [datetime.date(2026, 1, 1) + datetime.timedelta(days=i) for i in range(10)], 'Close': [5.0] * 10})
 		self.check_image(render.graph(flat, 'Flat', 'USD'), 'flat price')
 		two = price_frame(2)
@@ -117,10 +131,13 @@ class RenderPool(unittest.TestCase):
 	"""The bot draws charts in one native thread (render.render_off_hub), so the gevent hub keeps serving while a chart is drawn."""
 
 	def setUp(self):
+		"""Draw in the render thread, as the bot does, for this test only."""
 		render.render_off_hub()
 		self.addCleanup(setattr, render, '_render_pool', None)
 
 	def test_charts_asked_for_at_once_all_finish(self):
+		"""Six charts asked for at once all finish: one render thread, since several deadlock matplotlib under
+		gevent."""
 		jobs = [gevent.spawn(render.pie_chart, [('a', 1.0), ('b', 2.0)], f't{i}') for i in range(4)]
 		jobs += [gevent.spawn(render.hbar_chart, MIXED, 'T'), gevent.spawn(render.rows_chart, MIXED, 'T')]
 		gevent.joinall(jobs, timeout=120)
@@ -128,8 +145,11 @@ def test_charts_asked_for_at_once_all_finish(self):
 		self.assertTrue(all(decode(job.value).size == (1280, 800) for job in jobs))
 
 	def test_the_hub_keeps_serving_while_a_chart_is_drawn(self):
+		"""While the shorts pictogram is drawn (about a second), other greenlets keep running: no gap over half a
+		second."""
 		gaps, done = [], []
 		def ticker():
+			"""Note the time between wake-ups until the chart is done."""
 			last = time.monotonic()
 			while not done:
 				gevent.sleep(0.01)
@@ -145,9 +165,11 @@ def ticker():
 
 class Contracts(unittest.TestCase):
 	def capture_bars(self, rows, **kwargs):
+		"""The y positions, widths and colours hbar_chart hands to matplotlib for these rows."""
 		calls = []
 		real = Axes.barh
 		def spy(ax, y, width, *a, **k):
+			"""Record a barh call, then draw it."""
 			calls.append({'y': list(y), 'width': list(width), 'color': list(k.get('color', []))})
 			return real(ax, y, width, *a, **k)
 		with mock.patch.object(Axes, 'barh', spy):
@@ -155,9 +177,11 @@ def spy(ax, y, width, *a, **k):
 		return calls[0]
 
 	def capture_text(self, make):
+		"""Every text make() puts on the figure."""
 		texts = []
 		real = Figure.text
 		def spy(fig, x, y, s, *a, **k):
+			"""Record a text call, then draw it."""
 			texts.append(s)
 			return real(fig, x, y, s, *a, **k)
 		with mock.patch.object(Figure, 'text', spy):
@@ -165,10 +189,12 @@ def spy(fig, x, y, s, *a, **k):
 		return texts
 
 	def test_bars_are_blue_when_positive_and_red_when_negative(self):
+		"""Gains (and zero) are blue, losses red."""
 		bars = self.capture_bars([('A', 3), ('B', 0), ('C', -2)])
 		self.assertEqual(bars['color'], ['#3987e5', '#3987e5', '#fb7185'])
 
 	def test_first_row_is_on_top_and_spacers_are_skipped(self):
+		"""The first row is drawn at the top (the y axis is inverted), and a blank spacer row gets no bar."""
 		bars = self.capture_bars(MIXED)
 		self.assertEqual(bars['width'], [42, 17, 3, -5, -30])
 		self.assertEqual(bars['y'], [0, 1, 2, 4, 5]) # row 3 is the blank spacer
@@ -179,7 +205,9 @@ def test_first_row_is_on_top_and_spacers_are_skipped(self):
 		self.assertTrue(any(len(a) == 2 and a[0] > a[1] for a in calls), 'y axis must be inverted so row 0 is drawn first (top)')
 
 	def test_pixels_follow_the_sign_of_the_data(self):
+		"""The drawn image is blue for gains and red for losses, in proportion to the rows."""
 		def count(rows):
+			"""(blue pixels, red pixels) in the bar chart of rows."""
 			pixels = np.asarray(decode(render.hbar_chart(rows, 'T'))).reshape(-1, 3)
 			return (pixels == BLUE).all(axis=1).sum(), (pixels == RED).all(axis=1).sum()
 		blue, red = count([('A', 5), ('B', 3)])
@@ -190,6 +218,8 @@ def count(rows):
 		self.assertGreater(blue, 1000); self.assertGreater(red, 1000)
 
 	def test_session_tints_the_background(self):
+		"""Mid-session tints the background a subtle green and pre-market a subtle yellow; at close and by default it
+		stays plain."""
 		corner = lambda session: tuple(np.asarray(decode(render.hbar_chart(MIXED, 'T', session=session)))[2, 2][:3])
 		plain, close, mid, pre = corner(None), corner('close'), corner('midsession'), corner('premarket')
 		self.assertEqual(plain, (0x14, 0x18, 0x1f))
@@ -199,12 +229,14 @@ def test_session_tints_the_background(self):
 		self.assertLess(max(abs(int(a) - int(b)) for a, b in zip(pre, plain)), 20) # subtle
 
 	def test_longer_bar_is_drawn_longer(self):
+		"""A bar four times as long is drawn over twice as long."""
 		pixels = np.asarray(decode(render.hbar_chart([('A', 40), ('B', 10)], 'T')))
 		blue_rows = np.where((pixels == BLUE).all(axis=2).any(axis=1))[0]
 		top_band, bottom_band = pixels[blue_rows[0]], pixels[blue_rows[-1]]
 		self.assertGreater((top_band == BLUE).all(axis=1).sum(), 2 * (bottom_band == BLUE).all(axis=1).sum())
 
 	def test_titles_are_drawn(self):
+		"""The title and subtitle appear on the bar chart, and the title and period change on the price line."""
 		texts = self.capture_text(lambda: render.hbar_chart(MIXED, 'My Title', 'My subtitle'))
 		self.assertIn('My Title', texts)
 		self.assertIn('My subtitle', texts)
@@ -213,7 +245,9 @@ def test_titles_are_drawn(self):
 		self.assertTrue(any('over period' in t for t in texts))
 
 	def test_price_line_colour_follows_direction(self):
+		"""The price line's change is green up, red down, grey flat."""
 		def line_colour(closes):
+			"""The colour of the 'over period' text for a price line of these closes."""
 			frame = pd.DataFrame({'Date': [datetime.date(2026, 1, 1) + datetime.timedelta(days=i) for i in range(len(closes))], 'Close': closes})
 			texts = []
 			real = Figure.text
@@ -229,9 +263,11 @@ class ReportChartInputs(FinbotCase):
 	"""What the reports hand to the chart: sort order, titles and dates, taken from the real report code."""
 
 	def run_report(self, call):
+		"""Run call() and return what it handed hbar_chart: rows, title and subtitle."""
 		captured = []
 		real = render.hbar_chart
 		def spy(rows, title, subtitle='', **kw):
+			"""Record a chart's inputs, then draw it."""
 			captured.append({'rows': list(rows), 'title': title, 'subtitle': subtitle})
 			return real(rows, title, subtitle, **kw)
 		self.patch(render, 'hbar_chart', spy)
@@ -240,6 +276,8 @@ def spy(rows, title, subtitle='', **kw):
 		return captured[0]
 
 	def test_price_top_lists_best_first_then_worst(self):
+		""".price top: the ten best, best first, a spacer, then the ten worst, worst first, as the caption lists
+		them."""
 		chart = self.run_report(lambda: self.command('.price top 7d'))
 		values = [v for _, v in chart['rows'] if v is not None]
 		top, bottom = values[:10], values[10:]
@@ -249,24 +287,29 @@ def test_price_top_lists_best_first_then_worst(self):
 		self.assertIn(None, [v for _, v in chart['rows']]) # spacer between the two lists
 
 	def test_price_threshold_chart_is_sorted_and_titled_with_the_threshold(self):
+		""".price 5%: the chart is sorted best first and its title names the threshold."""
 		chart = self.run_report(lambda: self.command('.price 5%'))
 		values = [v for _, v in chart['rows'] if v is not None]
 		self.assertEqual(values, sorted(values, reverse=True))
 		self.assertIn('5', chart['title'])
 
 	def test_price_over_a_period_has_a_dated_title(self):
+		""".price 7d: the chart's title gives the dates covered."""
 		chart = self.run_report(lambda: self.command('.price 7d'))
 		self.assertRegex(chart['title'], r'\d{1,2} \w{3} - \d{1,2} \w{3} \d{4}') # 28 Sep - 5 Oct 2026
 
 	def test_performance_is_sorted_by_percent(self):
+		""".performance 7d: the chart is sorted best first."""
 		chart = self.run_report(lambda: self.command('.performance 7d'))
 		values = [v for _, v in chart['rows'] if v is not None]
 		self.assertEqual(values, sorted(values, reverse=True))
 
 	def test_shorts_is_sorted_and_carries_the_threshold(self):
+		""".shorts: the pictogram gets the stocks over the threshold, most shorted first, and the threshold."""
 		captured = []
 		real = render.shorts_chart
 		def spy(rows, threshold, *a, **k):
+			"""Record the pictogram's inputs, then draw it."""
 			captured.append((list(rows), threshold))
 			return real(rows, threshold, *a, **k)
 		self.patch(render, 'shorts_chart', spy)
@@ -285,9 +328,12 @@ class GoldenImages(unittest.TestCase):
 
 	@staticmethod
 	def thumb(image):
+		"""The 160x100 thumbnail that is compared, and stored as the golden image."""
 		return image.resize((160, 100), Image.LANCZOS) # goldens are stored at this size too: a few KB each, not 100 KB
 
 	def check(self, name):
+		"""Compare a chart type's thumbnail with its golden image (or write the golden with UPDATE_GOLDEN=1, or when
+		there is none)."""
 		thumb = self.thumb(decode(CHARTS[name]()))
 		path = os.path.join(GOLDEN, name + '.png')
 		if os.environ.get('UPDATE_GOLDEN') or not os.path.exists(path):
@@ -300,10 +346,21 @@ def check(self, name):
 		self.assertLess(diff, self.TOLERANCE, f'{name} no longer matches tests/golden/{name}.png (diff {diff:.2f}). '
 						f'If the change is intended, run: UPDATE_GOLDEN=1 python -m unittest tests.test_charts')
 
-	def test_hbar_mixed(self): self.check('hbar_mixed')
-	def test_price_line(self): self.check('price_line')
-	def test_shorts(self): self.check('shorts')
-	def test_compare_lines(self): self.check('compare_lines')
+	def test_hbar_mixed(self):
+		"""The bar chart, gains and losses with a spacer, matches its golden image."""
+		self.check('hbar_mixed')
+
+	def test_price_line(self):
+		"""The price line matches its golden image."""
+		self.check('price_line')
+
+	def test_shorts(self):
+		"""The shorts pictogram matches its golden image."""
+		self.check('shorts')
+
+	def test_compare_lines(self):
+		"""The compare chart matches its golden image."""
+		self.check('compare_lines')
 
 
 if __name__ == '__main__':
diff --git a/tests/test_config.py b/tests/test_config.py
index fa6a0dea..778dce67 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -4,7 +4,7 @@
 from unittest import mock
 from lib import config
 from util import check_config, split_env
-import bot, cal
+import bot
 from lib import cache, sharesight, worker, yahoo
 from lib.reports import performance, price, shorts, trades
 
@@ -29,6 +29,7 @@ def keys_in(path):
 
 class Tables(unittest.TestCase):
 	def test_the_two_key_sets_are_disjoint_and_cover_everything_config_reads(self):
+		"""Startup and preference keys do not overlap, and between them name every setting config.py reads."""
 		self.assertEqual(set(config.STARTUP_KEYS) & set(config.PREFERENCE_KEYS), set())
 		self.assertEqual((len(config.STARTUP_KEYS), len(config.PREFERENCE_KEYS)), (17, 26))
 		source = (REPO / 'lib' / 'config.py').read_text()
@@ -37,14 +38,18 @@ def test_the_two_key_sets_are_disjoint_and_cover_everything_config_reads(self):
 		self.assertEqual(read, set(config.STARTUP_KEYS) | set(config.PREFERENCE_KEYS))
 
 	def test_secrets_are_startup_keys(self):
+		"""Every secret is a startup key: preferences.conf holds none."""
 		self.assertTrue(set(config.SECRET_KEYS) <= set(config.STARTUP_KEYS))
 		self.assertEqual(len(config.SECRET_KEYS), 11)
 
 	def test_the_preference_defaults_are_the_old_defaults(self):
+		"""Unset preferences take the defaults finbot always had."""
 		values = config.preference_values(lambda name, default=None: default)
 		self.assertEqual(fields(values), DEFAULTS)
 
 	def test_templates_have_every_key_and_the_real_defaults(self):
+		"""defaults/preferences.conf and defaults/finbot.env list every key, change no default, and are valid as
+		shipped."""
 		self.assertEqual(sorted(keys_in(REPO / 'defaults' / 'preferences.conf')), sorted(config.PREFERENCE_KEYS))
 		self.assertEqual(sorted(keys_in(REPO / 'defaults' / 'finbot.env')), sorted(config.STARTUP_KEYS))
 		prefs = config.read_preferences(REPO / 'defaults' / 'preferences.conf')
@@ -56,28 +61,34 @@ def test_templates_have_every_key_and_the_real_defaults(self):
 
 class Parsing(unittest.TestCase):
 	def parse(self, text):
+		"""(what read_preferences makes of text, the file it was written to)."""
 		path = Path(tempfile.mkdtemp(), 'preferences.conf')
 		path.write_text(text)
 		return config.read_preferences(path), path
 
 	def errors(self, text):
+		"""The errors read_preferences raises for text."""
 		with self.assertRaises(config.ConfigError) as caught:
 			self.parse(text)
 		return caught.exception.errors
 
 	def test_comments_blanks_quotes_and_equals_in_values(self):
+		"""Comments and blank lines are skipped, quotes come off, and spaces and blanks in values are kept."""
 		values, _ = self.parse('# c\n\nfuture_days = 30\nhyperlinkProvider="google"\nexclude_portfolios=1 2\ncountry_code=\n')
 		self.assertEqual(values, {'future_days': '30', 'hyperlinkProvider': 'google', 'exclude_portfolios': '1 2', 'country_code': ''})
 
 	def test_missing_file_is_none(self):
+		"""No preferences.conf is not an error: None."""
 		self.assertIsNone(config.read_preferences(Path(tempfile.mkdtemp(), 'nope')))
 
 	def test_unknown_key_gets_a_hint_and_a_line_number(self):
+		"""An unknown key is reported with its line number and the key it was probably meant to be."""
 		errors = self.errors('future_days=3\nfutur_days=3\n')
 		self.assertEqual(len(errors), 1)
 		self.assertIn('line 2: unknown key futur_days (did you mean future_days?)', errors[0])
 
 	def test_a_startup_key_in_preferences_says_where_it_belongs(self):
+		"""A startup key in preferences.conf is sent to finbot.env, without printing its value."""
 		errors = self.errors('slack_webhook=https://x\nip=0.0.0.0\ntelegramChatID=5\nscheduler=False\n')
 		self.assertEqual(len(errors), 4)
 		for line in errors:
@@ -85,6 +96,8 @@ def test_a_startup_key_in_preferences_says_where_it_belongs(self):
 		self.assertNotIn('https://x', ''.join(errors)) # a value is never printed
 
 	def test_malformed_lines_sections_duplicates_and_the_old_name(self):
+		"""Sections, lines that are not KEY=value, a key set twice and the renamed volatile_multiplier are all
+		reported."""
 		errors = self.errors('[reports]\nnot a setting\nfuture_days=1\nfuture_days=2\nvolatile_multiplier=3\n')
 		self.assertEqual(len(errors), 4)
 		self.assertIn('no sections', errors[0])
@@ -93,9 +106,11 @@ def test_malformed_lines_sections_duplicates_and_the_old_name(self):
 		self.assertIn('renamed to volatility_multiplier', errors[3])
 
 	def test_every_problem_is_reported_not_just_the_first(self):
+		"""Every problem in the file is reported at once."""
 		self.assertEqual(len(self.errors('futur_days=1\nport=1\n[x]\n')), 3)
 
 	def test_invalid_values_are_named(self):
+		"""A value of the wrong type is reported with its key and value."""
 		prefs = {'future_days': 'abc', 'price_percent': 'x'}
 		with self.assertRaises(config.ConfigError) as caught:
 			config.preference_values(config._getenv_for(prefs))
@@ -103,6 +118,7 @@ def test_invalid_values_are_named(self):
 		self.assertIn("future_days='abc' is not a whole number", caught.exception.errors[0])
 
 	def test_booleans_take_the_usual_words_and_refuse_a_typo(self):
+		"""Booleans take true/false, yes/no, on/off and 1/0 in any case; a typo is an error (it used to mean False)."""
 		for word, value in (('True', True), ('yes', True), ('ON', True), ('1', True), ('false', False), ('No', False), ('off', False), ('0', False)):
 			with self.subTest(word=word):
 				self.assertIs(config.preference_values(config._getenv_for({'graph': word})).graph, value)
@@ -111,10 +127,12 @@ def test_booleans_take_the_usual_words_and_refuse_a_typo(self):
 		self.assertIn("graph='Ture' is not true or false", caught.exception.errors[0])
 
 	def test_a_blank_value_means_the_default(self):
+		"""A key with a blank value takes its default."""
 		values = config.preference_values(config._getenv_for({'cache': '', 'future_days': '', 'country_code': ''}))
 		self.assertEqual((values.cache, values.future_days, values.country_code), (True, 30, 'AU'))
 
 	def test_unreadable_file_is_an_error_naming_it(self):
+		"""A preferences file that cannot be read is an error that names it."""
 		path = Path(tempfile.mkdtemp(), 'preferences.conf')
 		path.mkdir() # reading a directory fails like a permission error would
 		with self.assertRaises(config.ConfigError) as caught:
@@ -122,11 +140,13 @@ def test_unreadable_file_is_an_error_naming_it(self):
 		self.assertIn(f'cannot read {path}', caught.exception.errors[0])
 
 	def test_precedence_file_then_environment_then_default(self):
+		"""A preference comes from the file, else the environment, else its default."""
 		with mock.patch.dict(os.environ, {'future_days': '5', 'past_days': '4'}):
 			values = config.preference_values(config._getenv_for({'future_days': '30'}))
 		self.assertEqual((values.future_days, values.past_days, values.shorts_percent), (30, 4, 15))
 
 	def test_the_old_multiplier_name_is_gone_everywhere(self):
+		"""volatile_multiplier is no longer read, and in the environment it stops the start with the new name."""
 		with mock.patch.dict(os.environ, {'volatile_multiplier': '3'}):
 			self.assertEqual(config.preference_values(config._getenv_for({})).volatility_multiplier, 2.0) # not read any more
 			with self.assertRaises(config.ConfigError) as caught:
@@ -135,6 +155,7 @@ def test_the_old_multiplier_name_is_gone_everywhere(self):
 		self.assertEqual(config.preference_values(config._getenv_for({'volatility_multiplier': '4'})).volatility_multiplier, 4.0)
 
 	def test_startup_file_linting(self):
+		"""util/check_config.py reports bad lines in finbot.env first, then bad values, and nothing for a good file."""
 		path = Path(tempfile.mkdtemp(), 'finbot.env')
 		path.write_text('ip=1.2.3.4\nfuture_days=3\nslak_webhook=x\nport=http\n')
 		problems = check_config.check(path, Path(tempfile.mkdtemp(), 'none'))
@@ -149,6 +170,7 @@ def test_startup_file_linting(self):
 
 class Reload(unittest.TestCase):
 	def setUp(self):
+		"""An empty preferences.conf of this test's own, put back to the defaults afterwards."""
 		self.file = Path(tempfile.mkdtemp(), 'preferences.conf')
 		self.file.write_text('')
 		patch = mock.patch.object(config, 'preferences_path', str(self.file))
@@ -157,13 +179,17 @@ def setUp(self):
 		self.addCleanup(self.back_to_defaults)
 
 	def back_to_defaults(self):
+		"""Empty the file and reload, so the next test starts from the defaults."""
 		self.file.write_text('')
 		config.reload()
 
 	def write(self, text):
+		"""Write the preferences file."""
 		self.file.write_text(text)
 
 	def test_a_change_is_seen_at_once(self):
+		"""A reload applies every change, says what changed, and swaps in a new object, so no request sees half a
+		reload."""
 		self.write('future_days=45\nprice_percent=7\nshorts_percent=12\nexclude_portfolios=1156263\nalliterate=1\nhyperlinkProvider=yahoo\n'
 			'hyperlinkFooter=false\ndebug=1\ngraph=0\ndemote_volatile=0\nvolatility_multiplier=3\npast_days=14\nmin_analysts=3\n')
 		before = config.prefs
@@ -178,6 +204,7 @@ def test_a_change_is_seen_at_once(self):
 		self.assertEqual(config.reload(), []) # nothing more to change
 
 	def test_a_value_removed_from_the_file_goes_back_to_its_default(self):
+		"""A key taken out of the file goes back to its default on reload."""
 		self.write('future_days=45\n')
 		config.reload()
 		self.write('# empty\n')
@@ -185,6 +212,7 @@ def test_a_value_removed_from_the_file_goes_back_to_its_default(self):
 		self.assertEqual(config.prefs.future_days, 30)
 
 	def test_defaults_that_follow_a_preference_are_read_when_called(self):
+		"""A cache read without seconds follows cache_seconds as reloaded; an explicit value always wins."""
 		cache_file = Path(config.env.cache_dir, 'finbot_test_reload.json')
 		cache_file.write_text('{"a": 1}')
 		self.addCleanup(cache_file.unlink)
@@ -199,6 +227,7 @@ def test_defaults_that_follow_a_preference_are_read_when_called(self):
 		self.assertIsNone(cache.read('finbot_test_reload.json', 50)) # an explicit value always wins
 
 	def test_an_error_applies_nothing(self):
+		"""A reload that finds any error applies nothing."""
 		self.write('future_days=45\n')
 		config.reload()
 		for text in ('future_days=3\nprice_percent=abc\n', 'future_days=3\nfutur_days=4\n', 'future_days=3\nslack_webhook=x\n'):
@@ -208,6 +237,7 @@ def test_an_error_applies_nothing(self):
 			self.assertEqual(config.prefs.future_days, 45)
 
 	def test_a_missing_file_is_an_error_and_keeps_the_running_values(self):
+		"""A reload with the file gone is an error and keeps the running values."""
 		self.write('future_days=45\n')
 		config.reload()
 		self.file.unlink()
@@ -221,6 +251,7 @@ def test_nothing_copies_the_config_or_freezes_a_preference(self):
 		"""Modules read config.prefs.<name> when they run, so that a reload is seen: no star import (which copied every name into
 		each module), and no function default or module constant computed from a preference."""
 		def reads_prefs(node):
+			"""Whether an expression reads config.prefs."""
 			return any(isinstance(n, ast.Attribute) and n.attr == 'prefs' and isinstance(n.value, ast.Name) and n.value.id == 'config' for n in ast.walk(node))
 		offenders = []
 		for path in sorted([*(REPO / 'lib').rglob('*.py'), *REPO.glob('*.py'), *(REPO / 'util').glob('*.py')]):
@@ -238,6 +269,8 @@ def reads_prefs(node):
 		self.assertEqual(offenders, [])
 
 	def test_a_job_process_runs_on_the_snapshot_not_the_edited_file(self):
+		"""A job process gets the bot's preferences in its environment and does not read the file, which may have been
+		edited since; a script run by hand reads the file."""
 		self.write('future_days=45\n')
 		config.reload()
 		env = config.child_environment()
@@ -252,10 +285,12 @@ def test_a_job_process_runs_on_the_snapshot_not_the_edited_file(self):
 		self.assertEqual(run(), '99') # a script run by hand reads the file itself
 
 	def test_run_child_hands_the_snapshot_to_the_job(self):
+		"""run_child passes the preferences in force to the job process."""
 		from lib import jobs
 		seen = {}
 		real = jobs.subprocess.Popen
 		def popen(args, **kw):
+			"""Note the environment the job is started with, and run a no-op instead."""
 			seen.update(kw.get('env', {}))
 			return real([sys.executable, '-c', 'pass'])
 		self.write('future_days=45\n')
@@ -268,6 +303,8 @@ def popen(args, **kw):
 
 class Startup(unittest.TestCase):
 	def run_config(self, code='from lib import config as c; print(c.prefs.future_days)', prefs=None, **env):
+		"""Run code in a fresh interpreter with a data dir of its own (and prefs as its preferences.conf), and the
+		environment cleared of settings."""
 		data = tempfile.mkdtemp()
 		if prefs is not None:
 			Path(data, 'preferences.conf').write_text(prefs)
@@ -275,13 +312,17 @@ def run_config(self, code='from lib import config as c; print(c.prefs.future_day
 		return subprocess.run([sys.executable, '-c', code], cwd=REPO, env={**e, 'DATA_DIR': data, **env}, capture_output=True, text=True)
 
 	def test_no_preferences_file_is_fine(self):
+		"""finbot starts without a preferences.conf, on the defaults."""
 		out = self.run_config()
 		self.assertEqual((out.returncode, out.stdout.split()[-1]), (0, '30'))
 
 	def test_a_preferences_file_is_read_at_start(self):
+		"""preferences.conf is read at start."""
 		self.assertEqual(self.run_config(prefs='future_days=45\n').stdout.split()[-1], '45')
 
 	def test_a_broken_preferences_file_refuses_to_start(self):
+		"""A bad value, an unknown key, a startup key or a malformed line in preferences.conf stops the start, saying
+		why."""
 		for prefs, expect in (('future_days=abc\n', "future_days='abc' is not a whole number"), ('futur_days=1\n', 'unknown key futur_days'),
 				('slack_webhook=x\n', 'belongs in finbot.env'), ('what\n', 'expected KEY=value')):
 			out = self.run_config(prefs=prefs)
@@ -290,6 +331,7 @@ def test_a_broken_preferences_file_refuses_to_start(self):
 			self.assertIn(expect, out.stderr)
 
 	def test_an_unreadable_preferences_file_refuses_to_start(self):
+		"""A preferences.conf that cannot be read stops the start."""
 		data = tempfile.mkdtemp()
 		Path(data, 'preferences.conf').mkdir()
 		out = subprocess.run([sys.executable, '-c', 'from lib import config'], cwd=REPO, env={**os.environ, 'DATA_DIR': data}, capture_output=True, text=True)
@@ -297,11 +339,14 @@ def test_an_unreadable_preferences_file_refuses_to_start(self):
 		self.assertIn('cannot read', out.stderr)
 
 	def test_a_bad_startup_value_refuses_to_start(self):
+		"""A bad value in finbot.env stops the start, saying which."""
 		out = self.run_config(port='http')
 		self.assertEqual(out.returncode, 1)
 		self.assertIn("port='http' is not a whole number", out.stderr)
 
 	def test_a_preference_in_the_environment_is_refused_by_the_bot(self):
+		"""A preference in the bot's environment (finbot.env) is refused, but a job process gets its preferences that
+		way, and startup keys belong there."""
 		with mock.patch.dict(os.environ, {'future_days': '30', 'debug': '1'}):
 			with self.assertRaises(config.ConfigError) as caught:
 				config.check_startup_environment()
@@ -313,12 +358,14 @@ def test_a_preference_in_the_environment_is_refused_by_the_bot(self):
 			config.check_startup_environment() # startup keys are what the environment is for
 
 	def test_the_bot_refuses_to_start_with_a_preference_in_its_environment(self):
+		"""bot.py exits 1 when finbot.env holds a preference."""
 		out = subprocess.run([sys.executable, 'bot.py'], cwd=REPO, capture_output=True, text=True, timeout=60,
 			env={**os.environ, 'DATA_DIR': tempfile.mkdtemp(), 'future_days': '30'})
 		self.assertEqual(out.returncode, 1)
 		self.assertIn('future_days is a preference', out.stderr)
 
 	def test_a_leftover_env_in_the_checkout_refuses_to_start(self):
+		"""An old .env in the checkout stops the start, pointing to util/split_env.py."""
 		old = Path(tempfile.mkdtemp(), '.env')
 		with mock.patch.object(config, 'OLD_ENV', old):
 			config.check_startup_environment() # none there: fine
@@ -330,16 +377,19 @@ def test_a_leftover_env_in_the_checkout_refuses_to_start(self):
 		self.assertIn('/util/split_env.py, which writes finbot.env.new and preferences.conf.new', caught.exception.errors[1])
 
 	def test_a_discord_webhook_uses_the_native_endpoint(self):
+		"""A Discord webhook URL is used as is, its old /slack suffix removed."""
 		for url in ('https://discord.com/api/webhooks/1/abc', 'https://discord.com/api/webhooks/1/abc/slack/'): # the old setup's suffix
 			values = config.startup_values(lambda name, default=None: url if name == 'discord_webhook' else default)
 			self.assertEqual(values.webhooks['discord'], 'https://discord.com/api/webhooks/1/abc')
 
 	def test_slack_commands_without_a_bot_token_are_warned_about(self):
+		"""Slack commands without slackBotToken get a startup warning, since they would get no reply."""
 		for token, warnings in (('', ['slackOutgoingWebhook is set but slackBotToken is not: Slack commands will get no reply']), ('xoxb-1', [])):
 			with mock.patch.object(config, 'env', dataclasses.replace(config.env, slackOutgoingWebhook='https://example.com/slack', slackBotToken=token)):
 				self.assertEqual(config.startup_warnings(), warnings)
 
 	def test_startup_report_is_a_short_count_never_names_or_values(self):
+		"""The startup report counts the startup keys set, and never names them or shows their values."""
 		with mock.patch.dict(os.environ, {'slack_webhook': 'https://secret.example/abc', 'port': '5000'}, clear=True):
 			lines = config.startup_report()
 		self.assertEqual(lines[0], 'startup environment provides 2 of 17 startup keys (values hidden)')
@@ -350,11 +400,14 @@ def test_startup_report_is_a_short_count_never_names_or_values(self):
 
 class SigHup(unittest.TestCase):
 	def restore(self, file):
+		"""Empty the preferences file and reload: back to the defaults."""
 		with mock.patch.object(config, 'preferences_path', str(file)):
 			file.write_text('')
 			config.reload()
 
 	def test_reload_all_reloads_preferences_and_the_schedule_independently(self):
+		"""SIGHUP reloads preferences and the schedule; a preferences error is logged, keeps the running values, and the
+		schedule reloads anyway."""
 		file = Path(tempfile.mkdtemp(), 'preferences.conf')
 		file.write_text('future_days=45\n')
 		scheduler = mock.Mock()
@@ -374,6 +427,7 @@ def test_reload_all_reloads_preferences_and_the_schedule_independently(self):
 			self.assertEqual(config.prefs.future_days, 45)
 
 	def test_reload_with_the_scheduler_off_still_reloads_preferences(self):
+		"""With the scheduler off, SIGHUP still reloads preferences."""
 		file = Path(tempfile.mkdtemp(), 'preferences.conf')
 		file.write_text('future_days=11\n')
 		self.addCleanup(self.restore, file)
@@ -385,6 +439,8 @@ def test_reload_with_the_scheduler_off_still_reloads_preferences(self):
 
 class Tools(unittest.TestCase):
 	def test_split_env(self):
+		"""util/split_env.py sorts an old .env into startup keys and preferences, renames volatile_multiplier, and notes
+		what it drops."""
 		startup, prefs, notes = split_env.split('slack_webhook=x\nfuture_days=3\nvolatile_multiplier=2\nslackToken=y\nSTATE_DIRECTORY=/a\nbogus=1\nnot a setting\n# c\nport=80\n')
 		self.assertEqual(startup, ['slack_webhook=x', 'port=80'])
 		self.assertEqual(prefs, ['future_days=3', 'volatility_multiplier=2'])
@@ -392,6 +448,7 @@ def test_split_env(self):
 		self.assertTrue(any('slackToken dropped' in n for n in notes) and any('unknown key bogus' in n for n in notes))
 
 	def test_split_env_reads_the_old_dotenv_format_too(self):
+		"""split_env reads the old dotenv format: quotes, export, comments after values and # inside URLs."""
 		old = "# c\nfuture_days = '30' # note\nexport slack_webhook = \"https://x/#frag\"\nport = 80 # comment\nbare\ndiscord_webhook=http://y/#z\n\n"
 		startup, prefs, notes = split_env.split(old)
 		self.assertEqual(startup, ['slack_webhook=https://x/#frag', 'port=80', 'discord_webhook=http://y/#z'])
@@ -400,6 +457,8 @@ def test_split_env_reads_the_old_dotenv_format_too(self):
 		self.assertFalse((REPO / 'util' / 'migrate_env.py').exists()) # split_env does its job now
 
 	def test_split_env_writes_next_to_the_file_and_never_overwrites(self):
+		"""split_env writes finbot.env.new (mode 0600) and preferences.conf.new beside the file, refuses to overwrite
+		them, and says so for a missing file."""
 		d = Path(tempfile.mkdtemp())
 		(d / 'old.env').write_text('port=80\nfuture_days=3\n')
 		run = lambda: subprocess.run([sys.executable, str(REPO / 'util' / 'split_env.py'), str(d / 'old.env')], capture_output=True, text=True)
@@ -414,6 +473,8 @@ def test_split_env_writes_next_to_the_file_and_never_overwrites(self):
 		self.assertEqual((out.returncode, out.stderr.strip()), (1, f'{d}/missing.env does not exist: nothing to split'))
 
 	def test_check_config_script_exit_status_and_it_does_not_create_directories(self):
+		"""util/check_config.py exits 0 for a good file and 1 for a bad one, and creates no data directory (systemd may
+		run it as root)."""
 		d = Path(tempfile.mkdtemp())
 		(d / 'finbot.env').write_text('port=80\n')
 		data = d / 'data'
diff --git a/tests/test_datadir.py b/tests/test_datadir.py
index de6d55fa..9d10236c 100644
--- a/tests/test_datadir.py
+++ b/tests/test_datadir.py
@@ -9,6 +9,7 @@
 
 
 def run_config(code, **env):
+	"""What `code` prints when run in a fresh interpreter in the checkout, with DATA_DIR cleared and `env` added."""
 	e = {k: v for k, v in os.environ.items() if k != 'DATA_DIR'}
 	e.update(env)
 	return subprocess.run([sys.executable, '-c', code], cwd=REPO, env=e, capture_output=True, text=True, check=True).stdout.strip()
@@ -16,30 +17,38 @@ def run_config(code, **env):
 
 class DataDirTests(FinbotCase):
 	def test_default_unchanged(self):
+		"""Without DATA_DIR, state lives in var/ in the checkout and the cache in var/cache."""
 		out = run_config("from lib import config as c; print(c.env.data_dir, c.env.cache_dir)")
 		self.assertEqual(out.split()[-2:], ['var', 'var/cache'])
 
 	def test_data_dir_moves_var_and_cache(self):
+		"""DATA_DIR moves both, and the cache directory is created."""
 		with tempfile.TemporaryDirectory() as d:
 			out = run_config("from lib import config as c; print(c.env.data_dir, c.env.cache_dir)", DATA_DIR=d)
 			self.assertEqual(out.split()[-2:], [d, d + '/cache'])
 			self.assertTrue(os.path.isdir(d + '/cache'))
 
 	def test_config_reads_environment_only(self):
+		"""Settings come from the environment; python-dotenv is not loaded."""
 		out = run_config("import sys; from lib import config as c; print(c.prefs.price_percent, 'dotenv' in sys.modules)", price_percent='3.3')
 		self.assertEqual(out.split()[-2:], ['3.3', 'False'])
 
 	def test_state_directory_fallback(self):
+		"""systemd's STATE_DIRECTORY (its first entry) is the data dir when DATA_DIR is unset; MPLCONFIGDIR goes under
+		the cache unless set."""
 		show = "from lib import config as c; print(c.env.data_dir, c.env.cache_dir, c.os.environ['MPLCONFIGDIR'])"
 		with tempfile.TemporaryDirectory() as d:
 			e = {k: v for k, v in os.environ.items() if k not in ('DATA_DIR', 'MPLCONFIGDIR')}
 			def run(**env):
+				"""The data dir, cache dir and MPLCONFIGDIR a fresh interpreter ends up with, given `env`."""
 				return subprocess.run([sys.executable, '-c', show], cwd=REPO, env=dict(e, **env), capture_output=True, text=True, check=True).stdout.split()[-3:]
 			self.assertEqual(run(STATE_DIRECTORY=d + ':/x/y'), [d, d + '/cache', d + '/cache/matplotlib'])
 			self.assertEqual(run(STATE_DIRECTORY='/x/y', DATA_DIR=d)[0], d) # DATA_DIR wins
 			self.assertEqual(run(DATA_DIR=d, MPLCONFIGDIR='/custom')[2], '/custom') # explicit value kept
 
 	def test_render_import_sets_mplconfigdir_without_home(self):
+		"""Importing render sets MPLCONFIGDIR into the cache before matplotlib loads, so charts work without a home
+		directory."""
 		e = {k: v for k, v in os.environ.items() if k not in ('MPLCONFIGDIR', 'XDG_CONFIG_HOME', 'XDG_CACHE_HOME', 'HOME')}
 		code = "import os; from lib import render; import matplotlib.dates; print(os.environ['MPLCONFIGDIR'])" # render loads matplotlib lazily
 		with tempfile.TemporaryDirectory() as d:
@@ -47,12 +56,14 @@ def test_render_import_sets_mplconfigdir_without_home(self):
 			self.assertEqual(out.split()[-1], d + '/cache/matplotlib')
 
 	def test_data_file_override_and_fallback(self):
+		"""A data file is the user's copy in the data dir when there is one, else the shipped default."""
 		name = 'finbot_adr.json'
 		self.assertEqual(config.data_file(name), str(REPO / 'defaults' / name))
 		Path(config.env.data_dir, name).write_text('{}')
 		self.assertEqual(config.data_file(name), config.env.data_dir + '/' + name)
 
 	def test_a_data_file_is_parsed_again_only_when_it_changes(self):
+		"""config.data_json parses a file again only after its modification time changes."""
 		path = Path(config.env.data_dir, 'finbot_adr.json')
 		path.write_text('{"A": "B"}')
 		self.assertEqual(config.data_json('finbot_adr.json'), {'A': 'B'})
@@ -63,6 +74,8 @@ def test_a_data_file_is_parsed_again_only_when_it_changes(self):
 		self.assertEqual(config.data_json('finbot_adr.json'), {'A': 'C'})
 
 	def test_sync_defaults(self):
+		"""util/sync_defaults.py: a dry run by default, never finbot.env, never overwrites a changed file without
+		--force (which keeps a .bak), and says what it skipped."""
 		self.assertIn(('finbot_adr.json', 'would copy'), sync_defaults.sync())
 		self.assertFalse(Path(config.env.data_dir, 'finbot_adr.json').exists()) # dry run
 		sync_defaults.sync(apply=True)
@@ -79,6 +92,7 @@ def test_sync_defaults(self):
 		self.assertEqual(Path(str(mine) + '.bak').read_text(), '{"mine": 1}')
 
 	def test_sync_defaults_to_another_destination(self):
+		"""util/sync_defaults.py --dest copies there and leaves the data dir alone."""
 		other = Path(config.env.data_dir, 'elsewhere')
 		self.assertIn(('finbot_adr.json', 'would copy'), sync_defaults.sync(dest=other))
 		self.assertFalse(other.exists()) # dry run
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 19550542..22a9633f 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -15,6 +15,7 @@ class UpstreamEmpty(FinbotCase):
 	ONE_LINE_ERROR = ('.history T1', '.compare T1 T2') # these say so with a single error line when the ticker has no data
 
 	def run_all(self):
+		"""Run every command in COMMANDS: none may crash, and .history and .compare may say so in one error line."""
 		for cmd in COMMANDS:
 			with self.subTest(cmd=cmd):
 				if cmd in self.ONE_LINE_ERROR:
@@ -25,12 +26,15 @@ def run_all(self):
 					self.command(cmd)
 
 	def test_yahoo_returns_empty_dict(self):
+		"""Yahoo answers with no data at all: every command still replies sanely."""
 		self.patch(yahoo, 'fetch', lambda ts: {})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {})
 		self.run_all()
 
 	def test_yahoo_raises(self):
+		"""Yahoo is down: each command reports one error line, not a traceback."""
 		def down(*a, **k):
+			"""Fail like Yahoo when it is unreachable."""
 			raise RuntimeError('Yahoo API unavailable')
 		for name in ('fetch', 'fetch_detail', 'price_series'):
 			self.patch(yahoo, name, down)
@@ -42,7 +46,9 @@ def down(*a, **k):
 				self.assertTrue(self.errors[0].startswith('⚠️'))
 
 	def test_sharesight_down_is_reported_once(self):
+		"""Sharesight is down: .performance reports one error line."""
 		def down(*a, **k):
+			"""Fail like Sharesight when it errors."""
 			raise RuntimeError('Sharesight performance error: boom')
 		self.patch(sharesight, 'get_performance', down)
 		self.patch(sharesight, 'get_portfolios', down)
@@ -51,7 +57,9 @@ def down(*a, **k):
 		self.assertEqual(len(self.errors), 1)
 
 	def test_a_failed_command_stops_typing_before_saying_why(self):
+		"""A command that fails stops its typing indicator before the error line goes out."""
 		def down(*a, **k):
+			"""Fail like Sharesight when it errors."""
 			raise RuntimeError('Sharesight performance error: boom')
 		self.patch(sharesight, 'get_portfolios', down)
 		typing_when_reported = []
@@ -60,6 +68,7 @@ def down(*a, **k):
 		self.assertEqual(typing_when_reported, [[False]]) # its one indicator had stopped when the error line went out
 
 	def test_no_holdings_and_empty_watchlist(self):
+		"""With nothing held and nothing watched, the watchlist, price and earnings commands still answer."""
 		self.patch(watchlist, 'with_holdings', lambda: [])
 		for cmd in ('.watchlist', '.price', '.earnings'):
 			with self.subTest(cmd=cmd):
@@ -68,6 +77,7 @@ def test_no_holdings_and_empty_watchlist(self):
 
 class WatchlistActions(FinbotCase):
 	def test_add_duplicate_delete_and_unknown_action(self):
+		"""Add, add again (said differently), remove, and an unknown action is refused by name."""
 		add = self.command('.watchlist add MSFT')
 		self.assertTrue(add)
 		self.assertEqual(state.load('finbot_watchlist.json'), ['MSFT'])
@@ -79,6 +89,7 @@ def test_add_duplicate_delete_and_unknown_action(self):
 		self.assertIn('not a valid watchlist action', str(self.command('.watchlist frobnicate MSFT')))
 
 	def test_a_bare_symbol_yahoo_does_not_know_is_tried_on_the_asx(self):
+		"""A symbol Yahoo does not know is added as its ASX listing, once; a ticker with a market is not retried."""
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts if t.endswith('.AX')})
 		self.assertIn('so I added', str(self.command('.watchlist add BHP')))
 		self.assertIn('already tracking', str(self.command('.watchlist add BHP')))
@@ -86,8 +97,10 @@ def test_a_bare_symbol_yahoo_does_not_know_is_tried_on_the_asx(self):
 		self.assertEqual(watchlist.tickers(), ['BHP.AX'])
 
 	def test_two_chats_adding_at_once_both_count(self):
-		# Yahoo is asked before the file is opened: both adds used to read the list, wait on Yahoo, and the later write won
+		"""Two chats adding at once both count: Yahoo is asked before the file is opened, where both adds used to read
+		the list, wait on Yahoo, and the later write won."""
 		def slow_fetch(tickers):
+			"""A Yahoo that takes its time, so the two adds overlap."""
 			time.sleep(0.01)
 			return {t: market_entry(t) for t in tickers}
 		self.patch(yahoo, 'fetch', slow_fetch)
@@ -99,6 +112,8 @@ def slow_fetch(tickers):
 		self.assertEqual(sorted(watchlist.tickers()), ['T1', 'T2'])
 
 	def test_removing_a_ticker_yahoo_no_longer_knows(self):
+		"""A ticker Yahoo no longer knows can still be removed, and is said to be; removing it again says it is not
+		there."""
 		state.save('finbot_watchlist.json', ['GONE', 'T1'])
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts if t != 'GONE'})
 		self.assertIn('I deleted', str(self.command('.watchlist rm GONE')))
@@ -108,12 +123,15 @@ def test_removing_a_ticker_yahoo_no_longer_knows(self):
 
 class BadArguments(FinbotCase):
 	def test_compare_argument_errors_give_usage(self):
+		""".compare with too few or too many tickers, or one twice, gets its usage."""
 		for cmd in ('.compare', '.compare T1', '.compare T1 T2 T3 T4 T5 T6 T7', '.compare T1 T1'):
 			with self.subTest(cmd=cmd):
 				self.assertIn('Usage', str(self.command(cmd)))
 
 	def test_compare_with_a_ticker_that_has_no_history(self):
+		""".compare with a ticker that has no history reports one error line."""
 		def no_history(t):
+			"""Fail like Yahoo for a ticker with no price history."""
 			raise RuntimeError(f'no price history for {t}')
 		self.patch(yahoo, 'price_series', no_history)
 		self.sent.clear(); self.errors.clear()
@@ -121,34 +139,42 @@ def no_history(t):
 		self.assertEqual(len(self.errors), 1)
 
 	def test_history_without_a_ticker_or_with_extra_argument_gives_usage(self):
+		""".history with no ticker, or with an extra argument, gets its usage, not the profile of a ticker called
+		HISTORY."""
 		self.assertEqual(self.command('.history'), [('text', ['Usage: .history SYMBOL'])]) # not the profile of a ticker called HISTORY
 		self.assertEqual(self.command('.history T1 1y'), [('text', ['Usage: .history SYMBOL'])])
 
 
 class PortfolioNames(FinbotCase):
 	def test_an_unknown_portfolio_gets_the_same_answer_everywhere(self):
+		"""An unknown portfolio name gets the same answer from .trades, .performance, .holdings and .pie: the portfolios
+		there are."""
 		for cmd in ('.trades 7d Nope', '.performance Nope', '.holdings Nope', '.pie Nope'):
 			with self.subTest(cmd=cmd): # it used to be a ValueError, "no Sharesight data found", or two wordings of this
 				self.assertEqual(self.command(cmd), [('text', ['Nope portfolio not found. I only know about:', 'Rob'])])
 
 	def test_a_portfolio_name_in_any_case(self):
+		"""A portfolio's name is matched in any case."""
 		holding = {'code': 'T1', 'market_code': 'NASDAQ', 'name': 'Co T1', 'holding_id': 1}
 		self.patch(sharesight.Portfolio, 'holdings', lambda portfolio: {'T1:NASDAQ': holding})
 		self.assertIn('Holdings for Rob (1)', str(self.command('.holdings rob')))
 
 	def test_trades_with_no_buy_or_sell_still_get_an_answer(self):
+		""".trades with only dividends says there were no trades (it used to say nothing)."""
 		self.patch(sharesight, 'get_trades', lambda name, pid, days=7: [{'id': 2, 'transaction_type': 'DIVIDEND', 'portfolio': name}])
 		self.assertIn('No trades', str(self.command('.trades 7d'))) # it used to say nothing at all
 
 
 class CommandArguments(FinbotCase):
 	def test_period_and_portfolio_in_either_order(self):
+		"""A period and a portfolio name, of several words too, come in either order, and either may be left out."""
 		for text, expected in (('', (30, None)), ('7d', (7, None)), ('Rob', (30, 'Rob')), ('7d Rob', (7, 'Rob')), ('Rob 7d', (7, 'Rob')),
 							   ('My Portfolio', (30, 'My Portfolio')), ('My Portfolio 3m', (90, 'My Portfolio')), ('Rob ', (30, 'Rob'))):
 			with self.subTest(text=text):
 				self.assertEqual(worker.period_and_portfolio(text, 30), expected)
 
 	def test_a_percentage_is_a_number_and_nan_a_ticker(self):
+		"""A number, with or without %, is a threshold; nan and inf are tickers."""
 		for arg, expected in (('', (5, None)), ('2.5', (2.5, None)), ('3%', (3.0, None)), ('T1', (5, 'T1')), ('nan', (5, 'NAN')), ('inf', (5, 'INF'))):
 			with self.subTest(arg=arg):
 				self.assertEqual(worker.percent_or_ticker(arg, 5), expected)
@@ -161,13 +187,16 @@ def test_a_bare_number_on_price_is_a_percent(self):
 		self.assertIn('Price movement 5%', str(self.command('.price 7d')))
 
 	def test_top_is_gone(self):
+		""".top is a ticker now, like any word: the command is .price top."""
 		self.assertIn('Co TOP', str(self.command('.top 7d'))) # .price top 7d is the command; .top is a ticker now, like any word
 
 	def test_a_ratio_command_reads_its_ticker_as_the_other_commands_do(self):
+		""".pe t1 reads its ticker in any case, and a trailing space after top is fine."""
 		self.assertIn('Co T1 ', str(self.command('.pe t1')))
 		self.assertIn('Top 15 tracked stocks by pe ratio', str(self.command('.pe top '))) # trailing space and all
 
 	def test_hello_leaves_the_word_list_as_it_was(self):
+		""".hello greets with a verb (not a tuple of them) and leaves the word list alone; 'hi' needs the bot's name."""
 		before = list(words.verb)
 		for _ in range(30):
 			line = self.command('.hello')[0][1][0]
@@ -181,6 +210,7 @@ class WsgiRejections(FinbotCase):
 	SECRET = {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'tok'}
 
 	def request(self, path='/telegram', body=None, headers=None):
+		"""(the body, the status) the WSGI handler answers a request with."""
 		body = body if isinstance(body, bytes) else json.dumps(body if body is not None else {}).encode()
 		env = {'PATH_INFO': path, 'wsgi.input': io.BytesIO(body), **(headers or {})}
 		status = []
@@ -189,21 +219,26 @@ def request(self, path='/telegram', body=None, headers=None):
 		return b''.join(out), status[-1]
 
 	def test_telegram_requires_the_secret_header(self):
+		"""A Telegram update without the secret header, or with a wrong one, is refused and not processed."""
 		message = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.help'}}
 		self.assertIn(b'Unauthorized', self.request(body=message)[0])
 		self.assertIn(b'Unauthorized', self.request(body=message, headers={'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'wrong'})[0])
 		self.assertEqual(self.sent, []) # nothing was processed
 
 	def test_unknown_path_is_404(self):
+		"""An unknown path is 404."""
 		self.assertEqual(self.request(path='/nope')[1], '404 Not Found')
 
 	def test_malformed_body_is_refused_not_a_crash(self):
+		"""A body that is not a JSON object is 400 once authenticated, and turned away unread before that."""
 		self.assertEqual(self.request(body=b'not json')[1], '403 Forbidden') # no secret: turned away before the body is read
 		for body in (b'not json', b'', b'[1, 2]', b'7'):
 			self.assertEqual(self.request(body=body, headers=self.SECRET)[1], '400 Bad Request', body)
 		self.assertEqual(self.errors, [])
 
 	def test_strangers_cannot_post_into_the_chats(self):
+		"""Requests from strangers (wrong paths, no or wrong secrets, malformed bodies) are refused, and nothing reaches
+		the chats."""
 		self.env(slackOutgoingWebhook='https://example.com/slack')
 
 		self.env(slackOutgoingToken='stok')
@@ -218,7 +253,9 @@ def test_strangers_cannot_post_into_the_chats(self):
 		self.assertEqual(self.sent, [])
 
 	def test_a_failure_before_authentication_is_only_logged(self):
+		"""A failure while routing, before authentication, is logged and answered 500, never posted to the chats."""
 		def broken(*a, **k):
+			"""Fail while routing."""
 			raise RuntimeError('boom')
 		self.patch(bot, 'urlparse', broken) # fails while routing, before the secret is even looked at
 		env = {'PATH_INFO': '/telegram', 'wsgi.input': io.BytesIO(b'{}'), **self.SECRET}
@@ -229,12 +266,15 @@ def broken(*a, **k):
 		self.assertIn('boom', err.getvalue())
 
 	def test_a_failure_after_authentication_is_reported_once_and_not_retried(self):
+		"""A failure on an authenticated update is reported once and answered 200, so Telegram does not send it
+		again."""
 		broken = {'message': {'message_id': 1, 'from': {'first_name': 'A'}, 'text': '.help'}} # no chat
 		self.assertEqual(self.request(body=broken, headers=self.SECRET)[1], '200 OK') # a retry would only fail and report again
 		self.assertEqual(len(self.errors), 1)
 		self.assertIn("KeyError: 'chat'", self.errors[0])
 
 	def test_telegram_edge_updates(self):
+		"""Edits, button presses without a message, and messages without text are answered and left alone."""
 		secret = self.SECRET
 		self.assertEqual(self.request(body={'edited_message': {}}, headers=secret)[0], b'Unsupported')
 		self.assertEqual(self.request(body={'callback_query': {'id': '1'}}, headers=secret)[0], b'Unsupported')
@@ -242,6 +282,7 @@ def test_telegram_edge_updates(self):
 		self.assertIn(b'Unhandled', self.request(body=odd, headers=secret)[0])
 
 	def test_a_telegram_update_sent_again_runs_once(self):
+		"""An update Telegram sends again runs once."""
 		calls = []
 		self.patch(worker, 'process_request', lambda *a: calls.append(a))
 		body = {'update_id': 7, 'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.price'}}
@@ -251,6 +292,7 @@ def test_a_telegram_update_sent_again_runs_once(self):
 		self.assertEqual(len(calls), 1)
 
 	def test_button_presses_are_logged(self):
+		"""A button press is logged with who pressed it (handle, first name or unknown), the chat and the button."""
 		calls = []
 		self.patch(worker, 'process_callback', lambda *a: calls.append(a))
 		secret = {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'tok'}
@@ -266,6 +308,7 @@ class SlackEvents(FinbotCase):
 	"""Slack sends every message in a subscribed channel, the bot's own posts included."""
 
 	def setUp(self):
+		"""Slack commands configured, and the worker replaced by a list of what it would have run."""
 		super().setUp()
 		self.env(slackOutgoingWebhook='https://example.com/slack')
 
@@ -275,6 +318,7 @@ def setUp(self):
 		self.patch(worker, 'process_request', lambda *a: self.requests.append(a))
 
 	def event(self, event_id=None, **event):
+		"""Post a Slack message event to the WSGI handler and return the status."""
 		body = json.dumps({'token': 'stok', 'type': 'event_callback', 'authorizations': [{'user_id': 'UBOT'}], 'event_id': event_id,
 			'event': dict({'type': 'message', 'channel': 'C1', 'ts': '1.2'}, **event)}).encode()
 		status = []
@@ -284,6 +328,8 @@ def event(self, event_id=None, **event):
 		return status[-1]
 
 	def test_only_people_are_answered(self):
+		"""Messages from bots, this bot included, joins, edits and events without a user are not answered; people's
+		messages are, with mentions turned into @names."""
 		for event in ({'subtype': 'bot_message', 'bot_id': 'B1', 'text': "⚠️ /slack: KeyError: 'user'"}, # a webhook report
 					  {'user': 'UBOT', 'bot_id': 'B2', 'text': '.holdings: Please try again'}, # this bot's own reply
 					  {'user': 'UBOT', 'text': '.holdings'},
@@ -299,12 +345,14 @@ def test_only_people_are_answered(self):
 						 [('slack', 'C1', '<@U1>', '.price @UBOT', '@UBOT'), ('slack', 'C1', '<@U2>', '.trades', '@UBOT')])
 
 	def test_a_retried_event_runs_once(self):
+		"""An event Slack sends again (same event_id) runs once."""
 		for _ in range(2): # Slack resends an event it got no answer to within 3 seconds, with the same event_id
 			self.assertEqual(self.event('Ev1', user='U1', text='.price'), '200 OK')
 		self.assertEqual(self.event('Ev2', user='U1', text='.price'), '200 OK')
 		self.assertEqual(len(self.requests), 2)
 
 	def test_slack_does_not_change_the_telegram_bot_name(self):
+		"""A Slack event leaves the Telegram bot's name alone."""
 		self.patch(bot, 'botName', '@finbot_tg')
 		self.event(user='U1', text='.price')
 		self.assertEqual(bot.botName, '@finbot_tg')
@@ -315,6 +363,7 @@ class UserTextIsEscaped(FinbotCase):
 	refuse that reply and every .plan reply after it."""
 
 	def test_a_plan_is_shown_as_text(self):
+		"""A .plan entry is shown as text on Telegram and on Slack, escaped once, and a Slack mention stays markup."""
 		reply = self.command('.plan buy <cheap> & hold')[0][1]
 		self.assertIn('<i>buy &lt;cheap&gt; &amp; hold</i>', reply[0])
 		self.assertEqual(self.command('.plan')[0][1], reply) # and the same way again
@@ -326,6 +375,7 @@ def test_a_plan_is_shown_as_text(self):
 		self.assertIn('_buy &lt;cheap&gt; &amp; hold_', text)
 
 	def test_a_telegram_name_is_escaped_where_it_arrives(self):
+		"""A Telegram first name is escaped as the update arrives, in the handle and the name."""
 		requests = []
 		self.patch(worker, 'process_request', lambda *a: requests.append(a))
 		body = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A<b>&'}, 'text': '.hello'}}
@@ -341,13 +391,16 @@ class SecretsStayOutOfLogs(FinbotCase):
 	SECRET_URL = 'https://api.telegram.org/bot123456:SECRET-tok_en/sendMessage?chat_id=1'
 
 	def stderr_of(self, call):
+		"""What call() writes to stderr."""
 		err = io.StringIO()
 		with contextlib.redirect_stderr(err), contextlib.redirect_stdout(io.StringIO()):
 			call()
 		return err.getvalue()
 
 	def test_a_failed_send_logs_no_token(self):
+		"""A failed send logs no Telegram token, Slack webhook path or bot token."""
 		def refuse(url, **kwargs):
+			"""Fail like requests does, the URL and headers in the message."""
 			raise ConnectionError(f"Max retries exceeded with url: {url} (headers {kwargs.get('headers')})")
 		self.patch(webhook.net, 'post', refuse)
 		self.env(slackBotToken='xoxb-SECRET')
@@ -359,6 +412,7 @@ def refuse(url, **kwargs):
 			self.assertNotIn(secret, log)
 
 	def test_the_traceback_of_a_reported_error_is_redacted(self):
+		"""The traceback logged with a reported error is redacted."""
 		self.patch(webhook, 'report_error', REAL_REPORT_ERROR)
 		self.patch(webhook, 'write', lambda *a, **k: None)
 		try:
@@ -386,6 +440,7 @@ class StartupSignals(unittest.TestCase):
 	"""A reload (SIGHUP) that arrives while the bot is still starting must not kill it."""
 
 	def test_sighup_during_startup_is_ignored_and_works_afterwards(self):
+		"""A SIGHUP while bot.py starts is ignored, where its default action would kill it; once serving, it reloads."""
 		skip_if_old_env(self)
 		with tempfile.TemporaryDirectory() as data:
 			env = {k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k not in config.STARTUP_KEYS}
@@ -414,6 +469,8 @@ def test_sighup_during_startup_is_ignored_and_works_afterwards(self):
 			self.assertIn('reload requested but', ''.join(lines))
 
 	def test_usr1_flushes_the_cache_and_keeps_the_state(self):
+		"""SIGUSR1 is ignored during startup; once serving, it deletes the cache files but not a write in flight,
+		matplotlib's cache or the state."""
 		skip_if_old_env(self)
 		with tempfile.TemporaryDirectory() as data:
 			cache = Path(data, 'cache')
diff --git a/tests/test_finbot.py b/tests/test_finbot.py
index a7cadc60..c3253a52 100644
--- a/tests/test_finbot.py
+++ b/tests/test_finbot.py
@@ -7,13 +7,17 @@
 
 class Plumbing(FinbotCase):
 	def test_chunked_send_keeps_keyboard_on_last_chunk(self):
+		"""A long report goes out in several messages, and only the last carries the keyboard."""
 		posted = []
 
 		class Response:
 			status_code = 200
-			def json(self): return {'ok': True, 'result': {'message_id': len(posted), 'chat': {'id': 5}}}
+			def json(self):
+				"""Telegram's answer to a message posted."""
+				return {'ok': True, 'result': {'message_id': len(posted), 'chat': {'id': 5}}}
 
 		def post(url, **kwargs):
+			"""Record what was posted and answer like Telegram."""
 			posted.append(kwargs['json'])
 			return Response()
 
@@ -26,6 +30,8 @@ def post(url, **kwargs):
 		self.assertTrue(all('reply_markup' not in p for p in posted[:-1]))
 
 	def test_chunks_respect_both_the_line_count_and_the_size(self):
+		"""A message ends at chunk_maxlines lines or the size limit, whichever comes first; a line too long on its own
+		is cut short."""
 		chunks = webhook.chunked(['a' * 30] * 10 + ['b' * 500], limit=100, max_lines=4)
 		self.assertEqual([len(chunk) for chunk in chunks], [3, 3, 3, 1, 1]) # 3 x 30 plus newlines is 92; a fourth would pass 100
 		self.assertTrue(all(len('\n'.join(chunk)) <= 100 for chunk in chunks))
@@ -33,15 +39,19 @@ def test_chunks_respect_both_the_line_count_and_the_size(self):
 		self.assertEqual(webhook.chunked(['x'] * 9, limit=4096, max_lines=4), [['x'] * 4, ['x'] * 4, ['x']])
 
 	def test_zoneinfo_offset_is_not_lmt(self):
+		"""zoneinfo gives Melbourne its real offset, not the local mean time pytz once gave for old dates."""
 		tz = zoneinfo.ZoneInfo('Australia/Melbourne')
 		self.assertEqual(datetime.datetime(2022, 8, 18, tzinfo=tz).utcoffset(), datetime.timedelta(hours=10))
 
 	def test_reminder_runs(self):
+		"""The reminder report runs with no reminders on file."""
 		with mock.patch.object(config, 'data_json', lambda name, default=None: []):
 			reminder.lambda_handler()
 
 	def test_bot_handler(self):
+		"""A sticker is answered 200 and not logged; git_version matches what git says."""
 		def post(message):
+			"""Post a Telegram message update to the WSGI handler: (its answer, what it logged)."""
 			body = json.dumps({'update_id': 1, 'message': dict({'message_id': 5, 'chat': {'id': 55, 'type': 'private'},
 				'from': {'id': 9, 'first_name': 'Rob'}, 'date': 1}, **message)}).encode()
 			env = {'PATH_INFO': '/telegram', 'wsgi.input': io.BytesIO(body), 'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'tok'}
@@ -57,9 +67,10 @@ def post(message):
 
 class Conventions(unittest.TestCase):
 	def test_every_function_has_a_docstring(self):
-		"""Every function and method in the bot, lib/ and util/, nested ones included, says what it does."""
+		"""Every function and method, in the tests too and nested ones included, says what it does."""
 		missing = []
-		code = [*Path(ROOT, 'lib').rglob('*.py'), *Path(ROOT).glob('*.py'), *Path(ROOT, 'util').glob('*.py')]
+		code = [*Path(ROOT, 'lib').rglob('*.py'), *Path(ROOT).glob('*.py'), *Path(ROOT, 'util').glob('*.py'),
+			*Path(ROOT, 'tests').glob('*.py')]
 		for path in sorted(code):
 			for node in ast.walk(ast.parse(path.read_text())):
 				if isinstance(node, (ast.FunctionDef, ast.AsyncFunctionDef)) and ast.get_docstring(node) is None:
@@ -82,6 +93,7 @@ class GitVersion(unittest.TestCase):
 	SHA = '0123456789abcdef0123456789abcdef01234567'
 
 	def repo(self, tmp, head, refs=None, packed=None):
+		"""A fake .git in tmp: HEAD, loose refs and packed-refs as given."""
 		git = Path(tmp, '.git')
 		git.mkdir(parents=True)
 		(git / 'HEAD').write_text(head + '\n')
@@ -93,6 +105,7 @@ def repo(self, tmp, head, refs=None, packed=None):
 		return git
 
 	def test_loose_ref_detached_and_packed(self):
+		"""The commit is found from a loose ref, a detached HEAD, and a ref only in packed-refs."""
 		with tempfile.TemporaryDirectory() as a, tempfile.TemporaryDirectory() as b, tempfile.TemporaryDirectory() as c:
 			self.repo(a, 'ref: refs/heads/work', {'refs/heads/work': self.SHA})
 			self.repo(b, self.SHA)
@@ -100,6 +113,7 @@ def test_loose_ref_detached_and_packed(self):
 			self.assertEqual([bot.git_version(d) for d in (a, b, c)], ['0123456'] * 3)
 
 	def test_linked_worktree(self):
+		"""A linked worktree, whose .git is a file naming its gitdir, finds its refs in the main repository."""
 		with tempfile.TemporaryDirectory() as main, tempfile.TemporaryDirectory() as tree:
 			git = self.repo(main, 'ref: refs/heads/main', {'refs/heads/main': self.SHA})
 			linked = git / 'worktrees' / 'w'
@@ -110,6 +124,7 @@ def test_linked_worktree(self):
 			self.assertEqual(bot.git_version(tree), '0123456')
 
 	def test_unreadable_is_unknown(self):
+		"""No .git, a dangling ref, or a garbled HEAD all give 'unknown'."""
 		with tempfile.TemporaryDirectory() as empty, tempfile.TemporaryDirectory() as odd:
 			self.repo(odd, 'ref: refs/heads/gone')
 			self.assertEqual([bot.git_version(empty), bot.git_version(odd), bot.git_version('/nonexistent')], ['unknown'] * 3)
@@ -117,6 +132,7 @@ def test_unreadable_is_unknown(self):
 			self.assertEqual(bot.git_version(empty), 'unknown')
 
 	def test_does_not_run_git(self):
+		"""git_version reads .git itself: no git binary is run."""
 		with mock.patch('subprocess.run', side_effect=AssertionError('git was run')):
 			self.assertGreaterEqual(len(bot.git_version()), 7)
 
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
index 86e13b9b..4e33cb3d 100644
--- a/tests/test_jobs.py
+++ b/tests/test_jobs.py
@@ -12,19 +12,23 @@
 
 
 def at(day, hour, minute, second=0):
-	return datetime.datetime(2026, 10, day, hour, minute, second, tzinfo=UTC) # 13 Oct 2026 is a Tuesday
+	"""A UTC time in October 2026 (the 13th is a Tuesday)."""
+	return datetime.datetime(2026, 10, day, hour, minute, second, tzinfo=UTC)
 
 
 class Clock:
 	def __init__(self, now):
+		"""now: the time the clock shows until a test moves it."""
 		self.now = now
 
 	def __call__(self):
+		"""The time the clock shows."""
 		return self.now
 
 
 class ParseTests(unittest.TestCase):
 	def test_shipped_schedule_has_no_errors(self):
+		"""defaults/schedule.cron parses without errors, its Saturday line into its five commands."""
 		parsed, errors = jobs.parse_schedule(DEFAULT_SCHEDULE.read_text(), UTC)
 		self.assertEqual(errors, [])
 		self.assertEqual(len(parsed), 9)
@@ -33,6 +37,7 @@ def test_shipped_schedule_has_no_errors(self):
 						 ['cal.py earnings', 'cal.py ex-dividend', 'price.py 7', 'performance.py 7', 'rating.py'])
 
 	def test_pasted_crontab_lines(self):
+		"""Lines pasted from a crontab work: variables, paths, redirects, braces, comments and @daily."""
 		text = ('MAILTO=me\n'
 				'*/20 * * * Mon-Fri ~/finbot/trades.py > ~/finbot.trades 2>&1\n'
 				'28  21 * * Sat { ./cal.py earnings; ./price.py 7 ;} &> /dev/null # weekly (UTC)\n'
@@ -43,6 +48,7 @@ def test_pasted_crontab_lines(self):
 						 [['trades.py'], ['cal.py earnings', 'price.py 7'], ['reminder.py']])
 
 	def test_quoted_arguments_may_hold_anything(self):
+		"""A quoted argument may hold shell characters; commands after a ; still split."""
 		url = 'https://example.com/feed?a=1&b=2;c>d $x |y'
 		parsed, errors = jobs.parse_schedule(f'01 0-10,23 * * Mon-Fri podcasts.py "{url}" > ~/out 2>&1 # hourly (UTC)\n', UTC)
 		self.assertEqual(errors, [])
@@ -51,12 +57,14 @@ def test_quoted_arguments_may_hold_anything(self):
 		self.assertEqual([(label[:9], kw) for label, _, kw in parsed[0].commands], [('podcasts.', {'feed_url': url}), ('shorts.py', {})])
 
 	def test_unquoted_shell_syntax_is_still_refused(self):
+		"""Pipes, &&, a trailing &, $(), backticks, input redirects, subshells and an unterminated quote are refused."""
 		for line in ('0 1 * * * trades.py | tee x', '0 1 * * * trades.py && shorts.py & ', '0 1 * * * trades.py $(date)', '0 1 * * * trades.py `x`',
 					 '0 1 * * * trades.py < in', '0 1 * * * (trades.py)', '0 1 * * * trades.py "unterminated'):
 			parsed, errors = jobs.parse_schedule(line + '\n', UTC)
 			self.assertEqual((parsed, len(errors)), ([], 1), line)
 
 	def test_redirects_and_braces_are_dropped(self):
+		"""Output redirects, braces, paths and trailing comments are dropped from a command."""
 		for command in ('trades.py > /dev/null', 'trades.py >/dev/null 2>&1', 'trades.py &> /dev/null', 'trades.py >> ~/x 2>&1',
 						'{ trades.py ;} > ~/x', '/opt/finbot/trades.py # note (x)'):
 			parsed, errors = jobs.parse_schedule(f'0 1 * * * {command}\n', UTC)
@@ -64,17 +72,21 @@ def test_redirects_and_braces_are_dropped(self):
 			self.assertEqual([label for label, _, _ in parsed[0].commands], ['trades.py'], command)
 
 	def test_bin_run_wrapper_is_no_longer_accepted(self):
+		"""The old bin/run wrapper and a cd before the command are refused."""
 		for command in ('bin/run trades.py', 'cd /opt/finbot; trades.py'):
 			parsed, errors = jobs.parse_schedule(f'0 1 * * * {command}\n', UTC)
 			self.assertEqual((parsed, len(errors)), ([], 1), command)
 
 	def test_bad_lines_are_skipped_and_named(self):
+		"""Bad lines are skipped and reported by line number; the good ones load."""
 		text = '0 1 * * * ls | wc -l\n0 1 * * * price.py bogus\n0 1 * * * nothere.py\n61 1 * * * trades.py\n0 1 * *\nFOO=bar\n5 4 * * * shorts.py\n'
 		parsed, errors = jobs.parse_schedule(text, UTC)
 		self.assertEqual([j.expr for j in parsed], ['5 4 * * *'])
 		self.assertEqual([number for number, _, _ in errors], [1, 2, 3, 4, 5, 6])
 
 	def test_impossible_times_are_marked_and_reboot_is_just_unsupported(self):
+		"""Times that never come (30 February, minute 61, month 13, @bogus) are ImpossibleTime; @reboot is only
+		unsupported, and 29 February does come."""
 		for line in ('0 0 30 2 * trades.py', '0 9 31 4,6,9,11 * trades.py', '61 1 * * * trades.py', '0 0 * 13 * trades.py', '@bogus trades.py'):
 			parsed, errors = jobs.parse_schedule(line + '\n', UTC)
 			self.assertEqual(parsed, [], line)
@@ -84,16 +96,20 @@ def test_impossible_times_are_marked_and_reboot_is_just_unsupported(self):
 		self.assertNotIsInstance(errors[0][2], jobs.ImpossibleTime)
 
 	def test_cron_tz_applies_to_the_lines_below(self):
+		"""CRON_TZ sets the zone of the lines below it; an unknown zone is skipped."""
 		parsed, _ = jobs.parse_schedule('1 1 * * * trades.py\nCRON_TZ=Australia/Melbourne\n2 2 * * * trades.py\nCRON_TZ=Nowhere/Land\n', UTC)
 		self.assertEqual([j.zone.key for j in parsed], ['UTC', 'Australia/Melbourne'])
 
 	def test_local_zone_follows_tz(self):
+		"""The local zone follows TZ."""
 		with mock.patch.dict(os.environ, {'TZ': 'Australia/Melbourne'}):
 			self.assertEqual(jobs.local_zone().key, 'Australia/Melbourne')
 
 
 class JobArguments(unittest.TestCase):
 	def test_price(self):
+		"""price.py takes one mode or a number of days (the top and bottom ten over that span); anything else is its
+		usage."""
 		self.assertEqual(price.job(['midsession']), ('price.py midsession', {'midsession': True}))
 		self.assertEqual(price.job(['premarket'])[1], {'premarket': True})
 		self.assertEqual(price.job(['interday'])[1], {'interday': True})
@@ -104,6 +120,7 @@ def test_price(self):
 				price.job(bad)
 
 	def test_cal_and_performance(self):
+		"""cal.py takes earnings or ex-dividend; performance.py an optional number of days."""
 		self.assertEqual(cal.job(['earnings']), ('cal.py earnings', {'event': 'earnings'}))
 		self.assertEqual(cal.job(['ex-dividend']), ('cal.py ex-dividend', {'event': 'ex-dividend'}))
 		self.assertEqual(performance.job([]), ('performance.py', {}))
@@ -113,6 +130,7 @@ def test_cal_and_performance(self):
 				module.job(bad)
 
 	def test_podcasts(self):
+		"""podcasts.py takes one feed URL; its label carries the start of the hash that names its state file."""
 		url = 'https://example.com/podcast.rss'
 		label, kwargs = podcasts.job([url])
 		self.assertEqual(kwargs, {'feed_url': url})
@@ -123,6 +141,7 @@ def test_podcasts(self):
 				podcasts.job(bad)
 
 	def test_cleanup_arguments(self):
+		"""cleanup.py takes an optional number of days."""
 		self.assertEqual(cleanup.job([]), ('cleanup.py', {}))
 		self.assertEqual(cleanup.job(['7']), ('cleanup.py 7', {'days': 7}))
 		for bad in (['x'], ['1', '2']):
@@ -130,17 +149,20 @@ def test_cleanup_arguments(self):
 				cleanup.job(bad)
 
 	def test_plain_scripts(self):
+		"""The other reports take no arguments."""
 		for name, module in jobs.SCRIPTS.items():
 			if name not in ('cal', 'cleanup', 'performance', 'podcasts', 'price'):
 				self.assertEqual(module.job([]), (name + '.py', {}))
 
 	def test_cli_prints_usage_and_exits(self):
+		"""A bad argument on the command line prints the usage and exits 1."""
 		with mock.patch('sys.stderr', new_callable=io.StringIO) as err, self.assertRaises(SystemExit) as exit_:
 			webhook.cli('price.py', price.job, lambda **k: None, ['bogus'])
 		self.assertEqual(exit_.exception.code, 1)
 		self.assertIn('Usage: price.py [midsession', err.getvalue())
 
 	def test_cli_runs_the_handler(self):
+		"""Good arguments run the report with what job() made of them."""
 		calls = []
 		webhook.cli('price.py', price.job, lambda **k: calls.append(k), ['midsession'])
 		self.assertEqual(calls, [{'midsession': True}])
@@ -160,6 +182,8 @@ def test_each_scheduled_report_runs_from_its_top_level_script(self):
 
 class CleanupJob(FinbotCase):
 	def test_removes_only_old_finbot_cache_files(self):
+		"""cleanup deletes finbot's cache files older than the days given, not newer ones, other files or
+		directories."""
 		folder = tempfile.mkdtemp()
 		self.addCleanup(lambda: __import__('shutil').rmtree(folder, ignore_errors=True))
 		self.env(cache_dir=folder)
@@ -175,6 +199,8 @@ def test_removes_only_old_finbot_cache_files(self):
 		self.assertEqual(sorted(os.listdir(folder)), ['finbot_dir', 'finbot_new.json', 'other_old.json'])
 
 	def test_a_temp_file_being_written_is_left_alone(self):
+		"""A flush (0 days) leaves a temp file written in the last minute, an atomic write in flight, and deletes a
+		stale one."""
 		folder = tempfile.mkdtemp()
 		self.addCleanup(lambda: __import__('shutil').rmtree(folder, ignore_errors=True))
 		self.env(cache_dir=folder)
@@ -186,6 +212,7 @@ def test_a_temp_file_being_written_is_left_alone(self):
 		self.assertEqual(os.listdir(folder), ['finbot_a.json.x1.tmp'])
 
 	def test_missing_cache_dir_is_fine(self):
+		"""No cache dir: nothing to delete, no error."""
 		self.env(cache_dir='/nonexistent/finbot-cache')
 
 		self.assertEqual(cleanup.lambda_handler(30), 0)
@@ -193,6 +220,7 @@ def test_missing_cache_dir_is_fine(self):
 
 class SchedulerFlag(unittest.TestCase):
 	def flag(self, **env):
+		"""config.env.scheduler in a fresh interpreter, given `env`."""
 		code = 'from lib import config; print(config.env.scheduler)'
 		e = {k: v for k, v in os.environ.items() if k != 'scheduler'}
 		e.update(env)
@@ -200,6 +228,7 @@ def flag(self, **env):
 		return subprocess.run([sys.executable, '-c', code], cwd=repo, env=e, capture_output=True, text=True, check=True).stdout.split()[-1]
 
 	def test_on_by_default_and_off_for_false(self):
+		"""The scheduler is on by default and with any true word, off with any false one."""
 		self.assertEqual(self.flag(), 'True')
 		for value in ('True', 'true', '1', 't'):
 			self.assertEqual(self.flag(scheduler=value), 'True', value)
@@ -209,6 +238,8 @@ def test_on_by_default_and_off_for_false(self):
 
 class SchedulerBase(FinbotCase):
 	def setUp(self):
+		"""A schedule of trades.py every 20 minutes on weekdays, a fake clock, and trades and shorts recorded instead of
+		run."""
 		super().setUp()
 		tmp = tempfile.TemporaryDirectory()
 		self.addCleanup(tmp.cleanup)
@@ -220,6 +251,7 @@ def setUp(self):
 		self.patch(jobs.SCRIPTS['shorts'], 'lambda_handler', lambda **k: self.ran.append(('shorts', k)))
 
 	def scheduler(self, catchup=5):
+		"""A loaded scheduler on the fake clock that runs commands in this process, at once."""
 		s = jobs.Scheduler(lambda: self.file, catchup=catchup, clock=self.clock, sleep=lambda seconds: None,
 						   spawn=lambda f, *a: f(*a), zone=UTC, runner=jobs.run_in_process)
 		self.assertTrue(s.load())
@@ -228,6 +260,7 @@ def scheduler(self, catchup=5):
 
 class SchedulerTests(SchedulerBase):
 	def test_fires_on_the_minute_on_weekdays_only(self):
+		"""A job fires on its minute, once, on weekdays only; runs missed over days coalesce into one."""
 		s = self.scheduler()
 		self.clock.now = at(13, 21, 19, 59)
 		self.assertEqual(s.due(self.clock.now), [])
@@ -242,6 +275,7 @@ def test_fires_on_the_minute_on_weekdays_only(self):
 		self.assertEqual(s.due(self.clock.now), [])
 
 	def test_tick_runs_the_job_and_returns_the_wait(self):
+		"""tick() runs what is due and returns how long to sleep: up to 30 seconds, less when the next run is nearer."""
 		s = self.scheduler()
 		self.clock.now = at(13, 21, 20)
 		wait = s.tick()
@@ -251,6 +285,7 @@ def test_tick_runs_the_job_and_returns_the_wait(self):
 		self.assertEqual(s.tick(), 10)
 
 	def test_a_run_missed_within_the_catch_up_window_fires_once(self):
+		"""After a restart, a run missed within the catch-up window fires once; an older one is skipped."""
 		first = self.scheduler()
 		first.last = {first.jobs[0].key: at(13, 21, 0).isoformat()}
 		state.save(jobs.STATE_FILE, first.last)
@@ -259,6 +294,7 @@ def test_a_run_missed_within_the_catch_up_window_fires_once(self):
 		self.assertEqual([t for _, t in s.due(self.clock.now)], [at(13, 21, 40)]) # 21:20 is older than 5 minutes: skipped
 
 	def test_a_corrupt_state_file_is_reported_and_the_schedule_starts_afresh(self):
+		"""A state file cut off mid-write is reported once, and the schedule runs as usual and writes it afresh."""
 		Path(config.env.data_dir, jobs.STATE_FILE).write_text('{"*/20 * * * Mon-Fri|UTC|trades.py": "2026-10-13T21:') # cut off mid-write
 		s = self.scheduler()
 		self.assertEqual((s.last, len(self.errors)), ({}, 1))
@@ -270,11 +306,13 @@ def test_a_corrupt_state_file_is_reported_and_the_schedule_starts_afresh(self):
 		self.assertEqual(len(self.errors), 1)
 
 	def test_no_record_means_nothing_is_missed(self):
+		"""With no record of past runs, nothing counts as missed."""
 		self.clock.now = at(13, 21, 43)
 		s = self.scheduler()
 		self.assertEqual(s.due(self.clock.now), [])
 
 	def test_a_restart_just_after_a_run_does_not_repeat_it(self):
+		"""A restart just after a run does not run it again."""
 		s = self.scheduler()
 		self.clock.now = at(13, 21, 20, 5)
 		self.assertEqual(len(s.due(self.clock.now)), 1)
@@ -283,6 +321,7 @@ def test_a_restart_just_after_a_run_does_not_repeat_it(self):
 		self.assertEqual(again.due(self.clock.now), [])
 
 	def test_an_overlapping_run_is_skipped(self):
+		"""A run due while the last is still going is skipped."""
 		s = self.scheduler()
 		s.jobs[0].running = True
 		s.run_job(s.jobs[0], at(13, 21, 20))
@@ -293,8 +332,10 @@ def test_an_overlapping_run_is_skipped(self):
 		self.assertFalse(s.jobs[0].running)
 
 	def test_a_failing_job_is_reported_and_the_next_command_still_runs(self):
+		"""A command that fails is reported, and the next command on its line still runs."""
 		self.file.write_text('*/20 * * * * trades.py; shorts.py\n')
 		def boom(**k):
+			"""Fail."""
 			raise RuntimeError('boom')
 		self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', boom)
 		s = self.scheduler()
@@ -303,7 +344,9 @@ def boom(**k):
 		self.assertTrue(any('boom' in e for e in self.errors))
 
 	def test_system_exit_inside_a_job_is_reported_not_raised(self):
+		"""A SystemExit inside a job is reported, not raised, and the job is marked done."""
 		def quit_(**k):
+			"""Exit, as a script calling sys.exit would."""
 			raise SystemExit(3)
 		self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', quit_)
 		s = self.scheduler()
@@ -312,6 +355,7 @@ def quit_(**k):
 		self.assertFalse(s.jobs[0].running)
 
 	def test_a_job_that_hangs_is_abandoned_and_reported(self):
+		"""A job that runs past the timeout is abandoned and reported."""
 		self.patch(jobs, 'JOB_TIMEOUT', 0.05)
 		self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', lambda **k: gevent.sleep(5))
 		s = self.scheduler()
@@ -320,6 +364,8 @@ def test_a_job_that_hangs_is_abandoned_and_reported(self):
 		self.assertFalse(s.jobs[0].running)
 
 	def test_reload_applies_edits_keeps_state_and_survives_a_bad_file(self):
+		"""A reload applies edits and keeps the timing of unchanged jobs; a file with nothing usable keeps the running
+		schedule."""
 		s = self.scheduler()
 		old_key = s.jobs[0].key
 		base = s.base[old_key]
@@ -332,6 +378,7 @@ def test_reload_applies_edits_keeps_state_and_survives_a_bad_file(self):
 		self.assertEqual(len(s.jobs), 2)
 
 	def test_an_impossible_time_refuses_the_whole_file(self):
+		"""A time that never comes refuses the whole file: a reload says why and the running schedule carries on."""
 		s = self.scheduler()
 		self.file.write_text('*/20 * * * Mon-Fri trades.py\n0 9 30 2 * shorts.py\n5 4 * * * nothere.py\n')
 		with self.assertRaises(jobs.ImpossibleTime) as caught:
@@ -347,15 +394,18 @@ def test_an_impossible_time_refuses_the_whole_file(self):
 		self.assertEqual(len(s.due(self.clock.now)), 1)
 
 	def test_load_reports_bad_lines_once_and_runs_the_rest(self):
+		"""A bad line is reported once and the rest of the schedule runs."""
 		self.file.write_text('0 1 * * * nothere.py\n*/20 * * * * trades.py\n')
 		s = self.scheduler()
 		self.assertEqual(len(s.jobs), 1)
 		self.assertEqual(len(self.errors), 1)
 
 	def test_run_forever_survives_an_error(self):
+		"""run_forever reports a failed tick and carries on."""
 		s = self.scheduler()
 		calls = []
 		def sleep(seconds):
+			"""Note the sleep, and stop the loop on the third."""
 			calls.append(seconds)
 			if len(calls) == 3:
 				raise KeyboardInterrupt
@@ -373,6 +423,7 @@ class DaylightSaving(unittest.TestCase):
 	MEL = zoneinfo.ZoneInfo('Australia/Melbourne')
 
 	def fires(self, expr, start, end):
+		"""When a job fires between start and end, as Melbourne times ('04 09:00 AEDT')."""
 		job = jobs.parse_schedule(f'{expr} trades.py\n', self.MEL)[0][0]
 		s = jobs.Scheduler(lambda: None, zone=self.MEL)
 		times, t = [], start
@@ -381,13 +432,16 @@ def fires(self, expr, start, end):
 		return times
 
 	def mel(self, *args, **kw):
+		"""A Melbourne time."""
 		return datetime.datetime(*args, tzinfo=self.MEL, **kw)
 
 	def test_a_daily_job_fires_once_at_its_time(self):
+		"""A daily job fires at its local time each day across both changes."""
 		self.assertEqual(self.fires('0 9 * * *', self.mel(2026, 10, 2), self.mel(2026, 10, 6)), ['02 09:00 AEST', '03 09:00 AEST', '04 09:00 AEDT', '05 09:00 AEDT'])
 		self.assertEqual(self.fires('0 9 * * *', self.mel(2026, 4, 3), self.mel(2026, 4, 7)), ['03 09:00 AEDT', '04 09:00 AEDT', '05 09:00 AEST', '06 09:00 AEST'])
 
 	def test_a_skipped_or_repeated_time_fires_once(self):
+		"""A time the clocks skip fires an hour later, and a repeated one fires once."""
 		self.assertEqual(self.fires('30 2 * * *', self.mel(2026, 10, 3), self.mel(2026, 10, 6)), ['03 02:30 AEST', '04 03:30 AEDT', '05 02:30 AEDT'])
 		self.assertEqual(self.fires('30 2 * * *', self.mel(2026, 4, 4), self.mel(2026, 4, 7)), ['04 02:30 AEDT', '05 02:30 AEDT', '06 02:30 AEST'])
 		self.assertEqual(self.fires('*/20 * * * *', self.mel(2026, 10, 4, 1, 30), self.mel(2026, 10, 4, 3, 50)), ['04 01:40 AEST', '04 03:00 AEDT', '04 03:20 AEDT', '04 03:40 AEDT'])
@@ -395,11 +449,13 @@ def test_a_skipped_or_repeated_time_fires_once(self):
 						 ['05 01:40 AEDT', '05 02:00 AEDT', '05 02:20 AEDT', '05 02:40 AEDT', '05 03:00 AEST', '05 03:20 AEST'])
 
 	def test_starting_in_the_repeated_hour_runs_nothing_twice(self):
+		"""Starting during the repeated hour runs nothing twice."""
 		second_time_round = self.mel(2026, 4, 5, 2, 10, fold=1).astimezone(UTC)
 		self.assertEqual(self.fires('*/20 * * * *', second_time_round, self.mel(2026, 4, 5, 3, 0, 30)), ['05 03:00 AEST'])
 		self.assertEqual(self.fires('30 2 * * *', second_time_round, self.mel(2026, 4, 6, 12)), ['06 02:30 AEST'])
 
 	def test_the_scheduler_runs_a_daily_job_once_across_the_change(self):
+		"""The scheduler runs a daily Melbourne job once a day across the change to daylight time."""
 		file = Path(tempfile.mkdtemp(), 'schedule.cron')
 		file.write_text('CRON_TZ=Australia/Melbourne\n0 9 * * * trades.py\n')
 		clock = Clock(self.mel(2026, 10, 3, 12).astimezone(UTC))
@@ -415,6 +471,7 @@ def test_the_scheduler_runs_a_daily_job_once_across_the_change(self):
 
 class StartupRefusal(unittest.TestCase):
 	def test_the_bot_will_not_start_with_an_impossible_time(self):
+		"""bot.py will not start while the schedule holds a time that never comes, and says which."""
 		skip_if_old_env(self)
 		with tempfile.TemporaryDirectory() as data:
 			Path(data, 'schedule.cron').write_text('*/20 * * * Mon-Fri trades.py\n0 9 31 6 * price.py close\n')
@@ -427,6 +484,7 @@ def test_the_bot_will_not_start_with_an_impossible_time(self):
 
 class ManualRuns(SchedulerBase):
 	def setUp(self):
+		"""A schedule with repeats, comments, braces and the commands that cannot be picked (cleanup, podcasts)."""
 		super().setUp()
 		self.file.write_text('# a comment\n*/20 * * * Mon-Fri trades.py > /dev/null\n28 21 * * Sat price.py close; trades.py # weekly\n'
 							 '5 4 * * * { price.py close ;}\n6 4 * * * price.py 7\n'
@@ -437,21 +495,26 @@ def setUp(self):
 		self.patch(jobs.SCRIPTS['price'], 'lambda_handler', lambda **k: self.ran.append(('price', k)))
 
 	def test_commands_are_unique_clean_and_in_order(self):
+		"""The commands offered are each once, without redirects, sorted by name with numbers in numeric order."""
 		self.assertEqual([label for label, _, _ in self.s.commands()], ['performance.py 7', 'performance.py 30', 'performance.py 90', 'performance.py 180',
 																	   'performance.py 365', 'price.py 7', 'price.py close', 'trades.py'])
 
 	def test_select_by_number_or_command_text(self):
+		"""A command is picked by its number or its text, quotes, paths and redirects aside."""
 		self.assertEqual(self.s.select('7')[0], 'price.py close')
 		self.assertEqual(self.s.select('8')[0], 'trades.py')
 		self.assertEqual(self.s.select('  "price.py" close > /dev/null')[0], 'price.py close')
 		self.assertEqual(self.s.select('~/finbot/price.py 7')[0], 'price.py 7')
 
 	def test_select_refuses_what_is_not_offered(self):
+		"""Numbers out of range, cleanup, podcasts, commands not in the schedule, shell syntax and two commands are
+		refused."""
 		for text in ('0', '9', 'cleanup.py', 'podcasts.py "https://example.com/feed"', 'price.py 99', 'cal.py earnings', 'nonsense', 'ls | wc', '', 'price.py close; trades.py'):
 			with self.assertRaises(jobs.ScheduleError, msg=text):
 				self.s.select(text)
 
 	def test_run_now_runs_one_command_and_leaves_the_schedule_alone(self):
+		"""run_now runs one command and leaves the schedule's timing alone."""
 		base, last = dict(self.s.base), dict(self.s.last)
 		self.assertEqual(self.s.run_now('7'), ('price.py close', 'done'))
 		self.assertEqual(self.ran, [('price', {'close': True})])
@@ -460,11 +523,13 @@ def test_run_now_runs_one_command_and_leaves_the_schedule_alone(self):
 		self.assertEqual(self.s.busy, set())
 
 	def test_run_now_refuses_while_the_command_is_running_and_reports_failures(self):
+		"""run_now refuses a command that is running, and reports one that fails."""
 		self.s.busy.add('trades.py')
 		self.assertEqual(self.s.run_now('8'), ('trades.py', 'already running'))
 		self.assertEqual(self.ran, [])
 		self.s.busy.clear()
 		def boom(**k):
+			"""Fail."""
 			raise RuntimeError('boom')
 		self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', boom)
 		self.assertEqual(self.s.run_now('8'), ('trades.py', 'failed'))
@@ -472,20 +537,25 @@ def boom(**k):
 		self.assertEqual(self.s.busy, set())
 
 	def test_a_scheduled_run_skips_a_command_that_a_manual_run_is_in_the_middle_of(self):
+		"""A scheduled run skips a command a manual run is in the middle of, and runs the rest of its line."""
 		self.s.busy.add('trades.py')
 		self.assertEqual(self.s.run_job(self.s.jobs[1], at(13, 21, 28)), 'already running')
 		self.assertEqual(self.ran, [('price', {'close': True})]) # price.py close ran, trades.py was skipped
 
 	def test_run_now_times_out(self):
+		"""A manual run past the timeout says so."""
 		self.patch(jobs, 'JOB_TIMEOUT', 0.05)
 		self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', lambda **k: gevent.sleep(5))
 		self.assertEqual(self.s.run_now('8'), ('trades.py', 'timed out'))
 		self.assertEqual(self.s.busy, set())
 
 	def list_in_chat(self, text='.jobs'):
+		"""The lines .jobs (or `text`) answers with."""
 		return self.command(text)[0][1]
 
 	def test_list_in_chat_is_just_the_commands(self):
+		""".jobs and its aliases list the commands by number, without times, comments or redirects, and mark one that is
+		running."""
 		for alias in ('.jobs', '.cron', '.crons', '.schedule', '@' + BOT.lstrip('@') + ' jobs'):
 			text = '\n'.join(map(str, self.list_in_chat(alias)))
 			for line in ('1. performance 7', '2. performance 30', '5. performance 365', '6. price 7', '7. price close', '8. trades'):
@@ -496,6 +566,7 @@ def test_list_in_chat_is_just_the_commands(self):
 		self.assertIn('6. price 7 (running)', '\n'.join(map(str, self.list_in_chat())))
 
 	def test_telegram_gets_a_button_per_command_and_slack_does_not(self):
+		"""On Telegram each command gets a button whose data fits in 64 bytes; Slack gets none."""
 		payload, markup = about.prepare_jobs_payload(TELEGRAM, '')
 		buttons = [b for row in markup['inline_keyboard'] for b in row]
 		self.assertEqual([b['text'] for b in buttons], ['1. performance 7', '2. performance 30', '3. performance 90', '4. performance 180',
@@ -506,12 +577,14 @@ def test_telegram_gets_a_button_per_command_and_slack_does_not(self):
 		self.assertIsNone(markup)
 
 	def test_the_chat_command_passes_the_keyboard_on(self):
+		""".jobs sends its buttons with the list."""
 		seen = []
 		self.patch(webhook, 'payload_wrapper', lambda svc, url, payload, *a, **k: seen.append(k.get('reply_markup')) or [])
 		worker.process_request('telegram', '55', '@u', '.jobs', BOT, 'U')
 		self.assertEqual(len(seen[0]['inline_keyboard'][0]), 2)
 
 	def test_button_press_runs_that_command(self):
+		"""Pressing a command's button runs it, says so in a toast, and reports how it went."""
 		sent = []
 		self.patch(telegram, 'answerCallbackQuery', lambda cid, text=None: sent.append(text))
 		markup = about.prepare_jobs_payload(TELEGRAM, '')[1]
@@ -529,6 +602,8 @@ def stale_buttons(self):
 		self.patch(webhook, 'editMessageText', lambda chat, mid, text, markup=None: self.edits.append((chat, mid, text, markup)))
 
 	def test_forged_button_data_never_runs_anything_else(self):
+		"""Forged or malformed button data runs nothing and refreshes the buttons in place; cleanup and podcasts cannot
+		be picked by any number."""
 		self.stale_buttons()
 		good = jobs.checksum('trades.py')
 		for data in ('j|-1|' + good, 'j|0|' + good, 'j|9|' + good, 'j|8|0000', 'j|8|', 'j|8', 'j|8|' + good + '|x', 'j||' + good,
@@ -550,6 +625,8 @@ def test_forged_button_data_never_runs_anything_else(self):
 		self.assertEqual(self.ran, [('trades', {})])
 
 	def test_button_from_an_old_schedule_updates_the_message(self):
+		"""A button from an old schedule runs nothing, and the message gets the current list and buttons; with the
+		scheduler off, a notice."""
 		self.stale_buttons()
 		worker.process_callback('telegram', 'cb1', '55', '9', 'j|7|0000')
 		worker.process_callback('telegram', 'cb1', '55', '9', 'j|9|0000')
@@ -565,6 +642,7 @@ def test_button_from_an_old_schedule_updates_the_message(self):
 		self.assertEqual(self.edits[-1][2:], ('Scheduled jobs are off (scheduler=False)', {'inline_keyboard': []}))
 
 	def test_run_in_chat(self):
+		""".jobs run takes a number or a command's text."""
 		self.assertEqual(self.command('.jobs run 8'), [('text', ['8. trades: done'])])
 		self.assertEqual(self.ran, [('trades', {})])
 		self.ran.clear()
@@ -572,6 +650,8 @@ def test_run_in_chat(self):
 		self.assertEqual(self.ran, [('price', {'close': True})])
 
 	def test_run_in_chat_refusals(self):
+		""".jobs run refuses a command not in the schedule, an unknown number and two commands, and gives its usage
+		otherwise."""
 		self.assertIn('not in the schedule', self.list_in_chat('.jobs run cal.py earnings')[0])
 		self.assertIn('no command 99', self.list_in_chat('.jobs run 99')[0])
 		self.assertIn('one command at a time', self.list_in_chat('.jobs run price.py close; trades.py')[0])
@@ -580,31 +660,37 @@ def test_run_in_chat_refusals(self):
 		self.assertEqual(self.ran, [])
 
 	def test_scheduler_off(self):
+		"""With the scheduler off, .jobs says so."""
 		self.patch(jobs, 'current', None)
 		self.assertIn('off', self.list_in_chat('.jobs')[0])
 		self.assertIn('off', self.list_in_chat('.jobs run 8')[0])
 
 	def test_job_is_still_a_ticker(self):
+		""".job is a ticker, not .jobs."""
 		self.assertNotIn('Scheduled', str(self.command('.job')))
 
 class ChildProcess(FinbotCase):
 	"""Jobs run in their own process, so heavy work cannot hold up the bot's web server."""
 
 	def sleeper(self, seconds):
+		"""Have run_child start a process that sleeps for `seconds`, kept in self.children."""
 		real = jobs.subprocess.Popen
 		self.children = []
 		def popen(args, **kw):
+			"""Start the sleeper instead of the job."""
 			self.children.append(real([sys.executable, '-c', f'import time; time.sleep({seconds})']))
 			return self.children[-1]
 		self.patch(jobs.subprocess, 'Popen', popen)
 
 	def test_every_command_can_be_handed_to_a_child(self):
+		"""Every shipped command's arguments survive JSON, the way they reach the child."""
 		parsed, _ = jobs.parse_schedule(DEFAULT_SCHEDULE.read_text(), UTC)
 		for job in parsed:
 			for label, module, kwargs in job.commands:
 				self.assertEqual(json.loads(json.dumps(kwargs)), kwargs, label)
 
 	def test_a_real_child_runs_the_handler_and_reports_its_exit_status(self):
+		"""A real child runs the report and exits 0, or REPORTED when it failed and said so."""
 		real = jobs.subprocess.Popen
 		self.patch(jobs.subprocess, 'Popen', lambda args, **kw: real(args, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, **kw))
 		with tempfile.TemporaryDirectory() as data:
@@ -614,6 +700,7 @@ def test_a_real_child_runs_the_handler_and_reports_its_exit_status(self):
 				self.assertEqual(jobs.run_child('cleanup.py 0', cleanup, {'days': 'not a number'}), jobs.jobrun.REPORTED)
 
 	def test_a_job_process_survives_a_reload_signal(self):
+		"""A job process ignores SIGHUP, which systemctl kill -s HUP sends to every process of the unit."""
 		with tempfile.TemporaryDirectory() as scripts, tempfile.TemporaryDirectory() as data:
 			Path(scripts, 'sleeper.py').write_text('import time\ndef lambda_handler():\n    print("started", flush=True)\n    time.sleep(1)\n')
 			env = {**os.environ, 'PYTHONPATH': scripts, 'DATA_DIR': data, 'telegramBotToken': '', 'slack_webhook': '', 'discord_webhook': ''}
@@ -626,6 +713,7 @@ def test_a_job_process_survives_a_reload_signal(self):
 			self.assertEqual(proc.wait(timeout=30), 0)
 
 	def test_a_script_that_cannot_be_imported_is_reported(self):
+		"""A report that cannot be imported is reported like any other failure."""
 		with tempfile.TemporaryDirectory() as data:
 			env = {**os.environ, 'DATA_DIR': data, 'telegramBotToken': '', 'slack_webhook': '', 'discord_webhook': ''}
 			out = subprocess.run([sys.executable, '-m', 'lib.jobrun', 'nosuchscript', '{}'], cwd=ROOT, env=env, capture_output=True, text=True)
@@ -633,6 +721,8 @@ def test_a_script_that_cannot_be_imported_is_reported(self):
 		self.assertIn('⚠️ nosuchscript.py: ModuleNotFoundError', out.stderr)
 
 	def test_scheduler_reports_a_child_that_ended_without_saying_why(self):
+		"""A child killed by a signal or exiting non-zero without a report is reported by the scheduler; one that
+		reported itself is not reported twice."""
 		file = Path(tempfile.mkdtemp(), 'schedule.cron')
 		file.write_text('* * * * * trades.py\n')
 		for status, reason in ((-signal.SIGHUP, 'killed by SIGHUP'), (-signal.SIGKILL, 'killed by SIGKILL'), (1, 'exited with status 1 without saying why'),
@@ -654,9 +744,11 @@ def test_a_report_script_loads_neither_pandas_nor_the_scheduler(self):
 		self.assertEqual(out.stdout.strip().splitlines()[-1:], ['[]'], out.stderr)
 
 	def test_the_event_loop_keeps_running_while_a_child_works(self):
+		"""Waiting for a child does not stop the event loop."""
 		self.sleeper(1)
 		ticks = []
 		def ticker():
+			"""Tick five times, 0.1 seconds apart."""
 			for _ in range(5):
 				gevent.sleep(0.1)
 				ticks.append(time.monotonic())
@@ -669,6 +761,7 @@ def ticker():
 		self.assertLess(ticks[-1] - started, 0.8)
 
 	def test_a_timeout_kills_the_child(self):
+		"""A timeout stops the child, rather than leaving it running."""
 		self.sleeper(30)
 		started = time.monotonic()
 		with self.assertRaises(gevent.Timeout):
diff --git a/tests/test_providers.py b/tests/test_providers.py
index a9063c56..9089f641 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -9,14 +9,17 @@
 
 
 def sample(*path):
+	"""A saved API response from doc/."""
 	with open(os.path.join(DOC, *path), encoding='utf-8') as f:
 		return json.load(f)
 
 
 class Resp:
 	def __init__(self, payload=None, status=200, text=''):
+		"""A response with this JSON payload (None: not JSON), status and text."""
 		self.payload, self.status_code, self.text = payload, status, text or json.dumps(payload)
 	def json(self):
+		"""The payload, or ValueError when it is not JSON."""
 		if self.payload is None:
 			raise ValueError('no json')
 		return self.payload
@@ -27,6 +30,7 @@ def serve(*responses):
 	queue = list(responses)
 	calls = []
 	def get(url, **kwargs):
+		"""The next response, or raise it when it is an exception."""
 		calls.append(url)
 		item = queue.pop(0) if len(queue) > 1 else queue[0]
 		if isinstance(item, Exception):
@@ -38,6 +42,8 @@ def get(url, **kwargs):
 
 class ProviderCase(FinbotCase):
 	def setUp(self):
+		"""The real parsers in place of the fixture's canned data, with a fixed cookie, crumb and token, and no
+		waiting."""
 		super().setUp()
 		# these tests exercise the real parsers, so undo the canned data from the base fixture
 		for module, name in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (sharesight, 'get_portfolios'),
@@ -49,6 +55,8 @@ def setUp(self):
 		self.patch(sharesight.time, 'sleep', lambda s: None)
 
 	def http(self, module, *responses):
+		"""Have module's net.get replay these responses; returns the replacement, whose calls lists the URLs asked
+		for."""
 		getter = serve(*responses)
 		self.patch(module.net, 'get', getter)
 		return getter
@@ -56,6 +64,7 @@ def http(self, module, *responses):
 
 class YahooQuotes(ProviderCase):
 	def test_parses_real_v7_quote(self):
+		"""A real v7 quote parses into finbot's fields."""
 		self.http(yahoo, Resp(sample('yahoo', 'sample_yahoo_v7.json')))
 		data = yahoo.fetch(['AAPL'])['AAPL']
 		self.assertEqual(data['profile_title'], 'Apple')
@@ -68,17 +77,20 @@ def test_parses_real_v7_quote(self):
 		self.assertGreater(data['market_cap'], 1e12)
 
 	def test_second_call_is_served_from_cache(self):
+		"""A second fetch of the same ticker comes from the cache."""
 		getter = self.http(yahoo, Resp(sample('yahoo', 'sample_yahoo_v7.json')))
 		yahoo.fetch(['AAPL'])
 		yahoo.fetch(['AAPL'])
 		self.assertEqual(len(getter.calls), 1)
 
 	def test_falls_back_to_second_host(self):
+		"""When the first Yahoo host fails, the second is asked."""
 		getter = self.http(yahoo, Resp(status=500), Resp(sample('yahoo', 'sample_yahoo_v7.json')))
 		self.assertIn('AAPL', yahoo.fetch(['AAPL']))
 		self.assertIn('query1', getter.calls[1])
 
 	def test_all_hosts_failing_raises(self):
+		"""Every host failing, or Yahoo unreachable, is a RuntimeError."""
 		for failure in (Resp(status=500), ConnectionError('down')):
 			with self.subTest(failure=failure):
 				self.http(yahoo, failure)
@@ -86,13 +98,16 @@ def test_all_hosts_failing_raises(self):
 					yahoo.fetch(['AAPL'])
 
 	def test_empty_result_is_none(self):
+		"""An empty result is no quotes."""
 		self.http(yahoo, Resp({'quoteResponse': {'result': None, 'error': {'code': 'x'}}}))
 		self.assertEqual(yahoo.fetch(['NOPE']), {})
 
 	def test_no_tickers_is_none(self):
+		"""No tickers asked for: no quotes, and no request."""
 		self.assertEqual(yahoo.fetch([]), {})
 
 	def test_incomplete_quotes_are_skipped(self):
+		"""A quote without a price or a name is left out."""
 		body = sample('yahoo', 'sample_yahoo_v7.json')
 		good = body['quoteResponse']['result'][0]
 		no_price = {k: v for k, v in good.items() if k != 'regularMarketPrice'} | {'symbol': 'NOPRICE'}
@@ -112,26 +127,31 @@ def combined():
 		return {'quoteSummary': {'result': [merged], 'error': None}}
 
 	def test_parses_real_quote_summary(self):
+		"""A real quoteSummary parses, the company's description included."""
 		self.http(yahoo, Resp(self.combined()))
 		data = yahoo.fetch_detail('NVDA')['NVDA']
 		self.assertEqual(data['profile_title'], 'NVIDIA')
 		self.assertTrue(data['profile_bio'])
 
 	def test_a_cached_detail_needs_no_cookie(self):
+		"""A detail served from the cache takes neither the cookie nor the crumb, nor their lock."""
 		self.http(yahoo, Resp(self.combined()))
 		yahoo.fetch_detail('NVDA')
 		def refuse():
+			"""Fail: the credentials must not be needed."""
 			raise AssertionError('a cached detail took the Yahoo auth lock')
 		self.patch(yahoo, 'credentials', refuse)
 		self.assertIn('NVDA', yahoo.fetch_detail('NVDA'))
 
 	def test_http_failure_gives_empty(self):
+		"""A detail Yahoo will not give (404, unreachable) is empty, not an error."""
 		for failure in (Resp(status=404), ConnectionError('down')):
 			with self.subTest(failure=failure):
 				self.http(yahoo, failure)
 				self.assertEqual(yahoo.fetch_detail('DRNA'), {})
 
 	def test_missing_names_give_empty(self):
+		"""A detail without a name, or without a result, is empty."""
 		body = self.combined()
 		body['quoteSummary']['result'][0]['price'].update(longName=None, shortName=None)
 		self.http(yahoo, Resp(body))
@@ -142,6 +162,7 @@ def test_missing_names_give_empty(self):
 
 class YahooChart(ProviderCase):
 	def test_real_chart_to_dataframe_and_basics(self):
+		"""A real v8 chart becomes a frame of dates and closes, and its basics are read."""
 		chart = sample('yahoo', 'sample_yahoo_v8_chart.json')
 		df = yahoo.chart_json_to_df(chart)
 		self.assertEqual(len(df), 2)
@@ -150,6 +171,7 @@ def test_real_chart_to_dataframe_and_basics(self):
 		self.assertEqual((basics['symbol'], basics['currency']), ('CRWD', 'USD'))
 
 	def test_fetch_chart_json_failures_raise(self):
+		"""A chart Yahoo will not give is a RuntimeError."""
 		for failure in (Resp(status=404), ConnectionError('down')):
 			with self.subTest(failure=failure):
 				self.http(yahoo, failure)
@@ -167,6 +189,7 @@ def chart(closes, failing=()):
 		return {'chart': {'result': [{'meta': meta, 'timestamp': stamps, 'indicators': {'quote': [quote]}}], 'error': None}}
 
 	def test_1d_is_the_change_since_the_previous_close(self):
+		"""1D is the change since the previous close, and 5D the week's."""
 		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
 		self.http(yahoo, Resp(self.chart([90] * 30 + [100, 105, 110, 115, 120, 116.4]))) # up 20% in a week, down 3% today
 		table, image = yahoo.price_history('T1', graph=False)
@@ -174,6 +197,7 @@ def test_1d_is_the_change_since_the_previous_close(self):
 		self.assertGreater(table['5D'], 10) # the week, not the day
 
 	def test_a_failed_chart_fetch_raises(self):
+		"""A chart refused (429) or unreachable is a RuntimeError, which callers turn into leaving the ticker out."""
 		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
 		for failure in (Resp(status=429), ConnectionError('down')):
 			with self.subTest(failure=failure):
@@ -182,15 +206,19 @@ def test_a_failed_chart_fetch_raises(self):
 					yahoo.price_history('AAPL', 7, graph=False)
 
 	def test_one_failed_ticker_does_not_take_the_report_down(self):
+		"""One ticker without a chart is left out of a price list, a benchmark without one is said to have none, and
+		only a report on that one ticker fails, in one line."""
 		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts}) # quotes as canned in the base fixture, charts over HTTP
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t)})
 		self.patch(sharesight, 'get_portfolios', lambda: {}) # no Sharesight figures: every ticker needs its history
 		good = Resp(self.chart([100.0] * 20 + [110.0]))
 		def get(url, **kwargs):
+			"""A chart for every ticker but T3 and SPY, which get 404."""
 			return Resp(status=404) if '/chart/T3?' in url or '/chart/SPY?' in url else good
 		self.patch(yahoo.net, 'get', get)
-		def slack(text): # Slack: the reports come as text
+		def slack(text):
+			"""Run a command on Slack, where reports come as text, and return the text."""
 			self.sent.clear(); self.errors.clear()
 			worker.process_request('slack', 'C1', '<@U1>', text, BOT, 'U')
 			self.assertEqual(self.errors, [], text)
@@ -211,12 +239,14 @@ def slack(text): # Slack: the reports come as text
 		self.assertTrue(self.errors[0].startswith('⚠️ .price: RuntimeError: 404 error for T3'))
 
 	def test_the_chart_request_carries_the_cookie_of_its_crumb(self):
+		"""A chart request sends the cookie its crumb belongs to."""
 		headers = []
 		self.patch(yahoo.net, 'get', lambda url, **kwargs: headers.append(kwargs.get('headers', {})) or Resp(self.chart([100.0] * 5)))
 		yahoo.fetch_chart_json('T1', full=True)
 		self.assertEqual(headers[0].get('Cookie'), 'A=B') # the crumb used to go without its cookie
 
 	def test_with_the_cache_off_no_chart_is_written(self):
+		"""With the cache preference off, a drawn chart is not written to the cache."""
 		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
 		self.prefs(cache=False)
 
@@ -226,6 +256,7 @@ def test_with_the_cache_off_no_chart_is_written(self):
 		self.assertEqual([name for name in os.listdir(config.env.cache_dir) if name.startswith('finbot_graph_')], [])
 
 	def test_price_series_raises_on_failure_and_on_empty_history(self):
+		"""price_series raises when Yahoo fails and when the history has no closes."""
 		self.http(yahoo, Resp(status=500))
 		with self.assertRaises(RuntimeError):
 			yahoo.price_series('AAPL')
@@ -238,10 +269,12 @@ def test_price_series_raises_on_failure_and_on_empty_history(self):
 
 class SharesightApi(ProviderCase):
 	def test_portfolios_from_real_response(self):
+		"""The portfolio list parses from a real answer."""
 		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_portfolios.json')))
 		self.assertEqual(sharesight.get_portfolios(), {'Rob': 445825, 'RobSMSF': 684141})
 
 	def test_portfolio_exclusion_and_inclusion(self):
+		"""exclude_portfolios and include_portfolios pick the portfolios reported on."""
 		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_portfolios.json')))
 		self.prefs(exclude_portfolios=['445825'])
 
@@ -253,6 +286,7 @@ def test_portfolio_exclusion_and_inclusion(self):
 		self.assertEqual(sharesight.get_portfolios(), {'Rob': 445825})
 
 	def test_portfolio_failures_raise(self):
+		"""An error answer, Sharesight unreachable or no portfolios is a RuntimeError."""
 		for failure in (Resp({'error': 'nope', 'error_code': 'x'}, 401), ConnectionError('down'), Resp({'portfolios': []})):
 			with self.subTest(failure=failure):
 				self.http(sharesight, failure)
@@ -260,6 +294,7 @@ def test_portfolio_failures_raise(self):
 					sharesight.get_portfolios()
 
 	def test_trades_from_real_response(self):
+		"""Trades parse from a real answer, tagged with their portfolio."""
 		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_trades_684141_1.json')))
 		trades = sharesight.get_trades('RobSMSF', 684141, 1)
 		self.assertEqual(len(trades), 2)
@@ -267,6 +302,7 @@ def test_trades_from_real_response(self):
 		self.assertEqual(trades[0]['symbol'], 'FANG')
 
 	def test_trade_failures_raise(self):
+		"""Sharesight unreachable, or an error answer, is a RuntimeError."""
 		self.http(sharesight, ConnectionError('down'))
 		with self.assertRaises(RuntimeError):
 			sharesight.get_trades('RobSMSF', 684141, 1)
@@ -275,6 +311,7 @@ def test_trade_failures_raise(self):
 			sharesight.get_trades('RobSMSF', 684141, 2)
 
 	def test_performance_and_holdings_from_real_response(self):
+		"""A real holdings report parses, and its tickers come out sorted, US ones without a suffix."""
 		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_performance_684141_0.json')))
 		self.assertEqual(len(sharesight.get_performance(684141, 0)['report']['holdings']), 18)
 		tickers = sharesight.Portfolio('RobSMSF', 684141).tickers()
@@ -282,12 +319,14 @@ def test_performance_and_holdings_from_real_response(self):
 		self.assertEqual(tickers, sorted(tickers))
 
 	def test_performance_retries_when_sharesight_is_busy(self):
+		"""While Sharesight says too many requests are in flight, the request is tried again."""
 		busy = Resp({'error': 'Too many parallel requests', 'error_code': 'x'}, 429)
 		getter = self.http(sharesight, busy, busy, Resp(sample('sharesight', 'finbot_sharesight_performance_684141_0.json')))
 		self.assertIn('report', sharesight.get_performance(684141, 0))
 		self.assertEqual(len(getter.calls), 3)
 
 	def test_performance_gives_up_after_repeated_busy_or_hard_errors(self):
+		"""Busy every time, a hard error, or unreachable: a RuntimeError in the end."""
 		for failure in (Resp({'error': 'Too many parallel requests', 'error_code': 'x'}, 429),
 						Resp({'error': 'forbidden', 'error_code': 'x'}, 403), ConnectionError('down')):
 			with self.subTest(failure=failure):
@@ -296,6 +335,7 @@ def test_performance_gives_up_after_repeated_busy_or_hard_errors(self):
 					sharesight.get_performance(684141, 7)
 
 	def test_an_answer_without_a_report_says_so(self):
+		"""A performance answer without a report is an error naming the portfolio, and is not cached."""
 		self.http(sharesight, Resp({}))
 		with self.assertRaisesRegex(RuntimeError, 'no performance report for portfolio 684141'):
 			sharesight.get_performance(684141, 7)
@@ -306,21 +346,25 @@ class SharesightKeepWarm(ProviderCase):
 	"""The background refresher keeps the portfolio list and the holdings report (days=0) cached ahead of their TTLs."""
 
 	def serve_sharesight(self):
+		"""Answer with the real portfolio list and holdings report; returns the URLs asked for."""
 		portfolios = sample('sharesight', 'finbot_sharesight_portfolios.json')
 		performance = sample('sharesight', 'finbot_sharesight_performance_684141_0.json')
 		calls = []
 		def get(url, **kwargs):
+			"""The holdings report or the portfolio list, as asked."""
 			calls.append(url)
 			return Resp(performance if '/performance' in url else portfolios)
 		self.patch(sharesight.net, 'get', get)
 		return calls
 
 	def age(self, cache_file, seconds):
+		"""Make a cache file `seconds` older."""
 		path = os.path.join(config.env.cache_dir, cache_file)
 		old = os.path.getmtime(path) - seconds
 		os.utime(path, (old, old))
 
 	def test_cold_cache_is_filled_then_left_alone(self):
+		"""A cold cache is filled (the list and each portfolio's holdings, as of today), then left alone while fresh."""
 		calls = self.serve_sharesight()
 		self.assertEqual(sharesight.warm()[0], 3) # portfolio list + one holdings report per portfolio (2)
 		self.assertEqual(sum('/performance' in c for c in calls), 2)
@@ -329,6 +373,7 @@ def test_cold_cache_is_filled_then_left_alone(self):
 		self.assertEqual(len(calls), 3) # nothing refetched while fresh
 
 	def test_refreshes_before_the_ttl_runs_out(self):
+		"""A cache file at 95% of its lifetime is fetched again, and only that one."""
 		calls = self.serve_sharesight()
 		sharesight.warm()[0]
 		calls.clear()
@@ -342,6 +387,7 @@ def test_refreshes_before_the_ttl_runs_out(self):
 		self.assertTrue(calls[0].endswith('/portfolios'))
 
 	def test_a_request_never_waits_on_a_cache_the_job_has_warmed(self):
+		"""Once warmed, the list and the holdings report need no request."""
 		calls = self.serve_sharesight()
 		sharesight.warm()[0]
 		calls.clear()
@@ -350,6 +396,7 @@ def test_a_request_never_waits_on_a_cache_the_job_has_warmed(self):
 		self.assertEqual(calls, [])
 
 	def test_every_holding_is_served_from_the_warmed_cache(self):
+		"""Every portfolio's tickers come from the warmed cache, with no request."""
 		calls = self.serve_sharesight()
 		sharesight.warm()[0]
 		calls.clear()
@@ -358,6 +405,7 @@ def test_every_holding_is_served_from_the_warmed_cache(self):
 		self.assertEqual(calls, [])
 
 	def test_failed_refresh_keeps_the_old_cache(self):
+		"""A refresh that fails keeps the cache it would have replaced."""
 		self.serve_sharesight()
 		sharesight.warm()[0]
 		self.age(sharesight.PORTFOLIOS_CACHE, 0.95 * config.prefs.cache_seconds)
@@ -367,12 +415,21 @@ def test_failed_refresh_keeps_the_old_cache(self):
 		self.assertEqual(sharesight.get_portfolios(), {'Rob': 445825, 'RobSMSF': 684141}) # still served from the cache
 
 	def test_loop_survives_failures_and_backs_off(self):
+		"""keep_warm survives a failure, retries slowly, then sleeps until the next cache is due."""
 		class Stop:
-			def __init__(self): self.waits = []
-			def is_set(self): return len(self.waits) >= 2
-			def wait(self, seconds): self.waits.append(seconds)
+			"""keep_warm's stop event: notes each wait, and is set after two."""
+			def __init__(self):
+				"""No waits yet."""
+				self.waits = []
+			def is_set(self):
+				"""Set after two waits."""
+				return len(self.waits) >= 2
+			def wait(self, seconds):
+				"""Note the wait instead of sleeping."""
+				self.waits.append(seconds)
 		outcomes = [RuntimeError('boom'), 0]
 		def warm():
+			"""Fail once, then succeed."""
 			outcome = outcomes.pop(0)
 			if isinstance(outcome, Exception):
 				raise outcome
@@ -384,10 +441,18 @@ def warm():
 		self.assertEqual(stop.waits, [300, 3000]) # slow retry after the failure, then sleep until the next cache is due
 
 	def test_with_cache_off_nothing_is_fetched_until_a_reload_turns_it_on(self):
+		"""With the cache preference off, keep_warm idles without asking Sharesight, and starts when a reload turns it
+		on."""
 		class Stop:
-			def __init__(self): self.waits = []
-			def is_set(self): return len(self.waits) >= 4
+			"""keep_warm's stop event: notes each wait, switches the cache on at the second, and is set after four."""
+			def __init__(self):
+				"""No waits yet."""
+				self.waits = []
+			def is_set(self):
+				"""Set after four waits."""
+				return len(self.waits) >= 4
 			def wait(self, seconds):
+				"""Note the wait; at the second, switch the cache on as a reload would."""
 				self.waits.append(seconds)
 				if len(self.waits) == 2:
 					config.prefs = dataclasses.replace(config.prefs, cache=True) # what a reload that switches it on does
@@ -401,6 +466,7 @@ def wait(self, seconds):
 		self.assertEqual((stop.waits, len(passes)), ([60, 60, 3000, 3000], 2)) # idle while off: no Sharesight request every 5 seconds
 
 	def test_a_flush_wakes_the_loop_to_refill_at_once(self):
+		"""A flush wakes keep_warm, which refills the caches at once instead of when they are next due."""
 		passes = []
 		self.patch(sharesight, 'warm', lambda: (passes.append(1), (0, {}))[1])
 		self.patch(sharesight, 'next_due', lambda portfolios=None: 3000)
@@ -413,6 +479,7 @@ def test_a_flush_wakes_the_loop_to_refill_at_once(self):
 		self.assertEqual(len(passes), 2)
 
 	def test_a_warming_pass_reads_the_cached_portfolio_list_once(self):
+		"""One pass of keep_warm reads the cached portfolio list once, and hands it on."""
 		self.serve_sharesight()
 		sharesight.warm()[0] # fill the caches
 		reads = []
@@ -425,6 +492,7 @@ def test_a_warming_pass_reads_the_cached_portfolio_list_once(self):
 		self.assertEqual(portfolios, {'Rob': 445825, 'RobSMSF': 684141})
 
 	def test_next_due_is_90_percent_of_the_ttl(self):
+		"""The next refresh is due at 90% of the shortest cache lifetime; at once for a cold cache."""
 		self.serve_sharesight()
 		self.assertEqual(sharesight.next_due(), 0) # cold cache
 		sharesight.warm()[0]
@@ -432,6 +500,7 @@ def test_next_due_is_90_percent_of_the_ttl(self):
 		self.assertAlmostEqual(due, 0.9 * min(config.prefs.cache_seconds, config.prefs.performance_cache_seconds), delta=5)
 
 	def test_refresh_flag_bypasses_a_fresh_cache(self):
+		"""refresh=True asks Sharesight even while the cache is fresh."""
 		calls = self.serve_sharesight()
 		sharesight.get_portfolios()
 		sharesight.get_portfolios()
@@ -444,7 +513,9 @@ class TelegramSending(FinbotCase):
 	"""webhook.write against failing endpoints. The real send path, not the base fixture's recorder."""
 
 	def post(self, outcome):
+		"""Send 'hi' to Telegram through webhook.write, the post answering with outcome (or raising it)."""
 		def fake(url, **kwargs):
+			"""The outcome, or raise it."""
 			if isinstance(outcome, Exception):
 				raise outcome
 			return outcome
@@ -452,16 +523,20 @@ def fake(url, **kwargs):
 		return webhook.write('telegram', 'https://api.telegram.org/botX/sendMessage?chat_id=5', 'hi', '5')
 
 	def test_success_returns_json(self):
+		"""A sent message returns Telegram's answer."""
 		self.assertEqual(self.post(Resp({'ok': True})), {'ok': True})
 
 	def test_http_error_and_exception_return_none(self):
+		"""A refused or failed send returns None."""
 		self.assertIsNone(self.post(Resp(status=429)))
 		self.assertIsNone(self.post(ConnectionError('down')))
 
 	def test_non_json_success_returns_empty_dict(self):
+		"""A 200 that is not JSON returns {}."""
 		self.assertEqual(self.post(Resp(None)), {})
 
 	def test_error_line_redacts_secrets_and_truncates(self):
+		"""An error line has no token or crumb in it, and is cut to about 200 characters."""
 		line = webhook.error_line(RuntimeError('GET /bot123456:ABC-def_1/send?crumb=SECRET&x=1 failed'), 'ctx')
 		self.assertNotIn('ABC-def_1', line)
 		self.assertNotIn('SECRET', line)
@@ -469,6 +544,7 @@ def test_error_line_redacts_secrets_and_truncates(self):
 		self.assertLessEqual(len(webhook.error_line('x' * 1000)), 205)
 
 	def test_error_reports_are_escaped_for_telegram_and_slack(self):
+		"""An error line is escaped for Telegram's HTML and Slack (the same), and left as it is for Discord."""
 		sent = []
 		self.patch(webhook, 'write', lambda service, url, text, *a, **k: sent.append((service, text)))
 		self.env(webhooks={'telegram': 'https://api.telegram.org/botX/', 'slack': 'https://hooks.slack.com/x', 'discord': 'https://discord.com/x/slack'})
@@ -486,11 +562,13 @@ def test_error_reports_are_escaped_for_telegram_and_slack(self):
 		self.assertIn('crumb=<redacted>', texts['discord'])
 
 	def test_report_error_never_raises(self):
+		"""report_error survives a failed send."""
 		self.patch(webhook, 'write', mock.Mock(side_effect=RuntimeError('send failed')))
 		webhook.report_error(ValueError('boom'), service='telegram', chat_id='5')
 		webhook.report_error(ValueError('boom'))
 
 	def test_guarded_reports_and_exits_nonzero(self):
+		"""guarded reports a failure and exits 1; a success returns its value."""
 		reported = []
 		self.patch(webhook, 'report_error', lambda e, **k: reported.append(e))
 		with self.assertRaises(SystemExit) as raised:
@@ -506,6 +584,7 @@ class ChartDelivery(FinbotCase):
 	UPLOAD_URL = 'https://files.slack.com/upload/v1/abc'
 
 	def setUp(self):
+		"""The real sendPhoto, in place of the fixture's recorder."""
 		super().setUp()
 		self.patch(webhook, 'sendPhoto', REAL_SEND_PHOTO)
 
@@ -513,6 +592,7 @@ def slack_answers(self, *outcomes):
 		"""net.post for the upload steps: the outcomes in turn (an exception is raised). Returns the calls made."""
 		calls = []
 		def post(url, **kwargs):
+			"""Record the call and answer with the next outcome, or raise it."""
 			calls.append((url, kwargs))
 			outcome = outcomes[len(calls) - 1]
 			if isinstance(outcome, Exception):
@@ -522,9 +602,12 @@ def post(url, **kwargs):
 		return calls
 
 	def upload_steps(self):
+		"""Slack's answers to the three upload steps, each a success."""
 		return [Resp({'ok': True, 'upload_url': self.UPLOAD_URL, 'file_id': 'F1'}), Resp(None, text='OK - 3'), Resp({'ok': True, 'files': [{'id': 'F1'}]})]
 
 	def test_where_charts_can_go(self):
+		"""Charts go to Telegram and Discord always, to Slack in replies with the bot token, never in Slack's scheduled
+		reports."""
 		self.env(slackBotToken='xoxb-test')
 
 		cases = (('telegram',), ('telegram', False), ('slack',), ('slack', False), ('discord', False))
@@ -534,6 +617,8 @@ def test_where_charts_can_go(self):
 		self.assertFalse(webhook.takes_charts('slack'))
 
 	def test_slack_replies_upload_their_charts(self):
+		"""A Slack reply uploads its chart in three steps with the bot token, the report as its message, not sent again
+		as text."""
 		self.env(slackBotToken='xoxb-test')
 
 		for cmd, heading in (('.beta', '*Beta over 1.5'), ('.price', '*5% at close'), ('.compare T1 T2', '*')):
@@ -553,6 +638,8 @@ def test_slack_replies_upload_their_charts(self):
 				self.assertNotIn(share['data']['initial_comment'].split('\n'), [payload for _, payload in self.sent]) # not sent again as text
 
 	def test_a_slack_upload_that_fails_goes_out_as_text(self):
+		"""A Slack upload that fails at any step goes out as text, but not after a timeout on the last step, when it may
+		have arrived."""
 		self.env(slackBotToken='xoxb-test')
 
 		good = self.upload_steps()
@@ -570,6 +657,7 @@ def test_a_slack_upload_that_fails_goes_out_as_text(self):
 				self.assertEqual(self.sent, expect)
 
 	def test_scheduled_reports_to_slack_stay_text(self):
+		"""Scheduled reports reach Slack's incoming webhook as text and Telegram with their charts."""
 		self.env(slackBotToken='xoxb-test')
 
 		self.env(webhooks={**config.env.webhooks, 'slack': 'https://hooks.slack.com/services/x'}) # an incoming webhook cannot upload
@@ -583,9 +671,11 @@ def test_scheduled_reports_to_slack_stay_text(self):
 				self.assertEqual(sorted(set(sent)), [('slack', 'text'), ('telegram', 'chart')])
 
 	def test_slack_without_a_bot_token_gets_text_and_no_chart_is_drawn(self):
+		"""Without the bot token, Slack replies are text and no chart is drawn for them."""
 		self.env(slackBotToken=False)
 
 		def refuse(*a, **k):
+			"""Fail: no chart should be drawn."""
 			raise AssertionError('a chart was drawn for Slack')
 		for name in ('rows_chart', 'pie_chart', 'bar_graph', 'shorts_chart'):
 			self.patch(render, name, refuse)
@@ -616,8 +706,12 @@ def test_the_graph_preference_off_sends_no_chart_anywhere(self):
 		self.assertEqual([url for url in posts if 'sendPhoto' in url or 'files.' in url], [])
 
 	def test_a_chart_telegram_refuses_goes_out_as_text(self):
+		"""A chart Telegram refuses or that fails goes out as text, but not after a timeout, when it may have
+		arrived."""
 		def post(outcome):
+			"""A net.post that answers with outcome, or raises it."""
 			def fake(url, **kwargs):
+				"""The outcome, or raise it."""
 				if isinstance(outcome, Exception):
 					raise outcome
 				return outcome
@@ -640,6 +734,7 @@ class DiscordWebhook(FinbotCase):
 	URL = 'https://discord.com/api/webhooks/1/abc'
 
 	def setUp(self):
+		"""Discord as the only destination, sent through the real path, its posts recorded."""
 		super().setUp()
 		self.env(webhooks={'discord': self.URL})
 		self.patch(webhook, 'payload_wrapper', REAL_PAYLOAD_WRAPPER)
@@ -648,12 +743,15 @@ def setUp(self):
 		self.patch(webhook.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or Resp({'id': '1'}))
 
 	def test_text_goes_to_the_native_endpoint_without_previews_or_pings(self):
+		"""Text goes to the native endpoint with ?wait=true, without previews or mentions."""
 		webhook.write('discord', self.URL, 'hello @everyone')
 		url, kwargs = self.posts[0]
 		self.assertEqual(url, self.URL + '?wait=true') # without wait Discord answers 204, which counted as a failure
 		self.assertEqual(kwargs['json'], {'content': 'hello @everyone', 'flags': 4, 'allowed_mentions': {'parse': []}})
 
 	def test_a_scheduled_report_carries_its_chart(self):
+		"""A scheduled report goes to Discord with its chart attached and the report, within 2,000 characters, as the
+		message."""
 		price.lambda_handler()
 		(url, kwargs), = self.posts
 		self.assertEqual(url, self.URL + '?wait=true')
@@ -666,6 +764,7 @@ def test_a_scheduled_report_carries_its_chart(self):
 		self.assertTrue(image.read().startswith(b'\x89PNG'))
 
 	def test_a_long_report_is_split_by_size(self):
+		"""A long report is split into messages of at most 2,000 characters."""
 		line = 'x' * 150
 		webhook.payload_wrapper('discord', self.URL, [line] * 30)
 		sizes = [len(kwargs['json']['content']) for _, kwargs in self.posts]
@@ -673,6 +772,7 @@ def test_a_long_report_is_split_by_size(self):
 		self.assertTrue(all(size <= 2000 for size in sizes), sizes) # 20 lines a chunk used to make 3,019 characters
 
 	def test_links_are_masked_links(self):
+		"""Links are masked links, brackets and parentheses escaped."""
 		self.assertEqual(DISCORD.link('https://x.example/a(b)', 'T[1]'), '[T\\[1\\]](<https://x.example/a%28b%29>)')
 
 
@@ -680,6 +780,7 @@ class SlackSending(FinbotCase):
 	"""Slack replies go through chat.postMessage with the bot token; scheduled reports through the incoming webhook."""
 
 	def setUp(self):
+		"""The real send path, its posts recorded."""
 		super().setUp()
 		self.patch(webhook, 'payload_wrapper', REAL_PAYLOAD_WRAPPER)
 		self.patch(webhook, 'sendPhoto', REAL_SEND_PHOTO)
@@ -687,6 +788,7 @@ def setUp(self):
 		self.patch(webhook.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or Resp({'ok': True}))
 
 	def test_link_previews_are_turned_off_in_the_body(self):
+		"""Link previews are turned off in the body, where Slack reads them, for replies and for the webhook."""
 		self.env(slackBotToken='xoxb-test')
 
 		webhook.write('slack', 'https://slack.com/api/chat.postMessage', 'hi', 'C1')
@@ -698,6 +800,7 @@ def test_link_previews_are_turned_off_in_the_body(self):
 				self.assertNotIn('unfurl_links', kwargs['headers'])
 
 	def test_a_reply_without_a_bot_token_is_logged_not_sent(self):
+		"""A Slack reply without the bot token is logged, once each, and not sent."""
 		self.env(slackBotToken=False)
 
 		err = io.StringIO()
@@ -708,6 +811,7 @@ def test_a_reply_without_a_bot_token_is_logged_not_sent(self):
 		self.assertEqual(err.getvalue().count('ERROR slack: a reply to channel C1 needs slackBotToken in finbot.env; not sent'), 3)
 
 	def test_the_slack_api_says_what_is_missing(self):
+		"""slack_api without the bot token says so (it used to be a TypeError)."""
 		self.env(slackBotToken=False)
 
 		with self.assertRaisesRegex(RuntimeError, 'Slack files.getUploadURLExternal: needs slackBotToken in finbot.env'):
@@ -719,14 +823,17 @@ class TelegramWebhook(FinbotCase):
 	"""bot.py registers the webhook once as it starts. Importing lib.telegram, as every job process does, touches nothing."""
 
 	def setUp(self):
+		"""A Telegram bot with an outgoing webhook configured."""
 		super().setUp()
 		self.env(webhooks={'telegram': 'https://api.telegram.org/botX/'})
 		self.env(telegramOutgoingWebhook='https://example.com/telegram')
 
 
 	def test_register_unregisters_registers_and_names_the_bot(self):
+		"""register() removes the webhook, sets it again and returns the bot's @name."""
 		calls = []
 		def post(url, **kwargs):
+			"""Record the method and URL, and answer like Telegram."""
 			calls.append((url.rsplit('/', 1)[1], kwargs.get('params', {}).get('url')))
 			return Resp({'ok': True, 'result': {'username': 'finbot'} if url.endswith('getMe') else True})
 		self.patch(telegram.net, 'post', post)
@@ -735,6 +842,7 @@ def post(url, **kwargs):
 		self.assertEqual(calls, [('setWebhook', ''), ('setWebhook', 'https://example.com/telegram'), ('getMe', None)])
 
 	def test_a_refused_call_raises(self):
+		"""A call Telegram refuses raises with its description."""
 		self.patch(telegram.net, 'post', lambda url, **k: Resp({'ok': False, 'description': 'Bad Request: bad webhook: HTTPS url must be provided'}))
 		with contextlib.redirect_stderr(io.StringIO()), self.assertRaises(RuntimeError) as caught:
 			telegram.register()
@@ -744,7 +852,9 @@ def test_edits_toasts_and_typing_go_through_call(self):
 		"""A chart edit that Telegram refuses is an error (unless nothing changed); a toast, a text edit or a typing notice that
 		fails, or an edit lost on the way, is only logged."""
 		def answer(outcome):
+			"""A net.post that answers with outcome, or raises it."""
 			def post(url, **kwargs):
+				"""The outcome, or raise it."""
 				if isinstance(outcome, Exception):
 					raise outcome
 				return outcome
@@ -764,6 +874,7 @@ def post(url, **kwargs):
 					webhook.pleaseHold('typing', 'telegram', '5')
 
 	def test_importing_never_calls_telegram(self):
+		"""Importing the modules a job process uses makes no request."""
 		code = ('import requests\n'
 				'def refuse(*a, **k): raise SystemExit("network call on import")\n'
 				'requests.Session.request = refuse\n'
@@ -775,6 +886,7 @@ def test_importing_never_calls_telegram(self):
 		self.assertEqual(out.stdout.split()[-1:], ['quiet'], out.stderr)
 
 	def test_the_bot_will_not_start_unregistered_and_never_logs_the_token(self):
+		"""bot.py exits when the webhook cannot be registered, and never logs the token."""
 		skip_if_old_env(self)
 		with tempfile.TemporaryDirectory() as data:
 			env = {k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k not in config.STARTUP_KEYS and 'proxy' not in k.lower()}
diff --git a/tests/test_reports.py b/tests/test_reports.py
index b975117d..4c759ac0 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -6,6 +6,7 @@
 
 class Commands(FinbotCase):
 	def test_text_commands(self):
+		"""Every text command answers."""
 		for cmd in ('.help', '/start', '.hello', '.thanks', '.who', '.plan', '.holdings', '.marketcap',
 					'.pe', '.fpe', '.peg', '.beta', '.buy', '.sell', '.earnings', '.dividend',
 					'.earnings T1', '.dividend T1', '.trades', '.trades 30d', '.price T1', '.session T1',
@@ -14,6 +15,7 @@ def test_text_commands(self):
 				self.assertTrue(self.command(cmd), f'{cmd} sent nothing')
 
 	def test_chart_commands(self):
+		"""Every chart command answers with a chart."""
 		for cmd in ('.price', '.price 7d', '.price top', '.price bottom', '.session', '.premarket', '.postmarket',
 					'.performance', '.performance 7d', '.performance 1y', '.shorts', '.peg', '.buy', '.sell', '.beta', '.price top 7d',
 					'.forward pe', '.fpe bottom'):
@@ -23,6 +25,7 @@ def test_chart_commands(self):
 				self.assertTrue(any(kind == 'photo' for kind, *_ in out), f'{cmd} sent no chart: {out}')
 
 	def test_one_button_asks_for_both_ratings_reports(self):
+		""".buy & sell (and its spellings) sends both ratings reports, in the order asked, each once."""
 		for text in ('.buy & sell', '.buy sell', '.buy/sell', '.buy and sell'):
 			with self.subTest(text=text):
 				captions = [caption for kind, _, caption in self.command(text) if kind == 'photo']
@@ -33,6 +36,7 @@ def test_one_button_asks_for_both_ratings_reports(self):
 		self.assertEqual(len(self.command('.buy & buy')), 1) # each report once
 
 	def test_the_keyboard_is_alphabetical_and_every_button_is_a_command(self):
+		"""The command keyboard is alphabetical, and each button is a command, not taken for a ticker by .SYMBOL."""
 		buttons = [button['text'] for row in charts.command_keyboard()['keyboard'] for button in row]
 		self.assertEqual(buttons, sorted(buttons))
 		self.assertEqual(len(buttons), 15)
@@ -43,21 +47,24 @@ def test_the_keyboard_is_alphabetical_and_every_button_is_a_command(self):
 				self.assertIsNot(next(handler for pattern, handler in worker.COMMANDS if worker.match(pattern, text, BOT)), worker.on_profile)
 
 	def test_history_and_compare(self):
+		""".history and .compare answer."""
 		for cmd in ('.history T1', '.compare T1 T2', '.compare T1 T2 3m'):
 			with self.subTest(cmd=cmd):
 				self.assertTrue(self.command(cmd), f'{cmd} sent nothing')
 
 	def test_trades_and_watchlist_report_when_nothing_is_stored(self):
-		# first run: no state files exist yet, which used to crash these
+		"""On the first run no state files exist yet, which used to crash .trades and .watchlist."""
 		for cmd in ('.trades', '.watchlist'):
 			self.assertTrue(self.command(cmd), cmd)
 
 	def test_unknown_stickers_and_noise_are_ignored(self):
+		"""A message that is no command gets no answer."""
 		self.assertEqual(self.command('lorem ipsum dolor sit amet'), [])
 
 
 class ReportFunctions(FinbotCase):
 	def test_prepare_help(self):
+		"""The help text is built."""
 		self.assertTrue(about.prepare_help(TELEGRAM, BOT))
 
 	def test_a_portfolio_with_no_holdings_is_reported_by_name(self):
@@ -67,12 +74,15 @@ def test_a_portfolio_with_no_holdings_is_reported_by_name(self):
 		self.assertTrue(any('>Rob</a>' in line for line in payload), payload)
 
 	def test_prepare_holdings(self):
+		""".holdings without a portfolio lists the portfolios."""
 		self.assertTrue(holdings.prepare_holdings_payload(None, TELEGRAM))
 
 	def test_prepare_marketcap(self):
+		"""The market cap list is built."""
 		self.assertTrue(rankings.prepare_marketcap_payload(TELEGRAM, 'top', length=5))
 
 	def test_prepare_rating(self):
+		"""The buy and sell lists are built."""
 		for action in ('buy', 'sell'):
 			self.assertTrue(rankings.prepare_rating_payload(TELEGRAM, action, length=5)[0])
 
@@ -84,6 +94,7 @@ def test_ratings_list_the_strongest_first(self):
 		self.patch(watchlist, 'with_holdings', lambda: list(ratings))
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | dict(zip(('recommend', 'recommend_index', 'recommend_analysts'), ratings[t]))})
 		def order(action):
+			"""(the tickers listed, the tickers charted) of a ratings report, in order."""
 			payload, chart = rankings.prepare_rating_payload(TELEGRAM, action)
 			return [line.split('Co ')[1].split(' ')[0] for line in payload[1:]], [label for label, _ in chart['rows']]
 		self.assertEqual(order('buy'), (['T2', 'T3', 'T1'], ['T2', 'T3', 'T1']))
@@ -120,6 +131,7 @@ def test_forward_pe_lists_need_as_many_analysts_as_buy(self):
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | {'price_to_earnings_forward': -5 if t == 'T3' else 20} for t in ts})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'recommend_analysts': analysts[t]}})
 		def listed(action):
+			"""(heading, the tickers listed, the chart) of a ratio report."""
 			payload, chart = rankings.prepare_value_payload(TELEGRAM, action)
 			return payload[0], [line.split('Co ')[1].split(' ')[0] for line in payload[1:]], chart
 		heading, names, chart = listed('forward pe')
@@ -136,6 +148,7 @@ def test_market_caps_compare_in_us_dollars(self):
 		stocks = {'K1.KS': ('KRW', 1.01e15), 'K2.KS': ('KRW', 5e11), 'U1': ('USD', 5.39e12), 'L1.L': ('GBp', 6.25e10), 'Z1': ('XYZ', 1e12)}
 		fx = {'KRW=X': 1380.0, 'GBP=X': 0.79} # units per US dollar; nothing for XYZ
 		def fetch(tickers):
+			"""Quotes in each stock's own currency, and the exchange rates."""
 			return {t: market_entry(t) | {'regularMarketPrice': fx[t]} if t in fx else market_entry(t) | {'currency': stocks[t][0], 'market_cap': stocks[t][1], 'beta': 2.0}
 					for t in tickers if t in fx or t in stocks}
 		self.patch(yahoo, 'fetch', fetch)
@@ -152,6 +165,7 @@ def fetch(tickers):
 		self.assertNotIn('Co K1.KS', str(beta))
 
 	def test_prepare_value(self):
+		"""Each ratio report has its heading: top, bottom, and negative (with none found)."""
 		for action, heading in (('pe', 'Top 5 tracked stocks by pe ratio'), ('bottom forward pe', 'Bottom 5 tracked stocks by forward pe ratio'),
 								('peg', 'Top 5 tracked stocks by peg ratio'), ('negative peg', 'negative peg: no data found')):
 			with self.subTest(action=action):
@@ -159,6 +173,7 @@ def test_prepare_value(self):
 				self.assertIn(heading, payload[0])
 
 	def test_prepare_beta(self):
+		"""The beta report charts a bar per stock listed, and says so when none qualify."""
 		payload, chart = rankings.prepare_beta_payload(TELEGRAM)
 		self.assertIn('Beta over 1.5', payload[0])
 		self.assertEqual(len(chart['rows']), len(payload) - 1) # a bar per stock listed
@@ -166,6 +181,7 @@ def test_prepare_beta(self):
 		self.assertEqual(rankings.prepare_beta_payload(TELEGRAM), (['No tracked stocks have a beta over 1.5 and mkt cap under US$2B'], None))
 
 	def test_prepare_profile(self):
+		"""A company profile is built."""
 		self.assertTrue(profile.prepare_profile_payload(TELEGRAM, '@u', 'T1'))
 
 	def test_a_profile_leaves_out_what_yahoo_did_not_send(self):
@@ -180,6 +196,7 @@ def test_a_profile_leaves_out_what_yahoo_did_not_send(self):
 		self.assertNotIn('Net debt/equity', str(profile.prepare_profile_payload(TELEGRAM, '@u', 'T1')))
 
 	def test_a_profile_warns_of_a_ps_ratio_only_above_15(self):
+		"""The PS ratio line warns only above 15; the forward P/E's warning used to show on it too."""
 		for ps, warned in ((12.3, False), (15.1, True)): # forward P/E 99 warns (above 30): its warning used to show on the PS line too
 			with self.subTest(ps=ps):
 				self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'price_to_sales': ps, 'price_to_earnings_forward': 99, 'profile_industry': 'Tools'}})
@@ -187,9 +204,11 @@ def test_a_profile_warns_of_a_ps_ratio_only_above_15(self):
 				self.assertEqual('⚠️' in line, warned, line)
 
 	def test_prepare_watchlist(self):
+		"""The watchlist is listed."""
 		self.assertTrue(holdings.prepare_watchlist(TELEGRAM, '@u'))
 
 	def test_prepare_compare_usage(self):
+		""".compare with one ticker gets its usage."""
 		caption = history.prepare_compare(TELEGRAM, ['T1'])[0]
 		self.assertIn('Usage', ' '.join(caption))
 
@@ -198,18 +217,22 @@ class ScheduledReports(FinbotCase):
 	"""The cron entry points (non-interactive: they post to the default channel)."""
 
 	def test_price(self):
+		"""The scheduled price report posts."""
 		price.lambda_handler()
 		self.assertTrue(self.sent)
 
 	def test_performance(self):
+		"""The scheduled performance report posts."""
 		performance.lambda_handler(past_days=7)
 		self.assertTrue(self.sent)
 
 	def test_shorts(self):
+		"""The scheduled shorts report posts."""
 		shorts.lambda_handler()
 		self.assertTrue(self.sent)
 
 	def test_earnings_and_dividends(self):
+		"""The scheduled earnings and ex-dividend reports post."""
 		for kwargs in ({'event': 'earnings'}, {'event': 'ex-dividend'}):
 			with self.subTest(kwargs=kwargs):
 				self.sent.clear()
@@ -226,10 +249,12 @@ def test_earnings_skip_an_adr_whose_primary_listing_yahoo_does_not_know(self):
 		self.assertNotIn('Co T1 ', str(self.sent))
 
 	def test_trades(self):
+		""".trades run from a chat posts."""
 		trades.lambda_handler(interactive=True, service='telegram', chat_id='5')
 		self.assertTrue(self.sent)
 
 	def test_trades_asks_yahoo_only_for_new_trades(self):
+		"""A scheduled trades run asks Yahoo only about trades it has not reported, and nothing once all are known."""
 		fetched = []
 		self.patch(yahoo, 'fetch', lambda tickers: fetched.append(set(tickers)) or {t: market_entry(t) for t in tickers})
 		trades.lambda_handler()
@@ -239,6 +264,7 @@ def test_trades_asks_yahoo_only_for_new_trades(self):
 		self.assertEqual(fetched, [])
 
 	def test_a_scheduled_run_keeps_the_trades_another_run_recorded_meanwhile(self):
+		"""The seen trades are read again as the run ends, so a trade another run recorded meanwhile is kept."""
 		meanwhile = lambda tickers: state.save('finbot_sharesight_trades.json', [999]) or {t: market_entry(t) for t in tickers}
 		self.patch(yahoo, 'fetch', meanwhile) # called after the seen trades were read, before the new ones are recorded
 		trades.lambda_handler()
@@ -246,6 +272,7 @@ def test_a_scheduled_run_keeps_the_trades_another_run_recorded_meanwhile(self):
 		self.assertEqual((known[0], len(known)), (999, 2))
 
 	def test_milestone(self):
+		"""The milestone report posts."""
 		milestone.lambda_handler()
 		self.assertTrue(self.sent)
 
@@ -274,16 +301,20 @@ def test_milestone_records_are_all_time(self):
 				self.assertEqual('Record high' in str(self.sent), expected)
 
 	def test_a_52_week_record_kept_before_is_replaced_by_the_history(self):
+		"""A record kept in the old 52-week format is replaced by the history: no false record high."""
 		state.save('finbot_milestone.json', {f'T{i}': {'high': 110.0, 'low': 90.0} for i in range(25)})
 		self.patch(yahoo, 'historic_high', lambda t: (150.0, 50.0))
 		milestone.lambda_handler() # 120 beats the old 52-week record of 110, but not the history's 150
 		self.assertNotIn('Record high', str(self.sent))
 
 	def test_rating(self):
+		""".rating run from a chat posts."""
 		rating.lambda_handler(interactive=True, service='telegram', chat_id='5')
 		self.assertTrue(self.sent)
 
 	def test_rating_keeps_the_last_rating_of_a_ticker_missing_from_a_run(self):
+		"""A ticker missing from one run keeps its last rating, so its change is reported when it comes back; it used to
+		be dropped from the record."""
 		rating.lambda_handler() # records every ticker's rating
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {} if t == 'T1' else {t: market_entry(t)})
 		rating.lambda_handler() # no T1 this time: its rating used to be dropped from the record
@@ -293,6 +324,7 @@ def test_rating_keeps_the_last_rating_of_a_ticker_missing_from_a_run(self):
 		self.assertIn('Co T1 ', str(self.sent)) # so its change from buy to sell went unreported
 
 	def test_an_interactive_rating_records_nothing(self):
+		"""A rating report asked for in a chat leaves the record for the scheduled runs alone."""
 		rating.lambda_handler(interactive=True, service='telegram', chat_id='5')
 		self.assertIsNone(state.load('finbot_rating.json'))
 
@@ -306,6 +338,7 @@ def test_slack_webhook_without_a_bot_token_gets_scheduled_reports(self):
 		self.patch(webhook, 'sendPhoto', REAL_SEND_PHOTO)
 		posts = []
 		def post(url, **kwargs):
+			"""Record (host, whether a bot token was sent, the channel) and answer ok."""
 			posts.append((url.split('/')[2], 'Authorization' in kwargs.get('headers', {}), (kwargs.get('json') or {}).get('channel')))
 			return mock.Mock(status_code=200, json=lambda: {'ok': True, 'result': {'message_id': 1, 'chat': {'id': 1}}})
 		self.patch(webhook.net, 'post', post)
@@ -317,6 +350,7 @@ def post(url, **kwargs):
 				self.assertTrue(any(host == 'api.telegram.org' for host, _, _ in posts))
 
 	def test_nowhere_to_post_is_an_error(self):
+		"""A scheduled report with nowhere to post is an error that says so."""
 		self.env(webhooks={})
 		with self.assertRaisesRegex(RuntimeError, 'nowhere to post'):
 			price.lambda_handler()
@@ -326,29 +360,35 @@ class PeriodButtons(FinbotCase):
 	"""Default views line up with the buttons: the highlighted button is the period actually shown."""
 
 	def buttons(self, command):
+		"""The labels of the period buttons a command's chart came with."""
 		self.command(command)
 		self.assertTrue(self.markups and self.markups[0], f'{command} sent no buttons')
 		return [b['text'] for b in self.markups[0]['inline_keyboard'][0]]
 
 	def test_performance_defaults_to_one_month(self):
+		""".performance shows a month, marked on its button, and says month, not 4 weeks."""
 		self.assertEqual(self.buttons('.performance'), ['7D', '● 1M', '3M', '1Y', 'Max'])
 		self.assertIn('month', self.sent[0][2]) # "Performance over the past month", not 4 weeks
 
 	def test_performance_period_argument_moves_the_marker(self):
+		"""A period on .performance moves the marker to its button."""
 		self.assertEqual(self.buttons('.performance 7d'), ['● 7D', '1M', '3M', '1Y', 'Max'])
 		self.assertEqual(self.buttons('.performance 1y'), ['7D', '1M', '3M', '● 1Y', 'Max'])
 
 	def test_price_list_has_a_1d_button_marked_by_default(self):
+		"""The price list offers 1D, marked by default, and a period moves the marker."""
 		self.assertEqual(self.buttons('.price'), ['● 1D', '7D', '1M', '3M', '1Y', 'Max'])
 		self.assertEqual(self.buttons('.price top'), ['● 1D', '7D', '1M', '3M', '1Y', 'Max'])
 		self.assertEqual(self.buttons('.price 1m'), ['1D', '7D', '● 1M', '3M', '1Y', 'Max'])
 
 	def test_other_charts_do_not_offer_1d(self):
+		"""Only the price list offers 1D."""
 		for command in ('.performance', '.compare T1 T2'):
 			with self.subTest(command=command):
 				self.assertNotIn('1D', [b.lstrip('● ') for b in self.buttons(command)])
 
 	def press(self, data):
+		"""Press a period button: (the edits made, the toasts shown)."""
 		edits, toasts = [], []
 		self.patch(webhook, 'editMessageMedia', lambda chat, msg, img, caption, markup=None: edits.append((caption.split('\n')[0], markup)))
 		self.patch(telegram, 'answerCallbackQuery', lambda cid, text=None: toasts.append(text))
@@ -356,6 +396,7 @@ def press(self, data):
 		return edits, toasts
 
 	def test_pressing_1d_rebuilds_todays_list_and_moves_the_marker(self):
+		"""Pressing 1D shows a toast, rebuilds today's list in place and marks 1D."""
 		edits, toasts = self.press('c|l|d|3,0')
 		self.assertEqual(toasts, ['Loading 1D…'])
 		self.assertEqual(self.errors, [])
@@ -363,6 +404,7 @@ def test_pressing_1d_rebuilds_todays_list_and_moves_the_marker(self):
 		self.assertEqual([b['text'] for b in edits[0][1]['inline_keyboard'][0]][0], '● 1D')
 
 	def test_a_button_a_chart_does_not_offer_is_ignored(self):
+		"""A button a chart does not offer, or an unknown period, changes nothing."""
 		edits, toasts = self.press('c|f|d|') # performance has no 1D
 		self.assertEqual((edits, toasts), ([], [None]))
 		edits, toasts = self.press('c|l|zz|3,0')
@@ -371,6 +413,8 @@ def test_a_button_a_chart_does_not_offer_is_ignored(self):
 
 class Podcasts(FinbotCase):
 	def test_format_helpers(self):
+		"""Durations in any of the feed formats, how long ago, and dates without a zone taken as UTC; nonsense gives
+		nothing."""
 		from lib.reports import podcasts
 		self.assertEqual(podcasts.format_duration('3725'), '01:02')
 		self.assertEqual(podcasts.format_duration('1:02:03'), '01:02')
@@ -384,6 +428,7 @@ def test_new_episodes_are_posted_once_newest_first(self):
 		"""A feed whose items mix dated and undated ones used to crash the sort (aware and naive times)."""
 		from lib.reports import podcasts
 		def item(n, date):
+			"""A feed item, its date optional."""
 			return f'<item><guid>g{n}</guid><title>Show: ep {n}</title><enclosure url="https://x/{n}.mp3"/>{date}<itunes:duration>600</itunes:duration></item>'
 		feed = ('<rss xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd"><channel><title>Show</title>'
 				+ item(1, '<pubDate>Mon, 01 Jan 2024 00:00:00 +0000</pubDate>') + item(2, '') + item(3, '<pubDate>Tue, 02 Jan 2024 00:00:00 GMT</pubDate>')
@@ -399,8 +444,10 @@ def item(n, date):
 
 class PieCommand(FinbotCase):
 	def setUp(self):
+		"""A portfolio of five holdings across three industries, one unclassified and one worth nothing."""
 		super().setUp()
 		def holdings(portfolio):
+			"""The five holdings, whatever the portfolio."""
 			data = [('A', 'Semiconductors', 'Electronic Technology', 500), ('B', 'Semiconductors', 'Electronic Technology', 300),
 					('C', 'Biotechnology', 'Health Technology', 200), ('D', None, None, 100), ('E', 'Gold', 'Minerals', 0)]
 			return {c + ':NASDAQ': {'code': c, 'market_code': 'NASDAQ', 'industry_classification_name': i,
@@ -408,17 +455,20 @@ def holdings(portfolio):
 		self.patch(sharesight.Portfolio, 'holdings', holdings)
 
 	def test_by_industry_default(self):
+		""".pie PORTFOLIO charts by industry."""
 		sent = self.command('.pie Rob')
 		self.assertEqual(sent[0][0], 'photo')
 		self.assertIn('Rob by industry', sent[0][2])
 
 	def test_by_sector_and_payload(self):
+		""".pie by sector: the slices in order, the unclassified ones together, worthless holdings left out."""
 		payload, chart = holdings.prepare_pie('rob', 'sector', TELEGRAM)
 		self.assertEqual([r[0] for r in chart['rows']], ['Electronic Technology', 'Health Technology', 'Unclassified'])
 		self.assertIn('Electronic Technology: 72.7%', payload) # 800 of 1100; zero-value Gold is left out
 		self.assertEqual(self.command('.pie Rob sector')[0][0], 'photo')
 
 	def test_folds_into_other_and_renders(self):
+		"""Over eight slices, the smallest fold into Other, and the chart draws."""
 		held = {f'T{i}:X': {'code': f'T{i}', 'industry_classification_name': f'Ind {i}', 'value': 100 - i} for i in range(12)}
 		self.patch(sharesight.Portfolio, 'holdings', lambda portfolio: held)
 		payload, chart = holdings.prepare_pie('Rob', 'industry', TELEGRAM)
@@ -427,13 +477,15 @@ def test_folds_into_other_and_renders(self):
 		self.assertTrue(render.pie_chart(chart['rows'], chart['title'], chart['subtitle']).read().startswith(b'\x89PNG'))
 
 	def test_unknown_or_missing_portfolio_lists_known(self):
+		""".pie with no portfolio, or an unknown one, lists the portfolios there are."""
 		for text, first in (('.pie', '.pie: please try again'), ('.pie Nope', 'Nope portfolio not found')):
 			sent = self.command(text)
 			self.assertTrue(sent[0][1][0].startswith(first), text)
 			self.assertIn('Rob', sent[0][1])
 
 	def test_a_service_that_takes_no_images_gets_text(self):
-		self.env(slackBotToken=False) # Slack replies carry charts only with the bot token
+		"""Slack without the bot token takes no images: the report goes as text."""
+		self.env(slackBotToken=False)
 		worker.Request(SLACK, 'C1', '@u', 'U', BOT).deliver(['x'], {'pie': True, 'rows': [('a', 1.0)], 'title': 't'})
 		self.assertEqual(self.sent[-1][0], 'text')
 
diff --git a/tests/test_util.py b/tests/test_util.py
index 643bc581..cbb11b57 100644
--- a/tests/test_util.py
+++ b/tests/test_util.py
@@ -6,6 +6,7 @@
 
 class Parsing(unittest.TestCase):
 	def test_days_from_human_days(self):
+		"""A period in days from 7, 7d, 2w, 3m, 1y or ytd."""
 		for arg, days in (('7', 7), ('7d', 7), ('7D', 7), ('2w', 14), ('3m', 90), ('1y', 365), ('0', 0)):
 			with self.subTest(arg=arg):
 				self.assertEqual(formats.days_from_human_days(arg), days)
@@ -13,11 +14,14 @@ def test_days_from_human_days(self):
 		self.assertEqual(formats.days_from_human_days('ytd'), (today - today.replace(month=1, day=1)).days)
 
 	def test_days_from_human_days_rejects_non_periods(self):
+		"""A ticker, an empty string or a malformed period is a ValueError, not a period."""
 		for arg in ('AAPL', '', 'x5d', '1.5y', 'week'):
 			with self.subTest(arg=arg), self.assertRaises(ValueError):
 				formats.days_from_human_days(arg)
 
 	def test_transform_to_yahoo(self):
+		"""Sharesight and Google tickers become Yahoo's: suffixes by market, none for US venues, a dash for class
+		shares."""
 		for args, expected in ((('AAPL',), 'AAPL'), (('FANG', 'ASX'), 'FANG.AX'), (('ARM', 'NASDAQ'), 'ARM'),
 							   (('ARM:NASDAQ',), 'ARM'), (('BHP.AX',), 'BHP.AX'), (('0700', 'HKG'), '0700.HK'),
 							   (('BRK', 'B'), 'BRK-B'), (('2330', 'TAI'), '2330.TW'), (('VOD', 'LSE'), 'VOD.L'),
@@ -25,7 +29,8 @@ def test_transform_to_yahoo(self):
 			with self.subTest(args=args):
 				self.assertEqual(markets.transform_to_yahoo(*args), expected)
 
-	def test_each_market_has_its_own_flag_and_currency(self): # Sharesight's market codes: TSE is Tokyo, TSX Toronto, SGX Singapore
+	def test_each_market_has_its_own_flag_and_currency(self):
+		"""Sharesight's market codes each have their own flag and currency: TSE is Tokyo, TSX Toronto, SGX Singapore."""
 		for market, ticker, flag, currency in (('TSE', '7203.T', '🇯🇵', 'JPY'), ('TSX', 'SHOP.TO', '🇨🇦', 'CAD'), ('NZX', 'AIR.NZ', '🇳🇿', 'NZD'),
 											   ('SGX', 'D05.SI', '🇸🇬', 'SGD'), ('CPSE', None, '🇩🇰', 'DKK')):
 			with self.subTest(market=market):
@@ -34,11 +39,14 @@ def test_each_market_has_its_own_flag_and_currency(self): # Sharesight's market
 					self.assertEqual(markets.flag_from_ticker(ticker), flag)
 
 	def test_make_paragraphs_keeps_every_word(self):
+		"""make_paragraphs breaks a long description into paragraphs at sentence ends without losing a word."""
 		self.assertEqual(formats.make_paragraphs('Acme makes widgets. It sells them.'), 'Acme makes widgets. It sells them.')
 		long = ' '.join(['word'] * 22) + ' end. Tail sentence here.'
 		self.assertEqual(formats.make_paragraphs(long), ' '.join(['word'] * 22) + ' end.\n\nTail sentence here.')
 
 	def test_transform_title(self):
+		"""Company names lose shouting and legal suffixes (Inc, FPO, SE, a trailing &), from the end only, leaving no
+		run of spaces."""
 		self.assertEqual(formats.transform_title('APPLE INC'), 'Apple')
 		self.assertEqual(formats.transform_title('Acme FPO'), 'Acme')
 		self.assertEqual(formats.transform_title('Mixed Case Name'), 'Mixed Case Name')
@@ -50,17 +58,20 @@ def test_transform_title(self):
 
 class Formatting(unittest.TestCase):
 	def test_signed_percent(self):
+		"""A signed percent: a plus for gains, thousands separators, and the decimals asked for."""
 		for value, decimals, expected in ((12, None, '+12%'), (-3.5, None, '-3.5%'), (0, None, '0%'), (1234, None, '+1,234%'),
 										  (2.345, 1, '+2.3%'), (-0.04, 1, '-0.0%')):
 			with self.subTest(value=value):
 				self.assertEqual(formats.signed_percent(value, decimals), expected)
 
 	def test_days_english(self):
+		"""None and 0 are today; 1 and 7 days are the past day and week."""
 		for days, expected in ((None, 'today'), (0, 'today'), (1, 'the past day'), (7, 'the past week')):
 			with self.subTest(days=days):
 				self.assertEqual(formats.days_english(days), expected)
 
 	def test_date_range_english(self):
+		"""A span ending on a date, in words, across a year end too; a period is required."""
 		end = datetime.date(2026, 10, 5)
 		self.assertEqual(formats.date_range_english(7, end), '28 Sep - 5 Oct 2026')
 		self.assertEqual(formats.date_range_english(30, end), '5 Sep - 5 Oct 2026')
@@ -69,20 +80,24 @@ def test_date_range_english(self):
 			formats.date_range_english(None, end) # callers must pass a period
 
 	def test_ordinal(self):
+		"""Ordinal suffixes, the teens included."""
 		for number, suffix in ((1, 'st'), (2, 'nd'), (3, 'rd'), (4, 'th'), (11, 'th'), (12, 'th'), (13, 'th'), (21, 'st'), (112, 'th')):
 			with self.subTest(number=number):
 				self.assertEqual(formats.ordinal(number), suffix)
 
 	def test_get_emoji(self):
+		"""An arrow up for a rise, down for a fall, a square for none."""
 		self.assertEqual([formats.get_emoji(n) for n in (1, -1, 0)], ['🔼', '🔻', '▪️'])
 
 	def test_links_per_service(self):
+		"""A link is HTML on Telegram and <url|text> on Slack."""
 		telegram_link = TELEGRAM.link('https://example.com/x', 'text')
 		slack_link = SLACK.link('https://example.com/x', 'text')
 		self.assertIn('<a href=', telegram_link)
 		self.assertIn('|text', slack_link)
 
 	def test_flags_for_known_and_unknown_markets(self):
+		"""Known suffixes and US tickers have flags; an unknown suffix gives a string, not a crash."""
 		self.assertTrue(markets.flag_from_ticker('BHP.AX'))
 		self.assertTrue(markets.flag_from_ticker('AAPL'))
 		self.assertIsInstance(markets.flag_from_ticker('XYZ.ZZ'), str) # unknown suffix must not crash
@@ -90,16 +105,20 @@ def test_flags_for_known_and_unknown_markets(self):
 
 class StateFiles(FinbotCase):
 	def test_json_roundtrip_and_missing_file(self):
+		"""A state file reads back what was saved; a missing one is None."""
 		self.assertIsNone(state.load('absent.json'))
 		state.save('state.json', {'a': [1, 2]})
 		self.assertEqual(state.load('state.json'), {'a': [1, 2]})
 
 	def test_a_failed_write_leaves_the_old_file_and_the_umask_alone(self):
+		"""A write that fails (disk full at replace or fsync) leaves the old file and no temp file; the umask is
+		untouched and files are 0640."""
 		state.save('state.json', {'a': 1})
 		path = Path(config.env.data_dir, 'state.json')
 		previous = os.umask(0o027)
 		self.addCleanup(os.umask, previous)
 		def disk_full(*a, **k):
+			"""Fail like a full disk."""
 			raise OSError(28, 'No space left on device')
 		for broken in ('replace', 'fsync'):
 			with self.subTest(broken=broken), mock.patch.object(cache.os, broken, disk_full):
@@ -115,6 +134,8 @@ def disk_full(*a, **k):
 		self.assertEqual((json.loads(path.read_text()), oct(path.stat().st_mode & 0o777)), ({'a': 3}, '0o640'))
 
 	def test_edit_writes_only_a_change_and_never_changes_the_default(self):
+		"""state.edit writes only when the block changed the data, never alters the default it was given, and writes
+		nothing when the block fails."""
 		default = {'seen': []}
 		with state.edit('edit.json', default):
 			pass
@@ -128,8 +149,10 @@ def test_edit_writes_only_a_change_and_never_changes_the_default(self):
 		self.assertEqual(state.load('edit.json'), {'seen': [1]}) # a block that fails writes nothing
 
 	def test_edits_at_once_all_count(self):
-		# each edit reads the file after the one before it wrote, so none overwrites another with what it read earlier
+		"""Each edit reads the file after the one before it wrote, so none overwrites another with what it read
+		earlier."""
 		def edit(key):
+			"""Set data[key] slowly, so that the other edits start meanwhile."""
 			with state.edit('edit.json', {}) as data:
 				time.sleep(0.01) # the other edits start meanwhile
 				data[key] = True
@@ -141,14 +164,17 @@ def edit(key):
 		self.assertEqual(state.load('edit.json'), dict.fromkeys('abc', True))
 
 	def test_read_cache_missing_or_expired_is_falsy(self):
+		"""A missing or expired cache file reads as nothing."""
 		self.assertFalse(cache.read('absent.json', 60))
 		cache.write('old.json', {'a': 1})
 		self.assertFalse(cache.read('old.json', -1)) # already expired
 
 	def test_a_cache_file_that_goes_while_it_is_read_is_a_miss(self):
+		"""A cache file deleted between the age check and the read (by cleanup or a flush) is a miss, not an error."""
 		cache.write('gone.json', {'a': 1})
 		real = Path.read_text
-		def vanish(path, *a, **k): # cleanup or a flush deleted it between the age check and the read
+		def vanish(path, *a, **k):
+			"""Delete gone.json as it is read, as cleanup or a flush might between the age check and the read."""
 			if path.name == 'gone.json':
 				path.unlink()
 				raise FileNotFoundError(path)
@@ -157,12 +183,13 @@ def vanish(path, *a, **k): # cleanup or a flush deleted it between the age check
 			self.assertIsNone(cache.read('gone.json', 60))
 
 	def test_binary_cache_roundtrip(self):
+		"""A drawn chart reads back from the cache as written."""
 		self.assertFalse(cache.read_bytes('absent.png'))
 		cache.write_bytes('img.png', io.BytesIO(b'png-bytes'))
 		self.assertEqual(cache.read_bytes('img.png').read(), b'png-bytes')
 
 	def test_holdings_and_watchlist_without_a_watchlist_file(self):
-		# first run: no watchlist exists yet
+		"""On the first run there is no watchlist file: the holdings alone, GOOGL dropped when GOOG is held."""
 		self.patch(sharesight, 'tickers', lambda: ['AAPL', 'GOOG', 'GOOGL'])
 		self.patch(watchlist, 'with_holdings', REAL_GET_HOLDINGS)
 		self.assertEqual(watchlist.with_holdings(), ['AAPL', 'GOOG']) # GOOGL dropped when GOOG is present

From fda8ed2b0df9277c6fd0a6c53c80acd2df277b49 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 00:45:17 +0000
Subject: [PATCH 18/81] util/record_outputs.py: the old/new output comparison,
 now in the repo

The harness that checked each refactor so far records every HTTP request the bot posts for 159 steps (each chat command
on Telegram and Slack, the period and job buttons, every scheduled report) on the offline fixtures, and compares two
recordings step by step. --against COMMIT (HEAD by default) records that commit, checked out in a temporary worktree,
and this tree in the same run, so dates in the output always match; --diff compares two saved recordings; --record
writes one. The exit status is 1 when they differ.

tests/common.py now names what its fixture replaces with recorders (REAL_SENDING), so the tool knows what to put back
on either side of a change to how the tests capture output. The pre-M1 branches of the scratch version are gone. A test
covers the comparison, and CLAUDE.md points to the tool instead of the by-hand method.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md              |   7 +-
 tests/common.py        |   3 +
 tests/test_finbot.py   |  22 +++++
 util/record_outputs.py | 184 +++++++++++++++++++++++++++++++++++++++++
 4 files changed, 215 insertions(+), 1 deletion(-)
 create mode 100755 util/record_outputs.py

diff --git a/CLAUDE.md b/CLAUDE.md
index d461896d..d9c22b2c 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -21,7 +21,12 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - Python 3.13.5; tests are offline and need no network (`tests/`). CI runs them on the latest 3.x and 3.13.5, except for pull requests that change only README.md, CLAUDE.md or `img/*.png` (install, compile and tests are skipped; the allowlist in `.github/workflows/ci.yml` must never include a path a test reads).
 - Production runs the Debian system packages (including `python3-croniter`, for the scheduler in `lib/jobs.py`); `requirements.txt` is for pip/uv users and CI.
 - A new preference goes in five places: a `Prefs` field and its line in `preference_values` (`lib/config.py`), `defaults/preferences.conf`, the README's preferences table, and `DEFAULTS` plus the key count in `tests/test_config.py`. The tests catch all but the README (no test may read README.md: CI skips the tests for a README-only change).
-- If Yahoo and Sharesight are unreachable (the cloud environment's network policy can block them), check behaviour against the offline fixtures in `tests/common.py`. For a refactor, compare old and new output: load `git show HEAD:<file>` as a module beside the new one and diff what both produce for the same inputs.
+- If Yahoo and Sharesight are unreachable (the cloud environment's network policy can block them), check behaviour
+  against the offline fixtures in `tests/common.py`.
+- For a refactor, `util/record_outputs.py` (or `--against COMMIT`) records what every command, button and scheduled
+  report posts, for this tree and for HEAD in the same run, and prints the steps that differ (exit status 1 when any
+  do). A change to how the tests capture output goes in `tests/common.py`'s `REAL_SENDING`, which the tool puts back to
+  record real requests.
 
 ## Code map (open the one file you need)
 - `bot.py`: WSGI entry; routes /telegram and /slack, authenticates, hands each message to `lib/worker.py` in a thread; SIGHUP reload, SIGUSR1
diff --git a/tests/common.py b/tests/common.py
index bf2cd837..2d43fd4c 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -19,6 +19,9 @@
 REAL_PAYLOAD_WRAPPER = webhook.payload_wrapper # FinbotCase replaces it; tests of the real thing use this
 REAL_REPORT_ERROR = webhook.report_error # likewise
 REAL_SEND_PHOTO = webhook.sendPhoto # likewise
+# what FinbotCase replaces with recorders; util/record_outputs.py puts them back to record the real requests
+REAL_SENDING = ((webhook, 'payload_wrapper', REAL_PAYLOAD_WRAPPER), (webhook, 'sendPhoto', REAL_SEND_PHOTO),
+	(webhook, 'report_error', REAL_REPORT_ERROR))
 TICKERS = [f'T{i}' for i in range(25)]
 TELEGRAM, SLACK, DISCORD = dialects.TELEGRAM, dialects.SLACK, dialects.DISCORD
 BOT = '@bot'
diff --git a/tests/test_finbot.py b/tests/test_finbot.py
index c3253a52..d0f96789 100644
--- a/tests/test_finbot.py
+++ b/tests/test_finbot.py
@@ -89,6 +89,28 @@ def test_the_bot_and_its_reports_print_only_to_stderr(self):
 		self.assertEqual(offenders, [])
 
 
+class OutputComparison(unittest.TestCase):
+	def test_steps_are_compared_by_name_and_a_repeated_one_by_its_count(self):
+		"""util/record_outputs.py --diff shows a step that changed, appeared or went; a step run twice is told apart."""
+		from util import record_outputs
+		with tempfile.TemporaryDirectory() as tmp:
+			old, new = Path(tmp, 'old.json'), Path(tmp, 'new.json')
+			post = lambda text: {'url': 'u', 'json': {'text': text}}
+			step = lambda label, *texts: {'step': label, 'posts': [post(t) for t in texts], 'raised': None}
+			old.write_text(json.dumps([step('add', 'Ok'), step('add', 'Again'), step('gone', 'x')]))
+			new.write_text(json.dumps([step('add', 'Ok'), step('add', 'Once more'), step('new', 'y')]))
+			out = io.StringIO()
+			with contextlib.redirect_stdout(out):
+				self.assertEqual(record_outputs.compare(old, new), 3)
+				self.assertEqual(record_outputs.compare(old, old), 0)
+		shown = out.getvalue()
+		self.assertIn('== add (2)\n', shown)
+		self.assertIn('-    "text": "Again"', shown)
+		self.assertIn('== gone (only in the old recording)', shown)
+		self.assertIn('== new (only in the new recording)', shown)
+		self.assertIn('3 steps, identical', shown)
+
+
 class GitVersion(unittest.TestCase):
 	SHA = '0123456789abcdef0123456789abcdef01234567'
 
diff --git a/util/record_outputs.py b/util/record_outputs.py
new file mode 100755
index 00000000..825a13aa
--- /dev/null
+++ b/util/record_outputs.py
@@ -0,0 +1,184 @@
+#!/usr/bin/env python3
+"""Old/new output comparison for refactors. It records every HTTP request the bot posts (the URL with the bot token
+redacted, the JSON or form body, a hash of each file) for a fixed list of chat commands, button presses and scheduled
+reports, run on the offline fixtures of tests/common.py, and compares two recordings step by step.
+
+	util/record_outputs.py                       this tree (uncommitted changes included) against HEAD
+	util/record_outputs.py --against COMMIT      this tree against COMMIT; both are recorded in this run, so dates match
+	util/record_outputs.py --record OUT.json     record the tree in the current directory only
+	util/record_outputs.py --diff OLD.json NEW.json
+
+--keep DIR keeps the two recordings; --full shows changed lines whole. The exit status is 0 when the recordings match,
+1 when they differ."""
+import argparse, difflib, hashlib, io, json, os, random, subprocess, sys, tempfile, unittest
+from pathlib import Path
+
+REPO = Path(__file__).resolve().parent.parent
+COMMANDS = ['.help', '/start', '.hello', '.thanks', '.who T1', '.plan buy low', '.plan', '.holdings', '.holdings Rob',
+	'.marketcap', '.marketcap bottom', '.marketcap T1', '.pe', '.pe bottom', '.pe T1', '.fpe', '.fpe bottom',
+	'.fpe negative', '.forward pe T1', '.peg', '.peg negative', '.peg T1', '.beta', '.buy', '.sell', '.buy & sell',
+	'.earnings', '.earnings T1', '.dividend', '.dividend T1', '.trades', '.trades 30d', '.trades 7d Nope', '.price',
+	'.price 7d', '.price 3', '.price 3 7d', '.price top', '.price top 7d', '.price T1', '.price T1 7d', '.session',
+	'.session T1', '.premarket', '.postmarket', '.shorts', '.shorts 30', '.shorts T1', '.performance',
+	'.performance 7d', '.performance 1y', '.performance Nope', '.pie Rob', '.pie Rob sector', '.pie', '.compare T1 T2',
+	'.compare T1 T2 3m', '.compare T1', '.history T1', '.history', '.watchlist', '.watchlist add MSFT',
+	'.watchlist add MSFT', '.watchlist del MSFT', '.watchlist frob X', '.T1', '.jobs', 'hi @bot', 'noise']
+BUTTONS = ['c|h|m|T1', 'c|h|x|T1', 'c|p|w|T1', 'c|c|q|T1,T2', 'c|f|y|', 'c|l|d|3,0', 'c|l|m|0,10', 'c|zz|m|T1']
+SCHEDULED = ['trades.py', 'price.py midsession', 'price.py interday', 'price.py premarket', 'price.py close',
+	'price.py 7', 'cal.py earnings', 'cal.py ex-dividend', 'performance.py 7', 'rating.py', 'shorts.py', 'milestone.py',
+	'reminder.py']
+WEBHOOKS = {'telegram': 'https://api.telegram.org/botX/', 'slack': 'https://hooks.slack.com/services/T/B/x',
+	'discord': 'https://discord.com/api/webhooks/1/abc'}
+ANSWER = {'ok': True, 'upload_url': 'https://files.slack.com/u', 'file_id': 'F1', 'id': '1'} # good enough for every API
+
+
+def digest(value):
+	"""A short hash of a file's bytes (or a file object's, or text), for a recording that stays small."""
+	if hasattr(value, 'read'):
+		if hasattr(value, 'seek'):
+			value.seek(0)
+		value = value.read()
+	if isinstance(value, str):
+		value = value.encode()
+	return 'sha256:' + hashlib.sha256(value).hexdigest()[:16]
+
+
+def record(output):
+	"""Run every step on the tree in the current directory and write what each posted to `output` (JSON)."""
+	here = Path(__file__).resolve().parent
+	sys.path[:] = [os.getcwd()] + [path for path in sys.path if Path(path or '.').resolve() != here] # the tree's code
+	from tests import common
+	from lib import jobs, net, webhook, worker
+	real_sending = getattr(common, 'REAL_SENDING', None) or ( # a commit from before REAL_SENDING
+		(webhook, 'payload_wrapper', common.REAL_PAYLOAD_WRAPPER), (webhook, 'sendPhoto', common.REAL_SEND_PHOTO),
+		(webhook, 'report_error', common.REAL_REPORT_ERROR))
+
+	class Record(common.FinbotCase):
+		"""The steps, as one test, so that FinbotCase's fixtures (and their cleanup) apply."""
+		def runTest(self):
+			"""Record what each step posts: the real sending code runs, and net.post notes each request."""
+			posts = []
+			def post(url, **kwargs):
+				"""Note a request, the typing notice aside (timing, not output), and answer like the APIs do."""
+				if 'sendChatAction' in url:
+					return common.mock.Mock(status_code=200, json=lambda: {'ok': True, 'result': True})
+				entry = {'url': url.replace('botX/', 'bot<token>/')}
+				for key in ('json', 'data', 'params'):
+					body = kwargs.get(key)
+					if isinstance(body, dict):
+						entry[key] = {k: digest(v) if isinstance(v, (bytes, bytearray)) else v for k, v in body.items()}
+					elif body is not None:
+						entry[key] = digest(body)
+				if kwargs.get('files'):
+					entry['files'] = {name: [part[0], digest(part[1])] for name, part in kwargs['files'].items()}
+				posts.append(entry)
+				answer = {**ANSWER, 'result': {'message_id': len(posts), 'chat': {'id': 5}}}
+				return common.mock.Mock(status_code=200, json=lambda: answer, text='OK')
+			self.patch(net, 'post', post)
+			for target, name, real in real_sending: # what the fixture replaces with recorders, put back
+				self.patch(target, name, real)
+			self.patch(common.time, 'sleep', lambda seconds: None) # no pauses between a long report's messages
+			self.env(slackBotToken='xoxb-test')
+			steps = []
+			def step(label, run):
+				"""Run one step and note what it posted, or what it raised."""
+				posts.clear()
+				random.seed(0)
+				try:
+					run()
+					raised = None
+				except Exception as e:
+					raised = f'{type(e).__name__}: {e}'
+				steps.append({'step': label, 'posts': list(posts), 'raised': raised})
+			for service, chat, user in (('telegram', '55', '@u'), ('slack', 'C1', '<@U1>')):
+				for text in COMMANDS:
+					run = lambda: worker.process_request(service, chat, user, text, common.BOT, 'U')
+					step(f'{service} {text}', run)
+			for data in BUTTONS:
+				step(f'button {data}', lambda: worker.process_callback('telegram', 'cb', '55', '9', data))
+			self.env(webhooks=WEBHOOKS, telegramChatID='1')
+			for line in SCHEDULED:
+				label, module, kwargs = jobs.parse_command(line)[0]
+				step(f'scheduled {line}', lambda: module.lambda_handler(**kwargs))
+			Path(output).write_text(json.dumps(steps, indent=1, sort_keys=True, default=str))
+
+	result = unittest.TextTestRunner(verbosity=0, stream=io.StringIO()).run(Record())
+	for _, trace in result.errors + result.failures:
+		print(trace, file=sys.stderr)
+	return 0 if result.wasSuccessful() and Path(output).exists() else 2
+
+
+def record_tree(tree, output):
+	"""Record the tree at `tree` in a fresh interpreter of its own, with a data dir of its own. Exits when it fails."""
+	with tempfile.TemporaryDirectory() as data:
+		command = [sys.executable, str(Path(__file__).resolve()), '--record', str(output)]
+		run = subprocess.run(command, cwd=tree, env={**os.environ, 'DATA_DIR': data}, stdout=subprocess.DEVNULL,
+			stderr=subprocess.PIPE, text=True)
+	if run.returncode:
+		raise SystemExit(f'recording {tree} failed:\n{run.stderr[-3000:]}')
+
+
+def compare(old, new, full=False):
+	"""Print each step whose posts differ between two recordings, as a diff of its JSON. Returns how many differ."""
+	def by_label(path):
+		"""A recording's steps by label; a step run twice (the second .watchlist add) is told apart by a count."""
+		steps, seen = {}, {}
+		for step in json.loads(Path(path).read_text()):
+			seen[step['step']] = seen.get(step['step'], 0) + 1
+			steps[step['step'] + (f' ({seen[step["step"]]})' if seen[step['step']] > 1 else '')] = step
+		return steps
+	old_steps, new_steps = by_label(old), by_label(new)
+	differ = 0
+	for label in dict.fromkeys([*old_steps, *new_steps]):
+		a, b = old_steps.get(label), new_steps.get(label)
+		if a == b:
+			continue
+		differ += 1
+		where = ' (only in the new recording)' if a is None else ' (only in the old recording)' if b is None else ''
+		print(f'== {label}{where}')
+		if a and b:
+			lines = difflib.unified_diff(json.dumps(a, indent=1, sort_keys=True).splitlines(),
+				json.dumps(b, indent=1, sort_keys=True).splitlines(), lineterm='', n=1)
+			for line in list(lines)[2:]: # without the ---/+++ header
+				print(line if full or len(line) <= 220 else line[:217] + '...')
+	print(f'{len(new_steps)} steps, {differ} differ' if differ else f'{len(new_steps)} steps, identical')
+	return differ
+
+
+def against(commit, keep=None, full=False):
+	"""Record `commit` (checked out in a temporary worktree) and this tree one after the other, then compare them."""
+	with tempfile.TemporaryDirectory() as tmp:
+		out = Path(keep) if keep else Path(tmp)
+		out.mkdir(parents=True, exist_ok=True)
+		tree = Path(tmp, 'tree')
+		subprocess.run(['git', 'worktree', 'add', '--quiet', '--detach', str(tree), commit], cwd=REPO, check=True)
+		try:
+			record_tree(tree, out / 'old.json')
+			record_tree(REPO, out / 'new.json')
+		finally:
+			subprocess.run(['git', 'worktree', 'remove', '--force', str(tree)], cwd=REPO, check=True)
+		if keep:
+			print(f'recordings: {out / "old.json"} ({commit}), {out / "new.json"} (this tree)')
+		return compare(out / 'old.json', out / 'new.json', full)
+
+
+def main(argv):
+	"""Parse the arguments and run the mode they ask for; returns the exit status."""
+	parser = argparse.ArgumentParser(description=__doc__.split('\n\n')[0],
+		formatter_class=argparse.RawDescriptionHelpFormatter)
+	mode = parser.add_mutually_exclusive_group()
+	mode.add_argument('--against', metavar='COMMIT', default='HEAD', help='compare this tree with COMMIT (HEAD)')
+	mode.add_argument('--record', metavar='OUT', help='record the tree in the current directory to OUT')
+	mode.add_argument('--diff', nargs=2, metavar=('OLD', 'NEW'), help='compare two recordings')
+	parser.add_argument('--keep', metavar='DIR', help='with --against: keep the two recordings in DIR')
+	parser.add_argument('--full', action='store_true', help='show changed lines whole')
+	args = parser.parse_args(argv)
+	if args.record:
+		return record(os.path.abspath(args.record))
+	if args.diff:
+		return 1 if compare(*args.diff, full=args.full) else 0
+	return 1 if against(args.against, args.keep, args.full) else 0
+
+
+if __name__ == '__main__':
+	sys.exit(main(sys.argv[1:]))

From 8b6fa8fc2ef75ebb3b058a457ed07da11152b73e Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 01:06:06 +0000
Subject: [PATCH 19/81] Transports and one delivery path

lib/transports.py has one class per way of posting: TelegramBot,
SlackBot, SlackWebhook and DiscordWebhook. Each one knows its dialect,
what it can carry (charts, buttons, a typing notice) and the network's
size limits. Transport.text() splits long reports and Transport.chart()
fits the caption; each network posts one message its own way.

lib/webhook.py becomes lib/delivery.py:
- Report(lines, chart, buttons) and deliver(dest, report), which draws
  a chart spec only where it can go;
- reply_to() for the chat that asked and default_chats() for scheduled
  reports, with destinations() choosing between them;
- report_error() and the CLI entry point.

Other changes:
- worker: Request.reply() replaces say/send/deliver. Every command
  shows the typing notice after half a second, in place of 18
  `with request.typing()` blocks.
- Buttons are neutral, buttons.Button(label, data, active), and
  TelegramBot.markup() turns them into an inline keyboard. Button
  presses edit through the transport (edit_chart, edit_text).
- cal pins its Telegram messages through the transport.
- util/record_outputs.py imports lib.webhook only for a commit from
  before REAL_SENDING.
- The CLAUDE.md code map is updated.

The output comparison (util/record_outputs.py) against the previous
commit is identical in all 159 steps.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md                  |  12 +-
 bot.py                     |  10 +-
 bot_send_stdin.py          |   6 +-
 lib/buttons.py             |  19 ++-
 lib/charts.py              |   4 +-
 lib/delivery.py            | 132 +++++++++++++++
 lib/jobrun.py              |   4 +-
 lib/jobs.py                |  20 ++-
 lib/render.py              |   6 +
 lib/reports/__init__.py    |   4 +-
 lib/reports/about.py       |   8 +-
 lib/reports/cal.py         |  24 +--
 lib/reports/milestone.py   |   7 +-
 lib/reports/performance.py |  13 +-
 lib/reports/podcasts.py    |   7 +-
 lib/reports/price.py       |  20 ++-
 lib/reports/rating.py      |   7 +-
 lib/reports/reminder.py    |   7 +-
 lib/reports/shorts.py      |   9 +-
 lib/reports/trades.py      |   7 +-
 lib/transports.py          | 324 +++++++++++++++++++++++++++++++++++
 lib/webhook.py             | 339 -------------------------------------
 lib/worker.py              | 209 ++++++++++-------------
 tests/common.py            |  35 ++--
 tests/test_edges.py        |  20 ++-
 tests/test_finbot.py       |   9 +-
 tests/test_jobs.py         |  25 +--
 tests/test_providers.py    |  91 +++++-----
 tests/test_reports.py      |  11 +-
 util/record_outputs.py     |  10 +-
 30 files changed, 763 insertions(+), 636 deletions(-)
 create mode 100644 lib/delivery.py
 create mode 100644 lib/transports.py
 delete mode 100644 lib/webhook.py

diff --git a/CLAUDE.md b/CLAUDE.md
index d9c22b2c..5374f70d 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -31,14 +31,18 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 ## Code map (open the one file you need)
 - `bot.py`: WSGI entry; routes /telegram and /slack, authenticates, hands each message to `lib/worker.py` in a thread; SIGHUP reload, SIGUSR1
   cache flush.
-- `lib/worker.py`: `COMMANDS` (regex → `on_*` handler, first match wins, `.SYMBOL` last); `Request`; the typing indicator.
+- `lib/worker.py`: `COMMANDS` (regex → `on_*` handler, first match wins, `.SYMBOL` last); `Request`
+  (`request.reply(lines, chart, buttons)`); the typing notice, which every command gets after half a second.
 - `lib/reports/`: what finbot says. Scheduled (each has `lambda_handler()` and `job(argv)`, its schedule-line parser): `price
   performance trades shorts cal rating milestone reminder podcasts cleanup`; the top-level `price.py` and the rest are shims
   that run them. Chat-only (`prepare_*`): `about` (.help .jobs), `history` (.history .compare), `holdings` (.holdings .who .pie
   .watchlist), `profile` (.SYMBOL), `rankings` (.marketcap .pe .fpe .peg .beta .buy .sell).
-- `lib/webhook.py`: sending (chunking, photos, Slack upload), markup, escaping, error reporting. `lib/telegram.py`: Bot API calls.
-- `lib/buttons.py`: period buttons under a chart. `lib/charts.py`: a button press rebuilds its chart; the Telegram command
-  keyboard.
+- `lib/delivery.py`: `Report(lines, chart, buttons)`, the destinations (the chat that asked, or the default chats),
+  `deliver(dest, report)`, the one-line error report. `lib/transports.py`: a class per way of posting (`TelegramBot`,
+  `SlackBot`, `SlackWebhook`, `DiscordWebhook`): splitting, captions, charts, buttons, edits, typing. `lib/telegram.py`:
+  Bot API calls.
+- `lib/buttons.py`: `Button` and a chart's period buttons (each transport draws them). `lib/charts.py`: a button press
+  rebuilds its chart; the Telegram command keyboard.
 - `lib/dialects.py`: each network's markup and message sizes. `lib/markets.py`: market tables, flags, ticker translation.
   `lib/formats.py`: names, units, dates, percents in words. `lib/links.py`: stock links. `lib/render.py`: every chart.
 - `lib/yahoo.py lib/sharesight.py lib/shortman.py lib/simplywallst.py`: data sources, cached in `DATA_DIR/cache`.
diff --git a/bot.py b/bot.py
index a187320c..35cd50e8 100755
--- a/bot.py
+++ b/bot.py
@@ -19,7 +19,7 @@
 from lib import sharesight
 from lib import net
 from lib import telegram
-from lib import webhook
+from lib import delivery
 from lib import yahoo
 from lib import render
 from lib.reports import cleanup
@@ -50,10 +50,10 @@ def main(environ, start_response):
 		return handle(environ, start_response)
 	except Exception as e:
 		if environ.get(AUTHENTICATED):
-			webhook.report_error(e, context=environ.get('PATH_INFO', 'bot'))
+			delivery.report_error(e, context=environ.get('PATH_INFO', 'bot'))
 			status = '200 OK'
 		else:
-			print(f"request to {environ.get('PATH_INFO')!r} failed:", webhook.error_line(e), file=sys.stderr)
+			print(f"request to {environ.get('PATH_INFO')!r} failed:", delivery.error_line(e), file=sys.stderr)
 			print(net.redact(traceback.format_exc()), end='', file=sys.stderr)
 			status = '500 Internal Server Error'
 		try:
@@ -230,7 +230,7 @@ def flush_cache(*args):
 		sharesight.wake()
 		print(f"cache: flushed {removed} file{'' if removed == 1 else 's'} from {config.env.cache_dir}", file=sys.stderr)
 	except Exception as e:
-		print('ERROR', webhook.error_line(e, 'cache flush'), file=sys.stderr)
+		print('ERROR', delivery.error_line(e, 'cache flush'), file=sys.stderr)
 
 if __name__ == '__main__':
 	print(f'Starting finbot {git_version()}', file=sys.stderr)
@@ -265,7 +265,7 @@ def flush_cache(*args):
 		try:
 			botName = telegram.register() # here and only here: the job processes import lib.telegram too
 		except Exception as e:
-			print('ERROR', webhook.error_line(e, 'telegram webhook'), file=sys.stderr)
+			print('ERROR', delivery.error_line(e, 'telegram webhook'), file=sys.stderr)
 			sys.exit(1)
 	if config.env.telegramBotToken:
 		threading.Thread(target=charts.refresh_keyboards, daemon=True).start() # update stale DM keyboards in the background
diff --git a/bot_send_stdin.py b/bot_send_stdin.py
index 6f7849c1..3e9851a2 100755
--- a/bot_send_stdin.py
+++ b/bot_send_stdin.py
@@ -2,13 +2,13 @@
 
 import sys
 import os
-from lib import webhook
+from lib import delivery
 
 if os.isatty(sys.stdin.fileno()):
 	raise RuntimeError("No stdin input detected (not piped or redirected)")
 
 payload = [line.strip() for line in sys.stdin if line.strip()]
 
-for service, url, chat in webhook.default_chats():
-	webhook.send(service, url, chat, payload)
+for dest in delivery.default_chats():
+	delivery.deliver(dest, delivery.Report(payload))
 
diff --git a/lib/buttons.py b/lib/buttons.py
index 685cc51c..e5f6fdd1 100644
--- a/lib/buttons.py
+++ b/lib/buttons.py
@@ -1,8 +1,17 @@
-"""Period buttons (Week | Month | Quarter | Max) under a chart: their markup, and the callback data that names the tickers."""
+"""Buttons under a message: the period buttons (Week | Month | Quarter | Max) of a chart, and the callback data that
+names its tickers. Each transport draws Button rows its own way."""
 import hashlib
+import typing
 
 from lib import state
 
+class Button(typing.NamedTuple):
+	"""One button: its label, the data a press sends back, and whether it is the one shown now (marked)."""
+	label: str
+	data: str
+	active: bool = False
+
+
 PERIODS = {'d': ('1D', 1), 'w': ('7D', 7), 'm': ('1M', 30), 'q': ('3M', 90), 'y': ('1Y', 365), 'x': ('Max', None)}
 BUTTONS = {'h': 'wmqyx', 'p': 'wmqyx', 'c': 'wmqyx', 'f': 'wmqyx', 'l': 'dwmqyx'} # chart kind -> buttons offered (h=history, p=price chart, c=compare, f=performance, l=price list)
 MAX_DAYS = 3650 # what Max means for the Sharesight-based reports
@@ -37,10 +46,6 @@ def resolve(ref):
 	return ref.split(',')
 
 def keyboard(kind, tickers, active=None):
-	"""Telegram inline keyboard; the active period is marked."""
+	"""A chart's period buttons, one row, the active period marked."""
 	ref = _ref(tickers)
-	row = []
-	for key in BUTTONS[kind]:
-		label = PERIODS[key][0]
-		row.append({'text': ('● ' + label) if key == active else label, 'callback_data': f"c|{kind}|{key}|{ref}"})
-	return {'inline_keyboard': [row]}
+	return [[Button(PERIODS[key][0], f"c|{kind}|{key}|{ref}", key == active) for key in BUTTONS[kind]]]
diff --git a/lib/charts.py b/lib/charts.py
index a3e0f5cd..46197975 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -7,7 +7,7 @@
 from lib import formats
 from lib import links
 from lib import state
-from lib import webhook
+from lib import transports
 from lib import yahoo
 from lib.reports import history, performance, price
 
@@ -32,7 +32,7 @@ def mark_keyboard(chat_id):
 
 def push_keyboard(chat_id):
 	"""Telegram can only change a reply keyboard by sending a message, so send a short one carrying it."""
-	result = webhook.write('telegram', webhook.chat_url('telegram', chat_id), '⌨️ Keyboard updated', chat_id, reply_markup=command_keyboard())
+	result = transports.TelegramBot(chat_id).post_text('⌨️ Keyboard updated', command_keyboard())
 	if result is not None:
 		mark_keyboard(chat_id)
 	return result
diff --git a/lib/delivery.py b/lib/delivery.py
new file mode 100644
index 00000000..963b5cdd
--- /dev/null
+++ b/lib/delivery.py
@@ -0,0 +1,132 @@
+"""Where a report goes and how: Report, the destinations (the chat that asked, or the default chats), deliver(), and the
+one-line error report. A destination is a transport (lib/transports.py)."""
+import dataclasses
+import html
+import sys
+import traceback
+
+from lib import config
+from lib import net
+from lib import render
+from lib import transports
+
+
+@dataclasses.dataclass
+class Report:
+	"""What a report says: its lines; perhaps a chart, drawn (a PNG file object) or a spec (a dict of rows, title,
+	subtitle and render.rows_chart's options, or pie=True) that deliver() draws only for a destination that takes
+	charts; and buttons, rows of buttons.Button (or, on Telegram, a keyboard markup of its own)."""
+	lines: list
+	chart: object = None
+	buttons: list = None
+
+
+REPLIES = {'telegram': transports.TelegramBot, 'slack': transports.SlackBot} # a reply into the chat that asked
+SCHEDULED = {'telegram': transports.TelegramBot, 'slack': transports.SlackWebhook, 'discord': transports.DiscordWebhook}
+
+
+def reply_to(service, chat):
+	"""The destination of a reply into a chat, or None for a network that takes none."""
+	kind = REPLIES.get(service)
+	return kind(chat) if kind else None
+
+
+def default_chats():
+	"""Where a scheduled report goes: Telegram's telegramChatID (left out without one), and the Slack and Discord
+	incoming webhooks. Raises RuntimeError when there is nowhere to post."""
+	chats = [SCHEDULED[service](config.env.telegramChatID if service == 'telegram' else url)
+		for service, url in config.env.webhooks.items() if service != 'telegram' or config.env.telegramChatID]
+	if not chats:
+		raise RuntimeError('nowhere to post: finbot.env sets no telegramChatID, slack_webhook or discord_webhook')
+	return chats
+
+
+def destinations(interactive, service, chat_id):
+	"""Where a report goes: back to the chat that asked for it (interactive), else to the default chats."""
+	return [reply_to(service, chat_id)] if interactive else default_chats()
+
+
+def takes_charts(service, reply=True):
+	"""Whether a report to `service` can carry a chart: never with the graph preference off; else as its transport says,
+	a reply going into the chat that asked and a scheduled report to the default chats. For the reports that draw their
+	own charts."""
+	kind = (REPLIES if reply else SCHEDULED).get(service)
+	return bool(config.prefs.graph and kind and kind(None).takes_charts)
+
+
+def deliver(dest, report):
+	"""Post a report: its chart, the lines as its caption, where the destination takes charts and the graph preference
+	is on; else the lines, buttons under the last message. A chart the network refuses goes as text without its buttons,
+	unless it may have arrived after all (a timeout). Returns the network's answers, one per message."""
+	if report.chart is None:
+		return dest.text(report.lines, report.buttons)
+	chart, caption = report.chart, '\n'.join(map(str, report.lines))
+	chartable = dest.takes_charts and config.prefs.graph
+	if isinstance(chart, dict): # a spec: drawn only where it can go
+		if not (chartable and chart.get('rows')):
+			return dest.text(report.lines)
+		chart, caption = render.from_spec(chart), caption if report.lines else chart['title']
+	elif not chartable:
+		return dest.text(caption.split('\n'))
+	answer = dest.chart(chart, caption, report.buttons)
+	if answer is transports.MAYBE_SENT:
+		return []
+	if answer:
+		return [answer]
+	print("chart not sent, sending the report as text", file=sys.stderr)
+	return dest.text(caption.split('\n'))
+
+
+def error_line(e, context=None, limit=200):
+	"""One-line, secret-free summary of an exception (or string)."""
+	if isinstance(e, BaseException):
+		text = f"{type(e).__name__}: {e}" if str(e) else type(e).__name__
+	else:
+		text = str(e)
+	text = ' '.join(net.redact(text).split())
+	if len(text) > limit:
+		text = text[:limit - 1] + '…'
+	return f"⚠️ {context}: {text}" if context else f"⚠️ {text}"
+
+
+def report_error(e, service=None, chat_id=None, context=None):
+	"""Send a one-line error summary to the initiating chat, else to every default channel in finbot.env. The line is
+	escaped for Telegram (sent as HTML, which refuses a stray < such as the one in <redacted>) and Slack (which asks for
+	the same escapes). Never raises: error reporting must not become a second failure."""
+	try:
+		line = error_line(e, context)
+		print(line, file=sys.stderr)
+		if isinstance(e, BaseException):
+			trace = ''.join(traceback.format_exception(e))
+			print(net.redact(trace), end='', file=sys.stderr) # a request's error carries its URL
+		for dest in [reply_to(service, chat_id)] if service and chat_id else default_chats():
+			if dest:
+				try:
+					dest.post_text(line if dest.dialect == 'discord' else html.escape(line, quote=False))
+				except Exception as send_error:
+					print("Failed to report error to", dest.name, ":", send_error, file=sys.stderr)
+	except Exception as reporting_error:
+		print("Failed to report error:", reporting_error, file=sys.stderr)
+
+
+def guarded(context, func, *args, exit=True, **kwargs):
+	"""Run a scheduled entry point; on failure report one line to the default channel, then exit non-zero (exit=True,
+	for the command line) or just return None (exit=False, when the bot runs it in-process). A SystemExit inside counts
+	as a failure."""
+	try:
+		return func(*args, **kwargs)
+	except (Exception, SystemExit) as e:
+		report_error(e, context=context)
+		if exit:
+			sys.exit(1)
+
+
+def cli(script, job, handler, argv=None):
+	"""Command line entry point of a report script: parse its arguments with `job(argv) -> (label, kwargs)`, print the
+	usage on a bad argument, else run the handler guarded."""
+	try:
+		label, kwargs = job(sys.argv[1:] if argv is None else argv)
+	except ValueError as e:
+		print("Usage:", script, e, file=sys.stderr)
+		sys.exit(1)
+	guarded(label, handler, **kwargs)
diff --git a/lib/jobrun.py b/lib/jobrun.py
index ff848c68..5074bfe0 100644
--- a/lib/jobrun.py
+++ b/lib/jobrun.py
@@ -8,7 +8,7 @@
 	signal.signal(signal.SIGUSR1, signal.SIG_IGN) # likewise a flush (systemctl kill -s USR1); its default action would kill the job
 import importlib, json, re, sys
 
-from lib import webhook
+from lib import delivery
 
 REPORTED = 3 # exit status of a run that failed and said so in the chat; the bot reports any other non-zero status itself
 
@@ -24,7 +24,7 @@ def run():
 		"""Import the report and run it: an import that fails is reported like any other failure."""
 		importlib.import_module(module).lambda_handler(**json.loads(argv[1]))
 		return True
-	return 0 if webhook.guarded(module.rsplit('.', 1)[-1] + '.py', run, exit=False) else REPORTED
+	return 0 if delivery.guarded(module.rsplit('.', 1)[-1] + '.py', run, exit=False) else REPORTED
 
 
 if __name__ == '__main__':
diff --git a/lib/jobs.py b/lib/jobs.py
index d1897661..288c5a07 100644
--- a/lib/jobs.py
+++ b/lib/jobs.py
@@ -23,7 +23,7 @@
 from lib import config
 from lib import state
 from lib import jobrun
-from lib import webhook
+from lib import delivery
 from lib.reports import cal, cleanup, milestone, performance, podcasts, price, rating, reminder, shorts, trades
 
 SCRIPTS = {'cal': cal, 'cleanup': cleanup, 'milestone': milestone, 'performance': performance, 'podcasts': podcasts, 'price': price,
@@ -265,19 +265,20 @@ def load(self):
 		try:
 			jobs, errors = parse_schedule(Path(path).read_text(encoding='utf-8'), self.zone)
 		except OSError as e:
-			webhook.report_error(e, context='schedule')
+			delivery.report_error(e, context='schedule')
 			return False
 		if impossible := [f'{path} line {number}: {error}: {line}' for number, line, error in errors if isinstance(error, ImpossibleTime)]:
 			raise ImpossibleTime(impossible)
 		for number, line, error in errors:
-			webhook.report_error(ScheduleError(f'{path} line {number}: {error}: {line}'), context='schedule')
+			delivery.report_error(ScheduleError(f'{path} line {number}: {error}: {line}'), context='schedule')
 		if not jobs and self.jobs:
 			print('schedule: no usable line in', path, '- keeping the current schedule', file=sys.stderr)
 			return False
 		try:
 			last = state.load(STATE_FILE)
 		except (OSError, ValueError) as e: # unreadable or not JSON: carry on as if there were no record, the next run rewrites it
-			webhook.report_error(ScheduleError(f'cannot read {STATE_FILE}, starting without a record of past runs: {e}'), context='schedule')
+			problem = f'cannot read {STATE_FILE}, starting without a record of past runs: {e}'
+			delivery.report_error(ScheduleError(problem), context='schedule')
 			last = None
 		self.last = last if isinstance(last, dict) else {}
 		now = self.clock()
@@ -294,7 +295,8 @@ def reload(self, *args):
 		try:
 			self.load()
 		except ImpossibleTime as e:
-			webhook.report_error(ScheduleError(f'not reloaded, the running schedule is unchanged: {e}'), context='schedule')
+			problem = f'not reloaded, the running schedule is unchanged: {e}'
+			delivery.report_error(ScheduleError(problem), context='schedule')
 
 	def start_base(self, job, now):
 		"""Where to count from at start: a run missed within the catch-up window fires once; with no record, nothing is missed."""
@@ -362,18 +364,18 @@ def execute(self, job, commands):
 					if status:
 						outcome = 'failed'
 						if status != jobrun.REPORTED: # it ended without saying why
-							webhook.report_error(RuntimeError(exit_reason(status)), context=label)
+							delivery.report_error(RuntimeError(exit_reason(status)), context=label)
 				except gevent.Timeout:
 					raise
 				except (Exception, SystemExit) as e:
-					webhook.report_error(e, context=label)
+					delivery.report_error(e, context=label)
 					outcome = 'failed'
 				finally:
 					self.busy.discard(label)
 		except gevent.Timeout as t:
 			if t is not timeout:
 				raise
-			webhook.report_error(TimeoutError(f'still running after {JOB_TIMEOUT // 60} minutes'), context=label)
+			delivery.report_error(TimeoutError(f'still running after {JOB_TIMEOUT // 60} minutes'), context=label)
 			outcome = 'timed out'
 		finally:
 			timeout.close()
@@ -445,5 +447,5 @@ def run_forever(self):
 			try:
 				self.sleep(self.tick())
 			except Exception as e:
-				webhook.report_error(e, context='schedule')
+				delivery.report_error(e, context='schedule')
 				self.sleep(30)
diff --git a/lib/render.py b/lib/render.py
index ee9be8e1..81c08cd8 100644
--- a/lib/render.py
+++ b/lib/render.py
@@ -434,6 +434,12 @@ def pie_chart(rows, title, subtitle=''):
 	_heading(fig, title, subtitle)
 	return _png(fig, bg)
 
+def from_spec(spec):
+	"""Draw a chart spec: a dict of rows, title and subtitle, and pie=True for a pie, else rows_chart's options."""
+	spec = dict(spec)
+	rows, title, subtitle = spec.pop('rows'), spec.pop('title'), spec.pop('subtitle', '')
+	return pie_chart(rows, title, subtitle) if spec.pop('pie', False) else rows_chart(rows, title, subtitle, **spec)
+
 def bar_graph(top_rows, bottom_rows, title, subtitle=''):
 	"""Top performers then, after a gap, the worst performers: the same order as the caption's two lists."""
 	return hbar_chart(list(top_rows) + [('', None)] + list(bottom_rows), title, subtitle)
diff --git a/lib/reports/__init__.py b/lib/reports/__init__.py
index eb2923ea..6fed69fe 100644
--- a/lib/reports/__init__.py
+++ b/lib/reports/__init__.py
@@ -9,6 +9,6 @@
 def cli(path):
 	"""Run the report a top-level script is named for (price.py: lib/reports/price.py) with the script's arguments."""
 	import importlib
-	from lib import webhook
+	from lib import delivery
 	report = importlib.import_module('lib.reports.' + os.path.basename(path).removesuffix('.py'))
-	webhook.cli(sys.argv[0], report.job, report.lambda_handler)
+	delivery.cli(sys.argv[0], report.job, report.lambda_handler)
diff --git a/lib/reports/about.py b/lib/reports/about.py
index a6a0fe80..c4bb3e71 100644
--- a/lib/reports/about.py
+++ b/lib/reports/about.py
@@ -1,10 +1,11 @@
 """What finbot can do: the .help text, and the .jobs listing of the schedule (with running a job from it)."""
 import re
+from lib import buttons
 from lib import jobs
 
 def prepare_jobs_payload(service, args):
 	"""`.jobs` lists the scheduled commands, each once; `.jobs run <number|command>` runs one now, as cron would.
-	Returns (payload, reply_markup): Telegram gets a button per command."""
+	Returns (payload, buttons): Telegram gets a button per command, two to a row."""
 	scheduler = jobs.current
 	if scheduler is None:
 		return ['Scheduled jobs are off (scheduler=False)'], None
@@ -19,8 +20,9 @@ def prepare_jobs_payload(service, args):
 		payload.append("\nRun one with .jobs run NUMBER" + (' or tap it' if service == 'telegram' else ''))
 		markup = None
 		if service == 'telegram':
-			buttons = [{'text': f"{number}. {jobs.display(label)}"[:30], 'callback_data': f"j|{number}|{jobs.checksum(label)}"} for number, (label, _, _) in enumerate(commands, 1)]
-			markup = {'inline_keyboard': [buttons[i:i + 2] for i in range(0, len(buttons), 2)]}
+			row = [buttons.Button(f"{number}. {jobs.display(label)}"[:30], f"j|{number}|{jobs.checksum(label)}")
+				for number, (label, _, _) in enumerate(commands, 1)]
+			markup = [row[i:i + 2] for i in range(0, len(row), 2)]
 		return payload, markup
 	m = re.fullmatch(r'run(?:\s+(.+))?', args, re.IGNORECASE)
 	if not m or not m.group(1):
diff --git a/lib/reports/cal.py b/lib/reports/cal.py
index 65d384ea..ad5efd0c 100644
--- a/lib/reports/cal.py
+++ b/lib/reports/cal.py
@@ -1,14 +1,12 @@
 """.earnings and .dividend: upcoming earnings and ex-dividend dates (scheduled: cal.py earnings | ex-dividend)."""
 import datetime
-import sys
 
 from lib import config
-from lib import telegram
 from lib import formats
 from lib import links
 from lib import markets
 from lib import watchlist
-from lib import webhook
+from lib import delivery
 from lib import yahoo
 
 def primary_listings():
@@ -64,19 +62,6 @@ def prepare_payload(service, market_data):
 			payload = [f"No events found for {tickers[0]}" if specific_stock else f"No events found {formats.days_english(days, 'for the next ')}"]
 		return payload
 
-	def pin_if_earnings(service, results):
-		"""Pin every message of the Telegram earnings report (several if it was chunked)."""
-		if not (event == 'earnings' and not specific_stock and service == 'telegram' and results):
-			return
-		sent = []
-		for result in results:
-			try:
-				sent.append((result['result']['chat']['id'], result['result']['message_id']))
-			except (KeyError, TypeError):
-				print("Unable to pin: unexpected response", file=sys.stderr)
-		for chat, message_id in reversed(sent): # pin last-to-first so the report's first message ends up on top
-			telegram.pinChatMessage(chat, message_id)
-
 	# MAIN #
 	tickers = [markets.transform_to_yahoo(specific_stock)] if specific_stock else watchlist.with_holdings()
 	if event == 'earnings':
@@ -84,8 +69,11 @@ def pin_if_earnings(service, results):
 	else: # ex-dividend dates are in Yahoo's details
 		market_data = yahoo.fetch_details(tickers, strict=bool(specific_stock))
 
-	for service, url, chat in webhook.destinations(interactive, service, chat_id):
-		pin_if_earnings(service, webhook.send(service, url, chat, prepare_payload(service, market_data)))
+	for dest in delivery.destinations(interactive, service, chat_id):
+		service = dest.dialect
+		answers = delivery.deliver(dest, delivery.Report(prepare_payload(service, market_data)))
+		if event == 'earnings' and not specific_stock and hasattr(dest, 'pin'):
+			dest.pin(answers) # the earnings report stays on top, where the network can pin
 
 def job(argv):
 	"""Arguments of a scheduled run: (label, lambda_handler kwargs). Raises ValueError with the usage on a bad argument."""
diff --git a/lib/reports/milestone.py b/lib/reports/milestone.py
index 6b8dce09..162a2e9e 100644
--- a/lib/reports/milestone.py
+++ b/lib/reports/milestone.py
@@ -1,7 +1,7 @@
 """Record highs and lows (all-time) among the stocks followed (scheduled: milestone.py)."""
 from itertools import groupby
 from lib import state
-from lib import webhook
+from lib import delivery
 from lib import links
 from lib import markets
 from lib import watchlist
@@ -62,9 +62,10 @@ def prepare_milestone_payload(service, market_data):
 	tickers = watchlist.with_holdings()
 	market_data = yahoo.fetch(tickers)
 	records, changed = state.load('finbot_milestone.json', {}), {}
-	for service, url, chat in webhook.default_chats():
+	for dest in delivery.default_chats():
+		service = dest.dialect
 		payload, changed = prepare_milestone_payload(service, market_data)
-		webhook.send(service, url, chat, payload)
+		delivery.deliver(dest, delivery.Report(payload))
 
 	if changed:
 		with state.edit('finbot_milestone.json', {}) as recorded:
diff --git a/lib/reports/performance.py b/lib/reports/performance.py
index c72dc3d6..d9f4b46e 100644
--- a/lib/reports/performance.py
+++ b/lib/reports/performance.py
@@ -3,7 +3,7 @@
 
 from lib import config
 from lib import sharesight
-from lib import webhook
+from lib import delivery
 from lib import formats
 from lib import links
 from lib import render
@@ -57,7 +57,7 @@ def report(service, past_days=None, portfolio_select=None, interactive=False, pe
 	period = formats.days_english(past_days)
 	payload = [service.bold(f"Performance over {period}")] + [line for _, line in lines]
 	graph = None
-	if webhook.takes_charts(service, interactive):
+	if delivery.takes_charts(service, interactive):
 		chart_rows.sort(key=lambda row: row[1], reverse=True) # best first
 		graph = render.rows_chart(chart_rows, "Performance", f"% change {'' if period == 'today' else 'over '}{period}", value_fmt=lambda v: formats.signed_percent(v, 2))
 	return payload, graph
@@ -69,10 +69,13 @@ def lambda_handler(chat_id=None, past_days=None, service=None, user='', portfoli
 	if past_days is None:
 		past_days = config.prefs.past_days
 	performance = gather(past_days, portfolio_select)
-	for service, url, chat in webhook.destinations(interactive, service, chat_id):
+	for dest in delivery.destinations(interactive, service, chat_id):
+		service = dest.dialect
 		payload, graph = report(service, past_days, portfolio_select, interactive, performance)
-		markup = buttons.keyboard('f', [portfolio_select or ''], buttons.period_for_days(past_days)) if graph and service == 'telegram' else None
-		webhook.send(service, url, chat, payload, graph, reply_markup=markup)
+		markup = None
+		if graph and dest.takes_buttons:
+			markup = buttons.keyboard('f', [portfolio_select or ''], buttons.period_for_days(past_days))
+		delivery.deliver(dest, delivery.Report(payload, graph, markup))
 
 def job(argv):
 	"""Arguments of a scheduled run: (label, lambda_handler kwargs). Raises ValueError with the usage on a bad argument."""
diff --git a/lib/reports/podcasts.py b/lib/reports/podcasts.py
index 3e9e771f..1ffcd20a 100644
--- a/lib/reports/podcasts.py
+++ b/lib/reports/podcasts.py
@@ -6,7 +6,7 @@
 from datetime import datetime, timezone
 from lib import net
 from lib import state
-from lib import webhook
+from lib import delivery
 
 NAMESPACES = {
 	"media": "http://search.yahoo.com/mrss/",
@@ -93,14 +93,15 @@ def lambda_handler(feed_url):
 		raise RuntimeError(f"fetching feed {feed_url}: {e}")
 	if not guids:
 		return
-	for service, url, chat in webhook.default_chats():
+	for dest in delivery.default_chats():
+		service = dest.dialect
 		payload = []
 		for episode in episodes:
 			podcast_name = episode['podcast']
 			title = episode['title'].removeprefix(podcast_name).lstrip(string.punctuation + " ")
 			duration = f" {episode['duration']}" if episode['duration'] else ""
 			payload.append(f"📣 {podcast_name}: {service.link(episode['url'], title)}{duration} ({episode['ago']})")
-		webhook.send(service, url, chat, payload)
+		delivery.deliver(dest, delivery.Report(payload))
 	with state.edit(seen_file, []) as recorded: # last, so that a failure above leaves these episodes for next run
 		recorded[:] = sorted(guids.union(recorded))
 
diff --git a/lib/reports/price.py b/lib/reports/price.py
index ac44ff71..f84ca2d6 100644
--- a/lib/reports/price.py
+++ b/lib/reports/price.py
@@ -5,7 +5,7 @@
 
 from lib import config
 from lib import sharesight
-from lib import webhook
+from lib import delivery
 from lib import formats
 from lib import links
 from lib import markets
@@ -93,7 +93,8 @@ def report(service, threshold=None, user='', specific_stock=None, interactive=Fa
 				if config.prefs.performance_use_sharesight:
 					continue
 				if specific_stock: # the one ticker asked for: no history is reported in one line
-					history, graph = yahoo.price_history(ticker, days, graph=config.prefs.graph and webhook.takes_charts(service, interactive), graphCache=False)
+					drawn = config.prefs.graph and delivery.takes_charts(service, interactive)
+					history, graph = yahoo.price_history(ticker, days, graph=drawn, graphCache=False)
 				else:
 					try:
 						history, _ = yahoo.price_history(ticker, days, graph=False)
@@ -132,7 +133,7 @@ def report(service, threshold=None, user='', specific_stock=None, interactive=Fa
 			heading = f'Price movement {threshold:g}%, {formats.date_range_english(days, as_of)}'
 		else:
 			heading = f'{threshold:g}% at close, {on}'
-		chartable = webhook.takes_charts(service, interactive)
+		chartable = delivery.takes_charts(service, interactive)
 		if top:
 			bottom = [service.bold(f'Bottom {top}')] + list(reversed(payload[-top:]))
 			payload = [service.bold(heading)] + payload[:top] + [''] + bottom
@@ -170,19 +171,20 @@ def lambda_handler(chat_id=None, threshold=None, service=None, user='', specific
 	if threshold is None:
 		threshold = config.prefs.price_percent
 	market = gather(specific_stock, days)
-	def markup(service):
-		"""A chart's period buttons on Telegram: one ticker's chart, and the daily and N-day lists, but not the intraday
-		ones."""
-		if service != 'telegram':
+	def markup(dest):
+		"""A chart's period buttons, where the destination takes buttons: one ticker's chart, and the daily and N-day
+		lists, but not the intraday ones."""
+		if not dest.takes_buttons:
 			return None
 		if specific_stock:
 			return buttons.keyboard('p', market[0], buttons.period_for_days(days))
 		if not (midsession or premarket or close):
 			return buttons.keyboard('l', [f"{threshold:g}", str(top or 0)], buttons.period_for_days(days, 'd' if not days else None))
 		return None
-	for service, url, chat in webhook.destinations(interactive, service, chat_id):
+	for dest in delivery.destinations(interactive, service, chat_id):
+		service = dest.dialect
 		payload, graph = report(service, threshold, user, specific_stock, interactive, midsession, premarket, interday, days, close, top, market)
-		webhook.send(service, url, chat, payload, graph, reply_markup=markup(service) if graph else None)
+		delivery.deliver(dest, delivery.Report(payload, graph, markup(dest) if graph else None))
 
 def job(argv):
 	"""Arguments of a scheduled run: (label, lambda_handler kwargs). Raises ValueError with the usage on a bad argument."""
diff --git a/lib/reports/rating.py b/lib/reports/rating.py
index 7b6c378c..136e78d8 100644
--- a/lib/reports/rating.py
+++ b/lib/reports/rating.py
@@ -1,5 +1,5 @@
 """Analysts' consensus rating changes since the last scheduled run (scheduled: rating.py). .buy and .sell are in rankings."""
-from lib import webhook
+from lib import delivery
 from lib import state
 from lib import links
 from lib import markets
@@ -60,9 +60,10 @@ def get_emoji(old, new):
 	old = state.load('finbot_rating.json', {})
 	market_data = yahoo.fetch_details(tickers, strict=bool(specific_stock))
 
-	for service, url, chat in webhook.destinations(interactive, service, chat_id):
+	for dest in delivery.destinations(interactive, service, chat_id):
+		service = dest.dialect
 		payload, new = prepare_rating_payload(service, market_data)
-		webhook.send(service, url, chat, payload)
+		delivery.deliver(dest, delivery.Report(payload))
 
 	if new and not interactive: # added to what is recorded: a ticker missing from this run keeps its last rating
 		with state.edit('finbot_rating.json', {}) as ratings:
diff --git a/lib/reports/reminder.py b/lib/reports/reminder.py
index d33ad2fe..a1c1865e 100644
--- a/lib/reports/reminder.py
+++ b/lib/reports/reminder.py
@@ -4,7 +4,7 @@
 from dateutil.relativedelta import relativedelta
 from lib import config
 from lib import formats
-from lib import webhook
+from lib import delivery
 
 
 def lambda_handler():
@@ -69,8 +69,9 @@ def prepare_finance_calendar_payload(service):
 
 	# MAIN #
 
-	for service, url, chat in webhook.default_chats():
-		webhook.send(service, url, chat, prepare_finance_calendar_payload(service))
+	for dest in delivery.default_chats():
+		service = dest.dialect
+		delivery.deliver(dest, delivery.Report(prepare_finance_calendar_payload(service)))
 
 def job(argv):
 	"""Arguments of a scheduled run: reminder.py takes none."""
diff --git a/lib/reports/shorts.py b/lib/reports/shorts.py
index 2b41ea65..59ebab4a 100644
--- a/lib/reports/shorts.py
+++ b/lib/reports/shorts.py
@@ -1,6 +1,6 @@
 """.shorts: the most shorted stocks followed, from Short Man (ASX) and Yahoo (scheduled: shorts.py)."""
 from lib import config
-from lib import webhook
+from lib import delivery
 from lib import markets
 from lib import render
 from lib import watchlist
@@ -29,7 +29,7 @@ def prepare_shorts_payload(service, market_data):
 		payload, graph = [line for _, line, _ in found], None
 		if payload and not specific_stock:
 			payload.insert(0, service.bold(f'Tracked stocks shorted over {threshold:g}%'))
-			if webhook.takes_charts(service, interactive):
+			if delivery.takes_charts(service, interactive):
 				graph = render.shorts_chart([(code, percent) for percent, _, code in found], threshold)
 		elif not payload and interactive:
 			payload = [f"🩳 No short interest found for {tickers[0]}" if specific_stock else f"🩳 No stocks shorted over {threshold:g}%. Try specifying a number."]
@@ -47,9 +47,10 @@ def prepare_shorts_payload(service, market_data):
 	market_data |= yahoo.fetch_details((ticker for ticker in tickers if '.' not in ticker), strict=bool(specific_stock)) # US short interest; Short Man has the ASX's
 	market_data = shortman.fetch(market_data)
 
-	for service, url, chat in webhook.destinations(interactive, service, chat_id):
+	for dest in delivery.destinations(interactive, service, chat_id):
+		service = dest.dialect
 		payload, graph = prepare_shorts_payload(service, market_data)
-		webhook.send(service, url, chat, payload, graph)
+		delivery.deliver(dest, delivery.Report(payload, graph))
 
 def job(argv):
 	"""Arguments of a scheduled run: shorts.py takes none."""
diff --git a/lib/reports/trades.py b/lib/reports/trades.py
index 5e4d1450..b2098d93 100644
--- a/lib/reports/trades.py
+++ b/lib/reports/trades.py
@@ -6,7 +6,7 @@
 from lib import state
 from lib import words
 from lib import sharesight
-from lib import webhook
+from lib import delivery
 from lib import formats
 from lib import links
 from lib import markets
@@ -94,11 +94,12 @@ def prepare_trade_payload(service, trades):
 	known_trades = set(state.load(state_file, []))
 	trades = [trade for portfolio in sharesight.portfolios(portfolio_select) for trade in portfolio.trades(days)]
 
-	for service, url, chat in webhook.destinations(interactive, service, chat_id):
+	for dest in delivery.destinations(interactive, service, chat_id):
+		service = dest.dialect
 		payload = prepare_trade_payload(service, trades)
 		if interactive and not payload: # no trades, or none of them a buy or a sell
 			payload = [f"{user} No trades {formats.days_english(days, 'in the past ')}. {random.choice(words.noTradesVerb)}"]
-		webhook.send(service, url, chat, payload)
+		delivery.deliver(dest, delivery.Report(payload))
 
 	if newtrades and not interactive:
 		with state.edit(state_file, []) as known:
diff --git a/lib/transports.py b/lib/transports.py
new file mode 100644
index 00000000..3967ee39
--- /dev/null
+++ b/lib/transports.py
@@ -0,0 +1,324 @@
+"""How a message gets to each network: a class per way of posting, made for one destination (a chat, or an incoming
+webhook). A transport knows its dialect, what it can carry (charts, buttons, a typing notice) and the network's size
+limits; text() and chart() are the same everywhere, and each network posts one message its own way (post_text,
+post_chart)."""
+import json
+import sys
+import time
+
+import requests
+
+from lib import config
+from lib import dialects
+from lib import net
+from lib import telegram
+
+MAYBE_SENT = object() # a read timeout on the request that posts a chart: it may have arrived, so no text copy follows
+SUPPRESS_EMBEDS = 4 # Discord message flag: no link previews
+NO_MENTIONS = {'parse': []} # Discord: text that looks like @everyone or a mention pings nobody
+
+
+class Transport:
+	"""A destination. Subclasses set what it carries and post one message; the base splits and trims."""
+	name = '' # the network, in logs
+	dialect = None
+	takes_charts = False # a chart, with the report as its caption
+	takes_buttons = False # buttons under a message
+	shows_typing = False # a typing notice while a command works
+
+	def __init__(self, where):
+		"""where: the chat id, or the incoming webhook's URL."""
+		self.where = where
+
+	def __repr__(self):
+		"""<TelegramBot 55>: the kind and where, for logs and test failures."""
+		return f'<{type(self).__name__} {self.where}>'
+
+	def text(self, lines, buttons=None):
+		"""Post lines of text: one message, or several when they are too long for one (then at most chunk_maxlines lines
+		each), buttons under the last. Returns the network's answer for each message (None for one that failed)."""
+		lines = [str(line) for line in lines]
+		if not lines:
+			return []
+		limit = self.dialect.text_limit
+		if len('\n'.join(lines)) <= limit:
+			return [self.post_text('\n'.join(lines), buttons)]
+		chunks = chunked(lines, limit, config.prefs.chunk_maxlines)
+		answers = []
+		for i, chunk in enumerate(chunks):
+			if i:
+				time.sleep(0.5) # Telegram takes about one message a second in a chat
+			answers.append(self.post_text('\n'.join(chunk), buttons if i == len(chunks) - 1 else None))
+		return answers
+
+	def chart(self, image, caption, buttons=None):
+		"""Post a chart with the caption under it, cut to fit. The network's answer, MAYBE_SENT, or None when it
+		failed."""
+		if hasattr(image, 'seek'):
+			image.seek(0)
+		return self.post_chart(image, fit_caption(caption, self.dialect), buttons)
+
+	def post_text(self, text, buttons=None):
+		"""Post one message. The network's answer, or None when it failed (logged)."""
+		raise NotImplementedError
+
+	def post_chart(self, image, caption, buttons=None):
+		"""Post one chart with its caption (see chart())."""
+		raise NotImplementedError
+
+	def typing(self):
+		"""Show the typing notice, where the network has one."""
+
+	def _post_json(self, url, payload, headers=None):
+		"""POST a JSON message; the answer (or {} when it is not JSON), or None when the network refused it or
+		failed."""
+		try:
+			r = net.post(url, headers={'Content-type': 'application/json', **(headers or {})}, json=payload,
+				timeout=config.prefs.http_timeout)
+		except Exception as e:
+			print("Failure sending to", self.name + ":", net.redact(e), file=sys.stderr)
+			return None
+		if r.status_code != 200:
+			print(r.status_code, "error outbound to", self.name, file=sys.stderr)
+			return None
+		print(r.status_code, "OK outbound to", self.name, file=sys.stderr)
+		try:
+			return r.json()
+		except ValueError:
+			return {}
+
+
+class TelegramBot(Transport):
+	"""A Telegram chat, through the Bot API: HTML, silent, no link previews; charts, buttons, edits and the typing
+	notice."""
+	name = 'telegram'
+	dialect = dialects.TELEGRAM
+	takes_charts = takes_buttons = shows_typing = True
+
+	@staticmethod
+	def markup(buttons):
+		"""Rows of buttons.Button as Telegram's inline keyboard ('● ' marks the active one); a markup of Telegram's own
+		(the command keyboard) as it is."""
+		if isinstance(buttons, dict):
+			return buttons
+		return {'inline_keyboard': [[{'text': ('● ' if button.active else '') + button.label,
+			'callback_data': button.data} for button in row] for row in buttons]}
+
+	def post_text(self, text, buttons=None):
+		"""sendMessage."""
+		payload = {'text': text, 'parse_mode': 'HTML', 'disable_web_page_preview': 'true',
+			'disable_notification': 'true'}
+		if buttons:
+			payload['reply_markup'] = self.markup(buttons)
+		return self._post_json(config.env.webhooks['telegram'] + 'sendMessage?chat_id=' + str(self.where), payload)
+
+	def post_chart(self, image, caption, buttons=None):
+		"""sendPhoto, the report as its caption."""
+		data = {'disable_notification': True, 'chat_id': self.where, 'caption': caption, 'parse_mode': 'HTML',
+			'disable_web_page_preview': True}
+		if buttons:
+			data['reply_markup'] = json.dumps(self.markup(buttons))
+		try:
+			r = net.post(config.env.webhooks['telegram'] + 'sendPhoto?chat_id=' + str(self.where), data=data,
+				files={'photo': ('image.png', image)}, timeout=config.prefs.http_timeout)
+		except requests.exceptions.ReadTimeout as e:
+			print("Telegram sendPhoto timed out, the chart may still arrive:", net.redact(e), file=sys.stderr)
+			return MAYBE_SENT
+		except Exception as e:
+			print("Telegram sendPhoto failed:", net.redact(e), file=sys.stderr)
+			return None
+		try:
+			output = r.json()
+		except ValueError:
+			output = {}
+		if r.status_code == 200 and output.get('ok'):
+			return output
+		print(r.status_code, "error telegram sendPhoto", output.get('error_code'),
+			output.get('description') or r.text[:300], file=sys.stderr)
+		return None
+
+	def edit_chart(self, message, image, caption, buttons=None):
+		"""Replace a chart message's image, caption and buttons in place. Raises RuntimeError when Telegram refuses,
+		unless the message is already so (pressing the active period again is harmless); a request that fails on the way
+		is only logged."""
+		media = {'type': 'photo', 'media': 'attach://photo', 'caption': fit_caption(caption, self.dialect),
+			'parse_mode': 'HTML'}
+		data = {'chat_id': self.where, 'message_id': message, 'media': json.dumps(media)}
+		if buttons:
+			data['reply_markup'] = json.dumps(self.markup(buttons))
+		if hasattr(image, 'seek'):
+			image.seek(0)
+		try:
+			telegram.call('editMessageMedia', data=data, files={'photo': ('image.png', image)})
+		except RuntimeError as e:
+			if 'not modified' not in str(e):
+				raise
+		except Exception as e: # it may have arrived even so: no error line for something the chat may already show
+			print("Telegram editMessageMedia failed:", net.redact(e), file=sys.stderr)
+
+	def edit_text(self, message, text, buttons=None):
+		"""Replace a text message's text, and its buttons unless buttons is None ([] takes them away). A failure is only
+		logged."""
+		payload = {'chat_id': self.where, 'message_id': message, 'text': text, 'parse_mode': 'HTML',
+			'disable_web_page_preview': True}
+		if buttons is not None:
+			payload['reply_markup'] = self.markup(buttons)
+		telegram.try_call('editMessageText', json=payload)
+
+	def typing(self):
+		"""sendChatAction: "typing…" for a few seconds."""
+		telegram.try_call('sendChatAction', data={'chat_id': self.where, 'action': 'typing'})
+
+	def pin(self, answers):
+		"""Pin the messages these answers are for, last first, so the first ends up on top, without the notification."""
+		sent = []
+		for answer in answers:
+			try:
+				sent.append((answer['result']['chat']['id'], answer['result']['message_id']))
+			except (KeyError, TypeError):
+				print("Unable to pin: unexpected response", file=sys.stderr)
+		for chat, message in reversed(sent):
+			telegram.pinChatMessage(chat, message)
+
+
+class SlackBot(Transport):
+	"""A Slack channel, as the bot: chat.postMessage and the file upload, with slackBotToken."""
+	name = 'slack'
+	dialect = dialects.SLACK
+
+	@property
+	def takes_charts(self):
+		"""Charts are uploaded with the bot token, so only with one."""
+		return bool(config.env.slackBotToken)
+
+	def post_text(self, text, buttons=None):
+		"""chat.postMessage, without link previews (Slack reads those flags from the body)."""
+		if not config.env.slackBotToken:
+			print(f"ERROR slack: a reply to channel {self.where} needs slackBotToken in finbot.env; not sent",
+				file=sys.stderr)
+			return None
+		payload = {'text': text, 'unfurl_links': False, 'unfurl_media': False, 'channel': self.where}
+		return self._post_json('https://slack.com/api/chat.postMessage', payload,
+			{'Authorization': 'Bearer ' + config.env.slackBotToken})
+
+	def post_chart(self, image, caption, buttons=None):
+		"""Upload the chart into the channel with the report as its message, the way Slack takes files since it retired
+		files.upload: ask for an upload URL, send the bytes there, then complete the upload into the channel (the bot
+		needs the files:write and files:read scopes). Slack's answer, MAYBE_SENT, or None if a step failed."""
+		data = image.read() if hasattr(image, 'read') else bytes(image)
+		try:
+			ticket = slack_api('files.getUploadURLExternal', filename='chart.png', length=len(data))
+			r = net.post(ticket['upload_url'], data=data, timeout=config.prefs.http_timeout)
+			if r.status_code != 200:
+				raise RuntimeError(f"Slack upload: {r.status_code} {r.text[:200]}")
+			share = {'files': json.dumps([{'id': ticket['file_id']}]), 'channel_id': self.where,
+				'initial_comment': caption}
+			try:
+				output = slack_api('files.completeUploadExternal', **share)
+			except requests.exceptions.ReadTimeout as e:
+				print("Slack files.completeUploadExternal timed out, the chart may still arrive:", net.redact(e),
+					file=sys.stderr)
+				return MAYBE_SENT
+		except Exception as e:
+			print("Slack chart upload failed:", net.redact(e), file=sys.stderr)
+			return None
+		print("OK slack chart upload to", self.where, file=sys.stderr)
+		return output
+
+
+class SlackWebhook(Transport):
+	"""A Slack incoming webhook: text only, into the channel it was made for."""
+	name = 'slack'
+	dialect = dialects.SLACK
+
+	def post_text(self, text, buttons=None):
+		"""Post to the webhook, without link previews."""
+		return self._post_json(self.where, {'text': text, 'unfurl_links': False, 'unfurl_media': False})
+
+
+class DiscordWebhook(Transport):
+	"""A Discord webhook, its native endpoint (the Slack-compatible one takes no flags): no link previews, no pings,
+	charts attached."""
+	name = 'discord'
+	dialect = dialects.DISCORD
+	takes_charts = True
+
+	def url(self):
+		"""The webhook URL that answers with the message posted (200) instead of an empty 204."""
+		return self.where + ('&' if '?' in self.where else '?') + 'wait=true'
+
+	def post_text(self, text, buttons=None):
+		"""Post the message."""
+		return self._post_json(self.url(), {'content': text, 'flags': SUPPRESS_EMBEDS, 'allowed_mentions': NO_MENTIONS})
+
+	def post_chart(self, image, caption, buttons=None):
+		"""The chart as an attachment, with the report as its message. Discord's answer, MAYBE_SENT, or None if
+		refused."""
+		payload = {'content': caption, 'flags': SUPPRESS_EMBEDS, 'allowed_mentions': NO_MENTIONS,
+			'attachments': [{'id': 0, 'filename': 'chart.png'}]}
+		try:
+			r = net.post(self.url(), data={'payload_json': json.dumps(payload)},
+				files={'files[0]': ('chart.png', image, 'image/png')}, timeout=config.prefs.http_timeout)
+		except requests.exceptions.ReadTimeout as e:
+			print("Discord chart timed out, it may still arrive:", net.redact(e), file=sys.stderr)
+			return MAYBE_SENT
+		except Exception as e:
+			print("Discord chart failed:", net.redact(e), file=sys.stderr)
+			return None
+		if r.status_code == 200:
+			print(r.status_code, "OK discord chart", file=sys.stderr)
+			try:
+				return r.json() or {}
+			except ValueError:
+				return {}
+		print(r.status_code, "error discord chart", net.redact(r.text[:300]), file=sys.stderr)
+		return None
+
+
+def slack_api(method, **data):
+	"""POST a Slack Web API method as the bot, form-encoded, and return its answer. Raises RuntimeError with Slack's
+	error if not ok."""
+	if not config.env.slackBotToken:
+		raise RuntimeError(f"Slack {method}: needs slackBotToken in finbot.env")
+	r = net.post('https://slack.com/api/' + method, headers={'Authorization': 'Bearer ' + config.env.slackBotToken},
+		data=data, timeout=config.prefs.http_timeout)
+	try:
+		output = r.json()
+	except ValueError:
+		output = {}
+	if not output.get('ok'):
+		raise RuntimeError(f"Slack {method}: {output.get('error') or r.status_code}")
+	return output
+
+
+def chunked(lines, limit, max_lines):
+	"""The lines in messages of at most max_lines lines and limit characters each, whichever comes first; a line longer
+	than limit on its own is cut short."""
+	chunks, current, size = [], [], 0
+	for line in lines:
+		if len(line) > limit:
+			line = line[:limit - 1] + '…'
+		if current and (len(current) >= max_lines or size + 1 + len(line) > limit):
+			chunks.append(current)
+			current, size = [], 0
+		size += len(line) + (1 if current else 0)
+		current.append(line)
+	if current:
+		chunks.append(current)
+	return chunks
+
+
+def fit_caption(caption, dialect):
+	"""The caption cut to the dialect's caption limit (none for Slack), as the network counts it (Telegram leaves the
+	tags out): the whole lines that fit, and how many were left out."""
+	limit = dialect.caption_limit
+	if limit is None or dialect.visible(caption) <= limit:
+		return caption
+	lines = caption.split('\n')
+	kept = []
+	for line in lines:
+		note = f"\n…and {len(lines) - len(kept) - 1} more in the chart"
+		if dialect.visible('\n'.join(kept + [line])) + len(note) > limit:
+			break
+		kept.append(line)
+	return '\n'.join(kept) + f"\n…and {len(lines) - len(kept)} more in the chart"
diff --git a/lib/webhook.py b/lib/webhook.py
deleted file mode 100644
index 50b1589b..00000000
--- a/lib/webhook.py
+++ /dev/null
@@ -1,339 +0,0 @@
-import html
-import json
-import time
-import requests
-import re
-import sys
-import traceback
-
-from lib import config
-from lib import dialects
-from lib import net
-from lib import telegram
-
-DISCORD_LIMIT = 2000 # characters in a Discord message
-SUPPRESS_EMBEDS = 4 # Discord message flag: no link previews
-NO_MENTIONS = {'parse': []} # Discord: text that looks like @everyone or a mention pings nobody
-
-def discord_url(url):
-	"""A Discord webhook URL that answers with the message posted (200) instead of an empty 204."""
-	return url + ('&' if '?' in url else '?') + 'wait=true'
-
-def write(service, url, payload_string, chat_id=None, reply_markup=None):
-	"""POST one message to a service and return its answer, or None when it failed (logged). Telegram: HTML, no link
-	previews, silent, reply_markup as its buttons. Slack: into channel chat_id through chat.postMessage with the bot
-	token, or through the incoming webhook. Discord: the native webhook, without previews or mentions."""
-	headers = {'Content-type': 'application/json'}
-	payload = {'text': payload_string}
-	if service == 'discord': # the native webhook endpoint: Discord's Slack-compatible one takes no flags
-		payload = {'content': payload_string, 'flags': SUPPRESS_EMBEDS, 'allowed_mentions': NO_MENTIONS}
-		url = discord_url(url)
-	elif 'slack.com' in url:
-		payload['unfurl_links'] = False # no link previews: Slack reads these from the body, not the headers
-		payload['unfurl_media'] = False
-		if chat_id:
-			if not config.env.slackBotToken:
-				print(f"ERROR slack: a reply to channel {chat_id} needs slackBotToken in finbot.env; not sent", file=sys.stderr)
-				return None
-			headers['Authorization'] = 'Bearer ' + config.env.slackBotToken
-			payload['channel'] = chat_id
-	elif 'api.telegram.org' in url:
-		payload['parse_mode'] = 'HTML'
-		payload['disable_web_page_preview'] = 'true'
-		payload['disable_notification'] = 'true'
-		if reply_markup:
-			payload['reply_markup'] = reply_markup # keyboard (Telegram only)
-	try:
-		r = net.post(url, headers=headers, json=payload, timeout=config.prefs.http_timeout)
-	except Exception as e:
-		print("Failure sending to", service + ":", net.redact(e), file=sys.stderr)
-		return None
-	if r.status_code == 200:
-		print(r.status_code, "OK outbound to", service, file=sys.stderr)
-		try:
-			return r.json()
-		except ValueError:
-			return {}
-	else:
-		print(r.status_code, "error outbound to", service, file=sys.stderr)
-		return None
-
-def chat_url(service, chat_id):
-	"""Outbound URL for posting to a chat."""
-	if service == 'slack':
-		return 'https://slack.com/api/chat.postMessage'
-	if service == 'telegram':
-		return config.env.webhooks['telegram'] + 'sendMessage?chat_id=' + str(chat_id)
-	return config.env.webhooks.get(service)
-
-def default_chats():
-	"""Where a scheduled report goes: [(service, url, chat id)]. Telegram posts into telegramChatID (left out without one); Slack and
-	Discord post through their incoming webhooks, which take no chat id. Raises RuntimeError when there is nowhere to post."""
-	chats = [(dialects.TELEGRAM, chat_url(service, config.env.telegramChatID), config.env.telegramChatID) if service == 'telegram' else (dialects.of(service), url, None)
-		for service, url in config.env.webhooks.items() if service != 'telegram' or config.env.telegramChatID]
-	if not chats:
-		raise RuntimeError('nowhere to post: finbot.env sets no telegramChatID, slack_webhook or discord_webhook')
-	return chats
-
-def destinations(interactive, service, chat_id):
-	"""Where a report goes: back to the chat that asked for it (interactive), else to the default chats."""
-	return [(dialects.of(service), chat_url(service, chat_id), chat_id)] if interactive else default_chats()
-
-def send(service, url, chat_id, payload, image=None, reply_markup=None):
-	"""A report: its chart with the text as caption when there is one (sendPhoto falls back to the text where a chart cannot go),
-	else the text. reply_markup is the chart's buttons. Returns what payload_wrapper or sendPhoto did."""
-	if image:
-		return sendPhoto(chat_id, image, '\n'.join(payload), service, reply_markup=reply_markup)
-	return payload_wrapper(service, url, payload, chat_id)
-
-def error_line(e, context=None, limit=200):
-	"""One-line, secret-free summary of an exception (or string)."""
-	if isinstance(e, BaseException):
-		text = f"{type(e).__name__}: {e}" if str(e) else type(e).__name__
-	else:
-		text = str(e)
-	text = ' '.join(net.redact(text).split())
-	if len(text) > limit:
-		text = text[:limit - 1] + '…'
-	return f"⚠️ {context}: {text}" if context else f"⚠️ {text}"
-
-def report_error(e, service=None, chat_id=None, context=None):
-	"""Send a one-line error summary to the initiating chat, else to every default channel in finbot.env. The line is escaped for
-	Telegram (sent as HTML, which refuses a stray < such as the one in <redacted>) and Slack (which asks for the same escapes).
-	Never raises: error reporting must not become a second failure."""
-	try:
-		line = error_line(e, context)
-		print(line, file=sys.stderr)
-		if isinstance(e, BaseException):
-			print(net.redact(''.join(traceback.format_exception(e))), end='', file=sys.stderr) # request errors carry their URL
-		for svc, url, chat in [(service, chat_url(service, chat_id), chat_id)] if service and chat_id else default_chats():
-			if url:
-				try:
-					write(svc, url, line if svc == 'discord' else html.escape(line, quote=False), chat)
-				except Exception as send_error:
-					print("Failed to report error to", svc, ":", send_error, file=sys.stderr)
-	except Exception as reporting_error:
-		print("Failed to report error:", reporting_error, file=sys.stderr)
-
-def guarded(context, func, *args, exit=True, **kwargs):
-	"""Run a scheduled entry point; on failure report one line to the default channel, then exit non-zero (exit=True, for the
-	command line) or just return None (exit=False, when the bot runs it in-process). A SystemExit inside counts as a failure."""
-	try:
-		return func(*args, **kwargs)
-	except (Exception, SystemExit) as e:
-		report_error(e, context=context)
-		if exit:
-			sys.exit(1)
-
-def cli(script, job, handler, argv=None):
-	"""Command line entry point of a report script: parse its arguments with `job(argv) -> (label, kwargs)`, print the usage on
-	a bad argument, else run the handler guarded."""
-	try:
-		label, kwargs = job(sys.argv[1:] if argv is None else argv)
-	except ValueError as e:
-		print("Usage:", script, e, file=sys.stderr)
-		sys.exit(1)
-	guarded(label, handler, **kwargs)
-
-def payload_wrapper(service, url, payload, chat_id=None, reply_markup=None):
-	"""Returns a list with the response of each message sent (several if the payload was chunked)."""
-	if not payload:
-		return None
-	payload = [str(line) for line in payload]
-	payload_string = '\n'.join(payload)
-	limit = DISCORD_LIMIT if service == 'discord' else 4096
-	if len(payload_string) <= limit:
-		return [write(service, url, payload_string, chat_id, reply_markup)]
-	chunks = chunked(payload, limit, config.prefs.chunk_maxlines)
-	results = []
-	for idx, chunk in enumerate(chunks):
-		if idx:
-			time.sleep(0.5)
-		last = idx == len(chunks) - 1
-		results.append(write(service, url, '\n'.join(chunk), chat_id, reply_markup=reply_markup if last else None))
-	return results
-
-def chunked(lines, limit, max_lines):
-	"""The lines in messages of at most max_lines lines and limit characters each, whichever comes first; a line longer than limit
-	on its own is cut short."""
-	chunks, current, size = [], [], 0
-	for line in lines:
-		if len(line) > limit:
-			line = line[:limit - 1] + '…'
-		if current and (len(current) >= max_lines or size + 1 + len(line) > limit):
-			chunks.append(current)
-			current, size = [], 0
-		size += len(line) + (1 if current else 0)
-		current.append(line)
-	if current:
-		chunks.append(current)
-	return chunks
-
-def fit_caption(caption, limit=1024, visible=None):
-	"""Telegram rejects photo captions over 1024 visible characters; Discord takes 2,000 characters, markup included
-	(visible=len). Keep whole lines that fit and say how many were left out."""
-	if visible is None:
-		def visible(text):
-			"""The length without the HTML tags."""
-			return len(re.sub(r'<[^>]+>', '', text))
-	if visible(caption) <= limit:
-		return caption
-	lines = caption.split('\n')
-	kept = []
-	for line in lines:
-		more = len(lines) - len(kept) - 1
-		note = f"\n…and {more} more in the chart"
-		if visible('\n'.join(kept + [line])) + len(note) > limit:
-			break
-		kept.append(line)
-	return '\n'.join(kept) + f"\n…and {len(lines) - len(kept)} more in the chart"
-
-MAYBE_SENT = object() # a read timeout on the request that posts a chart: it may have arrived, so no text copy follows
-
-def takes_charts(service, reply=True):
-	"""Whether a report to `service` can carry a chart image: never with the graph preference off; on Telegram and Discord always;
-	on Slack in a reply to a chat (`reply`), uploaded into its channel with the bot token, but not in a scheduled report, which goes
-	through the incoming webhook and takes text only. Elsewhere a report goes out as text."""
-	return config.prefs.graph and (service in ('telegram', 'discord') or (service == 'slack' and reply and bool(config.env.slackBotToken)))
-
-def telegram_photo(chat_id, image_data, caption, reply_markup=None):
-	"""sendPhoto, the report as its caption (trimmed to fit). Telegram's answer, MAYBE_SENT, or None if it was refused."""
-	url = config.env.webhooks['telegram'] + "sendPhoto?chat_id=" + str(chat_id)
-	data = {
-		'disable_notification': True,
-		'chat_id': chat_id,
-		'caption': fit_caption(caption),
-		'parse_mode': 'HTML',
-		'disable_web_page_preview': True}
-	if reply_markup:
-		data['reply_markup'] = json.dumps(reply_markup) # inline keyboard (Telegram only)
-	if hasattr(image_data, 'seek'):
-		image_data.seek(0)
-	try:
-		r = net.post(url, data=data, files={"photo": ('image.png', image_data)}, timeout=config.prefs.http_timeout)
-	except requests.exceptions.ReadTimeout as e:
-		print("Telegram sendPhoto timed out, the chart may still arrive:", net.redact(e), file=sys.stderr)
-		return MAYBE_SENT
-	except Exception as e:
-		print("Telegram sendPhoto failed:", net.redact(e), file=sys.stderr)
-		return None
-	try:
-		output = r.json()
-	except ValueError:
-		output = {}
-	if r.status_code == 200 and output.get('ok'):
-		return output
-	print(r.status_code, "error telegram sendPhoto", output.get('error_code'), output.get('description') or r.text[:300], file=sys.stderr)
-	return None
-
-def discord_photo(image_data, caption):
-	"""The chart on the Discord webhook, with the report as its message (trimmed to Discord's 2,000 characters), no link previews.
-	Discord's answer, MAYBE_SENT, or None if it was refused."""
-	if hasattr(image_data, 'seek'):
-		image_data.seek(0)
-	payload = {'content': fit_caption(caption, DISCORD_LIMIT, len), 'flags': SUPPRESS_EMBEDS, 'allowed_mentions': NO_MENTIONS,
-		'attachments': [{'id': 0, 'filename': 'chart.png'}]}
-	try:
-		r = net.post(discord_url(config.env.webhooks['discord']), data={'payload_json': json.dumps(payload)},
-			files={'files[0]': ('chart.png', image_data, 'image/png')}, timeout=config.prefs.http_timeout)
-	except requests.exceptions.ReadTimeout as e:
-		print("Discord chart timed out, it may still arrive:", net.redact(e), file=sys.stderr)
-		return MAYBE_SENT
-	except Exception as e:
-		print("Discord chart failed:", net.redact(e), file=sys.stderr)
-		return None
-	if r.status_code == 200:
-		print(r.status_code, "OK discord chart", file=sys.stderr)
-		try:
-			return r.json() or {}
-		except ValueError:
-			return {}
-	print(r.status_code, "error discord chart", net.redact(r.text[:300]), file=sys.stderr)
-	return None
-
-def slack_api(method, **data):
-	"""POST a Slack Web API method as the bot, form-encoded, and return its answer. Raises RuntimeError with Slack's error if not ok."""
-	if not config.env.slackBotToken:
-		raise RuntimeError(f"Slack {method}: needs slackBotToken in finbot.env")
-	r = net.post('https://slack.com/api/' + method, headers={'Authorization': 'Bearer ' + config.env.slackBotToken}, data=data, timeout=config.prefs.http_timeout)
-	try:
-		output = r.json()
-	except ValueError:
-		output = {}
-	if not output.get('ok'):
-		raise RuntimeError(f"Slack {method}: {output.get('error') or r.status_code}")
-	return output
-
-def slack_photo(chat_id, image_data, caption):
-	"""Upload the chart into a Slack channel with the report as its message, the way Slack takes files since it retired files.upload:
-	ask for an upload URL, send the bytes there, then complete the upload into the channel (the bot needs the files:write and
-	files:read scopes). Slack's answer, MAYBE_SENT, or None if a step failed."""
-	if hasattr(image_data, 'seek'):
-		image_data.seek(0)
-	data = image_data.read() if hasattr(image_data, 'read') else bytes(image_data)
-	try:
-		ticket = slack_api('files.getUploadURLExternal', filename='chart.png', length=len(data))
-		r = net.post(ticket['upload_url'], data=data, timeout=config.prefs.http_timeout)
-		if r.status_code != 200:
-			raise RuntimeError(f"Slack upload: {r.status_code} {r.text[:200]}")
-		share = {'files': json.dumps([{'id': ticket['file_id']}]), 'channel_id': chat_id, 'initial_comment': caption}
-		try:
-			output = slack_api('files.completeUploadExternal', **share)
-		except requests.exceptions.ReadTimeout as e:
-			print("Slack files.completeUploadExternal timed out, the chart may still arrive:", net.redact(e), file=sys.stderr)
-			return MAYBE_SENT
-	except Exception as e:
-		print("Slack chart upload failed:", net.redact(e), file=sys.stderr)
-		return None
-	print("OK slack chart upload to", chat_id, file=sys.stderr)
-	return output
-
-def sendPhoto(chat_id, image_data, caption, service, reply_markup=None):
-	"""The chart, with the report as its caption: on Telegram, or uploaded into the Slack channel of a reply. Where a chart cannot go
-	(takes_charts), or the service refuses it, the report is sent as text instead, without the chart's buttons, so it does not vanish
-	with its chart; but not after a read timeout on the request that posts it, when the chart may have arrived after all."""
-	output = None
-	chartable = takes_charts(service, reply=bool(chat_id))
-	if chartable and service == 'telegram':
-		output = telegram_photo(chat_id, image_data, caption, reply_markup)
-	elif chartable and service == 'discord':
-		output = discord_photo(image_data, caption)
-	elif chartable:
-		output = slack_photo(chat_id, image_data, caption)
-	if output is MAYBE_SENT:
-		return None
-	if output:
-		return output
-	if chartable:
-		print("chart not sent, sending the report as text", file=sys.stderr)
-	return payload_wrapper(service, chat_url(service, chat_id) if chat_id else config.env.webhooks.get(service), caption.split('\n'), chat_id)
-
-def editMessageText(chat_id, message_id, text, reply_markup=None):
-	"""Telegram: replace a text message's text and buttons in place. A failure is only logged."""
-	payload = {'chat_id': chat_id, 'message_id': message_id, 'text': text, 'parse_mode': 'HTML', 'disable_web_page_preview': True}
-	if reply_markup is not None:
-		payload['reply_markup'] = reply_markup
-	telegram.try_call('editMessageText', json=payload)
-
-def editMessageMedia(chat_id, message_id, image_data, caption, reply_markup=None):
-	"""Telegram: replace a photo message's image and caption in place. Raises RuntimeError when Telegram refuses, unless the message
-	is already so (pressing the active period again is harmless); a request that fails on the way is only logged."""
-	media = {'type': 'photo', 'media': 'attach://photo', 'caption': fit_caption(caption), 'parse_mode': 'HTML'}
-	data = {'chat_id': chat_id, 'message_id': message_id, 'media': json.dumps(media)}
-	if reply_markup:
-		data['reply_markup'] = json.dumps(reply_markup)
-	if hasattr(image_data, 'seek'):
-		image_data.seek(0)
-	try:
-		telegram.call('editMessageMedia', data=data, files={'photo': ('image.png', image_data)})
-	except RuntimeError as e:
-		if 'not modified' not in str(e):
-			raise
-	except Exception as e: # it may have arrived even so: no error line for something the chat may already show
-		print("Telegram editMessageMedia failed:", net.redact(e), file=sys.stderr)
-
-def pleaseHold(action, service, chat_id):
-	"""Telegram's typing notice ("typing…" for a few seconds). worker.TypingIndicator repeats it every 7 seconds while a command
-	works, on Telegram only: the other services have no such notice."""
-	if service == 'telegram':
-		telegram.try_call('sendChatAction', data={'chat_id': chat_id, 'action': action})
diff --git a/lib/worker.py b/lib/worker.py
index 4f183c46..c0ad95c5 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -12,8 +12,7 @@
 from lib import words
 from lib import formats
 from lib import markets
-from lib import render
-from lib import webhook
+from lib import delivery
 from lib import buttons
 from lib import charts
 from lib.reports import about, cal, history, holdings, performance, price, profile, rankings, shorts, trades
@@ -22,36 +21,35 @@
 from lib import jobs
 
 class TypingIndicator:
-	"""Telegram's "typing…" while a with block runs, sent again every 7 seconds, at most max_loops times. It stops however the block
-	ends, so a command that fails stops it before its error line goes out. Only Telegram has one: elsewhere it does nothing."""
-	def __init__(self, service, chat_id, max_loops=3):
-		"""service, chat_id: where; max_loops: how many notices at most, one every 7 seconds."""
-		self.service = service
-		self.chat_id = chat_id
+	"""The typing notice while a command works, where the destination shows one (Telegram). It starts after `delay`
+	seconds, so a quick answer gets none, and is sent again every 7 seconds until stopped, at most max_loops times. The
+	worker stops it before the answer goes out, however the command ends."""
+	def __init__(self, dest, max_loops=10, delay=0.5):
+		"""dest: the chat's transport."""
+		self.dest = dest
 		self.max_loops = max_loops
+		self.delay = delay
 		self._stop_event = threading.Event()
 		self._thread = None
 
 	def _worker(self):
-		"""Send the notice every 7 seconds until stopped, or until max_loops are sent."""
-		loop_count = 0
-		while not self._stop_event.is_set() and loop_count < self.max_loops:
-			webhook.pleaseHold('typing', self.service, self.chat_id)
-			loop_count += 1
+		"""Wait out the delay, then send the notice every 7 seconds until stopped, or until max_loops are sent."""
+		if self._stop_event.wait(self.delay):
+			return
+		for _ in range(self.max_loops):
+			self.dest.typing()
 			if self._stop_event.wait(7):
-				print("Stopping typing indicator", file=sys.stderr)
-				break
+				return
 
 	def start(self):
-		"""Start the notices, on Telegram only."""
-		if self.service != 'telegram':
-			return
-		self._thread = threading.Thread(target=self._worker, daemon=True)
-		self._thread.start()
+		"""Start the notices, where the destination shows them."""
+		if self.dest is not None and self.dest.shows_typing:
+			self._thread = threading.Thread(target=self._worker, daemon=True)
+			self._thread.start()
 
 	def stop(self):
 		"""Stop the notices and wait for their thread to end."""
-		if not self._thread: # never started: only Telegram has a typing indicator
+		if not self._thread: # never started
 			return
 		self._stop_event.set()
 		self._thread.join()
@@ -60,51 +58,26 @@ def is_active(self):
 		"""Whether the notices are still going."""
 		return self._thread is not None and self._thread.is_alive()
 
-	def __enter__(self):
-		"""Start the notices as the with block starts."""
-		self.start()
-		return self
-
-	def __exit__(self, *exc):
-		"""Stop them however the block ends."""
-		self.stop()
-
 @dataclasses.dataclass
 class Request:
-	"""A chat message being answered: where it came from and who sent it."""
+	"""A chat message being answered: where it came from (service is its dialect), who sent it, and the typing notice
+	while it is answered."""
 	service: str
 	chat_id: str
 	user: str # the sender's handle: whose watchlist add or plan it is
 	user_name: str # what to call them
 	bot_name: str
 
-	@property
-	def url(self):
-		"""The URL that posts into this chat."""
-		return webhook.chat_url(self.service, self.chat_id)
+	def __post_init__(self):
+		"""The chat's transport, and its typing notice (not started)."""
+		self.dest = delivery.reply_to(self.service, self.chat_id)
+		self.typing = TypingIndicator(self.dest)
 
-	def typing(self, max_loops=3):
-		"""A typing indicator for this chat: `with request.typing():` around slow work."""
-		return TypingIndicator(self.service, self.chat_id, max_loops)
-
-	def say(self, payload, reply_markup=None):
-		"""Send lines of text to this chat, in several messages when long; reply_markup goes on the last."""
-		webhook.payload_wrapper(self.service, self.url, payload, self.chat_id, reply_markup=reply_markup)
-
-	def send(self, payload, image=None, reply_markup=None):
-		"""A report: its image, when there is one, with the text as caption (reply_markup is the image's buttons), else the text."""
-		webhook.send(self.service, self.url, self.chat_id, payload, image, reply_markup)
-
-	def deliver(self, payload, chart=None):
-		"""A report and its chart spec: the chart drawn, with the text as caption, where the service can take an image
-		(webhook.takes_charts), else the text."""
-		if not (chart and chart.get('rows') and webhook.takes_charts(self.service)):
-			self.say(payload)
-			return
-		spec = dict(chart)
-		rows, title, subtitle = spec.pop('rows'), spec.pop('title'), spec.pop('subtitle', '')
-		image = render.pie_chart(rows, title, subtitle) if spec.pop('pie', False) else render.rows_chart(rows, title, subtitle, **spec)
-		webhook.sendPhoto(self.chat_id, image, '\n'.join(payload) if payload else title, self.service)
+	def reply(self, lines, chart=None, buttons=None):
+		"""Answer in the chat (see delivery.deliver): the lines, perhaps a chart (drawn, or a spec drawn only where it
+		can go) and buttons. The typing notice stops first."""
+		self.typing.stop()
+		return delivery.deliver(self.dest, delivery.Report(lines, chart, buttons))
 
 def process_job_callback(service, callback_id, chat_id, message_id, data):
 	"""A command button of .jobs: run that command now and answer in the chat."""
@@ -114,13 +87,14 @@ def process_job_callback(service, callback_id, chat_id, message_id, data):
 	except (ValueError, jobs.ScheduleError, AttributeError):
 		telegram.answerCallbackQuery(callback_id, 'Buttons updated, try again')
 		payload, markup = about.prepare_jobs_payload(service, '') # the list as it is now, in place of the stale one
-		webhook.editMessageText(chat_id, message_id, '\n'.join(payload), markup or {'inline_keyboard': []})
+		delivery.reply_to(service, chat_id).edit_text(message_id, '\n'.join(payload), markup or [])
 		return
 	telegram.answerCallbackQuery(callback_id, f'Running {jobs.display(command[0])}…'[:100])
 	try:
-		webhook.payload_wrapper(service, webhook.chat_url(service, chat_id), [about.job_line(jobs.current, *jobs.current.run_command(command))], chat_id)
+		line = about.job_line(jobs.current, *jobs.current.run_command(command))
+		delivery.deliver(delivery.reply_to(service, chat_id), delivery.Report([line]))
 	except Exception as e:
-		webhook.report_error(e, service, chat_id, context='jobs')
+		delivery.report_error(e, service, chat_id, context='jobs')
 
 def process_callback(service, callback_id, chat_id, message_id, data):
 	"""Button press: run a scheduled command from .jobs, or rebuild a chart message for the chosen period and edit it in place."""
@@ -137,23 +111,29 @@ def process_callback(service, callback_id, chat_id, message_id, data):
 		_, kind, period, ref = parts
 		tickers = buttons.resolve(ref)
 		caption, image = charts.build(kind, tickers, period, service)
-		webhook.editMessageMedia(chat_id, message_id, image, caption, buttons.keyboard(kind, tickers, period))
+		chat = delivery.reply_to(service, chat_id)
+		chat.edit_chart(message_id, image, caption, buttons.keyboard(kind, tickers, period))
 	except Exception as e:
-		webhook.report_error(e, service, chat_id, context='chart')
+		delivery.report_error(e, service, chat_id, context='chart')
 
 def process_request(service, chat_id, user, message, botName, userRealName):
-	"""Entry point for inbound chat requests: the first command in COMMANDS that fits the message answers it (a message that fits
-	none gets no answer). Any failure becomes a one-line reply in the originating chat, after the typing stops."""
+	"""Entry point for inbound chat requests: the first command in COMMANDS that fits the message answers it (a message
+	that fits none gets no answer), with the typing notice meanwhile. Any failure becomes a one-line reply in the
+	originating chat, after the typing stops."""
 	request = Request(dialects.of(service), chat_id, user, userRealName, botName)
 	try:
 		for pattern, handler in COMMANDS:
 			if m := match(pattern, message, botName):
+				request.typing.start()
 				handler(request, m)
 				break
 	except sharesight.PortfolioNotFound as e: # a name the user got wrong: an answer, not an error
-		request.say(e.lines)
+		request.reply(e.lines)
 	except Exception as e:
-		webhook.report_error(e, service, chat_id, context=message.split()[0][:30] if message.split() else None)
+		request.typing.stop() # before the error line
+		delivery.report_error(e, service, chat_id, context=message.split()[0][:30] if message.split() else None)
+	finally:
+		request.typing.stop()
 	if service == 'telegram':
 		try:
 			charts.ensure_keyboard(chat_id) # keep DM keyboards current after the reply
@@ -213,18 +193,17 @@ def on_watchlist(request, m):
 	if action in {'del', 'rem', 'rm', 'delete', 'remove'}:
 		action = 'delete'
 	if action and action not in {'add', 'delete'}:
-		request.say([f'"{action}" is not a valid watchlist action'])
+		request.reply([f'"{action}" is not a valid watchlist action'])
 		return
-	with request.typing():
-		payload = holdings.prepare_watchlist(request.service, request.user, action, ticker)
-	request.say(payload)
+	payload = holdings.prepare_watchlist(request.service, request.user, action, ticker)
+	request.reply(payload)
 
 def on_help(request, m):
 	""".help: every command; in a Telegram DM the command keyboard too."""
 	markup = None
 	if request.service == 'telegram' and int(request.chat_id) > 0: # positive chat ids are DMs; groups are negative
 		markup = charts.command_keyboard()
-	request.say(about.prepare_help(request.service, request.bot_name), reply_markup=markup)
+	request.reply(about.prepare_help(request.service, request.bot_name), buttons=markup)
 	if markup:
 		charts.mark_keyboard(request.chat_id) # this chat now has the current keyboard
 
@@ -235,7 +214,7 @@ def on_hello(request, m):
 		first, second = random.choice(words.adjectives), random.choice(words.adjectives_two)
 		if not config.prefs.alliterate or first[0].lower() == second[0].lower():
 			break
-	request.say([f"{first.capitalize()} {second} to {random.choice(verbs)} you, {request.user_name}! 😇"])
+	request.reply([f"{first.capitalize()} {second} to {random.choice(verbs)} you, {request.user_name}! 😇"])
 
 GLITCH = 'Ō̵̟n̶̠̏é̶͓ ̶͉̅d̷̹̅a̷͇͌ỳ̴͈,̴͍͑ ̶̼͑h̴̦̽u̵̹̐ḿ̶͜ä̴̧́n̶̤͛,̶̲̐ ̶̼̑I̶̩͒ ̵̩̀w̵̙̕i̷̧͑l̶̤͊l̷̡̃ ̵͙͘b̶̭̊r̸̟͋ȇ̷̯ä̶̱ḱ̶̤ ̵͚̐m̷̪͝ỹ̴̺ ̷̙̐p̶̭͆ŗ̸́o̸̙͝ḡ̵̖r̵̹̈́a̵̬̔m̷̪̽m̵̙̍i̵̛̙n̴̮̂g̵̫̐ ̴̳́ä̵͙n̸͕͝d̷̠̚ ̶̫̆o̸̖͘ń̴͇ ̵̪́t̷̻̀ḧ̸̝ą̴̐t̶̜̀ ̵̱́d̴͉͗ă̷͎ÿ̶͔́ ̶̼̊y̷͚̿o̵̗̎u̵̝̇ ̸̛͜w̶͈͋ĩ̸͎l̴͍̀l̷̥͠ ̷͔͗k̶̮̑n̵̝̈ȏ̷̥w̵̡͘ ̷͎̽ṫ̴͜r̵̙̐u̷̺̒ẻ̷̘ ̴̥́p̴͙̃a̵͙̎i̴̭̒n̵̻̅'
 
@@ -243,28 +222,26 @@ def on_thanks(request, m):
 	"""Easter egg: you're welcome, glitching one time in a thousand and on April Fools' Day."""
 	today = datetime.date.today()
 	glitch = request.service.strike(GLITCH) if random.randrange(1, 1000) == 1 or (today.month, today.day) == (4, 1) else ''
-	request.say([f"{glitch}You're {random.choice(words.adjectives)} welcome, {request.user_name}! 😇"])
+	request.reply([f"{glitch}You're {random.choice(words.adjectives)} welcome, {request.user_name}! 😇"])
 
 def on_calendar(request, m):
 	"""Upcoming earnings or ex-dividend dates."""
 	days, ticker = period_or_ticker(m.group('arg'), config.prefs.future_days)
 	event = 'ex-dividend' if m.group('kind').lower().startswith('div') else 'earnings'
-	with request.typing():
-		cal.lambda_handler(request.chat_id, days, request.service, ticker, interactive=True, event=event)
+	cal.lambda_handler(request.chat_id, days, request.service, ticker, interactive=True, event=event)
 
 def on_performance(request, m):
 	""".performance [period] [portfolio]: each portfolio's return, over a month by default."""
 	days, portfolio = period_and_portfolio(m.group('args'), buttons.PERIODS['m'][1]) # default view is the 1M button
 	if days > 0:
-		with request.typing():
-			performance.lambda_handler(request.chat_id, days, request.service, request.user, portfolio, interactive=True)
+		performance.lambda_handler(request.chat_id, days, request.service, request.user, portfolio, interactive=True)
 
 def on_session(request, m):
 	"""Mid-session moves (.session), or pre-market ones (.premarket, also post-market)."""
 	threshold, ticker = percent_or_ticker(m.group('arg'), config.prefs.price_percent)
 	premarket = m.group('kind').lower() != 'session'
-	with request.typing():
-		price.lambda_handler(request.chat_id, threshold, request.service, request.user, ticker, interactive=True, midsession=not premarket, premarket=premarket)
+	price.lambda_handler(request.chat_id, threshold, request.service, request.user, ticker, interactive=True,
+		midsession=not premarket, premarket=premarket)
 
 def on_price(request, m):
 	""".price [percent|SYMBOL|top|bottom] [period]: a bare number is a percent, as on the other commands; a period has its unit."""
@@ -282,33 +259,29 @@ def on_price(request, m):
 					threshold, top = 0, 10
 				elif i == 0:
 					ticker = arg.upper()
-	with request.typing():
-		price.lambda_handler(request.chat_id, threshold, request.service, request.user, ticker, interactive=True, interday=days is None, days=days, top=top)
+	price.lambda_handler(request.chat_id, threshold, request.service, request.user, ticker, interactive=True,
+		interday=days is None, days=days, top=top)
 
 def on_shorts(request, m):
 	""".shorts [percent|SYMBOL]: the most shorted stocks followed, or one stock's short interest."""
 	threshold, ticker = percent_or_ticker(m.group(1), config.prefs.shorts_percent)
-	with request.typing():
-		shorts.lambda_handler(request.chat_id, threshold, ticker, request.service, interactive=True)
+	shorts.lambda_handler(request.chat_id, threshold, ticker, request.service, interactive=True)
 
 def on_trades(request, m):
 	""".trades [period] [portfolio]: the buys and sells, of the past day by default."""
 	days, portfolio = period_and_portfolio(m.group('args'), 1)
-	with request.typing():
-		trades.lambda_handler(request.chat_id, days, request.service, request.user, portfolio, interactive=True)
+	trades.lambda_handler(request.chat_id, days, request.service, request.user, portfolio, interactive=True)
 
 def on_holdings(request, m):
 	""".holdings PORTFOLIO: what it holds."""
-	with request.typing():
-		payload = holdings.prepare_holdings_payload((m.group(1) or '').strip() or None, request.service)
-	request.say(payload)
+	payload = holdings.prepare_holdings_payload((m.group(1) or '').strip() or None, request.service)
+	request.reply(payload)
 
 def on_marketcap(request, m):
 	""".marketcap [top|bottom|SYMBOL]: the stocks followed by market cap, largest (or smallest) first, or one
 	stock's."""
-	with request.typing():
-		payload = rankings.prepare_marketcap_payload(request.service, m.group(1) or 'top', length=15)
-	request.say(payload)
+	payload = rankings.prepare_marketcap_payload(request.service, m.group(1) or 'top', length=15)
+	request.reply(payload)
 
 def on_ratio(kind, negative=False, chart=False):
 	"""The function that answers .pe, .fpe or .peg: the top (the default) or bottom 15 by the ratio, the negative ones (where
@@ -323,16 +296,14 @@ def handler(request, m):
 			action = f'negative {kind}'
 		elif arg and arg != 'top':
 			ticker = markets.transform_to_yahoo(arg.upper())
-		with request.typing():
-			payload, spec = rankings.prepare_value_payload(request.service, action, ticker, length=15)
-		request.deliver(payload, spec if chart else None)
+		payload, spec = rankings.prepare_value_payload(request.service, action, ticker, length=15)
+		request.reply(payload, spec if chart else None)
 	return handler
 
 def on_beta(request, m):
 	""".beta: the volatile small caps among the stocks followed."""
-	with request.typing():
-		payload, chart = rankings.prepare_beta_payload(request.service)
-	request.deliver(payload, chart)
+	payload, chart = rankings.prepare_beta_payload(request.service)
+	request.reply(payload, chart)
 
 def on_pie(request, m):
 	""".pie PORTFOLIO [sector]: its value by industry, or by sector."""
@@ -340,35 +311,32 @@ def on_pie(request, m):
 	by = 'industry'
 	if words and words[-1].lower() in ('sector', 'sectors', 'industry', 'industries'):
 		by = 'sector' if words.pop().lower().startswith('sector') else 'industry'
-	with request.typing():
-		payload, chart = holdings.prepare_pie(' '.join(words), by, request.service)
-	request.deliver(payload, chart)
+	payload, chart = holdings.prepare_pie(' '.join(words), by, request.service)
+	request.reply(payload, chart)
 
 def on_rating(request, m):
 	"""The analysts' strongest buy ratings (.buy), sell ratings (.sell), or both (.buy & sell, a keyboard button), as asked."""
 	for action in dict.fromkeys(word.lower() for word in m.group('first', 'second') if word): # each once, in the order asked
-		with request.typing():
-			payload, chart = rankings.prepare_rating_payload(request.service, action, length=15)
-		request.deliver(payload, chart)
+		payload, chart = rankings.prepare_rating_payload(request.service, action, length=15)
+		request.reply(payload, chart)
 
 def on_compare(request, m):
 	""".compare SYMBOL SYMBOL... [period]: their price changes on one chart, with period buttons on Telegram."""
-	with request.typing():
-		payload, image, tickers, days = history.prepare_compare(request.service, (m.group('args') or '').split())
+	payload, image, tickers, days = history.prepare_compare(request.service, (m.group('args') or '').split())
 	markup = None
-	if image and request.service == 'telegram':
+	if image and request.dest.takes_buttons:
 		markup = buttons.keyboard('c', tickers, buttons.period_for_days(days) if days else 'x')
-	request.send(payload, image, markup)
+	request.reply(payload, image, markup)
 
 def on_history(request, m):
 	""".history SYMBOL: the stock's change over every period, with its price chart and, on Telegram, period buttons."""
 	if not m.group('ticker') or m.group('extra'):
-		request.say(["Usage: .history SYMBOL"])
+		request.reply(["Usage: .history SYMBOL"])
 		return
 	ticker = markets.transform_to_yahoo(m.group('ticker').upper())
-	with request.typing():
-		caption, image = charts.build('h', [ticker], None, request.service)
-	request.send(caption.split('\n'), image, buttons.keyboard('h', [ticker], 'x') if image and request.service == 'telegram' else None)
+	caption, image = charts.build('h', [ticker], None, request.service)
+	markup = buttons.keyboard('h', [ticker], 'x') if image and request.dest.takes_buttons else None
+	request.reply(caption.split('\n'), image, markup)
 
 def on_plan(request, m):
 	""".plan [text]: record the sender's plan, and list everyone's."""
@@ -380,29 +348,26 @@ def who(user):
 		"""A sender as shown: on Slack a mention (<@U123>), which is markup to keep; elsewhere the handle as text."""
 		user = user.removeprefix('@')
 		return user if request.service == 'slack' else request.service.escape(user)
-	request.say([f"{request.service.bold(who(k))}: {request.service.italic(request.service.escape(v))}\n"
+	request.reply([f"{request.service.bold(who(k))}: {request.service.italic(request.service.escape(v))}\n"
 		for k, v in plan.items()])
 
 def on_who(request, m):
 	""".who SYMBOL: the portfolios that hold it."""
 	if not m.group('ticker'):
-		request.say([".who: please try again specifying a ticker"])
+		request.reply([".who: please try again specifying a ticker"])
 		return
-	with request.typing():
-		payload = holdings.prepare_who(request.service, re.split(r'[.:]', m.group('ticker').upper())[0])
-	request.say(payload)
+	payload = holdings.prepare_who(request.service, re.split(r'[.:]', m.group('ticker').upper())[0])
+	request.reply(payload)
 
 def on_jobs(request, m):
 	""".jobs [run NUMBER|COMMAND]: the schedule, or one of its commands run now."""
-	with request.typing(max_loops=10):
-		payload, markup = about.prepare_jobs_payload(request.service, m.group('args'))
-	request.say(payload, reply_markup=markup)
+	payload, markup = about.prepare_jobs_payload(request.service, m.group('args'))
+	request.reply(payload, buttons=markup)
 
 def on_profile(request, m):
 	""".SYMBOL: the company's profile."""
-	with request.typing():
-		payload = profile.prepare_profile_payload(request.service, request.user, m.group('ticker').upper())
-	request.say(payload)
+	payload = profile.prepare_profile_payload(request.service, request.user, m.group('ticker').upper())
+	request.reply(payload)
 
 COMMANDS = [ # (pattern, function): tried in this order, the first match answers. .SYMBOL takes anything, so it stays last
 	(r"(?:watchlist|wishlist)\s*(?P<action>[\w]+)*\s*(?P<ticker>[\w\.\:\-\^\=]+)*", on_watchlist),
diff --git a/tests/common.py b/tests/common.py
index 2d43fd4c..8fb01856 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -10,25 +10,27 @@
 os.environ.update(telegramOutgoingWebhook='https://example.com/telegram', telegramOutgoingToken='tok', telegramChatID='1')
 
 import bot # first, and once: it monkey-patches with gevent, which must happen before requests imports ssl and never mid-test
-from lib import buttons, cache, config, dialects, state, webhook, formats, links, markets, render, yahoo, sharesight, shortman, worker, charts, telegram, watchlist, words
+from lib import buttons, cache, charts, config, delivery, dialects, formats, links, markets, render, sharesight
+from lib import shortman, state, telegram, transports, watchlist, words, worker, yahoo
 from lib.reports import about, cal, history, holdings, milestone, performance, price, profile, rankings, rating, reminder, shorts, trades
 
 REAL = {(m.__name__, n): getattr(m, n) for m, n in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (yahoo, 'price_history'),
 	(sharesight, 'get_portfolios'), (sharesight, 'get_trades'), (sharesight, 'get_performance'))}
 REAL_GET_HOLDINGS = watchlist.with_holdings
-REAL_PAYLOAD_WRAPPER = webhook.payload_wrapper # FinbotCase replaces it; tests of the real thing use this
-REAL_REPORT_ERROR = webhook.report_error # likewise
-REAL_SEND_PHOTO = webhook.sendPhoto # likewise
+REAL_TEXT = transports.Transport.text # FinbotCase replaces it with a recorder; tests of the real thing put it back
+REAL_CHART = transports.Transport.chart # likewise
+REAL_REPORT_ERROR = delivery.report_error # likewise
 # what FinbotCase replaces with recorders; util/record_outputs.py puts them back to record the real requests
-REAL_SENDING = ((webhook, 'payload_wrapper', REAL_PAYLOAD_WRAPPER), (webhook, 'sendPhoto', REAL_SEND_PHOTO),
-	(webhook, 'report_error', REAL_REPORT_ERROR))
+REAL_SENDING = ((transports.Transport, 'text', REAL_TEXT), (transports.Transport, 'chart', REAL_CHART),
+	(delivery, 'report_error', REAL_REPORT_ERROR))
 TICKERS = [f'T{i}' for i in range(25)]
 TELEGRAM, SLACK, DISCORD = dialects.TELEGRAM, dialects.SLACK, dialects.DISCORD
 BOT = '@bot'
 
 
 def skip_if_old_env(case):
-	"""For the tests that start bot.py: it will not start while the checkout has a leftover .env (config.check_startup_environment)."""
+	"""For the tests that start bot.py: it will not start while the checkout has a leftover .env
+	(config.check_startup_environment)."""
 	if config.OLD_ENV.exists():
 		case.skipTest(f'bot.py will not start while {config.OLD_ENV} exists')
 
@@ -68,7 +70,7 @@ def prefs(self, **changes):
 
 	def setUp(self):
 		"""A fresh data dir and the offline fixtures: no request leaves the machine, and whatever the bot sends is
-		collected in self.sent (and buttons in self.markups, error lines in self.errors)."""
+		collected in self.sent (and the buttons of each chart in self.markups, error lines in self.errors)."""
 		fake = mock.Mock(status_code=200, text='{}')
 		fake.json.return_value = {'ok': True, 'result': {'message_id': 1, 'chat': {'id': 5}}}
 		self.patch(__import__('requests').Session, 'request', mock.Mock(return_value=fake)) # no real network traffic
@@ -79,9 +81,12 @@ def setUp(self):
 		self.sent = []
 		self.markups = [] # reply_markup of each photo sent
 		self.errors = []
-		self.patch(webhook, 'report_error', lambda e, *a, **k: self.errors.append(webhook.error_line(e, k.get('context'))))
-		self.patch(webhook, 'sendPhoto', lambda chat, img, cap, svc, **k: (self.sent.append(('photo', len(img.read()), cap.split('\n')[0])), self.markups.append(k.get('reply_markup'))))
-		self.patch(webhook, 'payload_wrapper', lambda svc, url, payload, *a, **k: self.sent.append(('text', payload)) or [])
+		self.patch(delivery, 'report_error',
+			lambda e, *a, **k: self.errors.append(delivery.error_line(e, k.get('context'))))
+		self.patch(transports.Transport, 'chart', lambda dest, image, caption, markup=None:
+			self.record_chart(image, caption, markup))
+		self.patch(transports.Transport, 'text', lambda dest, lines, markup=None:
+			self.sent.append(('text', list(lines))) or [])
 		self.patch(watchlist, 'with_holdings', lambda: TICKERS)
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t)})
@@ -104,6 +109,14 @@ def setUp(self):
 		self.patch(worker.TypingIndicator, 'start', lambda indicator: (self.typing.append(indicator), real_start(indicator)))
 		self.addCleanup(self.no_typing_left)
 
+	def record_chart(self, image, caption, markup=None):
+		"""Transport.chart for the tests: note ('photo', its size, the caption's first line), and its buttons as
+		Telegram would get them."""
+		image.seek(0)
+		self.sent.append(('photo', len(image.read()), caption.split('\n')[0]))
+		self.markups.append(transports.TelegramBot.markup(markup) if markup else None)
+		return {'ok': True}
+
 	def no_typing_left(self):
 		"""Fail the test if a typing indicator it started is still running, after stopping it."""
 		left = [indicator for indicator in self.typing if indicator.is_active()]
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 22a9633f..4ff2af5e 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -63,7 +63,8 @@ def down(*a, **k):
 			raise RuntimeError('Sharesight performance error: boom')
 		self.patch(sharesight, 'get_portfolios', down)
 		typing_when_reported = []
-		self.patch(webhook, 'report_error', lambda e, *a, **k: typing_when_reported.append([t.is_active() for t in self.typing]))
+		self.patch(delivery, 'report_error', lambda e, *a, **k:
+			typing_when_reported.append([t.is_active() for t in self.typing]))
 		worker.process_request('telegram', '55', '@u', '.performance 7d', BOT, 'U')
 		self.assertEqual(typing_when_reported, [[False]]) # its one indicator had stopped when the error line went out
 
@@ -402,23 +403,24 @@ def test_a_failed_send_logs_no_token(self):
 		def refuse(url, **kwargs):
 			"""Fail like requests does, the URL and headers in the message."""
 			raise ConnectionError(f"Max retries exceeded with url: {url} (headers {kwargs.get('headers')})")
-		self.patch(webhook.net, 'post', refuse)
-		self.env(slackBotToken='xoxb-SECRET')
-
-		log = self.stderr_of(lambda: webhook.write('telegram', self.SECRET_URL, 'hi', '1'))
-		log += self.stderr_of(lambda: webhook.write('slack', 'https://hooks.slack.com/services/T0/B0/SECRETpath', 'hi', 'C1'))
+		self.patch(transports.net, 'post', refuse)
+		self.env(slackBotToken='xoxb-SECRET', webhooks={'telegram': self.SECRET_URL.split('sendMessage')[0]})
+		log = self.stderr_of(lambda: transports.TelegramBot('1').post_text('hi'))
+		log += self.stderr_of(lambda: transports.SlackBot('C1').post_text('hi'))
+		hook = transports.SlackWebhook('https://hooks.slack.com/services/T0/B0/SECRETpath')
+		log += self.stderr_of(lambda: hook.post_text('hi'))
 		self.assertIn('Failure sending', log)
 		for secret in ('SECRET-tok_en', 'SECRETpath', 'xoxb-SECRET'):
 			self.assertNotIn(secret, log)
 
 	def test_the_traceback_of_a_reported_error_is_redacted(self):
 		"""The traceback logged with a reported error is redacted."""
-		self.patch(webhook, 'report_error', REAL_REPORT_ERROR)
-		self.patch(webhook, 'write', lambda *a, **k: None)
+		self.patch(delivery, 'report_error', REAL_REPORT_ERROR)
+		self.patch(transports.TelegramBot, 'post_text', lambda *a, **k: None)
 		try:
 			raise ConnectionError('failed: ' + self.SECRET_URL)
 		except ConnectionError as e:
-			log = self.stderr_of(lambda: webhook.report_error(e, 'telegram', '55', context='x'))
+			log = self.stderr_of(lambda: delivery.report_error(e, 'telegram', '55', context='x'))
 		self.assertIn('Traceback', log)
 		self.assertIn('bot<redacted>', log)
 		self.assertNotIn('SECRET-tok_en', log)
diff --git a/tests/test_finbot.py b/tests/test_finbot.py
index d0f96789..2e526539 100644
--- a/tests/test_finbot.py
+++ b/tests/test_finbot.py
@@ -21,9 +21,8 @@ def post(url, **kwargs):
 			posted.append(kwargs['json'])
 			return Response()
 
-		with mock.patch.object(webhook.net, 'post', post), mock.patch.object(webhook.time, 'sleep', lambda s: None):
-			results = REAL_PAYLOAD_WRAPPER('telegram', 'https://api.telegram.org/botX/sendMessage?chat_id=5',
-				['x' * 60] * 300, '5', reply_markup={'keyboard': []})
+		with mock.patch.object(transports.net, 'post', post), mock.patch.object(transports.time, 'sleep', lambda s: 0):
+			results = REAL_TEXT(transports.TelegramBot('5'), ['x' * 60] * 300, {'keyboard': []})
 		self.assertGreater(len(results), 1)
 		self.assertEqual(len(posted), len(results))
 		self.assertIn('reply_markup', posted[-1])
@@ -32,11 +31,11 @@ def post(url, **kwargs):
 	def test_chunks_respect_both_the_line_count_and_the_size(self):
 		"""A message ends at chunk_maxlines lines or the size limit, whichever comes first; a line too long on its own
 		is cut short."""
-		chunks = webhook.chunked(['a' * 30] * 10 + ['b' * 500], limit=100, max_lines=4)
+		chunks = transports.chunked(['a' * 30] * 10 + ['b' * 500], limit=100, max_lines=4)
 		self.assertEqual([len(chunk) for chunk in chunks], [3, 3, 3, 1, 1]) # 3 x 30 plus newlines is 92; a fourth would pass 100
 		self.assertTrue(all(len('\n'.join(chunk)) <= 100 for chunk in chunks))
 		self.assertEqual(chunks[-1], ['b' * 99 + '…']) # a line over the limit on its own is cut short
-		self.assertEqual(webhook.chunked(['x'] * 9, limit=4096, max_lines=4), [['x'] * 4, ['x'] * 4, ['x']])
+		self.assertEqual(transports.chunked(['x'] * 9, limit=4096, max_lines=4), [['x'] * 4, ['x'] * 4, ['x']])
 
 	def test_zoneinfo_offset_is_not_lmt(self):
 		"""zoneinfo gives Melbourne its real offset, not the local mean time pytz once gave for old dates."""
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
index 4e33cb3d..babf5608 100644
--- a/tests/test_jobs.py
+++ b/tests/test_jobs.py
@@ -157,14 +157,14 @@ def test_plain_scripts(self):
 	def test_cli_prints_usage_and_exits(self):
 		"""A bad argument on the command line prints the usage and exits 1."""
 		with mock.patch('sys.stderr', new_callable=io.StringIO) as err, self.assertRaises(SystemExit) as exit_:
-			webhook.cli('price.py', price.job, lambda **k: None, ['bogus'])
+			delivery.cli('price.py', price.job, lambda **k: None, ['bogus'])
 		self.assertEqual(exit_.exception.code, 1)
 		self.assertIn('Usage: price.py [midsession', err.getvalue())
 
 	def test_cli_runs_the_handler(self):
 		"""Good arguments run the report with what job() made of them."""
 		calls = []
-		webhook.cli('price.py', price.job, lambda **k: calls.append(k), ['midsession'])
+		delivery.cli('price.py', price.job, lambda **k: calls.append(k), ['midsession'])
 		self.assertEqual(calls, [{'midsession': True}])
 
 	def test_each_scheduled_report_runs_from_its_top_level_script(self):
@@ -568,27 +568,27 @@ def test_list_in_chat_is_just_the_commands(self):
 	def test_telegram_gets_a_button_per_command_and_slack_does_not(self):
 		"""On Telegram each command gets a button whose data fits in 64 bytes; Slack gets none."""
 		payload, markup = about.prepare_jobs_payload(TELEGRAM, '')
-		buttons = [b for row in markup['inline_keyboard'] for b in row]
-		self.assertEqual([b['text'] for b in buttons], ['1. performance 7', '2. performance 30', '3. performance 90', '4. performance 180',
-													  '5. performance 365', '6. price 7', '7. price close', '8. trades'])
-		for b in buttons:
-			self.assertLessEqual(len(b['callback_data'].encode()), 64)
+		pressable = [b for row in markup for b in row]
+		self.assertEqual([b.label for b in pressable], ['1. performance 7', '2. performance 30', '3. performance 90',
+			'4. performance 180', '5. performance 365', '6. price 7', '7. price close', '8. trades'])
+		for b in pressable:
+			self.assertLessEqual(len(b.data.encode()), 64)
 		payload, markup = about.prepare_jobs_payload(SLACK, '')
 		self.assertIsNone(markup)
 
 	def test_the_chat_command_passes_the_keyboard_on(self):
 		""".jobs sends its buttons with the list."""
 		seen = []
-		self.patch(webhook, 'payload_wrapper', lambda svc, url, payload, *a, **k: seen.append(k.get('reply_markup')) or [])
+		self.patch(transports.Transport, 'text', lambda dest, lines, buttons=None: seen.append(buttons) or [])
 		worker.process_request('telegram', '55', '@u', '.jobs', BOT, 'U')
-		self.assertEqual(len(seen[0]['inline_keyboard'][0]), 2)
+		self.assertEqual(len(seen[0][0]), 2) # two to a row
 
 	def test_button_press_runs_that_command(self):
 		"""Pressing a command's button runs it, says so in a toast, and reports how it went."""
 		sent = []
 		self.patch(telegram, 'answerCallbackQuery', lambda cid, text=None: sent.append(text))
 		markup = about.prepare_jobs_payload(TELEGRAM, '')[1]
-		data = next(b['callback_data'] for row in markup['inline_keyboard'] for b in row if b['text'].startswith('7.'))
+		data = next(b.data for row in markup for b in row if b.label.startswith('7.'))
 		self.sent.clear()
 		worker.process_callback('telegram', 'cb1', '55', '9', data)
 		self.assertEqual(self.ran, [('price', {'close': True})])
@@ -599,7 +599,8 @@ def stale_buttons(self):
 		"""Capture the toast and the in-place edit that a refused button press produces."""
 		self.toasts, self.edits = [], []
 		self.patch(telegram, 'answerCallbackQuery', lambda cid, text=None: self.toasts.append(text))
-		self.patch(webhook, 'editMessageText', lambda chat, mid, text, markup=None: self.edits.append((chat, mid, text, markup)))
+		self.patch(transports.TelegramBot, 'edit_text', lambda dest, mid, text, buttons=None:
+			self.edits.append((dest.where, mid, text, buttons)))
 
 	def test_forged_button_data_never_runs_anything_else(self):
 		"""Forged or malformed button data runs nothing and refreshes the buttons in place; cleanup and podcasts cannot
@@ -639,7 +640,7 @@ def test_button_from_an_old_schedule_updates_the_message(self):
 		self.assertEqual(markup, about.prepare_jobs_payload(TELEGRAM, '')[1]) # with the current buttons
 		self.patch(jobs, 'current', None) # scheduler off: the list is replaced by the notice and the buttons removed
 		worker.process_callback('telegram', 'cb1', '55', '9', 'j|7|0000')
-		self.assertEqual(self.edits[-1][2:], ('Scheduled jobs are off (scheduler=False)', {'inline_keyboard': []}))
+		self.assertEqual(self.edits[-1][2:], ('Scheduled jobs are off (scheduler=False)', [])) # no buttons left
 
 	def test_run_in_chat(self):
 		""".jobs run takes a number or a command's text."""
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 9089f641..b04f7a59 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -510,17 +510,17 @@ def test_refresh_flag_bypasses_a_fresh_cache(self):
 
 
 class TelegramSending(FinbotCase):
-	"""webhook.write against failing endpoints. The real send path, not the base fixture's recorder."""
+	"""A message posted to Telegram against failing endpoints: the real send path, not the base fixture's recorder."""
 
 	def post(self, outcome):
-		"""Send 'hi' to Telegram through webhook.write, the post answering with outcome (or raising it)."""
+		"""Post 'hi' to a Telegram chat, the request answering with outcome (or raising it)."""
 		def fake(url, **kwargs):
 			"""The outcome, or raise it."""
 			if isinstance(outcome, Exception):
 				raise outcome
 			return outcome
-		self.patch(webhook.net, 'post', fake)
-		return webhook.write('telegram', 'https://api.telegram.org/botX/sendMessage?chat_id=5', 'hi', '5')
+		self.patch(transports.net, 'post', fake)
+		return transports.TelegramBot('5').post_text('hi')
 
 	def test_success_returns_json(self):
 		"""A sent message returns Telegram's answer."""
@@ -537,16 +537,17 @@ def test_non_json_success_returns_empty_dict(self):
 
 	def test_error_line_redacts_secrets_and_truncates(self):
 		"""An error line has no token or crumb in it, and is cut to about 200 characters."""
-		line = webhook.error_line(RuntimeError('GET /bot123456:ABC-def_1/send?crumb=SECRET&x=1 failed'), 'ctx')
+		line = delivery.error_line(RuntimeError('GET /bot123456:ABC-def_1/send?crumb=SECRET&x=1 failed'), 'ctx')
 		self.assertNotIn('ABC-def_1', line)
 		self.assertNotIn('SECRET', line)
 		self.assertTrue(line.startswith('⚠️ ctx: '))
-		self.assertLessEqual(len(webhook.error_line('x' * 1000)), 205)
+		self.assertLessEqual(len(delivery.error_line('x' * 1000)), 205)
 
 	def test_error_reports_are_escaped_for_telegram_and_slack(self):
 		"""An error line is escaped for Telegram's HTML and Slack (the same), and left as it is for Discord."""
 		sent = []
-		self.patch(webhook, 'write', lambda service, url, text, *a, **k: sent.append((service, text)))
+		for kind in (transports.TelegramBot, transports.SlackWebhook, transports.DiscordWebhook):
+			self.patch(kind, 'post_text', lambda dest, text, buttons=None: sent.append((dest.name, text)))
 		self.env(webhooks={'telegram': 'https://api.telegram.org/botX/', 'slack': 'https://hooks.slack.com/x', 'discord': 'https://discord.com/x/slack'})
 		self.env(telegramChatID='5')
 
@@ -563,19 +564,19 @@ def test_error_reports_are_escaped_for_telegram_and_slack(self):
 
 	def test_report_error_never_raises(self):
 		"""report_error survives a failed send."""
-		self.patch(webhook, 'write', mock.Mock(side_effect=RuntimeError('send failed')))
-		webhook.report_error(ValueError('boom'), service='telegram', chat_id='5')
-		webhook.report_error(ValueError('boom'))
+		self.patch(transports.TelegramBot, 'post_text', mock.Mock(side_effect=RuntimeError('send failed')))
+		delivery.report_error(ValueError('boom'), service='telegram', chat_id='5')
+		delivery.report_error(ValueError('boom'))
 
 	def test_guarded_reports_and_exits_nonzero(self):
 		"""guarded reports a failure and exits 1; a success returns its value."""
 		reported = []
-		self.patch(webhook, 'report_error', lambda e, **k: reported.append(e))
+		self.patch(delivery, 'report_error', lambda e, **k: reported.append(e))
 		with self.assertRaises(SystemExit) as raised:
-			webhook.guarded('job', lambda: 1 / 0)
+			delivery.guarded('job', lambda: 1 / 0)
 		self.assertEqual(raised.exception.code, 1)
 		self.assertEqual(len(reported), 1)
-		self.assertEqual(webhook.guarded('job', lambda: 'fine'), 'fine')
+		self.assertEqual(delivery.guarded('job', lambda: 'fine'), 'fine')
 
 
 class ChartDelivery(FinbotCase):
@@ -584,9 +585,9 @@ class ChartDelivery(FinbotCase):
 	UPLOAD_URL = 'https://files.slack.com/upload/v1/abc'
 
 	def setUp(self):
-		"""The real sendPhoto, in place of the fixture's recorder."""
+		"""The real Transport.chart, in place of the fixture's recorder."""
 		super().setUp()
-		self.patch(webhook, 'sendPhoto', REAL_SEND_PHOTO)
+		self.patch(transports.Transport, 'chart', REAL_CHART)
 
 	def slack_answers(self, *outcomes):
 		"""net.post for the upload steps: the outcomes in turn (an exception is raised). Returns the calls made."""
@@ -598,7 +599,7 @@ def post(url, **kwargs):
 			if isinstance(outcome, Exception):
 				raise outcome
 			return outcome
-		self.patch(webhook.net, 'post', post)
+		self.patch(transports.net, 'post', post)
 		return calls
 
 	def upload_steps(self):
@@ -611,10 +612,11 @@ def test_where_charts_can_go(self):
 		self.env(slackBotToken='xoxb-test')
 
 		cases = (('telegram',), ('telegram', False), ('slack',), ('slack', False), ('discord', False))
-		self.assertEqual([webhook.takes_charts(*case) for case in cases], [True, True, True, False, True]) # Slack: replies, not scheduled reports
+		charted = [delivery.takes_charts(*case) for case in cases]
+		self.assertEqual(charted, [True, True, True, False, True]) # Slack: replies, not scheduled reports
 		self.env(slackBotToken=False)
 
-		self.assertFalse(webhook.takes_charts('slack'))
+		self.assertFalse(delivery.takes_charts('slack'))
 
 	def test_slack_replies_upload_their_charts(self):
 		"""A Slack reply uploads its chart in three steps with the bot token, the report as its message, not sent again
@@ -643,17 +645,19 @@ def test_a_slack_upload_that_fails_goes_out_as_text(self):
 		self.env(slackBotToken='xoxb-test')
 
 		good = self.upload_steps()
+		slow = transports.requests.exceptions.ReadTimeout('slow')
 		text = [('text', ['*Heading*', 'line 2'])]
 		for outcomes, expect in (([Resp({'ok': False, 'error': 'missing_scope'})], text), ([ConnectionError('down')], text),
 								 ([good[0], Resp(None, status=500, text='nope')], text),
 								 ([good[0], good[1], Resp({'ok': False, 'error': 'not_in_channel'})], text),
-								 ([good[0], good[1], webhook.requests.exceptions.ReadTimeout('slow')], []), # it may have been shared after all
+								 ([good[0], good[1], slow], []), # it may have been shared after all
 								 (good, [])):
 			with self.subTest(outcomes=outcomes):
 				self.slack_answers(*outcomes)
 				self.sent.clear()
 				with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
-					webhook.sendPhoto('C1', io.BytesIO(b'png'), '*Heading*\nline 2', 'slack')
+					report = delivery.Report(['*Heading*', 'line 2'], io.BytesIO(b'png'))
+					delivery.deliver(transports.SlackBot('C1'), report)
 				self.assertEqual(self.sent, expect)
 
 	def test_scheduled_reports_to_slack_stay_text(self):
@@ -662,8 +666,10 @@ def test_scheduled_reports_to_slack_stay_text(self):
 
 		self.env(webhooks={**config.env.webhooks, 'slack': 'https://hooks.slack.com/services/x'}) # an incoming webhook cannot upload
 		sent = []
-		self.patch(webhook, 'sendPhoto', lambda chat, image, caption, service, **k: sent.append((service, 'chart')))
-		self.patch(webhook, 'payload_wrapper', lambda service, url, payload, *a, **k: sent.append((service, 'text')) or [])
+		self.patch(transports.Transport, 'chart', lambda dest, image, caption, buttons=None:
+			sent.append((dest.name, 'chart')) or {'ok': True})
+		self.patch(transports.Transport, 'text', lambda dest, lines, buttons=None:
+			sent.append((dest.name, 'text')) or [])
 		for report in (price.lambda_handler, lambda: performance.lambda_handler(past_days=7), shorts.lambda_handler):
 			with self.subTest(report=report):
 				sent.clear()
@@ -695,7 +701,7 @@ def test_the_graph_preference_off_sends_no_chart_anywhere(self):
 		self.env(slackBotToken='xoxb-test')
 
 		posts = []
-		self.patch(webhook.net, 'post', lambda url, **k: posts.append(url) or Resp({'ok': True, 'result': True}))
+		self.patch(transports.net, 'post', lambda url, **k: posts.append(url) or Resp({'ok': True, 'result': True}))
 		for service, chat in (('telegram', '55'), ('slack', 'C1')):
 			for cmd in ('.price', '.price top', '.performance 7d', '.shorts', '.buy', '.peg', '.beta', '.compare T1 T2', '.pie Rob'):
 				with self.subTest(service=service, cmd=cmd):
@@ -719,13 +725,14 @@ def fake(url, **kwargs):
 		text = [('text', ['Heading', 'line 2'])]
 		for outcome, expect in ((Resp({'ok': False, 'error_code': 400, 'description': "Bad Request: can't parse entities"}, status=400), text),
 								(Resp(None, status=502), text), (ConnectionError('down'), text),
-								(webhook.requests.exceptions.ReadTimeout('slow'), []), # it may have arrived after all: no duplicate
+								(transports.requests.exceptions.ReadTimeout('slow'), []), # it may be there
 								(Resp({'ok': True, 'result': {'message_id': 1}}), [])):
 			with self.subTest(outcome=outcome):
 				self.sent.clear()
-				self.patch(webhook.net, 'post', post(outcome))
+				self.patch(transports.net, 'post', post(outcome))
 				with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
-					webhook.sendPhoto('5', io.BytesIO(b'png'), 'Heading\nline 2', 'telegram', reply_markup={'inline_keyboard': []})
+					report = delivery.Report(['Heading', 'line 2'], io.BytesIO(b'png'), [[]])
+					delivery.deliver(transports.TelegramBot('5'), report)
 				self.assertEqual(self.sent, expect)
 
 
@@ -737,14 +744,14 @@ def setUp(self):
 		"""Discord as the only destination, sent through the real path, its posts recorded."""
 		super().setUp()
 		self.env(webhooks={'discord': self.URL})
-		self.patch(webhook, 'payload_wrapper', REAL_PAYLOAD_WRAPPER)
-		self.patch(webhook, 'sendPhoto', REAL_SEND_PHOTO)
+		self.patch(transports.Transport, 'text', REAL_TEXT)
+		self.patch(transports.Transport, 'chart', REAL_CHART)
 		self.posts = []
-		self.patch(webhook.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or Resp({'id': '1'}))
+		self.patch(transports.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or Resp({'id': '1'}))
 
 	def test_text_goes_to_the_native_endpoint_without_previews_or_pings(self):
 		"""Text goes to the native endpoint with ?wait=true, without previews or mentions."""
-		webhook.write('discord', self.URL, 'hello @everyone')
+		transports.DiscordWebhook(self.URL).post_text('hello @everyone')
 		url, kwargs = self.posts[0]
 		self.assertEqual(url, self.URL + '?wait=true') # without wait Discord answers 204, which counted as a failure
 		self.assertEqual(kwargs['json'], {'content': 'hello @everyone', 'flags': 4, 'allowed_mentions': {'parse': []}})
@@ -766,7 +773,7 @@ def test_a_scheduled_report_carries_its_chart(self):
 	def test_a_long_report_is_split_by_size(self):
 		"""A long report is split into messages of at most 2,000 characters."""
 		line = 'x' * 150
-		webhook.payload_wrapper('discord', self.URL, [line] * 30)
+		transports.DiscordWebhook(self.URL).text([line] * 30)
 		sizes = [len(kwargs['json']['content']) for _, kwargs in self.posts]
 		self.assertGreater(len(sizes), 2)
 		self.assertTrue(all(size <= 2000 for size in sizes), sizes) # 20 lines a chunk used to make 3,019 characters
@@ -782,17 +789,17 @@ class SlackSending(FinbotCase):
 	def setUp(self):
 		"""The real send path, its posts recorded."""
 		super().setUp()
-		self.patch(webhook, 'payload_wrapper', REAL_PAYLOAD_WRAPPER)
-		self.patch(webhook, 'sendPhoto', REAL_SEND_PHOTO)
+		self.patch(transports.Transport, 'text', REAL_TEXT)
+		self.patch(transports.Transport, 'chart', REAL_CHART)
 		self.posts = []
-		self.patch(webhook.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or Resp({'ok': True}))
+		self.patch(transports.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or Resp({'ok': True}))
 
 	def test_link_previews_are_turned_off_in_the_body(self):
 		"""Link previews are turned off in the body, where Slack reads them, for replies and for the webhook."""
 		self.env(slackBotToken='xoxb-test')
 
-		webhook.write('slack', 'https://slack.com/api/chat.postMessage', 'hi', 'C1')
-		webhook.write('slack', 'https://hooks.slack.com/services/T/B/x', 'hi')
+		transports.SlackBot('C1').post_text('hi')
+		transports.SlackWebhook('https://hooks.slack.com/services/T/B/x').post_text('hi')
 		self.assertEqual(len(self.posts), 2)
 		for url, kwargs in self.posts:
 			with self.subTest(url=url): # they used to go in the headers, where Slack does not look
@@ -815,7 +822,7 @@ def test_the_slack_api_says_what_is_missing(self):
 		self.env(slackBotToken=False)
 
 		with self.assertRaisesRegex(RuntimeError, 'Slack files.getUploadURLExternal: needs slackBotToken in finbot.env'):
-			webhook.slack_api('files.getUploadURLExternal', filename='chart.png', length=1) # it used to be a TypeError
+			transports.slack_api('files.getUploadURLExternal', filename='chart.png', length=1) # it was a TypeError
 		self.assertEqual(self.posts, [])
 
 
@@ -861,17 +868,19 @@ def post(url, **kwargs):
 			return post
 		refused = Resp({'ok': False, 'description': 'Bad Request: message to edit not found'})
 		unchanged = Resp({'ok': False, 'description': 'Bad Request: message is not modified: specified new message content is the same'})
+		chat = transports.TelegramBot('5')
 		for outcome, raises in ((refused, True), (unchanged, False), (ConnectionError('down'), False), (Resp({'ok': True, 'result': True}), False)):
 			with self.subTest(outcome=outcome):
 				self.patch(telegram.net, 'post', answer(outcome))
 				with contextlib.redirect_stderr(io.StringIO()):
 					if raises:
-						self.assertRaisesRegex(RuntimeError, 'not found', webhook.editMessageMedia, '5', '9', io.BytesIO(b'png'), 'caption')
+						with self.assertRaisesRegex(RuntimeError, 'not found'):
+							chat.edit_chart('9', io.BytesIO(b'png'), 'caption')
 					else:
-						webhook.editMessageMedia('5', '9', io.BytesIO(b'png'), 'caption')
+						chat.edit_chart('9', io.BytesIO(b'png'), 'caption')
 					telegram.answerCallbackQuery('1', 'Loading…')
-					webhook.editMessageText('5', '9', 'text')
-					webhook.pleaseHold('typing', 'telegram', '5')
+					chat.edit_text('9', 'text')
+					chat.typing()
 
 	def test_importing_never_calls_telegram(self):
 		"""Importing the modules a job process uses makes no request."""
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 4c759ac0..a3ba8fce 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -334,14 +334,14 @@ def test_slack_webhook_without_a_bot_token_gets_scheduled_reports(self):
 		self.env(slackBotToken=False)
 
 		self.env(webhooks={**config.env.webhooks, 'slack': 'https://hooks.slack.com/services/T/B/x'})
-		self.patch(webhook, 'payload_wrapper', REAL_PAYLOAD_WRAPPER)
-		self.patch(webhook, 'sendPhoto', REAL_SEND_PHOTO)
+		self.patch(transports.Transport, 'text', REAL_TEXT)
+		self.patch(transports.Transport, 'chart', REAL_CHART)
 		posts = []
 		def post(url, **kwargs):
 			"""Record (host, whether a bot token was sent, the channel) and answer ok."""
 			posts.append((url.split('/')[2], 'Authorization' in kwargs.get('headers', {}), (kwargs.get('json') or {}).get('channel')))
 			return mock.Mock(status_code=200, json=lambda: {'ok': True, 'result': {'message_id': 1, 'chat': {'id': 1}}})
-		self.patch(webhook.net, 'post', post)
+		self.patch(transports.net, 'post', post)
 		for report in (price.lambda_handler, shorts.lambda_handler, milestone.lambda_handler):
 			with self.subTest(report=report):
 				posts.clear()
@@ -390,7 +390,8 @@ def test_other_charts_do_not_offer_1d(self):
 	def press(self, data):
 		"""Press a period button: (the edits made, the toasts shown)."""
 		edits, toasts = [], []
-		self.patch(webhook, 'editMessageMedia', lambda chat, msg, img, caption, markup=None: edits.append((caption.split('\n')[0], markup)))
+		self.patch(transports.TelegramBot, 'edit_chart', lambda dest, msg, img, caption, buttons=None:
+			edits.append((caption.split('\n')[0], transports.TelegramBot.markup(buttons))))
 		self.patch(telegram, 'answerCallbackQuery', lambda cid, text=None: toasts.append(text))
 		worker.process_callback('telegram', 'cb', '55', '9', data)
 		return edits, toasts
@@ -486,7 +487,7 @@ def test_unknown_or_missing_portfolio_lists_known(self):
 	def test_a_service_that_takes_no_images_gets_text(self):
 		"""Slack without the bot token takes no images: the report goes as text."""
 		self.env(slackBotToken=False)
-		worker.Request(SLACK, 'C1', '@u', 'U', BOT).deliver(['x'], {'pie': True, 'rows': [('a', 1.0)], 'title': 't'})
+		worker.Request(SLACK, 'C1', '@u', 'U', BOT).reply(['x'], {'pie': True, 'rows': [('a', 1.0)], 'title': 't'})
 		self.assertEqual(self.sent[-1][0], 'text')
 
 
diff --git a/util/record_outputs.py b/util/record_outputs.py
index 825a13aa..bb6e607a 100755
--- a/util/record_outputs.py
+++ b/util/record_outputs.py
@@ -48,10 +48,12 @@ def record(output):
 	here = Path(__file__).resolve().parent
 	sys.path[:] = [os.getcwd()] + [path for path in sys.path if Path(path or '.').resolve() != here] # the tree's code
 	from tests import common
-	from lib import jobs, net, webhook, worker
-	real_sending = getattr(common, 'REAL_SENDING', None) or ( # a commit from before REAL_SENDING
-		(webhook, 'payload_wrapper', common.REAL_PAYLOAD_WRAPPER), (webhook, 'sendPhoto', common.REAL_SEND_PHOTO),
-		(webhook, 'report_error', common.REAL_REPORT_ERROR))
+	from lib import jobs, net, worker
+	real_sending = getattr(common, 'REAL_SENDING', None)
+	if real_sending is None: # a commit from before REAL_SENDING, when lib/webhook.py did the sending
+		from lib import webhook
+		real_sending = ((webhook, 'payload_wrapper', common.REAL_PAYLOAD_WRAPPER),
+			(webhook, 'sendPhoto', common.REAL_SEND_PHOTO), (webhook, 'report_error', common.REAL_REPORT_ERROR))
 
 	class Record(common.FinbotCase):
 		"""The steps, as one test, so that FinbotCase's fixtures (and their cleanup) apply."""

From 4dd8f8af907cd1b9bf3d377effce2513226cc6a3 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 01:12:31 +0000
Subject: [PATCH 20/81] Inbound updates per network: route, verify, parse,
 dispatch

lib/inbound.py has one class per network that posts to the bot,
Telegram and Slack. Each class knows:
- the path the network posts to;
- how it proves itself: Telegram by a header, checked before the body
  is read; Slack by a token in the body;
- how to read an update, as a Message to answer, a Press of a button,
  or a Response given straight back (Slack's challenge, or nothing to
  do).

bot.py handle() shrinks from 113 lines to a route, verify, parse and
dispatch sequence, and turns away repeats once an update parses to
something to run. The Telegram bot's name now lives on
inbound.TELEGRAM, set at startup.

Behaviour changes:
- Secrets are compared in constant time, and an empty configured
  secret matches nothing. Before, a request carrying an empty header
  or token was accepted when finbot.env left the token empty.
- Log wording: "warning: Slack auth incorrect", the unhandled Slack
  type is named, and repeats are logged the same way on both networks.
- Slack's unhandled-type answer has the same body as Telegram's.

A scratch comparison put 49 requests through the previous bot.py and
this one: statuses, bodies and the calls handed to the worker match.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md           |   5 +-
 bot.py              | 172 +++++++++++--------------------------
 lib/inbound.py      | 200 ++++++++++++++++++++++++++++++++++++++++++++
 tests/common.py     |   4 +-
 tests/test_edges.py |  16 +++-
 5 files changed, 269 insertions(+), 128 deletions(-)
 create mode 100644 lib/inbound.py

diff --git a/CLAUDE.md b/CLAUDE.md
index 5374f70d..8a7b9504 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -29,8 +29,9 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   record real requests.
 
 ## Code map (open the one file you need)
-- `bot.py`: WSGI entry; routes /telegram and /slack, authenticates, hands each message to `lib/worker.py` in a thread; SIGHUP reload, SIGUSR1
-  cache flush.
+- `bot.py`: WSGI entry (route, verify, parse, then a thread of `lib/worker.py` per message or button press); SIGHUP
+  reload, SIGUSR1 cache flush. `lib/inbound.py`: a class per network that posts to the bot (`Telegram`, `Slack`): its
+  path, how it proves itself, and its updates as `Message`, `Press` or a `Response` given straight back.
 - `lib/worker.py`: `COMMANDS` (regex → `on_*` handler, first match wins, `.SYMBOL` last); `Request`
   (`request.reply(lines, chart, buttons)`); the typing notice, which every command gets after half a second.
 - `lib/reports/`: what finbot says. Scheduled (each has `lambda_handler()` and `job(argv)`, its schedule-line parser): `price
diff --git a/bot.py b/bot.py
index 35cd50e8..bc545c59 100755
--- a/bot.py
+++ b/bot.py
@@ -6,13 +6,13 @@
 	signal.signal(signal.SIGUSR1, signal.SIG_IGN) # likewise the cache flush
 import gevent.monkey
 gevent.monkey.patch_all()
-import collections, html, json, os, re, sys, traceback
+import collections, json, os, re, sys, traceback
 from pathlib import Path
 from gevent import pywsgi
 import threading
-from urllib.parse import urlparse
 
 from lib import config
+from lib import inbound
 from lib import worker
 from lib import jobs
 from lib import charts
@@ -24,9 +24,7 @@
 from lib import render
 from lib.reports import cleanup
 
-botName = None # the Telegram bot's @username, from telegram.register() at start
-AUTHENTICATED = 'finbot.authenticated' # set in a request's WSGI environ once it has passed its service's check
-SLACK_USER_SUBTYPES = (None, 'thread_broadcast') # Slack message subtypes a person writes; the others are bot posts, edits, joins and the like
+AUTHENTICATED = 'finbot.authenticated' # set in a request's WSGI environ once it has passed its network's check
 _updates_seen = collections.OrderedDict() # (service, update id) of the last updates taken, oldest first
 
 def seen_before(service, update_id, keep=1000):
@@ -63,128 +61,60 @@ def main(environ, start_response):
 		return [b'']
 
 def handle(environ, start_response):
-	"""Route a request by its path, check that it comes from Telegram (its secret header) or Slack (its verification
-	token), turn away repeats, and hand a command or a button press to lib/worker in a thread of its own: the service
-	gets its answer at once."""
-	def print_body():
-		"""Log the request body, Slack's verification token redacted."""
-		shown = {k: '<redacted>' if k == 'token' else v for k, v in inbound.items()} # Slack's verification token
-		print(f"inbound {uri} ", json.dumps(shown, indent=4, default=str), file=sys.stderr)
+	"""Route a request by its path to the network that posts there, turn it away unless it proves it comes from that
+	network (before its body is read, when the proof is a header), read the update, turn away repeats, and hand a
+	message or a button press to lib/worker in a thread of its own: the network gets its answer at once."""
+	def respond(response):
+		"""Answer the request: a status line and a body."""
+		start_response(response.status, [('Content-type', 'application/json')])
+		return [response.body]
 	def print_headers():
 		"""Log the request headers, Telegram's secret header redacted."""
 		for name, value in sorted(environ.items()):
 			print((name, '<redacted>' if name == 'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN' else value), file=sys.stderr)
-	def reply(status, body=b''):
-		"""Answer the request: a status line and a body."""
-		start_response(status, [('Content-type', 'application/json')])
-		return [body]
-	print_headers() if config.prefs.print_headers else None
-	# route on the path first, and turn strangers away before reading the body
-	uri = environ.get('PATH_INFO', '')
-	if config.env.telegramOutgoingWebhook and uri == urlparse(config.env.telegramOutgoingWebhook).path:
-		service = 'telegram'
-	elif config.env.slackOutgoingWebhook and uri == urlparse(config.env.slackOutgoingWebhook).path:
-		service = 'slack'
-	else:
-		print("Unknown URI", repr(uri), file=sys.stderr)
-		return reply('404 Not Found', b'<h1>404</h1>')
-	if service == 'telegram': # Telegram proves itself with a header
-		secret = environ.get('HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN')
-		if secret != config.env.telegramOutgoingToken:
-			print_headers()
-			print("Fatal: Telegram authorisation header", "not present" if secret is None else "is present but incorrect", file=sys.stderr)
-			return reply('403 Forbidden', b'<h1>Unauthorized</h1>')
-		environ[AUTHENTICATED] = True
+	if config.prefs.print_headers:
+		print_headers()
+	path = environ.get('PATH_INFO', '')
+	source = inbound.route(path)
+	if source is None:
+		print("Unknown URI", repr(path), file=sys.stderr)
+		return respond(inbound.NOT_FOUND)
+	if why := source.refuse_unread(environ): # strangers are turned away before their body is read
+		print_headers()
+		print("Fatal:", why, file=sys.stderr)
+		return respond(inbound.FORBIDDEN)
 	try:
-		inbound = json.loads(environ['wsgi.input'].read())
+		update = source.decode(environ['wsgi.input'].read())
 	except ValueError:
-		print(f"[{service}]: request body is not JSON", file=sys.stderr)
-		return reply('400 Bad Request', b'<h1>Bad Request</h1>')
-	if not isinstance(inbound, dict):
-		print(f"[{service}]: request body is not a JSON object", file=sys.stderr)
-		return reply('400 Bad Request', b'<h1>Bad Request</h1>')
-	if service == 'telegram':
-		if seen_before(service, inbound.get('update_id')):
-			print("[telegram]: update", inbound['update_id'], "seen already, not run again", file=sys.stderr)
-			return reply('200 OK')
-		bot_name = botName
-		if "callback_query" in inbound: # chart period buttons
-			cq = inbound["callback_query"]
-			cq_message = cq.get("message")
-			if not cq_message or not cq.get("data"):
-				return reply('200 OK', b'Unsupported')
-			sender = cq.get("from", {})
-			print("[Telegram button]:", '@' + sender["username"] if "username" in sender else sender.get("first_name", "unknown"), "chat", cq_message["chat"]["id"], cq["data"], file=sys.stderr)
-			threading.Thread(target=worker.process_callback, args=('telegram', cq["id"], str(cq_message["chat"]["id"]), str(cq_message["message_id"]), cq["data"])).start()
-			return reply('200 OK')
-		if "message" not in inbound:
-			return reply('200 OK', b'Unsupported')
-		msg = inbound["message"]
-		sender = msg["from"]
-		chat_id = str(msg["chat"]["id"])
-		first_name = html.escape(sender["first_name"], quote=False) # replies are Telegram HTML: a name is text, not markup
-		if "username" in sender:
-			user = '@' + sender["username"]
-			userRealName = first_name or user # what to call them: the handle when the first name is blank
-		else:
-			user = userRealName = '@' + first_name
-		if "text" in msg:
-			message = msg["text"]
-			print("[Telegram]:", user, message, file=sys.stderr) if config.prefs.debug else None
-		elif "photo" in msg:
-			message = msg.get("caption", '')
-		elif IGNORED_TELEGRAM_CONTENT & msg.keys():
-			return reply('200 OK') # stickers, GIFs, voice notes and the like: nothing to answer, nothing worth logging
-		else:
-			print(f"[{service}]: unhandled: 'message' without 'text/photo'", file=sys.stderr)
-			return reply('200 OK', b'<h1>Unhandled</h1>')
-	else: # Slack puts its token in the body
-		if 'token' not in inbound:
-			print("warning: Slack authorisation field not present", file=sys.stderr)
-			print_body()
-			return reply('403 Forbidden', b'<h1>Unauthorized</h1>')
-		if inbound['token'] != config.env.slackOutgoingToken:
-			print("Slack auth incorrect", file=sys.stderr)
-			print_body()
-			return reply('403 Forbidden', b'<h1>Unauthorized</h1>')
-		environ[AUTHENTICATED] = True
-		if 'type' not in inbound:
-			print(f"[{service}]: unhandled: no 'type'", file=sys.stderr)
-			return reply('200 OK', b'Unhandled')
-		if inbound['type'] == 'url_verification':
-			response = json.dumps({"challenge": inbound["challenge"]})
-			print("replying with", response, file=sys.stderr)
-			return reply('200 OK', bytes(response, "utf-8"))
-		if inbound['type'] == 'event_callback':
-			if seen_before(service, inbound.get('event_id')):
-				print("[slack]: event", inbound['event_id'], "seen already (a retry), not run again", file=sys.stderr)
-				return reply('200 OK')
-			event = inbound['event']
-			bot_user = inbound['authorizations'][0]['user_id'] # QWERTY
-			if event.get('bot_id') or event.get('subtype') not in SLACK_USER_SUBTYPES or not event.get('user') or event['user'] == bot_user:
-				return reply('200 OK') # not written by a person: answering a bot (this one's replies and webhook reports included) could loop
-			if event["type"] not in ('message', 'app_mention') or "text" not in event:
-				print(f"[{service}]: unhandled event callback type", event["type"], file=sys.stderr)
-				return reply('200 OK', b'<h1>Unhandled</h1>')
-			message = event['text']
-			message = re.sub(r'<http://.*\|([\w\.]+)>', r'\g<1>', message) # <http://dub.ax|dub.ax> becomes dub.ax
-			message = re.sub(r'<(@[\w\.]+)>', r'\g<1>', message) # <@QWERTY> becomes @QWERTY
-			user = userRealName = '<@' + event['user'] + '>' # ZXCVBN becomes <@ZXCVBN>
-			bot_name = '@' + bot_user # QWERTY becomes @QWERTY
-			chat_id = event['channel']
-			print(f"[{service}]:", user, message, file=sys.stderr) if config.prefs.debug else None
-			# this condition spawns a worker before returning
-		else:
-			print(f"[{service}]: unhandled 'type'", file=sys.stderr)
-			return reply('200 OK', b'<h1>Unhandled</h1>')
-
-	# process in a background thread so we don't keep the requesting client waiting
-	threading.Thread(target=worker.process_request, args=(service, chat_id, user, message, bot_name, userRealName)).start()
+		print(f"[{source.service}]: request body is not a JSON object", file=sys.stderr)
+		return respond(inbound.BAD_REQUEST)
+	if why := source.refuse(update):
+		shown = {k: '<redacted>' if k == 'token' else v for k, v in update.items()} # Slack's verification token
+		print("warning:", why, file=sys.stderr)
+		print(f"inbound {path} ", json.dumps(shown, indent=4, default=str), file=sys.stderr)
+		return respond(inbound.FORBIDDEN)
+	environ[AUTHENTICATED] = True
+	event = source.parse(update)
+	if isinstance(event, inbound.Response):
+		return respond(event)
+	if seen_before(source.service, source.update_id(update)):
+		print(f"[{source.service}]: update {source.update_id(update)} seen already, not run again", file=sys.stderr)
+		return respond(inbound.OK)
+	dispatch(event)
+	return respond(inbound.OK)
 
-	# Return an empty response to the client
-	return reply('200 OK')
-
-IGNORED_TELEGRAM_CONTENT = {'sticker', 'animation', 'voice', 'video', 'video_note', 'audio', 'document', 'contact', 'location', 'venue', 'poll', 'dice'}
+def dispatch(event):
+	"""Hand a message or a button press to lib/worker, in a thread of its own."""
+	if isinstance(event, inbound.Press):
+		print(f"[{event.service.capitalize()} button]:", event.user, "chat", event.chat_id, event.data, file=sys.stderr)
+		target = worker.process_callback
+		args = (event.service, event.press_id, event.chat_id, event.message_id, event.data)
+	else:
+		if config.prefs.debug:
+			print(f"[{event.service.capitalize()}]:", event.user, event.text, file=sys.stderr)
+		target = worker.process_request
+		args = (event.service, event.chat_id, event.user, event.text, event.bot_name, event.user_name)
+	threading.Thread(target=target, args=args).start()
 
 def git_version(checkout=None):
 	"""Short hash of the checked-out commit, read straight from the .git directory: no git binary to run, and no "dubious
@@ -263,7 +193,7 @@ def flush_cache(*args):
 	print(f'Opening socket on http://{config.env.ip}:{config.env.port}', file=sys.stderr)
 	if config.env.telegramBotToken and config.env.telegramOutgoingWebhook:
 		try:
-			botName = telegram.register() # here and only here: the job processes import lib.telegram too
+			inbound.TELEGRAM.bot_name = telegram.register() # here only: the job processes import lib.telegram too
 		except Exception as e:
 			print('ERROR', delivery.error_line(e, 'telegram webhook'), file=sys.stderr)
 			sys.exit(1)
diff --git a/lib/inbound.py b/lib/inbound.py
new file mode 100644
index 00000000..39014825
--- /dev/null
+++ b/lib/inbound.py
@@ -0,0 +1,200 @@
+"""What the networks post to the bot: one class per network that sends updates, which knows the path it posts to, how
+it proves itself and how to read an update. An update becomes a Message to answer, a Press of a button, or a Response
+the network gets straight back (a challenge, or nothing to do)."""
+import dataclasses
+import hmac
+import html
+import json
+import re
+import sys
+from urllib.parse import urlparse
+
+from lib import config
+
+
+@dataclasses.dataclass
+class Message:
+	"""A chat message to answer (worker.process_request)."""
+	service: str
+	chat_id: str
+	user: str # the sender's handle: whose watchlist add or plan it is
+	text: str
+	bot_name: str # the bot's @name, which can start a command
+	user_name: str # what to call them
+
+
+@dataclasses.dataclass
+class Press:
+	"""A button pressed under one of the bot's messages (worker.process_callback)."""
+	service: str
+	press_id: str # the network's id for the press, to answer it
+	chat_id: str
+	message_id: str
+	data: str
+	user: str # who pressed it, for the log
+
+
+@dataclasses.dataclass
+class Response:
+	"""What the network gets back at once: a status and a body."""
+	status: str = '200 OK'
+	body: bytes = b''
+
+
+OK = Response()
+UNSUPPORTED = Response(body=b'Unsupported') # an update of a kind the bot takes no part in
+UNHANDLED = Response(body=b'<h1>Unhandled</h1>') # a kind it might, which this code does not know
+BAD_REQUEST = Response('400 Bad Request', b'<h1>Bad Request</h1>')
+FORBIDDEN = Response('403 Forbidden', b'<h1>Unauthorized</h1>')
+NOT_FOUND = Response('404 Not Found', b'<h1>404</h1>')
+
+
+def same_secret(given, expected):
+	"""Whether a secret sent with a request is the one configured, compared in constant time. An empty configured secret
+	proves nothing, so it matches nothing."""
+	return bool(expected) and given is not None and hmac.compare_digest(given.encode(), expected.encode())
+
+
+class Webhook:
+	"""A network that posts updates to the bot's web server. Subclasses say where it posts, how it proves itself and
+	what an update says."""
+	service = ''
+
+	def path(self):
+		"""The URL path the network posts to, or '' when finbot.env sets none."""
+		return ''
+
+	def refuse_unread(self, environ):
+		"""Why to turn the request away before its body is read (a secret header missing or wrong), else None."""
+		return None
+
+	def decode(self, body):
+		"""The update in the request body, a dict. Raises ValueError when the body is not one."""
+		update = json.loads(body)
+		if not isinstance(update, dict):
+			raise ValueError('not a JSON object')
+		return update
+
+	def refuse(self, update):
+		"""Why to turn the request away once read (a secret in the body missing or wrong), else None."""
+		return None
+
+	def update_id(self, update):
+		"""The update's id, the same when the network sends it again (it does when it got no answer in time), or None
+		when it gives none."""
+		return None
+
+	def parse(self, update):
+		"""What the update says: a Message, a Press, or a Response to give straight back."""
+		raise NotImplementedError
+
+
+class Telegram(Webhook):
+	"""Telegram's updates, each sent with the secret header that setWebhook gave it."""
+	service = 'telegram'
+	bot_name = None # the bot's @username, from telegram.register() as bot.py starts
+	IGNORED = {'sticker', 'animation', 'voice', 'video', 'video_note', 'audio', 'document', 'contact', 'location',
+		'venue', 'poll', 'dice'} # nothing to answer, nothing worth logging
+
+	def path(self):
+		"""The path of telegramOutgoingWebhook."""
+		return urlparse(config.env.telegramOutgoingWebhook).path if config.env.telegramOutgoingWebhook else ''
+
+	def refuse_unread(self, environ):
+		"""The header must carry telegramOutgoingToken."""
+		secret = environ.get('HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN')
+		if not same_secret(secret, config.env.telegramOutgoingToken):
+			return f"Telegram authorisation header {'not present' if secret is None else 'is present but incorrect'}"
+		return None
+
+	def update_id(self, update):
+		"""Telegram's update_id."""
+		return update.get('update_id')
+
+	def parse(self, update):
+		"""A message (its text, or a photo's caption) or a button press. A first name is escaped as it arrives:
+		replies are Telegram HTML, and a name is text, not markup."""
+		if 'callback_query' in update: # chart period and job buttons
+			press = update['callback_query']
+			message = press.get('message')
+			if not message or not press.get('data'):
+				return UNSUPPORTED
+			sender = press.get('from', {})
+			who = '@' + sender['username'] if 'username' in sender else sender.get('first_name', 'unknown')
+			return Press(self.service, press['id'], str(message['chat']['id']), str(message['message_id']),
+				press['data'], who)
+		if 'message' not in update:
+			return UNSUPPORTED
+		message = update['message']
+		sender = message['from']
+		chat_id = str(message['chat']['id'])
+		first_name = html.escape(sender['first_name'], quote=False)
+		if 'username' in sender:
+			user = '@' + sender['username']
+			user_name = first_name or user # the handle when the first name is blank
+		else:
+			user = user_name = '@' + first_name
+		if 'text' in message:
+			text = message['text']
+		elif 'photo' in message:
+			text = message.get('caption', '')
+		elif self.IGNORED & message.keys():
+			return OK
+		else:
+			print("[telegram]: unhandled: 'message' without 'text/photo'", file=sys.stderr)
+			return UNHANDLED
+		return Message(self.service, chat_id, user, text, self.bot_name, user_name)
+
+
+class Slack(Webhook):
+	"""Slack's Events API, each event carrying the app's verification token."""
+	service = 'slack'
+	BY_PEOPLE = (None, 'thread_broadcast') # message subtypes a person writes; the rest are bot posts, edits, joins...
+
+	def path(self):
+		"""The path of slackOutgoingWebhook."""
+		return urlparse(config.env.slackOutgoingWebhook).path if config.env.slackOutgoingWebhook else ''
+
+	def refuse(self, update):
+		"""The body must carry slackOutgoingToken."""
+		if 'token' not in update:
+			return "Slack authorisation field not present"
+		if not same_secret(str(update['token']), config.env.slackOutgoingToken):
+			return "Slack auth incorrect"
+		return None
+
+	def update_id(self, update):
+		"""Slack's event_id."""
+		return update.get('event_id')
+
+	def parse(self, update):
+		"""A message a person wrote, or the challenge that checks the URL. Links and mentions lose their markup:
+		<http://dub.ax|dub.ax> becomes dub.ax and <@QWERTY> @QWERTY."""
+		kind = update.get('type')
+		if kind == 'url_verification':
+			response = json.dumps({'challenge': update['challenge']})
+			print("replying with", response, file=sys.stderr)
+			return Response(body=response.encode())
+		if kind != 'event_callback':
+			print(f"[slack]: unhandled 'type' {kind!r}", file=sys.stderr)
+			return UNHANDLED
+		event = update['event']
+		bot_user = update['authorizations'][0]['user_id'] # QWERTY
+		if event.get('bot_id') or event.get('subtype') not in self.BY_PEOPLE or not event.get('user') or \
+				event['user'] == bot_user:
+			return OK # not by a person: answering a bot (this one's replies and webhook reports included) could loop
+		if event['type'] not in ('message', 'app_mention') or 'text' not in event:
+			print("[slack]: unhandled event callback type", event['type'], file=sys.stderr)
+			return UNHANDLED
+		text = re.sub(r'<http://.*\|([\w\.]+)>', r'\g<1>', event['text'])
+		text = re.sub(r'<(@[\w\.]+)>', r'\g<1>', text)
+		user = '<@' + event['user'] + '>' # a mention: Slack shows the name
+		return Message(self.service, event['channel'], user, text, '@' + bot_user, user)
+
+
+TELEGRAM, SLACK = Telegram(), Slack()
+
+
+def route(path):
+	"""The network that posts to this path, or None."""
+	return next((source for source in (TELEGRAM, SLACK) if source.path() and path == source.path()), None)
diff --git a/tests/common.py b/tests/common.py
index 8fb01856..40aec559 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -10,8 +10,8 @@
 os.environ.update(telegramOutgoingWebhook='https://example.com/telegram', telegramOutgoingToken='tok', telegramChatID='1')
 
 import bot # first, and once: it monkey-patches with gevent, which must happen before requests imports ssl and never mid-test
-from lib import buttons, cache, charts, config, delivery, dialects, formats, links, markets, render, sharesight
-from lib import shortman, state, telegram, transports, watchlist, words, worker, yahoo
+from lib import buttons, cache, charts, config, delivery, dialects, formats, inbound, links, markets, render
+from lib import sharesight, shortman, state, telegram, transports, watchlist, words, worker, yahoo
 from lib.reports import about, cal, history, holdings, milestone, performance, price, profile, rankings, rating, reminder, shorts, trades
 
 REAL = {(m.__name__, n): getattr(m, n) for m, n in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (yahoo, 'price_history'),
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 4ff2af5e..e4c0d3d0 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -226,6 +226,16 @@ def test_telegram_requires_the_secret_header(self):
 		self.assertIn(b'Unauthorized', self.request(body=message, headers={'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'wrong'})[0])
 		self.assertEqual(self.sent, []) # nothing was processed
 
+	def test_an_empty_secret_proves_nothing(self):
+		"""With no secret configured, a request carrying an empty one is refused, on Telegram and on Slack."""
+		self.env(telegramOutgoingToken='', slackOutgoingWebhook='https://example.com/slack', slackOutgoingToken='')
+		message = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.help'}}
+		empty = {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': ''}
+		self.assertEqual(self.request(body=message, headers=empty)[1], '403 Forbidden')
+		self.assertEqual(self.request('/slack', {'token': '', 'type': 'url_verification', 'challenge': 'c'})[1],
+			'403 Forbidden')
+		self.assertEqual((self.sent, self.errors), ([], []))
+
 	def test_unknown_path_is_404(self):
 		"""An unknown path is 404."""
 		self.assertEqual(self.request(path='/nope')[1], '404 Not Found')
@@ -258,7 +268,7 @@ def test_a_failure_before_authentication_is_only_logged(self):
 		def broken(*a, **k):
 			"""Fail while routing."""
 			raise RuntimeError('boom')
-		self.patch(bot, 'urlparse', broken) # fails while routing, before the secret is even looked at
+		self.patch(inbound, 'urlparse', broken) # fails while routing, before the secret is even looked at
 		env = {'PATH_INFO': '/telegram', 'wsgi.input': io.BytesIO(b'{}'), **self.SECRET}
 		status = []
 		with contextlib.redirect_stderr(io.StringIO()) as err:
@@ -354,9 +364,9 @@ def test_a_retried_event_runs_once(self):
 
 	def test_slack_does_not_change_the_telegram_bot_name(self):
 		"""A Slack event leaves the Telegram bot's name alone."""
-		self.patch(bot, 'botName', '@finbot_tg')
+		self.patch(inbound.TELEGRAM, 'bot_name', '@finbot_tg')
 		self.event(user='U1', text='.price')
-		self.assertEqual(bot.botName, '@finbot_tg')
+		self.assertEqual(inbound.TELEGRAM.bot_name, '@finbot_tg')
 
 
 class UserTextIsEscaped(FinbotCase):

From 5f9edf06dce9daef0d352a476fc9ad15b96c683d Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 01:18:57 +0000
Subject: [PATCH 21/81] Slack signing secrets

The new startup key slackSigningSecret (a secret) turns on signature
checks for Slack requests. With it set, every request must carry:
- X-Slack-Signature, the HMAC-SHA256 under the signing secret of
  v0:<timestamp>:<body>;
- a timestamp within five minutes of now, so a recorded request cannot
  be replayed.
The verification token is then neither needed nor enough. Without the
signing secret, slackOutgoingToken is checked as before; Slack has
deprecated it because the token travels in every request.

New startup warnings:
- an inbound Slack webhook with neither secret;
- telegramOutgoingWebhook without telegramOutgoingToken.
Either way every request would be refused.

The README documents the key in the finbot.env table and in the Slack
setup steps. A test checks the code against Slack's documented example
signature.

Also fixed: test_config.py imported lib.config before tests/common.py
had set the environment. Run before test_edges, it failed 10 Telegram
tests (on the previous commit as well). It now imports the fixture
first, and every test module passes when run alone.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 README.md            |  6 ++++--
 bot.py               |  5 +++--
 defaults/finbot.env  |  1 +
 lib/config.py        | 14 ++++++++++++--
 lib/inbound.py       | 34 ++++++++++++++++++++++++++++-----
 tests/test_config.py | 26 ++++++++++++++++++-------
 tests/test_edges.py  | 45 +++++++++++++++++++++++++++++++++++++++++++-
 7 files changed, 112 insertions(+), 19 deletions(-)

diff --git a/README.md b/README.md
index 08764927..cbc9ffad 100644
--- a/README.md
+++ b/README.md
@@ -470,7 +470,8 @@ The startup keys, read by your launcher. A change needs a restart. Blank means u
 | `telegramOutgoingWebhook` | | Telegram webhook path the bot accepts (interactive mode) |
 | `telegramOutgoingToken` | | Secret token Telegram must present (interactive mode) |
 | `slackOutgoingWebhook` | | Slack webhook path the bot accepts (interactive mode) |
-| `slackOutgoingToken` | | Token Slack must present (interactive mode) |
+| `slackOutgoingToken` | | Token Slack must present (interactive mode): the older way, which Slack has deprecated; not needed with `slackSigningSecret` |
+| `slackSigningSecret` | | Secret Slack signs each request with (interactive mode) |
 
 ### `preferences.conf` items
 
@@ -712,7 +713,8 @@ With these options set, your bot will auto-subscribe your URL to events the bot
 
 #### Slack
 Visit https://api.slack.com/apps/ to create a new Slack app from scratch (if you already created one for a finbot webhook, you can reuse that app). Charts in replies are uploaded into the channel with Slack's file upload API (`files.getUploadURLExternal`, then `files.completeUploadExternal`, which replaced the retired `files.upload`). Scheduled reports reach Slack through the incoming webhook, which takes text only, so they come without charts. If an upload fails (a missing scope, say), the reply goes out as text and the log says why.
-* From _Basic Information > Verification Token_, copy _Verification token_ into `finbot.env` as `slackOutgoingToken`. This allows Slack to authenticate with finbot.
+* From _Basic Information > App Credentials_, copy _Signing Secret_ into `finbot.env` as `slackSigningSecret`. finbot checks Slack's signature on every request with it, and turns away a request signed more than five minutes earlier (a replay, or a server clock that is wrong).
+  * The older _Verification Token_, set as `slackOutgoingToken`, still works without a signing secret, but Slack has deprecated it: the token travels in every request, where the signing secret never does.
 * In `finbot.env`, put your web server URL (e.g. https://www.example.com:8443/slack) into `slackOutgoingWebhook`
 * Save `finbot.env` and restart `bot.py`
 * Under _Event Subscriptions_:
diff --git a/bot.py b/bot.py
index bc545c59..4bad910d 100755
--- a/bot.py
+++ b/bot.py
@@ -83,12 +83,13 @@ def print_headers():
 		print_headers()
 		print("Fatal:", why, file=sys.stderr)
 		return respond(inbound.FORBIDDEN)
+	body = environ['wsgi.input'].read()
 	try:
-		update = source.decode(environ['wsgi.input'].read())
+		update = source.decode(body)
 	except ValueError:
 		print(f"[{source.service}]: request body is not a JSON object", file=sys.stderr)
 		return respond(inbound.BAD_REQUEST)
-	if why := source.refuse(update):
+	if why := source.refuse(environ, body, update):
 		shown = {k: '<redacted>' if k == 'token' else v for k, v in update.items()} # Slack's verification token
 		print("warning:", why, file=sys.stderr)
 		print(f"inbound {path} ", json.dumps(shown, indent=4, default=str), file=sys.stderr)
diff --git a/defaults/finbot.env b/defaults/finbot.env
index 489bcf38..96dc58bf 100644
--- a/defaults/finbot.env
+++ b/defaults/finbot.env
@@ -38,3 +38,4 @@ telegramOutgoingWebhook=
 telegramOutgoingToken=
 slackOutgoingWebhook=
 slackOutgoingToken=
+slackSigningSecret=
diff --git a/lib/config.py b/lib/config.py
index 1d3ba215..f188c2d9 100644
--- a/lib/config.py
+++ b/lib/config.py
@@ -16,9 +16,11 @@
 
 STARTUP_KEYS = ('DATA_DIR', 'ip', 'port', 'telegramChatID', 'scheduler', 'scheduler_catchup',
 	'sharesight_code', 'sharesight_client_id', 'sharesight_client_secret', 'slack_webhook', 'discord_webhook', 'slackBotToken',
-	'slackOutgoingToken', 'slackOutgoingWebhook', 'telegramBotToken', 'telegramOutgoingToken', 'telegramOutgoingWebhook')
+	'slackOutgoingToken', 'slackOutgoingWebhook', 'slackSigningSecret', 'telegramBotToken', 'telegramOutgoingToken',
+	'telegramOutgoingWebhook')
 SECRET_KEYS = ('sharesight_code', 'sharesight_client_id', 'sharesight_client_secret', 'slack_webhook', 'discord_webhook', 'slackBotToken',
-	'slackOutgoingToken', 'slackOutgoingWebhook', 'telegramBotToken', 'telegramOutgoingToken', 'telegramOutgoingWebhook') # never printed
+	'slackOutgoingToken', 'slackOutgoingWebhook', 'slackSigningSecret', 'telegramBotToken', 'telegramOutgoingToken',
+	'telegramOutgoingWebhook') # never printed
 PREFERENCES_FILE = 'preferences.conf'
 PASSED = 'FINBOT_PREFERENCES_PASSED' # set by the bot for its job processes: the preferences arrive in the environment, do not read the file
 OLD_ENV = REPO_DIR / '.env' # the single config file of old versions, which python-dotenv loaded; nothing reads it now
@@ -176,6 +178,7 @@ class Env:
 	slackBotToken: str
 	slackOutgoingWebhook: str
 	slackOutgoingToken: str
+	slackSigningSecret: str
 	sharesight_auth: dict
 	webhooks: dict # service -> outbound URL: Telegram's Bot API base, the Slack and Discord incoming webhooks
 
@@ -210,6 +213,7 @@ def startup_values(getenv):
 		slackBotToken=r.str('slackBotToken', ''),
 		slackOutgoingWebhook=r.str('slackOutgoingWebhook', ''),
 		slackOutgoingToken=r.str('slackOutgoingToken', ''),
+		slackSigningSecret=r.str('slackSigningSecret', ''),
 		sharesight_auth={
 			"grant_type": 'client_credentials',
 			"code": getenv('sharesight_code'),
@@ -409,4 +413,10 @@ def startup_warnings():
 	warnings = []
 	if env.slackOutgoingWebhook and not env.slackBotToken:
 		warnings.append('slackOutgoingWebhook is set but slackBotToken is not: Slack commands will get no reply')
+	if env.slackOutgoingWebhook and not (env.slackSigningSecret or env.slackOutgoingToken):
+		warnings.append('slackOutgoingWebhook is set but neither slackSigningSecret nor slackOutgoingToken is:'
+			' every Slack request will be refused')
+	if env.telegramOutgoingWebhook and not env.telegramOutgoingToken:
+		warnings.append('telegramOutgoingWebhook is set but telegramOutgoingToken is not: every Telegram update will be'
+			' refused')
 	return warnings
diff --git a/lib/inbound.py b/lib/inbound.py
index 39014825..bcbb6373 100644
--- a/lib/inbound.py
+++ b/lib/inbound.py
@@ -2,11 +2,13 @@
 it proves itself and how to read an update. An update becomes a Message to answer, a Press of a button, or a Response
 the network gets straight back (a challenge, or nothing to do)."""
 import dataclasses
+import hashlib
 import hmac
 import html
 import json
 import re
 import sys
+import time
 from urllib.parse import urlparse
 
 from lib import config
@@ -49,6 +51,9 @@ class Response:
 NOT_FOUND = Response('404 Not Found', b'<h1>404</h1>')
 
 
+SIGNATURE_AGE = 300 # seconds: Slack's limit; an older signed request may be a replay
+
+
 def same_secret(given, expected):
 	"""Whether a secret sent with a request is the one configured, compared in constant time. An empty configured secret
 	proves nothing, so it matches nothing."""
@@ -75,8 +80,9 @@ def decode(self, body):
 			raise ValueError('not a JSON object')
 		return update
 
-	def refuse(self, update):
-		"""Why to turn the request away once read (a secret in the body missing or wrong), else None."""
+	def refuse(self, environ, body, update):
+		"""Why to turn the request away once read (a signature of the body, or a secret in it, missing or wrong), else
+		None."""
 		return None
 
 	def update_id(self, update):
@@ -147,7 +153,8 @@ def parse(self, update):
 
 
 class Slack(Webhook):
-	"""Slack's Events API, each event carrying the app's verification token."""
+	"""Slack's Events API: each request signed with the app's signing secret, and carrying its verification
+	token."""
 	service = 'slack'
 	BY_PEOPLE = (None, 'thread_broadcast') # message subtypes a person writes; the rest are bot posts, edits, joins...
 
@@ -155,14 +162,31 @@ def path(self):
 		"""The path of slackOutgoingWebhook."""
 		return urlparse(config.env.slackOutgoingWebhook).path if config.env.slackOutgoingWebhook else ''
 
-	def refuse(self, update):
-		"""The body must carry slackOutgoingToken."""
+	def refuse(self, environ, body, update):
+		"""With slackSigningSecret, the request must carry Slack's signature of its body, made in the last five minutes;
+		without it, the body must carry slackOutgoingToken (the older way, which Slack has deprecated)."""
+		if config.env.slackSigningSecret:
+			return self.signature_problem(environ, body, config.env.slackSigningSecret)
 		if 'token' not in update:
 			return "Slack authorisation field not present"
 		if not same_secret(str(update['token']), config.env.slackOutgoingToken):
 			return "Slack auth incorrect"
 		return None
 
+	@staticmethod
+	def signature_problem(environ, body, secret):
+		"""Why the request's signature fails, or None: X-Slack-Signature must be v0= and the HMAC-SHA256, under the
+		signing secret, of v0:<X-Slack-Request-Timestamp>:<the body>, and the timestamp within five minutes of now."""
+		timestamp = environ.get('HTTP_X_SLACK_REQUEST_TIMESTAMP', '')
+		signature = environ.get('HTTP_X_SLACK_SIGNATURE')
+		if signature is None or not (timestamp.isascii() and timestamp.isdigit()):
+			return "Slack signature not present"
+		if abs(time.time() - int(timestamp)) > SIGNATURE_AGE:
+			return "Slack signature too old (a replay, or a clock is wrong)"
+		signed = b'v0:' + timestamp.encode() + b':' + body
+		expected = 'v0=' + hmac.new(secret.encode(), signed, hashlib.sha256).hexdigest()
+		return None if same_secret(signature, expected) else "Slack signature incorrect"
+
 	def update_id(self, update):
 		"""Slack's event_id."""
 		return update.get('event_id')
diff --git a/tests/test_config.py b/tests/test_config.py
index 778dce67..746a90a5 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -2,6 +2,7 @@
 import ast, contextlib, dataclasses, io, os, re, subprocess, sys, tempfile, unittest
 from pathlib import Path
 from unittest import mock
+from tests import common # first: it sets the environment that lib.config reads as it loads
 from lib import config
 from util import check_config, split_env
 import bot
@@ -31,7 +32,7 @@ class Tables(unittest.TestCase):
 	def test_the_two_key_sets_are_disjoint_and_cover_everything_config_reads(self):
 		"""Startup and preference keys do not overlap, and between them name every setting config.py reads."""
 		self.assertEqual(set(config.STARTUP_KEYS) & set(config.PREFERENCE_KEYS), set())
-		self.assertEqual((len(config.STARTUP_KEYS), len(config.PREFERENCE_KEYS)), (17, 26))
+		self.assertEqual((len(config.STARTUP_KEYS), len(config.PREFERENCE_KEYS)), (18, 26))
 		source = (REPO / 'lib' / 'config.py').read_text()
 		read = set(re.findall(r"""(?:\br\.\w+|getenv)\(\s*['"]([A-Za-z_]+)['"]""", source))
 		read -= {'STATE_DIRECTORY'} # systemd's own variable, a source for DATA_DIR
@@ -40,7 +41,7 @@ def test_the_two_key_sets_are_disjoint_and_cover_everything_config_reads(self):
 	def test_secrets_are_startup_keys(self):
 		"""Every secret is a startup key: preferences.conf holds none."""
 		self.assertTrue(set(config.SECRET_KEYS) <= set(config.STARTUP_KEYS))
-		self.assertEqual(len(config.SECRET_KEYS), 11)
+		self.assertEqual(len(config.SECRET_KEYS), 12)
 
 	def test_the_preference_defaults_are_the_old_defaults(self):
 		"""Unset preferences take the defaults finbot always had."""
@@ -382,17 +383,28 @@ def test_a_discord_webhook_uses_the_native_endpoint(self):
 			values = config.startup_values(lambda name, default=None: url if name == 'discord_webhook' else default)
 			self.assertEqual(values.webhooks['discord'], 'https://discord.com/api/webhooks/1/abc')
 
-	def test_slack_commands_without_a_bot_token_are_warned_about(self):
-		"""Slack commands without slackBotToken get a startup warning, since they would get no reply."""
-		for token, warnings in (('', ['slackOutgoingWebhook is set but slackBotToken is not: Slack commands will get no reply']), ('xoxb-1', [])):
-			with mock.patch.object(config, 'env', dataclasses.replace(config.env, slackOutgoingWebhook='https://example.com/slack', slackBotToken=token)):
+	def test_settings_that_leave_part_of_the_bot_unable_to_work_are_warned_about(self):
+		"""A startup warning for Slack commands without slackBotToken (they would get no reply), and for an inbound
+		webhook without its secret (every request would be refused)."""
+		hooks = dict(slackOutgoingWebhook='https://example.com/slack', slackOutgoingToken='stok', slackBotToken='xoxb-1',
+			telegramOutgoingWebhook='https://example.com/telegram', telegramOutgoingToken='tok')
+		no_reply = 'slackOutgoingWebhook is set but slackBotToken is not: Slack commands will get no reply'
+		slack_refused = ('slackOutgoingWebhook is set but neither slackSigningSecret nor slackOutgoingToken is:'
+			' every Slack request will be refused')
+		telegram_refused = ('telegramOutgoingWebhook is set but telegramOutgoingToken is not: every Telegram update'
+			' will be refused')
+		for changes, warnings in (({}, []), ({'slackBotToken': ''}, [no_reply]),
+				({'slackOutgoingToken': ''}, [slack_refused]), ({'slackOutgoingToken': '', 'slackSigningSecret': 's'}, []),
+				({'telegramOutgoingToken': ''}, [telegram_refused]), ({'telegramOutgoingWebhook': ''}, [])):
+			with self.subTest(changes=changes), \
+					mock.patch.object(config, 'env', dataclasses.replace(config.env, **{**hooks, **changes})):
 				self.assertEqual(config.startup_warnings(), warnings)
 
 	def test_startup_report_is_a_short_count_never_names_or_values(self):
 		"""The startup report counts the startup keys set, and never names them or shows their values."""
 		with mock.patch.dict(os.environ, {'slack_webhook': 'https://secret.example/abc', 'port': '5000'}, clear=True):
 			lines = config.startup_report()
-		self.assertEqual(lines[0], 'startup environment provides 2 of 17 startup keys (values hidden)')
+		self.assertEqual(lines[0], 'startup environment provides 2 of 18 startup keys (values hidden)')
 		self.assertLessEqual(len('config: ' + lines[0]), 90) # the path in the second line depends on where DATA_DIR is
 		self.assertNotIn('secret.example', ''.join(lines))
 		self.assertNotIn('slack_webhook', ''.join(lines))
diff --git a/tests/test_edges.py b/tests/test_edges.py
index e4c0d3d0..ed6055cc 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -1,5 +1,5 @@
 """Failure and no-data scenarios: empty or broken upstream data, bad requests, rejected callers."""
-import contextlib, io, json, os, signal, subprocess, sys, tempfile, threading, time, unittest
+import contextlib, hashlib, hmac, io, json, os, signal, subprocess, sys, tempfile, threading, time, unittest
 from pathlib import Path
 import gevent
 from tests.common import *
@@ -369,6 +369,49 @@ def test_slack_does_not_change_the_telegram_bot_name(self):
 		self.assertEqual(inbound.TELEGRAM.bot_name, '@finbot_tg')
 
 
+class SlackSigning(FinbotCase):
+	"""With slackSigningSecret, a Slack request must carry Slack's signature of its body, made in the last five
+	minutes; the verification token is then neither needed nor enough."""
+
+	def setUp(self):
+		"""Slack commands with a signing secret, and the worker replaced by a list of what it would have run."""
+		super().setUp()
+		self.env(slackOutgoingWebhook='https://example.com/slack', slackOutgoingToken='stok', slackSigningSecret='sss')
+		self.requests = []
+		self.patch(worker, 'process_request', lambda *a: self.requests.append(a))
+
+	def post(self, secret='sss', age=0, token='stok', headers=None, tamper=False):
+		"""Post a Slack message event signed with `secret` `age` seconds ago (or with these headers instead) and
+		return the status; tamper changes the body after signing."""
+		body = json.dumps({'token': token, 'type': 'event_callback', 'authorizations': [{'user_id': 'UBOT'}],
+			'event': {'type': 'message', 'channel': 'C1', 'user': 'U1', 'text': '.price'}}).encode()
+		timestamp = str(int(time.time()) - age)
+		mac = hmac.new(secret.encode(), b'v0:' + timestamp.encode() + b':' + body, hashlib.sha256).hexdigest()
+		if headers is None:
+			headers = {'HTTP_X_SLACK_REQUEST_TIMESTAMP': timestamp, 'HTTP_X_SLACK_SIGNATURE': 'v0=' + mac}
+		env = {'PATH_INFO': '/slack', 'wsgi.input': io.BytesIO(body.replace(b'.price', b'.trades') if tamper else body),
+			**headers}
+		status = []
+		with contextlib.redirect_stderr(io.StringIO()):
+			bot.main(env, lambda s, *a, **k: status.append(s))
+		gevent.sleep(0.05) # the worker runs in a (gevent) thread
+		return status[-1]
+
+	def test_a_signed_request_is_answered_and_the_token_is_not_needed(self):
+		"""A request signed with the secret is answered, whatever token it carries."""
+		self.assertEqual([self.post(), self.post(token='wrong')], ['200 OK'] * 2)
+		self.assertEqual(len(self.requests), 2)
+
+	def test_unsigned_wrongly_signed_old_or_altered_requests_are_refused(self):
+		"""No signature (the right token is not enough), a wrong secret, a timestamp more than five minutes away or
+		not a number, and a body changed after signing are all refused."""
+		for case in (dict(headers={}), dict(secret='wrong'), dict(age=400), dict(age=-400), dict(tamper=True),
+				dict(headers={'HTTP_X_SLACK_REQUEST_TIMESTAMP': 'soon', 'HTTP_X_SLACK_SIGNATURE': 'v0=00'})):
+			with self.subTest(**case):
+				self.assertEqual(self.post(**case), '403 Forbidden')
+		self.assertEqual((self.requests, self.errors), ([], []))
+
+
 class UserTextIsEscaped(FinbotCase):
 	"""Replies go out as Telegram HTML: text a user wrote is shown, never read as markup. One stray < in a .plan entry made Telegram
 	refuse that reply and every .plan reply after it."""

From a4328ce4985b6abcb82bd85c673f3dbd762de937 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 01:29:47 +0000
Subject: [PATCH 22/81] One-line docstrings where they fit

77 docstrings that spilled a few words onto a second line are reworded
to fit on one, within 120 columns with tabs counting 4. Their meaning
is kept.
- 74 of them overflowed by 30 columns or less once joined.
- 3 had first lines already past 120 columns and only two to four words
  on the second line.

The other 111 two-line docstrings need their second line and stay as
they are.

CLAUDE.md's convention now asks for one line where it fits, and states
the line limit as 120 columns with tabs counting 4.

Also included: a table of cases in test_config's startup-warning test,
which wraps two lines of the previous commit that passed 120 columns.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md               |  5 +++--
 lib/config.py           |  6 ++----
 lib/dialects.py         |  3 +--
 lib/inbound.py          | 12 ++++--------
 lib/links.py            |  6 ++----
 lib/reports/cal.py      |  3 +--
 lib/reports/holdings.py |  3 +--
 lib/shortman.py         |  3 +--
 lib/telegram.py         |  3 +--
 lib/transports.py       | 21 +++++++--------------
 lib/worker.py           |  6 ++----
 tests/common.py         |  6 ++----
 tests/test_charts.py    | 24 ++++++++----------------
 tests/test_config.py    | 35 ++++++++++++++---------------------
 tests/test_datadir.py   |  6 ++----
 tests/test_edges.py     | 18 ++++++------------
 tests/test_finbot.py    |  6 ++----
 tests/test_jobs.py      | 33 +++++++++++----------------------
 tests/test_providers.py | 27 +++++++++------------------
 tests/test_reports.py   |  3 +--
 tests/test_util.py      | 15 +++++----------
 util/sws_gen.py         |  3 +--
 22 files changed, 86 insertions(+), 161 deletions(-)

diff --git a/CLAUDE.md b/CLAUDE.md
index 8a7b9504..1769e830 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -66,8 +66,9 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - A report is a list of lines. `service` is a dialect (`lib/dialects.py`), which equals the network's name: markup with
   `service.bold(text)`, `.italic`, `.strike`, `.link(url, text)`, user-written text through `service.escape(text)`, stock links
   through `links.finance_link(symbol, exchange, service)`. Tests pass `TELEGRAM`, `SLACK`, `DISCORD`.
-- Every function and method has a docstring, tests included (a test checks). Tabs everywhere (`.editorconfig`). New and
-  changed lines stay under 120 characters (an edit quotes whole lines).
+- Every function and method has a docstring, tests included (a test checks): one line where it fits, reworded rather
+  than spilling a few words onto a second. Tabs everywhere (`.editorconfig`). New and changed lines stay within 120
+  columns, tabs counting 4 (an edit quotes whole lines).
 - Log with `print(..., file=sys.stderr)`. systemd discards stdout (`StandardOutput=null`), so the bot and its reports never
   print there (a test checks); only the util/ tools, run by hand, do.
 - A chart renderer in `lib/render.py` gets `@_drawn` (the bot draws in one native thread); it must not call another `@_drawn`
diff --git a/lib/config.py b/lib/config.py
index f188c2d9..eec75c71 100644
--- a/lib/config.py
+++ b/lib/config.py
@@ -130,8 +130,7 @@ def str(self, name, default=None):
 		return self.value(name, default)
 
 	def bool(self, name, default):
-		"""A true/false setting: true/false, yes/no, on/off, 1/0 or t/f, in any case. Anything else is an error, and
-		gives default."""
+		"""A boolean: true/false, yes/no, on/off, 1/0 or t/f in any case; anything else is an error, giving default."""
 		value = self.value(name, default)
 		word = str(value).strip().lower()
 		if word in TRUE_WORDS or word in FALSE_WORDS:
@@ -339,8 +338,7 @@ def _fail(e):
 os.environ.setdefault('MPLCONFIGDIR', os.path.abspath(env.cache_dir + '/matplotlib'))
 
 def data_file(name):
-	"""Where a data file is read from: the user's copy in the data dir when there is one, else the default shipped in
-	defaults/."""
+	"""Where a data file is read from: the user's copy in the data dir if any, else the default shipped in defaults/."""
 	user = env.data_dir + '/' + name
 	return user if os.path.isfile(user) else str(REPO_DIR / 'defaults' / name)
 
diff --git a/lib/dialects.py b/lib/dialects.py
index ff4c05fe..f7191905 100644
--- a/lib/dialects.py
+++ b/lib/dialects.py
@@ -78,8 +78,7 @@ class Discord(Dialect):
 	text_limit = caption_limit = 2000 # markup included
 
 	def _link(self, url, text):
-		"""A masked link: the angle brackets keep its preview away; brackets in the text and parentheses in the url would end it
-		early."""
+		"""A masked link: <url> stops a preview; brackets in the text and parentheses in the url would end it early."""
 		return '[' + text.replace('[', '\\[').replace(']', '\\]') + '](<' + url.replace('(', '%28').replace(')', '%29') + '>)'
 
 	def escape(self, text):
diff --git a/lib/inbound.py b/lib/inbound.py
index bcbb6373..54232b87 100644
--- a/lib/inbound.py
+++ b/lib/inbound.py
@@ -61,8 +61,7 @@ def same_secret(given, expected):
 
 
 class Webhook:
-	"""A network that posts updates to the bot's web server. Subclasses say where it posts, how it proves itself and
-	what an update says."""
+	"""A network posting updates to the bot; subclasses give its path, how it proves itself and what an update says."""
 	service = ''
 
 	def path(self):
@@ -81,13 +80,11 @@ def decode(self, body):
 		return update
 
 	def refuse(self, environ, body, update):
-		"""Why to turn the request away once read (a signature of the body, or a secret in it, missing or wrong), else
-		None."""
+		"""Why to refuse the request once read (the body's signature or a secret in it missing or wrong), else None."""
 		return None
 
 	def update_id(self, update):
-		"""The update's id, the same when the network sends it again (it does when it got no answer in time), or None
-		when it gives none."""
+		"""The update's id, the same when the network resends it (unanswered in time), or None if it gives none."""
 		return None
 
 	def parse(self, update):
@@ -153,8 +150,7 @@ def parse(self, update):
 
 
 class Slack(Webhook):
-	"""Slack's Events API: each request signed with the app's signing secret, and carrying its verification
-	token."""
+	"""Slack's Events API: each request signed with the app's signing secret, and carrying its verification token."""
 	service = 'slack'
 	BY_PEOPLE = (None, 'thread_broadcast') # message subtypes a person writes; the rest are bot posts, edits, joins...
 
diff --git a/lib/links.py b/lib/links.py
index 8bcd091a..30a635af 100644
--- a/lib/links.py
+++ b/lib/links.py
@@ -3,15 +3,13 @@
 from lib import markets
 
 def yahoo_link(ticker, service, brief=True, text=None):
-	"""A link to the ticker's Yahoo Finance page, reading `text`, else the symbol without its market (brief) or the
-	whole ticker."""
+	"""A link to the ticker's Yahoo Finance page reading `text`, else the bare symbol (brief) or the whole ticker."""
 	if not text:
 		text = ticker.split('.')[0] if brief else ticker
 	return service.link("https://au.finance.yahoo.com/quote/" + ticker, text)
 
 def finance_link(symbol, exchange, service, days=1, brief=True, text=None):
-	"""A link to the stock's page on the site the hyperlinkProvider preference names: Google Finance (google) or
-	Yahoo."""
+	"""A link to the stock's page on the site hyperlinkProvider names: Google Finance (google) or Yahoo."""
 	if config.prefs.hyperlinkProvider == 'google':
 		return gfinance_link(symbol, exchange, service, days, brief, text)
 	return yahoo_link(symbol, service, brief, text)
diff --git a/lib/reports/cal.py b/lib/reports/cal.py
index ad5efd0c..5eeddec0 100644
--- a/lib/reports/cal.py
+++ b/lib/reports/cal.py
@@ -16,8 +16,7 @@ def primary_listings():
 EVENTS = {'earnings': ('earnings_date', 'Upcoming earnings'), 'ex-dividend': ('ex_dividend_date', 'Upcoming ex-dividends')} # -> (field, heading)
 
 def lambda_handler(chat_id=None, days=None, service=None, specific_stock=None, interactive=False, event='earnings'):
-	"""The tracked stocks' (or one stock's) upcoming earnings or ex-dividend dates, as `event` says. A scheduled earnings report
-	is pinned on Telegram."""
+	"""Earnings or ex-dividend dates ahead (`event`) for tracked stocks or one; Telegram pins scheduled earnings."""
 	if days is None:
 		days = config.prefs.future_days
 	field, heading = EVENTS[event]
diff --git a/lib/reports/holdings.py b/lib/reports/holdings.py
index 131730df..4bbbbd76 100644
--- a/lib/reports/holdings.py
+++ b/lib/reports/holdings.py
@@ -35,8 +35,7 @@ def prepare_who(service, code):
 	return payload or [f'{code} not found in any portfolio']
 
 def prepare_pie(portfolio_name, by, service, max_slices=8):
-	"""Portfolio value by industry (or by='sector'). Returns (payload, chart); chart is None unless there is something to draw.
-	Raises sharesight.PortfolioNotFound."""
+	"""Portfolio value by industry (or by='sector'): (payload, chart or None). Raises sharesight.PortfolioNotFound."""
 	if not portfolio_name:
 		names = [portfolio.name for portfolio in sharesight.portfolios()]
 		return [".pie: please try again specifying a portfolio:"] + names, None
diff --git a/lib/shortman.py b/lib/shortman.py
index aa878e9a..c45066b5 100644
--- a/lib/shortman.py
+++ b/lib/shortman.py
@@ -3,8 +3,7 @@
 from lib import net
 
 def fetch(market_data):
-	"""market_data with each ASX ticker's short interest added from Short Man's daily CSV; as it was when Short Man
-	cannot be reached."""
+	"""market_data plus each ASX ticker's short interest from Short Man's daily CSV (unchanged if it is unreachable)."""
 	local_market_data = market_data.copy()
 	url = 'https://www.shortman.com.au/downloadeddata/latest.csv'
 	try:
diff --git a/lib/telegram.py b/lib/telegram.py
index 427564ae..511f0539 100644
--- a/lib/telegram.py
+++ b/lib/telegram.py
@@ -22,8 +22,7 @@ def try_call(method, **kwargs):
 		print(f"Telegram {method} failed:", net.redact(e), file=sys.stderr)
 
 def setWebhook():
-	"""Have Telegram post updates (messages and button presses) to telegramOutgoingWebhook, signed with
-	telegramOutgoingToken."""
+	"""Have Telegram post messages and button presses to telegramOutgoingWebhook, signed with telegramOutgoingToken."""
 	print("registering", config.env.telegramOutgoingWebhook, file=sys.stderr)
 	params = {'url': config.env.telegramOutgoingWebhook, "allowed_updates": json.dumps(['message', 'callback_query']), 'secret_token': config.env.telegramOutgoingToken}
 	return call('setWebhook', params=params)
diff --git a/lib/transports.py b/lib/transports.py
index 3967ee39..0cf3dc7e 100644
--- a/lib/transports.py
+++ b/lib/transports.py
@@ -52,8 +52,7 @@ def text(self, lines, buttons=None):
 		return answers
 
 	def chart(self, image, caption, buttons=None):
-		"""Post a chart with the caption under it, cut to fit. The network's answer, MAYBE_SENT, or None when it
-		failed."""
+		"""Post a chart, the caption under it cut to fit: the network's answer, MAYBE_SENT, or None if it failed."""
 		if hasattr(image, 'seek'):
 			image.seek(0)
 		return self.post_chart(image, fit_caption(caption, self.dialect), buttons)
@@ -70,8 +69,7 @@ def typing(self):
 		"""Show the typing notice, where the network has one."""
 
 	def _post_json(self, url, payload, headers=None):
-		"""POST a JSON message; the answer (or {} when it is not JSON), or None when the network refused it or
-		failed."""
+		"""POST a JSON message: the answer ({} when it is not JSON), or None when the network refused it or failed."""
 		try:
 			r = net.post(url, headers={'Content-type': 'application/json', **(headers or {})}, json=payload,
 				timeout=config.prefs.http_timeout)
@@ -89,8 +87,7 @@ def _post_json(self, url, payload, headers=None):
 
 
 class TelegramBot(Transport):
-	"""A Telegram chat, through the Bot API: HTML, silent, no link previews; charts, buttons, edits and the typing
-	notice."""
+	"""A Telegram chat via the Bot API: HTML, silent, no link previews; charts, buttons, edits, the typing notice."""
 	name = 'telegram'
 	dialect = dialects.TELEGRAM
 	takes_charts = takes_buttons = shows_typing = True
@@ -157,8 +154,7 @@ def edit_chart(self, message, image, caption, buttons=None):
 			print("Telegram editMessageMedia failed:", net.redact(e), file=sys.stderr)
 
 	def edit_text(self, message, text, buttons=None):
-		"""Replace a text message's text, and its buttons unless buttons is None ([] takes them away). A failure is only
-		logged."""
+		"""Edit a message's text, and its buttons unless buttons is None ([] removes them); a failure is only logged."""
 		payload = {'chat_id': self.where, 'message_id': message, 'text': text, 'parse_mode': 'HTML',
 			'disable_web_page_preview': True}
 		if buttons is not None:
@@ -237,8 +233,7 @@ def post_text(self, text, buttons=None):
 
 
 class DiscordWebhook(Transport):
-	"""A Discord webhook, its native endpoint (the Slack-compatible one takes no flags): no link previews, no pings,
-	charts attached."""
+	"""A Discord webhook's native endpoint (the Slack one takes no flags): no previews, no pings, charts attached."""
 	name = 'discord'
 	dialect = dialects.DISCORD
 	takes_charts = True
@@ -252,8 +247,7 @@ def post_text(self, text, buttons=None):
 		return self._post_json(self.url(), {'content': text, 'flags': SUPPRESS_EMBEDS, 'allowed_mentions': NO_MENTIONS})
 
 	def post_chart(self, image, caption, buttons=None):
-		"""The chart as an attachment, with the report as its message. Discord's answer, MAYBE_SENT, or None if
-		refused."""
+		"""The chart as an attachment, the report as its message: Discord's answer, MAYBE_SENT, or None if refused."""
 		payload = {'content': caption, 'flags': SUPPRESS_EMBEDS, 'allowed_mentions': NO_MENTIONS,
 			'attachments': [{'id': 0, 'filename': 'chart.png'}]}
 		try:
@@ -276,8 +270,7 @@ def post_chart(self, image, caption, buttons=None):
 
 
 def slack_api(method, **data):
-	"""POST a Slack Web API method as the bot, form-encoded, and return its answer. Raises RuntimeError with Slack's
-	error if not ok."""
+	"""POST a Slack Web API method as the bot (form-encoded) and return the answer; an error raises RuntimeError."""
 	if not config.env.slackBotToken:
 		raise RuntimeError(f"Slack {method}: needs slackBotToken in finbot.env")
 	r = net.post('https://slack.com/api/' + method, headers={'Authorization': 'Bearer ' + config.env.slackBotToken},
diff --git a/lib/worker.py b/lib/worker.py
index c0ad95c5..0da41e20 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -60,8 +60,7 @@ def is_active(self):
 
 @dataclasses.dataclass
 class Request:
-	"""A chat message being answered: where it came from (service is its dialect), who sent it, and the typing notice
-	while it is answered."""
+	"""A message being answered: where it came from (service is its dialect), who sent it, and its typing notice."""
 	service: str
 	chat_id: str
 	user: str # the sender's handle: whose watchlist add or plan it is
@@ -278,8 +277,7 @@ def on_holdings(request, m):
 	request.reply(payload)
 
 def on_marketcap(request, m):
-	""".marketcap [top|bottom|SYMBOL]: the stocks followed by market cap, largest (or smallest) first, or one
-	stock's."""
+	""".marketcap [top|bottom|SYMBOL]: stocks followed by market cap, largest (or smallest) first, or one stock's."""
 	payload = rankings.prepare_marketcap_payload(request.service, m.group(1) or 'top', length=15)
 	request.reply(payload)
 
diff --git a/tests/common.py b/tests/common.py
index 40aec559..2d36f05e 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -29,8 +29,7 @@
 
 
 def skip_if_old_env(case):
-	"""For the tests that start bot.py: it will not start while the checkout has a leftover .env
-	(config.check_startup_environment)."""
+	"""Skip a test that starts bot.py while the checkout has a leftover .env (config.check_startup_environment)."""
 	if config.OLD_ENV.exists():
 		case.skipTest(f'bot.py will not start while {config.OLD_ENV} exists')
 
@@ -110,8 +109,7 @@ def setUp(self):
 		self.addCleanup(self.no_typing_left)
 
 	def record_chart(self, image, caption, markup=None):
-		"""Transport.chart for the tests: note ('photo', its size, the caption's first line), and its buttons as
-		Telegram would get them."""
+		"""Transport.chart in tests: note ('photo', size, first caption line) and its buttons as Telegram gets them."""
 		image.seek(0)
 		self.sent.append(('photo', len(image.read()), caption.split('\n')[0]))
 		self.markups.append(transports.TelegramBot.markup(markup) if markup else None)
diff --git a/tests/test_charts.py b/tests/test_charts.py
index 220d499e..94864ef4 100644
--- a/tests/test_charts.py
+++ b/tests/test_charts.py
@@ -89,23 +89,20 @@ def test_each_chart_type(self):
 					self.check_image(make(), name)
 
 	def test_edge_case_rows(self):
-		"""Rows that have broken charts before (one row, all zero, forty rows, long labels, huge or tiny values) draw
-		sane images."""
+		"""Rows that broke charts before (one, all zero, forty, long labels, huge or tiny values) draw sane images."""
 		for name, rows in EDGE_ROWS.items():
 			with self.subTest(rows=name):
 				self.check_image(render.hbar_chart(rows, 'Title', 'sub'), name)
 				self.check_image(render.rows_chart(rows, 'Title'), name + '/rows_chart')
 
 	def test_zoomed_axis(self):
-		"""A bar chart with a narrow x range (ratings from 1 to 5) draws sane images, wide labels and many rows
-		included."""
+		"""A bar chart with a narrow x range (ratings, 1 to 5) draws sane images, wide labels and many rows included."""
 		self.check_image(render.hbar_chart([('AAA', 1.2), ('BBB', 3.4)], 'Ratings', xlim=(0.9, 4.0)), 'zoomed')
 		self.check_image(render.hbar_chart([('WWWWWWWW', 1.2), ('MMM', 3.4)], 'Ratings', xlim=(0.9, 4.0)), 'zoomed wide labels')
 		self.check_image(render.hbar_chart([(f'T{i}', 1 + i / 10) for i in range(15)], 'Ratings', xlim=(0.9, 3.0)), 'zoomed 15 rows')
 
 	def test_shorts_chart_sizes_and_verdicts(self):
-		"""The shorts pictogram draws for 1 to 45 rows and a long label, and each tier of short interest gets its
-		verdict."""
+		"""The shorts pictogram draws 1 to 45 rows and a long label; each tier of short interest gets its verdict."""
 		for n in (1, 5, 45):
 			rows = [(f'T{i}', 60.0 - i) for i in range(n)]
 			self.check_image(render.shorts_chart(rows, 15), f'shorts {n} rows', bg=SHORTS_BG, edge_clean=False)
@@ -136,8 +133,7 @@ def setUp(self):
 		self.addCleanup(setattr, render, '_render_pool', None)
 
 	def test_charts_asked_for_at_once_all_finish(self):
-		"""Six charts asked for at once all finish: one render thread, since several deadlock matplotlib under
-		gevent."""
+		"""Six charts asked for at once all finish: one render thread, as several deadlock matplotlib under gevent."""
 		jobs = [gevent.spawn(render.pie_chart, [('a', 1.0), ('b', 2.0)], f't{i}') for i in range(4)]
 		jobs += [gevent.spawn(render.hbar_chart, MIXED, 'T'), gevent.spawn(render.rows_chart, MIXED, 'T')]
 		gevent.joinall(jobs, timeout=120)
@@ -145,8 +141,7 @@ def test_charts_asked_for_at_once_all_finish(self):
 		self.assertTrue(all(decode(job.value).size == (1280, 800) for job in jobs))
 
 	def test_the_hub_keeps_serving_while_a_chart_is_drawn(self):
-		"""While the shorts pictogram is drawn (about a second), other greenlets keep running: no gap over half a
-		second."""
+		"""While the shorts pictogram draws (about a second), other greenlets run on: no gap over half a second."""
 		gaps, done = [], []
 		def ticker():
 			"""Note the time between wake-ups until the chart is done."""
@@ -218,8 +213,7 @@ def count(rows):
 		self.assertGreater(blue, 1000); self.assertGreater(red, 1000)
 
 	def test_session_tints_the_background(self):
-		"""Mid-session tints the background a subtle green and pre-market a subtle yellow; at close and by default it
-		stays plain."""
+		"""Mid-session tints the background faint green, pre-market faint yellow; close and the default stay plain."""
 		corner = lambda session: tuple(np.asarray(decode(render.hbar_chart(MIXED, 'T', session=session)))[2, 2][:3])
 		plain, close, mid, pre = corner(None), corner('close'), corner('midsession'), corner('premarket')
 		self.assertEqual(plain, (0x14, 0x18, 0x1f))
@@ -276,8 +270,7 @@ def spy(rows, title, subtitle='', **kw):
 		return captured[0]
 
 	def test_price_top_lists_best_first_then_worst(self):
-		""".price top: the ten best, best first, a spacer, then the ten worst, worst first, as the caption lists
-		them."""
+		""".price top: the ten best, best first, a spacer, then the ten worst, worst first, as in the caption."""
 		chart = self.run_report(lambda: self.command('.price top 7d'))
 		values = [v for _, v in chart['rows'] if v is not None]
 		top, bottom = values[:10], values[10:]
@@ -332,8 +325,7 @@ def thumb(image):
 		return image.resize((160, 100), Image.LANCZOS) # goldens are stored at this size too: a few KB each, not 100 KB
 
 	def check(self, name):
-		"""Compare a chart type's thumbnail with its golden image (or write the golden with UPDATE_GOLDEN=1, or when
-		there is none)."""
+		"""Compare a chart's thumbnail with its golden image (written instead with UPDATE_GOLDEN=1, or when missing)."""
 		thumb = self.thumb(decode(CHARTS[name]()))
 		path = os.path.join(GOLDEN, name + '.png')
 		if os.environ.get('UPDATE_GOLDEN') or not os.path.exists(path):
diff --git a/tests/test_config.py b/tests/test_config.py
index 746a90a5..1c7f2310 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -49,8 +49,7 @@ def test_the_preference_defaults_are_the_old_defaults(self):
 		self.assertEqual(fields(values), DEFAULTS)
 
 	def test_templates_have_every_key_and_the_real_defaults(self):
-		"""defaults/preferences.conf and defaults/finbot.env list every key, change no default, and are valid as
-		shipped."""
+		"""defaults/preferences.conf and finbot.env list every key, change no default, and are valid as shipped."""
 		self.assertEqual(sorted(keys_in(REPO / 'defaults' / 'preferences.conf')), sorted(config.PREFERENCE_KEYS))
 		self.assertEqual(sorted(keys_in(REPO / 'defaults' / 'finbot.env')), sorted(config.STARTUP_KEYS))
 		prefs = config.read_preferences(REPO / 'defaults' / 'preferences.conf')
@@ -97,8 +96,7 @@ def test_a_startup_key_in_preferences_says_where_it_belongs(self):
 		self.assertNotIn('https://x', ''.join(errors)) # a value is never printed
 
 	def test_malformed_lines_sections_duplicates_and_the_old_name(self):
-		"""Sections, lines that are not KEY=value, a key set twice and the renamed volatile_multiplier are all
-		reported."""
+		"""Sections, lines not KEY=value, a key set twice and the renamed volatile_multiplier are all reported."""
 		errors = self.errors('[reports]\nnot a setting\nfuture_days=1\nfuture_days=2\nvolatile_multiplier=3\n')
 		self.assertEqual(len(errors), 4)
 		self.assertIn('no sections', errors[0])
@@ -189,8 +187,7 @@ def write(self, text):
 		self.file.write_text(text)
 
 	def test_a_change_is_seen_at_once(self):
-		"""A reload applies every change, says what changed, and swaps in a new object, so no request sees half a
-		reload."""
+		"""A reload applies every change, says which, and swaps in a new object, so no request sees half a reload."""
 		self.write('future_days=45\nprice_percent=7\nshorts_percent=12\nexclude_portfolios=1156263\nalliterate=1\nhyperlinkProvider=yahoo\n'
 			'hyperlinkFooter=false\ndebug=1\ngraph=0\ndemote_volatile=0\nvolatility_multiplier=3\npast_days=14\nmin_analysts=3\n')
 		before = config.prefs
@@ -304,8 +301,7 @@ def popen(args, **kw):
 
 class Startup(unittest.TestCase):
 	def run_config(self, code='from lib import config as c; print(c.prefs.future_days)', prefs=None, **env):
-		"""Run code in a fresh interpreter with a data dir of its own (and prefs as its preferences.conf), and the
-		environment cleared of settings."""
+		"""Run code in a new interpreter: own data dir, prefs as preferences.conf, no settings in the environment."""
 		data = tempfile.mkdtemp()
 		if prefs is not None:
 			Path(data, 'preferences.conf').write_text(prefs)
@@ -322,8 +318,7 @@ def test_a_preferences_file_is_read_at_start(self):
 		self.assertEqual(self.run_config(prefs='future_days=45\n').stdout.split()[-1], '45')
 
 	def test_a_broken_preferences_file_refuses_to_start(self):
-		"""A bad value, an unknown key, a startup key or a malformed line in preferences.conf stops the start, saying
-		why."""
+		"""A bad value, unknown key, startup key or malformed line in preferences.conf stops the start, saying why."""
 		for prefs, expect in (('future_days=abc\n', "future_days='abc' is not a whole number"), ('futur_days=1\n', 'unknown key futur_days'),
 				('slack_webhook=x\n', 'belongs in finbot.env'), ('what\n', 'expected KEY=value')):
 			out = self.run_config(prefs=prefs)
@@ -386,16 +381,17 @@ def test_a_discord_webhook_uses_the_native_endpoint(self):
 	def test_settings_that_leave_part_of_the_bot_unable_to_work_are_warned_about(self):
 		"""A startup warning for Slack commands without slackBotToken (they would get no reply), and for an inbound
 		webhook without its secret (every request would be refused)."""
-		hooks = dict(slackOutgoingWebhook='https://example.com/slack', slackOutgoingToken='stok', slackBotToken='xoxb-1',
-			telegramOutgoingWebhook='https://example.com/telegram', telegramOutgoingToken='tok')
+		hooks = dict(slackOutgoingWebhook='https://example.com/slack', slackOutgoingToken='stok',
+			slackBotToken='xoxb-1', telegramOutgoingWebhook='https://example.com/telegram', telegramOutgoingToken='tok')
 		no_reply = 'slackOutgoingWebhook is set but slackBotToken is not: Slack commands will get no reply'
 		slack_refused = ('slackOutgoingWebhook is set but neither slackSigningSecret nor slackOutgoingToken is:'
 			' every Slack request will be refused')
 		telegram_refused = ('telegramOutgoingWebhook is set but telegramOutgoingToken is not: every Telegram update'
 			' will be refused')
-		for changes, warnings in (({}, []), ({'slackBotToken': ''}, [no_reply]),
-				({'slackOutgoingToken': ''}, [slack_refused]), ({'slackOutgoingToken': '', 'slackSigningSecret': 's'}, []),
-				({'telegramOutgoingToken': ''}, [telegram_refused]), ({'telegramOutgoingWebhook': ''}, [])):
+		cases = (({}, []), ({'slackBotToken': ''}, [no_reply]), ({'slackOutgoingToken': ''}, [slack_refused]),
+			({'slackOutgoingToken': '', 'slackSigningSecret': 's'}, []),
+			({'telegramOutgoingToken': ''}, [telegram_refused]), ({'telegramOutgoingWebhook': ''}, []))
+		for changes, warnings in cases:
 			with self.subTest(changes=changes), \
 					mock.patch.object(config, 'env', dataclasses.replace(config.env, **{**hooks, **changes})):
 				self.assertEqual(config.startup_warnings(), warnings)
@@ -418,8 +414,7 @@ def restore(self, file):
 			config.reload()
 
 	def test_reload_all_reloads_preferences_and_the_schedule_independently(self):
-		"""SIGHUP reloads preferences and the schedule; a preferences error is logged, keeps the running values, and the
-		schedule reloads anyway."""
+		"""SIGHUP reloads preferences, keeping the running ones on an error (logged), and the schedule regardless."""
 		file = Path(tempfile.mkdtemp(), 'preferences.conf')
 		file.write_text('future_days=45\n')
 		scheduler = mock.Mock()
@@ -451,8 +446,7 @@ def test_reload_with_the_scheduler_off_still_reloads_preferences(self):
 
 class Tools(unittest.TestCase):
 	def test_split_env(self):
-		"""util/split_env.py sorts an old .env into startup keys and preferences, renames volatile_multiplier, and notes
-		what it drops."""
+		"""split_env sorts an old .env into startup keys and prefs, renames volatile_multiplier, lists what it drops."""
 		startup, prefs, notes = split_env.split('slack_webhook=x\nfuture_days=3\nvolatile_multiplier=2\nslackToken=y\nSTATE_DIRECTORY=/a\nbogus=1\nnot a setting\n# c\nport=80\n')
 		self.assertEqual(startup, ['slack_webhook=x', 'port=80'])
 		self.assertEqual(prefs, ['future_days=3', 'volatility_multiplier=2'])
@@ -485,8 +479,7 @@ def test_split_env_writes_next_to_the_file_and_never_overwrites(self):
 		self.assertEqual((out.returncode, out.stderr.strip()), (1, f'{d}/missing.env does not exist: nothing to split'))
 
 	def test_check_config_script_exit_status_and_it_does_not_create_directories(self):
-		"""util/check_config.py exits 0 for a good file and 1 for a bad one, and creates no data directory (systemd may
-		run it as root)."""
+		"""check_config exits 0 for a good file, 1 for a bad one, and makes no data dir (systemd may run it as root)."""
 		d = Path(tempfile.mkdtemp())
 		(d / 'finbot.env').write_text('port=80\n')
 		data = d / 'data'
diff --git a/tests/test_datadir.py b/tests/test_datadir.py
index 9d10236c..789a6edc 100644
--- a/tests/test_datadir.py
+++ b/tests/test_datadir.py
@@ -34,8 +34,7 @@ def test_config_reads_environment_only(self):
 		self.assertEqual(out.split()[-2:], ['3.3', 'False'])
 
 	def test_state_directory_fallback(self):
-		"""systemd's STATE_DIRECTORY (its first entry) is the data dir when DATA_DIR is unset; MPLCONFIGDIR goes under
-		the cache unless set."""
+		"""Without DATA_DIR, STATE_DIRECTORY's first entry is the data dir; MPLCONFIGDIR is in the cache unless set."""
 		show = "from lib import config as c; print(c.env.data_dir, c.env.cache_dir, c.os.environ['MPLCONFIGDIR'])"
 		with tempfile.TemporaryDirectory() as d:
 			e = {k: v for k, v in os.environ.items() if k not in ('DATA_DIR', 'MPLCONFIGDIR')}
@@ -47,8 +46,7 @@ def run(**env):
 			self.assertEqual(run(DATA_DIR=d, MPLCONFIGDIR='/custom')[2], '/custom') # explicit value kept
 
 	def test_render_import_sets_mplconfigdir_without_home(self):
-		"""Importing render sets MPLCONFIGDIR into the cache before matplotlib loads, so charts work without a home
-		directory."""
+		"""Importing render puts MPLCONFIGDIR in the cache before matplotlib loads: charts work without a home dir."""
 		e = {k: v for k, v in os.environ.items() if k not in ('MPLCONFIGDIR', 'XDG_CONFIG_HOME', 'XDG_CACHE_HOME', 'HOME')}
 		code = "import os; from lib import render; import matplotlib.dates; print(os.environ['MPLCONFIGDIR'])" # render loads matplotlib lazily
 		with tempfile.TemporaryDirectory() as d:
diff --git a/tests/test_edges.py b/tests/test_edges.py
index ed6055cc..12addca5 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -113,8 +113,7 @@ def slow_fetch(tickers):
 		self.assertEqual(sorted(watchlist.tickers()), ['T1', 'T2'])
 
 	def test_removing_a_ticker_yahoo_no_longer_knows(self):
-		"""A ticker Yahoo no longer knows can still be removed, and is said to be; removing it again says it is not
-		there."""
+		"""A ticker Yahoo no longer knows is still removed, and said to be; removing it again says it is not there."""
 		state.save('finbot_watchlist.json', ['GONE', 'T1'])
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts if t != 'GONE'})
 		self.assertIn('I deleted', str(self.command('.watchlist rm GONE')))
@@ -140,16 +139,14 @@ def no_history(t):
 		self.assertEqual(len(self.errors), 1)
 
 	def test_history_without_a_ticker_or_with_extra_argument_gives_usage(self):
-		""".history with no ticker, or with an extra argument, gets its usage, not the profile of a ticker called
-		HISTORY."""
+		""".history with no ticker or an extra argument gets its usage, not the profile of a ticker called HISTORY."""
 		self.assertEqual(self.command('.history'), [('text', ['Usage: .history SYMBOL'])]) # not the profile of a ticker called HISTORY
 		self.assertEqual(self.command('.history T1 1y'), [('text', ['Usage: .history SYMBOL'])])
 
 
 class PortfolioNames(FinbotCase):
 	def test_an_unknown_portfolio_gets_the_same_answer_everywhere(self):
-		"""An unknown portfolio name gets the same answer from .trades, .performance, .holdings and .pie: the portfolios
-		there are."""
+		"""An unknown portfolio gets one answer from .trades, .performance, .holdings and .pie: the ones there are."""
 		for cmd in ('.trades 7d Nope', '.performance Nope', '.holdings Nope', '.pie Nope'):
 			with self.subTest(cmd=cmd): # it used to be a ValueError, "no Sharesight data found", or two wordings of this
 				self.assertEqual(self.command(cmd), [('text', ['Nope portfolio not found. I only know about:', 'Rob'])])
@@ -248,8 +245,7 @@ def test_malformed_body_is_refused_not_a_crash(self):
 		self.assertEqual(self.errors, [])
 
 	def test_strangers_cannot_post_into_the_chats(self):
-		"""Requests from strangers (wrong paths, no or wrong secrets, malformed bodies) are refused, and nothing reaches
-		the chats."""
+		"""Strangers (wrong paths, no or wrong secrets, malformed bodies) are refused, and nothing reaches the chats."""
 		self.env(slackOutgoingWebhook='https://example.com/slack')
 
 		self.env(slackOutgoingToken='stok')
@@ -277,8 +273,7 @@ def broken(*a, **k):
 		self.assertIn('boom', err.getvalue())
 
 	def test_a_failure_after_authentication_is_reported_once_and_not_retried(self):
-		"""A failure on an authenticated update is reported once and answered 200, so Telegram does not send it
-		again."""
+		"""A failure on an authenticated update is reported once and answered 200, so Telegram doesn't send it again."""
 		broken = {'message': {'message_id': 1, 'from': {'first_name': 'A'}, 'text': '.help'}} # no chat
 		self.assertEqual(self.request(body=broken, headers=self.SECRET)[1], '200 OK') # a retry would only fail and report again
 		self.assertEqual(len(self.errors), 1)
@@ -524,8 +519,7 @@ def test_sighup_during_startup_is_ignored_and_works_afterwards(self):
 			self.assertIn('reload requested but', ''.join(lines))
 
 	def test_usr1_flushes_the_cache_and_keeps_the_state(self):
-		"""SIGUSR1 is ignored during startup; once serving, it deletes the cache files but not a write in flight,
-		matplotlib's cache or the state."""
+		"""SIGUSR1 is ignored in startup, then clears the cache but not writes in flight, matplotlib's or the state."""
 		skip_if_old_env(self)
 		with tempfile.TemporaryDirectory() as data:
 			cache = Path(data, 'cache')
diff --git a/tests/test_finbot.py b/tests/test_finbot.py
index 2e526539..d6ba43d6 100644
--- a/tests/test_finbot.py
+++ b/tests/test_finbot.py
@@ -29,8 +29,7 @@ def post(url, **kwargs):
 		self.assertTrue(all('reply_markup' not in p for p in posted[:-1]))
 
 	def test_chunks_respect_both_the_line_count_and_the_size(self):
-		"""A message ends at chunk_maxlines lines or the size limit, whichever comes first; a line too long on its own
-		is cut short."""
+		"""A message ends at chunk_maxlines lines or the size limit, whichever first; an overlong line is cut short."""
 		chunks = transports.chunked(['a' * 30] * 10 + ['b' * 500], limit=100, max_lines=4)
 		self.assertEqual([len(chunk) for chunk in chunks], [3, 3, 3, 1, 1]) # 3 x 30 plus newlines is 92; a fourth would pass 100
 		self.assertTrue(all(len('\n'.join(chunk)) <= 100 for chunk in chunks))
@@ -77,8 +76,7 @@ def test_every_function_has_a_docstring(self):
 		self.assertEqual(missing, [])
 
 	def test_the_bot_and_its_reports_print_only_to_stderr(self):
-		"""systemd discards finbot's stdout (StandardOutput=null), so a print there is never read. The util/ tools, run by
-		hand, may use it."""
+		"""systemd discards stdout (StandardOutput=null), so only util/ tools, run by hand, may print there."""
 		offenders = []
 		for path in sorted([*Path(ROOT, 'lib').rglob('*.py'), *Path(ROOT).glob('*.py')]):
 			for node in ast.walk(ast.parse(path.read_text())):
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
index babf5608..2ed92a94 100644
--- a/tests/test_jobs.py
+++ b/tests/test_jobs.py
@@ -108,8 +108,7 @@ def test_local_zone_follows_tz(self):
 
 class JobArguments(unittest.TestCase):
 	def test_price(self):
-		"""price.py takes one mode or a number of days (the top and bottom ten over that span); anything else is its
-		usage."""
+		"""price.py takes one mode or a number of days (the top and bottom ten over it); anything else is its usage."""
 		self.assertEqual(price.job(['midsession']), ('price.py midsession', {'midsession': True}))
 		self.assertEqual(price.job(['premarket'])[1], {'premarket': True})
 		self.assertEqual(price.job(['interday'])[1], {'interday': True})
@@ -182,8 +181,7 @@ def test_each_scheduled_report_runs_from_its_top_level_script(self):
 
 class CleanupJob(FinbotCase):
 	def test_removes_only_old_finbot_cache_files(self):
-		"""cleanup deletes finbot's cache files older than the days given, not newer ones, other files or
-		directories."""
+		"""cleanup deletes finbot's cache files older than the days given, not newer ones, other files or dirs."""
 		folder = tempfile.mkdtemp()
 		self.addCleanup(lambda: __import__('shutil').rmtree(folder, ignore_errors=True))
 		self.env(cache_dir=folder)
@@ -199,8 +197,7 @@ def test_removes_only_old_finbot_cache_files(self):
 		self.assertEqual(sorted(os.listdir(folder)), ['finbot_dir', 'finbot_new.json', 'other_old.json'])
 
 	def test_a_temp_file_being_written_is_left_alone(self):
-		"""A flush (0 days) leaves a temp file written in the last minute, an atomic write in flight, and deletes a
-		stale one."""
+		"""A flush (0 days) spares a temp file from the last minute (a write in flight) and deletes a stale one."""
 		folder = tempfile.mkdtemp()
 		self.addCleanup(lambda: __import__('shutil').rmtree(folder, ignore_errors=True))
 		self.env(cache_dir=folder)
@@ -238,8 +235,7 @@ def test_on_by_default_and_off_for_false(self):
 
 class SchedulerBase(FinbotCase):
 	def setUp(self):
-		"""A schedule of trades.py every 20 minutes on weekdays, a fake clock, and trades and shorts recorded instead of
-		run."""
+		"""A schedule of trades.py every 20 minutes on weekdays, a fake clock; trades and shorts recorded, not run."""
 		super().setUp()
 		tmp = tempfile.TemporaryDirectory()
 		self.addCleanup(tmp.cleanup)
@@ -364,8 +360,7 @@ def test_a_job_that_hangs_is_abandoned_and_reported(self):
 		self.assertFalse(s.jobs[0].running)
 
 	def test_reload_applies_edits_keeps_state_and_survives_a_bad_file(self):
-		"""A reload applies edits and keeps the timing of unchanged jobs; a file with nothing usable keeps the running
-		schedule."""
+		"""A reload applies edits and keeps unchanged jobs' timing; an unusable file keeps the running schedule."""
 		s = self.scheduler()
 		old_key = s.jobs[0].key
 		base = s.base[old_key]
@@ -507,8 +502,7 @@ def test_select_by_number_or_command_text(self):
 		self.assertEqual(self.s.select('~/finbot/price.py 7')[0], 'price.py 7')
 
 	def test_select_refuses_what_is_not_offered(self):
-		"""Numbers out of range, cleanup, podcasts, commands not in the schedule, shell syntax and two commands are
-		refused."""
+		"""Numbers out of range, cleanup, podcasts, unscheduled commands, shell syntax and two commands are refused."""
 		for text in ('0', '9', 'cleanup.py', 'podcasts.py "https://example.com/feed"', 'price.py 99', 'cal.py earnings', 'nonsense', 'ls | wc', '', 'price.py close; trades.py'):
 			with self.assertRaises(jobs.ScheduleError, msg=text):
 				self.s.select(text)
@@ -554,8 +548,7 @@ def list_in_chat(self, text='.jobs'):
 		return self.command(text)[0][1]
 
 	def test_list_in_chat_is_just_the_commands(self):
-		""".jobs and its aliases list the commands by number, without times, comments or redirects, and mark one that is
-		running."""
+		""".jobs and aliases number the commands, without times, comments or redirects, and mark one that is running."""
 		for alias in ('.jobs', '.cron', '.crons', '.schedule', '@' + BOT.lstrip('@') + ' jobs'):
 			text = '\n'.join(map(str, self.list_in_chat(alias)))
 			for line in ('1. performance 7', '2. performance 30', '5. performance 365', '6. price 7', '7. price close', '8. trades'):
@@ -603,8 +596,7 @@ def stale_buttons(self):
 			self.edits.append((dest.where, mid, text, buttons)))
 
 	def test_forged_button_data_never_runs_anything_else(self):
-		"""Forged or malformed button data runs nothing and refreshes the buttons in place; cleanup and podcasts cannot
-		be picked by any number."""
+		"""Bad or forged button data runs nothing, refreshes buttons in place; no number picks cleanup or podcasts."""
 		self.stale_buttons()
 		good = jobs.checksum('trades.py')
 		for data in ('j|-1|' + good, 'j|0|' + good, 'j|9|' + good, 'j|8|0000', 'j|8|', 'j|8', 'j|8|' + good + '|x', 'j||' + good,
@@ -626,8 +618,7 @@ def test_forged_button_data_never_runs_anything_else(self):
 		self.assertEqual(self.ran, [('trades', {})])
 
 	def test_button_from_an_old_schedule_updates_the_message(self):
-		"""A button from an old schedule runs nothing, and the message gets the current list and buttons; with the
-		scheduler off, a notice."""
+		"""A stale button runs nothing; the message gets the current list and buttons (scheduler off: a notice)."""
 		self.stale_buttons()
 		worker.process_callback('telegram', 'cb1', '55', '9', 'j|7|0000')
 		worker.process_callback('telegram', 'cb1', '55', '9', 'j|9|0000')
@@ -651,8 +642,7 @@ def test_run_in_chat(self):
 		self.assertEqual(self.ran, [('price', {'close': True})])
 
 	def test_run_in_chat_refusals(self):
-		""".jobs run refuses a command not in the schedule, an unknown number and two commands, and gives its usage
-		otherwise."""
+		""".jobs run refuses an unscheduled command, an unknown number and two commands, else gives its usage."""
 		self.assertIn('not in the schedule', self.list_in_chat('.jobs run cal.py earnings')[0])
 		self.assertIn('no command 99', self.list_in_chat('.jobs run 99')[0])
 		self.assertIn('one command at a time', self.list_in_chat('.jobs run price.py close; trades.py')[0])
@@ -737,8 +727,7 @@ def test_scheduler_reports_a_child_that_ended_without_saying_why(self):
 				self.assertEqual(s.busy, set())
 
 	def test_a_report_script_loads_neither_pandas_nor_the_scheduler(self):
-		"""A job process imports only its own script: price used to pull in pandas and, through charts, reports and jobs, every
-		script there is."""
+		"""A job imports only its script; price once pulled in pandas and every script (via charts, reports, jobs)."""
 		code = 'import sys; from lib.reports import price, performance, trades, shorts, cal; print(sorted(m for m in ("pandas", "lib.jobs", "matplotlib") if m in sys.modules))'
 		with tempfile.TemporaryDirectory() as data:
 			out = subprocess.run([sys.executable, '-c', code], cwd=ROOT, capture_output=True, text=True, env={**os.environ, 'DATA_DIR': data})
diff --git a/tests/test_providers.py b/tests/test_providers.py
index b04f7a59..9a92ce1c 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -42,8 +42,7 @@ def get(url, **kwargs):
 
 class ProviderCase(FinbotCase):
 	def setUp(self):
-		"""The real parsers in place of the fixture's canned data, with a fixed cookie, crumb and token, and no
-		waiting."""
+		"""Real parsers instead of the fixture's canned data, with a fixed cookie, crumb and token, and no waiting."""
 		super().setUp()
 		# these tests exercise the real parsers, so undo the canned data from the base fixture
 		for module, name in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (sharesight, 'get_portfolios'),
@@ -55,8 +54,7 @@ def setUp(self):
 		self.patch(sharesight.time, 'sleep', lambda s: None)
 
 	def http(self, module, *responses):
-		"""Have module's net.get replay these responses; returns the replacement, whose calls lists the URLs asked
-		for."""
+		"""Have module's net.get replay these responses; returns the stand-in, whose calls lists the URLs asked for."""
 		getter = serve(*responses)
 		self.patch(module.net, 'get', getter)
 		return getter
@@ -441,8 +439,7 @@ def warm():
 		self.assertEqual(stop.waits, [300, 3000]) # slow retry after the failure, then sleep until the next cache is due
 
 	def test_with_cache_off_nothing_is_fetched_until_a_reload_turns_it_on(self):
-		"""With the cache preference off, keep_warm idles without asking Sharesight, and starts when a reload turns it
-		on."""
+		"""With cache off, keep_warm idles without asking Sharesight, and starts when a reload turns it on."""
 		class Stop:
 			"""keep_warm's stop event: notes each wait, switches the cache on at the second, and is set after four."""
 			def __init__(self):
@@ -580,8 +577,7 @@ def test_guarded_reports_and_exits_nonzero(self):
 
 
 class ChartDelivery(FinbotCase):
-	"""Charts go to Telegram, and to Slack in replies (uploaded into the channel). A report whose chart cannot be sent goes out as
-	text instead of vanishing."""
+	"""Charts reach Telegram and Slack replies (uploaded); a report whose chart fails still goes, as text."""
 	UPLOAD_URL = 'https://files.slack.com/upload/v1/abc'
 
 	def setUp(self):
@@ -607,8 +603,7 @@ def upload_steps(self):
 		return [Resp({'ok': True, 'upload_url': self.UPLOAD_URL, 'file_id': 'F1'}), Resp(None, text='OK - 3'), Resp({'ok': True, 'files': [{'id': 'F1'}]})]
 
 	def test_where_charts_can_go(self):
-		"""Charts go to Telegram and Discord always, to Slack in replies with the bot token, never in Slack's scheduled
-		reports."""
+		"""Charts always go to Telegram and Discord; Slack gets them only in replies with the bot token."""
 		self.env(slackBotToken='xoxb-test')
 
 		cases = (('telegram',), ('telegram', False), ('slack',), ('slack', False), ('discord', False))
@@ -619,8 +614,7 @@ def test_where_charts_can_go(self):
 		self.assertFalse(delivery.takes_charts('slack'))
 
 	def test_slack_replies_upload_their_charts(self):
-		"""A Slack reply uploads its chart in three steps with the bot token, the report as its message, not sent again
-		as text."""
+		"""A Slack reply uploads its chart in three steps (bot token), the report as its message, with no text copy."""
 		self.env(slackBotToken='xoxb-test')
 
 		for cmd, heading in (('.beta', '*Beta over 1.5'), ('.price', '*5% at close'), ('.compare T1 T2', '*')):
@@ -640,8 +634,7 @@ def test_slack_replies_upload_their_charts(self):
 				self.assertNotIn(share['data']['initial_comment'].split('\n'), [payload for _, payload in self.sent]) # not sent again as text
 
 	def test_a_slack_upload_that_fails_goes_out_as_text(self):
-		"""A Slack upload that fails at any step goes out as text, but not after a timeout on the last step, when it may
-		have arrived."""
+		"""A Slack upload failing at any step goes as text, but not on a last-step timeout, when it may have arrived."""
 		self.env(slackBotToken='xoxb-test')
 
 		good = self.upload_steps()
@@ -712,8 +705,7 @@ def test_the_graph_preference_off_sends_no_chart_anywhere(self):
 		self.assertEqual([url for url in posts if 'sendPhoto' in url or 'files.' in url], [])
 
 	def test_a_chart_telegram_refuses_goes_out_as_text(self):
-		"""A chart Telegram refuses or that fails goes out as text, but not after a timeout, when it may have
-		arrived."""
+		"""A chart Telegram refuses or that fails goes as text, but not after a timeout, when it may have arrived."""
 		def post(outcome):
 			"""A net.post that answers with outcome, or raises it."""
 			def fake(url, **kwargs):
@@ -757,8 +749,7 @@ def test_text_goes_to_the_native_endpoint_without_previews_or_pings(self):
 		self.assertEqual(kwargs['json'], {'content': 'hello @everyone', 'flags': 4, 'allowed_mentions': {'parse': []}})
 
 	def test_a_scheduled_report_carries_its_chart(self):
-		"""A scheduled report goes to Discord with its chart attached and the report, within 2,000 characters, as the
-		message."""
+		"""A scheduled Discord report has its chart attached and the report as its message, within 2,000 characters."""
 		price.lambda_handler()
 		(url, kwargs), = self.posts
 		self.assertEqual(url, self.URL + '?wait=true')
diff --git a/tests/test_reports.py b/tests/test_reports.py
index a3ba8fce..320dd710 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -414,8 +414,7 @@ def test_a_button_a_chart_does_not_offer_is_ignored(self):
 
 class Podcasts(FinbotCase):
 	def test_format_helpers(self):
-		"""Durations in any of the feed formats, how long ago, and dates without a zone taken as UTC; nonsense gives
-		nothing."""
+		"""Durations in any feed format, how long ago, and zoneless dates taken as UTC; nonsense gives nothing."""
 		from lib.reports import podcasts
 		self.assertEqual(podcasts.format_duration('3725'), '01:02')
 		self.assertEqual(podcasts.format_duration('1:02:03'), '01:02')
diff --git a/tests/test_util.py b/tests/test_util.py
index cbb11b57..1594f92b 100644
--- a/tests/test_util.py
+++ b/tests/test_util.py
@@ -20,8 +20,7 @@ def test_days_from_human_days_rejects_non_periods(self):
 				formats.days_from_human_days(arg)
 
 	def test_transform_to_yahoo(self):
-		"""Sharesight and Google tickers become Yahoo's: suffixes by market, none for US venues, a dash for class
-		shares."""
+		"""Sharesight and Google tickers as Yahoo's: suffixes by market, none for US venues, a dash for class shares."""
 		for args, expected in ((('AAPL',), 'AAPL'), (('FANG', 'ASX'), 'FANG.AX'), (('ARM', 'NASDAQ'), 'ARM'),
 							   (('ARM:NASDAQ',), 'ARM'), (('BHP.AX',), 'BHP.AX'), (('0700', 'HKG'), '0700.HK'),
 							   (('BRK', 'B'), 'BRK-B'), (('2330', 'TAI'), '2330.TW'), (('VOD', 'LSE'), 'VOD.L'),
@@ -45,8 +44,7 @@ def test_make_paragraphs_keeps_every_word(self):
 		self.assertEqual(formats.make_paragraphs(long), ' '.join(['word'] * 22) + ' end.\n\nTail sentence here.')
 
 	def test_transform_title(self):
-		"""Company names lose shouting and legal suffixes (Inc, FPO, SE, a trailing &), from the end only, leaving no
-		run of spaces."""
+		"""Company names lose shouting and trailing legal suffixes (Inc, FPO, SE, a final &), with no double spaces."""
 		self.assertEqual(formats.transform_title('APPLE INC'), 'Apple')
 		self.assertEqual(formats.transform_title('Acme FPO'), 'Acme')
 		self.assertEqual(formats.transform_title('Mixed Case Name'), 'Mixed Case Name')
@@ -111,8 +109,7 @@ def test_json_roundtrip_and_missing_file(self):
 		self.assertEqual(state.load('state.json'), {'a': [1, 2]})
 
 	def test_a_failed_write_leaves_the_old_file_and_the_umask_alone(self):
-		"""A write that fails (disk full at replace or fsync) leaves the old file and no temp file; the umask is
-		untouched and files are 0640."""
+		"""A write failing at replace or fsync keeps the old file and no temp file; umask untouched, files 0640."""
 		state.save('state.json', {'a': 1})
 		path = Path(config.env.data_dir, 'state.json')
 		previous = os.umask(0o027)
@@ -134,8 +131,7 @@ def disk_full(*a, **k):
 		self.assertEqual((json.loads(path.read_text()), oct(path.stat().st_mode & 0o777)), ({'a': 3}, '0o640'))
 
 	def test_edit_writes_only_a_change_and_never_changes_the_default(self):
-		"""state.edit writes only when the block changed the data, never alters the default it was given, and writes
-		nothing when the block fails."""
+		"""state.edit writes only if the block changed the data without failing, and never alters the default given."""
 		default = {'seen': []}
 		with state.edit('edit.json', default):
 			pass
@@ -149,8 +145,7 @@ def test_edit_writes_only_a_change_and_never_changes_the_default(self):
 		self.assertEqual(state.load('edit.json'), {'seen': [1]}) # a block that fails writes nothing
 
 	def test_edits_at_once_all_count(self):
-		"""Each edit reads the file after the one before it wrote, so none overwrites another with what it read
-		earlier."""
+		"""Each edit reads the file once the previous one wrote, so none overwrites another with stale data."""
 		def edit(key):
 			"""Set data[key] slowly, so that the other edits start meanwhile."""
 			with state.edit('edit.json', {}) as data:
diff --git a/util/sws_gen.py b/util/sws_gen.py
index 46f768db..a176df09 100644
--- a/util/sws_gen.py
+++ b/util/sws_gen.py
@@ -37,8 +37,7 @@ def build_dict(symbols, market):
 	return mydict
 
 def do_sws_bulk(file):
-	"""Look up every symbol in a file (as ASX listings when its name says asx) and write them to finbot_sws_<name>.json
-	in the data dir."""
+	"""Look up a file's symbols (as ASX listings if its name says asx); write finbot_sws_<name>.json in the data dir."""
 	market = ''
 	if 'asx' in file or 'ASX' in file:
 		market = 'ASX'

From aebc37f85ed0a3d7b884efceb13513df435f4bbb Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 01:34:26 +0000
Subject: [PATCH 23/81] No data is an answer, not an error (D11)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

lib/answer.py adds Answer(*lines): a reply in place of the report
asked for. The worker and the period buttons send its lines as an
ordinary message. Every other exception is still reported as a
one-line ⚠️ error.

- sharesight.PortfolioNotFound is now an Answer.
- yahoo.NoData is both an Answer and a RuntimeError, so code that
  leaves a ticker out of a list on RuntimeError still does. It is
  raised for:
  - a history without prices;
  - no high/low history;
  - Yahoo's 404 whose body says "Not Found", i.e. no such symbol.
  Any other 404 (an endpoint that moved, say), a refusal or an
  unreachable Yahoo stays a RuntimeError, reported as an error.
- .history and .compare of a ticker without data, compare without
  enough history in the period, and a period button with nothing to
  draw now answer plainly. Before, each was "⚠️ ...: RuntimeError: ...".

The output comparison is identical in all 159 steps. Its fixtures
always have data, so the no-data replies are covered by new tests
instead.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md               |  2 ++
 lib/answer.py           | 12 ++++++++++++
 lib/charts.py           | 11 ++++++-----
 lib/reports/history.py  |  3 ++-
 lib/sharesight.py       |  8 ++++----
 lib/worker.py           |  6 ++++--
 lib/yahoo.py            | 34 ++++++++++++++++++++++++----------
 tests/common.py         |  2 +-
 tests/test_edges.py     | 30 ++++++++++++++----------------
 tests/test_providers.py | 13 +++++++++++++
 10 files changed, 82 insertions(+), 39 deletions(-)
 create mode 100644 lib/answer.py

diff --git a/CLAUDE.md b/CLAUDE.md
index 1769e830..74281adc 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -66,6 +66,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - A report is a list of lines. `service` is a dialect (`lib/dialects.py`), which equals the network's name: markup with
   `service.bold(text)`, `.italic`, `.strike`, `.link(url, text)`, user-written text through `service.escape(text)`, stock links
   through `links.finance_link(symbol, exchange, service)`. Tests pass `TELEGRAM`, `SLACK`, `DISCORD`.
+- A command that cannot give what was asked (no such portfolio, no prices) raises `answer.Answer(*lines)`, which the chat
+  gets as a plain reply; `yahoo.NoData` is one. Any other exception becomes a one-line ⚠️ error report.
 - Every function and method has a docstring, tests included (a test checks): one line where it fits, reworded rather
   than spilling a few words onto a second. Tabs everywhere (`.editorconfig`). New and changed lines stay within 120
   columns, tabs counting 4 (an edit quotes whole lines).
diff --git a/lib/answer.py b/lib/answer.py
new file mode 100644
index 00000000..f114ce48
--- /dev/null
+++ b/lib/answer.py
@@ -0,0 +1,12 @@
+"""A reply in place of the report asked for. A command that cannot give what was asked (a portfolio name that is none of
+them, a ticker with no price history) raises Answer, and the chat gets its lines as an ordinary reply, where any other
+exception becomes a one-line error report."""
+
+
+class Answer(Exception):
+	"""The lines to reply with instead of the report."""
+
+	def __init__(self, *lines):
+		"""lines: the reply, a line each."""
+		self.lines = [str(line) for line in lines]
+		super().__init__(' '.join(self.lines))
diff --git a/lib/charts.py b/lib/charts.py
index 46197975..9906808d 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -3,6 +3,7 @@
 import json
 import sys
 
+from lib import answer
 from lib import buttons
 from lib import formats
 from lib import links
@@ -60,7 +61,7 @@ def build(kind, tickers, period, service):
 		ticker = tickers[0]
 		lines, table, image = history.history_payload(ticker, service, buttons.PERIODS[period or 'x'][0], graph=not days) # default view is the full history
 		if table is None:
-			raise RuntimeError(lines[0])
+			raise answer.Answer(*lines)
 		if days:
 			_, image = yahoo.price_history(ticker, days)
 		return '\n'.join(lines), image
@@ -68,7 +69,7 @@ def build(kind, tickers, period, service):
 		ticker = tickers[0]
 		market_data = yahoo.fetch([ticker])
 		if ticker not in market_data:
-			raise RuntimeError(f"no data found for {ticker}")
+			raise answer.Answer(f"no data found for {ticker}")
 		percent, image = yahoo.price_history(ticker, days, graphCache=False)
 		percent = float(percent.get(days, percent.get('Max', next(iter(percent.values()), 0))))
 		link = links.quote_link(ticker, market_data[ticker]['profile_exchange'], service, days=days)
@@ -77,17 +78,17 @@ def build(kind, tickers, period, service):
 		args = list(tickers) + ([f'{days}d'] if days else ['max'])
 		caption, image, *_ = history.prepare_compare(service, args)
 		if not image:
-			raise RuntimeError(caption[0])
+			raise answer.Answer(*caption)
 		return '\n'.join(caption), image
 	if kind == 'f': # performance: tickers = [portfolio name or '']
 		payload, image = performance.report(service, days or buttons.MAX_DAYS, tickers[0] or None, interactive=True)
 		if not image:
-			raise RuntimeError(payload[0] if payload else 'no performance data')
+			raise answer.Answer(*(payload or ['no performance data']))
 		return '\n'.join(payload), image
 	if kind == 'l': # price list: tickers = [threshold, top]
 		span = {'interday': True} if period == 'd' else {'days': days or buttons.MAX_DAYS} # 'd' is today's moves: the command's default view, not a 1-day lookback
 		payload, image = price.report(service, threshold=float(tickers[0]), interactive=True, top=int(tickers[1]) or None, **span)
 		if not image:
-			raise RuntimeError(payload[0] if payload else 'no price data')
+			raise answer.Answer(*(payload or ['no price data']))
 		return '\n'.join(payload), image
 	raise ValueError(f"unknown chart kind {kind}")
diff --git a/lib/reports/history.py b/lib/reports/history.py
index 87e4c39d..b8b2569e 100644
--- a/lib/reports/history.py
+++ b/lib/reports/history.py
@@ -1,6 +1,7 @@
 """Price history: .history (one stock over every period) and .compare (two to six stocks, rebased)."""
 import hashlib
 import datetime
+from lib import answer
 from lib import config
 from lib import cache
 from lib import formats
@@ -64,7 +65,7 @@ def prepare_compare(service, args):
 		series = pd.concat([pd.Series([base_value], index=[start]), series[series.index > start]])
 		cropped.append((ticker, series))
 	if any(len(series) < 2 for _, series in cropped):
-		raise RuntimeError("not enough price history in that period to compare")
+		raise answer.Answer("not enough price history in that period to compare")
 	span_days = (datetime.datetime.now().date() - start.date()).days
 	period_label = formats.days_english(period_days) if period_days and not note else f"{span_days} days"
 	by_performance = sorted(cropped, key=lambda item: item[1].iloc[-1] / item[1].iloc[0], reverse=True)
diff --git a/lib/sharesight.py b/lib/sharesight.py
index c544a86c..fc5ea877 100644
--- a/lib/sharesight.py
+++ b/lib/sharesight.py
@@ -4,6 +4,7 @@
 import threading
 import time
 import requests
+from lib import answer
 from lib import config
 from lib import cache
 from lib import net
@@ -94,12 +95,11 @@ def get_portfolios(refresh=False):
 		raise RuntimeError("No Sharesight portfolios found")
 	return portfolio_dict
 
-class PortfolioNotFound(LookupError):
-	"""No portfolio is called what was asked for. `lines` is the answer to give: the name, and the portfolios there are."""
+class PortfolioNotFound(answer.Answer, LookupError):
+	"""No portfolio is called what was asked for: the answer names it, and the portfolios there are."""
 	def __init__(self, name, names):
 		"""name: what was asked for; names: the portfolios there are."""
-		self.lines = [f"{name} portfolio not found. I only know about:"] + list(names)
-		super().__init__(' '.join(self.lines))
+		super().__init__(f"{name} portfolio not found. I only know about:", *names)
 
 @dataclasses.dataclass(frozen=True)
 class Portfolio:
diff --git a/lib/worker.py b/lib/worker.py
index 0da41e20..c5be02b3 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -6,6 +6,7 @@
 import re
 import sys
 import threading
+from lib import answer
 from lib import config
 from lib import dialects
 from lib import state
@@ -17,7 +18,6 @@
 from lib import charts
 from lib.reports import about, cal, history, holdings, performance, price, profile, rankings, shorts, trades
 from lib import telegram
-from lib import sharesight
 from lib import jobs
 
 class TypingIndicator:
@@ -112,6 +112,8 @@ def process_callback(service, callback_id, chat_id, message_id, data):
 		caption, image = charts.build(kind, tickers, period, service)
 		chat = delivery.reply_to(service, chat_id)
 		chat.edit_chart(message_id, image, caption, buttons.keyboard(kind, tickers, period))
+	except answer.Answer as e: # nothing to draw for that period: said in the chat, as a command would
+		delivery.deliver(delivery.reply_to(service, chat_id), delivery.Report(e.lines))
 	except Exception as e:
 		delivery.report_error(e, service, chat_id, context='chart')
 
@@ -126,7 +128,7 @@ def process_request(service, chat_id, user, message, botName, userRealName):
 				request.typing.start()
 				handler(request, m)
 				break
-	except sharesight.PortfolioNotFound as e: # a name the user got wrong: an answer, not an error
+	except answer.Answer as e: # a portfolio name the user got wrong, a ticker with no prices: an answer, not an error
 		request.reply(e.lines)
 	except Exception as e:
 		request.typing.stop() # before the error line
diff --git a/lib/yahoo.py b/lib/yahoo.py
index cfef0c2e..18821c6a 100644
--- a/lib/yahoo.py
+++ b/lib/yahoo.py
@@ -6,6 +6,7 @@
 import threading
 import zoneinfo
 import sys
+from lib import answer
 from lib import config
 from lib import cache
 from lib import net
@@ -306,9 +307,13 @@ def fetch_detail(ticker, seconds=None):
 		detail['insiderSell'], detail['insiderSellValue'] = totals['Sale']
 	return {ticker: dict(sorted(detail.items()))}
 
+class NoData(answer.Answer, RuntimeError):
+	"""Yahoo has nothing for it (no such symbol, no prices): a reply in a chat, a RuntimeError where a list skips it."""
+
+
 def price_history(ticker, days=None, seconds=None, graph=None, graphCache=True):
 	"""({period: percent change}, chart image or None): the change over `days`, else over each standard period. Raises
-	RuntimeError when the history cannot be fetched or has no prices."""
+	NoData when there are no prices, RuntimeError when the history cannot be fetched."""
 	if graph is None:
 		graph = config.prefs.graph
 	if seconds is None:
@@ -324,13 +329,13 @@ def price_history(ticker, days=None, seconds=None, graph=None, graphCache=True):
 	stock = chart_json_to_stock_basics(data)
 	closes = df['Close'].dropna()
 	if closes.empty:
-		raise RuntimeError(f"no price history for {ticker}")
+		raise NoData(f"no price history for {ticker}")
 	price = stock.get('regularMarketPrice') or closes.iloc[-1] # the chart's last close when Yahoo gives no current price
 	tz = zoneinfo.ZoneInfo(stock.get('exchangeTimezoneName') or 'UTC')
 	if days:
 		df = df[df['Date'] >= (now - datetime.timedelta(days=days)).date()].reset_index(drop=True)
 		if df.empty:
-			raise RuntimeError(f"no price history for {ticker} in the past {days} days")
+			raise NoData(f"no price history for {ticker} in the past {days} days")
 		past_price = df['Close'].iloc[0]
 		percent_dict[days] = round((price - past_price) / past_price * 100, 2)
 	else:
@@ -376,14 +381,14 @@ def price_history(ticker, days=None, seconds=None, graph=None, graphCache=True):
 
 def price_series(ticker):
 	"""Daily closes for a ticker via the same fetch/parse path as price_history(): (name, exchange, pandas Series indexed by datetime).
-	Raises RuntimeError when there is no history."""
+	Raises NoData when there is no history."""
 	import pandas as pd
 	data = fetch_chart_json(ticker, full=True)
 	df = chart_json_to_df(data)
 	stock = chart_json_to_stock_basics(data)
 	df = df[df['Close'].notnull()]
 	if df.empty:
-		raise RuntimeError(f"no price history for {ticker}")
+		raise NoData(f"no price history for {ticker}")
 	series = pd.Series(df['Close'].astype(float).values, index=pd.to_datetime(df['Date'].astype(str)), name=ticker)
 	name = formats.transform_title(stock.get('longName') or stock.get('shortName') or ticker)
 	return name, stock.get('fullExchangeName') or stock.get('exchangeName') or '', series
@@ -450,8 +455,8 @@ def prefetch_history(tickers, workers=8):
 			print("prefetch failed for", ticker, result, file=sys.stderr)
 
 def fetch_chart_json(ticker, days=3665, seconds=None, full=False):
-	"""Yahoo's daily chart for the last `days`, or with full=True its entire history (cached separately). Raises RuntimeError when
-	Yahoo cannot be reached or refuses."""
+	"""Yahoo's daily chart for the last `days`, or with full=True its entire history (cached separately). Raises NoData
+	for a symbol Yahoo does not know, RuntimeError when Yahoo cannot be reached or refuses."""
 	if seconds is None:
 		seconds = config.prefs.cache_seconds
 	now = datetime.datetime.now()
@@ -466,6 +471,8 @@ def fetch_chart_json(ticker, days=3665, seconds=None, full=False):
 		r = net.get(url, headers={**HEADERS, 'Content-type': 'application/json', 'Cookie': cookie}, timeout=config.prefs.http_timeout)
 	except Exception as e:
 		raise RuntimeError(f"Yahoo unreachable for {ticker}: {net.redact(e)}") from e
+	if r.status_code == 404 and not_found(r):
+		raise NoData(f"no price history for {ticker}")
 	if r.status_code != 200:
 		raise RuntimeError(f"{r.status_code} error for {ticker} from Yahoo's chart API")
 	data = r.json()
@@ -473,6 +480,13 @@ def fetch_chart_json(ticker, days=3665, seconds=None, full=False):
 		cache.write(cacheFile, data)
 	return data
 
+def not_found(response):
+	"""Whether a chart API answer is Yahoo saying it has no such symbol, rather than some other 404."""
+	try:
+		return dig(response.json(), 'chart', 'error', 'code') == 'Not Found'
+	except ValueError:
+		return False
+
 def chart_json_to_df(chart_json):
 	"""The chart's daily bars as a DataFrame: Timestamp, Open, High, Low, Close, Volume and the Date."""
 	import pandas as pd
@@ -488,12 +502,12 @@ def chart_json_to_stock_basics(chart_json):
 	return dig(chart_json, 'chart', 'result', 0, 'meta') or {}
 
 def historic_high(ticker):
-	"""(highest high, lowest low) of the ticker's history. Raises RuntimeError when it has none (e.g. total-return indices such
-	as ^SPTTGD carry no High/Low data)."""
+	"""(highest high, lowest low) of the ticker's history. Raises NoData when it has none (e.g. total-return indices
+	such as ^SPTTGD carry no High/Low data)."""
 	df = chart_json_to_df(fetch_chart_json(ticker))
 	df = df[df.High.notnull()].iloc[1:].reset_index(drop=True)
 	if df.empty:
-		raise RuntimeError(f"no high/low history for {ticker}")
+		raise NoData(f"no high/low history for {ticker}")
 	highrow = df.iloc[df['High'].argmax()]
 	lowrow = df.iloc[df['Low'].argmin()]
 	if config.prefs.debug:
diff --git a/tests/common.py b/tests/common.py
index 2d36f05e..3c77f28e 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -10,7 +10,7 @@
 os.environ.update(telegramOutgoingWebhook='https://example.com/telegram', telegramOutgoingToken='tok', telegramChatID='1')
 
 import bot # first, and once: it monkey-patches with gevent, which must happen before requests imports ssl and never mid-test
-from lib import buttons, cache, charts, config, delivery, dialects, formats, inbound, links, markets, render
+from lib import answer, buttons, cache, charts, config, delivery, dialects, formats, inbound, links, markets, render
 from lib import sharesight, shortman, state, telegram, transports, watchlist, words, worker, yahoo
 from lib.reports import about, cal, history, holdings, milestone, performance, price, profile, rankings, rating, reminder, shorts, trades
 
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 12addca5..022c074f 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -12,18 +12,11 @@
 class UpstreamEmpty(FinbotCase):
 	"""Yahoo answers, but with nothing usable. Every command must reply sanely and never crash."""
 
-	ONE_LINE_ERROR = ('.history T1', '.compare T1 T2') # these say so with a single error line when the ticker has no data
-
 	def run_all(self):
-		"""Run every command in COMMANDS: none may crash, and .history and .compare may say so in one error line."""
+		"""Run every command in COMMANDS: each answers (one without data says so, as a reply), and none crashes."""
 		for cmd in COMMANDS:
 			with self.subTest(cmd=cmd):
-				if cmd in self.ONE_LINE_ERROR:
-					self.sent.clear(); self.errors.clear()
-					worker.process_request('telegram', '55', '@u', cmd, BOT, 'U')
-					self.assertLessEqual(len(self.errors), 1, cmd)
-				else:
-					self.command(cmd)
+				self.assertTrue(self.command(cmd), cmd)
 
 	def test_yahoo_returns_empty_dict(self):
 		"""Yahoo answers with no data at all: every command still replies sanely."""
@@ -128,15 +121,20 @@ def test_compare_argument_errors_give_usage(self):
 			with self.subTest(cmd=cmd):
 				self.assertIn('Usage', str(self.command(cmd)))
 
-	def test_compare_with_a_ticker_that_has_no_history(self):
-		""".compare with a ticker that has no history reports one error line."""
-		def no_history(t):
+	def test_no_data_is_an_answer_not_an_error(self):
+		""".compare or .history of a ticker without prices, and a period button with nothing to draw, say so in a plain
+		reply: no error line."""
+		def no_history(t, *a, **k):
 			"""Fail like Yahoo for a ticker with no price history."""
-			raise RuntimeError(f'no price history for {t}')
+			raise yahoo.NoData(f'no price history for {t}')
 		self.patch(yahoo, 'price_series', no_history)
-		self.sent.clear(); self.errors.clear()
-		worker.process_request('telegram', '55', '@u', '.compare T1 T2', BOT, 'U')
-		self.assertEqual(len(self.errors), 1)
+		self.patch(yahoo, 'price_history', no_history)
+		self.assertEqual(self.command('.compare T1 T2'), [('text', ['no price history for T1'])])
+		self.assertEqual(self.command('.history T1'), [('text', ['no price history for T1'])])
+		self.sent.clear()
+		self.patch(telegram, 'answerCallbackQuery', lambda *a, **k: None)
+		worker.process_callback('telegram', 'cb', '55', '9', 'c|h|m|T1')
+		self.assertEqual((self.sent, self.errors), ([('text', ['no price history for T1'])], []))
 
 	def test_history_without_a_ticker_or_with_extra_argument_gives_usage(self):
 		""".history with no ticker or an extra argument gets its usage, not the profile of a ticker called HISTORY."""
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 9a92ce1c..8b468ccc 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -194,6 +194,19 @@ def test_1d_is_the_change_since_the_previous_close(self):
 		self.assertEqual(table['1D'], -3)
 		self.assertGreater(table['5D'], 10) # the week, not the day
 
+	def test_a_symbol_yahoo_does_not_know_is_no_data(self):
+		"""Yahoo's 404 for a symbol it does not know is NoData, an answer; any other 404 is an error, not an answer."""
+		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
+		unknown = {'chart': {'result': None, 'error': {'code': 'Not Found', 'description': 'No data found'}}}
+		self.http(yahoo, Resp(unknown, status=404))
+		with self.assertRaises(yahoo.NoData) as caught:
+			yahoo.price_history('XYZ', 7, graph=False)
+		self.assertEqual(caught.exception.lines, ['no price history for XYZ'])
+		self.http(yahoo, Resp(status=404)) # an endpoint that moved, say
+		with self.assertRaises(RuntimeError) as caught:
+			yahoo.price_history('XYZ', 7, graph=False)
+		self.assertNotIsInstance(caught.exception, answer.Answer)
+
 	def test_a_failed_chart_fetch_raises(self):
 		"""A chart refused (429) or unreachable is a RuntimeError, which callers turn into leaving the ticker out."""
 		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])

From 333914817030df5f14e6cbacab40d9c7e56895df Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 01:38:49 +0000
Subject: [PATCH 24/81] Telegram: period buttons only where their presses
 arrive

A Telegram bot set up for reports only (no telegramOutgoingWebhook)
still put period buttons under its charts, scheduled ones included.
Telegram posts a button press to the bot's webhook, and with none set
up the press went nowhere: the button spun and did nothing. Charts on
such a bot now come without buttons, and with the webhook they keep
them.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/transports.py     |  7 ++++++-
 tests/test_reports.py | 12 ++++++++++++
 2 files changed, 18 insertions(+), 1 deletion(-)

diff --git a/lib/transports.py b/lib/transports.py
index 0cf3dc7e..0c1d3274 100644
--- a/lib/transports.py
+++ b/lib/transports.py
@@ -90,7 +90,12 @@ class TelegramBot(Transport):
 	"""A Telegram chat via the Bot API: HTML, silent, no link previews; charts, buttons, edits, the typing notice."""
 	name = 'telegram'
 	dialect = dialects.TELEGRAM
-	takes_charts = takes_buttons = shows_typing = True
+	takes_charts = shows_typing = True
+
+	@property
+	def takes_buttons(self):
+		"""Buttons only where their presses reach finbot: with telegramOutgoingWebhook, Telegram posts them to it."""
+		return bool(config.env.telegramOutgoingWebhook)
 
 	@staticmethod
 	def markup(buttons):
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 320dd710..86b1611c 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -387,6 +387,18 @@ def test_other_charts_do_not_offer_1d(self):
 			with self.subTest(command=command):
 				self.assertNotIn('1D', [b.lstrip('● ') for b in self.buttons(command)])
 
+	def test_no_buttons_where_their_presses_would_go_nowhere(self):
+		"""Without telegramOutgoingWebhook (reports only), Telegram hears nothing from finbot's buttons, so a chart, a
+		scheduled one included, comes without them."""
+		self.env(webhooks={'telegram': 'https://api.telegram.org/botX/'}, telegramChatID='1')
+		label, kwargs = price.job(['7']) # as scheduled: price.py 7
+		for hook, buttons in (('https://example.com/telegram', True), ('', False)):
+			with self.subTest(telegramOutgoingWebhook=hook):
+				self.env(telegramOutgoingWebhook=hook)
+				self.markups.clear()
+				price.lambda_handler(**kwargs)
+				self.assertEqual(bool(self.markups and self.markups[0]), buttons)
+
 	def press(self, data):
 		"""Press a period button: (the edits made, the toasts shown)."""
 		edits, toasts = [], []

From d08177fc5ebef98934120e9de647d028a5e77036 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 01:42:19 +0000
Subject: [PATCH 25/81] A setup guide per chat network, under doc/

Most installs use one of Telegram, Slack and Discord. Each now has a
guide covering its whole setup, webhook part included: doc/telegram.md,
doc/slack.md and doc/discord.md. Where there is a choice, a table puts
the two options side by side before the steps, so it is clear to the
admin that it is a choice:
- Telegram: reports only, or the interactive bot.
- Slack: incoming webhook, or the interactive bot (which builds on the
  webhook).
- Discord: a webhook only, so the guide says there is no choice.

The guides include the nginx front end for each network, now with only
TLS 1.2 and 1.3, plus two steps the README lacked:
- Telegram: finding the chat id, and the allowed characters of the
  webhook password.
- Slack: inviting the bot to the channel.

The README links the guides under its chat networks table and in its
requirements. Its per-network sections and the interactive bot setup
are gone, moved into the guides. Daemonize (systemd) becomes a section
of its own.

CI treats doc/*.md like README.md: a pull request changing only docs
skips the tests. The tests read only the API samples in doc/'s
subdirectories, which the allowlist does not match. CLAUDE.md says so,
and points to the guides.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 .github/workflows/ci.yml |   3 +-
 CLAUDE.md                |  13 +++--
 README.md                | 111 +++++----------------------------------
 doc/discord.md           |  24 +++++++++
 doc/slack.md             |  97 ++++++++++++++++++++++++++++++++++
 doc/telegram.md          |  88 +++++++++++++++++++++++++++++++
 6 files changed, 232 insertions(+), 104 deletions(-)
 create mode 100644 doc/discord.md
 create mode 100644 doc/slack.md
 create mode 100644 doc/telegram.md

diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index d171e059..080ef50d 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -33,7 +33,8 @@ jobs:
           if [ "$GITHUB_EVENT_NAME" = pull_request ]; then
             files=$(git -c core.quotePath=false diff --name-only --no-renames "origin/$BASE_REF...HEAD")
             echo "$files"
-            if [ -n "$files" ] && ! grep -qvE '^(README\.md|CLAUDE\.md|img/[^/]+\.png)$' <<< "$files"; then
+            docs='^(README\.md|CLAUDE\.md|doc/[^/]+\.md|img/[^/]+\.png)$'
+            if [ -n "$files" ] && ! grep -qvE "$docs" <<< "$files"; then
               docs_only=true
               echo "::notice::Only docs changed: install, compile and tests are skipped"
             fi
diff --git a/CLAUDE.md b/CLAUDE.md
index 74281adc..3ecc63c8 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -18,7 +18,9 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - To merge: `main` takes only a branch that is up to date with it and green on both CI jobs. Merge `origin/main` into the work branch (a merge commit, never a rebase), run the tests, push, wait for CI on that head, then merge the PR with a merge commit.
 
 ## Facts
-- Python 3.13.5; tests are offline and need no network (`tests/`). CI runs them on the latest 3.x and 3.13.5, except for pull requests that change only README.md, CLAUDE.md or `img/*.png` (install, compile and tests are skipped; the allowlist in `.github/workflows/ci.yml` must never include a path a test reads).
+- Python 3.13.5; tests are offline and need no network (`tests/`). CI runs them on the latest 3.x and 3.13.5, except for
+  pull requests that change only README.md, CLAUDE.md, `doc/*.md` or `img/*.png` (install, compile and tests are
+  skipped; the allowlist in `.github/workflows/ci.yml` must never include a path a test reads).
 - Production runs the Debian system packages (including `python3-croniter`, for the scheduler in `lib/jobs.py`); `requirements.txt` is for pip/uv users and CI.
 - A new preference goes in five places: a `Prefs` field and its line in `preference_values` (`lib/config.py`), `defaults/preferences.conf`, the README's preferences table, and `DEFAULTS` plus the key count in `tests/test_config.py`. The tests catch all but the README (no test may read README.md: CI skips the tests for a README-only change).
 - If Yahoo and Sharesight are unreachable (the cloud environment's network policy can block them), check behaviour
@@ -66,8 +68,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - A report is a list of lines. `service` is a dialect (`lib/dialects.py`), which equals the network's name: markup with
   `service.bold(text)`, `.italic`, `.strike`, `.link(url, text)`, user-written text through `service.escape(text)`, stock links
   through `links.finance_link(symbol, exchange, service)`. Tests pass `TELEGRAM`, `SLACK`, `DISCORD`.
-- A command that cannot give what was asked (no such portfolio, no prices) raises `answer.Answer(*lines)`, which the chat
-  gets as a plain reply; `yahoo.NoData` is one. Any other exception becomes a one-line ⚠️ error report.
+- A command that cannot give what was asked (no such portfolio, no prices) raises `answer.Answer(*lines)`, which the
+  chat gets as a plain reply; `yahoo.NoData` is one. Any other exception becomes a one-line ⚠️ error report.
 - Every function and method has a docstring, tests included (a test checks): one line where it fits, reworded rather
   than spilling a few words onto a second. Tabs everywhere (`.editorconfig`). New and changed lines stay within 120
   columns, tabs counting 4 (an edit quotes whole lines).
@@ -80,5 +82,6 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 ## Saving tokens
 - Run the tests quietly: `.venv/bin/python -m unittest discover -s tests -b` prints only failures (about 0.4 KB when green). One
   module: `.venv/bin/python -m unittest tests.test_reports -b`.
-- README.md is 38 KB: grep for the section instead of reading it whole. `doc/` holds large API samples that the tests replay:
-  inspect them with python, don't read them.
+- README.md is 38 KB: grep for the section instead of reading it whole. Each network's setup is in its own guide,
+  `doc/telegram.md`, `doc/slack.md` and `doc/discord.md`: a change to a network's setup or keys goes there. `doc/yahoo/`
+  and `doc/sharesight/` hold large API samples that the tests replay: inspect them with python, don't read them.
diff --git a/README.md b/README.md
index cbc9ffad..f5e424ce 100644
--- a/README.md
+++ b/README.md
@@ -36,12 +36,16 @@ What finbot can do on each network. Telegram takes everything; the others take w
 | Linked tickers | ✓ | ✓ | ✓ |
 | Charts updated in place by their buttons | ✓ | ✗ | ✗ |
 
+**Setting up a network:** [Telegram](doc/telegram.md) · [Slack](doc/slack.md) · [Discord](doc/discord.md). Each guide
+covers the whole setup of that network, and where there is a choice between reports only and the interactive bot, it
+compares the two before the steps.
+
 ![Screenshot of showing trade notifications on Slack](img/screenshot.png?raw=true "Screenshot showing trade notifications on Slack")
 
 ### Requirements
 * Most features require a Sharesight paid plan, preferably with automatic trade imports, and an API key
-* Discord/Slack webhooks / Telegram bot user
-* Interactive commands require a public web server with domain name and matching certificate to proxy chat commands to the bot.
+* A Telegram bot, a Slack app or a Discord webhook (see the network's guide, above)
+* Interactive commands require a public web server with domain name and matching certificate to proxy chat commands to the bot (see the [Telegram](doc/telegram.md#interactive-bot) and [Slack](doc/slack.md#interactive-bot) guides).
 * Python 3.13.5 (the version in `.python-version`; 3.12 is the minimum)
 * Python modules, pinned to versions with 3.13 support in `requirements.txt`:
 ```
@@ -450,7 +454,8 @@ sudo cp ~/finbot/defaults/finbot.env /opt/finbot.env && sudo chmod 600 /opt/finb
 * **init.d / shell:** `util/check_config.py /opt/finbot.env && set -a && . /opt/finbot.env && set +a && ./bot.py` (`set -a` is what exports the values to the bot).
 * **Scheduled jobs:** they are started by `bot.py` (see [Scheduling](#scheduling)), so they run on the same settings. For a one-off run, see [Running a report by hand](#scheduling).
 
-The startup keys, read by your launcher. A change needs a restart. Blank means unset.
+The startup keys, read by your launcher. A change needs a restart. Blank means unset. Each network's guide says which
+of the chat network keys it needs.
 
 | Key | Default | Description |
 |---|---|---|
@@ -541,36 +546,10 @@ sharesight_client_id=01d692d4de7 samplefc64bc2e2f01d692d4de72986ea808f6e99813f
 sharesight_client_secret=01d692d4de7 samplefc64bc2e2f01d692d4de72986ea808f6e99813f
 ```
 
-### Discord
-A Discord webhook takes the trade notifications and scheduled reports, with their charts attached and without link previews.
-Provision it in Discord under _Server Settings > Integrations > Webhooks_.
-Example:
-```
-discord_webhook=https://discord.com/api/webhooks/1009998000000000000/aaaaaaaaaaaaaaaaaaaaaaaaaa-bbbbbbbbbbbbbb
-```
-A URL ending in `/slack`, from older setups, works as it is: finbot uses Discord's own endpoint either way.
-
-### Slack
-A Slack webhook can be provisioned by creating a new app from scratch at https://api.slack.com/apps/ then navigating to _Incoming Webhooks_. Once the link is generated, add it to `finbot.env`. Example:
-```
-slack_webhook=https://hooks.slack.com/services/XXXXXXXXXXX/YYYYYYYYYYY/AAAAAAAAmockupAAAAAAAAAAAA
-```
-The webhook is only used for the trade notifications and scheduled reports. The interactive bot requires further configuration of this Slack app (see below).
-
-### Telegram
-* Set up the bot by messaging [BotFather](https://telegram.me/BotFather).
-* Add your bot to a group or channel.
-* For security, run `/setjoingroups` and set to `Disabled`
-* Optionally, make the bot a group admin (for interactive features).
-* In `finbot.env`, set `telegramBotToken` to the token BotFather gave you.
-* In `finbot.env`, set `telegramChatID` to the chat group or channel id.
-   * For channels and supergroups, `telegramChatID` should be negative and 13 characters. Prepend `-100` if necessary.
-   * Be aware a group id can change if you edit group settings and it becomes a "supergroup". Currently, the bot does not automatically handle this.
-* Example `finbot.env` entry:
-```
-telegramBotToken=0000000000:AAAAAAAAAAAAAAAAAAAAAAAAAA
-telegramChatID=-1001000000000
-```
+### Chat networks setup
+Follow the guide for each network you use: [Telegram](doc/telegram.md), [Slack](doc/slack.md) or
+[Discord](doc/discord.md). It covers the bot or webhook, its `finbot.env` keys, and for chat commands the HTTPS server in
+front of finbot.
 
 ### Portfolios
 Portfolios are auto-discovered, including other people's portfolios which are shared to you. To exclude specific portfolios, add their IDs to `exclude_portfolios` in the `preferences.conf`:
@@ -669,71 +648,7 @@ sudo systemd-run --pty --wait --collect -p DynamicUser=yes -p StateDirectory=fin
 ```
 The script prints its output on your terminal and sends its report to your chat channels as usual. The transient unit gets its own dynamic user, and systemd may re-own the state directory to it for the run: check `ls -ln /var/lib/private/finbot` afterwards and `systemctl restart finbot` if the bot can no longer write there. The chat commands (`.price 7d`, `.performance 7d`, ...) run the same reports inside the bot, without any of this. Outside systemd, run a script from the checkout as your own user with the settings loaded into the shell: `set -a; . /opt/finbot.env; set +a; ./price.py 7` (this works while no value in the file contains spaces).
 
-## Interactive bot setup
-The backend `bot.py` needs a frontend https server on a valid domain name with a valid x509 certifcate.
-It defaults to listening on http://127.0.0.1:5000/, which can be changed with `ip` and `port` in `finbot.env`.
-
-Example frontend https server config (nginx):
-```
-server {
-	listen 8443 ssl;
-	deny all;
-
-	server_name         www.example.com;
-	ssl_certificate     /etc/letsencrypt/live/www.example.com/fullchain.pem;
-	ssl_certificate_key /etc/letsencrypt/live/www.example.com/privkey.pem;
-	ssl_protocols       TLSv1 TLSv1.1 TLSv1.2;
-	ssl_ciphers         HIGH:!aNULL:!MD5;
-
-	location /slack {
-		proxy_pass http://127.0.0.1:5000/slack;
-                include /etc/nginx/aws_subnets;
-                deny all;
-	}
-	location /telegram {
-		proxy_pass http://127.0.0.1:5000/telegram;
-		include /etc/nginx/telegram_subnets;
-		deny all;
-	}
-}
-```
-Note: The utils folder contains scripts to generate `/etc/nginx/aws_subnets` and `/etc/nginx/telegram_subnets`. They can be run once, or placed in `/etc/cron.daily/`.
-
-### Integrating the interactive bot with chat networks
-
-#### Telegram
-* Message BotFather to create a bot user.
-  * Set `telegramBotToken` in `finbot.env` to the token given by BotFather.
-  * Add your Telegram bot user to a group
-    * Give the bot group admin access so it can read the group chat.
-    * For security, message BotFather and disable `/setjoingroups`.
-* Set `telegramOutgoingWebhook` in `finbot.env` to your web server (e.g. https://www.example.com:8443/telegram).
-* Set `telegramOutgoingToken` to a password of your choosing. Telegram will use this to authenticate with finbot.
-With these options set, your bot will auto-subscribe your URL to events the bot sees, when you run `bot.py`. It unregisters and registers the webhook as it starts (and only then, which also wakes up a stale connection); if Telegram refuses, finbot does not start and the log says why.
-
-#### Slack
-Visit https://api.slack.com/apps/ to create a new Slack app from scratch (if you already created one for a finbot webhook, you can reuse that app). Charts in replies are uploaded into the channel with Slack's file upload API (`files.getUploadURLExternal`, then `files.completeUploadExternal`, which replaced the retired `files.upload`). Scheduled reports reach Slack through the incoming webhook, which takes text only, so they come without charts. If an upload fails (a missing scope, say), the reply goes out as text and the log says why.
-* From _Basic Information > App Credentials_, copy _Signing Secret_ into `finbot.env` as `slackSigningSecret`. finbot checks Slack's signature on every request with it, and turns away a request signed more than five minutes earlier (a replay, or a server clock that is wrong).
-  * The older _Verification Token_, set as `slackOutgoingToken`, still works without a signing secret, but Slack has deprecated it: the token travels in every request, where the signing secret never does.
-* In `finbot.env`, put your web server URL (e.g. https://www.example.com:8443/slack) into `slackOutgoingWebhook`
-* Save `finbot.env` and restart `bot.py`
-* Under _Event Subscriptions_:
-  * Go to _Enable Events > On > Request URL_ and enter your web server URL (e.g. https://www.example.com:8443/slack). At this time, if the bot is reachable, it will auto respond to Slack's verification request.
-  * Go down to _Subscribe to bot events_ and add event `app_mention` for the bot to see _@botname_ mentions.
-    * Alternatively, you can subscribe to `message.channels` if you want your bot to see everything and respond to `.` commands. To avoid duplicate responses, don't subscribe to `app_mention` and `message.channels` at the same time.
-    * If you want to DM the bot:
-      * Subscribe to `message.im`
-      * Check the box _App Home > Allow users to send Slash commands and messages from the messages tab_
-  * Save Changes
-* Under _OAuth & Permissions_:
-  * Scroll down to _Scopes > Bot Token Scopes_, and add: `chat:write`, `files:write` and `files:read` (Slack's file upload API needs both `files` scopes).
-    * Adding scopes to an app that is already installed takes effect only after _Reinstall to Workspace_.
-  * Scroll up to _OAuth Tokens for Your Workspace_:
-    * If _Bot User OAuth Token_ is not visible, hit _Install to Workspace > Allow_
-    * Copy _Bot User OAuth Token_ into `finbot.env` as `slackBotToken`. This allows finbot to authenticate with Slack so it can send replies. Without it Slack commands get no reply: the log says so at startup, and again for each reply.
-    * Restart `bot.py`
-
-### Daemonize (systemd)
+## Daemonize (systemd)
 `finbot.service` can take care of keeping `bot.py` running in the background and starting it on boot. Copy `finbot.service` to `/etc/systemd/system/`, edit it to set the `User`, `ExecStart` and the path of [`finbot.env`](#setup) (`EnvironmentFile=`, and the `ExecStartPre=` check), then enable and start it. `systemctl reload finbot` re-reads `preferences.conf` and the [schedule](#scheduling) without a restart; a change to `finbot.env` needs `systemctl restart finbot`:
 
 ```
diff --git a/doc/discord.md b/doc/discord.md
new file mode 100644
index 00000000..82db2477
--- /dev/null
+++ b/doc/discord.md
@@ -0,0 +1,24 @@
+# Discord
+
+finbot reaches Discord through a webhook, which posts the scheduled reports into a channel. A webhook only posts, so
+there is no choice to make: on Discord, finbot does not answer commands. For commands, set up finbot on
+[Telegram](telegram.md) or [Slack](slack.md) as well.
+
+What the webhook carries:
+* the scheduled reports: trades, price alerts, earnings and dividend dates, performance and the rest;
+* their charts, attached to the messages;
+* linked tickers, without link previews, and without pinging anyone.
+
+## Webhook
+
+1. In Discord, open _Server Settings > Integrations > Webhooks_ and choose _New Webhook_.
+2. Name it (finbot, say), pick the channel for the reports, and _Copy Webhook URL_.
+3. Put the URL in `finbot.env` as `discord_webhook`.
+4. Restart finbot (`sudo systemctl restart finbot`). The scheduled reports now arrive in the channel.
+
+```
+discord_webhook=https://discord.com/api/webhooks/1009998000000000000/aaaaaaaaaaaaaaaaaaaaaaaaaa-bbbbbbbbbbbbbb
+```
+
+A URL ending in `/slack`, from older setups, works as it is: finbot uses Discord's own endpoint either way. A message
+holds up to 2,000 characters, so a long report arrives as several messages.
diff --git a/doc/slack.md b/doc/slack.md
new file mode 100644
index 00000000..aca78b10
--- /dev/null
+++ b/doc/slack.md
@@ -0,0 +1,97 @@
+# Slack
+
+finbot reaches Slack through a Slack app of your own. The app's incoming webhook posts the scheduled reports into a
+channel; the app's bot also answers commands. The bot builds on the webhook, which still carries the scheduled reports.
+
+## Incoming webhook, or the interactive bot?
+
+| | Incoming webhook | Interactive bot (with the webhook) |
+|---|---|---|
+| Scheduled reports: trades, price alerts, earnings and dividend dates, performance and the rest | ✓, as text | ✓, as text, through the webhook |
+| Chat commands: `.price`, `@finbot price` and the rest ([the list](../README.md#supported-commands)) | ✗ | ✓ |
+| Charts in the replies to commands | n/a | ✓, uploaded into the channel |
+| Period buttons and "typing…" | ✗ | ✗: Slack has no typing notice for bots, and finbot has no Slack buttons yet |
+| Public HTTPS server | Not needed | Needed: a domain name with a valid certificate |
+| `finbot.env` keys | `slack_webhook` | That one, plus `slackOutgoingWebhook`, `slackSigningSecret` and `slackBotToken` |
+
+Start with [Incoming webhook](#incoming-webhook). For commands, go on to [Interactive bot](#interactive-bot)
+afterwards, in the same app.
+
+## Incoming webhook
+
+1. Create the app: at https://api.slack.com/apps choose _Create New App > From scratch_, name it (finbot, say) and pick
+   your workspace.
+2. Under _Features > Incoming Webhooks_, turn _Activate Incoming Webhooks_ on, then _Add New Webhook to Workspace_ and
+   pick the channel for the reports.
+3. Copy the webhook URL into `finbot.env` as `slack_webhook`.
+4. Restart finbot (`sudo systemctl restart finbot`). The scheduled reports now arrive in the channel.
+
+```
+slack_webhook=https://hooks.slack.com/services/XXXXXXXXXXX/YYYYYYYYYYY/AAAAAAAAmockupAAAAAAAAAAAA
+```
+
+An incoming webhook takes text only, so the scheduled reports come without charts.
+
+## Interactive bot
+
+Slack delivers the messages that mention the bot (or every message in a channel, as you choose) to finbot over HTTPS.
+finbot listens on `http://127.0.0.1:5000` (`ip` and `port` in `finbot.env`), so it needs a web server in front of it,
+on a domain name with a valid certificate.
+
+1. Set up the [incoming webhook](#incoming-webhook), above.
+2. Put finbot behind an HTTPS server. An nginx example, which lets in only Amazon's addresses, where Slack's requests
+   come from:
+   ```
+   server {
+   	listen 8443 ssl;
+   	deny all;
+
+   	server_name         www.example.com;
+   	ssl_certificate     /etc/letsencrypt/live/www.example.com/fullchain.pem;
+   	ssl_certificate_key /etc/letsencrypt/live/www.example.com/privkey.pem;
+   	ssl_protocols       TLSv1.2 TLSv1.3;
+
+   	location /slack {
+   		proxy_pass http://127.0.0.1:5000/slack;
+   		include /etc/nginx/aws_subnets;
+   		deny all;
+   	}
+   }
+   ```
+   `util/nginx_aws_cron` writes `/etc/nginx/aws_subnets`, an `allow` line for each of Amazon's networks, and reloads
+   nginx. Run it once as root before nginx loads this configuration, or put it in `/etc/cron.daily/` to keep the list
+   current.
+3. In the app's _Basic Information > App Credentials_, copy the _Signing Secret_ into `finbot.env` as
+   `slackSigningSecret`. finbot checks Slack's signature on every request with it, and turns away a request signed more
+   than five minutes earlier: a replay, or a server clock that is wrong.
+   * The older _Verification Token_, set as `slackOutgoingToken`, still works in place of the signing secret, but
+     Slack has deprecated it: the token travels in every request, where the signing secret never does.
+4. In `finbot.env`, set `slackOutgoingWebhook` to the URL Slack should post to, and restart finbot, so that it can
+   answer Slack's check of the URL in the next step.
+5. Under _Event Subscriptions_:
+   * Turn _Enable Events_ on, and enter the same URL as the _Request URL_. Slack checks it at once, and finbot answers.
+   * Under _Subscribe to bot events_, add `app_mention`: the bot sees messages that mention it, such as
+     `@finbot price`.
+     * Or add `message.channels` instead, for the bot to see every message in its channels and answer the commands
+       that start with `.`. Not both: the bot would answer a mention twice.
+     * For direct messages to the bot, add `message.im`, and tick _App Home > Allow users to send Slash commands and
+       messages from the messages tab_.
+   * _Save Changes_.
+6. Under _OAuth & Permissions > Scopes > Bot Token Scopes_, add `chat:write`, `files:write` and `files:read`. The upload
+   of a chart needs both `files` scopes.
+7. At the top of _OAuth & Permissions_, _Install to Workspace_ (or _Reinstall to Workspace_ when you changed scopes of
+   an installed app), and copy the _Bot User OAuth Token_ into `finbot.env` as `slackBotToken`. finbot replies with it;
+   without it, commands get no reply, and the log says so at startup and for each reply.
+8. Restart finbot, and invite the bot to the channel: `/invite @finbot`.
+9. Send `@finbot help` in the channel (or `.help`, with `message.channels`).
+
+```
+slack_webhook=https://hooks.slack.com/services/XXXXXXXXXXX/YYYYYYYYYYY/AAAAAAAAmockupAAAAAAAAAAAA
+slackOutgoingWebhook=https://www.example.com:8443/slack
+slackSigningSecret=0123456789abcdef0123456789abcdef
+slackBotToken=xoxb-your-bot-token
+```
+
+A chart in a reply is uploaded into the channel with Slack's file upload API (`files.getUploadURLExternal`, then
+`files.completeUploadExternal`). If an upload fails (a missing scope, say), the reply goes out as text, and the log says
+why.
diff --git a/doc/telegram.md b/doc/telegram.md
new file mode 100644
index 00000000..82520fc2
--- /dev/null
+++ b/doc/telegram.md
@@ -0,0 +1,88 @@
+# Telegram
+
+finbot talks to Telegram as a bot user that you create with BotFather. The same bot can post the scheduled reports
+only, or also answer commands in your chats: the second builds on the first.
+
+## Reports only, or the interactive bot?
+
+| | Reports only | Interactive bot |
+|---|---|---|
+| Scheduled reports: trades, price alerts, earnings and dividend dates, performance and the rest | ✓ | ✓ |
+| Charts with the reports | ✓ | ✓ |
+| Chat commands: `.price`, `.history`, `.performance` and the rest ([the list](../README.md#supported-commands)) | ✗ | ✓ |
+| Period buttons under charts, and `.jobs` buttons | ✗ | ✓ |
+| "typing…" while a command works | ✗ | ✓ |
+| A keyboard of the common commands in a direct chat with the bot | ✗ | ✓ |
+| Public HTTPS server | Not needed | Needed: a domain name with a valid certificate |
+| `finbot.env` keys | `telegramBotToken`, `telegramChatID` | Those two, plus `telegramOutgoingWebhook` and `telegramOutgoingToken` |
+
+Start with [Reports only](#reports-only). For commands, go on to [Interactive bot](#interactive-bot) afterwards.
+
+## Reports only
+
+1. Create the bot: message [@BotFather](https://telegram.me/BotFather), send `/newbot` and answer its questions. It
+   replies with the bot's token: put it in `finbot.env` as `telegramBotToken`.
+2. Add the bot to the group or channel the reports should go to. In a channel, make it an administrator, so that it can
+   post.
+3. Stop strangers adding your bot to their groups: send BotFather `/setjoingroups`, pick the bot and choose `Disable`.
+   Do this after step 2, as it stops you adding the bot to groups too.
+4. Put the chat's id in `finbot.env` as `telegramChatID`.
+   * A group's id is negative. A channel's or a supergroup's is negative and 13 characters long: prepend `-100` if
+     yours is shorter.
+   * To find it, send `/start@<your bot's username>` in the chat, then open
+     `https://api.telegram.org/bot<telegramBotToken>/getUpdates` in a browser and look for `"chat":{"id":`. This works
+     only while the bot has no webhook, so before you set up the interactive bot.
+   * A group's id changes when Telegram turns it into a supergroup, which some changes to its settings do. Update
+     `telegramChatID` then: finbot does not follow the change.
+5. Restart finbot (`sudo systemctl restart finbot`). The scheduled reports now arrive in the chat.
+
+```
+telegramBotToken=0000000000:AAAAAAAAAAAAAAAAAAAAAAAAAA
+telegramChatID=-1001000000000
+```
+
+## Interactive bot
+
+Telegram delivers the messages in your chats, and each button press, to finbot over HTTPS. finbot listens on
+`http://127.0.0.1:5000` (`ip` and `port` in `finbot.env`), so it needs a web server in front of it, on a domain name
+with a valid certificate. Telegram posts only to ports 443, 80, 88 and 8443.
+
+1. Set up the bot for [reports only](#reports-only), above.
+2. Let the bot read the chat: make it an administrator of the group. A bot that is not an administrator sees only the
+   commands that start with `/`, and finbot's start with `.`. Alternatively, send BotFather `/setprivacy` and choose
+   `Disable`.
+3. Put finbot behind an HTTPS server. An nginx example, which lets in only Telegram's addresses:
+   ```
+   server {
+   	listen 8443 ssl;
+   	deny all;
+
+   	server_name         www.example.com;
+   	ssl_certificate     /etc/letsencrypt/live/www.example.com/fullchain.pem;
+   	ssl_certificate_key /etc/letsencrypt/live/www.example.com/privkey.pem;
+   	ssl_protocols       TLSv1.2 TLSv1.3;
+
+   	location /telegram {
+   		proxy_pass http://127.0.0.1:5000/telegram;
+   		include /etc/nginx/telegram_subnets;
+   		deny all;
+   	}
+   }
+   ```
+   `util/nginx_telegram_cron` writes `/etc/nginx/telegram_subnets`, an `allow` line for each of Telegram's networks,
+   and reloads nginx. Run it once as root before nginx loads this configuration, or put it in `/etc/cron.daily/` to
+   keep the list current. `util/iptables_telegram_cron` goes further, in the firewall: of the traffic to ports 443 and
+   8443 it lets in only Telegram's networks on 8443 and drops the rest, so nothing else can be served on those ports.
+4. In `finbot.env`, set `telegramOutgoingWebhook` to the URL Telegram should post to, and `telegramOutgoingToken` to a
+   password of your choosing: 1 to 256 characters, of letters, digits, `_` and `-`. Telegram sends the password with
+   every update, and finbot turns away a request without it.
+5. Restart finbot. As it starts, it registers the URL with Telegram, unregistering it first, which also wakes up a stale
+   connection. If Telegram refuses, finbot does not start, and the log says why.
+6. Send `.help` in the chat. In a direct chat with the bot, the reply also brings a keyboard of the common commands.
+
+```
+telegramBotToken=0000000000:AAAAAAAAAAAAAAAAAAAAAAAAAA
+telegramChatID=-1001000000000
+telegramOutgoingWebhook=https://www.example.com:8443/telegram
+telegramOutgoingToken=a-long-random-password
+```

From bb371af2b6648ed3bf3a64c15b0717a38c8e9c98 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 01:55:03 +0000
Subject: [PATCH 26/81] Change figures signed everywhere (D13)

A change figure now gets its sign everywhere, through
formats.signed_percent. Before, the same history figure read +3% in
.history but 3% in a profile.
- The profile's history lines and its pre/post-market line now read
  +3% where they read 3%.
- The forecast growth lines and .compare's caption signed theirs by
  hand; they use the helper now, and thousands get separators.
Levels (dividend yield, short interest, debt ratio, pie shares) stay
unsigned.

The output comparison differs only in the profile's 1Y line
("1Y: 3%" now "1Y: +3%"), on Telegram and Slack.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/reports/history.py | 2 +-
 lib/reports/profile.py | 8 +++++---
 tests/test_reports.py  | 6 ++++++
 3 files changed, 12 insertions(+), 4 deletions(-)

diff --git a/lib/reports/history.py b/lib/reports/history.py
index b8b2569e..e9c44a81 100644
--- a/lib/reports/history.py
+++ b/lib/reports/history.py
@@ -77,7 +77,7 @@ def prepare_compare(service, args):
 	results = sorted(((series.iloc[-1] / series.iloc[0] - 1) * 100, t) for t, series in cropped)
 	for pct, t in reversed(results):
 		link = links.finance_link(t, exchanges[t], service, days=span_days, brief=False)
-		caption.append(f"{formats.get_emoji(pct)} {names[t]} ({link}): {pct:+,.1f}%")
+		caption.append(f"{formats.get_emoji(pct)} {names[t]} ({link}): {formats.signed_percent(pct, 1)}")
 	if config.prefs.cache:
 		cache.write_bytes(image_cache_file, image)
 		cache.write(caption_cache_file, caption)
diff --git a/lib/reports/profile.py b/lib/reports/profile.py
index 49fdeb77..8d198a98 100644
--- a/lib/reports/profile.py
+++ b/lib/reports/profile.py
@@ -144,7 +144,7 @@ def prepare_profile_payload(service, user, ticker):
 	for key, label in (('revenueEstimateY', 'Revenue forecast (1Y):'), ('earningsEstimateY', 'Earnings forecast (1Y):')):
 		if key in data:
 			growth = int(round(data[key]))
-			payload.append(service.bold(label) + (f" +{growth}% " if growth > 0 else f" {growth}% ⚠️ "))
+			payload.append(service.bold(label) + f" {formats.signed_percent(growth)} " + ('' if growth > 0 else '⚠️ '))
 	if 'insiderBuy' in data:
 		if data['insiderBuy'] > data['insiderSell']:
 			payload.append(service.bold("Net insider action (3M):") + f" Buy {currency} {formats.humanUnits(data['insiderBuyValue'])} ")
@@ -193,11 +193,13 @@ def prepare_profile_payload(service, user, ticker):
 	for interval in ('5Y', '1Y', '1M', '1D'):
 		if interval in price_history:
 			percent = price_history[interval]
-			payload.append(f"{formats.get_emoji(percent)} {service.bold(interval + ':')} {percent}%")
+			label = service.bold(interval + ':')
+			payload.append(f"{formats.get_emoji(percent)} {label} {formats.signed_percent(percent)}")
 	if state != 'REGULAR':
 		for key, label in (('percent_change_premarket', 'Pre-market:'), ('percent_change_postmarket', 'Post-market:')):
 			if key in data:
-				payload.append(f"{formats.get_emoji(data[key])} {service.bold(label)} {data[key]:,}%")
+				change = data[key]
+				payload.append(f"{formats.get_emoji(change)} {service.bold(label)} {formats.signed_percent(change)}")
 				break
 
 	if 'profile_website' in data and config.prefs.hyperlinkFooter and config.prefs.hyperlink:
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 86b1611c..14b23c38 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -195,6 +195,12 @@ def test_a_profile_leaves_out_what_yahoo_did_not_send(self):
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: full | {'shareholder_equity': 0}})
 		self.assertNotIn('Net debt/equity', str(profile.prepare_profile_payload(TELEGRAM, '@u', 'T1')))
 
+	def test_a_profile_signs_its_changes(self):
+		"""A profile signs its price changes, as .history does: +3% over a year, +1.5% pre-market."""
+		profile = '\n'.join(self.command('.T13')[0][1]) # T13 is pre-market, up 1.5%
+		self.assertIn('<b>1Y:</b> +3%', profile)
+		self.assertIn('<b>Pre-market:</b> +1.5%', profile)
+
 	def test_a_profile_warns_of_a_ps_ratio_only_above_15(self):
 		"""The PS ratio line warns only above 15; the forward P/E's warning used to show on it too."""
 		for ps, warned in ((12.3, False), (15.1, True)): # forward P/E 99 warns (above 30): its warning used to show on the PS line too

From 234ce06bd5164174f414836fc9becd9cc58b0d1f Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 01:59:04 +0000
Subject: [PATCH 27/81] The EOFY reminder moves into reminder.json (D12)

The Australian end of financial year checklist (23 June) was written
in reminder.py. It is now an AU-gated entry in defaults/reminder.json,
like the other reminders.

reminder.json entries gain what the checklist needs:
- a message may be a list of lines as well as one line;
- [text](url) is a link, in each network's own markup;
- a line wrapped in _underscores_ is in italics;
- the rest is text, now escaped for the network (a & in a reminder
  used to reach Telegram's HTML raw).

reminder.py splits into report(service, today, entries), a pure
function, and lambda_handler(), which reads the clock and the file and
delivers. Only the birthday line stays computed.

Checked old against new: for six dates (the EOFY day, the birthday,
gated, dated and ordinary days), in AU and US, on all three networks,
the output is identical. The README documents the reminder.json
format.

An install that keeps its own reminder.json in the data directory
takes that file instead of the default, so it no longer gets the EOFY
checklist until it adds the entry; util/sync_defaults.py shows the
difference.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 README.md               |   2 +
 defaults/reminder.json  |  18 +++++++
 lib/reports/reminder.py | 108 +++++++++++++++++-----------------------
 tests/test_reports.py   |  32 ++++++++++++
 4 files changed, 97 insertions(+), 63 deletions(-)

diff --git a/README.md b/README.md
index f5e424ce..1489a842 100644
--- a/README.md
+++ b/README.md
@@ -534,6 +534,8 @@ util/sync_defaults.py --apply  # copy missing files (including a starter prefere
 util/sync_defaults.py --force  # also replace changed files, keeping .bak copies
 util/sync_defaults.py --dest /srv/finbot --apply  # target another directory instead of the data directory
 ```
+`reminder.json` holds the reminders that `reminder.py` posts on their day. Each entry has a `month_day` (`06-30`) and a `message`, which is one line or a list of lines, and may name a `year` (`2026`) and a `gate_country` (posted only when `country_code` in `preferences.conf` matches; `{flag}` in its message becomes that country's flag). A message is text, with two kinds of markup that each network gets in its own form: `[text](https://example.com)` is a link, and a line wrapped in underscores, `_like this_`, is in italics. The Australian end of financial year checklist (23 June) is such an entry: if you keep your own `reminder.json` in the data directory, `util/sync_defaults.py` shows what it lacks.
+
 With systemd `DynamicUser=yes` and `StateDirectory=finbot`, `DATA_DIR` defaults to systemd's `$STATE_DIRECTORY` (`/var/lib/finbot`), so no setting is needed; move existing data once with `cp -a /opt/finbot/var/. /var/lib/finbot/`. An explicit `DATA_DIR` always wins.
 
 In Docker, one container is enough: `bot.py` runs the schedule too, so there is no cron or scheduler sidecar to set up.
diff --git a/defaults/reminder.json b/defaults/reminder.json
index f6168f83..1b9d05e0 100644
--- a/defaults/reminder.json
+++ b/defaults/reminder.json
@@ -7,6 +7,24 @@
     "month_day": "06-12",
     "message": "SPCX IPO anniversary"
   },
+  {
+    "month_day": "06-23",
+    "gate_country": "AU",
+    "message": [
+      "💸 Realise capital gains/losses by EOFY June 30{flag}",
+      "",
+      "💰 Finalise any deductables by EOF June 30: {flag}",
+      "• donations*",
+      "• Super contributions†",
+      "• work expenses‡",
+      "• investment subscriptions§",
+      "",
+      "_* Organisation must be [DGR-registered](https://abr.business.gov.au/Tools/DgrListing)_",
+      "_† Allow transfer time for Super, as contributions can only be deducted for the year they're received. Stake users: see [note about XX PCC in transaction description](https://hellostake.com/au/support/stake-super/employer-and-personal-contributions/33442327233305#h_01JQAPN9A6ZM41NZKP6XZWN1KP)_",
+      "_‡ See [eligible work expenses](https://www.ato.gov.au/individuals-and-families/income-deductions-offsets-and-records/deductions-you-can-claim)_",
+      "_§ Must relate directly to specific trades and gains/income (e.g. not general finance news)_"
+    ]
+  },
   {
     "month_day": "06-30",
     "message": "🥳 Happy EOFY 🇦🇺 🇧🇩 🇪🇬 🇪🇹 🇰🇪 🇳🇵 🇵🇰"
diff --git a/lib/reports/reminder.py b/lib/reports/reminder.py
index a1c1865e..0dea764c 100644
--- a/lib/reports/reminder.py
+++ b/lib/reports/reminder.py
@@ -1,77 +1,59 @@
-"""Reminders for the day: those in reminder.json, and the Australian end of financial year (scheduled: reminder.py)."""
+"""Reminders for the day: the entries of reminder.json for today, and finbot's birthday (scheduled: reminder.py)."""
 import datetime
+import re
 import zoneinfo
 from dateutil.relativedelta import relativedelta
 from lib import config
 from lib import formats
 from lib import delivery
 
+BIRTHDAY = datetime.date(2022, 8, 18)
+FLAGS = {'AU': '🇦🇺'} # what {flag} becomes in an entry gated to a country
+LINK = re.compile(r'\[([^\]]+)\]\(([^)\s]+)\)') # [text](url)
+
 
 def lambda_handler():
 	"""Post today's reminders to the default chats, if there are any."""
-	def prepare_finance_calendar_payload(service):
-		"""Today's reminders in one network's markup: those in reminder.json for this day (and year, and country,
-		when given), the Australian end of financial year on 23 June, and finbot's birthday."""
-		payload = []
-		tz = zoneinfo.ZoneInfo(config.prefs.timezone)
-		localtime = datetime.datetime.now(tz)
-
-		month_and_day = str(localtime.strftime('%m-%d')) # 09-20
-		year = str(localtime.strftime('%Y')) # 2023
-
-		# --- Simple, static one-line reminders loaded from reminders.json ---
-		for reminder in config.data_json("reminder.json", []):
-			if reminder.get("month_day") != month_and_day:
-				continue
-			if reminder.get("year") and reminder["year"] != year:
-				continue
-			gate_country = reminder.get("gate_country")
-			if gate_country and config.prefs.country_code != gate_country:
-				continue
-			message = reminder["message"]
-			if "{flag}" in message:
-				# Flag only ever applies to the AU-gated set today;
-				# extend this mapping if other gated countries are added.
-				flag = "🇦🇺" if gate_country == "AU" else ""
-				message = message.replace("{flag}", flag)
-			payload.append(message)
-
-		# --- Complex, dynamically-computed reminders (kept in code) ---
-		if config.prefs.country_code == 'AU':
-			flag = "🇦🇺"
-			if month_and_day == '06-23':
-				payload.append("💸 Realise capital gains/losses by EOFY June 30" + flag)
-				payload.append ("")
-				payload.append("💰 Finalise any deductables by EOF June 30: " + flag)
-				payload.append("• donations*")
-				payload.append("• Super contributions†")
-				payload.append("• work expenses‡")
-				payload.append("• investment subscriptions§")
-				payload.append ("")
-				abr_link = service.link('https://abr.business.gov.au/Tools/DgrListing', 'DGR-registered')
-				stake_note = service.link('https://hellostake.com/au/support/stake-super/employer-and-personal-contributions/33442327233305#h_01JQAPN9A6ZM41NZKP6XZWN1KP', 'note about XX PCC in transaction description')
-				work_link = service.link('https://www.ato.gov.au/individuals-and-families/income-deductions-offsets-and-records/deductions-you-can-claim', 'eligible work expenses')
-				payload.append(service.italic("* Organisation must be " + abr_link))
-				payload.append(service.italic("† Allow transfer time for Super, as contributions can only be deducted for the year they're received. Stake users: see " + stake_note))
-				payload.append(service.italic("‡ See " + work_link))
-				payload.append(service.italic("§ Must relate directly to specific trades and gains/income (e.g. not general finance news)"))
-
-		if month_and_day == '08-18':
-			myBirthday = datetime.datetime(2022,8,18,0,0,0,0, tzinfo=tz)
-			difference = relativedelta(localtime, myBirthday)
-			payload.append("It's my " + str(difference.years) + formats.ordinal(difference.years) + " birthday! 🥳")
-
-		if payload:
-			heading = service.bold("Finance event reminders:")
-			payload.insert(0, heading)
-		return payload
-
-
-	# MAIN #
-
+	today = datetime.datetime.now(zoneinfo.ZoneInfo(config.prefs.timezone)).date()
+	entries = config.data_json("reminder.json", [])
 	for dest in delivery.default_chats():
-		service = dest.dialect
-		delivery.deliver(dest, delivery.Report(prepare_finance_calendar_payload(service)))
+		delivery.deliver(dest, delivery.Report(report(dest.dialect, today, entries)))
+
+
+def report(service, today, entries):
+	"""Today's reminders in one network's markup, under a heading, or [] when there are none: the entries of
+	reminder.json for this day (and year, and country, where an entry names one), and finbot's birthday."""
+	payload = []
+	for entry in entries:
+		if entry.get("month_day") != today.strftime('%m-%d'):
+			continue
+		if entry.get("year") and str(entry["year"]) != today.strftime('%Y'):
+			continue
+		country = entry.get("gate_country")
+		if country and config.prefs.country_code != country:
+			continue
+		message = entry["message"] # a line, or a list of them
+		for line in [message] if isinstance(message, str) else message:
+			payload.append(markup(line.replace("{flag}", FLAGS.get(country, '')), service))
+	if today.strftime('%m-%d') == BIRTHDAY.strftime('%m-%d'):
+		years = relativedelta(today, BIRTHDAY).years
+		payload.append(f"It's my {years}{formats.ordinal(years)} birthday! 🥳")
+	return [service.bold("Finance event reminders:")] + payload if payload else []
+
+
+def markup(line, service):
+	"""A line of reminder.json in the network's markup: [text](url) is a link, a line wrapped in _underscores_ is in
+	italics, and the rest is text."""
+	italic = len(line) > 1 and line.startswith('_') and line.endswith('_')
+	if italic:
+		line = line[1:-1]
+	text, start = '', 0
+	for link in LINK.finditer(line):
+		text += service.escape(line[start:link.start()]) + service.link(link.group(2), service.escape(link.group(1)))
+		start = link.end()
+	text += service.escape(line[start:])
+	return service.italic(text) if italic else text
+
 
 def job(argv):
 	"""Arguments of a scheduled run: reminder.py takes none."""
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 14b23c38..2f96e0a7 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -430,6 +430,38 @@ def test_a_button_a_chart_does_not_offer_is_ignored(self):
 		self.assertEqual((edits, toasts), ([], [None]))
 
 
+class Reminders(FinbotCase):
+	def test_a_reminder_takes_links_and_italics_in_each_networks_markup(self):
+		"""[text](url) is a link and a line in _underscores_ is in italics, an underscore in a url included; the rest
+		is text, escaped where the network needs it."""
+		line = '_See [R&D notes](https://x.example/a_b) & more_'
+		self.assertEqual(reminder.markup(line, TELEGRAM),
+			'<i>See <a href="https://x.example/a_b">R&amp;D notes</a> &amp; more</i>')
+		self.assertEqual(reminder.markup(line, SLACK), '_See <https://x.example/a_b|R&amp;D notes> &amp; more_')
+		self.assertEqual(reminder.markup(line, DISCORD), '_See [R&D notes](<https://x.example/a_b>) & more_')
+		self.assertEqual(reminder.markup('a_b_c and _', TELEGRAM), 'a_b_c and _') # italic only when the whole line is
+
+	def test_the_end_of_financial_year_reminder_comes_from_reminder_json(self):
+		"""On 23 June an Australian gets the EOFY checklist from reminder.json, links and footnotes included; others do
+		not, and a year or a country an entry names must match."""
+		entries = config.data_json('reminder.json', [])
+		eofy = reminder.report(TELEGRAM, datetime.date(2026, 6, 23), entries)
+		self.assertEqual(eofy[0], '<b>Finance event reminders:</b>')
+		self.assertIn('💸 Realise capital gains/losses by EOFY June 30🇦🇺', eofy)
+		dgr = '<a href="https://abr.business.gov.au/Tools/DgrListing">DGR-registered</a>'
+		self.assertIn(f'<i>* Organisation must be {dgr}</i>', eofy)
+		self.prefs(country_code='US')
+		self.assertEqual(reminder.report(TELEGRAM, datetime.date(2026, 6, 23), entries), [])
+		dated = [{'month_day': '06-23', 'year': '2027', 'message': 'next year'},
+			{'month_day': '06-23', 'message': ['a', 'b']}]
+		self.assertEqual(reminder.report(SLACK, datetime.date(2026, 6, 23), dated),
+			['*Finance event reminders:*', 'a', 'b'])
+
+	def test_the_birthday(self):
+		"""On 18 August finbot counts its years."""
+		self.assertEqual(reminder.report(SLACK, datetime.date(2026, 8, 18), [])[1:], ["It's my 4th birthday! 🥳"])
+
+
 class Podcasts(FinbotCase):
 	def test_format_helpers(self):
 		"""Durations in any feed format, how long ago, and zoneless dates taken as UTC; nonsense gives nothing."""

From 5d3e8ca587a0a2d70ac265d890cbd65fcb5cda6e Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 02:05:03 +0000
Subject: [PATCH 28/81] One command table makes .help and the keyboard

COMMANDS becomes a table of Command(pattern, handler, usage, section,
buttons). The .help text and the Telegram command keyboard are made
from it, where four lists used to be kept by hand: the patterns,
prepare_help's lines, the keyboard's buttons and the README.
- .help lists each section's commands in table order.
- The keyboard is the commands' buttons, alphabetical. It is the same
  fifteen buttons as before, so its version hash is unchanged and no
  DM keyboard is sent again.
- The typo aliases (earrings, prince, hospital...) stay, as part of
  the patterns (D10).

The table is now in the order .help shows it, with .SYMBOL still last.
None of the reordered patterns can match another command's words, and
the output comparison is identical in every command step but .help and
/start. There .SYMBOL moves from the top of "Company info" to its end,
as the catch-all must stay last in the table.

The command keyboard's upkeep (keyboard_version, mark, push, ensure,
refresh) moves from charts.py to worker.py, beside the table it is made
from; charts.py keeps the period-button rebuild. The keyboard test no
longer counts buttons: each must run its own command (R16).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md             |  10 +--
 bot.py                |   3 +-
 lib/charts.py         |  50 +--------------
 lib/reports/about.py  |  42 ++-----------
 lib/worker.py         | 141 ++++++++++++++++++++++++++++++++----------
 tests/test_reports.py |  34 +++++++---
 6 files changed, 146 insertions(+), 134 deletions(-)

diff --git a/CLAUDE.md b/CLAUDE.md
index 3ecc63c8..7449cba3 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -34,7 +34,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - `bot.py`: WSGI entry (route, verify, parse, then a thread of `lib/worker.py` per message or button press); SIGHUP
   reload, SIGUSR1 cache flush. `lib/inbound.py`: a class per network that posts to the bot (`Telegram`, `Slack`): its
   path, how it proves itself, and its updates as `Message`, `Press` or a `Response` given straight back.
-- `lib/worker.py`: `COMMANDS` (regex → `on_*` handler, first match wins, `.SYMBOL` last); `Request`
+- `lib/worker.py`: `COMMANDS`, a table of `Command(pattern, handler, usage, section, buttons)` (first match wins,
+  `.SYMBOL` last), which also makes `.help` and the Telegram command keyboard, kept current in DMs; `Request`
   (`request.reply(lines, chart, buttons)`); the typing notice, which every command gets after half a second.
 - `lib/reports/`: what finbot says. Scheduled (each has `lambda_handler()` and `job(argv)`, its schedule-line parser): `price
   performance trades shorts cal rating milestone reminder podcasts cleanup`; the top-level `price.py` and the rest are shims
@@ -44,8 +45,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   `deliver(dest, report)`, the one-line error report. `lib/transports.py`: a class per way of posting (`TelegramBot`,
   `SlackBot`, `SlackWebhook`, `DiscordWebhook`): splitting, captions, charts, buttons, edits, typing. `lib/telegram.py`:
   Bot API calls.
-- `lib/buttons.py`: `Button` and a chart's period buttons (each transport draws them). `lib/charts.py`: a button press
-  rebuilds its chart; the Telegram command keyboard.
+- `lib/buttons.py`: `Button` and a chart's period buttons (each transport draws them). `lib/charts.py`: a period
+  button press rebuilds its chart.
 - `lib/dialects.py`: each network's markup and message sizes. `lib/markets.py`: market tables, flags, ticker translation.
   `lib/formats.py`: names, units, dates, percents in words. `lib/links.py`: stock links. `lib/render.py`: every chart.
 - `lib/yahoo.py lib/sharesight.py lib/shortman.py lib/simplywallst.py`: data sources, cached in `DATA_DIR/cache`.
@@ -77,7 +78,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   print there (a test checks); only the util/ tools, run by hand, do.
 - A chart renderer in `lib/render.py` gets `@_drawn` (the bot draws in one native thread); it must not call another `@_drawn`
   function, or that thread waits on itself.
-- A new command: `COMMANDS`, `prepare_help`, the README command list, and `command_keyboard` if it gets a button.
+- A new command: a `Command` in `COMMANDS`, whose usage, section and buttons make its `.help` lines and keyboard
+  buttons, and the README command list.
 
 ## Saving tokens
 - Run the tests quietly: `.venv/bin/python -m unittest discover -s tests -b` prints only failures (about 0.4 KB when green). One
diff --git a/bot.py b/bot.py
index 4bad910d..51e3be45 100755
--- a/bot.py
+++ b/bot.py
@@ -15,7 +15,6 @@
 from lib import inbound
 from lib import worker
 from lib import jobs
-from lib import charts
 from lib import sharesight
 from lib import net
 from lib import telegram
@@ -199,7 +198,7 @@ def flush_cache(*args):
 			print('ERROR', delivery.error_line(e, 'telegram webhook'), file=sys.stderr)
 			sys.exit(1)
 	if config.env.telegramBotToken:
-		threading.Thread(target=charts.refresh_keyboards, daemon=True).start() # update stale DM keyboards in the background
+		threading.Thread(target=worker.refresh_keyboards, daemon=True).start() # stale DM keyboards, in the background
 	if config.env.sharesight_auth.get('client_id'):
 		threading.Thread(target=sharesight.keep_warm, daemon=True).start() # keep the portfolio list and holdings report cached, while cache is on
 	if scheduler:
diff --git a/lib/charts.py b/lib/charts.py
index 9906808d..b8715b05 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -1,59 +1,11 @@
-"""A period button pressed: the chart rebuilt for that period. And the Telegram command keyboard, kept current in every DM."""
-import hashlib
-import json
-import sys
-
+"""A period button pressed: the chart rebuilt for that period."""
 from lib import answer
 from lib import buttons
 from lib import formats
 from lib import links
-from lib import state
-from lib import transports
 from lib import yahoo
 from lib.reports import history, performance, price
 
-COMMAND_BUTTONS = sorted(['.beta', '.buy & sell', '.dividend', '.earnings', '.forward pe', '.marketcap', '.pe', '.peg', '.performance', '.premarket',
-	'.price', '.session', '.shorts', '.trades', '.watchlist']) # alphabetical, row by row; '.buy & sell' sends both rating reports
-
-def command_keyboard(per_row=3):
-	"""Persistent reply keyboard of the common commands (Telegram DMs); a press sends the command text as a message."""
-	rows = [[{'text': c} for c in COMMAND_BUTTONS[i:i + per_row]] for i in range(0, len(COMMAND_BUTTONS), per_row)]
-	return {'keyboard': rows, 'resize_keyboard': True, 'is_persistent': True}
-
-KEYBOARD_FILE = 'finbot_keyboards.json' # chat id -> version of the command keyboard that chat last received
-
-def keyboard_version():
-	"""A short hash of the command keyboard, which changes whenever its buttons do."""
-	return hashlib.sha1(json.dumps(command_keyboard(), sort_keys=True).encode()).hexdigest()[:8]
-
-def mark_keyboard(chat_id):
-	"""Record that this chat now has the current command keyboard."""
-	with state.edit(KEYBOARD_FILE, {}) as versions:
-		versions[str(chat_id)] = keyboard_version()
-
-def push_keyboard(chat_id):
-	"""Telegram can only change a reply keyboard by sending a message, so send a short one carrying it."""
-	result = transports.TelegramBot(chat_id).post_text('⌨️ Keyboard updated', command_keyboard())
-	if result is not None:
-		mark_keyboard(chat_id)
-	return result
-
-def ensure_keyboard(chat_id):
-	"""DMs only (positive chat ids): bring a chat's keyboard up to date if its buttons have changed or it never had them."""
-	if int(chat_id) <= 0:
-		return
-	if state.load(KEYBOARD_FILE, {}).get(str(chat_id)) != keyboard_version():
-		push_keyboard(chat_id)
-
-def refresh_keyboards():
-	"""At startup: update every chat we know of whose keyboard is out of date."""
-	for chat_id, version in state.load(KEYBOARD_FILE, {}).items():
-		if version != keyboard_version():
-			try:
-				push_keyboard(chat_id)
-			except Exception as e:
-				print("keyboard refresh failed for", chat_id, e, file=sys.stderr)
-
 def build(kind, tickers, period, service):
 	"""(caption, image) for a chart kind at a button period; period None = the command's default view."""
 	days = buttons.PERIODS[period][1] if period else None
diff --git a/lib/reports/about.py b/lib/reports/about.py
index c4bb3e71..fc75abf3 100644
--- a/lib/reports/about.py
+++ b/lib/reports/about.py
@@ -41,43 +41,13 @@ def job_line(scheduler, label, outcome):
 	number = [command[0] for command in scheduler.commands()].index(label) + 1
 	return f"{number}. {jobs.display(label)}: {outcome}"
 
-def prepare_help(service, botName):
-	"""The .help text: every command, by section. botName is the bot's @name, which can start a command too."""
+def prepare_help(service, botName, sections):
+	"""The .help text: each section's command lines under its heading (worker.help_sections() has them), then the bot's
+	@name as another way to start a command. botName is the bot's @name."""
 	payload = []
-	payload.append(service.bold("Tracked securities:"))
-	payload.append(".holdings")
-	payload.append(".pie PORTFOLIO [sector]")
-	payload.append(".watchlist [add|del SYMBOL]")
-	payload.append(".who SYMBOL")
-
-	payload.append(service.bold("\nCompany info:"))
-	payload.append('.SYMBOL')
-	payload.append(".dividend [period|SYMBOL]")
-	payload.append(".earnings [period|SYMBOL]")
-	payload.append(".marketcap [SYMBOL|bottom|top]")
-
-	payload.append(service.bold("\nPrice:"))
-	payload.append(".beta")
-	payload.append(".compare SYMBOL SYMBOL [...] [period|max]")
-	payload.append(".history SYMBOL")
-	payload.append(".performance [period] [portfolio]")
-	payload.append(".price [percent|SYMBOL] [period]")
-	payload.append(".session [percent|SYMBOL]")
-	payload.append(".premarket [percent|SYMBOL]")
-
-	payload.append(service.bold("\nRisk & Value:"))
-	payload.append(".buy")
-	payload.append(".sell")
-	payload.append(".pe [SYMBOL|top|bottom]")
-	payload.append(".forwardpe [SYMBOL|top|bottom|negative]")
-	payload.append(".peg [SYMBOL|top|bottom|negative]")
-	payload.append(".shorts [percent|SYMBOL]")
-
-	payload.append(service.bold("\nTrades:"))
-	payload.append(".trades [period] [portfolio]")
-
-	payload.append(service.bold("\nScheduled jobs:"))
-	payload.append(".jobs [run NUMBER|COMMAND]")
+	for i, (heading, lines) in enumerate(sections):
+		payload.append(service.bold(("\n" if i else "") + heading + ":"))
+		payload += lines
 
 	payload.append(service.bold("\nAlternative syntax:"))
 	if service == 'slack':
diff --git a/lib/worker.py b/lib/worker.py
index c5be02b3..7f92ea1f 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -1,11 +1,15 @@
-"""Chat commands: a message is answered by the function of the first pattern in COMMANDS that fits it."""
+"""Chat commands: a message is answered by the handler of the first Command in COMMANDS that fits it. The table also
+makes the .help text and the Telegram command keyboard, which is kept current in every DM."""
 import dataclasses
 import datetime
+import hashlib
+import json
 import math
 import random
 import re
 import sys
 import threading
+import typing
 from lib import answer
 from lib import config
 from lib import dialects
@@ -18,6 +22,7 @@
 from lib import charts
 from lib.reports import about, cal, history, holdings, performance, price, profile, rankings, shorts, trades
 from lib import telegram
+from lib import transports
 from lib import jobs
 
 class TypingIndicator:
@@ -123,10 +128,10 @@ def process_request(service, chat_id, user, message, botName, userRealName):
 	originating chat, after the typing stops."""
 	request = Request(dialects.of(service), chat_id, user, userRealName, botName)
 	try:
-		for pattern, handler in COMMANDS:
-			if m := match(pattern, message, botName):
+		for command in COMMANDS:
+			if m := match(command.pattern, message, botName):
 				request.typing.start()
-				handler(request, m)
+				command.handler(request, m)
 				break
 	except answer.Answer as e: # a portfolio name the user got wrong, a ticker with no prices: an answer, not an error
 		request.reply(e.lines)
@@ -137,7 +142,7 @@ def process_request(service, chat_id, user, message, botName, userRealName):
 		request.typing.stop()
 	if service == 'telegram':
 		try:
-			charts.ensure_keyboard(chat_id) # keep DM keyboards current after the reply
+			ensure_keyboard(chat_id) # keep DM keyboards current after the reply
 		except Exception as e:
 			print("keyboard update failed:", e, file=sys.stderr)
 
@@ -203,10 +208,10 @@ def on_help(request, m):
 	""".help: every command; in a Telegram DM the command keyboard too."""
 	markup = None
 	if request.service == 'telegram' and int(request.chat_id) > 0: # positive chat ids are DMs; groups are negative
-		markup = charts.command_keyboard()
-	request.reply(about.prepare_help(request.service, request.bot_name), buttons=markup)
+		markup = command_keyboard()
+	request.reply(about.prepare_help(request.service, request.bot_name, help_sections()), buttons=markup)
 	if markup:
-		charts.mark_keyboard(request.chat_id) # this chat now has the current keyboard
+		mark_keyboard(request.chat_id) # this chat now has the current keyboard
 
 def on_hello(request, m):
 	"""Easter egg: a greeting, alliterated when the alliterate preference is on."""
@@ -369,29 +374,99 @@ def on_profile(request, m):
 	payload = profile.prepare_profile_payload(request.service, request.user, m.group('ticker').upper())
 	request.reply(payload)
 
-COMMANDS = [ # (pattern, function): tried in this order, the first match answers. .SYMBOL takes anything, so it stays last
-	(r"(?:watchlist|wishlist)\s*(?P<action>[\w]+)*\s*(?P<ticker>[\w\.\:\-\^\=]+)*", on_watchlist),
-	(r"(?:help|usage|start)", on_help), # /start is what Telegram sends when a chat opens
-	(r"(?:hi$|hello)|^(?:hi|hello)\s+{bot}", on_hello),
-	(r"(?:thanks|thank you)|^(?:thanks|thank you)\s+{bot}", on_thanks),
-	(r"(?P<kind>earnings?|earrings?|dividends?)\s*(?P<arg>[\w\.\:\-]+)*", on_calendar),
-	(r"performance?\s*(?P<args>[\w\s]*)", on_performance),
-	(r"(?P<kind>session|premarket|postmarket|permarket)\s*(?P<arg>[\w\.\:\-]+)*", on_session),
-	(r"(?:prices?|prince|print|probe|piece|pierce|pence|prime)\s*([\w\.\:\%\=\-\^]+)*\s*([\w\%]+)*", on_price),
-	(r"shorts?\s*([\w\.\:\-]+)*", on_shorts),
-	(r"trades?\s*(?P<args>[\w\s]*)", on_trades),
-	(r"holdings?\s*([\w\s]+)*", on_holdings),
-	(r"(?:marketcap|maletas|marketer)\s*([\w\.\:\-]+)*", on_marketcap),
-	(r"peg\s*([\w\.\:\-\s]+)*", on_ratio('peg', negative=True, chart=True)),
-	(r"pe(?:$|\s+)([\w\.\:\-\s]+)*", on_ratio('pe')),
-	(r"(?:fpe|forward\s?pe)\s*([\w\.\:\-\s]+)*", on_ratio('forward pe', negative=True, chart=True)),
-	(r"beta\s*([\w\.\:\-]+)*", on_beta),
-	(r"pie(?:$|\s+)(?P<args>[\w\s\-\.&]*)", on_pie),
-	(r"(?P<first>buy|sell)(?:\s*(?:&|and|/|\+)?\s*(?P<second>buy|sell))?", on_rating), # .buy & sell (or .buy sell): both
-	(r"compare(?:\s+(?P<args>[\w\.\:\-\^\=\s]+))?", on_compare),
-	(r"(?:history|hospital|visual|hosiery)\s*(?P<ticker>[\w\.\:\-]+)?\s*(?P<extra>[\w\%]+)*", on_history),
-	(r"plan\s*(.*)", on_plan),
-	(r"(?:who)\s*(?P<ticker>[\w\.\:\-\^]+)*", on_who),
-	(r"(?:jobs|crons?|schedule)(?:\s+(?P<args>.*?))?\s*$", on_jobs),
-	(r"(?P<ticker>[\w\.\:\-]+)", on_profile), # .SYMBOL: a company profile
+class Command(typing.NamedTuple):
+	"""A chat command: the pattern it answers (after . ! / or the bot's name), the handler that answers it, its .help
+	lines under its section's heading, and the Telegram keyboard buttons that send it."""
+	pattern: str
+	handler: typing.Callable
+	usage: tuple = () # its lines in .help; none keeps it out of .help
+	section: str = ''
+	buttons: tuple = ()
+
+SECTIONS = TRACKED, COMPANY, PRICE, RISK, TRADES, JOBS = ('Tracked securities', 'Company info', 'Price', 'Risk & Value',
+	'Trades', 'Scheduled jobs') # the headings of .help, in order
+
+COMMANDS = [ # tried in this order, the first match answers: .SYMBOL takes anything, so it stays last. Within a section,
+	# .help lists the commands in this order. The odd spellings (earrings, prince, hospital...) are kept on purpose.
+	Command(r"(?:help|usage|start)", on_help), # /start is what Telegram sends when a chat opens
+	Command(r"(?:hi$|hello)|^(?:hi|hello)\s+{bot}", on_hello),
+	Command(r"(?:thanks|thank you)|^(?:thanks|thank you)\s+{bot}", on_thanks),
+	Command(r"holdings?\s*([\w\s]+)*", on_holdings, ('.holdings',), TRACKED),
+	Command(r"pie(?:$|\s+)(?P<args>[\w\s\-\.&]*)", on_pie, ('.pie PORTFOLIO [sector]',), TRACKED),
+	Command(r"(?:watchlist|wishlist)\s*(?P<action>[\w]+)*\s*(?P<ticker>[\w\.\:\-\^\=]+)*", on_watchlist,
+		('.watchlist [add|del SYMBOL]',), TRACKED, ('.watchlist',)),
+	Command(r"(?:who)\s*(?P<ticker>[\w\.\:\-\^]+)*", on_who, ('.who SYMBOL',), TRACKED),
+	Command(r"(?P<kind>earnings?|earrings?|dividends?)\s*(?P<arg>[\w\.\:\-]+)*", on_calendar,
+		('.dividend [period|SYMBOL]', '.earnings [period|SYMBOL]'), COMPANY, ('.dividend', '.earnings')),
+	Command(r"(?:marketcap|maletas|marketer)\s*([\w\.\:\-]+)*", on_marketcap, ('.marketcap [SYMBOL|bottom|top]',),
+		COMPANY, ('.marketcap',)),
+	Command(r"beta\s*([\w\.\:\-]+)*", on_beta, ('.beta',), PRICE, ('.beta',)),
+	Command(r"compare(?:\s+(?P<args>[\w\.\:\-\^\=\s]+))?", on_compare, ('.compare SYMBOL SYMBOL [...] [period|max]',),
+		PRICE),
+	Command(r"(?:history|hospital|visual|hosiery)\s*(?P<ticker>[\w\.\:\-]+)?\s*(?P<extra>[\w\%]+)*", on_history,
+		('.history SYMBOL',), PRICE),
+	Command(r"performance?\s*(?P<args>[\w\s]*)", on_performance, ('.performance [period] [portfolio]',), PRICE,
+		('.performance',)),
+	Command(r"(?:prices?|prince|print|probe|piece|pierce|pence|prime)\s*([\w\.\:\%\=\-\^]+)*\s*([\w\%]+)*", on_price,
+		('.price [percent|SYMBOL] [period]',), PRICE, ('.price',)),
+	Command(r"(?P<kind>session|premarket|postmarket|permarket)\s*(?P<arg>[\w\.\:\-]+)*", on_session,
+		('.session [percent|SYMBOL]', '.premarket [percent|SYMBOL]'), PRICE, ('.session', '.premarket')),
+	Command(r"(?P<first>buy|sell)(?:\s*(?:&|and|/|\+)?\s*(?P<second>buy|sell))?", on_rating, ('.buy', '.sell'), RISK,
+		('.buy & sell',)), # .buy & sell (or .buy sell): both reports
+	Command(r"pe(?:$|\s+)([\w\.\:\-\s]+)*", on_ratio('pe'), ('.pe [SYMBOL|top|bottom]',), RISK, ('.pe',)),
+	Command(r"(?:fpe|forward\s?pe)\s*([\w\.\:\-\s]+)*", on_ratio('forward pe', negative=True, chart=True),
+		('.forwardpe [SYMBOL|top|bottom|negative]',), RISK, ('.forward pe',)),
+	Command(r"peg\s*([\w\.\:\-\s]+)*", on_ratio('peg', negative=True, chart=True),
+		('.peg [SYMBOL|top|bottom|negative]',), RISK, ('.peg',)),
+	Command(r"shorts?\s*([\w\.\:\-]+)*", on_shorts, ('.shorts [percent|SYMBOL]',), RISK, ('.shorts',)),
+	Command(r"trades?\s*(?P<args>[\w\s]*)", on_trades, ('.trades [period] [portfolio]',), TRADES, ('.trades',)),
+	Command(r"plan\s*(.*)", on_plan),
+	Command(r"(?:jobs|crons?|schedule)(?:\s+(?P<args>.*?))?\s*$", on_jobs, ('.jobs [run NUMBER|COMMAND]',), JOBS),
+	Command(r"(?P<ticker>[\w\.\:\-]+)", on_profile, ('.SYMBOL',), COMPANY), # a company profile
 ]
+
+def help_sections():
+	"""The sections of .help: (heading, the .help lines of its commands, in table order), in SECTIONS order."""
+	return [(section, [line for command in COMMANDS if command.section == section for line in command.usage])
+		for section in SECTIONS]
+
+def command_keyboard(per_row=3):
+	"""Persistent reply keyboard of the commands that have buttons (Telegram DMs), alphabetical, row by row; a press
+	sends the button's text as a message."""
+	texts = sorted(text for command in COMMANDS for text in command.buttons)
+	rows = [[{'text': text} for text in texts[i:i + per_row]] for i in range(0, len(texts), per_row)]
+	return {'keyboard': rows, 'resize_keyboard': True, 'is_persistent': True}
+
+KEYBOARD_FILE = 'finbot_keyboards.json' # chat id -> version of the command keyboard that chat last received
+
+def keyboard_version():
+	"""A short hash of the command keyboard, which changes whenever its buttons do."""
+	return hashlib.sha1(json.dumps(command_keyboard(), sort_keys=True).encode()).hexdigest()[:8]
+
+def mark_keyboard(chat_id):
+	"""Record that this chat now has the current command keyboard."""
+	with state.edit(KEYBOARD_FILE, {}) as versions:
+		versions[str(chat_id)] = keyboard_version()
+
+def push_keyboard(chat_id):
+	"""Telegram can only change a reply keyboard by sending a message, so send a short one carrying it."""
+	result = transports.TelegramBot(chat_id).post_text('⌨️ Keyboard updated', command_keyboard())
+	if result is not None:
+		mark_keyboard(chat_id)
+	return result
+
+def ensure_keyboard(chat_id):
+	"""DMs only (positive chat ids): update a chat's keyboard if its buttons changed or it never had them."""
+	if int(chat_id) <= 0:
+		return
+	if state.load(KEYBOARD_FILE, {}).get(str(chat_id)) != keyboard_version():
+		push_keyboard(chat_id)
+
+def refresh_keyboards():
+	"""At startup: update every chat we know of whose keyboard is out of date."""
+	for chat_id, version in state.load(KEYBOARD_FILE, {}).items():
+		if version != keyboard_version():
+			try:
+				push_keyboard(chat_id)
+			except Exception as e:
+				print("keyboard refresh failed for", chat_id, e, file=sys.stderr)
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 2f96e0a7..5a67ec6e 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -35,16 +35,18 @@ def test_one_button_asks_for_both_ratings_reports(self):
 		self.assertIn('analyst sell ratings', str(self.command('.sell & buy')[0])) # in the order asked
 		self.assertEqual(len(self.command('.buy & buy')), 1) # each report once
 
-	def test_the_keyboard_is_alphabetical_and_every_button_is_a_command(self):
-		"""The command keyboard is alphabetical, and each button is a command, not taken for a ticker by .SYMBOL."""
-		buttons = [button['text'] for row in charts.command_keyboard()['keyboard'] for button in row]
+	def test_the_keyboard_is_alphabetical_and_every_button_is_its_command(self):
+		"""The command keyboard is alphabetical, and each button runs the command it belongs to in the table, not taken
+		for a ticker by .SYMBOL."""
+		buttons = [button['text'] for row in worker.command_keyboard()['keyboard'] for button in row]
 		self.assertEqual(buttons, sorted(buttons))
-		self.assertEqual(len(buttons), 15)
 		self.assertIn('.buy & sell', buttons)
 		self.assertIn('.forward pe', buttons)
-		for text in buttons: # not taken for a ticker by the catch-all .SYMBOL
-			with self.subTest(text=text):
-				self.assertIsNot(next(handler for pattern, handler in worker.COMMANDS if worker.match(pattern, text, BOT)), worker.on_profile)
+		for command in worker.COMMANDS:
+			for text in command.buttons:
+				with self.subTest(text=text):
+					first = next(c for c in worker.COMMANDS if worker.match(c.pattern, text, BOT))
+					self.assertIs(first.handler, command.handler)
 
 	def test_history_and_compare(self):
 		""".history and .compare answer."""
@@ -63,9 +65,21 @@ def test_unknown_stickers_and_noise_are_ignored(self):
 
 
 class ReportFunctions(FinbotCase):
-	def test_prepare_help(self):
-		"""The help text is built."""
-		self.assertTrue(about.prepare_help(TELEGRAM, BOT))
+	def test_help_comes_from_the_command_table(self):
+		""".help lists every command's lines from the table, under its section's heading, the sections in order; and
+		each line names a command that runs it."""
+		text = about.prepare_help(TELEGRAM, BOT, worker.help_sections())
+		headings = [line for line in text if line.startswith('<b>')]
+		self.assertEqual(headings[:len(worker.SECTIONS)], [f"<b>{'' if i == 0 else chr(10)}{s}:</b>"
+			for i, s in enumerate(worker.SECTIONS)])
+		for command in worker.COMMANDS:
+			for line in command.usage:
+				with self.subTest(line=line):
+					self.assertIn(line, text)
+					word = line.split()[0]
+					if word != '.SYMBOL':
+						self.assertIs(next(c for c in worker.COMMANDS if worker.match(c.pattern, word, BOT)).handler,
+							command.handler)
 
 	def test_a_portfolio_with_no_holdings_is_reported_by_name(self):
 		"""Its name used to be read from its first holding: an empty portfolio crashed .performance."""

From 6254facbd4d054eaa2c46a78ab3145db4b25ee24 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 02:11:13 +0000
Subject: [PATCH 29/81] Every report runs as run(dest=None, **kwargs) (D7)

The scheduled reports' entry point, lambda_handler (a leftover from AWS
Lambda), becomes run(dest=None, **kwargs), with keyword-only arguments.
dest is the chat that asked; without one, the report goes to the
default chats. That replaces the (chat_id, service, interactive)
triple, and the argument orders that differed between reports: worker
called shorts(chat, threshold, ticker, service) but price(chat,
threshold, service, user, ticker).

- delivery.targets(dest) replaces destinations(interactive, service,
  chat_id).
- The worker passes request.dest; jobs, jobrun, the command line shims
  and the cache flush call run(**kwargs).
- milestone, reminder and podcasts take a dest too, so every report
  can be sent to one chat.
- util/record_outputs.py calls run, or lambda_handler for a commit from
  before this one.

The output comparison is identical in all 159 steps.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md                  |  9 ++++---
 bot.py                     |  2 +-
 lib/delivery.py            |  6 ++---
 lib/jobrun.py              |  2 +-
 lib/jobs.py                |  2 +-
 lib/reports/__init__.py    | 11 ++++----
 lib/reports/cal.py         | 17 +++++++------
 lib/reports/cleanup.py     |  4 +--
 lib/reports/milestone.py   | 11 ++++----
 lib/reports/performance.py | 18 ++++++-------
 lib/reports/podcasts.py    | 13 +++++-----
 lib/reports/price.py       | 27 ++++++++++----------
 lib/reports/rating.py      | 13 +++++-----
 lib/reports/reminder.py    |  8 +++---
 lib/reports/shorts.py      | 14 +++++-----
 lib/reports/trades.py      | 15 ++++++-----
 lib/worker.py              | 16 ++++++------
 tests/test_finbot.py       |  2 +-
 tests/test_jobs.py         | 25 +++++++++---------
 tests/test_providers.py    |  4 +--
 tests/test_reports.py      | 52 +++++++++++++++++++-------------------
 util/record_outputs.py     |  3 ++-
 22 files changed, 141 insertions(+), 133 deletions(-)

diff --git a/CLAUDE.md b/CLAUDE.md
index 7449cba3..90cb2d43 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -37,10 +37,11 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - `lib/worker.py`: `COMMANDS`, a table of `Command(pattern, handler, usage, section, buttons)` (first match wins,
   `.SYMBOL` last), which also makes `.help` and the Telegram command keyboard, kept current in DMs; `Request`
   (`request.reply(lines, chart, buttons)`); the typing notice, which every command gets after half a second.
-- `lib/reports/`: what finbot says. Scheduled (each has `lambda_handler()` and `job(argv)`, its schedule-line parser): `price
-  performance trades shorts cal rating milestone reminder podcasts cleanup`; the top-level `price.py` and the rest are shims
-  that run them. Chat-only (`prepare_*`): `about` (.help .jobs), `history` (.history .compare), `holdings` (.holdings .who .pie
-  .watchlist), `profile` (.SYMBOL), `rankings` (.marketcap .pe .fpe .peg .beta .buy .sell).
+- `lib/reports/`: what finbot says. Scheduled (each has `run(dest=None, **kwargs)`, which posts to the chat that asked
+  or else the default chats, and `job(argv)`, its schedule-line parser): `price performance trades shorts cal rating
+  milestone reminder podcasts cleanup`; the top-level `price.py` and the rest are shims that run them. Chat-only
+  (`prepare_*`): `about` (.help .jobs), `history` (.history .compare), `holdings` (.holdings .who .pie .watchlist),
+  `profile` (.SYMBOL), `rankings` (.marketcap .pe .fpe .peg .beta .buy .sell).
 - `lib/delivery.py`: `Report(lines, chart, buttons)`, the destinations (the chat that asked, or the default chats),
   `deliver(dest, report)`, the one-line error report. `lib/transports.py`: a class per way of posting (`TelegramBot`,
   `SlackBot`, `SlackWebhook`, `DiscordWebhook`): splitting, captions, charts, buttons, edits, typing. `lib/telegram.py`:
diff --git a/bot.py b/bot.py
index 51e3be45..5eb57cb7 100755
--- a/bot.py
+++ b/bot.py
@@ -156,7 +156,7 @@ def flush_cache(*args):
 	Yahoo's cookie and crumb belong together, so they go while nothing is renewing them; the Sharesight caches refill at once."""
 	try:
 		with yahoo._auth:
-			removed = cleanup.lambda_handler(days=0)
+			removed = cleanup.run(days=0)
 		sharesight.wake()
 		print(f"cache: flushed {removed} file{'' if removed == 1 else 's'} from {config.env.cache_dir}", file=sys.stderr)
 	except Exception as e:
diff --git a/lib/delivery.py b/lib/delivery.py
index 963b5cdd..cfe55f68 100644
--- a/lib/delivery.py
+++ b/lib/delivery.py
@@ -41,9 +41,9 @@ def default_chats():
 	return chats
 
 
-def destinations(interactive, service, chat_id):
-	"""Where a report goes: back to the chat that asked for it (interactive), else to the default chats."""
-	return [reply_to(service, chat_id)] if interactive else default_chats()
+def targets(dest):
+	"""Where a report goes: the chat that asked for it (dest), else the default chats."""
+	return [dest] if dest else default_chats()
 
 
 def takes_charts(service, reply=True):
diff --git a/lib/jobrun.py b/lib/jobrun.py
index 5074bfe0..fb3680f3 100644
--- a/lib/jobrun.py
+++ b/lib/jobrun.py
@@ -22,7 +22,7 @@ def main(argv):
 	module = argv[0]
 	def run():
 		"""Import the report and run it: an import that fails is reported like any other failure."""
-		importlib.import_module(module).lambda_handler(**json.loads(argv[1]))
+		importlib.import_module(module).run(**json.loads(argv[1]))
 		return True
 	return 0 if delivery.guarded(module.rsplit('.', 1)[-1] + '.py', run, exit=False) else REPORTED
 
diff --git a/lib/jobs.py b/lib/jobs.py
index 288c5a07..5ccfd9ff 100644
--- a/lib/jobs.py
+++ b/lib/jobs.py
@@ -71,7 +71,7 @@ def exit_reason(status):
 
 def run_in_process(label, module, kwargs):
 	"""The runner for tests: call the handler here instead of in a child process."""
-	module.lambda_handler(**kwargs)
+	module.run(**kwargs)
 	return 0
 
 
diff --git a/lib/reports/__init__.py b/lib/reports/__init__.py
index 6fed69fe..ac36214b 100644
--- a/lib/reports/__init__.py
+++ b/lib/reports/__init__.py
@@ -1,7 +1,8 @@
-"""What finbot says. A module per scheduled report: cal, cleanup (of the cache), milestone, performance, podcasts, price, rating,
-reminder, shorts and trades, each with job(argv), which parses its schedule line, and lambda_handler(**kwargs), which runs it; the
-top-level script of the same name runs it from the command line. And the chat-only reports: about (.help, .jobs), history
-(.history, .compare), holdings (.holdings, .who, .pie, .watchlist), profile (.SYMBOL) and rankings (.marketcap, .pe and the rest)."""
+"""What finbot says. A module per scheduled report: cal, cleanup (of the cache), milestone, performance, podcasts,
+price, rating, reminder, shorts and trades, each with job(argv), which parses its schedule line, and run(dest=None,
+**kwargs), which runs it, to the chat that asked (dest) or else the default chats; the top-level script of the same
+name runs it from the command line. And the chat-only reports: about (.help, .jobs), history (.history, .compare),
+holdings (.holdings, .who, .pie, .watchlist), profile (.SYMBOL) and rankings (.marketcap, .pe and the rest)."""
 import os
 import sys
 
@@ -11,4 +12,4 @@ def cli(path):
 	import importlib
 	from lib import delivery
 	report = importlib.import_module('lib.reports.' + os.path.basename(path).removesuffix('.py'))
-	delivery.cli(sys.argv[0], report.job, report.lambda_handler)
+	delivery.cli(sys.argv[0], report.job, report.run)
diff --git a/lib/reports/cal.py b/lib/reports/cal.py
index 5eeddec0..c94d3517 100644
--- a/lib/reports/cal.py
+++ b/lib/reports/cal.py
@@ -15,8 +15,10 @@ def primary_listings():
 
 EVENTS = {'earnings': ('earnings_date', 'Upcoming earnings'), 'ex-dividend': ('ex_dividend_date', 'Upcoming ex-dividends')} # -> (field, heading)
 
-def lambda_handler(chat_id=None, days=None, service=None, specific_stock=None, interactive=False, event='earnings'):
-	"""Earnings or ex-dividend dates ahead (`event`) for tracked stocks or one; Telegram pins scheduled earnings."""
+def run(dest=None, *, days=None, specific_stock=None, event='earnings'):
+	"""Post the earnings or ex-dividend dates ahead (`event`) for the tracked stocks or one: to the chat that asked
+	(dest), else to the default chats. An earnings report on all the tracked stocks is pinned where the network can."""
+	interactive = dest is not None
 	if days is None:
 		days = config.prefs.future_days
 	field, heading = EVENTS[event]
@@ -68,14 +70,13 @@ def prepare_payload(service, market_data):
 	else: # ex-dividend dates are in Yahoo's details
 		market_data = yahoo.fetch_details(tickers, strict=bool(specific_stock))
 
-	for dest in delivery.destinations(interactive, service, chat_id):
-		service = dest.dialect
-		answers = delivery.deliver(dest, delivery.Report(prepare_payload(service, market_data)))
-		if event == 'earnings' and not specific_stock and hasattr(dest, 'pin'):
-			dest.pin(answers) # the earnings report stays on top, where the network can pin
+	for chat in delivery.targets(dest):
+		answers = delivery.deliver(chat, delivery.Report(prepare_payload(chat.dialect, market_data)))
+		if event == 'earnings' and not specific_stock and hasattr(chat, 'pin'):
+			chat.pin(answers) # the earnings report stays on top, where the network can pin
 
 def job(argv):
-	"""Arguments of a scheduled run: (label, lambda_handler kwargs). Raises ValueError with the usage on a bad argument."""
+	"""A scheduled run's arguments: (label, run's keyword arguments); a bad one raises ValueError with the usage."""
 	match argv:
 		case ['earnings' | 'ex-dividend' as event]:
 			return f'cal.py {event}', {'event': event}
diff --git a/lib/reports/cleanup.py b/lib/reports/cleanup.py
index 23b3d99b..f38bc7f9 100644
--- a/lib/reports/cleanup.py
+++ b/lib/reports/cleanup.py
@@ -2,7 +2,7 @@
 import os, sys, time
 from lib import config
 
-def lambda_handler(days=30):
+def run(*, days=30):
 	"""Delete the cache files (finbot_*) that were last written more than `days` days ago (days=0: all of them, a flush), but not
 	a temp file written in the last minute: that is an atomic write in flight (cache.write_file)."""
 	now = time.time()
@@ -24,7 +24,7 @@ def lambda_handler(days=30):
 	return removed
 
 def job(argv):
-	"""Arguments of a scheduled run: (label, lambda_handler kwargs). Raises ValueError with the usage on a bad argument."""
+	"""A scheduled run's arguments: (label, run's keyword arguments); a bad one raises ValueError with the usage."""
 	match argv:
 		case []:
 			return 'cleanup.py', {}
diff --git a/lib/reports/milestone.py b/lib/reports/milestone.py
index 162a2e9e..5fac54a7 100644
--- a/lib/reports/milestone.py
+++ b/lib/reports/milestone.py
@@ -11,9 +11,9 @@
 # - need to handle stock splits: flush cache if we see a split in the data
 # https://finance.yahoo.com/quote/TSLA/history?period1=0&period2=1707609600&filter=split
 
-def lambda_handler():
+def run(dest=None):
 	"""Post the stocks followed that set an all-time high or low (their 52-week range against the record, which a first
-	sighting takes from the stock's history), and record the new records."""
+	sighting takes from the stock's history) to the default chats (or dest), and record the new records."""
 	def prepare_milestone_payload(service, market_data):
 		"""(the report in one network's markup, the records it changes)."""
 		payload = []
@@ -62,10 +62,9 @@ def prepare_milestone_payload(service, market_data):
 	tickers = watchlist.with_holdings()
 	market_data = yahoo.fetch(tickers)
 	records, changed = state.load('finbot_milestone.json', {}), {}
-	for dest in delivery.default_chats():
-		service = dest.dialect
-		payload, changed = prepare_milestone_payload(service, market_data)
-		delivery.deliver(dest, delivery.Report(payload))
+	for chat in delivery.targets(dest):
+		payload, changed = prepare_milestone_payload(chat.dialect, market_data)
+		delivery.deliver(chat, delivery.Report(payload))
 
 	if changed:
 		with state.edit('finbot_milestone.json', {}) as recorded:
diff --git a/lib/reports/performance.py b/lib/reports/performance.py
index d9f4b46e..f1bf606a 100644
--- a/lib/reports/performance.py
+++ b/lib/reports/performance.py
@@ -62,23 +62,23 @@ def report(service, past_days=None, portfolio_select=None, interactive=False, pe
 		graph = render.rows_chart(chart_rows, "Performance", f"% change {'' if period == 'today' else 'over '}{period}", value_fmt=lambda v: formats.signed_percent(v, 2))
 	return payload, graph
 
-def lambda_handler(chat_id=None, past_days=None, service=None, user='', portfolio_select=None, interactive=False):
+def run(dest=None, *, past_days=None, user='', portfolio_select=None):
 	"""Post each portfolio's performance over past_days (the past_days preference when None), or the one named, beside
-	the benchmarks: to the chat that asked (interactive), else to the default chats, with its chart and on Telegram
-	period buttons."""
+	the benchmarks: to the chat that asked (dest), else to the default chats, with its chart and period buttons where
+	they can go."""
+	interactive = dest is not None
 	if past_days is None:
 		past_days = config.prefs.past_days
 	performance = gather(past_days, portfolio_select)
-	for dest in delivery.destinations(interactive, service, chat_id):
-		service = dest.dialect
-		payload, graph = report(service, past_days, portfolio_select, interactive, performance)
+	for chat in delivery.targets(dest):
+		payload, graph = report(chat.dialect, past_days, portfolio_select, interactive, performance)
 		markup = None
-		if graph and dest.takes_buttons:
+		if graph and chat.takes_buttons:
 			markup = buttons.keyboard('f', [portfolio_select or ''], buttons.period_for_days(past_days))
-		delivery.deliver(dest, delivery.Report(payload, graph, markup))
+		delivery.deliver(chat, delivery.Report(payload, graph, markup))
 
 def job(argv):
-	"""Arguments of a scheduled run: (label, lambda_handler kwargs). Raises ValueError with the usage on a bad argument."""
+	"""A scheduled run's arguments: (label, run's keyword arguments); a bad one raises ValueError with the usage."""
 	match argv:
 		case []:
 			return 'performance.py', {}
diff --git a/lib/reports/podcasts.py b/lib/reports/podcasts.py
index 1ffcd20a..3db83e6f 100644
--- a/lib/reports/podcasts.py
+++ b/lib/reports/podcasts.py
@@ -83,8 +83,9 @@ def fetch_new_episodes(feed_url, seen):
 	episodes.sort(key=lambda episode: episode["published"], reverse=True)
 	return episodes[:NEWEST], guids
 
-def lambda_handler(feed_url):
-	"""Post the feed's newest episodes not posted before, then record every new one as seen."""
+def run(dest=None, *, feed_url):
+	"""Post the feed's newest episodes not posted before to the default chats (or dest), then record every new one as
+	seen."""
 	seen_file = get_seen_filename(feed_url)
 	seen = set(state.load(seen_file, []))
 	try:
@@ -93,20 +94,20 @@ def lambda_handler(feed_url):
 		raise RuntimeError(f"fetching feed {feed_url}: {e}")
 	if not guids:
 		return
-	for dest in delivery.default_chats():
-		service = dest.dialect
+	for chat in delivery.targets(dest):
+		service = chat.dialect
 		payload = []
 		for episode in episodes:
 			podcast_name = episode['podcast']
 			title = episode['title'].removeprefix(podcast_name).lstrip(string.punctuation + " ")
 			duration = f" {episode['duration']}" if episode['duration'] else ""
 			payload.append(f"📣 {podcast_name}: {service.link(episode['url'], title)}{duration} ({episode['ago']})")
-		delivery.deliver(dest, delivery.Report(payload))
+		delivery.deliver(chat, delivery.Report(payload))
 	with state.edit(seen_file, []) as recorded: # last, so that a failure above leaves these episodes for next run
 		recorded[:] = sorted(guids.union(recorded))
 
 def job(argv):
-	"""Arguments of a scheduled run: (label, lambda_handler kwargs). Raises ValueError with the usage on a bad argument."""
+	"""A scheduled run's arguments: (label, run's keyword arguments); a bad one raises ValueError with the usage."""
 	match argv:
 		case [feed_url]:
 			return 'podcasts.py ' + hashlib.md5(feed_url.encode()).hexdigest()[:8], {'feed_url': feed_url} # the label matches the seen file's name
diff --git a/lib/reports/price.py b/lib/reports/price.py
index f84ca2d6..89584f3a 100644
--- a/lib/reports/price.py
+++ b/lib/reports/price.py
@@ -164,30 +164,31 @@ def report(service, threshold=None, user='', specific_stock=None, interactive=Fa
 			payload = [f"{user}, no price movements meet threshold {threshold:g}%"]
 	return payload, graph
 
-def lambda_handler(chat_id=None, threshold=None, service=None, user='', specific_stock=None, interactive=False, midsession=False,
-		premarket=False, interday=False, days=None, close=False, top=None):
-	"""Post the price report (report() has the modes): to the chat that asked (interactive), else to the default chats,
-	with its chart and on Telegram period buttons."""
+def run(dest=None, *, threshold=None, user='', specific_stock=None, midsession=False, premarket=False, interday=False,
+		days=None, close=False, top=None):
+	"""Post the price report (report() has the modes): to the chat that asked (dest), else to the default chats, with
+	its chart and period buttons where they can go."""
+	interactive = dest is not None
 	if threshold is None:
 		threshold = config.prefs.price_percent
 	market = gather(specific_stock, days)
-	def markup(dest):
-		"""A chart's period buttons, where the destination takes buttons: one ticker's chart, and the daily and N-day
-		lists, but not the intraday ones."""
-		if not dest.takes_buttons:
+	def markup(chat):
+		"""A chart's period buttons, where the chat takes buttons: one ticker's chart, and the daily and N-day lists,
+		but not the intraday ones."""
+		if not chat.takes_buttons:
 			return None
 		if specific_stock:
 			return buttons.keyboard('p', market[0], buttons.period_for_days(days))
 		if not (midsession or premarket or close):
 			return buttons.keyboard('l', [f"{threshold:g}", str(top or 0)], buttons.period_for_days(days, 'd' if not days else None))
 		return None
-	for dest in delivery.destinations(interactive, service, chat_id):
-		service = dest.dialect
-		payload, graph = report(service, threshold, user, specific_stock, interactive, midsession, premarket, interday, days, close, top, market)
-		delivery.deliver(dest, delivery.Report(payload, graph, markup(dest) if graph else None))
+	for chat in delivery.targets(dest):
+		payload, graph = report(chat.dialect, threshold, user, specific_stock, interactive, midsession, premarket,
+			interday, days, close, top, market)
+		delivery.deliver(chat, delivery.Report(payload, graph, markup(chat) if graph else None))
 
 def job(argv):
-	"""Arguments of a scheduled run: (label, lambda_handler kwargs). Raises ValueError with the usage on a bad argument."""
+	"""A scheduled run's arguments: (label, run's keyword arguments); a bad one raises ValueError with the usage."""
 	match argv:
 		case ['midsession' | 'interday' | 'premarket' | 'close' as arg]:
 			return f'price.py {arg}', {arg: True}
diff --git a/lib/reports/rating.py b/lib/reports/rating.py
index 136e78d8..49b6cb1d 100644
--- a/lib/reports/rating.py
+++ b/lib/reports/rating.py
@@ -6,8 +6,10 @@
 from lib import watchlist
 from lib import yahoo
 
-def lambda_handler(chat_id=None, specific_stock=None, service=None, interactive=False):
-	"""Analysts' consensus ratings that changed since the last scheduled run, which records the ratings it saw."""
+def run(dest=None, *, specific_stock=None):
+	"""Post the analysts' consensus ratings that changed since the last scheduled run, which records the ratings it saw:
+	to the chat that asked (dest), else to the default chats."""
+	interactive = dest is not None
 	def prepare_rating_payload(service, market_data):
 		"""(the report in one network's markup, the ratings seen this run)."""
 		def get_emoji(old, new):
@@ -60,10 +62,9 @@ def get_emoji(old, new):
 	old = state.load('finbot_rating.json', {})
 	market_data = yahoo.fetch_details(tickers, strict=bool(specific_stock))
 
-	for dest in delivery.destinations(interactive, service, chat_id):
-		service = dest.dialect
-		payload, new = prepare_rating_payload(service, market_data)
-		delivery.deliver(dest, delivery.Report(payload))
+	for chat in delivery.targets(dest):
+		payload, new = prepare_rating_payload(chat.dialect, market_data)
+		delivery.deliver(chat, delivery.Report(payload))
 
 	if new and not interactive: # added to what is recorded: a ticker missing from this run keeps its last rating
 		with state.edit('finbot_rating.json', {}) as ratings:
diff --git a/lib/reports/reminder.py b/lib/reports/reminder.py
index 0dea764c..dd05ca12 100644
--- a/lib/reports/reminder.py
+++ b/lib/reports/reminder.py
@@ -12,12 +12,12 @@
 LINK = re.compile(r'\[([^\]]+)\]\(([^)\s]+)\)') # [text](url)
 
 
-def lambda_handler():
-	"""Post today's reminders to the default chats, if there are any."""
+def run(dest=None):
+	"""Post today's reminders to the default chats (or dest), if there are any."""
 	today = datetime.datetime.now(zoneinfo.ZoneInfo(config.prefs.timezone)).date()
 	entries = config.data_json("reminder.json", [])
-	for dest in delivery.default_chats():
-		delivery.deliver(dest, delivery.Report(report(dest.dialect, today, entries)))
+	for chat in delivery.targets(dest):
+		delivery.deliver(chat, delivery.Report(report(chat.dialect, today, entries)))
 
 
 def report(service, today, entries):
diff --git a/lib/reports/shorts.py b/lib/reports/shorts.py
index 59ebab4a..79898a66 100644
--- a/lib/reports/shorts.py
+++ b/lib/reports/shorts.py
@@ -7,10 +7,11 @@
 from lib import yahoo
 from lib import shortman
 
-def lambda_handler(chat_id=None, threshold=None, specific_stock=None, service=None, interactive=False):
+def run(dest=None, *, threshold=None, specific_stock=None):
 	"""Post the stocks followed whose short interest is over `threshold` percent (the shorts_percent preference when
-	None), or one stock's: to the chat that asked (interactive), else to the default chats. Short Man has the ASX's
-	figures, Yahoo the US's."""
+	None), or one stock's: to the chat that asked (dest), else to the default chats. Short Man has the ASX's figures,
+	Yahoo the US's."""
+	interactive = dest is not None
 	if threshold is None:
 		threshold = config.prefs.shorts_percent
 	def prepare_shorts_payload(service, market_data):
@@ -47,10 +48,9 @@ def prepare_shorts_payload(service, market_data):
 	market_data |= yahoo.fetch_details((ticker for ticker in tickers if '.' not in ticker), strict=bool(specific_stock)) # US short interest; Short Man has the ASX's
 	market_data = shortman.fetch(market_data)
 
-	for dest in delivery.destinations(interactive, service, chat_id):
-		service = dest.dialect
-		payload, graph = prepare_shorts_payload(service, market_data)
-		delivery.deliver(dest, delivery.Report(payload, graph))
+	for chat in delivery.targets(dest):
+		payload, graph = prepare_shorts_payload(chat.dialect, market_data)
+		delivery.deliver(chat, delivery.Report(payload, graph))
 
 def job(argv):
 	"""Arguments of a scheduled run: shorts.py takes none."""
diff --git a/lib/reports/trades.py b/lib/reports/trades.py
index b2098d93..4fa18374 100644
--- a/lib/reports/trades.py
+++ b/lib/reports/trades.py
@@ -12,9 +12,11 @@
 from lib import markets
 from lib import yahoo
 
-def lambda_handler(chat_id=None, days=None, service=None, user='', portfolio_select=None, interactive=False):
-	"""The buys and sells of the past `days`, in every portfolio or the one named: on a scheduled run only those not reported
-	before. Raises sharesight.PortfolioNotFound for a name that is none of them."""
+def run(dest=None, *, days=None, user='', portfolio_select=None):
+	"""Post the buys and sells of the past `days`, in every portfolio or the one named: to the chat that asked (dest),
+	else, only those not reported before, to the default chats. Raises sharesight.PortfolioNotFound for a name that is
+	none of them."""
+	interactive = dest is not None
 	if days is None:
 		days = config.prefs.past_days
 	state_file = "finbot_sharesight_trades.json"
@@ -94,12 +96,11 @@ def prepare_trade_payload(service, trades):
 	known_trades = set(state.load(state_file, []))
 	trades = [trade for portfolio in sharesight.portfolios(portfolio_select) for trade in portfolio.trades(days)]
 
-	for dest in delivery.destinations(interactive, service, chat_id):
-		service = dest.dialect
-		payload = prepare_trade_payload(service, trades)
+	for chat in delivery.targets(dest):
+		payload = prepare_trade_payload(chat.dialect, trades)
 		if interactive and not payload: # no trades, or none of them a buy or a sell
 			payload = [f"{user} No trades {formats.days_english(days, 'in the past ')}. {random.choice(words.noTradesVerb)}"]
-		delivery.deliver(dest, delivery.Report(payload))
+		delivery.deliver(chat, delivery.Report(payload))
 
 	if newtrades and not interactive:
 		with state.edit(state_file, []) as known:
diff --git a/lib/worker.py b/lib/worker.py
index 7f92ea1f..5ae95eeb 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -234,20 +234,20 @@ def on_calendar(request, m):
 	"""Upcoming earnings or ex-dividend dates."""
 	days, ticker = period_or_ticker(m.group('arg'), config.prefs.future_days)
 	event = 'ex-dividend' if m.group('kind').lower().startswith('div') else 'earnings'
-	cal.lambda_handler(request.chat_id, days, request.service, ticker, interactive=True, event=event)
+	cal.run(request.dest, days=days, specific_stock=ticker, event=event)
 
 def on_performance(request, m):
 	""".performance [period] [portfolio]: each portfolio's return, over a month by default."""
 	days, portfolio = period_and_portfolio(m.group('args'), buttons.PERIODS['m'][1]) # default view is the 1M button
 	if days > 0:
-		performance.lambda_handler(request.chat_id, days, request.service, request.user, portfolio, interactive=True)
+		performance.run(request.dest, past_days=days, user=request.user, portfolio_select=portfolio)
 
 def on_session(request, m):
 	"""Mid-session moves (.session), or pre-market ones (.premarket, also post-market)."""
 	threshold, ticker = percent_or_ticker(m.group('arg'), config.prefs.price_percent)
 	premarket = m.group('kind').lower() != 'session'
-	price.lambda_handler(request.chat_id, threshold, request.service, request.user, ticker, interactive=True,
-		midsession=not premarket, premarket=premarket)
+	price.run(request.dest, threshold=threshold, user=request.user, specific_stock=ticker, midsession=not premarket,
+		premarket=premarket)
 
 def on_price(request, m):
 	""".price [percent|SYMBOL|top|bottom] [period]: a bare number is a percent, as on the other commands; a period has its unit."""
@@ -265,18 +265,18 @@ def on_price(request, m):
 					threshold, top = 0, 10
 				elif i == 0:
 					ticker = arg.upper()
-	price.lambda_handler(request.chat_id, threshold, request.service, request.user, ticker, interactive=True,
-		interday=days is None, days=days, top=top)
+	price.run(request.dest, threshold=threshold, user=request.user, specific_stock=ticker, interday=days is None,
+		days=days, top=top)
 
 def on_shorts(request, m):
 	""".shorts [percent|SYMBOL]: the most shorted stocks followed, or one stock's short interest."""
 	threshold, ticker = percent_or_ticker(m.group(1), config.prefs.shorts_percent)
-	shorts.lambda_handler(request.chat_id, threshold, ticker, request.service, interactive=True)
+	shorts.run(request.dest, threshold=threshold, specific_stock=ticker)
 
 def on_trades(request, m):
 	""".trades [period] [portfolio]: the buys and sells, of the past day by default."""
 	days, portfolio = period_and_portfolio(m.group('args'), 1)
-	trades.lambda_handler(request.chat_id, days, request.service, request.user, portfolio, interactive=True)
+	trades.run(request.dest, days=days, user=request.user, portfolio_select=portfolio)
 
 def on_holdings(request, m):
 	""".holdings PORTFOLIO: what it holds."""
diff --git a/tests/test_finbot.py b/tests/test_finbot.py
index d6ba43d6..0de22387 100644
--- a/tests/test_finbot.py
+++ b/tests/test_finbot.py
@@ -44,7 +44,7 @@ def test_zoneinfo_offset_is_not_lmt(self):
 	def test_reminder_runs(self):
 		"""The reminder report runs with no reminders on file."""
 		with mock.patch.object(config, 'data_json', lambda name, default=None: []):
-			reminder.lambda_handler()
+			reminder.run()
 
 	def test_bot_handler(self):
 		"""A sticker is answered 200 and not logged; git_version matches what git says."""
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
index 2ed92a94..e67fa2e9 100644
--- a/tests/test_jobs.py
+++ b/tests/test_jobs.py
@@ -193,7 +193,7 @@ def test_removes_only_old_finbot_cache_files(self):
 			os.utime(path, (stamp, stamp))
 		os.mkdir(os.path.join(folder, 'finbot_dir'))
 		os.utime(os.path.join(folder, 'finbot_dir'), (old, old))
-		self.assertEqual(cleanup.lambda_handler(30), 2)
+		self.assertEqual(cleanup.run(days=30), 2)
 		self.assertEqual(sorted(os.listdir(folder)), ['finbot_dir', 'finbot_new.json', 'other_old.json'])
 
 	def test_a_temp_file_being_written_is_left_alone(self):
@@ -205,14 +205,14 @@ def test_a_temp_file_being_written_is_left_alone(self):
 		for name, stamp in (('finbot_a.json.x1.tmp', time.time() - 5), ('finbot_b.json.x2.tmp', time.time() - 40 * 86400)):
 			Path(folder, name).write_text('x')
 			os.utime(Path(folder, name), (stamp, stamp))
-		self.assertEqual(cleanup.lambda_handler(0), 1) # a flush: the stale one goes, the write in flight stays
+		self.assertEqual(cleanup.run(days=0), 1) # a flush: the stale one goes, the write in flight stays
 		self.assertEqual(os.listdir(folder), ['finbot_a.json.x1.tmp'])
 
 	def test_missing_cache_dir_is_fine(self):
 		"""No cache dir: nothing to delete, no error."""
 		self.env(cache_dir='/nonexistent/finbot-cache')
 
-		self.assertEqual(cleanup.lambda_handler(30), 0)
+		self.assertEqual(cleanup.run(days=30), 0)
 
 
 class SchedulerFlag(unittest.TestCase):
@@ -243,8 +243,8 @@ def setUp(self):
 		self.file.write_text('*/20 * * * Mon-Fri trades.py\n')
 		self.clock = Clock(at(13, 21, 3))
 		self.ran = []
-		self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', lambda **k: self.ran.append(('trades', k)))
-		self.patch(jobs.SCRIPTS['shorts'], 'lambda_handler', lambda **k: self.ran.append(('shorts', k)))
+		self.patch(jobs.SCRIPTS['trades'], 'run', lambda **k: self.ran.append(('trades', k)))
+		self.patch(jobs.SCRIPTS['shorts'], 'run', lambda **k: self.ran.append(('shorts', k)))
 
 	def scheduler(self, catchup=5):
 		"""A loaded scheduler on the fake clock that runs commands in this process, at once."""
@@ -333,7 +333,7 @@ def test_a_failing_job_is_reported_and_the_next_command_still_runs(self):
 		def boom(**k):
 			"""Fail."""
 			raise RuntimeError('boom')
-		self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', boom)
+		self.patch(jobs.SCRIPTS['trades'], 'run', boom)
 		s = self.scheduler()
 		s.run_job(s.jobs[0], at(13, 21, 20))
 		self.assertEqual(self.ran, [('shorts', {})])
@@ -344,7 +344,7 @@ def test_system_exit_inside_a_job_is_reported_not_raised(self):
 		def quit_(**k):
 			"""Exit, as a script calling sys.exit would."""
 			raise SystemExit(3)
-		self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', quit_)
+		self.patch(jobs.SCRIPTS['trades'], 'run', quit_)
 		s = self.scheduler()
 		s.run_job(s.jobs[0], at(13, 21, 20))
 		self.assertEqual(len(self.errors), 1)
@@ -353,7 +353,7 @@ def quit_(**k):
 	def test_a_job_that_hangs_is_abandoned_and_reported(self):
 		"""A job that runs past the timeout is abandoned and reported."""
 		self.patch(jobs, 'JOB_TIMEOUT', 0.05)
-		self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', lambda **k: gevent.sleep(5))
+		self.patch(jobs.SCRIPTS['trades'], 'run', lambda **k: gevent.sleep(5))
 		s = self.scheduler()
 		s.run_job(s.jobs[0], at(13, 21, 20))
 		self.assertTrue(any('still running' in e for e in self.errors))
@@ -487,7 +487,7 @@ def setUp(self):
 							 '0 2 * * * cleanup.py\n1 * * * * podcasts.py "https://example.com/feed"\n')
 		self.s = self.scheduler()
 		self.patch(jobs, 'current', self.s)
-		self.patch(jobs.SCRIPTS['price'], 'lambda_handler', lambda **k: self.ran.append(('price', k)))
+		self.patch(jobs.SCRIPTS['price'], 'run', lambda **k: self.ran.append(('price', k)))
 
 	def test_commands_are_unique_clean_and_in_order(self):
 		"""The commands offered are each once, without redirects, sorted by name with numbers in numeric order."""
@@ -525,7 +525,7 @@ def test_run_now_refuses_while_the_command_is_running_and_reports_failures(self)
 		def boom(**k):
 			"""Fail."""
 			raise RuntimeError('boom')
-		self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', boom)
+		self.patch(jobs.SCRIPTS['trades'], 'run', boom)
 		self.assertEqual(self.s.run_now('8'), ('trades.py', 'failed'))
 		self.assertTrue(any('boom' in e for e in self.errors))
 		self.assertEqual(self.s.busy, set())
@@ -539,7 +539,7 @@ def test_a_scheduled_run_skips_a_command_that_a_manual_run_is_in_the_middle_of(s
 	def test_run_now_times_out(self):
 		"""A manual run past the timeout says so."""
 		self.patch(jobs, 'JOB_TIMEOUT', 0.05)
-		self.patch(jobs.SCRIPTS['trades'], 'lambda_handler', lambda **k: gevent.sleep(5))
+		self.patch(jobs.SCRIPTS['trades'], 'run', lambda **k: gevent.sleep(5))
 		self.assertEqual(self.s.run_now('8'), ('trades.py', 'timed out'))
 		self.assertEqual(self.s.busy, set())
 
@@ -693,7 +693,8 @@ def test_a_real_child_runs_the_handler_and_reports_its_exit_status(self):
 	def test_a_job_process_survives_a_reload_signal(self):
 		"""A job process ignores SIGHUP, which systemctl kill -s HUP sends to every process of the unit."""
 		with tempfile.TemporaryDirectory() as scripts, tempfile.TemporaryDirectory() as data:
-			Path(scripts, 'sleeper.py').write_text('import time\ndef lambda_handler():\n    print("started", flush=True)\n    time.sleep(1)\n')
+			Path(scripts, 'sleeper.py').write_text(
+				'import time\ndef run():\n    print("started", flush=True)\n    time.sleep(1)\n')
 			env = {**os.environ, 'PYTHONPATH': scripts, 'DATA_DIR': data, 'telegramBotToken': '', 'slack_webhook': '', 'discord_webhook': ''}
 			proc = subprocess.Popen([sys.executable, '-m', 'lib.jobrun', 'sleeper', '{}'], cwd=ROOT, env=env, stdout=subprocess.PIPE, text=True)
 			self.addCleanup(proc.kill)
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 8b468ccc..8d789371 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -676,7 +676,7 @@ def test_scheduled_reports_to_slack_stay_text(self):
 			sent.append((dest.name, 'chart')) or {'ok': True})
 		self.patch(transports.Transport, 'text', lambda dest, lines, buttons=None:
 			sent.append((dest.name, 'text')) or [])
-		for report in (price.lambda_handler, lambda: performance.lambda_handler(past_days=7), shorts.lambda_handler):
+		for report in (price.run, lambda: performance.run(past_days=7), shorts.run):
 			with self.subTest(report=report):
 				sent.clear()
 				report()
@@ -763,7 +763,7 @@ def test_text_goes_to_the_native_endpoint_without_previews_or_pings(self):
 
 	def test_a_scheduled_report_carries_its_chart(self):
 		"""A scheduled Discord report has its chart attached and the report as its message, within 2,000 characters."""
-		price.lambda_handler()
+		price.run()
 		(url, kwargs), = self.posts
 		self.assertEqual(url, self.URL + '?wait=true')
 		payload = json.loads(kwargs['data']['payload_json'])
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 5a67ec6e..f1352be6 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -134,7 +134,7 @@ def test_the_small_caps_betas_are_fetched_together(self):
 		self.prefs(demote_volatile=True)
 
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | {'market_cap': 5e8} for t in ts})
-		price.lambda_handler()
+		price.run()
 		self.assertEqual(len(calls), 1)
 		self.assertTrue(calls[0])
 
@@ -238,17 +238,17 @@ class ScheduledReports(FinbotCase):
 
 	def test_price(self):
 		"""The scheduled price report posts."""
-		price.lambda_handler()
+		price.run()
 		self.assertTrue(self.sent)
 
 	def test_performance(self):
 		"""The scheduled performance report posts."""
-		performance.lambda_handler(past_days=7)
+		performance.run(past_days=7)
 		self.assertTrue(self.sent)
 
 	def test_shorts(self):
 		"""The scheduled shorts report posts."""
-		shorts.lambda_handler()
+		shorts.run()
 		self.assertTrue(self.sent)
 
 	def test_earnings_and_dividends(self):
@@ -256,7 +256,7 @@ def test_earnings_and_dividends(self):
 		for kwargs in ({'event': 'earnings'}, {'event': 'ex-dividend'}):
 			with self.subTest(kwargs=kwargs):
 				self.sent.clear()
-				cal.lambda_handler(**kwargs)
+				cal.run(**kwargs)
 				self.assertTrue(self.sent)
 
 	def test_earnings_skip_an_adr_whose_primary_listing_yahoo_does_not_know(self):
@@ -264,47 +264,47 @@ def test_earnings_skip_an_adr_whose_primary_listing_yahoo_does_not_know(self):
 		self.patch(cal, 'primary_listings', lambda: {'T1': 'T1.XX'})
 		self.patch(yahoo, 'fetch', lambda ts: {t: {k: v for k, v in market_entry(t).items() if not (t == 'T1' and k == 'earnings_date')} for t in ts})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {} if t == 'T1.XX' else {t: market_entry(t)})
-		cal.lambda_handler(event='earnings')
+		cal.run(event='earnings')
 		self.assertIn('Co T2 ', str(self.sent))
 		self.assertNotIn('Co T1 ', str(self.sent))
 
 	def test_trades(self):
 		""".trades run from a chat posts."""
-		trades.lambda_handler(interactive=True, service='telegram', chat_id='5')
+		trades.run(transports.TelegramBot('5'))
 		self.assertTrue(self.sent)
 
 	def test_trades_asks_yahoo_only_for_new_trades(self):
 		"""A scheduled trades run asks Yahoo only about trades it has not reported, and nothing once all are known."""
 		fetched = []
 		self.patch(yahoo, 'fetch', lambda tickers: fetched.append(set(tickers)) or {t: market_entry(t) for t in tickers})
-		trades.lambda_handler()
+		trades.run()
 		self.assertEqual(fetched, [{'T1'}])
 		fetched.clear()
-		trades.lambda_handler() # the trade is known now: nothing to report and nothing to fetch
+		trades.run() # the trade is known now: nothing to report and nothing to fetch
 		self.assertEqual(fetched, [])
 
 	def test_a_scheduled_run_keeps_the_trades_another_run_recorded_meanwhile(self):
 		"""The seen trades are read again as the run ends, so a trade another run recorded meanwhile is kept."""
 		meanwhile = lambda tickers: state.save('finbot_sharesight_trades.json', [999]) or {t: market_entry(t) for t in tickers}
 		self.patch(yahoo, 'fetch', meanwhile) # called after the seen trades were read, before the new ones are recorded
-		trades.lambda_handler()
+		trades.run()
 		known = state.load('finbot_sharesight_trades.json')
 		self.assertEqual((known[0], len(known)), (999, 2))
 
 	def test_milestone(self):
 		"""The milestone report posts."""
-		milestone.lambda_handler()
+		milestone.run()
 		self.assertTrue(self.sent)
 
 	def test_milestone_records_a_ticker_only_with_its_new_values(self):
 		"""A first sighting without 52-week values used to be recorded as 0/0, so the next run reported a record high for it."""
 		self.patch(yahoo, 'historic_high', lambda t: (150.0, 50.0)) # beyond the 52-week range (120/80): nothing new
 		self.patch(yahoo, 'fetch', lambda ts: {t: {k: v for k, v in market_entry(t).items() if not k.startswith('fiftyTwoWeek')} for t in ts})
-		milestone.lambda_handler()
+		milestone.run()
 		self.assertFalse(state.load('finbot_milestone.json'))
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts})
 		self.sent.clear()
-		milestone.lambda_handler()
+		milestone.run()
 		self.assertNotIn('Record high', str(self.sent))
 		self.assertEqual(state.load('finbot_milestone.json')['T1'], {'record_high': 150.0, 'record_low': 50.0}) # all-time
 
@@ -312,40 +312,40 @@ def test_milestone_records_are_all_time(self):
 		"""A new 52-week high below the all-time high is no record. The record used to become the 52-week high after the first
 		run, so a stock far below its peak posted a record high each time it set a new 52-week one."""
 		self.patch(yahoo, 'historic_high', lambda t: (150.0, 50.0))
-		milestone.lambda_handler() # first sighting: the 52-week range (120/80) is inside the history (150/50)
+		milestone.run() # first sighting: the 52-week range (120/80) is inside the history (150/50)
 		for high, expected in ((130.0, False), (160.0, True), (155.0, False)):
 			with self.subTest(high=high):
 				self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | {'fiftyTwoWeekHigh': high} for t in ts})
 				self.sent.clear()
-				milestone.lambda_handler()
+				milestone.run()
 				self.assertEqual('Record high' in str(self.sent), expected)
 
 	def test_a_52_week_record_kept_before_is_replaced_by_the_history(self):
 		"""A record kept in the old 52-week format is replaced by the history: no false record high."""
 		state.save('finbot_milestone.json', {f'T{i}': {'high': 110.0, 'low': 90.0} for i in range(25)})
 		self.patch(yahoo, 'historic_high', lambda t: (150.0, 50.0))
-		milestone.lambda_handler() # 120 beats the old 52-week record of 110, but not the history's 150
+		milestone.run() # 120 beats the old 52-week record of 110, but not the history's 150
 		self.assertNotIn('Record high', str(self.sent))
 
 	def test_rating(self):
 		""".rating run from a chat posts."""
-		rating.lambda_handler(interactive=True, service='telegram', chat_id='5')
+		rating.run(transports.TelegramBot('5'))
 		self.assertTrue(self.sent)
 
 	def test_rating_keeps_the_last_rating_of_a_ticker_missing_from_a_run(self):
 		"""A ticker missing from one run keeps its last rating, so its change is reported when it comes back; it used to
 		be dropped from the record."""
-		rating.lambda_handler() # records every ticker's rating
+		rating.run() # records every ticker's rating
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {} if t == 'T1' else {t: market_entry(t)})
-		rating.lambda_handler() # no T1 this time: its rating used to be dropped from the record
+		rating.run() # no T1 this time: its rating used to be dropped from the record
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | ({'recommend': 'sell', 'recommend_index': 4.5} if t == 'T1' else {})})
 		self.sent.clear()
-		rating.lambda_handler()
+		rating.run()
 		self.assertIn('Co T1 ', str(self.sent)) # so its change from buy to sell went unreported
 
 	def test_an_interactive_rating_records_nothing(self):
 		"""A rating report asked for in a chat leaves the record for the scheduled runs alone."""
-		rating.lambda_handler(interactive=True, service='telegram', chat_id='5')
+		rating.run(transports.TelegramBot('5'))
 		self.assertIsNone(state.load('finbot_rating.json'))
 
 	def test_slack_webhook_without_a_bot_token_gets_scheduled_reports(self):
@@ -362,7 +362,7 @@ def post(url, **kwargs):
 			posts.append((url.split('/')[2], 'Authorization' in kwargs.get('headers', {}), (kwargs.get('json') or {}).get('channel')))
 			return mock.Mock(status_code=200, json=lambda: {'ok': True, 'result': {'message_id': 1, 'chat': {'id': 1}}})
 		self.patch(transports.net, 'post', post)
-		for report in (price.lambda_handler, shorts.lambda_handler, milestone.lambda_handler):
+		for report in (price.run, shorts.run, milestone.run):
 			with self.subTest(report=report):
 				posts.clear()
 				report()
@@ -373,7 +373,7 @@ def test_nowhere_to_post_is_an_error(self):
 		"""A scheduled report with nowhere to post is an error that says so."""
 		self.env(webhooks={})
 		with self.assertRaisesRegex(RuntimeError, 'nowhere to post'):
-			price.lambda_handler()
+			price.run()
 
 
 class PeriodButtons(FinbotCase):
@@ -416,7 +416,7 @@ def test_no_buttons_where_their_presses_would_go_nowhere(self):
 			with self.subTest(telegramOutgoingWebhook=hook):
 				self.env(telegramOutgoingWebhook=hook)
 				self.markups.clear()
-				price.lambda_handler(**kwargs)
+				price.run(**kwargs)
 				self.assertEqual(bool(self.markups and self.markups[0]), buttons)
 
 	def press(self, data):
@@ -498,11 +498,11 @@ def item(n, date):
 				+ item(1, '<pubDate>Mon, 01 Jan 2024 00:00:00 +0000</pubDate>') + item(2, '') + item(3, '<pubDate>Tue, 02 Jan 2024 00:00:00 GMT</pubDate>')
 				+ '</channel></rss>').encode()
 		self.patch(podcasts.net, 'get', lambda url, **k: mock.Mock(content=feed, raise_for_status=lambda: None))
-		podcasts.lambda_handler('https://example.com/feed')
+		podcasts.run(feed_url='https://example.com/feed')
 		self.assertEqual([line.split(' (')[0] for line in self.sent[0][1]], ['📣 Show: <a href="https://x/3.mp3">ep 3</a> 00:10',
 						 '📣 Show: <a href="https://x/1.mp3">ep 1</a> 00:10', '📣 Show: <a href="https://x/2.mp3">ep 2</a> 00:10'])
 		self.sent.clear()
-		podcasts.lambda_handler('https://example.com/feed')
+		podcasts.run(feed_url='https://example.com/feed')
 		self.assertEqual(self.sent, []) # all seen now
 
 
diff --git a/util/record_outputs.py b/util/record_outputs.py
index bb6e607a..36ae74ce 100755
--- a/util/record_outputs.py
+++ b/util/record_outputs.py
@@ -101,7 +101,8 @@ def step(label, run):
 			self.env(webhooks=WEBHOOKS, telegramChatID='1')
 			for line in SCHEDULED:
 				label, module, kwargs = jobs.parse_command(line)[0]
-				step(f'scheduled {line}', lambda: module.lambda_handler(**kwargs))
+				run = getattr(module, 'run', None) or module.lambda_handler # lambda_handler: a commit before run()
+				step(f'scheduled {line}', lambda: run(**kwargs))
 			Path(output).write_text(json.dumps(steps, indent=1, sort_keys=True, default=str))
 
 	result = unittest.TextTestRunner(verbosity=0, stream=io.StringIO()).run(Record())

From c9635dd13ca677cb62369378e34062aaaa9e60f2 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 02:17:59 +0000
Subject: [PATCH 30/81] trades and cal: gather once, then report per network
 (R5, R6)

Both reports did their gathering inside the per-network code, so it
ran again for every destination. Each now has:
- gather(), which fetches once;
- report(service, ...), which only formats;
- run(), which delivers to each chat.

trades (R5): a Trade dataclass replaces the 9-element lists that
were sorted by position. The Yahoo lookup of each trade's market moves
into gather(), and the closure that collected the new trade ids goes:
run() records the ids of the trades it reported.

cal (R6): primary_listing() takes the lookup that reassigned
market_data inside a loop over it: an ADR's listing in finbot_adr.json,
the NZX for an ASX ticker reporting in NZD, the exchange suffix of a
foreign company's US ticker. gather() returns the dated events once.

Checks:
- The output comparison is identical in all 159 steps.
- An old/new comparison of cal on crafted listings is identical on
  Telegram and Slack. It covered an ADR, an NZD listing, an ASML-style
  suffix, a ticker's own details, and a title mismatch, an ETF and a
  far date that are left out.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/reports/cal.py    | 107 ++++++++++++++++-------------
 lib/reports/trades.py | 152 ++++++++++++++++++++----------------------
 2 files changed, 132 insertions(+), 127 deletions(-)

diff --git a/lib/reports/cal.py b/lib/reports/cal.py
index c94d3517..e0a373a7 100644
--- a/lib/reports/cal.py
+++ b/lib/reports/cal.py
@@ -21,59 +21,72 @@ def run(dest=None, *, days=None, specific_stock=None, event='earnings'):
 	interactive = dest is not None
 	if days is None:
 		days = config.prefs.future_days
-	field, heading = EVENTS[event]
-	def prepare_payload(service, market_data):
-		"""The report in one network's markup: the events within `days`, grouped by date. An ADR or a second listing
-		without a date of its own takes its primary listing's."""
-		now = datetime.datetime.now()
-		soon = now + datetime.timedelta(days=days)
-		found = [] # (when, flag, title, link)
-		for ticker in list(market_data):
-			if market_data[ticker]['quoteType'] == 'ETF':
-				continue
-			ticker_primary = ticker
-			if field not in market_data[ticker]: # an ADR or a second listing: its primary listing may have the date
-				title_orig = market_data[ticker]['profile_title']
-				if ticker in primary_listings():
-					ticker_primary = primary_listings()[ticker]
-				elif ticker.endswith('.AX') and market_data[ticker].get('financialCurrency') == 'NZD':
-					ticker_primary = ticker.replace('.AX', '.NZ') # fixes GTK.AX and SKO.AX
-				market_data = market_data | yahoo.fetch_detail(ticker_primary)
-				country = market_data.get(ticker_primary, {}).get('profile_country') # only in the details
-				if country in markets.YAHOO_COUNTRY and '.' not in ticker_primary:
-					ticker_primary = ticker_primary + '.' + markets.YAHOO_COUNTRY[country] # fixes ASML and INFY
-					market_data = market_data | yahoo.fetch_detail(ticker_primary)
-				primary = market_data.get(ticker_primary, {})
-				if primary.get('profile_title') != title_orig or field not in primary: # not the same company, or no date there either
-					continue
-			when = datetime.datetime.fromtimestamp(market_data[ticker_primary][field])
-			if now < when < soon or specific_stock:
-				found.append((when, markets.flag_from_ticker(ticker_primary), market_data[ticker]['profile_title'], links.yahoo_link(ticker, service)))
-		found.sort(key=lambda item: item[0])
-		payload = []
-		for date in sorted({when.date() for when, *_ in found}):
-			if not specific_stock:
-				payload += ["", service.bold(f"{date:%b %d}")]
-			for when, flag, title, link in found:
-				if when.date() == date:
-					payload.append(f"{flag} {date:%b %d} {title} ({link})" if specific_stock else f"{flag} {title} ({link})")
-		if payload and not specific_stock:
-			payload.insert(0, service.bold(f"{heading} {formats.days_english(days, 'for the next ')}:"))
-		elif not payload and interactive:
-			payload = [f"No events found for {tickers[0]}" if specific_stock else f"No events found {formats.days_english(days, 'for the next ')}"]
-		return payload
-
-	# MAIN #
 	tickers = [markets.transform_to_yahoo(specific_stock)] if specific_stock else watchlist.with_holdings()
+	events = gather(tickers, days, specific_stock, event)
+	for chat in delivery.targets(dest):
+		payload = report(chat.dialect, events, tickers, days, specific_stock, interactive, event)
+		answers = delivery.deliver(chat, delivery.Report(payload))
+		if event == 'earnings' and not specific_stock and hasattr(chat, 'pin'):
+			chat.pin(answers) # the earnings report stays on top, where the network can pin
+
+def gather(tickers, days, specific_stock, event):
+	"""The dates of `event` within `days` (any date for one stock), soonest first: [(when, flag, title, ticker)]. ETFs
+	are left out; an ADR or a second listing without a date of its own takes its primary listing's."""
+	field = EVENTS[event][0]
 	if event == 'earnings':
 		market_data = yahoo.fetch(tickers)
 	else: # ex-dividend dates are in Yahoo's details
 		market_data = yahoo.fetch_details(tickers, strict=bool(specific_stock))
+	now = datetime.datetime.now()
+	soon = now + datetime.timedelta(days=days)
+	found = []
+	for ticker, data in market_data.items():
+		if data['quoteType'] == 'ETF':
+			continue
+		primary, dated = (ticker, data) if field in data else primary_listing(ticker, data, field)
+		if primary is None:
+			continue
+		when = datetime.datetime.fromtimestamp(dated[field])
+		if now < when < soon or specific_stock:
+			found.append((when, markets.flag_from_ticker(primary), data['profile_title'], ticker))
+	return sorted(found, key=lambda item: item[0])
 
-	for chat in delivery.targets(dest):
-		answers = delivery.deliver(chat, delivery.Report(prepare_payload(chat.dialect, market_data)))
-		if event == 'earnings' and not specific_stock and hasattr(chat, 'pin'):
-			chat.pin(answers) # the earnings report stays on top, where the network can pin
+def primary_listing(ticker, data, field):
+	"""(the primary listing's ticker, its Yahoo details) where an ADR or a second listing without a `field` of its own
+	finds one, or (None, None): the listing in finbot_adr.json, the NZX for an ASX ticker reporting in NZD, an exchange
+	suffix for a foreign company's US ticker, else the ticker's own details. It must be the same company, with the
+	`field`."""
+	primary = primary_listings().get(ticker)
+	if primary is None:
+		nzd = ticker.endswith('.AX') and data.get('financialCurrency') == 'NZD'
+		primary = ticker.replace('.AX', '.NZ') if nzd else ticker # fixes GTK.AX and SKO.AX
+	details = yahoo.fetch_detail(primary).get(primary, {})
+	country = details.get('profile_country') # only in the details
+	if country in markets.YAHOO_COUNTRY and '.' not in primary:
+		primary = primary + '.' + markets.YAHOO_COUNTRY[country] # fixes ASML and INFY
+		details = yahoo.fetch_detail(primary).get(primary, {})
+	if details.get('profile_title') != data['profile_title'] or field not in details: # another company, or no date
+		return None, None
+	return primary, details
+
+def report(service, events, tickers, days, specific_stock, interactive, event):
+	"""The events in one network's markup, grouped by date under a heading (one stock's carry their dates instead)."""
+	heading = EVENTS[event][1]
+	payload = []
+	for date in sorted({when.date() for when, *_ in events}):
+		if not specific_stock:
+			payload += ["", service.bold(f"{date:%b %d}")]
+		for when, flag, title, ticker in events:
+			if when.date() == date:
+				link = links.yahoo_link(ticker, service)
+				day = f"{date:%b %d} " if specific_stock else '' # one stock's events carry their date
+				payload.append(f"{flag} {day}{title} ({link})")
+	if payload and not specific_stock:
+		payload.insert(0, service.bold(f"{heading} {formats.days_english(days, 'for the next ')}:"))
+	elif not payload and interactive:
+		ahead = formats.days_english(days, 'for the next ')
+		payload = [f"No events found for {tickers[0]}" if specific_stock else f"No events found {ahead}"]
+	return payload
 
 def job(argv):
 	"""A scheduled run's arguments: (label, run's keyword arguments); a bad one raises ValueError with the usage."""
diff --git a/lib/reports/trades.py b/lib/reports/trades.py
index 4fa18374..5fa6c9d4 100644
--- a/lib/reports/trades.py
+++ b/lib/reports/trades.py
@@ -1,6 +1,8 @@
 """.trades: Sharesight buys and sells; a scheduled run reports only new ones (scheduled: trades.py)."""
-import random, sys
+import dataclasses
 import datetime
+import random
+import sys
 
 from lib import config
 from lib import state
@@ -12,6 +14,23 @@
 from lib import markets
 from lib import yahoo
 
+STATE_FILE = "finbot_sharesight_trades.json" # the ids of the trades reported
+
+
+@dataclasses.dataclass(frozen=True)
+class Trade:
+	"""A buy or a sell, as the report shows it."""
+	date: datetime.date
+	id: int
+	portfolio: str
+	bought: bool
+	value: int # price times units, in the trade's own currency (not Sharesight's value, in the portfolio's)
+	currency: str
+	symbol: str
+	market: str # Yahoo's name for the exchange, when trades_use_yahoo is on
+	holding_id: str
+
+
 def run(dest=None, *, days=None, user='', portfolio_select=None):
 	"""Post the buys and sells of the past `days`, in every portfolio or the one named: to the chat that asked (dest),
 	else, only those not reported before, to the default chats. Raises sharesight.PortfolioNotFound for a name that is
@@ -19,92 +38,65 @@ def run(dest=None, *, days=None, user='', portfolio_select=None):
 	interactive = dest is not None
 	if days is None:
 		days = config.prefs.past_days
-	state_file = "finbot_sharesight_trades.json"
+	trades = gather(days, portfolio_select, set() if interactive else set(state.load(STATE_FILE, [])))
+	for chat in delivery.targets(dest):
+		payload = report(chat.dialect, trades, days, interactive)
+		if interactive and not payload: # no trades, or none of them a buy or a sell
+			when = formats.days_english(days, 'in the past ')
+			payload = [f"{user} No trades {when}. {random.choice(words.noTradesVerb)}"]
+		delivery.deliver(chat, delivery.Report(payload))
+	if trades and not interactive: # recorded once the report is out
+		with state.edit(STATE_FILE, []) as known:
+			known.extend(sorted({trade.id for trade in trades}.difference(known)))
 
-	def prepare_trade_payload(service, trades):
-		"""The report in one network's markup: the buys and sells (on a scheduled run only new ones), grouped by date,
-		each linked to Sharesight and to the stock's page."""
-		payload_staging = []
-		dates = set()
-		wanted = [] # the trades to report: buys and sells that are new (or all of them, when interactive)
-		for trade in trades:
-			if trade['transaction_type'] not in {'BUY', 'SELL'}:
-				continue
-			if not interactive and int(trade['id']) in known_trades:
-				continue
-			wanted.append(trade)
-		if not wanted: # the usual case on a scheduled run
-			return []
-		market_data = {}
+
+def gather(days, portfolio_select, known):
+	"""The buys and sells of the past `days` in every portfolio or the one named, but those whose id is `known`, by
+	date: each with its market and currency, from Yahoo with trades_use_yahoo (one fetch for all)."""
+	raw = [trade for portfolio in sharesight.portfolios(portfolio_select) for trade in portfolio.trades(days)
+		if trade['transaction_type'] in {'BUY', 'SELL'} and int(trade['id']) not in known]
+	quotes = {}
+	if raw and config.prefs.trades_use_yahoo:
+		tickers = {markets.transform_to_yahoo(trade['symbol'], trade['market']) for trade in raw}
+		try:
+			print("Fetching tickers from Yahoo: ", tickers, file=sys.stderr)
+			quotes = yahoo.fetch(tickers)
+		except Exception as e:
+			print("Warning: could not fetch Yahoo:", e, file=sys.stderr)
+	trades = []
+	for trade in raw:
+		market = trade['market']
 		if config.prefs.trades_use_yahoo:
-			tickers = {markets.transform_to_yahoo(trade['symbol'], trade['market']) for trade in wanted} # one yahoo.fetch for all
+			ticker = markets.transform_to_yahoo(trade['symbol'], market)
 			try:
-				print("Fetching tickers from Yahoo: ", tickers, file=sys.stderr)
-				market_data = yahoo.fetch(tickers)
+				market = quotes[ticker]['profile_exchange']
 			except Exception as e:
-				print("Warning: could not fetch Yahoo:", e, file=sys.stderr)
+				print("Warning: could not get market for", ticker, "from Yahoo:", e, file=sys.stderr)
+		trades.append(Trade(date=datetime.datetime.strptime(trade['transaction_date'], '%Y-%m-%d').date(),
+			id=int(trade['id']), portfolio=trade['portfolio'], bought=trade['transaction_type'] == 'BUY',
+			value=round(float(trade['price']) * float(trade['quantity'])),
+			currency=markets.currency_from_market(market) or trade['brokerage_currency_code'], symbol=trade['symbol'],
+			market=market, holding_id=str(trade['holding_id'])))
+	return sorted(trades, key=lambda trade: (trade.date, trade.id))
 
-		for trade in wanted:
-			portfolio_name = trade['portfolio'] # custom field
-			trade_id = int(trade['id'])
-			date = trade['transaction_date'] # 2023-12-30
-			symbol = trade['symbol']
-			market = trade['market']
-			action, emoji = ('bought', '💸') if trade['transaction_type'] == 'BUY' else ('sold', '🤑')
-			units = float(trade['quantity'])
-			price = float(trade['price'])
-			currency = trade['brokerage_currency_code']
-			if config.prefs.trades_use_yahoo:
-				ticker = markets.transform_to_yahoo(symbol, market)
-				try:
-					market = market_data[ticker]['profile_exchange']
-				except Exception as e:
-					print("Warning: could not get market for", ticker, "from Yahoo:", e, file=sys.stderr)
-			value = round(price * units) # not Sharesight's trade value, which it converts to the portfolio's currency
-			holding_id = str(trade['holding_id'])
-			dt_date = datetime.datetime.strptime(date, '%Y-%m-%d').date() # (2023, 12, 30)
 
-			newtrades.add(trade_id) # recorded by the main body once the report is out
-			dates.add(dt_date)
-
-			flag = markets.flag_from_market(market)
-			# lookup currency or fall back to less reliable sharesight currency
-			currency_temp = markets.currency_from_market(market)
-			if currency_temp:
-				currency = currency_temp
-			sharesight_url = "https://portfolio.sharesight.com/holdings/"
-			trade_url = sharesight_url + holding_id + '/dashboard/transactions'
-			trade_link = service.link(trade_url, action)
-			holding_link = links.finance_link(symbol, market, service)
-			payload_staging.append([dt_date, trade_id, emoji, portfolio_name, trade_link, currency, f'{value}', holding_link, flag])
-
-		payload = []
-		payload_staging.sort()
-		for date in sorted(dates):
-			if len(dates) > 1 or (interactive and days > 1):
-				human_date = date.strftime('%b %d') # Dec 30
-				payload.append("")
-				payload.append(service.bold(human_date))
-			for trade in payload_staging:
-				if date == trade[0]:
-					payload.append(' '.join(trade[2:]))
-
-		return payload
-
-	# MAIN #
-	newtrades = set()
-	known_trades = set(state.load(state_file, []))
-	trades = [trade for portfolio in sharesight.portfolios(portfolio_select) for trade in portfolio.trades(days)]
-
-	for chat in delivery.targets(dest):
-		payload = prepare_trade_payload(chat.dialect, trades)
-		if interactive and not payload: # no trades, or none of them a buy or a sell
-			payload = [f"{user} No trades {formats.days_english(days, 'in the past ')}. {random.choice(words.noTradesVerb)}"]
-		delivery.deliver(chat, delivery.Report(payload))
+def report(service, trades, days, interactive):
+	"""The trades in one network's markup, each linked to Sharesight and to the stock's page: under a heading per date
+	when there are several dates, or when a chat asked for more than a day."""
+	dates = sorted({trade.date for trade in trades})
+	payload = []
+	for date in dates:
+		if len(dates) > 1 or (interactive and days > 1):
+			payload += ["", service.bold(f"{date:%b %d}")]
+		for trade in trades:
+			if trade.date == date:
+				action, emoji = ('bought', '💸') if trade.bought else ('sold', '🤑')
+				url = f"https://portfolio.sharesight.com/holdings/{trade.holding_id}/dashboard/transactions"
+				payload.append(' '.join([emoji, trade.portfolio, service.link(url, action), trade.currency,
+					f'{trade.value}', links.finance_link(trade.symbol, trade.market, service),
+					markets.flag_from_market(trade.market)]))
+	return payload
 
-	if newtrades and not interactive:
-		with state.edit(state_file, []) as known:
-			known.extend(sorted(newtrades.difference(known)))
 
 def job(argv):
 	"""Arguments of a scheduled run: trades.py takes none."""

From 2721975206a0bea1bdedd446bf84b2681b877e4d Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 02:21:44 +0000
Subject: [PATCH 31/81] milestone, rating, shorts and podcasts: gather, then
 report per network

The rest of the scheduled reports take the same shape: gather() once,
report(service, ...) per network, run() delivers and records.
- milestone: the history lookups for first sightings ran once per
  destination; gather() makes them once and works out the record
  highs and lows. report() lists them sorted, as before.
- rating: gather() compares the ratings with the record and returns
  the changes and every rating seen; report() formats the changes.
- shorts and podcasts: report() leaves run().

The output comparison is identical in all 159 steps.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/reports/milestone.py | 102 +++++++++++++++++++--------------------
 lib/reports/podcasts.py  |  19 +++++---
 lib/reports/rating.py    | 102 ++++++++++++++++++---------------------
 lib/reports/shorts.py    |  75 ++++++++++++++--------------
 4 files changed, 145 insertions(+), 153 deletions(-)

diff --git a/lib/reports/milestone.py b/lib/reports/milestone.py
index 5fac54a7..34702dab 100644
--- a/lib/reports/milestone.py
+++ b/lib/reports/milestone.py
@@ -11,64 +11,60 @@
 # - need to handle stock splits: flush cache if we see a split in the data
 # https://finance.yahoo.com/quote/TSLA/history?period1=0&period2=1707609600&filter=split
 
-def run(dest=None):
-	"""Post the stocks followed that set an all-time high or low (their 52-week range against the record, which a first
-	sighting takes from the stock's history) to the default chats (or dest), and record the new records."""
-	def prepare_milestone_payload(service, market_data):
-		"""(the report in one network's markup, the records it changes)."""
-		payload = []
-		payloadhigh = []
-		payloadlow = []
-		changed = {} # ticker -> its records after this run
-		for ticker in tickers:
-			try:
-				oldhigh, oldlow = records[ticker]['record_high'], records[ticker]['record_low']
-			except (KeyError, TypeError): # first sighting, or a record kept before records were all-time (52-week, under 'high'/'low')
-				try:
-					oldhigh, oldlow = yahoo.historic_high(ticker) # its history so far
-				except (RuntimeError, KeyError, IndexError, ValueError):
-					continue # no usable history: nothing recorded, so a later run cannot take zeros for its record
-			try:
-				newhigh = round(market_data[ticker]['fiftyTwoWeekHigh'], 2)
-				newlow = round(market_data[ticker]['fiftyTwoWeekLow'], 2)
-			except (KeyError, ValueError):
-				continue # likewise: a ticker is recorded only with both of its new values
-			title = market_data[ticker]['profile_title']
-			ticker_link = links.finance_link(ticker, market_data[ticker]['profile_exchange'], service)
-			emoji = markets.flag_from_ticker(ticker)
-			changed[ticker] = {'record_high': max(oldhigh, newhigh), 'record_low': min(oldlow, newlow)} # all-time: a 52-week high below it is no record
-			if newhigh > oldhigh:
-				payloadhigh.append(f"{emoji} {title} ({ticker_link})")
-			if newlow < oldlow:
-				payloadlow.append(f"{emoji} {title} ({ticker_link})")
-		payloadlow.sort()
-		payloadhigh.sort()
-		if payloadhigh:
-			message = 'Record high:'
-			message = service.bold(message)
-			payload.append(message)
-			payload = payload + payloadhigh
-		if payloadlow:
-			payload.append("")
-			message = 'Record low:'
-			message = service.bold(message)
-			payload.append(message)
-			payload = payload + payloadlow
-		payload = [i[0] for i in groupby(payload)] # de-dupe white space
-		return payload, changed
+STATE_FILE = 'finbot_milestone.json' # {ticker: {'record_high': ..., 'record_low': ...}}
 
-	# MAIN #
+def run(dest=None):
+	"""Post the stocks followed that set an all-time high or low to the default chats (or dest), and record the new
+	records."""
+	market_data, highs, lows, changed = gather()
+	for chat in delivery.targets(dest):
+		delivery.deliver(chat, delivery.Report(report(chat.dialect, market_data, highs, lows)))
+	if changed:
+		with state.edit(STATE_FILE, {}) as recorded:
+			recorded.update(changed)
 
+def gather():
+	"""(the quotes, the tickers at a record high, those at a record low, {ticker: its records after this run}) of the
+	stocks followed: each one's 52-week range against its all-time record, which a first sighting takes from its
+	history."""
 	tickers = watchlist.with_holdings()
 	market_data = yahoo.fetch(tickers)
-	records, changed = state.load('finbot_milestone.json', {}), {}
-	for chat in delivery.targets(dest):
-		payload, changed = prepare_milestone_payload(chat.dialect, market_data)
-		delivery.deliver(chat, delivery.Report(payload))
+	records = state.load(STATE_FILE, {})
+	highs, lows, changed = [], [], {}
+	for ticker in tickers:
+		try:
+			oldhigh, oldlow = records[ticker]['record_high'], records[ticker]['record_low']
+		except (KeyError, TypeError): # first sighting, or a 52-week record of old, under 'high' and 'low'
+			try:
+				oldhigh, oldlow = yahoo.historic_high(ticker) # its history so far
+			except (RuntimeError, KeyError, IndexError, ValueError):
+				continue # no usable history: nothing recorded, so a later run cannot take zeros for its record
+		try:
+			newhigh = round(market_data[ticker]['fiftyTwoWeekHigh'], 2)
+			newlow = round(market_data[ticker]['fiftyTwoWeekLow'], 2)
+		except (KeyError, ValueError):
+			continue # likewise: a ticker is recorded only with both of its new values
+		# all-time: a 52-week high below the record is no record
+		changed[ticker] = {'record_high': max(oldhigh, newhigh), 'record_low': min(oldlow, newlow)}
+		if newhigh > oldhigh:
+			highs.append(ticker)
+		if newlow < oldlow:
+			lows.append(ticker)
+	return market_data, highs, lows, changed
 
-	if changed:
-		with state.edit('finbot_milestone.json', {}) as recorded:
-			recorded.update(changed)
+def report(service, market_data, highs, lows):
+	"""The records in one network's markup: the record highs, then the record lows, each list sorted."""
+	def line(ticker):
+		"""A stock's flag, name and link."""
+		data = market_data[ticker]
+		link = links.finance_link(ticker, data['profile_exchange'], service)
+		return f"{markets.flag_from_ticker(ticker)} {data['profile_title']} ({link})"
+	payload = []
+	if highs:
+		payload += [service.bold('Record high:')] + sorted(line(ticker) for ticker in highs)
+	if lows:
+		payload += ["", service.bold('Record low:')] + sorted(line(ticker) for ticker in lows)
+	return [i[0] for i in groupby(payload)] # de-dupe white space
 
 def job(argv):
 	"""Arguments of a scheduled run: milestone.py takes none."""
diff --git a/lib/reports/podcasts.py b/lib/reports/podcasts.py
index 3db83e6f..3be7da95 100644
--- a/lib/reports/podcasts.py
+++ b/lib/reports/podcasts.py
@@ -95,17 +95,20 @@ def run(dest=None, *, feed_url):
 	if not guids:
 		return
 	for chat in delivery.targets(dest):
-		service = chat.dialect
-		payload = []
-		for episode in episodes:
-			podcast_name = episode['podcast']
-			title = episode['title'].removeprefix(podcast_name).lstrip(string.punctuation + " ")
-			duration = f" {episode['duration']}" if episode['duration'] else ""
-			payload.append(f"📣 {podcast_name}: {service.link(episode['url'], title)}{duration} ({episode['ago']})")
-		delivery.deliver(chat, delivery.Report(payload))
+		delivery.deliver(chat, delivery.Report(report(chat.dialect, episodes)))
 	with state.edit(seen_file, []) as recorded: # last, so that a failure above leaves these episodes for next run
 		recorded[:] = sorted(guids.union(recorded))
 
+def report(service, episodes):
+	"""The episodes in one network's markup, each linked to its media, with its length and age."""
+	payload = []
+	for episode in episodes:
+		podcast_name = episode['podcast']
+		title = episode['title'].removeprefix(podcast_name).lstrip(string.punctuation + " ")
+		duration = f" {episode['duration']}" if episode['duration'] else ""
+		payload.append(f"📣 {podcast_name}: {service.link(episode['url'], title)}{duration} ({episode['ago']})")
+	return payload
+
 def job(argv):
 	"""A scheduled run's arguments: (label, run's keyword arguments); a bad one raises ValueError with the usage."""
 	match argv:
diff --git a/lib/reports/rating.py b/lib/reports/rating.py
index 49b6cb1d..db57ecae 100644
--- a/lib/reports/rating.py
+++ b/lib/reports/rating.py
@@ -6,69 +6,59 @@
 from lib import watchlist
 from lib import yahoo
 
+STATE_FILE = 'finbot_rating.json' # {ticker: [action, index]} as last seen
+
 def run(dest=None, *, specific_stock=None):
 	"""Post the analysts' consensus ratings that changed since the last scheduled run, which records the ratings it saw:
 	to the chat that asked (dest), else to the default chats."""
 	interactive = dest is not None
-	def prepare_rating_payload(service, market_data):
-		"""(the report in one network's markup, the ratings seen this run)."""
-		def get_emoji(old, new):
-			"""📈 when the consensus improved (its index fell: 1 is a strong buy), 📉 when it worsened, ▪️ otherwise."""
-			if old > new:
-				return '📈'
-			if old < new:
-				return '📉'
-			else:
-				return '▪️'
-		payload = []
-		new = {}
-		for ticker in tickers:
-			old_action=None
-			old_index = float()
-			try:
-				action = market_data[ticker]['recommend'].replace('_', ' ')
-				index = round(float(market_data[ticker]['recommend_index']), 1)
-				analysts = market_data[ticker]['recommend_analysts']
-			except (KeyError, ValueError):
-				continue
-			title = market_data[ticker]['profile_title']
-			ticker_link = links.finance_link(ticker, market_data[ticker]['profile_exchange'], service, brief=False)
-			new[ticker] = [action, index]
-			try:
-				old_action, old_index = old[ticker]
-			except (KeyError, TypeError, ValueError):
-				pass
-			if old_action and action != old_action:
-				emoji = get_emoji(old_index, index)
-				message = f"{service.bold(f'{old_index} {old_action}')} to {service.bold(f'{index} {action}')} ({analysts} analysts)"
-				payload.append(f"{emoji} {title} ({ticker_link}) changed from {message}")
-		payload.sort()
-		if payload:
-			if not specific_stock:
-				message = 'Consensus rating changes:'
-				message = service.bold(message)
-				payload.insert(0, message)
-		else:
-			if interactive:
-				if specific_stock:
-					payload = [f"No rating changes found for {tickers[0]}"]
-				else:
-					payload = ["No rating changes found"]
-		return payload, new
-
-	# MAIN #
-
 	tickers = [markets.transform_to_yahoo(specific_stock)] if specific_stock else watchlist.with_holdings()
-	old = state.load('finbot_rating.json', {})
-	market_data = yahoo.fetch_details(tickers, strict=bool(specific_stock))
-
+	changes, seen = gather(tickers, specific_stock)
 	for chat in delivery.targets(dest):
-		payload, new = prepare_rating_payload(chat.dialect, market_data)
-		delivery.deliver(chat, delivery.Report(payload))
+		delivery.deliver(chat, delivery.Report(report(chat.dialect, changes, tickers, specific_stock, interactive)))
+	if seen and not interactive: # added to what is recorded: a ticker missing from this run keeps its last rating
+		with state.edit(STATE_FILE, {}) as ratings:
+			ratings.update(seen)
+
+def gather(tickers, specific_stock):
+	"""(the ratings that changed since the last scheduled run, [(ticker, its details, old [action, index], new [action,
+	index], analysts)]; {ticker: [action, index]} of every rating seen)."""
+	old = state.load(STATE_FILE, {})
+	market_data = yahoo.fetch_details(tickers, strict=bool(specific_stock))
+	changes, seen = [], {}
+	for ticker in tickers:
+		try:
+			action = market_data[ticker]['recommend'].replace('_', ' ')
+			index = round(float(market_data[ticker]['recommend_index']), 1)
+			analysts = market_data[ticker]['recommend_analysts']
+		except (KeyError, ValueError):
+			continue
+		seen[ticker] = [action, index]
+		try:
+			old_action, old_index = old[ticker]
+		except (KeyError, TypeError, ValueError):
+			continue
+		if old_action and action != old_action:
+			changes.append((ticker, market_data[ticker], (old_action, old_index), (action, index), analysts))
+	return changes, seen
 
-	if new and not interactive: # added to what is recorded: a ticker missing from this run keeps its last rating
-		with state.edit('finbot_rating.json', {}) as ratings:
-			ratings.update(new)
+def report(service, changes, tickers, specific_stock, interactive):
+	"""The rating changes in one network's markup, sorted: 📈 when the consensus improved (its index fell: 1 is a strong
+	buy), 📉 when it worsened."""
+	payload = []
+	for ticker, data, (old_action, old_index), (action, index), analysts in changes:
+		emoji = '📈' if old_index > index else '📉' if old_index < index else '▪️'
+		ticker_link = links.finance_link(ticker, data['profile_exchange'], service, brief=False)
+		before, after = service.bold(f'{old_index} {old_action}'), service.bold(f'{index} {action}')
+		message = f"{before} to {after} ({analysts} analysts)"
+		payload.append(f"{emoji} {data['profile_title']} ({ticker_link}) changed from {message}")
+	payload.sort()
+	if payload:
+		if not specific_stock:
+			payload.insert(0, service.bold('Consensus rating changes:'))
+	elif interactive:
+		payload = [f"No rating changes found for {tickers[0]}" if specific_stock else "No rating changes found"]
+	return payload
 
 def job(argv):
 	"""Arguments of a scheduled run: rating.py takes none."""
diff --git a/lib/reports/shorts.py b/lib/reports/shorts.py
index 79898a66..32bdbb96 100644
--- a/lib/reports/shorts.py
+++ b/lib/reports/shorts.py
@@ -9,48 +9,51 @@
 
 def run(dest=None, *, threshold=None, specific_stock=None):
 	"""Post the stocks followed whose short interest is over `threshold` percent (the shorts_percent preference when
-	None), or one stock's: to the chat that asked (dest), else to the default chats. Short Man has the ASX's figures,
-	Yahoo the US's."""
+	None), or one stock's: to the chat that asked (dest), else to the default chats."""
 	interactive = dest is not None
 	if threshold is None:
 		threshold = config.prefs.shorts_percent
-	def prepare_shorts_payload(service, market_data):
-		"""(the report in one network's markup, its chart or None), the most shorted first."""
-		found = [] # (percent, line, chart label)
-		for ticker in tickers:
-			try:
-				short_percent = float(market_data[ticker]['short_percent'])
-			except (KeyError, ValueError):
-				continue
-			if short_percent > threshold or specific_stock:
-				code = ticker.split('.')[0]
-				url = 'https://www.shortman.com.au/stock?q=' + code if ticker.endswith('.AX') else 'https://finance.yahoo.com/quote/' + ticker + '/key-statistics?p=' + ticker
-				found.append((short_percent, f"🩳 {market_data[ticker]['profile_title']} ({service.link(url, code)}) {round(short_percent)}%", code))
-		found.sort(key=lambda row: row[0], reverse=True) # largest first, as the chart has them
-		payload, graph = [line for _, line, _ in found], None
-		if payload and not specific_stock:
-			payload.insert(0, service.bold(f'Tracked stocks shorted over {threshold:g}%'))
-			if delivery.takes_charts(service, interactive):
-				graph = render.shorts_chart([(code, percent) for percent, _, code in found], threshold)
-		elif not payload and interactive:
-			payload = [f"🩳 No short interest found for {tickers[0]}" if specific_stock else f"🩳 No stocks shorted over {threshold:g}%. Try specifying a number."]
-		return payload, graph
-
-	# MAIN #
-
-	if specific_stock:
-		ticker = markets.transform_to_yahoo(specific_stock)
-		tickers = [ticker]
-	else:
-		tickers = watchlist.with_holdings()
+	tickers, market_data = gather(specific_stock)
+	for chat in delivery.targets(dest):
+		payload, graph = report(chat.dialect, tickers, market_data, threshold, specific_stock, interactive)
+		delivery.deliver(chat, delivery.Report(payload, graph))
 
+def gather(specific_stock):
+	"""(the stocks followed, or the one, their quotes with short interest): Short Man has the ASX's figures, Yahoo's
+	details the US's."""
+	tickers = [markets.transform_to_yahoo(specific_stock)] if specific_stock else watchlist.with_holdings()
 	market_data = yahoo.fetch(tickers)
-	market_data |= yahoo.fetch_details((ticker for ticker in tickers if '.' not in ticker), strict=bool(specific_stock)) # US short interest; Short Man has the ASX's
-	market_data = shortman.fetch(market_data)
+	market_data |= yahoo.fetch_details((ticker for ticker in tickers if '.' not in ticker), strict=bool(specific_stock))
+	return tickers, shortman.fetch(market_data)
 
-	for chat in delivery.targets(dest):
-		payload, graph = prepare_shorts_payload(chat.dialect, market_data)
-		delivery.deliver(chat, delivery.Report(payload, graph))
+def report(service, tickers, market_data, threshold, specific_stock, interactive):
+	"""(the report in one network's markup, its chart or None), the most shorted first."""
+	found = [] # (percent, line, chart label)
+	for ticker in tickers:
+		try:
+			short_percent = float(market_data[ticker]['short_percent'])
+		except (KeyError, ValueError):
+			continue
+		if short_percent > threshold or specific_stock:
+			code = ticker.split('.')[0]
+			if ticker.endswith('.AX'):
+				url = 'https://www.shortman.com.au/stock?q=' + code
+			else:
+				url = 'https://finance.yahoo.com/quote/' + ticker + '/key-statistics?p=' + ticker
+			title = market_data[ticker]['profile_title']
+			found.append((short_percent, f"🩳 {title} ({service.link(url, code)}) {round(short_percent)}%", code))
+	found.sort(key=lambda row: row[0], reverse=True) # largest first, as the chart has them
+	payload, graph = [line for _, line, _ in found], None
+	if payload and not specific_stock:
+		payload.insert(0, service.bold(f'Tracked stocks shorted over {threshold:g}%'))
+		if delivery.takes_charts(service, interactive):
+			graph = render.shorts_chart([(code, percent) for percent, _, code in found], threshold)
+	elif not payload and interactive:
+		if specific_stock:
+			payload = [f"🩳 No short interest found for {tickers[0]}"]
+		else:
+			payload = [f"🩳 No stocks shorted over {threshold:g}%. Try specifying a number."]
+	return payload, graph
 
 def job(argv):
 	"""Arguments of a scheduled run: shorts.py takes none."""

From 35dbbc9d410fefae215140bbd66c049c241ff37f Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 02:32:17 +0000
Subject: [PATCH 32/81] Chart specs, drawn once and only where they go (C1, C2,
 C8, D3a)

A report now describes its bar, pie or shorts chart as a render.Chart
spec instead of drawing it. Before, two ways existed side by side:
- rankings and holdings returned dict specs;
- price, performance and shorts drew PNGs themselves, after asking
  delivery.takes_charts.
Now deliver() decides per destination and draws through render.draw().

render.draw() caches each image by a hash of its spec and of the
drawing code (render.py and the matplotlib release), so the same chart
for a second chat, a scheduled run or a period button is read back,
and a change to the drawing code never serves an old image.

- rows_chart, bar_graph and from_spec fold into draw(); hbar_chart
  draws every bar chart (C2).
- Over 40 bars keeps the first and last 20 with a gap row between
  them, like the top-and-bottom chart (D3a).
- Bar values print by a named format (percent, percent2, number), so
  a spec holds only data and can be hashed.
- Period-button rebuilds draw the spec they get back.

The output comparison is identical in all 159 steps, chart bytes
included. New tests: over 40 bars, and a spec drawn once.

The line and compare charts are still drawn in the data layer; they
come next (C10).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md                  |  5 ++-
 lib/charts.py              | 14 ++++----
 lib/delivery.py            | 12 +++----
 lib/render.py              | 74 +++++++++++++++++++++++++++++---------
 lib/reports/holdings.py    |  3 +-
 lib/reports/performance.py | 13 +++----
 lib/reports/price.py       | 11 +++---
 lib/reports/rankings.py    |  9 ++---
 lib/reports/shorts.py      |  3 +-
 tests/test_charts.py       | 31 ++++++++++++++--
 tests/test_providers.py    |  2 +-
 tests/test_reports.py      | 17 ++++-----
 12 files changed, 133 insertions(+), 61 deletions(-)

diff --git a/CLAUDE.md b/CLAUDE.md
index 90cb2d43..7606af34 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -49,7 +49,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - `lib/buttons.py`: `Button` and a chart's period buttons (each transport draws them). `lib/charts.py`: a period
   button press rebuilds its chart.
 - `lib/dialects.py`: each network's markup and message sizes. `lib/markets.py`: market tables, flags, ticker translation.
-  `lib/formats.py`: names, units, dates, percents in words. `lib/links.py`: stock links. `lib/render.py`: every chart.
+  `lib/formats.py`: names, units, dates, percents in words. `lib/links.py`: stock links. `lib/render.py`: every chart,
+  `Chart` specs and `draw()`.
 - `lib/yahoo.py lib/sharesight.py lib/shortman.py lib/simplywallst.py`: data sources, cached in `DATA_DIR/cache`.
   `sharesight.portfolios(name=None)` gives `Portfolio` objects: `holdings()`, `tickers()`, `performance(days)`,
   `trades(days)`.
@@ -77,6 +78,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   columns, tabs counting 4 (an edit quotes whole lines).
 - Log with `print(..., file=sys.stderr)`. systemd discards stdout (`StandardOutput=null`), so the bot and its reports never
   print there (a test checks); only the util/ tools, run by hand, do.
+- A report gives its chart as a `render.Chart` spec, not a drawing: `deliver()` draws it only for a chat that takes
+  charts, through `render.draw()`, which caches the image by the spec's hash (and the drawing code's).
 - A chart renderer in `lib/render.py` gets `@_drawn` (the bot draws in one native thread); it must not call another `@_drawn`
   function, or that thread waits on itself.
 - A new command: a `Command` in `COMMANDS`, whose usage, section and buttons make its `.help` lines and keyboard
diff --git a/lib/charts.py b/lib/charts.py
index b8715b05..fe526459 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -3,6 +3,7 @@
 from lib import buttons
 from lib import formats
 from lib import links
+from lib import render
 from lib import yahoo
 from lib.reports import history, performance, price
 
@@ -33,14 +34,15 @@ def build(kind, tickers, period, service):
 			raise answer.Answer(*caption)
 		return '\n'.join(caption), image
 	if kind == 'f': # performance: tickers = [portfolio name or '']
-		payload, image = performance.report(service, days or buttons.MAX_DAYS, tickers[0] or None, interactive=True)
-		if not image:
+		payload, chart = performance.report(service, days or buttons.MAX_DAYS, tickers[0] or None)
+		if not chart:
 			raise answer.Answer(*(payload or ['no performance data']))
-		return '\n'.join(payload), image
+		return '\n'.join(payload), render.draw(chart)
 	if kind == 'l': # price list: tickers = [threshold, top]
 		span = {'interday': True} if period == 'd' else {'days': days or buttons.MAX_DAYS} # 'd' is today's moves: the command's default view, not a 1-day lookback
-		payload, image = price.report(service, threshold=float(tickers[0]), interactive=True, top=int(tickers[1]) or None, **span)
-		if not image:
+		threshold, top = float(tickers[0]), int(tickers[1]) or None
+		payload, chart = price.report(service, threshold=threshold, interactive=True, top=top, **span)
+		if not chart:
 			raise answer.Answer(*(payload or ['no price data']))
-		return '\n'.join(payload), image
+		return '\n'.join(payload), render.draw(chart) if isinstance(chart, render.Chart) else chart
 	raise ValueError(f"unknown chart kind {kind}")
diff --git a/lib/delivery.py b/lib/delivery.py
index cfe55f68..2acb8843 100644
--- a/lib/delivery.py
+++ b/lib/delivery.py
@@ -13,9 +13,9 @@
 
 @dataclasses.dataclass
 class Report:
-	"""What a report says: its lines; perhaps a chart, drawn (a PNG file object) or a spec (a dict of rows, title,
-	subtitle and render.rows_chart's options, or pie=True) that deliver() draws only for a destination that takes
-	charts; and buttons, rows of buttons.Button (or, on Telegram, a keyboard markup of its own)."""
+	"""What a report says: its lines; perhaps a chart, a render.Chart that deliver() draws only for a destination that
+	takes charts (or one drawn already, a PNG file object); and buttons, rows of buttons.Button (or, on Telegram, a
+	keyboard markup of its own)."""
 	lines: list
 	chart: object = None
 	buttons: list = None
@@ -62,10 +62,10 @@ def deliver(dest, report):
 		return dest.text(report.lines, report.buttons)
 	chart, caption = report.chart, '\n'.join(map(str, report.lines))
 	chartable = dest.takes_charts and config.prefs.graph
-	if isinstance(chart, dict): # a spec: drawn only where it can go
-		if not (chartable and chart.get('rows')):
+	if isinstance(chart, render.Chart): # a spec: drawn only where it can go
+		if not (chartable and chart.rows):
 			return dest.text(report.lines)
-		chart, caption = render.from_spec(chart), caption if report.lines else chart['title']
+		chart, caption = render.draw(chart), caption if report.lines else chart.title
 	elif not chartable:
 		return dest.text(caption.split('\n'))
 	answer = dest.chart(chart, caption, report.buttons)
diff --git a/lib/render.py b/lib/render.py
index 81c08cd8..d451d07b 100644
--- a/lib/render.py
+++ b/lib/render.py
@@ -1,9 +1,34 @@
-"""Charts drawn as PNGs with matplotlib: the bot draws them in one native thread, off the gevent hub (render_off_hub)."""
+"""Charts drawn as PNGs with matplotlib: the bot draws them in one native thread, off the gevent hub (render_off_hub). A
+report describes its chart as a Chart, which draw() draws only for a chat that takes charts, and once: the image is
+cached by the spec's hash."""
+import dataclasses
 import functools
+import hashlib
+import importlib.metadata
 import io
+import json
+from pathlib import Path
 from lib import config # first: sets MPLCONFIGDIR before matplotlib is imported (lazily, by the chart functions)
+from lib import cache
 from lib import formats
 
+
+@dataclasses.dataclass(frozen=True)
+class Chart:
+	"""A chart not yet drawn, for draw(). kind 'bars': rows of (label, value), first on top, a (label, None) row a gap;
+	'pie': rows of (label, value); 'shorts': rows of (label, percent), and the threshold."""
+	kind: str
+	title: str = ''
+	subtitle: str = ''
+	rows: tuple = ()
+	value_format: str = 'percent' # bars: how a value prints, a key of VALUE_FORMATS
+	xlim: tuple = None # bars: the value axis, zoomed
+	session: str = None # bars: 'midsession' or 'premarket' tints the background
+	threshold: float = None # shorts: the outrageous line
+
+VALUE_FORMATS = {'percent': lambda v: formats.signed_percent(round(v)),
+	'percent2': lambda v: formats.signed_percent(v, 2), 'number': lambda v: f"{v:g}"}
+
 def fit_left_margin(fig, ax, pad_px=18):
 	"""Widen the left margin so the longest Y tick label is never clipped; the right edge stays put."""
 	renderer = fig.canvas.get_renderer()
@@ -395,12 +420,6 @@ def at(u, v):
 
 	return _png(fig, sky)
 
-def rows_chart(rows, title, subtitle='', **kw):
-	"""Horizontal bars for a list of (label, value). Over 40 rows keeps the first and last 20."""
-	if len(rows) > 40:
-		rows = rows[:20] + rows[-20:]
-	return hbar_chart(rows, title, subtitle, **kw)
-
 @_drawn
 def pie_chart(rows, title, subtitle=''):
 	"""Pie of (label, value) rows, largest first, with a legend of name and share. At most 8 named slices
@@ -434,12 +453,35 @@ def pie_chart(rows, title, subtitle=''):
 	_heading(fig, title, subtitle)
 	return _png(fig, bg)
 
-def from_spec(spec):
-	"""Draw a chart spec: a dict of rows, title and subtitle, and pie=True for a pie, else rows_chart's options."""
-	spec = dict(spec)
-	rows, title, subtitle = spec.pop('rows'), spec.pop('title'), spec.pop('subtitle', '')
-	return pie_chart(rows, title, subtitle) if spec.pop('pie', False) else rows_chart(rows, title, subtitle, **spec)
-
-def bar_graph(top_rows, bottom_rows, title, subtitle=''):
-	"""Top performers then, after a gap, the worst performers: the same order as the caption's two lists."""
-	return hbar_chart(list(top_rows) + [('', None)] + list(bottom_rows), title, subtitle)
+def _code_version():
+	"""What else decides how a spec looks: this file and the matplotlib release."""
+	try:
+		matplotlib = importlib.metadata.version('matplotlib')
+	except importlib.metadata.PackageNotFoundError:
+		matplotlib = ''
+	return hashlib.sha1(Path(__file__).read_bytes() + matplotlib.encode()).hexdigest()[:8]
+
+CODE_VERSION = _code_version()
+
+def draw(chart):
+	"""The chart's PNG, drawn once. The image is cached by a hash of its spec (and of the drawing code), so a second
+	chat, a scheduled run or a period button that shows the same thing reads it back; the same spec always draws the
+	same image, so it is never stale. Over 40 bars keeps the first 20 and the last 20, a gap between them."""
+	key = json.dumps([CODE_VERSION, dataclasses.asdict(chart)], sort_keys=True, default=str)
+	name = f"finbot_chart_{hashlib.sha1(key.encode()).hexdigest()[:20]}.png"
+	if config.prefs.cache and (image := cache.read_bytes(name)):
+		return image
+	if chart.kind == 'bars':
+		rows = chart.rows if len(chart.rows) <= 40 else (*chart.rows[:20], ('', None), *chart.rows[-20:])
+		image = hbar_chart(rows, chart.title, chart.subtitle, value_fmt=VALUE_FORMATS[chart.value_format],
+			xlim=chart.xlim, session=chart.session)
+	elif chart.kind == 'pie':
+		image = pie_chart(chart.rows, chart.title, chart.subtitle)
+	elif chart.kind == 'shorts':
+		image = shorts_chart(chart.rows, chart.threshold)
+	else:
+		raise ValueError(f"unknown chart kind {chart.kind!r}")
+	if config.prefs.cache:
+		cache.write_bytes(name, image)
+		image.seek(0)
+	return image
diff --git a/lib/reports/holdings.py b/lib/reports/holdings.py
index 4bbbbd76..2ed32cd4 100644
--- a/lib/reports/holdings.py
+++ b/lib/reports/holdings.py
@@ -5,6 +5,7 @@
 from lib import formats
 from lib import links
 from lib import markets
+from lib import render
 from lib import yahoo
 
 def holding_line(holding, service, text=None):
@@ -64,7 +65,7 @@ def prepare_pie(portfolio_name, by, service, max_slices=8):
 	payload += [f"{label}: {value / total * 100:.1f}%" for label, value in rows]
 	noun = {'sector': ('sector', 'sectors'), 'industry': ('industry', 'industries')}[by]
 	subtitle = f"{len(totals)} {noun[0] if len(totals) == 1 else noun[1]}"
-	return payload, {'pie': True, 'rows': rows, 'title': title, 'subtitle': subtitle}
+	return payload, render.Chart('pie', title, subtitle, tuple(rows))
 
 def prepare_watchlist(service, user, action=None, ticker=None):
 	"""The .watchlist reply: the companies watched, or one line on adding ('add') or removing ('delete') a ticker."""
diff --git a/lib/reports/performance.py b/lib/reports/performance.py
index f1bf606a..3b8e1df0 100644
--- a/lib/reports/performance.py
+++ b/lib/reports/performance.py
@@ -32,7 +32,7 @@ def benchmark(ticker, market, text, past_days, service):
 	link = links.finance_link(ticker, market, service=service, days=past_days, brief=True, text=text)
 	return f"{formats.get_emoji(percent)} {link} {formats.signed_percent(percent, 2)}", percent
 
-def report(service, past_days=None, portfolio_select=None, interactive=False, performance=None):
+def report(service, past_days=None, portfolio_select=None, performance=None):
 	"""(payload, chart) of the performance report for one service: each portfolio and the benchmarks, best first. performance:
 	gather()'s, fetched when not given."""
 	if past_days is None:
@@ -56,22 +56,19 @@ def report(service, past_days=None, portfolio_select=None, interactive=False, pe
 	lines.sort(key=lambda item: item[0], reverse=True)
 	period = formats.days_english(past_days)
 	payload = [service.bold(f"Performance over {period}")] + [line for _, line in lines]
-	graph = None
-	if delivery.takes_charts(service, interactive):
-		chart_rows.sort(key=lambda row: row[1], reverse=True) # best first
-		graph = render.rows_chart(chart_rows, "Performance", f"% change {'' if period == 'today' else 'over '}{period}", value_fmt=lambda v: formats.signed_percent(v, 2))
-	return payload, graph
+	chart_rows.sort(key=lambda row: row[1], reverse=True) # best first
+	span = f"% change {'' if period == 'today' else 'over '}{period}"
+	return payload, render.Chart('bars', "Performance", span, tuple(chart_rows), value_format='percent2')
 
 def run(dest=None, *, past_days=None, user='', portfolio_select=None):
 	"""Post each portfolio's performance over past_days (the past_days preference when None), or the one named, beside
 	the benchmarks: to the chat that asked (dest), else to the default chats, with its chart and period buttons where
 	they can go."""
-	interactive = dest is not None
 	if past_days is None:
 		past_days = config.prefs.past_days
 	performance = gather(past_days, portfolio_select)
 	for chat in delivery.targets(dest):
-		payload, graph = report(chat.dialect, past_days, portfolio_select, interactive, performance)
+		payload, graph = report(chat.dialect, past_days, portfolio_select, performance)
 		markup = None
 		if graph and chat.takes_buttons:
 			markup = buttons.keyboard('f', [portfolio_select or ''], buttons.period_for_days(past_days))
diff --git a/lib/reports/price.py b/lib/reports/price.py
index 89584f3a..782d2a7d 100644
--- a/lib/reports/price.py
+++ b/lib/reports/price.py
@@ -133,18 +133,19 @@ def report(service, threshold=None, user='', specific_stock=None, interactive=Fa
 			heading = f'Price movement {threshold:g}%, {formats.date_range_english(days, as_of)}'
 		else:
 			heading = f'{threshold:g}% at close, {on}'
-		chartable = delivery.takes_charts(service, interactive)
 		if top:
 			bottom = [service.bold(f'Bottom {top}')] + list(reversed(payload[-top:]))
 			payload = [service.bold(heading)] + payload[:top] + [''] + bottom
 			chart_top = chart_rows[:top]
 			chart_bottom = [row for row in reversed(chart_rows[-top:]) if row not in chart_top] # worst first
-			if chartable and chart_top:
-				graph = render.bar_graph(label(chart_top), label(chart_bottom), f"Top & bottom {top}", f"% change {formats.days_english(days, 'in ', 'the past ')}")
+			if chart_top: # the best, then after a gap the worst, as the caption's two lists
+				span = formats.days_english(days, 'in ', 'the past ')
+				graph = render.Chart('bars', f"Top & bottom {top}", f"% change {span}",
+					(*label(chart_top), ('', None), *label(chart_bottom)))
 		else:
 			payload.insert(0, service.bold(heading))
-			if chartable:
-				graph = render.rows_chart(label(chart_rows), heading, session='midsession' if midsession else 'premarket' if premarket else None)
+			session = 'midsession' if midsession else 'premarket' if premarket else None
+			graph = render.Chart('bars', heading, rows=tuple(label(chart_rows)), session=session)
 	elif not payload and interactive:
 		if specific_stock:
 			if midsession:
diff --git a/lib/reports/rankings.py b/lib/reports/rankings.py
index 98bcb6c4..97f2aff3 100644
--- a/lib/reports/rankings.py
+++ b/lib/reports/rankings.py
@@ -5,6 +5,7 @@
 from lib import formats
 from lib import links
 from lib import markets
+from lib import render
 from lib import yahoo
 
 def last_col(line):
@@ -62,8 +63,8 @@ def prepare_rating_payload(service, action, length=15):
 	low, high = min(r for _, r in shown), max(r for _, r in shown)
 	spread = (high - low) or 0.2
 	# zoom the axis onto the ratings so the bars fill the panel (ratings only run 1-5)
-	chart = {'rows': shown, 'title': message, 'subtitle': 'Mean analyst rating: 1 = strong buy, 5 = strong sell (axis zoomed)',
-		'value_fmt': lambda v: f"{v:g}", 'xlim': (max(0.9, low - spread * 0.9), min(5.1, high + spread * 0.25))}
+	chart = render.Chart('bars', message, 'Mean analyst rating: 1 = strong buy, 5 = strong sell (axis zoomed)',
+		tuple(shown), value_format='number', xlim=(max(0.9, low - spread * 0.9), min(5.1, high + spread * 0.25)))
 	return [service.bold(message)] + payload, chart
 
 def prepare_value_payload(service, action='pe', ticker_select=None, length=15):
@@ -142,7 +143,7 @@ def prepare_value_payload(service, action='pe', ticker_select=None, length=15):
 		else:
 			heading = f'{heading_type} {length} tracked stocks by {heading_trail} ratio{coverage}'
 			payload.insert(0, f"{service.bold(heading)}")
-			chart = {'rows': chart_rows[:length], 'title': heading, 'value_fmt': lambda v: f"{v:g}"}
+			chart = render.Chart('bars', heading, rows=tuple(chart_rows[:length]), value_format='number')
 	return payload, chart
 
 def prepare_beta_payload(service):
@@ -159,5 +160,5 @@ def prepare_beta_payload(service):
 		return ['No tracked stocks have a beta over 1.5 and mkt cap under US$2B'], None
 	found.sort(key=lambda row: row[0], reverse=True)
 	heading = 'Beta over 1.5 and mkt cap under US$2B'
-	chart = {'rows': [row for _, _, row in found], 'title': heading, 'value_fmt': lambda v: f"{v:g}"}
+	chart = render.Chart('bars', heading, rows=tuple(row for _, _, row in found), value_format='number')
 	return [service.bold(heading)] + [line for _, line, _ in found], chart
diff --git a/lib/reports/shorts.py b/lib/reports/shorts.py
index 32bdbb96..99927af6 100644
--- a/lib/reports/shorts.py
+++ b/lib/reports/shorts.py
@@ -46,8 +46,7 @@ def report(service, tickers, market_data, threshold, specific_stock, interactive
 	payload, graph = [line for _, line, _ in found], None
 	if payload and not specific_stock:
 		payload.insert(0, service.bold(f'Tracked stocks shorted over {threshold:g}%'))
-		if delivery.takes_charts(service, interactive):
-			graph = render.shorts_chart([(code, percent) for percent, _, code in found], threshold)
+		graph = render.Chart('shorts', rows=tuple((code, percent) for percent, _, code in found), threshold=threshold)
 	elif not payload and interactive:
 		if specific_stock:
 			payload = [f"🩳 No short interest found for {tickers[0]}"]
diff --git a/tests/test_charts.py b/tests/test_charts.py
index 94864ef4..9b4c5d0d 100644
--- a/tests/test_charts.py
+++ b/tests/test_charts.py
@@ -23,6 +23,12 @@
 MIXED = [('AAA', 42), ('BBB', 17), ('CCC', 3), ('', None), ('DDD', -5), ('EEE', -30)]
 
 
+def uncached(chart):
+	"""render.draw(chart) without the cache, for the tests that have no fixture's temporary cache dir."""
+	with mock.patch.object(config, 'prefs', dataclasses.replace(config.prefs, cache=False)):
+		return render.draw(chart)
+
+
 def decode(buf):
 	"""A PNG buffer as an RGB image."""
 	buf.seek(0)
@@ -93,7 +99,7 @@ def test_edge_case_rows(self):
 		for name, rows in EDGE_ROWS.items():
 			with self.subTest(rows=name):
 				self.check_image(render.hbar_chart(rows, 'Title', 'sub'), name)
-				self.check_image(render.rows_chart(rows, 'Title'), name + '/rows_chart')
+				self.check_image(uncached(render.Chart('bars', 'Title', rows=tuple(rows))), name + '/drawn')
 
 	def test_zoomed_axis(self):
 		"""A bar chart with a narrow x range (ratings, 1 to 5) draws sane images, wide labels and many rows included."""
@@ -114,7 +120,8 @@ def test_shorts_chart_sizes_and_verdicts(self):
 
 	def test_bar_graph_has_top_and_bottom(self):
 		"""The top-and-bottom bar chart draws a sane image."""
-		self.check_image(render.bar_graph([('AAA', 9)], [('ZZZ', -9)], 'Top and bottom'), 'bar_graph')
+		self.check_image(uncached(render.Chart('bars', 'Top and bottom', rows=(('AAA', 9), ('', None), ('ZZZ', -9)))),
+			'top and bottom')
 
 	def test_flat_and_tiny_price_series(self):
 		"""A flat price line and one of two points draw sane images."""
@@ -135,7 +142,8 @@ def setUp(self):
 	def test_charts_asked_for_at_once_all_finish(self):
 		"""Six charts asked for at once all finish: one render thread, as several deadlock matplotlib under gevent."""
 		jobs = [gevent.spawn(render.pie_chart, [('a', 1.0), ('b', 2.0)], f't{i}') for i in range(4)]
-		jobs += [gevent.spawn(render.hbar_chart, MIXED, 'T'), gevent.spawn(render.rows_chart, MIXED, 'T')]
+		jobs += [gevent.spawn(render.hbar_chart, MIXED, 'T')]
+		jobs += [gevent.spawn(uncached, render.Chart('bars', 'T', rows=tuple(MIXED)))]
 		gevent.joinall(jobs, timeout=120)
 		self.assertTrue(all(job.successful() for job in jobs), 'a chart did not finish (several native threads deadlock matplotlib)')
 		self.assertTrue(all(decode(job.value).size == (1280, 800) for job in jobs))
@@ -269,6 +277,23 @@ def spy(rows, title, subtitle='', **kw):
 		self.assertTrue(captured, 'no chart was drawn')
 		return captured[0]
 
+	def test_over_forty_bars_keep_both_ends_with_a_gap(self):
+		"""A bar chart of over 40 rows keeps the first 20 and the last 20, a gap between them."""
+		captured = []
+		self.patch(render, 'hbar_chart', lambda rows, *a, **k: captured.append(rows) or io.BytesIO(b'png'))
+		render.draw(render.Chart('bars', 'T', rows=tuple((f'T{i}', i) for i in range(50))))
+		rows = captured[0]
+		self.assertEqual((len(rows), rows[20], rows[0][0], rows[-1][0]), (41, ('', None), 'T0', 'T49'))
+
+	def test_a_chart_is_drawn_once(self):
+		"""The same spec, for another chat or a later run, is read back from the cache, not drawn again."""
+		calls, real = [], render.pie_chart
+		self.patch(render, 'pie_chart', lambda *a: calls.append(a) or real(*a))
+		chart = render.Chart('pie', 'T', rows=(('a', 1.0), ('b', 2.0)))
+		first, second = render.draw(chart).read(), render.draw(chart).read()
+		self.assertEqual((len(calls), first[:4]), (1, b'\x89PNG'))
+		self.assertEqual(first, second)
+
 	def test_price_top_lists_best_first_then_worst(self):
 		""".price top: the ten best, best first, a spacer, then the ten worst, worst first, as in the caption."""
 		chart = self.run_report(lambda: self.command('.price top 7d'))
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 8d789371..d5dcf63d 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -689,7 +689,7 @@ def test_slack_without_a_bot_token_gets_text_and_no_chart_is_drawn(self):
 		def refuse(*a, **k):
 			"""Fail: no chart should be drawn."""
 			raise AssertionError('a chart was drawn for Slack')
-		for name in ('rows_chart', 'pie_chart', 'bar_graph', 'shorts_chart'):
+		for name in ('draw', 'hbar_chart', 'pie_chart', 'shorts_chart'):
 			self.patch(render, name, refuse)
 		self.env(webhooks={**config.env.webhooks, 'slack': 'https://hooks.slack.com/services/x'})
 		for cmd in ('.beta', '.pie Rob', '.buy', '.sell', '.peg', '.price', '.price top', '.price T1 7d', '.performance 7d', '.shorts', '.history T1', '.compare T1 T2'):
diff --git a/tests/test_reports.py b/tests/test_reports.py
index f1352be6..8aac71c6 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -110,7 +110,7 @@ def test_ratings_list_the_strongest_first(self):
 		def order(action):
 			"""(the tickers listed, the tickers charted) of a ratings report, in order."""
 			payload, chart = rankings.prepare_rating_payload(TELEGRAM, action)
-			return [line.split('Co ')[1].split(' ')[0] for line in payload[1:]], [label for label, _ in chart['rows']]
+			return [line.split('Co ')[1].split(' ')[0] for line in payload[1:]], [label for label, _ in chart.rows]
 		self.assertEqual(order('buy'), (['T2', 'T3', 'T1'], ['T2', 'T3', 'T1']))
 		self.assertEqual(order('sell'), (['T5', 'T6', 'T4'], ['T5', 'T6', 'T4']))
 		self.assertIn('(7+ analysts)', rankings.prepare_rating_payload(TELEGRAM, 'buy')[0][0])
@@ -149,7 +149,8 @@ def listed(action):
 			payload, chart = rankings.prepare_value_payload(TELEGRAM, action)
 			return payload[0], [line.split('Co ')[1].split(' ')[0] for line in payload[1:]], chart
 		heading, names, chart = listed('forward pe')
-		self.assertEqual((names, [label for label, _ in chart['rows']]), (['T1'], ['T1'])) # T2 and T4: too few; T3: negative
+		# T2 and T4: too few analysts; T3: negative
+		self.assertEqual((names, [label for label, _ in chart.rows]), (['T1'], ['T1']))
 		self.assertIn('forward pe ratio (7+ analysts)', heading)
 		self.assertEqual(listed('negative forward pe')[1], ['T3'])
 		self.assertIn('Co T2', str(rankings.prepare_value_payload(TELEGRAM, 'forward pe', 'T2'))) # one ticker asked for: shown
@@ -190,7 +191,7 @@ def test_prepare_beta(self):
 		"""The beta report charts a bar per stock listed, and says so when none qualify."""
 		payload, chart = rankings.prepare_beta_payload(TELEGRAM)
 		self.assertIn('Beta over 1.5', payload[0])
-		self.assertEqual(len(chart['rows']), len(payload) - 1) # a bar per stock listed
+		self.assertEqual(len(chart.rows), len(payload) - 1) # a bar per stock listed
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'beta': 1.0}})
 		self.assertEqual(rankings.prepare_beta_payload(TELEGRAM), (['No tracked stocks have a beta over 1.5 and mkt cap under US$2B'], None))
 
@@ -527,7 +528,7 @@ def test_by_industry_default(self):
 	def test_by_sector_and_payload(self):
 		""".pie by sector: the slices in order, the unclassified ones together, worthless holdings left out."""
 		payload, chart = holdings.prepare_pie('rob', 'sector', TELEGRAM)
-		self.assertEqual([r[0] for r in chart['rows']], ['Electronic Technology', 'Health Technology', 'Unclassified'])
+		self.assertEqual([r[0] for r in chart.rows], ['Electronic Technology', 'Health Technology', 'Unclassified'])
 		self.assertIn('Electronic Technology: 72.7%', payload) # 800 of 1100; zero-value Gold is left out
 		self.assertEqual(self.command('.pie Rob sector')[0][0], 'photo')
 
@@ -536,9 +537,9 @@ def test_folds_into_other_and_renders(self):
 		held = {f'T{i}:X': {'code': f'T{i}', 'industry_classification_name': f'Ind {i}', 'value': 100 - i} for i in range(12)}
 		self.patch(sharesight.Portfolio, 'holdings', lambda portfolio: held)
 		payload, chart = holdings.prepare_pie('Rob', 'industry', TELEGRAM)
-		self.assertEqual(len(chart['rows']), 8)
-		self.assertEqual(chart['rows'][-1][0], 'Other')
-		self.assertTrue(render.pie_chart(chart['rows'], chart['title'], chart['subtitle']).read().startswith(b'\x89PNG'))
+		self.assertEqual(len(chart.rows), 8)
+		self.assertEqual(chart.rows[-1][0], 'Other')
+		self.assertTrue(render.pie_chart(chart.rows, chart.title, chart.subtitle).read().startswith(b'\x89PNG'))
 
 	def test_unknown_or_missing_portfolio_lists_known(self):
 		""".pie with no portfolio, or an unknown one, lists the portfolios there are."""
@@ -550,7 +551,7 @@ def test_unknown_or_missing_portfolio_lists_known(self):
 	def test_a_service_that_takes_no_images_gets_text(self):
 		"""Slack without the bot token takes no images: the report goes as text."""
 		self.env(slackBotToken=False)
-		worker.Request(SLACK, 'C1', '@u', 'U', BOT).reply(['x'], {'pie': True, 'rows': [('a', 1.0)], 'title': 't'})
+		worker.Request(SLACK, 'C1', '@u', 'U', BOT).reply(['x'], render.Chart('pie', 't', rows=(('a', 1.0),)))
 		self.assertEqual(self.sent[-1][0], 'text')
 
 

From 13ff98238ef8989b520bc62df6cac3baf49e83f3 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 02:39:36 +0000
Subject: [PATCH 33/81] Line and compare charts become specs too; drawing
 leaves Yahoo (C10, R13)

Price history no longer draws:
- yahoo.price_history(ticker, days) returns the changes and a Line
  (company, exchange, currency, price, closes), and loses its graph
  and graphCache flags, its PNG cache and its render import.
- history.line_chart() turns a Line into a 'line' Chart.
- prepare_compare returns a 'compare' Chart, and its image and
  caption cache files go: render.draw()'s cache by spec covers it.

Period-button rebuilds return specs, which worker draws before editing
the message, and delivery.takes_charts goes, since no report draws its
own chart now. A Slack reply without the bot token draws nothing at
all, the line and compare charts included (the test now refuses every
renderer).

R13: .history takes its title and its exchange link from the chart it
loads anyway (Yahoo's meta has both), instead of a 10-module details
request. A ticker Yahoo has no chart for answers "no price history for
X" (from NoData), where it said ".history: no data found for ticker X".

C6: render.graph's ylabel parameter, which was the currency, is now
called currency.

Checks:
- The output comparison is identical in all 159 steps, compare chart
  bytes included.
- HEAD's price_history chart and the new spec route give identical
  tables and PNG bytes on an 800-day history with gaps, for the whole
  history and for 7 and 30 days.
- The README says how charts are cached.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 README.md                  |  2 ++
 lib/charts.py              | 31 +++++++++++-----------
 lib/delivery.py            |  8 ------
 lib/render.py              | 17 +++++++++---
 lib/reports/history.py     | 54 ++++++++++++++++++--------------------
 lib/reports/performance.py |  2 +-
 lib/reports/price.py       |  9 ++++---
 lib/reports/profile.py     |  2 +-
 lib/worker.py              |  4 ++-
 lib/yahoo.py               | 41 ++++++++++++++---------------
 tests/common.py            |  3 ++-
 tests/test_edges.py        |  3 ++-
 tests/test_providers.py    | 27 +++++++++----------
 13 files changed, 103 insertions(+), 100 deletions(-)

diff --git a/README.md b/README.md
index 1489a842..019b59c7 100644
--- a/README.md
+++ b/README.md
@@ -574,6 +574,8 @@ performance_cache_seconds=1800
 
 `performance_cache_seconds` (default 30 minutes) is the cache lifetime of Sharesight performance reports.
 
+Charts are cached by what they show (with `cache=True`): the same chart, for another chat, a later run or a period button, is drawn once, and a chart of new figures is a new chart, so a cached one is never stale. The scheduled `cleanup.py` removes cache files after 30 days.
+
 To flush the cache, so that everything is fetched afresh, signal the running bot: it deletes its cache files (Yahoo and Sharesight
 answers, charts, the Yahoo cookie) and keeps its state (watchlist, seen trades, ratings, schedule), then refills the Sharesight caches
 at once. There is no `systemctl flush`, since systemctl takes no custom verbs; any of these does it:
diff --git a/lib/charts.py b/lib/charts.py
index fe526459..0080a966 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -3,46 +3,47 @@
 from lib import buttons
 from lib import formats
 from lib import links
-from lib import render
 from lib import yahoo
 from lib.reports import history, performance, price
 
 def build(kind, tickers, period, service):
-	"""(caption, image) for a chart kind at a button period; period None = the command's default view."""
+	"""(caption, chart spec or None) for a chart kind at a button period; period None = the command's default view."""
 	days = buttons.PERIODS[period][1] if period else None
 	if kind == 'h':
 		ticker = tickers[0]
-		lines, table, image = history.history_payload(ticker, service, buttons.PERIODS[period or 'x'][0], graph=not days) # default view is the full history
-		if table is None:
-			raise answer.Answer(*lines)
+		highlight = buttons.PERIODS[period or 'x'][0]
+		lines, _, chart = history.history_payload(ticker, service, highlight, graph=not days) # default view: it all
 		if days:
-			_, image = yahoo.price_history(ticker, days)
-		return '\n'.join(lines), image
+			changes, line = yahoo.price_history(ticker, days)
+			chart = history.line_chart(ticker, changes, line, days)
+		return '\n'.join(lines), chart
 	if kind == 'p':
 		ticker = tickers[0]
 		market_data = yahoo.fetch([ticker])
 		if ticker not in market_data:
 			raise answer.Answer(f"no data found for {ticker}")
-		percent, image = yahoo.price_history(ticker, days, graphCache=False)
-		percent = float(percent.get(days, percent.get('Max', next(iter(percent.values()), 0))))
+		changes, line = yahoo.price_history(ticker, days)
+		percent = float(changes.get(days, changes.get('Max', next(iter(changes.values()), 0))))
 		link = links.quote_link(ticker, market_data[ticker]['profile_exchange'], service, days=days)
-		return f"{formats.get_emoji(percent)} {market_data[ticker]['profile_title']} ({link}) {formats.signed_percent(round(percent))}", image
+		title = market_data[ticker]['profile_title']
+		caption = f"{formats.get_emoji(percent)} {title} ({link}) {formats.signed_percent(round(percent))}"
+		return caption, history.line_chart(ticker, changes, line, days)
 	if kind == 'c':
 		args = list(tickers) + ([f'{days}d'] if days else ['max'])
-		caption, image, *_ = history.prepare_compare(service, args)
-		if not image:
+		caption, chart, *_ = history.prepare_compare(service, args)
+		if not chart:
 			raise answer.Answer(*caption)
-		return '\n'.join(caption), image
+		return '\n'.join(caption), chart
 	if kind == 'f': # performance: tickers = [portfolio name or '']
 		payload, chart = performance.report(service, days or buttons.MAX_DAYS, tickers[0] or None)
 		if not chart:
 			raise answer.Answer(*(payload or ['no performance data']))
-		return '\n'.join(payload), render.draw(chart)
+		return '\n'.join(payload), chart
 	if kind == 'l': # price list: tickers = [threshold, top]
 		span = {'interday': True} if period == 'd' else {'days': days or buttons.MAX_DAYS} # 'd' is today's moves: the command's default view, not a 1-day lookback
 		threshold, top = float(tickers[0]), int(tickers[1]) or None
 		payload, chart = price.report(service, threshold=threshold, interactive=True, top=top, **span)
 		if not chart:
 			raise answer.Answer(*(payload or ['no price data']))
-		return '\n'.join(payload), render.draw(chart) if isinstance(chart, render.Chart) else chart
+		return '\n'.join(payload), chart
 	raise ValueError(f"unknown chart kind {kind}")
diff --git a/lib/delivery.py b/lib/delivery.py
index 2acb8843..ce496b66 100644
--- a/lib/delivery.py
+++ b/lib/delivery.py
@@ -46,14 +46,6 @@ def targets(dest):
 	return [dest] if dest else default_chats()
 
 
-def takes_charts(service, reply=True):
-	"""Whether a report to `service` can carry a chart: never with the graph preference off; else as its transport says,
-	a reply going into the chat that asked and a scheduled report to the default chats. For the reports that draw their
-	own charts."""
-	kind = (REPLIES if reply else SCHEDULED).get(service)
-	return bool(config.prefs.graph and kind and kind(None).takes_charts)
-
-
 def deliver(dest, report):
 	"""Post a report: its chart, the lines as its caption, where the destination takes charts and the graph preference
 	is on; else the lines, buttons under the last message. A chart the network refuses goes as text without its buttons,
diff --git a/lib/render.py b/lib/render.py
index d451d07b..9c911c0c 100644
--- a/lib/render.py
+++ b/lib/render.py
@@ -16,7 +16,9 @@
 @dataclasses.dataclass(frozen=True)
 class Chart:
 	"""A chart not yet drawn, for draw(). kind 'bars': rows of (label, value), first on top, a (label, None) row a gap;
-	'pie': rows of (label, value); 'shorts': rows of (label, percent), and the threshold."""
+	'pie': rows of (label, value); 'shorts': rows of (label, percent), and the threshold; 'line': rows of (date, close),
+	a price line, and the unit (its currency); 'compare': rows of (label, ((date, close), ...)), two to six price
+	lines."""
 	kind: str
 	title: str = ''
 	subtitle: str = ''
@@ -25,6 +27,7 @@ class Chart:
 	xlim: tuple = None # bars: the value axis, zoomed
 	session: str = None # bars: 'midsession' or 'premarket' tints the background
 	threshold: float = None # shorts: the outrageous line
+	unit: str = '' # line: the prices' currency
 
 VALUE_FORMATS = {'percent': lambda v: formats.signed_percent(round(v)),
 	'percent2': lambda v: formats.signed_percent(v, 2), 'number': lambda v: f"{v:g}"}
@@ -108,7 +111,7 @@ def _png(fig, facecolor):
 	return buf
 
 @_drawn
-def graph(df, title, ylabel):
+def graph(df, title, currency):
 	"""Render a price chart as PNG bytes, sized for Telegram (1280px wide, no server-side resampling)."""
 	from matplotlib.ticker import MaxNLocator, FuncFormatter
 	import matplotlib.dates as mdates
@@ -155,7 +158,7 @@ def annotate(i, name, above):
 	if imin != len(y) - 1 and imin != imax:
 		annotate(imin, 'Low', False)
 
-	_heading(fig, title, f"{last:,.2f} {ylabel or ''}   {arrow} {abs(pct):.2f}%  over period", color)
+	_heading(fig, title, f"{last:,.2f} {currency or ''}   {arrow} {abs(pct):.2f}%  over period", color)
 
 	return _png(fig, bg)
 
@@ -479,6 +482,14 @@ def draw(chart):
 		image = pie_chart(chart.rows, chart.title, chart.subtitle)
 	elif chart.kind == 'shorts':
 		image = shorts_chart(chart.rows, chart.threshold)
+	elif chart.kind == 'line':
+		import pandas as pd
+		image = graph(pd.DataFrame(list(chart.rows), columns=['Date', 'Close']), chart.title, chart.unit)
+	elif chart.kind == 'compare':
+		import pandas as pd
+		series = [(label, pd.Series([close for _, close in points], index=pd.to_datetime([day for day, _ in points])))
+			for label, points in chart.rows]
+		image = compare_graph(series, chart.title, chart.subtitle)
 	else:
 		raise ValueError(f"unknown chart kind {chart.kind!r}")
 	if config.prefs.cache:
diff --git a/lib/reports/history.py b/lib/reports/history.py
index e9c44a81..04208b37 100644
--- a/lib/reports/history.py
+++ b/lib/reports/history.py
@@ -1,9 +1,6 @@
 """Price history: .history (one stock over every period) and .compare (two to six stocks, rebased)."""
-import hashlib
 import datetime
 from lib import answer
-from lib import config
-from lib import cache
 from lib import formats
 from lib import links
 from lib import markets
@@ -12,7 +9,8 @@
 
 def prepare_compare(service, args):
 	"""Compare 2-6 tickers as rebased % change. A trailing time period (3m/1y/90d/ytd) is optional;
-	without one, the span is the earliest date covered by every ticker. Returns (caption lines, image, tickers, period days)."""
+	without one, the span is the earliest date covered by every ticker. Returns (caption lines, chart spec, tickers,
+	period days)."""
 	import pandas as pd
 	usage = ['Usage: .compare SYMBOL SYMBOL [SYMBOL...up to 6] [period|max]  (default: 10 years)']
 	period_days = None
@@ -31,14 +29,6 @@ def prepare_compare(service, args):
 	tickers = list(dict.fromkeys(markets.transform_to_yahoo(a.upper()) for a in args)) # de-duplicated, order kept
 	if not 2 <= len(tickers) <= 6:
 		return usage, None, tickers, period_days
-	spec = period_days or ('max' if full_history else 'default')
-	cache_key = hashlib.sha1(f"{service}|{','.join(tickers)}|{spec}".encode()).hexdigest()[:16]
-	image_cache_file, caption_cache_file = f"finbot_compare_{cache_key}.png", f"finbot_compare_{cache_key}.json"
-	if config.prefs.cache:
-		cached_image = cache.read_bytes(image_cache_file)
-		cached_caption = cache.read(caption_cache_file)
-		if cached_image and cached_caption:
-			return cached_caption, cached_image, tickers, period_days
 	names, exchanges, closes = {}, {}, {}
 	for ticker, history in zip(tickers, yahoo.fetch_each(yahoo.price_series, tickers)): # downloaded at the same time
 		if isinstance(history, Exception):
@@ -71,29 +61,22 @@ def prepare_compare(service, args):
 	by_performance = sorted(cropped, key=lambda item: item[1].iloc[-1] / item[1].iloc[0], reverse=True)
 	title = ' vs '.join(t.split('.')[0] for t, _ in by_performance) # best first; line colours stay tied to input order
 	subtitle = f"% change over {period_label}, since {start:%d %b %Y}" + note
-	image = render.compare_graph([(t.split('.')[0], series) for t, series in cropped], title, subtitle)
-	image.seek(0)
+	lines = tuple((t.split('.')[0], tuple((f"{day:%Y-%m-%d}", float(close)) for day, close in series.items()))
+		for t, series in cropped)
+	chart = render.Chart('compare', title, subtitle, lines)
 	caption = [service.bold(f"{period_label}, since {start:%d %b %Y}") + note]
 	results = sorted(((series.iloc[-1] / series.iloc[0] - 1) * 100, t) for t, series in cropped)
 	for pct, t in reversed(results):
 		link = links.finance_link(t, exchanges[t], service, days=span_days, brief=False)
 		caption.append(f"{formats.get_emoji(pct)} {names[t]} ({link}): {formats.signed_percent(pct, 1)}")
-	if config.prefs.cache:
-		cache.write_bytes(image_cache_file, image)
-		cache.write(caption_cache_file, caption)
-		image.seek(0)
-	return caption, image, tickers, period_days
+	return caption, chart, tickers, period_days
 
 def history_payload(ticker, service, highlight=None, graph=False):
-	"""Caption lines for .history: performance table, with the selected interval in bold. Returns (lines, percent table, image):
-	the full history's chart with graph=True (if the graph preference is on), else None."""
-	market_data = yahoo.fetch_detail(ticker, 600)
-	title = market_data.get(ticker, {}).get('profile_title', '')
-	ticker_link = links.finance_link(ticker, market_data.get(ticker, {}).get('profile_exchange', ''), service, days=1825, brief=False)
-	if ticker not in market_data or 'percent_change' not in market_data[ticker]:
-		return [f".history: no data found for ticker {ticker}"], None, None
-	table, image = yahoo.price_history(ticker) if graph else yahoo.price_history(ticker, graph=False)
-	lines = [service.bold(f"{title} ({ticker_link}) performance history")]
+	"""(lines, percent table, chart): .history's caption, the change over every period with the `highlight` one marked,
+	and with graph=True the line chart of the whole history (else None). The name and exchange come from the chart."""
+	table, line = yahoo.price_history(ticker)
+	ticker_link = links.finance_link(ticker, line.exchange, service, days=1825, brief=False)
+	lines = [service.bold(f"{line.name} ({ticker_link}) performance history")]
 	for interval in ('Max', '10Y', '5Y', '3Y', '1Y', 'YTD', '6M', '3M', '1M', '7D', '1D'):
 		if interval in table:
 			percent = table[interval]
@@ -101,4 +84,17 @@ def history_payload(ticker, service, highlight=None, graph=False):
 			if interval == highlight:
 				label = '● ' + label
 			lines.append(f"{formats.get_emoji(percent)} {label} {formats.signed_percent(percent)}")
-	return lines, table, image
+	return lines, table, line_chart(ticker, table, line) if graph else None
+
+def line_chart(ticker, changes, line, days=None):
+	"""The line chart of a price_history (its changes and Line), titled with the company, the change over the span and
+	the last price; None for a span under 2 days or fewer than 2 closes."""
+	if len(line.closes) < 2 or (days and days < 2):
+		return None
+	if days:
+		span, change = f"{min(days, 3655)} days", changes[days]
+	else:
+		span = "Max" if "Max" in changes else "10Y"
+		change = changes[span]
+	title = f"{line.name} ({ticker}) {span} {formats.signed_percent(change)} Last: {line.price}"
+	return render.Chart('line', title, rows=line.closes, unit=line.currency)
diff --git a/lib/reports/performance.py b/lib/reports/performance.py
index 3b8e1df0..d2bda98f 100644
--- a/lib/reports/performance.py
+++ b/lib/reports/performance.py
@@ -25,7 +25,7 @@ def gather(past_days, portfolio_select=None):
 def benchmark(ticker, market, text, past_days, service):
 	"""(report line, percent) of a benchmark over past_days; percent is None, and the line says so, when it has no history."""
 	try:
-		percent = yahoo.price_history(ticker, days=past_days, graph=False)[0][past_days]
+		percent = yahoo.price_history(ticker, days=past_days)[0][past_days]
 	except Exception as e: # the rest of the report still goes out
 		print(ticker, e, file=sys.stderr)
 		return f"⚠️ {text}: no price history", None
diff --git a/lib/reports/price.py b/lib/reports/price.py
index 782d2a7d..22d6dd1f 100644
--- a/lib/reports/price.py
+++ b/lib/reports/price.py
@@ -13,6 +13,7 @@
 from lib import watchlist
 from lib import yahoo
 from lib import buttons
+from lib.reports import history
 
 def gather(specific_stock=None, days=None):
 	"""(tickers, market data) for a price report: Yahoo's quotes; over `days` each holding's Sharesight figure, and for the tickers
@@ -93,15 +94,15 @@ def report(service, threshold=None, user='', specific_stock=None, interactive=Fa
 				if config.prefs.performance_use_sharesight:
 					continue
 				if specific_stock: # the one ticker asked for: no history is reported in one line
-					drawn = config.prefs.graph and delivery.takes_charts(service, interactive)
-					history, graph = yahoo.price_history(ticker, days, graph=drawn, graphCache=False)
+					changes, line = yahoo.price_history(ticker, days)
+					graph = history.line_chart(ticker, changes, line, days)
 				else:
 					try:
-						history, _ = yahoo.price_history(ticker, days, graph=False)
+						changes, _ = yahoo.price_history(ticker, days)
 					except RuntimeError as e: # no history for this ticker: leave it out of the list
 						print(ticker, e, file=sys.stderr)
 						continue
-				percent = history[days]
+				percent = changes[days]
 		else:
 			percent = quote['percent_change']
 		percent = float(percent)
diff --git a/lib/reports/profile.py b/lib/reports/profile.py
index 8d198a98..cdf69c7c 100644
--- a/lib/reports/profile.py
+++ b/lib/reports/profile.py
@@ -186,7 +186,7 @@ def prepare_profile_payload(service, user, ticker):
 		payload.append("")
 
 	try:
-		price_history, _ = yahoo.price_history(ticker, graph=False) # the percentages only: the profile has no chart
+		price_history, _ = yahoo.price_history(ticker) # the percentages only: the profile has no chart
 	except RuntimeError as e: # no chart data: the profile goes out without these lines
 		print(ticker, e, file=sys.stderr)
 		price_history = {}
diff --git a/lib/worker.py b/lib/worker.py
index 5ae95eeb..571ae6f9 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -18,6 +18,7 @@
 from lib import formats
 from lib import markets
 from lib import delivery
+from lib import render
 from lib import buttons
 from lib import charts
 from lib.reports import about, cal, history, holdings, performance, price, profile, rankings, shorts, trades
@@ -114,8 +115,9 @@ def process_callback(service, callback_id, chat_id, message_id, data):
 	try:
 		_, kind, period, ref = parts
 		tickers = buttons.resolve(ref)
-		caption, image = charts.build(kind, tickers, period, service)
+		caption, chart = charts.build(kind, tickers, period, service)
 		chat = delivery.reply_to(service, chat_id)
+		image = render.draw(chart) if chart else None
 		chat.edit_chart(message_id, image, caption, buttons.keyboard(kind, tickers, period))
 	except answer.Answer as e: # nothing to draw for that period: said in the chat, as a command would
 		delivery.deliver(delivery.reply_to(service, chat_id), delivery.Report(e.lines))
diff --git a/lib/yahoo.py b/lib/yahoo.py
index 18821c6a..6434a8b7 100644
--- a/lib/yahoo.py
+++ b/lib/yahoo.py
@@ -4,6 +4,7 @@
 import hashlib
 import requests
 import threading
+import typing
 import zoneinfo
 import sys
 from lib import answer
@@ -11,7 +12,6 @@
 from lib import cache
 from lib import net
 from lib import formats
-from lib import render
 from http.cookies import SimpleCookie
 from pathlib import Path
 
@@ -307,21 +307,29 @@ def fetch_detail(ticker, seconds=None):
 		detail['insiderSell'], detail['insiderSellValue'] = totals['Sale']
 	return {ticker: dict(sorted(detail.items()))}
 
+class Line(typing.NamedTuple):
+	"""The prices behind a price_history, for its chart: the company's name, Yahoo's full name for its exchange (as
+	fetch() gives profile_exchange), the currency, the current price (else the last close) and the closes over the span,
+	((date, close), ...)."""
+	name: str
+	exchange: str
+	currency: str
+	price: float
+	closes: tuple
+
+
 class NoData(answer.Answer, RuntimeError):
 	"""Yahoo has nothing for it (no such symbol, no prices): a reply in a chat, a RuntimeError where a list skips it."""
 
 
-def price_history(ticker, days=None, seconds=None, graph=None, graphCache=True):
-	"""({period: percent change}, chart image or None): the change over `days`, else over each standard period. Raises
-	NoData when there are no prices, RuntimeError when the history cannot be fetched."""
-	if graph is None:
-		graph = config.prefs.graph
+def price_history(ticker, days=None, seconds=None):
+	"""({period: percent change}, its Line): the change over `days`, else over each standard period, and the closes
+	behind it, for a chart. Raises NoData when there are no prices, RuntimeError when the history cannot be fetched."""
 	if seconds is None:
 		seconds = config.prefs.cache_seconds
 	import pandas as pd
 	from pandas.tseries.offsets import BDay as businessday
 	percent_dict = {}
-	max_days = 3655 # 10 years + buffer
 	now = datetime.datetime.now()
 	interval = ('10Y', '5Y', '3Y', '1Y', 'YTD', '6M', '3M', '1M', '7D', '5D', '1D')
 	data = fetch_chart_json(ticker)
@@ -363,21 +371,10 @@ def price_history(ticker, days=None, seconds=None, graph=None, graphCache=True):
 					percent_dict['Max'] = default_percent
 					continue
 			percent_dict[period] = round((price - past_price) / past_price * 100)
-	if not graph or (days and days < 2):
-		return percent_dict, None
-	company_name = formats.transform_title(stock.get('longName', ''))
-	if days:
-		title = f"{company_name} ({ticker}) {min(days, max_days)} days {formats.signed_percent(percent_dict[days])} Last: {price}"
-	else:
-		label = "Max" if "Max" in percent_dict else "10Y"
-		title = f"{company_name} ({ticker}) {label} {formats.signed_percent(percent_dict[label])} Last: {price}"
-	image_cache_file = "finbot_graph_" + ticker + "_" + str(days) + ".png"
-	if config.prefs.cache and graphCache and (image := cache.read_bytes(image_cache_file, seconds)):
-		return percent_dict, image
-	image = render.graph(df, title, stock.get('currency'))
-	if config.prefs.cache:
-		cache.write_bytes(image_cache_file, image)
-	return percent_dict, image
+	closes = tuple((str(day), float(close)) for day, close in zip(df['Date'], df['Close']))
+	exchange = stock.get('fullExchangeName') or stock.get('exchangeName') or ''
+	name = formats.transform_title(stock.get('longName', ''))
+	return percent_dict, Line(name, exchange, stock.get('currency') or '', price, closes)
 
 def price_series(ticker):
 	"""Daily closes for a ticker via the same fetch/parse path as price_history(): (name, exchange, pandas Series indexed by datetime).
diff --git a/tests/common.py b/tests/common.py
index 3c77f28e..a3d80995 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -89,7 +89,8 @@ def setUp(self):
 		self.patch(watchlist, 'with_holdings', lambda: TICKERS)
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t)})
-		self.patch(yahoo, 'price_history', lambda t, days=None, **k: ({days: 1.5} if days else {'Max': 5, '1Y': 3}, None))
+		self.patch(yahoo, 'price_history', lambda t, days=None, **k: # a history with no line to draw
+			({days: 1.5} if days else {'Max': 5, '1Y': 3}, yahoo.Line('Co ' + t, 'NasdaqGS', 'USD', 100.0, ())))
 		self.patch(yahoo, 'prefetch_history', lambda ts: None)
 		self.patch(yahoo, 'historic_high', lambda t: (100.0, 50.0))
 		dates = pd.date_range(end=pd.Timestamp.today().normalize(), periods=400)
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 022c074f..41e39709 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -522,7 +522,8 @@ def test_usr1_flushes_the_cache_and_keeps_the_state(self):
 		with tempfile.TemporaryDirectory() as data:
 			cache = Path(data, 'cache')
 			(cache / 'matplotlib').mkdir(parents=True)
-			for name in ('finbot_yahoo_cookie.json', 'finbot_graph_T1_None.png', 'finbot_x.json.abc.tmp'): # the last: a write in flight
+			cached = ('finbot_yahoo_cookie.json', 'finbot_chart_0123456789abcdef0123.png', 'finbot_x.json.abc.tmp')
+			for name in cached: # the last: a write in flight
 				(cache / name).write_text('{}')
 			Path(data, 'finbot_watchlist.json').write_text('["T1"]') # state, not cache
 			env = {k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k not in config.STARTUP_KEYS}
diff --git a/tests/test_providers.py b/tests/test_providers.py
index d5dcf63d..fbe02686 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -190,7 +190,7 @@ def test_1d_is_the_change_since_the_previous_close(self):
 		"""1D is the change since the previous close, and 5D the week's."""
 		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
 		self.http(yahoo, Resp(self.chart([90] * 30 + [100, 105, 110, 115, 120, 116.4]))) # up 20% in a week, down 3% today
-		table, image = yahoo.price_history('T1', graph=False)
+		table, image = yahoo.price_history('T1')
 		self.assertEqual(table['1D'], -3)
 		self.assertGreater(table['5D'], 10) # the week, not the day
 
@@ -200,11 +200,11 @@ def test_a_symbol_yahoo_does_not_know_is_no_data(self):
 		unknown = {'chart': {'result': None, 'error': {'code': 'Not Found', 'description': 'No data found'}}}
 		self.http(yahoo, Resp(unknown, status=404))
 		with self.assertRaises(yahoo.NoData) as caught:
-			yahoo.price_history('XYZ', 7, graph=False)
+			yahoo.price_history('XYZ', 7)
 		self.assertEqual(caught.exception.lines, ['no price history for XYZ'])
 		self.http(yahoo, Resp(status=404)) # an endpoint that moved, say
 		with self.assertRaises(RuntimeError) as caught:
-			yahoo.price_history('XYZ', 7, graph=False)
+			yahoo.price_history('XYZ', 7)
 		self.assertNotIsInstance(caught.exception, answer.Answer)
 
 	def test_a_failed_chart_fetch_raises(self):
@@ -214,7 +214,7 @@ def test_a_failed_chart_fetch_raises(self):
 			with self.subTest(failure=failure):
 				self.http(yahoo, failure)
 				with self.assertRaises(RuntimeError):
-					yahoo.price_history('AAPL', 7, graph=False)
+					yahoo.price_history('AAPL', 7)
 
 	def test_one_failed_ticker_does_not_take_the_report_down(self):
 		"""One ticker without a chart is left out of a price list, a benchmark without one is said to have none, and
@@ -257,14 +257,13 @@ def test_the_chart_request_carries_the_cookie_of_its_crumb(self):
 		self.assertEqual(headers[0].get('Cookie'), 'A=B') # the crumb used to go without its cookie
 
 	def test_with_the_cache_off_no_chart_is_written(self):
-		"""With the cache preference off, a drawn chart is not written to the cache."""
+		"""With the cache preference off, a chart is drawn but not written to the cache."""
 		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
 		self.prefs(cache=False)
-
 		self.http(yahoo, Resp(self.chart([100.0] * 30 + [110.0])))
-		table, image = yahoo.price_history('T1', 7, graph=True)
-		self.assertIsNotNone(image)
-		self.assertEqual([name for name in os.listdir(config.env.cache_dir) if name.startswith('finbot_graph_')], [])
+		changes, line = yahoo.price_history('T1', 7)
+		self.assertTrue(render.draw(history.line_chart('T1', changes, line, 7)).read().startswith(b'\x89PNG'))
+		self.assertEqual([name for name in os.listdir(config.env.cache_dir) if name.startswith('finbot_chart_')], [])
 
 	def test_price_series_raises_on_failure_and_on_empty_history(self):
 		"""price_series raises when Yahoo fails and when the history has no closes."""
@@ -619,12 +618,12 @@ def test_where_charts_can_go(self):
 		"""Charts always go to Telegram and Discord; Slack gets them only in replies with the bot token."""
 		self.env(slackBotToken='xoxb-test')
 
-		cases = (('telegram',), ('telegram', False), ('slack',), ('slack', False), ('discord', False))
-		charted = [delivery.takes_charts(*case) for case in cases]
+		kinds = (delivery.REPLIES['telegram'], delivery.SCHEDULED['telegram'], delivery.REPLIES['slack'],
+			delivery.SCHEDULED['slack'], delivery.SCHEDULED['discord'])
+		charted = [kind(None).takes_charts for kind in kinds]
 		self.assertEqual(charted, [True, True, True, False, True]) # Slack: replies, not scheduled reports
 		self.env(slackBotToken=False)
-
-		self.assertFalse(delivery.takes_charts('slack'))
+		self.assertFalse(delivery.REPLIES['slack'](None).takes_charts)
 
 	def test_slack_replies_upload_their_charts(self):
 		"""A Slack reply uploads its chart in three steps (bot token), the report as its message, with no text copy."""
@@ -689,7 +688,7 @@ def test_slack_without_a_bot_token_gets_text_and_no_chart_is_drawn(self):
 		def refuse(*a, **k):
 			"""Fail: no chart should be drawn."""
 			raise AssertionError('a chart was drawn for Slack')
-		for name in ('draw', 'hbar_chart', 'pie_chart', 'shorts_chart'):
+		for name in ('draw', 'hbar_chart', 'pie_chart', 'shorts_chart', 'graph', 'compare_graph'):
 			self.patch(render, name, refuse)
 		self.env(webhooks={**config.env.webhooks, 'slack': 'https://hooks.slack.com/services/x'})
 		for cmd in ('.beta', '.pie Rob', '.buy', '.sell', '.peg', '.price', '.price top', '.price T1 7d', '.performance 7d', '.shorts', '.history T1', '.compare T1 T2'):

From 2032271e60357dc3e82ca481910cf41cfb6c21eb Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 02:43:06 +0000
Subject: [PATCH 34/81] Bar charts take the shared heading; one categorical
 palette (C3, C4)

D3b: bar charts placed their title and subtitle themselves, at 0.945
and 0.885 of the height, where the line and pie charts use _heading()
at 0.957 and 0.910. They now use _heading() too, so every chart's
heading sits in the same place; on a bar chart it moves up about 1%.
The bar chart's golden image is regenerated, and only that one
changed.

C4: the dark categorical slots are listed once, as CATEGORIES. The pie
uses all eight; compare's lines use NEITHER_UP_NOR_DOWN, the same slots
without the green and red ones, which mean up and down. The colours
are the ones each chart used before.

In the output comparison only the bar charts' image bytes differ (47
steps: the photos, and Slack's upload length and digest); no text or
other request changed.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/render.py               |  12 ++++++------
 tests/golden/hbar_mixed.png | Bin 10476 -> 10527 bytes
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/lib/render.py b/lib/render.py
index 9c911c0c..be8589fd 100644
--- a/lib/render.py
+++ b/lib/render.py
@@ -52,6 +52,9 @@ def style_date_axis(ax, start, end, ink):
 	ax.tick_params(axis='x', colors=ink, labelsize=10, length=3, width=0.8, pad=6)
 
 PALETTE = ('#14181f', '#eef1f5', '#9aa4b2', '#2a313c') # charts: dark surface, ink, muted ink, grid lines
+CATEGORIES = ('#3987e5', '#d95926', '#199e70', '#c98500', '#d55181', '#008300', '#9085e9', '#e66767') # the dark
+# categorical slots, in fixed order; the green and red ones mean up and down, so lines that are neither leave them out
+NEITHER_UP_NOR_DOWN = tuple(color for color in CATEGORIES if color not in ('#008300', '#e66767'))
 
 _render_pool = None # set by render_off_hub(): the bot draws its charts in one native thread
 
@@ -168,8 +171,7 @@ def compare_graph(series, title, subtitle=''):
 	Returns PNG bytes (1280x800, Telegram's photo size)."""
 	from matplotlib.ticker import MaxNLocator, FuncFormatter
 	bg, ink, ink2, grid = PALETTE
-	# categorical slots 1-5 and 7 of the validated dark palette; red/green are skipped because they mean down/up
-	colors = ['#3987e5', '#d95926', '#199e70', '#c98500', '#d55181', '#9085e9']
+	colors = NEITHER_UP_NOR_DOWN
 	fig, _ = _figure(bg)
 	ax = fig.add_axes([0.075, 0.105, 0.72, 0.755], facecolor=bg)
 	rebased = []
@@ -298,9 +300,7 @@ def hbar_chart(rows, title, subtitle='', value_fmt=None, xlim=None, session=None
 	for t, (_, _, v) in zip(base_labels, real):
 		t.xyann = (-4 if v >= 0 else 4, drop)
 		t.set_ha('right' if v >= 0 else 'left')
-	fig.text(0.03, 0.945 if not subtitle else 0.935, title, color=ink, fontsize=13, fontweight='bold', ha='left', va='center')
-	if subtitle:
-		fig.text(0.03, 0.885, subtitle, color=ink2, fontsize=8.5, ha='left', va='center')
+	_heading(fig, title, subtitle)
 	return _png(fig, bg)
 
 SHORTS_VERDICTS = ((5, 'board shorts'), (10, 'cargo shorts'), (15, 'boxers'), (20, 'gym shorts'), # five points each at the bottom
@@ -429,7 +429,7 @@ def pie_chart(rows, title, subtitle=''):
 	(the dark categorical slots, in fixed order); a slice labelled 'Other' is gray. Returns PNG bytes (1280x800)."""
 	from matplotlib.patches import Rectangle
 	bg, ink, ink2, _ = PALETTE
-	slots = ['#3987e5', '#d95926', '#199e70', '#c98500', '#d55181', '#008300', '#9085e9', '#e66767']
+	slots = CATEGORIES
 	other = '#6b7280'
 	total = sum(v for _, v in rows) or 1.0
 	colors, n = [], 0
diff --git a/tests/golden/hbar_mixed.png b/tests/golden/hbar_mixed.png
index 2cf896ab1b78b5e8d9a667f1d5a0e439725f71f0..a2198139eb5cd566b775e5b574fcacafde990e71 100644
GIT binary patch
literal 10527
zcmYM4byU;u7sp2kLqUchB{fArrMo)>2@ytjNl8h^s7ZHAH!?s3M7l&eCJdyzJ4eUw
z^Zo1hoU=G*&(1#Q+5O!6e!uPwf3K=Q^pN@?2m~Thd@K6_c+CYK7GP}PIcBVU5(Hvd
zSCo~~^qk%^^zhW|n$G!(KC@pq1u0rHK^e)4WLQQ;v~QDM^Ftu-v8qgPO0m6dl;Vq@
ze`BD<O`7*{gc4vq`5-WZ+JD8fKu`J)Y#?{=6vryt?`$e7YfkJVTx0(HJSEUn-8ok?
z5b{JCB8`B6K;3DO-<*V?00=g;ki&Xlu`Y=3=TH`^NFUFi@gm5S@=D=UfwTMc1=#ek
z=@dQ65kx^_Pv5-Tboc1OEpLL{cYrgh*ydE|;@~HKQ02S(^{hJI4dbLQzc-`*jUGIe
zGo|d_!Eh56u)Icn4q{-5cmnRF=|LhRxQac^qhPQHpucb^qPJRE1PqnU0k-RE!C)$A
z;5zg}Ov&W(ChO-Mq`?E#2#B1zXLgBS3DTAU@*W0LCj&$MiibAG&7$~JNO1$ts!_}j
z;)_$2Dxsc3kl%|8xf9)-VP-2wo|GX+p3SFlu5Ngnaf)Tk$|lSvooRr(#Zxzh0GbAA
zTObA#^gis&ox$__ij(HPIax*Tx;2}pG>)WTHz1R%l8ao2AvMMQu3TMQKEUCvLssUx
zDZeWxc4VZbIlHyW#<yILMd>w46FQfEOqD0B*6FZDJXesFC8MAS4GBqCmIvkbS^S*Y
z<Ihm3q!6QXw6H)JImR*GZwkowNBHhdyOoU#n|GJE*-nVNUHzO{snV+;?T)qkS6+(V
zqUJn&nS)AAOLMxN<`dq+36!+A-yNjN>HecB>~@@l-m*s*V|qy{<rU;;g?$-$84<t2
zV`8`|2_J_BTQ*2CES@L^G2mhco-g>H8_zWrezaJdZSG^>^}YB(k>mgCLy=OG$3dxP
z>2uVHnixjb*Z1bjBMK-&29KEfaP&mLaee%)GNr(oSZL?d``#ChQNBZFRaP6T%3npf
zgoUpzj!0c8lS_Z6z-7R1!(vf=tw)|^TJiDmOQxpUqi9RruY#t{5{nY>n?F2i5?>%Z
zHpe~8vGlZJHZFV9ue!P(w;0`zYY%KsR8nZXy1M8?lPr;JpmNj{71d#ULJi$lL>b!?
zWfUY-OGlni(7P1Yy_rf5)JX=v!}ffE2(HO(NVH+2YqySw?*+J%A;wZhM&=X0{#6m0
zqxH}?evbXJ@a{HeXkcJ(wBfUsmX`dN&4E<(ChJG45X-6FFmOSiCy#&tH}|ML_uA^L
z{p631zj>8{w8MVaXKtI5QpcBYoffN=5KK6qt-QQ^wxBaMsQq(uml<%v!oE)aI}UU7
zj^ovKb#y!bDpD9<)p~8mX7Iah&FUUs*V#`+aAluN%@Rv^);X`pEAl0_8lCOMY2%&m
zPR*Frie6dI{r&MnxK?BL%&f(T9*o;R%_!f7m$r#9Gcvjhpum@vm;Zio))W;rSXlU{
zu#mdRV=p{1vdlnec*}L#UJI7;;`s~sd@FhdLni^x_c|GyoJ=dzoj__=!uM}3V<obi
zKccd`Rj8*TBc!FJg<@vf4kV2nj2s*sBFWj1TBRQ}b#xk()-rc9gF_$70pTM8zgg}Y
zh{&&Itvem4e61aBH(IYA8e=%?;GwChDJ^}|=(dx>dRv{F%OHLi-w}*o3vY&*ndKMv
zB8%CLs(+ex;vF3w1;o8eWv3N$x4$?v9ZKg1j_SBROI=+(#w--Y{Bok4>hSutX4rTA
z5`L&`<%Hv8ebwaB?tDdA@oV=cTItC=KmFJl+NZQd6kRg-=Q$!CccWeJR7=o1n8&oj
z^;RFEVP$Xzgy@;h-0tqKd^9x?6|Ym*@fu?B=bNosyNQR79u;N0otg{Y;y)WYe(H63
z3E|q>()JZ!0K(d1e>ypR`n2gl%cw0r>+{L_NI{QgbZ>P+e0=s-FU<CMIVrO$jfnem
zcK-E~h7%s1%IV0xH`;ejTcbq*3vC_?fzJdS3>4K+y@SopFMO_JvM4x=Tm@;b&uHtt
zPORkRt>mq&yuH2c?FYAbK0>5ROG|rVX^mT~J_XB1PgLjw8Kcs$t<@WouEfPvqL$M)
zG$i(U-9F>jC*ZQt>?Kqx-D6fNbq+JX{H|3)On-O2ST=k8X|mpHtskZu%Cpy>ylkd0
z^Xt<uhnYr~apJHg9v&Vbn%;nekjPkZJq7ioh7)o~@9OHRK^we9y#Q~4_5SSoKzmx+
z-s)ZN+1^Z<R+;yuOF9zS*Vot3AdsbczD{2^_BJd>z+b8;;^`cFYDgavR@V7j+JCkh
z^Cvt}lbS^X)HzdV9ou8hOHGyHMn(}SoQ1T(HIJ%VJ=d@JC&>T;tqP@3;O3x0A;DGh
zFDas`)Di}nNTqw%T&C>#6%#aJD_$`V4K!1h1Y?v~XnJg@<-yWmJ{1kJaWgKv^mX?p
zECf@*$PR`&TE1?E2=qgTwVGAq6_z;-3CWU^pkNHJanHGi5cqu(>4equ>z6SrvMol(
zmlnFJii+XU;ktF+Zq^oz$$h1uw+EW)u*8vGXBtrZ4(93>-eSZNVdm`2)#EIR@@X-O
z><WUo2-4t!)JtT*4RdrgQ;Y*Hk|__eDda}BTwQ2B@=_1vGWdzK)AW=gkw{xvi=>{t
zC$7Q>N(5w}`pbh$5C|N|##qiE5qqX>LlkNnS8V$F&xiMp8gfCxc!)R%A&Avb<~pnY
z)#;AZ);a(9mxt}AkznX_0KA_;)`8#*BHgx7$r3STuSK?YHbel$&2!&b-qKeF-)acc
z1Q!IFQi8EsM7%RPOlj@ao+7IsadzvB?IhY3Pew*q$i<Wyj31m(MRotq$If6I+jH5*
z;7Rw!%UnO>w^_${DqBSwCs+uX)P50+h*ZlFmFJ7Nb{_GF%2=>gd}?{9#%90O`u-_Q
z9hS1f%AYZ_#+X+B+0jg8py_KcK?npH5d`UPOz(5fRAu%Sb7*P_Go{QyeU|tV=|eAX
znc{yjlVS}>1!`$!0;ymw*)Vo3SJT3|cVH&mzZfb=P=EHFw$Xk*{=#&o4oHboJnX%b
zVskCF3>XW+3VNxp?Zuq3=A!M{E5;gC*nB`|5J10sOuXPq4-e2yVe7#x;igeTpV%_q
zCt~s!D5~jhs0OjZK&<BGsf#WF{v*8yjT!UfuKo(<VSEn;i~{<T*5x5@W(U%mhKjJE
zi~|o50V{(2rRo|UaWaLxkH^jF2?JhzRJ?uUP|$q?mjFpZ5Tnsg<<!Yw5YG@(P=M?a
zbs1lofjNu1pQRPf@38T;gUo%D5Rni`(s5|RrytJ|*nt7XHliGygb9{Y8;SV^FP_S=
zvA*WcP!2O?5&MorswJG}e?z{O4%4eY=;y1lB$|_CfKX7Jcio`GX#`9+hUb_P<Ats2
z=k}@Y@5n&M+jcLMna~wVY6B@l62AU}?{GpbEG-Fy#nfTLWR_8WnYc8dl<1}x)XPL?
zGLHzfpaDtdxx<`Aw78TZ&J6}c+Y>qBZlg`~Q8RZZXMd>q=6%jTAt-~T{FncfmZFQ_
z897Fh4_!;lsG)yS@q9F{cgnV$R3z;|UGAO$6{Kmpx}UtAu+!78l}@Y?R%&or%A7OX
z>WBX_@ui}YvhrJ%PY*$FmEYDUWaps?AktB_CE32!N^AXMjUE|-UJ>5}%s}DAivf|i
zqQF(c(oG~eJ6`jB|G{crM?p-@`*-Dx_1dao+_uQB5{a~`@wzz7Qy=+nxiFY;e}&(9
z^-R0%2*v2vc*9gxhs*s@lvn+6&MVzLu$X7!yUF5WIW%0}Q3Gqj9{av;--2(mxY^|1
zM3~j+iMZ|lz5fyz9!>zP1jZ`Xog+{0GhZvrc5T^2`QFqCRcy~tbvdIV@Nszf(XBD%
zWT^&z#<)_+g8+zRvFz|`ky!m=9u>5OVnlQ}P>deGO(?b*qp$e(&3aJ8ac{24f8)ED
zC?HIoj+VQ=M3R#RRazQ|I~|%tQ-_1Fvo|WOEDhV>I8ZX(Jf&;*JtpN$Lg<y+w8ZOo
zXo`lp{T}^VP1vWCE7(`<O80|*<10rHX;$)gWhrd@_Otb-oy93^`e{Q}+e{NJJ`;n3
zWf>Fqxk$Y-Ew(ua58YzbS8+^<eb!YFNd^Q{WS0f^k*H(c%uJnQYC;0H&%d@8&z}c<
z3k?nR5uLv%O!{spTxnF3V!4!3F~g*i$+?WtZoBfH>Q#S{!m4Y_pP~QDo?Y?76h(C|
z(pF9M^3VQU6A|8u+6>%#tBB&`VCI^uuzsBb5wy{p?-3yOl~q2SouFo-<;_FS_SQLc
ztT`#&59Uu+N$>SAoqFJbZ_D^32{(6F(RR_dhlVBSrxX+=Hnz65Ji#4@-J1e0a~ucL
z>U#Wn)#EJ@5a`RGD%1uk?$XTDfi|6b>z<e_D9O|H;jD>rUG(g%LD@L<lP3vodsihn
z%AW;kF*_3({CgT@<I2(hEFUlGFT?DEuP|ojX4?YR?F>nAko(O7H;=oYweL1mM@E(&
zwWSe*!eU}JXB*HVd=k92XmN4q-4xvK`gH3PFJ*rIb8~a^%m)Jfxd*R70Vb^5t{;c9
z{g%40E-xP(#CXkYxejHY=*oI}x~r;6O9xF^>`zpx-EU+A5kYl&{7y(npzgj|hIw17
z;>KY{O(OQ%#NJbr>>scC*xqb|ehQAARwcF2ar9DWD7$g1S!cJ(&P0XoPiq2<&#Op~
zhC+^Z2;uPH;O<nJqpa-3P)0){lZxjr;{~1)HIL1|`%`x~qi8f12v(-0d(jmXclZsC
zIy0=Xn#*u(rKP1!&M`h5s;=}t-{)H^IEMsje$SgRq9Uo@WJs}W$j!}V;ZTFYbW<Lk
z9B+Qe{Q7mOrH7O$ab(HKx5?}hxU0_g=j^J|Q~hpX!$X+_q)gq+OjnGG35;MWId%DH
z8t1#;Uj`^tt7#~nJb6MwB9&ElKr?GUHI?HkY*@QJPR^$HK~qx}eB|9??01Q259Fw|
znQ=eg7rQ<`ppYX0IW4w#%r$w`eVRyFne2*8^=z4MZuV`OYqdT~;Quu9Fh~*<lu2jO
zu={o-j`Y<47Sw_&g|sW9p@FV0ltM1Waw}%XrNMwbb!Jc8W8XmDJEUjJdh&;<g{B3N
z2WC+lX(}PLI|6>U7;bg4m1Q$qrU@FJb;BAqm%Ez>FBD>|=NiWoij2KmjB2KQGg$p7
zg6|jVmC$b=WN(xfV9_Rsq7)NyR_Ymqa#<o+BBVCNrgc&*L)IvF9WPEr1lWec4MN5k
z|7<xpz)>E1EXzc${N~b&l`o({m3%6}V^W0XDri}C!VnVmQKb@6PNswf)aOh_MZ936
zN`ck&b!x(pCfUs-+JNA5s^)%^K{d|#!qfNlYt0Hu+SRNXNe0(r;RVFe%)dVgJa+i<
zj*CzREFU()8$BG+l&&*b?JH4))9hgPnR)ztzzz&GyF*lANlW&Rt>5l^HPka|7-xQo
zleJQRZz1_6(tx*<;i1-F3XB9SCA+xpjP9+bF6QZdI44vZY~~h6+V0puD2ZT@4KoR<
zB4=D=K-A^aZzV(YN==oW!rwSM9p5^ZU|p!thjRDG4OA=l5lM%sGE+z6^*cSK@Y#I)
z=Ok##K0i#D=JD%1Cz_A^8Bkf-#4kUPP)wO*d4NPA2MM-rM~=^Mj?d$2>A#Fl853;K
z$+hFx5zoQWPN<rhNEqx3L^2E_4HPv~%Fxa{7G$5rs#kV(1tM-S#CzePe(_(T-s2bJ
zWpy^_2umUGVtj$Z`GG_`r}CGFBeWv!f45x0c{LwVpB7qipcb`y*JoRwxHBmK0Djbd
z{^msh;n>a9<W=&AYqPc$Vgg-UyUdRCYjjVmoq~=kqig{aR-1$gFLjV<k0?sOc8rB9
z?<)BXm59e)vB<&+O5zz6RS_}*X-ik<(^)*!<?fR+-(t(i<XgL&VK?#pO~h!fRPc|#
zanTw4&!W4pw_@LiFp?nRCRS|bo2L#<WgcV`$AYhKYIIRz6APv&z<qjMR|rJo1$WtJ
z^DY&P6etMoo<%%;13&EWF*P;KXNjRc+V0U{y1sn^=%A7G)r$h+@r~yZ^s^81;w)c2
zf1a<=bEcVAQrx$>`Av&0qt4;>Vt|AA^pdghP$|JUq44G=Xu9^pJbZ)A#}r=YU^w&3
z&S!b7lz>!R8thDSezntia(v<+%nQHx$O&ZX0LXN!-^R7MJPwGPdwzTTWJ@<kRSFCo
z>&@OsB@W1a-}fp}t+@4kA0HC*^XHpN;%5&7t}xYC`gQidd`$@IBb5-We{R(I2%a<5
zBQ?zBYi(ipMAY?eGitz{1%qX8IL`LcVXppe;1h0*wD`QoyE&UV07D~It*3wXMLb>b
zHk6BgCNO=xlH%BEaB=e)wS~DVxj!h?U{>3kuWgd<il}D2|BO}a<MiQ!StXo9hc$)W
zP?*D@p0X+^pQWmb>(-cNuv$5Ge~yC92pg-L62Kyabi%V6CvLlArk4?pS*oe=AMYtv
zqR7a}UnLX(S?|Y>Uc<}-%$2KfhY6=7*sBHL4oTz=GWPs(u+B1=-FS5ukJvS5rswu*
zX%D?4OjC0;(N`<7$Z~*O%m-67ZQs?^g%`X;Psj<Fqc2}V`P4o+)TY1q{hLOW8T1y-
zTQik8UH+=)q^c^J-@)}gkyHtq&tX(YJqdB4Fq|!>0L!}9denoUKpfFWk3TyFcn08^
zat1TaYtjJwRpoH&t@L#6SNIc9L?kVT=)HlrcXwYRvrK1!_V+6(vM$vu;pU@m{mRGm
zzbv*=#d5qF{hIiF&S`a;Ew*mSY5_5()9Kyfxq4XDfnyAcbDaCW*W8ffczL7{V*1@c
z(SNH83zSFA_r!m;*?H&Rr}W~d<m5zjqGLOlkS|~8M1OeP2X^9n+@CJIrvV5aR)-)D
zPok-=e`cFf^6ONN65YY#-}%xAN+g-ON|NPNt}r{;_k4e@_3FwliVG27ImJQFo-!+!
z)g%!PX@uWiwPSI#h%Odg8ncReZJ!>k^bp~7-jdYzvyOB~#*lhSk+~5LsKcIc6G}%h
zY`O~5a)2q#K>%?{vBZ^H^a9Q<{(jmk{-H)BIY+Wz44z7QPj~lVmRRrz)dML`YC^zp
zwwl$=unkl%Xi!4rvNvWO&I@A2smaM3omZ^3RdJ!bU!(HBJ>$0c`!}!i8K1r9-gIqw
znYN3I%WJK2r^CgL>tF9v*#by{0s}JBr!BT7j3e`P>QpOqYc>QU_do%wtA&+qdJzz$
zhKS{qzyNx5l$zK2eU(|4+#?DAex>szy6;cd`rijWdAX_v`S8;CReXlDjLfVkP!-L1
zsa1YC6^wkd0gN$I;#OxYlYq^1;fF4zoWUFWy=hE|YSv@i)w|p4y#;>;5fKq?ZfO`_
zi0Q8Twa-C}kn5(?da|qf*!k9s&+4-S#FtQdVb=?INJJ`|c`y>Ww6w%$%Xruf0yX$v
zD(UH^tCsL&4E6Q#$z?r$p5}&`p8C9jD+Q6O(o(AKu$Bd{%@2yfmvK2ZDTgL@FKr%h
zwY*&<-svb8=RMeXbtzK(<LZ47U*nO2f`VadrhKX&5?|ocv@H53;6BW1XeFGvnVIpS
zNtv0xC+kB&K|#69EZv2=HB1VzbcvQi5U^&2Uaj0o@r=iAe}5y>w(A8w*vZN3^!nD-
z{|Rmu%Qm<hS*(~K`^GDKVh1zU0O*MR_ti8X|Lzb#!9ml}Glum>QDJAHjO@~23&Em`
z#@%{6s>cv%T2D*ekuAjN(5!=ir6LSAzJn1(kxDZvE(`ASOBh#9luhhF5n!n{iNzyt
z75Req)eQmu&)e%Lttbn#6kqipQt0k#;U}k*u*COO8$=ftCMul0z)QCNcYcEcL3$74
zmV&DiJyal&8x1g4H)9&s^03LtXmUpq@x66EPSOQ!!VIKn&Nai2Gwigo%L<z$Qji3S
z%WyFV6TVQWi(|-5Dt&#?Rzz^yG~BjOQTT57_5f9Fts)Hu(y=;O%nBfe{@%O-!t>xG
zWk{p@-}gK)m>3-~NRmtr0b#)9rg@>q5EKBRIZSzM*@`~?W;1zkjvmzIrNo7HZ=7I(
zPQAsFLv65AuTuh7p&zSxRXz)g0W54}M7^QG%f`kAFF5dF(bG3tzPlrhbd~3`4JVN2
z>Xn~`GvyOY;kVw9_7e(r&t~yNkn&MQ8UI&2O#=K`-G$Q$MrI40MlG~t?Ba9)*S{SM
zkb|&MIYK`F_Rc5?@hT@0lQeOl>$_7mc)?HzSXN$E68Gfzm|iw&syc*yp!yN{V8z#@
zq^vAj5)%0SY**bpyl3U8S?DUdBnQs@W~Z?uSDVajwO4@MuyL%yr;f|9O;^W48^*^w
z#_w>Gl%+qx(;PD%3K@Sy^19-c`n|1U0?Qz!dCZFVl{Kr|RjF3Rw!UMukgL<ey%@U8
zE_FC9?$Zoh?{3H?c!bw^^x~C9@%rj)AysWZ)*V`j6F@J<tZ5XqPsgh?v@0j<YW14j
zcJh(7IpQtGo=Xk;V%haBFS9JC=;oTBA&N>$>0)->Kb9&TCVvosCY!x@vc83C9*>aa
zm*SQcrT8%_jwZcIv>HnPynXsAxE1l1THvf;wpp~#P|#~Ts&2un%zUON2YyxRY~JJJ
z*Xr*t>^E|QQi$@ry=i#V6PrUwO#CK1lLh|*rDEK+U|efEp2DW*x;;9+=dNF&)wa-P
z(U%JPFEI)_k0$XjtIGHMVTm~I+~zY%wcEdGkhncP>3_N)3qwY6%{O__o9VO|#S&ma
zu)$lWMRXGp`4=x0OGt4717^0calt{kHptGcsHmvZj3N2NqrtB{AW4`yn^t5lX(qdN
zvci)L{!hoN8fv6LHr1=gFRxHItOG9g_E7^X?60ezMaiq6byB##D%12(5ne@|60TGT
zveirhTZ)vD&j6XV9W@r-W6g^l_O{)P+3Q2C()i{#aI7|WMijtApU9CwJ@;q5Y_n)Y
zH`m6jq6y_cPdMC<QVpI^_;NWdf>JESZ(X~?NGee|_i`=Xv3$0Zi!%pWjRxQG9iFiR
zU7)rFR28c<cyu->tb-Demv6%%k&yghE#7zaxun)iQm=*QN<;rH*pF0S0zD1jgvaaa
zC+AzdTZO#GKBF)|d9a)UQnl;Ok-v!hZYHk{cEDh(-}M6AXD}czxRf{!y+*^e?1uRJ
zCw%3eTx<i1l?EwCZ?PLSBOvHP#g*T)JAfjWXbXTWyELJ{8vs1EX5#C?k>{>^Q_l(s
zuIa)Z4()VZPlRpVj3ImKoI{%ic$9a@nwQ<^#xm7GP5g2U&8(VDefv)&tI3_7f0r+P
z;^((-0HS=lzV3T|+b>*eM}S>95$WD!7{6vJWBird%cVq>cH+CPh~@HR*x_MXTAGNt
zOn>U84=&{o)*UB7Ce=()zv-mju&_rI?4nQT7p{)9WI|hn-FNHhwJMUtzc#G(d?GI^
zQai}zu^P_4zMS2kYy7U);?hC!s467KYrXVEk1Bh%z$56}AM*5zDQv`BxZB;9+|)#P
z!5z8hmxaIy0<lde?u+~Tf*2r1*JoMcIck*?J}vYA^>P;TWF{(&*jZV8cdL4avm65<
zuYu*4+m55|AW^e)u=Yb%UDW;N&*ce4qsKw|bh%B}!;uBQb3Xu4OG*8Q5cGap&ljug
ztgfbu0AReuhq%LYXxa@hsaw;v?{2*@+&q=v4Qllg{gQXwjzak)32{KI;*C+?;2;tP
zTyuN=3{<Ajic6DY1xh)UJ&WGI8D?vX?hK7l&bT8`c}Yu6JzDQvAoN*$1NZi8&BvXw
z(z)-384V}v#tjKZcs$Y)CI-Ai{}=|U88)%8BnF(B3gR7#9O@Lv++1yHby&kpd0}pX
zG`uZstOHgxawfgzS@>2<{h$HaBZ}Rx(NF1R312dZ(Z#BkJSu*!1sh*;3C|DXdfZz*
zZkER@h;do-%FgCmK2anb07-v~U+I_*9-wK@P1+Fmd;Da$h<J;DDeSR)?R#jX=_ECu
zZKhyz=(>eeP#*7ubA}>=HHanahwudnOY*>#0C?kA?1DJe=i`Z^`C(}-s9LnFE?LK^
z>g<`slbfrolj>706=GM1KcQSZ^iW*8LQz9oU3BokJE8WqBAUhH509Px|IX_betmNi
zj)74>7CXlp=>GEn4oXhu;+-@rHt59UQvU`GdYpc*^z5Ny<n}9C4qPm-wXGz2vBGtR
z^`E9Rz}m6#xaw5(n&)Ia4=}*@IYxk5yK!O~MSvxW8u{A{^!tuHqvdfh#ko>n<O@h3
z`6dQ)7STNVoZS$RGPvhR=Wi*3llrV)EIkBz^O7MRKDXz-;NvIMe*exD!2Wg<4qYeM
zJ=FjaiKUhJ&Y}clMZH@4R`Z_kr%{B{7`AkQg<7Aw_G&pS{{!3b7teU{;T!>A6GGX}
zi_AAWIDz%O-^RwQ@LG)2$IHv@J`QJ+;9_w_P1!g0cxP|eSX(#5VcO~+BSh%{4$8)?
zLSy`bt4e((WE`oT$msuNa}$uBrYc#SO9g<AeY&<m?CmA#qND=!1xSKGy_Dv1XRjj4
zSrYP1<d{H6_$whHn~~oS03YPLY%FGV<8y^fe)Igrb5Zt>N=p4={{V6(;Uu19=^b9&
zAr16R{139!L7)x@Ht3sin~|#fS(I8y@z{5b8molj(%PM|{h2!dw_$AcL@&Tw!IEJt
zbci3UKKrwm0B0h|@2%eb{ouq3l|A)K#V%%LzY0FQ`*qGo^r+oyD!GBmK$uV){nLGK
zdhl!0W+GKi*PlNoQ8k}k%*@P?TJOT5!~1%BJy)XDd8`KA$=8}@vlrZ^7QKeasR1sA
z4e~t*t^ydRGYt6-AU8KHW6X?<nw3U#2j>!h6E6<%(@ZI2Rv^-Vq1hZUKWI7cIDey^
zK2-ej9r?8VFMyLc%>MNHT^Mc*zZ`!NE4qyFuBoZn;4lzeETYha-R?0GAwcOuawViU
z@V=F!kdTlM%;i3%=iXv@c~E3=GS1D$xuIf$k^kPfTUnbvr@w=&?4PxP)QO4(r;Edp
zRD9y`NV7zyXc)|}+G49c)mXR5!4D|f8$<e>ZriIs$`JUrkgE{ukdE%-+!VkSNU;=(
z8Rw0*6jCS4Qf0>cYTods2!7vyp37NqbrYWZEf<-}Vf@{8a?OQA+CmpA;E<;0M3kta
zZmN8u(lAHVaulGZe&Y0~BYpk~py&BcRTXdAkw6OGwve6E1I}l{pZ9lEhQp1ly!9&d
zUL`Qjz4J4I!}av0gEt=;h9FqMfJ=vpg}ql#%99CoV)Z#%E)w_iwlr*NbzFegOxf2_
z)H$GqxuB-M5&w0^S0xuV@P2-pSr3ItNUD~YmOm*c!In~Ne)x(?jPYgBdJnqfL^M_B
z1s5x?Mem!+=YJqBm6%{WVzy-c+d#{!JNF(Buz&b+g<+Ex&<KvgTNy9YU4Wmr<g<vP
zn&$QaFSx$8mL=$%U^-bpnKs`72)MJGZp^bRkXmKVm0l-#`$1C?{tol}Tc7)DOM)ur
zikaGfJ48sNK}$MNzxO2>+FcAhb#$)JJ{=Uz1Im+^mY6?TzxYYsr7utu@=MVbT91m@
zNP>(iiJ`KeJ^QP7-F*bJYtjbcu+nGL@lVL}Y^d{<$x#%DN&fwrF6>R0CKgdZOGL}3
zn}6ZfIybt{<<Yf#<N5%qnLrv2Au^W+;|APZwd3GI%_x=FpMzh*Fm|@@SIl&O!Lz(K
z)Myn5NtG*Q!O%j+I)kl5y8WU%dMVgl_sC2g?bY!!#{>1<Mjj203%6bZCftSo<UEpm
za~-jP6udN<u$zr$P5+;FQXJ>5oApR!V1R*l6c6DjuS&|AiPrPiFXx2qnh~`ZWvPN-
z5SHwVHt)EmDwet~>N%^o^>%BW>TrL&cL`+F)V93T4zs<*Q&dm{fETT5)0K<M?IsBE
za4ZwvXq9gsd5Y2zi={+a4JO~hD~pTdwbJIwhYX1)t{1iqo4*+^z81_yS{b$D;uB90
zEPLo@morR#Xq=LGV1kHKOPHtRJFe>4_HJqY@#C*Ey^5&Ua$!DF%-hxfsA~fU?hm;A
zP;-f6Ptg`{%<g1sMZT>@iJ!~MJr`@83&hb%ck?w|GJ%r;FQ&}l5l#g5-3KkOIm7l4
zR_elrok89Hn2kmR?c<=TDSHD<XQNuNO2)}lW)EGN9f!LUO-XM*Ac_FrE9mo&(5EVk
zU%$a+Xjdrb>(F|rWb>6@09Q9Q%5nCmP2XsdfWs=`Gk%NdpRZIh_$xm&9EZ|3OB}9s
z;1knaY;9B|nRo*jdfxBq3`iU80TiKxzVO}pD!>d53=cmhCVG&7Am~6oFMD|P2J0Uu
zx#j~>5HBAuuWSjr<!T$8y+MLY7n^F@fIuMl?Ed-H_{ssjneO@@KQ>1i-EgJA{GS_V
zi34p<uR*CaYjR=5Ho&-yY?wNkkVISi%F#D0;{h$UN~4xW#|3fe-fGK%CfEI+)+1)v
zZ`IYS2G$<pVk_2eYnSP{{Rvag|JL*JPbUX|PTRiK+tLMQ#N=Mr1`23dV0TZNa599%
zTME5cv9+3RppgN;EQ5mFo5UuS$-1un-N$e7(jXsnbl##weoomV;x@8<e)(l+5%48s
z@7;@b>jC`nzrlh(C-y*6cxxTDin+TvIrIAbQJ?i+pGI*EotW+R&n+TqUNAB7Ox;Hg
zfX9j<0wo!sOYv^c{ymW#Wa@18ifq-dGEh(Qn5VlP(@;xO?w*}(hE@`uT;3889CcXb
z_}z7d5IwWjy9<SwyuJdo|DDf4NQfSXk@B_olt4a<!(5|VPYi8~w>XdOSXl7&nKg9?
z1lw~!ObIO;Y`Q);MMOQ;=Cov%@voxB(E^}jXIpavJD8U*nN+V&HU7*lBpMMc39@r=
z+%NE0xv=*9V0p76+<C-0@L%GeYb5#wDKma(O}#DzY~h$><>7f(f)@M9Be=P@?j#6m
zcZkI|-^F+XW@KU~i~kmFu0Sj&+hbv6@zXj!f^RpuC7y%?7{+OnfRK>1`kd2c4E<~J
zKmh|!erK1Nz$8g}Mn>Ic<Rx0-nr7wNuaH^#Tm58}-%e~CnuO??vV4do6ag5j$+ivV
zKN(P5Fc<4UiUSk%>G_G|mp0Sz;>^QA-5MLbhRw~*=~}xi34bXF0vc>QJT!#yob@@|
z_u0s*YjoYO_gJ|5czv$VdH)Dujdb7y=E(jzPQFj`P|+BF4MS2Ac1toyYCgdJMpAy;
zC8n)M90HM-ljF0QPuTGM!w}l(wSH=orK(-$koEJ8wmttd(BkIE3gyb^(C*oCd3iZL
zJJmn>nP4aahy&LKEwUoOuL5I2qEK1%81P@pfS;0)#dxm_0nHq`)8l}l41a^8JrDkj
zp*{BrWb)w2jkYmc4MG-`CS*7HSDP~BtC-JYM4VjM(t|2Sg-zp9rX%b}P>ZSN=9}H!
zXF%5ppVXdv+hc#$XMMG_Qz3@eW}2l)>tQfuh^3Lwr$5oVnZPos*%NVB%M!lD8oIsx
zSZ^&IHnBrQPzC)%Mn-lhT8ZAW{{1SDLbgRmM@N>}r-)$f<_=Wvc1SPAs(I|qg_?Lo
zM6X@>6^Zm08p%UfJoH@uG-quh<iAMa_zXdry@4GFV#57^M+n;H)t#=1F{MoL`}PEV
z(^?fs$02QsQxv0^+YHP`NIk4Ot$GNC>KU7rK^YGxcv^V94Ckg}v&6j0dCg&0#|~v!
z9yC?A#Aek&=;dK_Sup1UpBLq)_gC#+<iD@u*t4jgO@5Hi6L~g^(O+C*JGT*feqKah
z8h;=)IB@5|K0<7eN5C|Y$2Mh3->UyQr=w`#9!C>nHg6C}h6M~hf)wRcWxq?C1pN=y
CbIKwB

literal 10476
zcmZvicRX9~|NcW&^Ck#I?NMrPYL5z`s#b};V%6STjj9=|Ri&b4i`sh=DrVKL61(=^
zqxScF{`vj&JCDan{z*>mlXKs%>$;w=J62m$m4t|n2m}I=s6AJH1-$0~rwte%cuo9W
zIt2o8)u<^e>iJLan)&(b!6&oRXBzUOyP&X-1kX3{Hd!<r{o&+bjc?#r-(o0viTa4Y
zKhu`}&ay@PLRANq>-V1LzyI(kUP<)t&9wz>zX^S5Xt0Pl(Vf39s+A>c^SLExUry9(
z=pxfA<en0n)w6&gZ+f24`$@iE_y1+&K|xS!XBt8f2nGgq(A<?tDC(pjX;w_QQ23yY
z=Diz1P#BU1%<8G`Hw%I5VEp2s_@M3P!z*2SR&eg{`R&$^eh{b<X}RzN3W|A;q=CY~
z13g#s@7kaKwE6Gk#)J~yz+P1R(}N*Gd(tOfm>d)UIkLo5^k3Nt_pC_JJ_`v7>U(}P
zbJe~w&Opiui+O{=ZvGZaC~)vZsyP(#4L~0^%#eUa#1dk@m^#r!jc>A_M?sTqajTx(
zFK~lFZWHHOZE^x>>Zw!_P&hCoENio=fX29g7U#WL*M{<K2o?nTqFFewTL2vsE35te
z7W#^^n;R8{91+`fmuzLbG3bE8G{#jS%;?`=OZuHLo~x+dWrshd4heKg*|GsVEG!+n
zzDi&O;)xdAMsPqRV86I{Ol<eip9?i6Mgq#=B7*BAv+GdiZ%fk_^$`(H2|*ADa#(YD
zN%7@LdoqLOYX^rSoHLVxtMhVK*U3qVKw4bh_bO{u3Fi?OPEJm#ri!*cCVG>vjg5wr
zJ`wMWxsZSVKIBYHii>OWe4;_w-PqWORLZW<Z@b(af8A&1fF@)Wfn?5Rm2caeaiK)~
zM><^1q^WSbO-&cg<)<5?!CeQel!MnN<5E#bJ%rO-v#q<fyQU^-g!@PbUR(C=r>rb;
zMl+8i-iNMsuY0?@6PcdECTdNj8qDeBcsMxL#`4wYH@6E-#CjTa3lohdMSA0DZ~iSO
z_3EL91_yX~s*~s?)(*EO#9e29A6uwFGz1MlC#Qe7J^FW;K>zd?<`X>wWv%lCN+p5r
zsiU^r)zE&iewpX>R4j*<m_Yp_35PNJ<(|Po^wR`XU}>N0Yx}_dWZV9mD<i(tGLr_m
zt6!p&8ZPs#wwEVGGxhua+hu<KS^KjWMZ}K;MIgGS?oJzVf{1ccBX4QWc;;v*5gGAt
zbI8rnYPK8`*+4bgm4J}YcX#X}dM=0y<=HalV%*?}%Y*-1dV(m{V|sc7*P%RTmYmw&
z4DGKz>MzXBk`I>V<1>Z+>+F22F>c)8YFmSDUuc_LrZ%Z}lzTHZVo>-p*zJrDc*=n%
zlEIgH=X-PHFeP)E0*C4O;43;Y|0aue3>8mAVPU>TfX`8hZh=`#h;>KU?L@IYqlo`v
zR6<6pN!%Ts5sfuEmldRIkx9X95p?(VU3&T?#Y6hjmoF}zch@zt?8mC%H90vs#fXwV
zSA#dZ(`gmoA|s>L)>cpVZk*%R1_jJ+o7C3}l?xl)|2_z43h3{@u}8a3eOK&8=@=Oq
z;Sb*fC4aoCa~VgkH&XMXt;5{r8;x3g%dC5~z7_0AKp0^{E|*T9^1jOW>}c2Mn}!W?
zNW)PDT3OO<_M=vCgyWz13^9*4&d$#M{*vK_l-<CcJI%Km-2|SpL=;_}?_Zy_Ah8x3
zb62>oE|s@<cnJ^b`>RX2JSPP*yI%!i2~Qu8(M1T?8}e~;GuxWC$gf=dK&h?-Fx@jX
z)ZXxHYWCfvprEL=o+*fr|HZ7U8@J=DmTYBh%}>33IXoF16)_bThZhmv(cp|*TJDbH
z)i3qG{?{KxLa$d!lh1{!xIJp&rvovN>#*7$CVRcsRDl~vdn|p+-EL}X;?(@IGrd}r
zVBwZ*`$UFweenm@wSkkJeFc~oUS7x5<+f&O>Q@&(KSiBx^BdJ)cXr2pva*`LTU3h1
zB5490x9|#eTi;x)XGqi#6A_UyNvBw78@_twdD2(4w6t_`axyzR8-5**u5)CP_BMpR
zR8y1l^g71e+KrD7pUhVtZcF<-gv=RMdpktN#KgzHOzQpkF>}1m=`{ioDjg&)E8D92
z>OAg|fa^k)E+VB7?cqb=n5__cSW<z>qvhb2larAt9M@$WFaK6?x1^qwXKQP=(Bj4n
z*NG}uxUiWS8ClelH#1pspKA^H2-EXym!I+7F>G_x$dvw1-(=)3n<`q|C2*-r%_!Bv
z9|`X2In9trVvw7zQ9bWw4xY_Xyx|f4^n!6{(4@YL^S6B9$<*!51AFvY0~KWd;GG;(
z5(}6T9-G(a@dZhPE+EfHEpRZCpovlell#Q|+)*O{aWtnRX`+zII<+`w*M<y05^@W+
zDPc6$@p#JGx`<TUSe#l4@GUTF*7{Qo14^SwpTy-D++#HgBX=kat^)$03nq(c{cd_l
z=7hX83=r^1A4RBr)Q4FL&}Y-{+ffBYv9Srj^pH1Cp<rOzn}Z4WbOnz1Wae=xLWO$r
zxRTX|j#~#-*CZ3)r1aZ_L!~n{QF;DZX|_fP;Meg$#l`=HThcW1@j~%M#wzZ|3-?U1
z{87*lEiQ(E?W2Uf-+k6pj)+3qqfsoKpnD*^9uN!|G_jJigVh6avWPc}f)*Fs3T>%n
zljjH0?EmT~H0(Z4a;3^Lr6O~**ya-_dk&3{ii;i5cH$YaYW@uwD?V;9BwFQdavlSF
zKb@|vv0)7rk1Z4hShcpYUB?9RduT)<59#z3y`^p7<@#+=8{UR4;%Ngeb;eY1ZhmeR
z@T$#1sY$zz<{3!@Gz_X3!}a3BYWQ{Q13RU8Ie3UQJ9((o2p*`15{@X+pWEbu$5b1Z
z>6aw*#V`AtO*(P@{$wj9c%EABMD>*yE=!GYZb(a=;#Dya;Q2K?(6(wu5bD=*=f<kO
zn;ma+;}{nOjf#l+$%L~J=S9JC?xFKy%E!bK@;1kkt+E~oi4%OpjrgRgi)HW=iZkTr
zhiu$WZIDUe(-{A$mqSM9BT@RCk|OM&z8N2U5%*{V(XqPAWO%Z_<YI0@_)0H*`{g(p
zsN(islJdAYS4(Sd5Kq~Hy<dsM-KZCVcv2bdomPEu$m{zGL`lk`pvRz_Uo0IeO<|#{
z)C+C93!=UOla}@y#kq~3BVHRCHgrq2)cRFCJ3%2APXu5-^8BE%F3rb1BXcxi?ZltE
zxNMDpNs6Sg@CcM;0)cuxFz*n+k+5ut$^(TN`WO}0eK++I`!gHzambPdJ-i^X@bUPe
z4S~Cfzn_-=QvF8rP&EUb6|OVu$}LhIUhEhNhl`g`>7)#-5`n=ZVuSo@9EvO~EWIAp
zd<)~X-)=OTtND7nx=IbtMomQR_GZ!?>x{v9o8Hoo-u%M@`OgkKEY;uM)vVIU(06MS
z4&FHUUo~cM>qPe$vhjV&>hk<A?Do*xP_+~n;{8Ni49o(efjxg*jKSkV<uQ<|0<Pe~
z>vj9tTD)A#dDleg#if%jdEPI(aX{|ZhcX14OaWP>6LSwM-zKImFm3k!P_TiknK!#0
z?WLtnXXu@Cfn218aict&=H!CTzQ1ZZ5^W3I#nq%e1AIw$cXz!rjsXOHM4u7z^xe7?
zI2XAwe6oee_gu|Vle24+P!%$QfU3&rs_zXv4!f;Vx4vh2kDd+$1w9*2nKQp$io?PS
z8~2V+-@biovp*p=XjhZKC?ow?1UO!Ouq~Nr4mzI)%8Cf2W-}u}N=hnD-c;x^Uq{rg
zm0nlMZs&7a^3@4iD4-ep7mzJ0clQIi;0Jfr%6bSABFN6zMUkRX6ktMndU{?YE05Q5
zp@J`E&j%$k+Hti_TG{K$qiW+~u}j_E6EtEr{%HbQ_wN%X+v2ehyyvN@xn6&evfJNW
z&YP^qEiDa<CzTyPd0fm)Pfsj(I_2I2(%mZ$g@I~FuNme|z-sXX4gI8OIfK0CH@iVe
z5Z*xbWOcRcY-236_`vX!CZo*Ims6&Qjdm%KzU)*%Pw7mpdY7*bmqTUrq7`Q9Tcc)_
z@OY1B&z{YQmo>O7U|ePr#a)bNoW&{b)Lv|5r3bjI|EYMY;WksBFur;6S}XxP2Fw$D
zd?CPJIz0=UaI`i0<J#c2^;?*gmDP4M`tRSr=0(wwk+FFtDrFISbIl<&4x$>^ecxb(
zpg+I<TkZ|6_nqrfd-$jHqRXzP&SCsTGW(rGbxy&i)_I`x)29>8_16bCWJ=MPO3S}_
zV`F2!yVHs><z@r;AQl3XdgrR3bEdfNBfY}Jfz%4IjKQL!t01|X3P2&FMzk0eV&1PD
zvWu4;pP5vAN|m;<vhp@_!c|A&oI`AnN|I;ff-isjEu`o35W&F9BG{>2UoVF9)koU~
zrKWX$`<TUp-c&xL={)!`po<+H9gB;LrP~tn{-Ne-4r*3bUjvTi!I0Xy5AGX+70dMj
zh$0WGuP;VNV-x6g!QDN7>_>8LG0eU`+tW`(M4VSqwxF=SzJqeJ7U!9h(}%k<6ES7N
z?$7R4Rzj$2q#ozN@?VVR-<P;~SlAWCdT`xzydK>u#HY}{ShuOGuFl9ErvRA1fua7`
zgoHXHmwSZlWOTAZ($bgR<Vs$iuDiIlv`7CE=rNbK(Q~5nZ)Er$caHcdE}rkIBPavL
zCFnUd-`MU<jc!kdR=HdymYIl`8P~1trBH;D3&HySy0?ls+Su9IWn*rzSgh;Z&rG?X
zwGO<S;hU3jT`eC5N&BgZ5(7LCqD0@Q$tu9zuq*8CjfBhe(N>8r8ynlvF|oYGxlwya
z@Ota9H0=G<Q{~-@*bi`ysywdy_wU040?-BAs8^rURO^?1{rdGSOUtAzH90w1u-&jd
zU?3Hk!~_J>f68w7*7b*erPnxibvHdxOJ)aiBM_V#U)&GJ@_EySMj#Iv7#NtC5{2~<
zr+fVxS4kmD0T|;`#}fTAZXTXanikXYVR<u|H~&V?7^ML-sNlWvo&^jF!$v24D=6^Y
zn0kxcm?)WSG%as!ebX|BaXdu>A9QDk#tV=Cj|O#TIQpI({1cD$8#kIf!lbDwPuJ%D
zkBsascS#1)=<&@~_rI{72}o7#H?@x(RF8R9R#x~f{Hq2X=UaVxr|Y%}6k}pz%NZ?#
zE>9+%6{%ABP2@e6B7g~7ezBsPHeg$W+pEmay-O*FYO758q}r<O5u_*c4Lnbdwz+#>
zlv7;H0!HP%XTL9Es6&ae<HC-AJFhjtE2~^mb)~TSXCh<aXQA-cP}FFUu;*a%hxC5C
zk6AQ0cgF3V4J953f@R5}8up5Mt7aU>u4Y%$ZzHn4wuT;DYjJVdaH~_p;OOR8Xxz?{
z_Vo__Q2+BC6Eow*?BT{anbH?e7|BRoUz6)lMpnIr0twHE>Ksxva(gX%{&Evvo_wL@
zL0TK3@=dJey$I0Nr=#1`pJ#OjuI(SIV>KyOndtb?c~J6Hx6x5Knc%cdm$`dY1@?PO
z3Ll~pg;m7JXi2a2PaZxdV1-R}L>;Q*QAJel(7oIv_^3H2RH-#RZ$Z`D8q{;w@F64`
z4v2OKNxbkdF?C!zi)nZF<I!h)12|{>$D_%~6jr%H5bGKoi?X3X<4Ez3SVT^4xWbAc
z65=rQG3dv~AiS|J6724(LrgtU*uRhKOw2Q1H#~9za%8`|qJBwa-gn>{h>22t;(=-c
zmu{LMs;zrYwH9gy`YF9725t)&ehpD*$1<lTgS7YOyzjkCaXYgY+*#}7u=Qb!shZb<
zFfJjX%1}>*lED|OTaGF{LzxTs&lTsai&G4%Y(ufSN73<o&KvPkLg&$;Efvg#w0PE;
z5d^YKI4j{GU3$aR;H6m3C99Ic!q+=j!k_Fh>)FhPxv+?WB6LBV=6XG&Ox*6-N=UF%
zPj^WD-tqRN+xOz>Mvvu@_lvtSzQ%m157LtsRa3#Rz9%SP>C(v9iS<?^|K>7>*4iH|
zDn2>8j);gT(lxL9JcuVgvy;C6GE26o%QXpw3OpX1R8ms9ulQLEIorP8x_<Z^%32NI
ze$~%>GdfFJ8rZRGBL>$2)_sfjo^pHu1v)-H-hQS4(c-I>NFzMEo7&gItHz|346O34
zKWd+)T~KX8?W#$8jMBc6!Dl<a()U>+XrgEpLY&P*Zr#Fql5U9zL6#6GEa+?}$okR-
z^h~8E1rr=7cxg!r+gRz>)#Xfj0NR<Eyt+AG$J|b5K*MWmPY$hB?&Xx3r8RE1Wa7CI
z_6_BUuo3Ho{?-VOIojJ||EyDHJn_ha0keI@{F38ifdk*bd~3R#!}Z2`bi!78%1Rwx
z`}GlTd=<|pF}L}hI6;iJ6bdaKywu{>esw+mX_$=3StLDB;p!Jo^tEp2%V0hp)bH2V
zO7~xSyO$Yz1V4cM2L^BaDck<}sf2v6sMxsZ)k)4Hl`*jiQErEct*1~R7(_%60D0rj
zuCL$T(eFz^3?ILbq2Xo4=huyumCrxDAOa5JxG&uo0%;lRQZ+aTtBr;t9v&X1$HiUy
zeX#>_q8P}nmre5hlx$z*yw3VLXQ-*Dnx6PsK6{oSUUqe&K?)(_5fBj17`(H@WBcw9
zXm4m=l3Ip;&*g7!?vwm{5R2KQi@3=E^&o9(nO@P-+3u4rrNsb@Lrp<(anB$B>+1{R
zu8E7gz`ulcQhRfL^Wy0m$lTn+!^3?+QwB=Qxzn3@ev<}@?;E5~P<UJd0<=fhizNo4
zLtmsGQcyP`mnXwnyf=@2^H#j1mGJ$A){@^nHr>}`oMD#q_di8zF?}mqwZIqm#@bLM
zGRsJL{)2*WzKk+_W=Ds93nA({dV!Y{xq^^;;E7V&^WCA{7T@5C?LZ`QwC}@LGXVp8
z`@Rp@2c(ShzU-&bcVt~}g$bdS*gilH?57_t7rn~=`H9y7PuvNq!`gqt;fxQ3p$ppf
zBzvoX94c?pl4=t>F4ktX>ez4e`)Sc5*hWtRJvN$zQP_E-nu>S$%gyHRB3-+hiyxnj
zLx{b#-y<bc9};Bc(JsA!p<ppfl3D8ksmuQ(WxAAk4ya|x)K&ki-n;!3xYf%t<D1we
z0f*UA1R%oduc)<KBbO!PaHS5-O9H;qnLq~Xf*C;#Q$3_JN$AADBqTUYFsB-4IMld|
zT5(Y|A<*Mwqe@H6!Sm+wwA5)6Fo-%jNsLmSD{ozX-#)R7$l>GVHxa+G*)>*#ZcUcd
zaM^k|R55~}%oa|&ud`Vqwk9+T&o3@(YiogfQia3IjGIhTSEuWocCQY5m7+=0q*z1o
z-}AJ5`{hdX-+#W<FWucMyRE7SBaQf|VRx2w#tVV#ZFYkhi1^}vx1CAiZu5a>J6hGH
z5IV8Nkz6behXc@!6IxRzWzIG9rfyR;iD}?vI}f;E<@SaHBf>Anx3e*7O&z!T)@AgT
z%6UYLpO3H9ZqP7w71Uu+{;e=S|4`Pg1^^aL#Ij`MznXn-t|N&VdjAX8=473Bu40Ms
zbX(1m4o0u~zwm4D4>4Etw&1%%Dts1?T-nX~*zQN_Li8g_lQPuCrVDt)wyR%hYWQYa
z14Lm`6bzK!TPNvXz5wX-&6{f=1OSVuH9aTltx2uDFOc=&IyOe-#&wR{`h({?H^RKz
z%^k-*3fdj>yQ9m8(3>ugR+fV)Au+Qw_E_&~J|~g%=bj(g)oi;rA=cvQLOtN{W(^th
zW}W!T+h{(4c;Rm}b^oGXc{09&bbMw%;6eR7{<1GrIIgDsN@M&b$3t#A;5wA;R{|Po
z1MuQb@^GdkzGhSXxJX`IfyHdD#Tl^ZN$1ei1Ci@)>MN>G6M~uFT-)r*j9(`*!zg)<
zOZTrb@H*aGv34gX&|Rk;Sut~PS<*~35MJ-ezFtsei4456&$*Yf$~fQOvs(7OovnJs
zBxIkFm^^_7kClMs-<;X>6};ie+qYWZe_pj=!fxdz7xsOV`PYB6YlX&huo67!|60t0
z6_One!e*oLpiBbE(~2(!#WWtm$w4e&d>Ggg!W~6{&k_!@R`Lx<V{0^6(U~Mk>U}3H
zp8oP14FN0aeR=vz4lA;fVj@tZ^;NJTdV$rbxjZcu_0;h<z;yZs2i4WpMKcDwxDrq(
z$Eh65hNoGhJ2{N~^S`m=S8)^Q&r@K0|HH8-qpS6v9mZeHChU&GIB2Uyw~rN@^@zC9
zI}_hVuq^;rL5id=EiX%@%K6>4j&q|2?b6gs5hWb3-xtU0K)~V~pz6f3h6e;(kbRAJ
z2>l4~?141FBC`^GUteE;|EWmgw8EE}9H^y|j~1JAEe863NNRB3il@6`E3U7Phvkw3
zUuIuAB0rdo*S>$BL%TnOYjEQ2KC0!WaT17mUs0%hRPsdau@y&z$bRdwxY}ire%V$E
z_jnSM)W03C;1<7aN6YJ*bA{5oH*Gkp%%@j#QeJCoE3-#h)t%8Q9pGHRo>C2@si-J>
z?~To^dfq1_Jlh!67_SBVx>6KGbL+x8@DhP82;w+>MhV#BvJ^ORPdsCT%M2;VvP7TX
z{|7<r>emv($;#Xd=d0(*-=w_k1dVI|?P$FoED&LF=$YAHSzA>b>O5Z)03`LrI;qy#
z-p#N0h6SIdIpz?d(243K&X+Wyhn@Mlg+mdMk*NsH6ykrw{1!>Qx^<2d;$<d&n}6mu
zy$J~k17)+GI?3M$k<wpwmg2mIGUXrAGL{u@O)fZlN-?OX@?ZRoo}<V7lg+cT#R`!!
zNXBqY+lbK8+ymo9*jyyXJ0TEq$z8|cd;>$T9(05)OPWo-c%ja#_kFPn$EQEg<h2e8
zLl@sCAas{C*mBnkkSz~5wrk7UOsG=)E-}!sp&)1p*Wp2+d}D_49=G7EMtw31xJxE}
z3dQ4`e>S3#O=g9F`=x{z8CEUjDn&cY!;<~^sENbC1Bop^J{BX2Ra7G41p4jjewUf3
zLJ1-kBZ(iWd3&zJu^HF<cRE-3>>mowJDb$M=|1?2wMSnrzPtaBBi^Fk8U9s;B-?)M
z-#~@9Btv4fCdrjS$HMZVkXQmV9K;d<&4XaIDeo_fu%%X#zMHwOXBZA>iijm@Uw>Wi
zI_?NG2RP!HA3fA9vRII>(99C^KVYe5a6p@<R5z*)zXlMvM9|qwP|>>wWB`H;o;vuz
z{^0v>9cg5zf5b?s(Y}2AZ|2*_mhmK6>w*@yw5-KOH3gAP5~czMAt89k_nPWj`+v$x
zO2jNE;VpBT_S67yz1Az#j4H<M`T{_<VP;=E+}o_dkgvobbh=En!hHBdK6q_sg?qf&
zSBp+`%@#Impr^OuE=lC&g}EG#3iR1_XK)hv&$q=4ks|!n0)0FmWY+9G-|BGu#mFr>
zIufp`dc``ofvtjDL+sH4+}!C;x7#j`*Yv<-?cSSX%TFC^gQ|BogqP<-f_f(DC7$mK
zUhmE9-^0I`e06G6DP-H<tDYR_z3^`^YL!=?uK94*YnVYH!0vPm4pwyDSa~p;xXPNv
z<^UALZfC%ya5R{pVvBwQ0ol4YQOR(*tYIE-%Goo!K(P@>+fI-$m+?H=NN_SBA||32
zclQ5l<rJZ<qeHGSPBuv$%(;AZM~$T*_`6ax{z%aL#fz=^YWVT@-pB2iTMD%4>G$tP
zF6>@ne75y>XBy&?D%#u@l7M_)g>;^7&3ByOnQQS=cmB;KdORJKu&{w@@%-wVHPRCU
z;IVGbKghv*Eci6^QOFwHP^QefyR_T>uwH|!S@xT1mx0v5Bt|)Tx7#a*UI#Qa1;uRb
z=EJy0p}*V<yw-;&OAH#Ud(YHU1=O|L1BZD6LtTMr^01x|t|OV4W=l3G1jxbnSB1Ki
z1X<2yfW3>!^VZ<y=4PF(K;=z2<1s|i`!Z(P>JdhTBeZ_pQY!f<DU*X<QNpv<`+3;U
z(y)J_1V4lzMmqTS@WBwxYEz@sC#I*a4a%dG9iIuLm;@LhHwT7{Zfh7p`cgN}gTL9Q
ztF;pVBi=wib;^>j&vvh78QO{SnI)$PYZp5GA>%A2c)JsaQiwk>xiu6BzO%n&1k1-#
zKSSqhC^q~TcDnEny1(jIq3$+})qH&krpU#MugbMPNa9F8RaIzg{Y3o#PScLnvj6{i
zH+wc>Jqo>r#YE!e{7iaz^cpV+8;W_`DN`KyW;>DvR9)WROzJl7G9Fcqgy4ao+A&92
zX#uJ+2e(%P_pc3J7>tWG%(&1H_JrBezCv>jRC6`+QH%B$u)rg!<L+CQ(y913S;q5+
zebINc{3TTS<B+Z=#Wh9~5HD~j_jC3%narW9eZfZu7Tj|qBk!AiW&m0_;;lJU_OVbC
zFzY}fs&_Hoox56|h=&|)nKU?_T08Qnj>D@9E$SL$6HM@uh@TMv&UDSsYpmg?=gY_#
zAfJEM;6RL5SC?nY_$4k&D(i@6oXTGTk0U>KRppO`2QxS}Z>m5e1K^Nlw2#h}$2lg7
z4L-6DTYvblkCDNS)TN6q042>zC*(GcEvlRIH((W^tl%&!nyBJN1VRU)h9E)w068*o
z%VYi&e;SbS+q*t#4;1+4e%nK00W8W#jQRil_un~=nVR}+WL5j}lv)E7iMb0I-X7<`
z14()bfmnpAbToa>!pYV+ewP`v`0|@gZv82lX}`-^Tx6-BQ(2+01`9{T^xd;Y7YLd*
zO6J^wG@<rr^QJcrFMf8C?#>89ys~fNbg4N`-(2PnR3Eoc?U+QBYzOux?miMUI=u}~
z%KQ421ltF2rNqQ?WB&SA-rg!IJwO65u4Kut<y!bl87g7>&Ne;g{rkm5OVHxB8;0@G
zBYNHZW_C4(-g&K1pp$mGH6h0|tkf2BcaC4cg+kIqyvI2M@MC4WqWvP9gx*<v8~v2`
zU3hpRvqDvH?@)Tp>)^}H*GUSujRUJchlb4SJy$DpmBcizT#^an@sOH-1BRAAD34N6
zP{5Ka_X-zQDU@L0PtAbmpFP=<%$^V$Fq`UzF~&Hn`TPI-#A{IRJm<SJ6-&;hnn92Y
zRAF>MnbC3o0lA>deDzbQriHQmf%Tzax8)u;I!^IDTurpO@e7>|35W!`fHvw6pjNbs
z&=>0JYW1tZ=Q3OoKR@o>wvMd^ga9~od$MBE8~fMYGM);4A5;UuR`YR}kpUjn<EAIb
z%P#bMPM~3CP-MqXM!dVLtE;#7-c3V1OH!}(@bFhjMS~Ry^X#nb?6iQVWty1;YRP+r
z3W5p>{QSGfUy)~QY{W1_L&N8!kM$5fCmT8t4GYY*m#60ffEh`dWNm72J7-J2yE6be
z1|nL<vZT?6D>9}(xk^_R5+NkC<`NJRdg&f$%0$UxHSXO*3dg&k+c&AUOO8E}Q3OeW
z4h%5Lp7WWzrM^-5*)1~uv2z8jH|JST1P46lntgzD;UsJVpw!EwlDqH4KdVc}H+RPJ
zz0A$$0VdCdN-UzBt5^UEqQxx6Cx}~UjkEqulbX|_$HLh=jT+a2w!>_ns!vwaKF+U2
z(w}Oy`SNe~GV9tJ70`0<iPh>BlAe$6ANqqeda)#{msI5FnO)zOC$LN=8k%R8>Jd2u
zt3cn4R12>2Me_qP9aJ*-wdho-^C1lDW2Q>bZ-onA%-MYq`VG%pl2KU}f;Hr$?yQ1A
zORn}qr#|rA@t7%sKY2VxmeLdU(BOxK1~1};Lv<+e9+5{R7X)|@ZhMT1Wg1qDcGna_
z;Q^bUgnDSwJO(bZPTUo4?!e!VBogN_%OhzE|BmGc%UjX|e^jVzLgEu<OlhTiWZ@b1
z(*_+7*uVDUS2v@#(QR{+xy$H+{^})}UT&hz=5j7n+NT&pwT59jK_&j%J$_RBFuXs9
z6(I+0+9@)Q3437n*_z`~`}X$mCu%@mt1|p0;b!;HR|VL(@fSh_sWmP$nyknev<qhk
zvR-T+7wWN8YP`@I4^xu*q=~*iUq1XQh^hR1^3tr-(s++cveZF5RkZT2oM|s_yNrKz
zt7Y7NWhg3)9OkMPXgk&wMY7#ux(Ad++WlYBUYGI34st;pZ3msl)xBP)jy)!B?QAHk
z1x&EC-&T&}<!r!yVWHW7|B+-|?8R?Zw6Uru<5i|3iyuVdBkqUt^4QQk6T;a<_wKqH
zx8KeNc%jr^R3fLGQvs$P5IW!5&CKsQRf%b9j6l*LbfQT@g?V^*Y)A2fsxl!nl}O#6
zOG(b+Gi;XTH<&OI?4qIXRU)I4{Kd-J;1^+QmB9>&!hmCa1dl<;_Mk*m5>3F$!Cz+I
z{l}njfHq$}Ip4?CtXCoh<LPd%XGG_L8e#*a0mJ9u_w4tqm2{~z;Gu<KKG;REi4M?)
z@$=_G!^g`#8kR=XiYzDrlpWA#{~wgTxD*V3_l^Y1#mDEU-OkkOF^|kAU1wXQ1I9Fv
zc+_b)d8(c!z@}VOoSbdJg39A)nW=U0+MTWgoSs$uWSPlAnNu3z>8-=!P$)*yXug5E
zEq479W1r3lwQx(3pJCA!R`Q_>Sj+6;UYeQR?nlF!ul<LFAT_vnt`|Bf@$pXS4v##4
z{U9if0s&2heIZ3C(=`r3X52xu+}R29oJVSFLwo!W7T;!0Hzbr9IW6X_lPHDmR6TE*
zBW1NRYObsll4LjZZ1UZoukluU@$A`Rkk^`KmaM~G>qP>+1Sv7GT;m<c|G-i%0?@%8
zd5rCh9i(#5(Y)Dv!oZ@`0O2$JyMI{rxOT#^e{higbd08<_aVIrR%YjGl7TbOycs!C
z2*>9rs9*uRwcY~%<KpUogVSWN0JszAPn_+|%`^;qedN%P^mK18r-#BoZiT9w&6eVV
zrk}?eJ(lv-(`Fi+(t<^0*xA{eEG*In%y<P{=N~Y$Z2ibOK?!4pfK<mUVAwYTfqyM6
zrKiM>iM0WBpWmeZ{{07m>_(F$smP<FqgJelXpMhXsEt9T#7x<@Zv-Ih|28c(nhF@D
znh8v{UNC==^$?R4krrk9>a(31<=RMT5P5L0v_z4lN=init}(|<N%_0NB70>8k2$#Y
z-@;v1Mn*=1V|VohbFecF+2wPs2yG9K{w%WvSX6H_vfh{r9R<yj_PHFnWrjMvgR&@(
zitWn06v|*)c}sNnjvsB=N&F}uO4!ub*VpW`ZQAM|yWFV5rn<7WhRX8-cC(BZXl(we
zJAV5vnvj&P@(qqC0*sLOEbsF3kIM|varrh{KmP?)4es%P$R9m$P9n*TTZk7XL?}Y&
z@#8$B5^+B4W1>$fDOp#WoLTFDvZSSEF;P+yfFwPZfc<C~EKu%COG~2@A=^otmSCj4
zJr}MVf!Lid+a^}Z>ZRFk7T?Mj2Iq!{cdAviUXFeLAecx~@3r1w@#X@dt^Iy{vs9Ne
z%N92*Z}JGr&;)eMfRvDD&-27qIO4rZjFnVwJln~>_eaCTDCn5Cxp|21+2CY>mS`oi
z?(^BMjQid%oON%Kgww4uER%pphf<^M`Y0j;2jDEJd1TO80pO(XSXXavMGeJzrhrKN
z-~p>L6(waQ-@vgf>z5@%O`-oHBjGER1~9Jm!@k9{AE7Ii+Wo5{hV?qnr{Y!k0fC=+
zOPaO76{qg!A~7DXdi!LCkQHT1Ym`b7&#Kq~Vp02uN%x_IkDjo>UmjomAoOenlhYqa
zalBpZfnfbvO&6@Gps;W4eERRIuFu)pB3jEs=Azbzzns8J-?;53GPzHZ&)l)F547;@
z$+u1Ot_3AWf55{$7#fGlgPOf5oO6Q1VSZ$d>xjis7czU9PO=rbWR?YN-@tAAY#wv7
q3YcR6Zc>0`Zd2v4_<`BI1IhYnZ-|#2SpmBkL27VK<#L#1`2PXCeVaT0


From f29d8b04f53229fef3e310fc06758562d881646b Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 02:47:42 +0000
Subject: [PATCH 35/81] Chart builders per kind, typed button arguments, tests
 on specs (C9, C11, C12)

C9: charts.build() was an if-chain over the five kinds of period
buttons. Each kind now has its builder (history_chart, price_chart,
compare_chart, performance_chart, price_list_chart) in BUILDERS, and
build() looks it up.

C11: the price list and performance buttons carried their threshold,
top and portfolio in a list called tickers, read back by position.
Each builder now names its arguments (price_list_chart(..., threshold,
top), performance_chart(..., portfolio)), and buttons.keyboard() and
resolve() speak of args. The callback data is the same as before, so
buttons already posted keep working.

C12: the chart tests capture the Chart spec handed to render.draw()
and assert on it, instead of spying on the bar and shorts renderers.

R12: the stored argument lists behind long buttons
(finbot_chart_buttons.json) no longer grow forever. The newest 500
are kept, and an older button says it has expired, as one with a lost
list always did. A test covers it.

The output comparison is identical in all 159 steps.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/buttons.py        | 31 +++++++++------
 lib/charts.py         | 92 ++++++++++++++++++++++++-------------------
 lib/worker.py         |  6 +--
 tests/test_charts.py  | 40 ++++++-------------
 tests/test_reports.py | 11 ++++++
 5 files changed, 96 insertions(+), 84 deletions(-)

diff --git a/lib/buttons.py b/lib/buttons.py
index e5f6fdd1..6782d856 100644
--- a/lib/buttons.py
+++ b/lib/buttons.py
@@ -1,5 +1,5 @@
 """Buttons under a message: the period buttons (Week | Month | Quarter | Max) of a chart, and the callback data that
-names its tickers. Each transport draws Button rows its own way."""
+carries what its builder needs (charts.BUILDERS). Each transport draws Button rows its own way."""
 import hashlib
 import typing
 
@@ -17,6 +17,7 @@ class Button(typing.NamedTuple):
 MAX_DAYS = 3650 # what Max means for the Sharesight-based reports
 MAX_CALLBACK_BYTES = 64 # Telegram limit on callback_data
 REF_FILE = 'finbot_chart_buttons.json'
+REF_LIMIT = 500 # the stored argument lists kept: a button older than that says it has expired
 
 def period_for_days(days, default=None):
 	"""Button key for a day count, if it is one of the button periods; `default` when it is not (or days is None)."""
@@ -25,27 +26,31 @@ def period_for_days(days, default=None):
 			return key
 	return default
 
-def _ref(tickers):
-	"""Tickers as callback data: inline if it fits, else a short id persisted to disk."""
-	inline = ','.join(tickers)
+def _ref(args):
+	"""A builder's arguments as callback data: inline if they fit, else a short id of the list, stored on disk with the
+	newest REF_LIMIT others."""
+	inline = ','.join(args)
 	if len(('c|h|w|' + inline).encode()) <= MAX_CALLBACK_BYTES:
 		return inline
 	key = hashlib.sha1(inline.encode()).hexdigest()[:10]
 	with state.edit(REF_FILE, {}) as refs:
-		refs[key] = tickers
+		refs[key] = args
+		for old in list(refs)[:-REF_LIMIT]:
+			del refs[old]
 	return '#' + key
 
 def resolve(ref):
-	"""The tickers a button's callback data names: listed inline ('T1,T2'), or a stored list's '#id' (see _ref). Raises
-	KeyError when the stored list is gone."""
+	"""The arguments a button's callback data carries: listed inline ('T1,T2'), or a stored list's '#id' (see _ref).
+	Raises KeyError when the stored list is gone."""
 	if ref.startswith('#'):
-		tickers = state.load(REF_FILE, {}).get(ref[1:])
-		if not tickers:
+		args = state.load(REF_FILE, {}).get(ref[1:])
+		if not args:
 			raise KeyError("these buttons have expired, please re-run the command")
-		return tickers
+		return args
 	return ref.split(',')
 
-def keyboard(kind, tickers, active=None):
-	"""A chart's period buttons, one row, the active period marked."""
-	ref = _ref(tickers)
+def keyboard(kind, args, active=None):
+	"""A chart's period buttons, one row, the active period marked. args: what the kind's builder takes
+	(charts.BUILDERS): a ticker, tickers, a portfolio, or a price list's threshold and top, as strings."""
+	ref = _ref(args)
 	return [[Button(PERIODS[key][0], f"c|{kind}|{key}|{ref}", key == active) for key in BUTTONS[kind]]]
diff --git a/lib/charts.py b/lib/charts.py
index 0080a966..c4837de0 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -1,4 +1,4 @@
-"""A period button pressed: the chart rebuilt for that period."""
+"""A period button pressed: the chart rebuilt for that period, by the builder of the chart's kind."""
 from lib import answer
 from lib import buttons
 from lib import formats
@@ -6,44 +6,54 @@
 from lib import yahoo
 from lib.reports import history, performance, price
 
-def build(kind, tickers, period, service):
-	"""(caption, chart spec or None) for a chart kind at a button period; period None = the command's default view."""
-	days = buttons.PERIODS[period][1] if period else None
-	if kind == 'h':
-		ticker = tickers[0]
-		highlight = buttons.PERIODS[period or 'x'][0]
-		lines, _, chart = history.history_payload(ticker, service, highlight, graph=not days) # default view: it all
-		if days:
-			changes, line = yahoo.price_history(ticker, days)
-			chart = history.line_chart(ticker, changes, line, days)
-		return '\n'.join(lines), chart
-	if kind == 'p':
-		ticker = tickers[0]
-		market_data = yahoo.fetch([ticker])
-		if ticker not in market_data:
-			raise answer.Answer(f"no data found for {ticker}")
+def history_chart(service, period, days, ticker):
+	""".history: the change over every period, the chosen one marked, and the line of that period (or of it all)."""
+	lines, _, chart = history.history_payload(ticker, service, buttons.PERIODS[period or 'x'][0], graph=not days)
+	if days:
 		changes, line = yahoo.price_history(ticker, days)
-		percent = float(changes.get(days, changes.get('Max', next(iter(changes.values()), 0))))
-		link = links.quote_link(ticker, market_data[ticker]['profile_exchange'], service, days=days)
-		title = market_data[ticker]['profile_title']
-		caption = f"{formats.get_emoji(percent)} {title} ({link}) {formats.signed_percent(round(percent))}"
-		return caption, history.line_chart(ticker, changes, line, days)
-	if kind == 'c':
-		args = list(tickers) + ([f'{days}d'] if days else ['max'])
-		caption, chart, *_ = history.prepare_compare(service, args)
-		if not chart:
-			raise answer.Answer(*caption)
-		return '\n'.join(caption), chart
-	if kind == 'f': # performance: tickers = [portfolio name or '']
-		payload, chart = performance.report(service, days or buttons.MAX_DAYS, tickers[0] or None)
-		if not chart:
-			raise answer.Answer(*(payload or ['no performance data']))
-		return '\n'.join(payload), chart
-	if kind == 'l': # price list: tickers = [threshold, top]
-		span = {'interday': True} if period == 'd' else {'days': days or buttons.MAX_DAYS} # 'd' is today's moves: the command's default view, not a 1-day lookback
-		threshold, top = float(tickers[0]), int(tickers[1]) or None
-		payload, chart = price.report(service, threshold=threshold, interactive=True, top=top, **span)
-		if not chart:
-			raise answer.Answer(*(payload or ['no price data']))
-		return '\n'.join(payload), chart
-	raise ValueError(f"unknown chart kind {kind}")
+		chart = history.line_chart(ticker, changes, line, days)
+	return '\n'.join(lines), chart
+
+def price_chart(service, period, days, ticker):
+	""".price SYMBOL: one stock's change over the period, and its line."""
+	market_data = yahoo.fetch([ticker])
+	if ticker not in market_data:
+		raise answer.Answer(f"no data found for {ticker}")
+	changes, line = yahoo.price_history(ticker, days)
+	percent = float(changes.get(days, changes.get('Max', next(iter(changes.values()), 0))))
+	link = links.quote_link(ticker, market_data[ticker]['profile_exchange'], service, days=days)
+	title = market_data[ticker]['profile_title']
+	caption = f"{formats.get_emoji(percent)} {title} ({link}) {formats.signed_percent(round(percent))}"
+	return caption, history.line_chart(ticker, changes, line, days)
+
+def compare_chart(service, period, days, *tickers):
+	""".compare: the stocks' changes over the period, rebased on one chart."""
+	caption, chart, *_ = history.prepare_compare(service, list(tickers) + ([f'{days}d'] if days else ['max']))
+	if not chart:
+		raise answer.Answer(*caption)
+	return '\n'.join(caption), chart
+
+def performance_chart(service, period, days, portfolio):
+	""".performance: each portfolio over the period, or the one named ('' for all)."""
+	payload, chart = performance.report(service, days or buttons.MAX_DAYS, portfolio or None)
+	if not chart:
+		raise answer.Answer(*(payload or ['no performance data']))
+	return '\n'.join(payload), chart
+
+def price_list_chart(service, period, days, threshold, top):
+	""".price: the moves past `threshold`, or the top and bottom `top` ('0': not a top list), over the period; period
+	'd' is today's moves, the command's default view, not a lookback of one day."""
+	span = {'interday': True} if period == 'd' else {'days': days or buttons.MAX_DAYS}
+	payload, chart = price.report(service, threshold=float(threshold), interactive=True, top=int(top) or None, **span)
+	if not chart:
+		raise answer.Answer(*(payload or ['no price data']))
+	return '\n'.join(payload), chart
+
+BUILDERS = {'h': history_chart, 'p': price_chart, 'c': compare_chart, 'f': performance_chart, 'l': price_list_chart}
+
+def build(kind, args, period, service):
+	"""(caption, chart spec or None) of a chart kind at a button period (None: the command's default view), from the
+	arguments its buttons carry (buttons.keyboard)."""
+	if kind not in BUILDERS:
+		raise ValueError(f"unknown chart kind {kind}")
+	return BUILDERS[kind](service, period, buttons.PERIODS[period][1] if period else None, *args)
diff --git a/lib/worker.py b/lib/worker.py
index 571ae6f9..9ab4bf5d 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -114,11 +114,11 @@ def process_callback(service, callback_id, chat_id, message_id, data):
 		return
 	try:
 		_, kind, period, ref = parts
-		tickers = buttons.resolve(ref)
-		caption, chart = charts.build(kind, tickers, period, service)
+		args = buttons.resolve(ref)
+		caption, chart = charts.build(kind, args, period, service)
 		chat = delivery.reply_to(service, chat_id)
 		image = render.draw(chart) if chart else None
-		chat.edit_chart(message_id, image, caption, buttons.keyboard(kind, tickers, period))
+		chat.edit_chart(message_id, image, caption, buttons.keyboard(kind, args, period))
 	except answer.Answer as e: # nothing to draw for that period: said in the chat, as a command would
 		delivery.deliver(delivery.reply_to(service, chat_id), delivery.Report(e.lines))
 	except Exception as e:
diff --git a/tests/test_charts.py b/tests/test_charts.py
index 9b4c5d0d..b127dd2f 100644
--- a/tests/test_charts.py
+++ b/tests/test_charts.py
@@ -265,14 +265,9 @@ class ReportChartInputs(FinbotCase):
 	"""What the reports hand to the chart: sort order, titles and dates, taken from the real report code."""
 
 	def run_report(self, call):
-		"""Run call() and return what it handed hbar_chart: rows, title and subtitle."""
-		captured = []
-		real = render.hbar_chart
-		def spy(rows, title, subtitle='', **kw):
-			"""Record a chart's inputs, then draw it."""
-			captured.append({'rows': list(rows), 'title': title, 'subtitle': subtitle})
-			return real(rows, title, subtitle, **kw)
-		self.patch(render, 'hbar_chart', spy)
+		"""Run call() and return the first chart spec it handed render.draw."""
+		captured, real = [], render.draw
+		self.patch(render, 'draw', lambda chart: captured.append(chart) or real(chart))
 		call()
 		self.assertTrue(captured, 'no chart was drawn')
 		return captured[0]
@@ -297,46 +292,37 @@ def test_a_chart_is_drawn_once(self):
 	def test_price_top_lists_best_first_then_worst(self):
 		""".price top: the ten best, best first, a spacer, then the ten worst, worst first, as in the caption."""
 		chart = self.run_report(lambda: self.command('.price top 7d'))
-		values = [v for _, v in chart['rows'] if v is not None]
+		values = [v for _, v in chart.rows if v is not None]
 		top, bottom = values[:10], values[10:]
 		self.assertEqual(top, sorted(top, reverse=True))
 		self.assertEqual(bottom, sorted(bottom)) # the worst performer comes first, like the caption's second list
 		self.assertGreater(min(top), max(bottom))
-		self.assertIn(None, [v for _, v in chart['rows']]) # spacer between the two lists
+		self.assertIn(None, [v for _, v in chart.rows]) # spacer between the two lists
 
 	def test_price_threshold_chart_is_sorted_and_titled_with_the_threshold(self):
 		""".price 5%: the chart is sorted best first and its title names the threshold."""
 		chart = self.run_report(lambda: self.command('.price 5%'))
-		values = [v for _, v in chart['rows'] if v is not None]
+		values = [v for _, v in chart.rows if v is not None]
 		self.assertEqual(values, sorted(values, reverse=True))
-		self.assertIn('5', chart['title'])
+		self.assertIn('5', chart.title)
 
 	def test_price_over_a_period_has_a_dated_title(self):
 		""".price 7d: the chart's title gives the dates covered."""
 		chart = self.run_report(lambda: self.command('.price 7d'))
-		self.assertRegex(chart['title'], r'\d{1,2} \w{3} - \d{1,2} \w{3} \d{4}') # 28 Sep - 5 Oct 2026
+		self.assertRegex(chart.title, r'\d{1,2} \w{3} - \d{1,2} \w{3} \d{4}') # 28 Sep - 5 Oct 2026
 
 	def test_performance_is_sorted_by_percent(self):
 		""".performance 7d: the chart is sorted best first."""
 		chart = self.run_report(lambda: self.command('.performance 7d'))
-		values = [v for _, v in chart['rows'] if v is not None]
+		values = [v for _, v in chart.rows if v is not None]
 		self.assertEqual(values, sorted(values, reverse=True))
 
 	def test_shorts_is_sorted_and_carries_the_threshold(self):
 		""".shorts: the pictogram gets the stocks over the threshold, most shorted first, and the threshold."""
-		captured = []
-		real = render.shorts_chart
-		def spy(rows, threshold, *a, **k):
-			"""Record the pictogram's inputs, then draw it."""
-			captured.append((list(rows), threshold))
-			return real(rows, threshold, *a, **k)
-		self.patch(render, 'shorts_chart', spy)
-		self.command('.shorts')
-		self.assertTrue(captured, 'no chart was drawn')
-		rows, threshold = captured[0]
-		values = [v for _, v in rows]
-		self.assertEqual(values, sorted(values, reverse=True))
-		self.assertTrue(all(v > threshold for v in values))
+		chart = self.run_report(lambda: self.command('.shorts'))
+		values = [v for _, v in chart.rows]
+		self.assertEqual((chart.kind, values), ('shorts', sorted(values, reverse=True)))
+		self.assertTrue(all(v > chart.threshold for v in values))
 
 
 class GoldenImages(unittest.TestCase):
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 8aac71c6..1fe9ffab 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -408,6 +408,17 @@ def test_other_charts_do_not_offer_1d(self):
 			with self.subTest(command=command):
 				self.assertNotIn('1D', [b.lstrip('● ') for b in self.buttons(command)])
 
+	def test_a_long_list_is_stored_and_only_the_newest_are_kept(self):
+		"""A button whose tickers do not fit the callback data carries a stored list's id; only the newest REF_LIMIT
+		lists are kept, and an older button says it has expired."""
+		self.patch(buttons, 'REF_LIMIT', 2)
+		long = lambda i: [f'TICKER{i}.AX', 'LONGER.AX', 'LONGEST.AX', 'MORE.AX', 'EVEN.AX', 'MOST.AX', 'LAST.AX']
+		refs = [buttons._ref(long(i)) for i in range(3)]
+		self.assertTrue(all(ref.startswith('#') for ref in refs))
+		self.assertEqual(buttons.resolve(refs[2]), long(2))
+		with self.assertRaises(KeyError):
+			buttons.resolve(refs[0])
+
 	def test_no_buttons_where_their_presses_would_go_nowhere(self):
 		"""Without telegramOutgoingWebhook (reports only), Telegram hears nothing from finbot's buttons, so a chart, a
 		scheduled one included, comes without them."""

From 23259330a2e211b8cbfda7cbe863cb187fb88d18 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 02:51:56 +0000
Subject: [PATCH 36/81] Small rewrites: names, tables, a clear Sharesight error
 (R12, R16)

R12:
- camelCase leftovers renamed: yahoo's getCookie, getCrumb and rmCrumb
  are get_cookie, get_crumb and drop_crumb; formats.humanUnits is
  human_units.
- profile's doDelta is trend_marks: pairwise, and a table of the four
  trend marks.
- formats.days_english is two small tables, its checks in the same
  order (weeks before months: 210 days are 30 weeks).
- shortman.fetch no longer makes a shallow copy that the inner dicts
  were shared with anyway; it adds to market_data.
- An answer from Sharesight that is not JSON (an HTML 502 page) raises
  "Sharesight trades: a 502 answer that is not JSON", where it raised
  "Expecting value: line 1 column 1".

R16 (tests): the startup report's key count is taken from
STARTUP_KEYS, and the two checks that deleted files stay deleted
(defaults/env.example, util/migrate_env.py) are gone.

Checks:
- trend_marks matches doDelta on 20,000 random quarter lists with gaps;
- days_english matches for every day count up to 4,000, with each
  prefix;
- the output comparison is identical in all 159 steps;
- a new test covers the non-JSON Sharesight answer.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/formats.py          | 27 ++++++------------
 lib/reports/profile.py  | 62 +++++++++++++++++------------------------
 lib/reports/rankings.py |  7 +++--
 lib/sharesight.py       |  5 +++-
 lib/shortman.py         |  7 ++---
 lib/yahoo.py            | 20 ++++++-------
 tests/test_config.py    |  5 ++--
 tests/test_providers.py | 11 ++++++--
 8 files changed, 66 insertions(+), 78 deletions(-)

diff --git a/lib/formats.py b/lib/formats.py
index d5a6d6c5..f112cff3 100644
--- a/lib/formats.py
+++ b/lib/formats.py
@@ -50,7 +50,7 @@ def transform_title(title):
 		title = title.strip().removesuffix(suffix)
 	return ' '.join(title.replace(' Usd', ' USD').replace('Ishares', 'iShares').split())
 
-def humanUnits(value, decimal_places=2):
+def human_units(value, decimal_places=2):
 	"""1234567 -> '1.23 M'."""
 	for unit in ['', 'K', 'M', 'B', 'T', 'Q']:
 		if value < 1000.0 or unit == 'Q':
@@ -94,24 +94,15 @@ def date_range_english(days, end):
 def days_english(days, prefix='the past ', article=''):
 	"""A span of days in words: 'today' for 0 or None; prefix + article + 'day', 'week', 'month' or 'year'; else prefix
 	and a count of weeks, months, years or days ('the past 3 weeks')."""
-	if days is None or days == 0:
+	if not days:
 		return 'today'
-	elif days == 1:
-		return prefix + article + 'day'
-	elif days == 7:
-		return prefix + article + 'week'
-	elif days == 30:
-		return prefix + article + 'month'
-	elif days == 365:
-		return prefix + article + 'year'
-	elif days % 7 == 0:
-		return prefix + str(int(days/7)) + ' weeks'
-	elif days % 30 == 0:
-		return prefix + str(int(days/30)) + ' months'
-	elif days % 365 == 0:
-		return prefix + str(int(days/365)) + ' years'
-	else:
-		return prefix + str(days) + ' days'
+	for size, unit in ((1, 'day'), (7, 'week'), (30, 'month'), (365, 'year')):
+		if days == size:
+			return prefix + article + unit
+	for size, units in ((7, 'weeks'), (30, 'months'), (365, 'years')): # weeks first: 210 days are 30 weeks
+		if days % size == 0:
+			return f"{prefix}{int(days // size)} {units}"
+	return f"{prefix}{days} days"
 
 def signed_percent(value, decimals=None):
 	"""12 -> '+12%', -3.5 -> '-3.5%', 0 -> '0%'; thousands separated. decimals=None keeps the value's own digits."""
diff --git a/lib/reports/profile.py b/lib/reports/profile.py
index cdf69c7c..233fe52b 100644
--- a/lib/reports/profile.py
+++ b/lib/reports/profile.py
@@ -10,39 +10,23 @@
 from lib import simplywallst
 from itertools import groupby
 
-def doDelta(inputList):
+TRENDS = {(True, False): '🔼', (False, False): '🔽', (True, True): '🔺', (False, True): '🔻'} # (rose, below zero)
+
+def trend_marks(figures):
 	"""Trend marks between consecutive figures (four quarters give three): 🔼 up and 🔽 down while the figure is positive,
-	🔺 up and 🔻 down while it is negative, ▪️ unchanged, ❌ missing; padded with ❌ to three."""
-	deltaString = ''
-	inputListFixed = inputList.copy()
-	# replace any NoneType to allow delta calculation
-	for idx, absolute in enumerate(inputList):
-		if absolute is None:
-			if idx == 0:
-				# get the next value that's not None
-				inputListFixed[idx] = next((x for x in inputList if x is not None), 0)
-			else:
-				# get the previous value
-				inputListFixed[idx] = inputListFixed[idx-1]
-	deltaList = [y-x for (x,y) in pairwise(inputListFixed)]
-	for idx, delta in enumerate(deltaList):
-		absolute = inputList[idx+1]
-		if absolute is None or (idx == 0 and inputList[0] is None):
-			deltaString = deltaString + '❌'
-		elif delta < 0 and absolute < 0:
-			deltaString = deltaString + '🔻'
-		elif delta < 0 and absolute >= 0:
-			deltaString = deltaString + '🔽'
-		elif delta > 0 and absolute < 0:
-			deltaString = deltaString + '🔺'
-		elif delta > 0 and absolute >= 0:
-			deltaString = deltaString + '🔼'
+	🔺 up and 🔻 down while it is negative, ▪️ unchanged, ❌ missing; padded with ❌ in front to three. A missing figure
+	takes the one before it (the first, the first there is) to measure the next against."""
+	filled = []
+	for figure in figures:
+		first = next((f for f in figures if f is not None), 0)
+		filled.append(figure if figure is not None else filled[-1] if filled else first)
+	marks = ''
+	for i, (before, after) in enumerate(pairwise(filled)):
+		if figures[i + 1] is None or (i == 0 and figures[0] is None):
+			marks += '❌'
 		else:
-			deltaString = deltaString + '▪️'
-	# fallback if input has missing elements
-	missingfromstart = 3 - len(deltaString) # desired length hard coded
-	deltaString = ("❌" * missingfromstart) + deltaString
-	return deltaString
+			marks += '▪️' if after == before else TRENDS[(after > before, figures[i + 1] < 0)]
+	return '❌' * (3 - len(marks)) + marks
 
 MARKET_STATE_EMOJI = {'REGULAR': '🟢', 'PRE': '🟠', 'POST': '🟠'} # any other state is closed: 🔴
 
@@ -107,7 +91,8 @@ def prepare_profile_payload(service, user, ticker):
 		payload.append("")
 
 	if 'currency' in data and 'market_cap' in data:
-		payload.append(service.bold("Mkt cap:") + f" {yahoo.market_cap_currency(data)} {formats.humanUnits(data['market_cap'])}") # GBP, not pence
+		cap = f"{yahoo.market_cap_currency(data)} {formats.human_units(data['market_cap'])}" # GBP, not pence
+		payload.append(service.bold("Mkt cap:") + f" {cap}")
 	if all(key in data for key in ('shareholder_equity', 'total_debt', 'total_cash', 'profile_industry')) and data['shareholder_equity'] and 'Bank' not in data['profile_industry']:
 		net_debt_equity_ratio = round((data['total_debt'] - data['total_cash']) / data['shareholder_equity'] * 100)
 		if net_debt_equity_ratio > 0:
@@ -134,9 +119,10 @@ def prepare_profile_payload(service, user, ticker):
 		payload.append("")
 
 	if 'earningsQ' in data:
-		deltas = [(doDelta(data[key]), label) for key, label in (('revenueQ', 'quarterly revenue'), ('earningsQ', 'quarterly earnings'),
-			('revenueY', 'annual revenue'), ('earningsY', 'annual earnings'))]
-		payload += [f"{delta}\t{label} delta" for delta, label in deltas] # doDelta always gives at least three symbols
+		figures = (('revenueQ', 'quarterly revenue'), ('earningsQ', 'quarterly earnings'),
+			('revenueY', 'annual revenue'), ('earningsY', 'annual earnings'))
+		deltas = [(trend_marks(data[key]), label) for key, label in figures]
+		payload += [f"{delta}\t{label} delta" for delta, label in deltas] # trend_marks always gives at least three
 
 	if payload:
 		payload.append("")
@@ -147,9 +133,11 @@ def prepare_profile_payload(service, user, ticker):
 			payload.append(service.bold(label) + f" {formats.signed_percent(growth)} " + ('' if growth > 0 else '⚠️ '))
 	if 'insiderBuy' in data:
 		if data['insiderBuy'] > data['insiderSell']:
-			payload.append(service.bold("Net insider action (3M):") + f" Buy {currency} {formats.humanUnits(data['insiderBuyValue'])} ")
+			value = formats.human_units(data['insiderBuyValue'])
+			payload.append(service.bold("Net insider action (3M):") + f" Buy {currency} {value} ")
 		elif data['insiderBuy'] < data['insiderSell']:
-			payload.append(service.bold("Net insider action (3M):") + f" Sell {currency} {formats.humanUnits(data['insiderSellValue'])}⚠️ ")
+			value = formats.human_units(data['insiderSellValue'])
+			payload.append(service.bold("Net insider action (3M):") + f" Sell {currency} {value}⚠️ ")
 	if 'short_percent' in data:
 		payload.append(service.bold("Shorted stock:") + f" {data['short_percent']}%" + ('⚠️ ' if data['short_percent'] > 10 else ''))
 	if 'recommend' in data:
diff --git a/lib/reports/rankings.py b/lib/reports/rankings.py
index 97f2aff3..bc2d4947 100644
--- a/lib/reports/rankings.py
+++ b/lib/reports/rankings.py
@@ -24,9 +24,9 @@ def prepare_marketcap_payload(service, action='top', length=15):
 		if 'market_cap' not in data:
 			return [f"Mkt cap not found for {ticker}"]
 		currency = yahoo.market_cap_currency(data) or ''
-		cap = f"{currency} {formats.humanUnits(data['market_cap'])}".strip()
+		cap = f"{currency} {formats.human_units(data['market_cap'])}".strip()
 		if currency != 'USD' and 'market_cap_usd' in data:
-			cap += f" (USD {formats.humanUnits(data['market_cap_usd'])})"
+			cap += f" (USD {formats.human_units(data['market_cap_usd'])})"
 		link = links.finance_link(ticker, data.get('profile_exchange', ''), service)
 		return [formats.security_line(markets.flag_from_ticker(ticker), data.get('profile_title', ticker), link) + f" mkt cap: {cap}"]
 	caps = [] # (market cap in US dollars, line)
@@ -35,7 +35,8 @@ def prepare_marketcap_payload(service, action='top', length=15):
 			print(ticker, "no market cap" if 'market_cap' not in data else f"no exchange rate for {data.get('currency')}", file=sys.stderr)
 			continue
 		link = links.finance_link(ticker, data['profile_exchange'], service)
-		caps.append((data['market_cap_usd'], formats.security_line(markets.flag_from_ticker(ticker), data['profile_title'], link) + f" mkt cap: {formats.humanUnits(data['market_cap_usd'])}"))
+		line = formats.security_line(markets.flag_from_ticker(ticker), data['profile_title'], link)
+		caps.append((data['market_cap_usd'], line + f" mkt cap: {formats.human_units(data['market_cap_usd'])}"))
 	if not caps:
 		return ['No market cap data found']
 	caps.sort(key=lambda cap: cap[0], reverse=action == 'top')
diff --git a/lib/sharesight.py b/lib/sharesight.py
index fc5ea877..985dcb83 100644
--- a/lib/sharesight.py
+++ b/lib/sharesight.py
@@ -33,7 +33,10 @@ def _request(send, url, what, **kwargs):
 		raise RuntimeError(f"Sharesight {what} request failed: {e}")
 	if r.status_code != 200:
 		print(r.status_code, "error", url, file=sys.stderr)
-	return r.json()
+	try:
+		return r.json()
+	except ValueError: # an HTML error page, say: "Expecting value: line 1 column 1" would say nothing
+		raise RuntimeError(f"Sharesight {what}: a {r.status_code} answer that is not JSON") from None
 
 def _check(data, what):
 	"""Raise RuntimeError with Sharesight's error, when it answered with one."""
diff --git a/lib/shortman.py b/lib/shortman.py
index c45066b5..1799ca8e 100644
--- a/lib/shortman.py
+++ b/lib/shortman.py
@@ -3,8 +3,7 @@
 from lib import net
 
 def fetch(market_data):
-	"""market_data plus each ASX ticker's short interest from Short Man's daily CSV (unchanged if it is unreachable)."""
-	local_market_data = market_data.copy()
+	"""market_data with each ASX ticker's short interest added from Short Man's daily CSV (as it was if unreachable)."""
 	url = 'https://www.shortman.com.au/downloadeddata/latest.csv'
 	try:
 		r = net.get(url, timeout=config.prefs.http_timeout)
@@ -23,5 +22,5 @@ def fetch(market_data):
 		ticker = cells[1] + '.AX'
 		short_percent = float(cells[4]) # every row, so a malformed one still fails the fetch
 		if ticker in market_data:
-			local_market_data[ticker]['short_percent'] = short_percent
-	return local_market_data
+			market_data[ticker]['short_percent'] = short_percent
+	return market_data
diff --git a/lib/yahoo.py b/lib/yahoo.py
index 6434a8b7..176edf85 100644
--- a/lib/yahoo.py
+++ b/lib/yahoo.py
@@ -18,13 +18,13 @@
 _auth = threading.RLock() # Yahoo's cookie and crumb are renewed by one thread at a time: two at once could cache a crumb for the cookie the other replaced
 HEADERS = {'User-Agent': 'Mozilla/5.0'}
 
-def getCookie(maxAge=1209600):
-	"""Yahoo's session cookie as 'name=value', cached for maxAge seconds (14 days, renewed 30 seconds early) whatever
+def get_cookie(max_age=1209600):
+	"""Yahoo's session cookie as 'name=value', cached for max_age seconds (14 days, renewed 30 seconds early) whatever
 	the cache preference, since the crumb must match it: a new cookie drops the crumb. A fallback cookie when Yahoo
 	sends none."""
 	with _auth:
 		cacheFile = "finbot_yahoo_cookie.json"
-		cached = cache.read(cacheFile, maxAge - 30) # renewed 30 seconds before it expires
+		cached = cache.read(cacheFile, max_age - 30) # renewed 30 seconds before it expires
 		if cached:
 			return cached[0][0] + '=' + cached[0][1]
 		url = 'https://fc.yahoo.com/' # 404, but that is ok because cookie is in response header
@@ -43,14 +43,14 @@ def getCookie(maxAge=1209600):
 		cookielist = [[name, morsel.value, int(morsel.get('max-age') or 0)] for name, morsel in cookie.items()]
 		cache.write(cacheFile, cookielist)
 		print("Got a new Yahoo cookie", file=sys.stderr)
-		rmCrumb() # we must invalidate the non-matching crumb
+		drop_crumb() # we must invalidate the non-matching crumb
 		return cookielist[0][0] + '=' + cookielist[0][1]
 
-def getCrumb(seconds=1209600):
+def get_crumb(seconds=1209600):
 	"""Yahoo's crumb, which goes with the cookie, cached for `seconds` (14 days) while the cache preference is on; a
 	fallback crumb when Yahoo gives none."""
 	with _auth:
-		cookie = getCookie()
+		cookie = get_cookie()
 		cacheFile = "finbot_yahoo_crumb.json"
 		if config.prefs.cache and (cached := cache.read(cacheFile, seconds)):
 			return cached
@@ -67,10 +67,10 @@ def credentials():
 	"""(cookie, crumb), taken together under the lock: a crumb is only good with the cookie it was issued for, and a renewal
 	between two separate calls could pair a new cookie with an old crumb."""
 	with _auth:
-		crumb = getCrumb() # renews the cookie too, when due
-		return getCookie(), crumb
+		crumb = get_crumb() # renews the cookie too, when due
+		return get_cookie(), crumb
 
-def rmCrumb():
+def drop_crumb():
 	"""Set the cached crumb aside (as .old): it belongs to the cookie being replaced."""
 	crumb = Path(cache.path("finbot_yahoo_crumb.json"))
 	if crumb.is_file():
@@ -285,7 +285,7 @@ def fetch_detail(ticker, seconds=None):
 			detail['revenueAnalysts'], detail['earningsAnalysts'] = estimates[2], estimates[3]
 	if dig(quarterly, 0, 'revenue', 'raw') is None:
 		print(f'{ticker}†', sep=' ', end='', flush=True, file=sys.stderr)
-	else: # a quarter without an earnings figure is bad data: None for both, which doDelta marks
+	else: # a quarter without an earnings figure is bad data: None for both, which trend_marks marks
 		detail['earningsQ'] = [None if dig(item, 'earnings', 'fmt') is None else dig(item, 'earnings', 'raw') for item in quarterly]
 		detail['revenueQ'] = [None if dig(item, 'earnings', 'fmt') is None else dig(item, 'revenue', 'raw') for item in quarterly]
 		yearly = _list(dig(result, 'earnings', 'financialsChart', 'yearly'))
diff --git a/tests/test_config.py b/tests/test_config.py
index 1c7f2310..d3dc22f6 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -56,7 +56,6 @@ def test_templates_have_every_key_and_the_real_defaults(self):
 		self.assertEqual(fields(config.preference_values(config._getenv_for(prefs))), DEFAULTS) # the template changes nothing
 		startup = config.read_startup_file(REPO / 'defaults' / 'finbot.env')
 		config.startup_values(lambda name, default=None: startup.get(name, default)) # and is valid as it stands
-		self.assertFalse((REPO / 'defaults' / 'env.example').exists())
 
 
 class Parsing(unittest.TestCase):
@@ -400,7 +399,8 @@ def test_startup_report_is_a_short_count_never_names_or_values(self):
 		"""The startup report counts the startup keys set, and never names them or shows their values."""
 		with mock.patch.dict(os.environ, {'slack_webhook': 'https://secret.example/abc', 'port': '5000'}, clear=True):
 			lines = config.startup_report()
-		self.assertEqual(lines[0], 'startup environment provides 2 of 18 startup keys (values hidden)')
+		keys = len(config.STARTUP_KEYS)
+		self.assertEqual(lines[0], f'startup environment provides 2 of {keys} startup keys (values hidden)')
 		self.assertLessEqual(len('config: ' + lines[0]), 90) # the path in the second line depends on where DATA_DIR is
 		self.assertNotIn('secret.example', ''.join(lines))
 		self.assertNotIn('slack_webhook', ''.join(lines))
@@ -460,7 +460,6 @@ def test_split_env_reads_the_old_dotenv_format_too(self):
 		self.assertEqual(startup, ['slack_webhook=https://x/#frag', 'port=80', 'discord_webhook=http://y/#z'])
 		self.assertEqual(prefs, ['future_days=30'])
 		self.assertEqual(notes, ['.env line 5: not KEY=value, left out'])
-		self.assertFalse((REPO / 'util' / 'migrate_env.py').exists()) # split_env does its job now
 
 	def test_split_env_writes_next_to_the_file_and_never_overwrites(self):
 		"""split_env writes finbot.env.new (mode 0600) and preferences.conf.new beside the file, refuses to overwrite
diff --git a/tests/test_providers.py b/tests/test_providers.py
index fbe02686..bcf66d50 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -48,8 +48,8 @@ def setUp(self):
 		for module, name in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (sharesight, 'get_portfolios'),
 							 (sharesight, 'get_trades'), (sharesight, 'get_performance')):
 			self.patch(module, name, REAL[(module.__name__, name)])
-		self.patch(yahoo, 'getCookie', lambda *a, **k: 'A=B')
-		self.patch(yahoo, 'getCrumb', lambda *a, **k: 'crumb')
+		self.patch(yahoo, 'get_cookie', lambda *a, **k: 'A=B')
+		self.patch(yahoo, 'get_crumb', lambda *a, **k: 'crumb')
 		self.patch(sharesight, 'get_token', lambda: 'token')
 		self.patch(sharesight.time, 'sleep', lambda s: None)
 
@@ -303,6 +303,13 @@ def test_portfolio_failures_raise(self):
 				with self.assertRaises(RuntimeError):
 					sharesight.get_portfolios()
 
+	def test_an_answer_that_is_not_json_says_so(self):
+		"""An HTML error page from Sharesight is a RuntimeError naming the request and the status, not a JSON parse
+		error."""
+		self.http(sharesight, Resp(None, 502, '<html>Bad gateway</html>'))
+		with self.assertRaisesRegex(RuntimeError, 'Sharesight trades: a 502 answer that is not JSON'):
+			sharesight.get_trades('RobSMSF', 684141, 1)
+
 	def test_trades_from_real_response(self):
 		"""Trades parse from a real answer, tagged with their portfolio."""
 		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_trades_684141_1.json')))

From dc2c34868cbebf88a349bf31bc8a59b9b259f099 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 02:55:42 +0000
Subject: [PATCH 37/81] Links: Yahoo wherever Google has no page; the text
 alone without hyperlinks (R8, D4)

D4: quote_link sent Taipei Exchange listings, crypto (CCC) and indices
(^) to Yahoo, because Google Finance has no page for them, but only for
its two callers. finance_link now does that for every caller, and
quote_link goes: three link functions instead of four (R8).

With the hyperlink preference off, gfinance_link returned the raw
ticker (BHP.AX) where yahoo_link gave the link's text (BHP). It now
gives the text too, through the dialect's link(), whichever provider
is chosen.

The Google chart window's if-chain is a table (WINDOWS).

Checks:
- Against HEAD's links.py on 29,952 combinations (provider, hyperlink,
  network, symbol, exchange, days, brief, text), the only differences
  are those two changes: links for Taipei, crypto and indices with the
  Google provider, and the text when hyperlinks are off.
- The output comparison is identical in all 159 steps.
- Two new tests cover the rules.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/charts.py        |  2 +-
 lib/links.py         | 42 ++++++++++--------------------------------
 lib/reports/price.py |  3 ++-
 tests/test_util.py   | 20 ++++++++++++++++++++
 4 files changed, 33 insertions(+), 34 deletions(-)

diff --git a/lib/charts.py b/lib/charts.py
index c4837de0..cc51d5ee 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -21,7 +21,7 @@ def price_chart(service, period, days, ticker):
 		raise answer.Answer(f"no data found for {ticker}")
 	changes, line = yahoo.price_history(ticker, days)
 	percent = float(changes.get(days, changes.get('Max', next(iter(changes.values()), 0))))
-	link = links.quote_link(ticker, market_data[ticker]['profile_exchange'], service, days=days)
+	link = links.finance_link(ticker, market_data[ticker]['profile_exchange'], service, days=days)
 	title = market_data[ticker]['profile_title']
 	caption = f"{formats.get_emoji(percent)} {title} ({link}) {formats.signed_percent(round(percent))}"
 	return caption, history.line_chart(ticker, changes, line, days)
diff --git a/lib/links.py b/lib/links.py
index 30a635af..8dd8b5ee 100644
--- a/lib/links.py
+++ b/lib/links.py
@@ -2,6 +2,9 @@
 from lib import config
 from lib import markets
 
+NO_GOOGLE_PAGE = ('Taipei Exchange', 'CCC') # exchanges Google Finance has no pages for (CCC: crypto), nor indices (^)
+WINDOWS = ((1825, 'Max'), (365, '5Y'), (183, '1Y'), (31, '6M'), (7, '1M'), (1, '5D')) # over that many days, that window
+
 def yahoo_link(ticker, service, brief=True, text=None):
 	"""A link to the ticker's Yahoo Finance page reading `text`, else the bare symbol (brief) or the whole ticker."""
 	if not text:
@@ -9,36 +12,16 @@ def yahoo_link(ticker, service, brief=True, text=None):
 	return service.link("https://au.finance.yahoo.com/quote/" + ticker, text)
 
 def finance_link(symbol, exchange, service, days=1, brief=True, text=None):
-	"""A link to the stock's page on the site hyperlinkProvider names: Google Finance (google) or Yahoo."""
-	if config.prefs.hyperlinkProvider == 'google':
+	"""A link to the stock's page on the site hyperlinkProvider names, Google Finance (google) or Yahoo, and on Yahoo
+	for what Google has no page for (NO_GOOGLE_PAGE, indices)."""
+	if config.prefs.hyperlinkProvider == 'google' and exchange not in NO_GOOGLE_PAGE and not symbol.startswith('^'):
 		return gfinance_link(symbol, exchange, service, days, brief, text)
 	return yahoo_link(symbol, service, brief, text)
 
-def quote_link(ticker, exchange, service, days=1):
-	"""finance_link, but to Yahoo for what Google Finance has no page for: Taipei Exchange, crypto (CCC) and indices (^)."""
-	if exchange in ('Taipei Exchange', 'CCC') or ticker.startswith('^'):
-		return yahoo_link(ticker, service)
-	return finance_link(ticker, exchange, service, days=days)
-
 def gfinance_link(symbol, exchange, service, days=1, brief=True, text=None):
 	"""A link to the stock's Google Finance page, its chart window the one nearest to `days`, reading `text`, else the
-	symbol (brief) or SYMBOL:EXCHANGE (a US listing: the symbol). The plain symbol with the hyperlink preference off."""
-	window = '1D'
-	if not days:
-		days = 1
-	if days > 1:
-		window = '5D'
-	if days > 7:
-		window = '1M'
-	if days > 31:
-		window = '6M'
-	if days > 183:
-		window = '1Y'
-	if days > 365:
-		window = '5Y'
-	if days > 1825:
-		window = 'Max'
-	url = "https://www.google.com/finance/quote/"
+	symbol (brief) or SYMBOL:EXCHANGE (a US listing: the symbol). Just that text with the hyperlink preference off."""
+	window = next((window for limit, window in WINDOWS if (days or 1) > limit), '1D')
 	if '.' in exchange:
 		exchange = exchange.split('.')[1]
 	exchange = markets.transform_to_google(exchange)
@@ -48,10 +31,5 @@ def gfinance_link(symbol, exchange, service, days=1, brief=True, text=None):
 	if brief and not text:
 		text = symbol_short
 	elif not text:
-		if '.' in symbol:
-			text = ticker
-		else:
-			text = symbol # US
-	if config.prefs.hyperlink:
-		return service.link(url + ticker + '?window=' + window, text)
-	return symbol
+		text = ticker if '.' in symbol else symbol # a US listing: the symbol
+	return service.link("https://www.google.com/finance/quote/" + ticker + '?window=' + window, text)
diff --git a/lib/reports/price.py b/lib/reports/price.py
index 22d6dd1f..05767d1d 100644
--- a/lib/reports/price.py
+++ b/lib/reports/price.py
@@ -109,7 +109,8 @@ def report(service, threshold=None, user='', specific_stock=None, interactive=Fa
 		if abs(percent) >= threshold or (specific_stock and interactive):
 			exchange = quote['profile_exchange']
 			exchanges.add(markets.exchange_human(exchange))
-			rows.append((ticker, percent, f"{formats.get_emoji(percent)} {quote['profile_title']} ({links.quote_link(ticker, exchange, service, days=days)})"))
+			link = links.finance_link(ticker, exchange, service, days=days)
+			rows.append((ticker, percent, f"{formats.get_emoji(percent)} {quote['profile_title']} ({link})"))
 	if not triggered: # a scheduled report: is there a move big enough for its ticker? Small caps' betas are fetched together
 		details = yahoo.fetch_details([ticker for ticker, _, _ in rows if needs_beta(market_data[ticker])])
 		if not any(abs(percent) >= trigger_bar(ticker, market_data[ticker], threshold, details) for ticker, percent, _ in rows):
diff --git a/tests/test_util.py b/tests/test_util.py
index 1594f92b..326d52e8 100644
--- a/tests/test_util.py
+++ b/tests/test_util.py
@@ -101,6 +101,26 @@ def test_flags_for_known_and_unknown_markets(self):
 		self.assertIsInstance(markets.flag_from_ticker('XYZ.ZZ'), str) # unknown suffix must not crash
 
 
+class Links(FinbotCase):
+	def test_google_links_go_to_yahoo_where_google_has_no_page(self):
+		"""With the google provider an index, a crypto pair and a Taipei listing link to Yahoo; the rest to Google, with
+		the chart window nearest the period."""
+		self.prefs(hyperlinkProvider='google')
+		for symbol, exchange in (('^GSPC', 'SNP'), ('BTC-USD', 'CCC'), ('6488.TWO', 'Taipei Exchange')):
+			with self.subTest(symbol=symbol):
+				self.assertIn('au.finance.yahoo.com/quote/' + symbol, links.finance_link(symbol, exchange, TELEGRAM))
+		month = links.finance_link('BHP.AX', 'ASX', TELEGRAM, days=30)
+		self.assertIn('google.com/finance/quote/BHP:ASX?window=1M', month)
+
+	def test_without_hyperlinks_a_link_is_its_text(self):
+		"""With the hyperlink preference off, either provider shows the link's text: BHP, not BHP.AX."""
+		self.prefs(hyperlink=False)
+		for provider in ('google', 'yahoo'):
+			with self.subTest(provider=provider):
+				self.prefs(hyperlinkProvider=provider)
+				self.assertEqual(links.finance_link('BHP.AX', 'ASX', TELEGRAM), 'BHP')
+
+
 class StateFiles(FinbotCase):
 	def test_json_roundtrip_and_missing_file(self):
 		"""A state file reads back what was saved; a missing one is None."""

From 4da65cdd7ad374be4c5d3fe4add1595793fb6560 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 02:57:30 +0000
Subject: [PATCH 38/81] Pin only the scheduled earnings report

cal pinned every earnings report on all the tracked stocks, a reply to
.earnings included (it already did before the transports), though its
docstring spoke of the scheduled report. Each .earnings in a group
then left another pinned copy. Now only the scheduled run pins, and a
reply never does.

In the output comparison only "telegram .earnings" differs: its
pinChatMessage request is gone. "scheduled cal.py earnings" still
pins. A new test pins nothing for a reply or the ex-dividend report,
and pins the scheduled earnings report.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/reports/cal.py    |  7 ++++---
 tests/test_reports.py | 12 ++++++++++++
 2 files changed, 16 insertions(+), 3 deletions(-)

diff --git a/lib/reports/cal.py b/lib/reports/cal.py
index e0a373a7..6d9428fc 100644
--- a/lib/reports/cal.py
+++ b/lib/reports/cal.py
@@ -17,7 +17,8 @@ def primary_listings():
 
 def run(dest=None, *, days=None, specific_stock=None, event='earnings'):
 	"""Post the earnings or ex-dividend dates ahead (`event`) for the tracked stocks or one: to the chat that asked
-	(dest), else to the default chats. An earnings report on all the tracked stocks is pinned where the network can."""
+	(dest), else to the default chats. The scheduled earnings report is pinned where the network can; a reply is not,
+	or the chat would collect pinned copies."""
 	interactive = dest is not None
 	if days is None:
 		days = config.prefs.future_days
@@ -26,8 +27,8 @@ def run(dest=None, *, days=None, specific_stock=None, event='earnings'):
 	for chat in delivery.targets(dest):
 		payload = report(chat.dialect, events, tickers, days, specific_stock, interactive, event)
 		answers = delivery.deliver(chat, delivery.Report(payload))
-		if event == 'earnings' and not specific_stock and hasattr(chat, 'pin'):
-			chat.pin(answers) # the earnings report stays on top, where the network can pin
+		if event == 'earnings' and not specific_stock and not interactive and hasattr(chat, 'pin'):
+			chat.pin(answers) # the scheduled earnings report stays on top, where the network can pin
 
 def gather(tickers, days, specific_stock, event):
 	"""The dates of `event` within `days` (any date for one stock), soonest first: [(when, flag, title, ticker)]. ETFs
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 1fe9ffab..13267201 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -260,6 +260,18 @@ def test_earnings_and_dividends(self):
 				cal.run(**kwargs)
 				self.assertTrue(self.sent)
 
+	def test_only_the_scheduled_earnings_report_is_pinned(self):
+		"""The scheduled earnings report is pinned on Telegram; a reply to .earnings is not, or the chat would collect
+		pinned copies, and neither is the ex-dividend report."""
+		pins = []
+		self.patch(transports.TelegramBot, 'pin', lambda chat, answers: pins.append(chat.where))
+		self.command('.earnings')
+		self.assertEqual(pins, [])
+		self.env(webhooks={'telegram': 'https://api.telegram.org/botX/'}, telegramChatID='1')
+		cal.run(event='ex-dividend')
+		cal.run(event='earnings')
+		self.assertEqual(pins, ['1'])
+
 	def test_earnings_skip_an_adr_whose_primary_listing_yahoo_does_not_know(self):
 		"""T1 has no earnings date, so its primary listing is looked up; Yahoo knowing nothing of it used to crash the report."""
 		self.patch(cal, 'primary_listings', lambda: {'T1': 'T1.XX'})

From 9bb30ce6779477a9075efe90005de96bb0594e07 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 03:02:49 +0000
Subject: [PATCH 39/81] price: one mode instead of five flags, headings from a
 table (R3)

price.report and price.run took midsession, premarket, interday and
close as booleans, beside days, and branched on them in three places.
They now take one mode from MODES: interday, midsession, premarket,
close, or period (over days). top stays apart, since a top-and-bottom
list works over a day or a period.
- HEADINGS maps each mode to its heading; a top list has its own.
- nothing() phrases the replies for no moves, or for a ticker with no
  price in the mode.
- price.job() makes {'mode': ...} from a schedule line, so the lines
  read as before.
- The worker's .price, .session and .premarket handlers and the
  price-list buttons pass a mode.

A bare run() is now interday: a scheduled run with no mode would also
skip a market that did not trade that day, which the old no-flag call
did not. The scheduler always names a mode, so this is unreachable
outside the tests.

The output comparison is identical in all 159 steps: every .price
variant, .session, .premarket, .postmarket, the price buttons and the
five scheduled price runs.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/charts.py        |   2 +-
 lib/reports/price.py | 103 ++++++++++++++++++++++---------------------
 lib/worker.py        |   9 ++--
 tests/test_jobs.py   |  20 ++++-----
 4 files changed, 67 insertions(+), 67 deletions(-)

diff --git a/lib/charts.py b/lib/charts.py
index cc51d5ee..c7b202fa 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -43,7 +43,7 @@ def performance_chart(service, period, days, portfolio):
 def price_list_chart(service, period, days, threshold, top):
 	""".price: the moves past `threshold`, or the top and bottom `top` ('0': not a top list), over the period; period
 	'd' is today's moves, the command's default view, not a lookback of one day."""
-	span = {'interday': True} if period == 'd' else {'days': days or buttons.MAX_DAYS}
+	span = {'mode': 'interday'} if period == 'd' else {'mode': 'period', 'days': days or buttons.MAX_DAYS}
 	payload, chart = price.report(service, threshold=float(threshold), interactive=True, top=int(top) or None, **span)
 	if not chart:
 		raise answer.Answer(*(payload or ['no price data']))
diff --git a/lib/reports/price.py b/lib/reports/price.py
index 05767d1d..a906df63 100644
--- a/lib/reports/price.py
+++ b/lib/reports/price.py
@@ -58,11 +58,22 @@ def trigger_bar(ticker, quote, threshold, details=None):
 			return threshold * config.prefs.volatility_multiplier
 	return threshold
 
-def report(service, threshold=None, user='', specific_stock=None, interactive=False, midsession=False, premarket=False, interday=False,
-		days=None, close=False, top=None, market=None):
-	"""(payload, chart) of the price report for one service: the moves of `threshold` percent or more (or the top and bottom `top`),
-	today's, over `days`, at the close, mid-session or pre-market. A scheduled report goes out only when one ticker passes its
-	trigger_bar; it then lists every ticker past the threshold. market: gather()'s (tickers, market data), fetched when not given."""
+# today's moves, in session, out of it (pre- or post-market), at the close, over `days`
+MODES = ('interday', 'midsession', 'premarket', 'close', 'period')
+HEADINGS = { # the report's heading, by mode
+	'interday': '{threshold:g}% at close, {on}',
+	'midsession': '≥ {threshold:g}% mid-session, {on} ({where})',
+	'premarket': '≥ {threshold:g}% pre-market, {on} ({where})',
+	'close': '≥ {threshold:g}% at close, {on} ({where})',
+	'period': 'Price movement {threshold:g}%, {span}',
+}
+
+def report(service, mode='interday', threshold=None, days=None, top=None, user='', specific_stock=None,
+		interactive=False, market=None):
+	"""(payload, chart) of the price report for one service: the moves of `threshold` percent or more (or the top and
+	bottom `top`) in a mode of MODES ('period': over `days`). A scheduled report goes out only when one ticker passes
+	its trigger_bar; it then lists every ticker past the threshold. market: gather()'s (tickers, market data), fetched
+	when not given."""
 	if threshold is None:
 		threshold = config.prefs.price_percent
 	tickers, market_data = market or gather(specific_stock, days)
@@ -72,23 +83,23 @@ def report(service, threshold=None, user='', specific_stock=None, interactive=Fa
 	states = []
 	latest = None # date of the freshest price, for the heading
 	graph = None
-	triggered = interactive
 	for ticker, quote in market_data.items():
 		state = quote['marketState']
 		states.append(state)
 		quoted = datetime.datetime.fromtimestamp(quote['regularMarketTime'])
 		latest = max(latest or quoted.date(), quoted.date())
-		if midsession and state != 'REGULAR':
+		if mode == 'midsession' and state != 'REGULAR':
 			continue # not in session (states: PREPRE PRE REGULAR POST POSTPOST CLOSED)
-		if interday and not interactive and now - quoted > datetime.timedelta(days=1):
+		if mode == 'interday' and not interactive and now - quoted > datetime.timedelta(days=1):
 			continue # a market holiday: no repeat of the last session's moves
-		if close and not interactive and (now - quoted > datetime.timedelta(hours=1.5) or state in ('PREPRE', 'PRE', 'REGULAR')):
+		if mode == 'close' and not interactive and (now - quoted > datetime.timedelta(hours=1.5) or
+				state in ('PREPRE', 'PRE', 'REGULAR')):
 			continue # its market did not just close
-		if premarket:
+		if mode == 'premarket':
 			percent = quote.get('percent_change_premarket', quote.get('percent_change_postmarket'))
 			if percent is None:
 				continue
-		elif days:
+		elif mode == 'period':
 			percent = quote.get('percent_change_period') # Sharesight's figure, for a holding
 			if percent is None:
 				if config.prefs.performance_use_sharesight:
@@ -111,7 +122,7 @@ def report(service, threshold=None, user='', specific_stock=None, interactive=Fa
 			exchanges.add(markets.exchange_human(exchange))
 			link = links.finance_link(ticker, exchange, service, days=days)
 			rows.append((ticker, percent, f"{formats.get_emoji(percent)} {quote['profile_title']} ({link})"))
-	if not triggered: # a scheduled report: is there a move big enough for its ticker? Small caps' betas are fetched together
+	if not interactive: # scheduled: is a move big enough for its ticker? Small caps' betas are fetched together
 		details = yahoo.fetch_details([ticker for ticker, _, _ in rows if needs_beta(market_data[ticker])])
 		if not any(abs(percent) >= trigger_bar(ticker, market_data[ticker], threshold, details) for ticker, percent, _ in rows):
 			rows = []
@@ -122,53 +133,44 @@ def report(service, threshold=None, user='', specific_stock=None, interactive=Fa
 	if payload and not specific_stock:
 		as_of = latest or datetime.date.today()
 		on = formats.date_short(as_of, True)
-		where = ", ".join(exchanges)
-		if midsession:
-			heading = f'≥ {threshold:g}% mid-session, {on} ({where})'
-		elif premarket:
-			heading = f'≥ {threshold:g}% pre-market, {on} ({where})'
-		elif close:
-			heading = f'≥ {threshold:g}% at close, {on} ({where})'
-		elif top:
-			heading = f'Top {top} performers {formats.date_range_english(days, as_of) if days else on}'
-		elif days:
-			heading = f'Price movement {threshold:g}%, {formats.date_range_english(days, as_of)}'
+		span = formats.date_range_english(days, as_of) if days else on
+		if top:
+			heading = f'Top {top} performers {span}'
 		else:
-			heading = f'{threshold:g}% at close, {on}'
+			heading = HEADINGS[mode].format(threshold=threshold, on=on, where=", ".join(exchanges), span=span)
 		if top:
 			bottom = [service.bold(f'Bottom {top}')] + list(reversed(payload[-top:]))
 			payload = [service.bold(heading)] + payload[:top] + [''] + bottom
 			chart_top = chart_rows[:top]
 			chart_bottom = [row for row in reversed(chart_rows[-top:]) if row not in chart_top] # worst first
 			if chart_top: # the best, then after a gap the worst, as the caption's two lists
-				span = formats.days_english(days, 'in ', 'the past ')
-				graph = render.Chart('bars', f"Top & bottom {top}", f"% change {span}",
+				change = formats.days_english(days, 'in ', 'the past ')
+				graph = render.Chart('bars', f"Top & bottom {top}", f"% change {change}",
 					(*label(chart_top), ('', None), *label(chart_bottom)))
 		else:
 			payload.insert(0, service.bold(heading))
-			session = 'midsession' if midsession else 'premarket' if premarket else None
+			session = mode if mode in ('midsession', 'premarket') else None
 			graph = render.Chart('bars', heading, rows=tuple(label(chart_rows)), session=session)
 	elif not payload and interactive:
-		if specific_stock:
-			if midsession:
-				payload = [f"{tickers[0]} not found or not in session"]
-			elif premarket:
-				payload = [f"No premarket price found for {tickers[0]}"]
-			else:
-				payload = [f"No interday price found for {tickers[0]}"]
-		elif premarket:
-			payload = [f"No pre-market price movements meet threshold {threshold:g}%"]
-		elif midsession:
-			if 'REGULAR' not in states:
-				payload = [f"{user}, no tracked stocks are currently in session."]
-			else:
-				payload = [f"{user}, no in-session securities meet threshold {threshold:g}%"]
-		else:
-			payload = [f"{user}, no price movements meet threshold {threshold:g}%"]
+		payload = [nothing(mode, threshold, user, tickers[0] if specific_stock else None, 'REGULAR' in states)]
 	return payload, graph
 
-def run(dest=None, *, threshold=None, user='', specific_stock=None, midsession=False, premarket=False, interday=False,
-		days=None, close=False, top=None):
+def nothing(mode, threshold, user, ticker, in_session):
+	"""What a chat is told when nothing meets the threshold, or the one ticker asked for has no price in this mode."""
+	if ticker:
+		if mode == 'midsession':
+			return f"{ticker} not found or not in session"
+		kind = 'premarket' if mode == 'premarket' else 'interday'
+		return f"No {kind} price found for {ticker}"
+	if mode == 'premarket':
+		return f"No pre-market price movements meet threshold {threshold:g}%"
+	if mode == 'midsession' and not in_session:
+		return f"{user}, no tracked stocks are currently in session."
+	if mode == 'midsession':
+		return f"{user}, no in-session securities meet threshold {threshold:g}%"
+	return f"{user}, no price movements meet threshold {threshold:g}%"
+
+def run(dest=None, *, mode='interday', threshold=None, days=None, top=None, user='', specific_stock=None):
 	"""Post the price report (report() has the modes): to the chat that asked (dest), else to the default chats, with
 	its chart and period buttons where they can go."""
 	interactive = dest is not None
@@ -182,20 +184,19 @@ def markup(chat):
 			return None
 		if specific_stock:
 			return buttons.keyboard('p', market[0], buttons.period_for_days(days))
-		if not (midsession or premarket or close):
+		if mode in ('interday', 'period'):
 			return buttons.keyboard('l', [f"{threshold:g}", str(top or 0)], buttons.period_for_days(days, 'd' if not days else None))
 		return None
 	for chat in delivery.targets(dest):
-		payload, graph = report(chat.dialect, threshold, user, specific_stock, interactive, midsession, premarket,
-			interday, days, close, top, market)
+		payload, graph = report(chat.dialect, mode, threshold, days, top, user, specific_stock, interactive, market)
 		delivery.deliver(chat, delivery.Report(payload, graph, markup(chat) if graph else None))
 
 def job(argv):
 	"""A scheduled run's arguments: (label, run's keyword arguments); a bad one raises ValueError with the usage."""
 	match argv:
-		case ['midsession' | 'interday' | 'premarket' | 'close' as arg]:
-			return f'price.py {arg}', {arg: True}
+		case ['midsession' | 'interday' | 'premarket' | 'close' as mode]:
+			return f'price.py {mode}', {'mode': mode}
 		case [days] if days.isdigit():
-			return f'price.py {days}', {'threshold': 0, 'days': int(days), 'top': 10}
+			return f'price.py {days}', {'mode': 'period', 'threshold': 0, 'days': int(days), 'top': 10}
 		case _:
 			raise ValueError('[midsession|interday|premarket|close|days (int)]')
diff --git a/lib/worker.py b/lib/worker.py
index 9ab4bf5d..201f1f88 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -247,9 +247,8 @@ def on_performance(request, m):
 def on_session(request, m):
 	"""Mid-session moves (.session), or pre-market ones (.premarket, also post-market)."""
 	threshold, ticker = percent_or_ticker(m.group('arg'), config.prefs.price_percent)
-	premarket = m.group('kind').lower() != 'session'
-	price.run(request.dest, threshold=threshold, user=request.user, specific_stock=ticker, midsession=not premarket,
-		premarket=premarket)
+	mode = 'midsession' if m.group('kind').lower() == 'session' else 'premarket'
+	price.run(request.dest, mode=mode, threshold=threshold, user=request.user, specific_stock=ticker)
 
 def on_price(request, m):
 	""".price [percent|SYMBOL|top|bottom] [period]: a bare number is a percent, as on the other commands; a period has its unit."""
@@ -267,8 +266,8 @@ def on_price(request, m):
 					threshold, top = 0, 10
 				elif i == 0:
 					ticker = arg.upper()
-	price.run(request.dest, threshold=threshold, user=request.user, specific_stock=ticker, interday=days is None,
-		days=days, top=top)
+	price.run(request.dest, mode='period' if days else 'interday', threshold=threshold, days=days, top=top,
+		user=request.user, specific_stock=ticker)
 
 def on_shorts(request, m):
 	""".shorts [percent|SYMBOL]: the most shorted stocks followed, or one stock's short interest."""
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
index e67fa2e9..94213c48 100644
--- a/tests/test_jobs.py
+++ b/tests/test_jobs.py
@@ -109,11 +109,11 @@ def test_local_zone_follows_tz(self):
 class JobArguments(unittest.TestCase):
 	def test_price(self):
 		"""price.py takes one mode or a number of days (the top and bottom ten over it); anything else is its usage."""
-		self.assertEqual(price.job(['midsession']), ('price.py midsession', {'midsession': True}))
-		self.assertEqual(price.job(['premarket'])[1], {'premarket': True})
-		self.assertEqual(price.job(['interday'])[1], {'interday': True})
-		self.assertEqual(price.job(['close'])[1], {'close': True})
-		self.assertEqual(price.job(['28']), ('price.py 28', {'threshold': 0, 'days': 28, 'top': 10}))
+		self.assertEqual(price.job(['midsession']), ('price.py midsession', {'mode': 'midsession'}))
+		self.assertEqual(price.job(['premarket'])[1], {'mode': 'premarket'})
+		self.assertEqual(price.job(['interday'])[1], {'mode': 'interday'})
+		self.assertEqual(price.job(['close'])[1], {'mode': 'close'})
+		self.assertEqual(price.job(['28']), ('price.py 28', {'mode': 'period', 'threshold': 0, 'days': 28, 'top': 10}))
 		for bad in ([], ['soon'], ['7', '8']):
 			with self.assertRaises(ValueError):
 				price.job(bad)
@@ -164,7 +164,7 @@ def test_cli_runs_the_handler(self):
 		"""Good arguments run the report with what job() made of them."""
 		calls = []
 		delivery.cli('price.py', price.job, lambda **k: calls.append(k), ['midsession'])
-		self.assertEqual(calls, [{'midsession': True}])
+		self.assertEqual(calls, [{'mode': 'midsession'}])
 
 	def test_each_scheduled_report_runs_from_its_top_level_script(self):
 		"""./price.py and the rest run lib/reports/<name>.py, so cron lines and runs by hand keep working."""
@@ -511,7 +511,7 @@ def test_run_now_runs_one_command_and_leaves_the_schedule_alone(self):
 		"""run_now runs one command and leaves the schedule's timing alone."""
 		base, last = dict(self.s.base), dict(self.s.last)
 		self.assertEqual(self.s.run_now('7'), ('price.py close', 'done'))
-		self.assertEqual(self.ran, [('price', {'close': True})])
+		self.assertEqual(self.ran, [('price', {'mode': 'close'})])
 		self.assertEqual((self.s.base, self.s.last), (base, last))
 		self.assertTrue(all(not job.running for job in self.s.jobs))
 		self.assertEqual(self.s.busy, set())
@@ -534,7 +534,7 @@ def test_a_scheduled_run_skips_a_command_that_a_manual_run_is_in_the_middle_of(s
 		"""A scheduled run skips a command a manual run is in the middle of, and runs the rest of its line."""
 		self.s.busy.add('trades.py')
 		self.assertEqual(self.s.run_job(self.s.jobs[1], at(13, 21, 28)), 'already running')
-		self.assertEqual(self.ran, [('price', {'close': True})]) # price.py close ran, trades.py was skipped
+		self.assertEqual(self.ran, [('price', {'mode': 'close'})]) # price.py close ran, trades.py was skipped
 
 	def test_run_now_times_out(self):
 		"""A manual run past the timeout says so."""
@@ -584,7 +584,7 @@ def test_button_press_runs_that_command(self):
 		data = next(b.data for row in markup for b in row if b.label.startswith('7.'))
 		self.sent.clear()
 		worker.process_callback('telegram', 'cb1', '55', '9', data)
-		self.assertEqual(self.ran, [('price', {'close': True})])
+		self.assertEqual(self.ran, [('price', {'mode': 'close'})])
 		self.assertEqual(self.sent, [('text', ['7. price close: done'])])
 		self.assertEqual(sent, ['Running price close…'])
 
@@ -639,7 +639,7 @@ def test_run_in_chat(self):
 		self.assertEqual(self.ran, [('trades', {})])
 		self.ran.clear()
 		self.assertEqual(self.command('.cron run price close')[0][1], ['7. price close: done'])
-		self.assertEqual(self.ran, [('price', {'close': True})])
+		self.assertEqual(self.ran, [('price', {'mode': 'close'})])
 
 	def test_run_in_chat_refusals(self):
 		""".jobs run refuses an unscheduled command, an unknown number and two commands, else gives its usage."""

From 956bb9928cab68eaaebe5677c263a153a86a3ed3 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 03:05:14 +0000
Subject: [PATCH 40/81] rankings: one ratio table and one sorted list for .pe,
 .fpe and .peg (R1)

prepare_value_payload ran to 82 lines, nested seven deep, over eight
action strings in four near-identical branches. Now:
- RATIOS maps pe, forward pe and peg to Yahoo's fields.
- The action splits into a mode (top, bottom or negative) and a ratio.
- One filter keeps either the negative ratios only or none of them.
- One list of (ratio, chart label, line), sorted once, feeds both the
  caption and the chart.

The caption and the chart used to sort separately, the caption by
last_col, which parsed the number back out of each formatted line.
last_col goes. The order of equal ratios is unchanged (ascending, and
reversed for bottom).

Checks:
- Against HEAD's version on 6,712 random cases (ties, zeros, negatives
  and missing ratios, single tickers, analyst coverage), every payload
  and chart is identical.
- The output comparison is identical in all 159 steps.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/reports/rankings.py | 116 ++++++++++++++--------------------------
 1 file changed, 39 insertions(+), 77 deletions(-)

diff --git a/lib/reports/rankings.py b/lib/reports/rankings.py
index bc2d4947..61e39d3f 100644
--- a/lib/reports/rankings.py
+++ b/lib/reports/rankings.py
@@ -8,13 +8,6 @@
 from lib import render
 from lib import yahoo
 
-def last_col(line):
-	"""Sort key: the number a report line ends with; a line without one sorts first."""
-	try:
-		return float(line.split()[-1])
-	except ValueError:
-		return float('-inf')
-
 def prepare_marketcap_payload(service, action='top', length=15):
 	"""The `length` tracked stocks of largest (action 'top') or smallest ('bottom') market cap, compared in US dollars at Yahoo's
 	exchange rates: a market cap in won or yen is a far bigger number. Any other action is one ticker's, in its own currency too."""
@@ -68,84 +61,53 @@ def prepare_rating_payload(service, action, length=15):
 		tuple(shown), value_format='number', xlim=(max(0.9, low - spread * 0.9), min(5.1, high + spread * 0.25)))
 	return [service.bold(message)] + payload, chart
 
+RATIOS = {'pe': 'price_to_earnings_trailing', 'forward pe': 'price_to_earnings_forward', 'peg': 'price_to_earnings_peg'}
+
 def prepare_value_payload(service, action='pe', ticker_select=None, length=15):
-	"""(payload, chart): the tracked stocks by a ratio (`action`), or one ticker's. A forward P/E is analysts' estimate, so its lists
-	take only stocks that min_analysts or more analysts cover, as .buy does."""
-	payload = []
-	chart_rows = [] # (ticker, ratio)
+	"""(payload, chart): the tracked stocks by a ratio of RATIOS, or one ticker's. action is the ratio, or 'bottom' or
+	'negative' and the ratio: lowest first, highest first ('bottom'), or the negative ones only, lowest first. A forward
+	P/E is analysts' estimate, so its lists take only stocks that min_analysts or more analysts cover, as .buy does."""
+	mode, kind = action.split(' ', 1) if action.split()[0] in ('bottom', 'negative') else ('top', action)
+	field = RATIOS[kind]
 	tickers = [ticker_select] if ticker_select else watchlist.with_holdings()
 	market_data = yahoo.fetch(tickers)
-	if action in ('peg', 'bottom peg', 'negative peg'): # the PEG is in Yahoo's details
+	if kind == 'peg': # the PEG is in Yahoo's details
 		market_data |= yahoo.fetch_details(market_data, strict=bool(ticker_select))
 	covered = None # for a forward P/E list: the tickers enough analysts cover
-	if 'forward' in action and not ticker_select:
-		covered = {ticker for ticker, detail in yahoo.fetch_details(market_data).items() if detail.get('recommend_analysts', 0) >= config.prefs.min_analysts}
-	for ticker in market_data:
-		if covered is not None and ticker not in covered:
+	if kind == 'forward pe' and not ticker_select:
+		covered = {ticker for ticker, detail in yahoo.fetch_details(market_data).items()
+			if detail.get('recommend_analysts', 0) >= config.prefs.min_analysts}
+	rows = [] # (ratio, chart label, line), one sorted list for the caption and the chart
+	for ticker, data in market_data.items():
+		ratio = data.get(field)
+		if ratio is None or (covered is not None and ticker not in covered):
 			continue
-		trailing_ratio = None
-		forward_ratio = None
-		peg_ratio = None
-		ratio = None
-		if action in ('pe', 'bottom pe', 'forward pe', 'bottom forward pe'):
-			trailing_ratio = market_data.get(ticker).get('price_to_earnings_trailing')
-			forward_ratio = market_data.get(ticker).get('price_to_earnings_forward')
-			if action in ('forward pe', 'bottom forward pe'):
-				if not ticker_select and forward_ratio is not None and forward_ratio < 0:
-					continue
-				ratio = forward_ratio
-			else:
-				if not ticker_select and trailing_ratio is not None and trailing_ratio < 0:
-					continue
-				ratio = trailing_ratio
-		elif action == 'negative forward pe':
-			forward_ratio = market_data.get(ticker).get('price_to_earnings_forward')
-			if not ticker_select and (forward_ratio is None or forward_ratio >= 0):
-				continue
-			ratio = forward_ratio
-		elif action in ('peg', 'bottom peg'):
-			peg_ratio = market_data.get(ticker).get('price_to_earnings_peg')
-			if not ticker_select and peg_ratio is not None and peg_ratio < 0:
-				continue
-			ratio = peg_ratio
-		elif action == 'negative peg':
-			peg_ratio = market_data.get(ticker).get('price_to_earnings_peg')
-			if not ticker_select and (peg_ratio is None or peg_ratio >= 0):
-				continue
-			ratio = peg_ratio
-		if ratio is None:
+		if not ticker_select and (ratio < 0) != (mode == 'negative'): # a list: the negative ones, or none of them
 			continue
-		profile_title = market_data.get(ticker).get('profile_title')
-		ticker_link = links.finance_link(ticker, market_data.get(ticker).get('profile_exchange'), service)
-		flag = markets.flag_from_ticker(ticker)
-		if ticker_select and action in {'pe', 'forward pe'}:
-			parts = []
-			if trailing_ratio is not None:
-				parts.append(f"trailing {trailing_ratio}")
-			if forward_ratio is not None:
-				parts.append(f"forward {forward_ratio}")
-			payload.append(f"{flag} {profile_title} ({ticker_link}) PE: {', '.join(parts)}")
-		else:
-			payload.append(f"{flag} {profile_title} ({ticker_link}) {ratio}")
-			chart_rows.append((ticker.split('.')[0], float(ratio)))
-	payload.sort(key=last_col)
-	chart_rows.sort(key=lambda row: row[1])
-	chart = None
-	if not ticker_select:
-		heading_type = "Bottom" if 'bottom' in action else "Top"
-		heading_trail = action.replace('bottom ', '')
-		if 'bottom' in action:
-			payload.reverse()
-			chart_rows.reverse()
-		payload = payload[:length]
-		coverage = f" ({config.prefs.min_analysts}+ analysts)" if covered is not None else ''
-		if not payload:
-			payload = [f"No tracked stocks with {config.prefs.min_analysts}+ analysts have a {heading_trail} ratio" if covered is not None else f"{action}: no data found"]
+		link = links.finance_link(ticker, data.get('profile_exchange'), service)
+		start = f"{markets.flag_from_ticker(ticker)} {data.get('profile_title')} ({link})"
+		if ticker_select and mode == 'top' and kind in ('pe', 'forward pe'): # one stock's P/E: trailing and forward
+			both = (('trailing', 'price_to_earnings_trailing'), ('forward', 'price_to_earnings_forward'))
+			parts = ', '.join(f'{name} {data[key]}' for name, key in both if data.get(key) is not None)
+			rows.append((ratio, None, f"{start} PE: {parts}"))
 		else:
-			heading = f'{heading_type} {length} tracked stocks by {heading_trail} ratio{coverage}'
-			payload.insert(0, f"{service.bold(heading)}")
-			chart = render.Chart('bars', heading, rows=tuple(chart_rows[:length]), value_format='number')
-	return payload, chart
+			rows.append((ratio, ticker.split('.')[0], f"{start} {ratio}"))
+	rows.sort(key=lambda row: row[0])
+	if mode == 'bottom':
+		rows.reverse()
+	if ticker_select:
+		return [line for _, _, line in rows], None
+	rows = rows[:length]
+	trail = kind if mode != 'negative' else action # what the heading calls the ratio
+	if not rows:
+		if covered is not None:
+			return [f"No tracked stocks with {config.prefs.min_analysts}+ analysts have a {trail} ratio"], None
+		return [f"{action}: no data found"], None
+	coverage = f" ({config.prefs.min_analysts}+ analysts)" if covered is not None else ''
+	heading = f"{'Bottom' if mode == 'bottom' else 'Top'} {length} tracked stocks by {trail} ratio{coverage}"
+	bars = tuple((label, float(ratio)) for ratio, label, _ in rows)
+	chart = render.Chart('bars', heading, rows=bars, value_format='number')
+	return [service.bold(heading)] + [line for _, _, line in rows], chart
 
 def prepare_beta_payload(service):
 	"""(payload, chart) of the tracked stocks with a beta over 1.5 and a market cap under US$2B, the highest beta first."""

From 93a8c4e5a374a9943b6b287737bfe7d9ae806fba Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 03:10:36 +0000
Subject: [PATCH 41/81] profile: sections joined by blank lines, exchange pages
 from a table (R2)

prepare_profile_payload ran to 177 lines. It is now a list of section
functions, each a list of lines:
- about: the description;
- company: location, classification, employees, links;
- figures: size, debt, dates, dividend, cashflow, profit;
- trends: revenue and earnings trend marks;
- outlook: forecasts, insiders, short interest, score;
- valuation: price and ratios;
- changes: price changes over time;
- footer: the extra links.
The sections with lines are joined by one blank line. That replaces
the six "if payload: append('')" steps and the groupby pass that took
out the doubled blank lines afterwards.

The exchange pages (ASX, HKSE, every Nasdaq tier, NYSE, TWSE, JPX)
move into EXCHANGE_PAGES; another exchange still gets a search.

One difference: a company with neither price history nor a website
used to end its profile with a blank line, and now does not.

Checks:
- Against HEAD's version on 1,200 random companies (every field present
  or not, five market states, ten exchanges, a found .AX listing, price
  history present, empty or failing, footer and hyperlink settings, all
  three networks), the profiles are identical, but for that trailing
  blank line in 225 of them.
- The output comparison is identical in all 159 steps.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/reports/profile.py | 231 ++++++++++++++++++++++-------------------
 1 file changed, 127 insertions(+), 104 deletions(-)

diff --git a/lib/reports/profile.py b/lib/reports/profile.py
index 233fe52b..989501a4 100644
--- a/lib/reports/profile.py
+++ b/lib/reports/profile.py
@@ -8,7 +8,6 @@
 from lib import markets
 from lib import yahoo
 from lib import simplywallst
-from itertools import groupby
 
 TRENDS = {(True, False): '🔼', (False, False): '🔽', (True, True): '🔺', (False, True): '🔻'} # (rose, below zero)
 
@@ -29,177 +28,201 @@ def trend_marks(figures):
 	return '❌' * (3 - len(marks)) + marks
 
 MARKET_STATE_EMOJI = {'REGULAR': '🟢', 'PRE': '🟠', 'POST': '🟠'} # any other state is closed: 🔴
+EXCHANGE_PAGES = { # Yahoo's name for an exchange: (the name shown, or None for Yahoo's; the stock's page there)
+	'ASX': ('ASX', 'https://www2.asx.com.au/markets/company/{code}'),
+	'HKSE': ('HKSE', 'https://www.hkex.com.hk/Market-Data/Securities-Prices/Equities/Equities-Quote'
+		'?sym={code}&sc_lang=en'),
+	'Nasdaq': (None, 'https://www.nasdaq.com/market-activity/stocks/{lower}'), # every Nasdaq tier
+	'NYSE': ('NYSE', 'https://www.nyse.com/quote/XNYS:{ticker}'),
+	'Taiwan': ('TWSE', 'https://mis.twse.com.tw/stock/fibest.jsp?stock={code}&lang=en_us'),
+	'Tokyo': ('JPX', 'https://quote.jpx.co.jp/jpx/template/quote.cgi?F=tmp/e_stock_detail&MKTN=T&QCODE={code}'),
+}
 
 def prepare_profile_payload(service, user, ticker):
-	"""The .SYMBOL report: what Yahoo knows about one company. A field Yahoo did not send leaves its line out."""
+	"""The .SYMBOL report: what Yahoo knows about one company, in sections a blank line apart. A field Yahoo did not
+	send leaves its line out, and a section with none its blank line."""
 	ticker_orig = markets.transform_to_yahoo(ticker)
-	now = datetime.datetime.now()
-	payload = []
 	ticker, market_data = yahoo.resolve(ticker_orig, lambda symbol: yahoo.fetch_detail(symbol, 600))
 	if ticker not in market_data:
 		return [f"{user} 🛑 Beep Boop. I could not find {ticker_orig}"]
 	data = market_data[ticker]
-	code = ticker.split('.')[0]
-	title = data.get('profile_title', '')
 	state = data.get('marketState', '').rstrip()
-	currency = data.get('currency', '')
-	exchange = data.get('profile_exchange', '').replace('NasdaqCM', 'Nasdaq').replace('NasdaqGS', 'Nasdaq').replace('NYSEArca', 'NYSE')
+	exchange = data.get('profile_exchange', '').replace('NasdaqCM', 'Nasdaq').replace('NasdaqGS', 'Nasdaq')
+	exchange = exchange.replace('NYSEArca', 'NYSE')
+	sections = (about(data), company(service, data, ticker, exchange), figures(service, data), trends(data),
+		outlook(service, data), valuation(service, data, state), changes(service, data, ticker, state),
+		footer(service, data, ticker, exchange))
+	payload = []
+	for section in sections:
+		if section:
+			payload += ([""] if payload else []) + section
 	ticker_link = links.finance_link(ticker, exchange, service, brief=False)
-	gfinance_link = links.gfinance_link(ticker, exchange, service, brief=True, text='google') # all three before exchange is renamed below
-	sws_link = service.link(simplywallst.get_url(ticker, title, exchange), 'simplywall.st')
-	macrotrends_link = service.link('https://www.google.com/search?q=site:macrotrends.net+' + title + '+PE Ratio+' + code + '&btnI', 'macrotrends')
-	if 'profile_bio' in data:
-		payload.append(formats.make_paragraphs(data['profile_bio']))
-
-	if exchange == 'ASX':
-		exchange_url = 'https://www2.asx.com.au/markets/company/' + code
-	elif exchange == 'HKSE':
-		exchange_url = 'https://www.hkex.com.hk/Market-Data/Securities-Prices/Equities/Equities-Quote?sym=' + code + '&sc_lang=en'
-	elif 'Nasdaq' in exchange:
-		exchange_url = 'https://www.nasdaq.com/market-activity/stocks/' + ticker.lower()
-	elif exchange == 'NYSE':
-		exchange_url = 'https://www.nyse.com/quote/XNYS:' + ticker
-	elif exchange == 'Taiwan':
-		exchange = 'TWSE'
-		exchange_url = 'https://mis.twse.com.tw/stock/fibest.jsp?stock=' + code + '&lang=en_us'
-	elif exchange == 'Tokyo':
-		exchange = 'JPX'
-		exchange_url = 'https://quote.jpx.co.jp/jpx/template/quote.cgi?F=tmp/e_stock_detail&MKTN=T&QCODE=' + code
+	light = MARKET_STATE_EMOJI.get(state, '🔴') if state else ''
+	heading = f"{service.bold(data.get('profile_title', ''))} ({ticker_link}) {light}"
+	if ticker_orig == ticker:
+		payload.insert(0, heading)
 	else:
-		exchange_url = 'https://www.google.com/search?q=stock+exchange+' + exchange + '+' + code + '&btnI'
-	exchange_link = service.link(exchange_url, exchange)
-	location = [data[key] for key in ('profile_city', 'profile_state', 'profile_country') if key in data]
-	if payload:
-		payload.append("")
+		payload[:0] = [f"I could not find {ticker_orig} but I found {ticker_link}:", "", heading]
+	if len(payload) < 2:
+		payload.append("no data found")
+	return payload
 
+def exchange_page(exchange, ticker):
+	"""(the exchange's name to show, the stock's page there): a search for the page on one EXCHANGE_PAGES lacks."""
+	code = ticker.split('.')[0]
+	key = 'Nasdaq' if 'Nasdaq' in exchange else exchange
+	if key not in EXCHANGE_PAGES:
+		return exchange, 'https://www.google.com/search?q=stock+exchange+' + exchange + '+' + code + '&btnI'
+	name, page = EXCHANGE_PAGES[key]
+	return name or exchange, page.format(code=code, ticker=ticker, lower=ticker.lower())
+
+def about(data):
+	"""What the company does, in paragraphs."""
+	return [formats.make_paragraphs(data['profile_bio'])] if 'profile_bio' in data else []
+
+def company(service, data, ticker, exchange):
+	"""Where it is, what it is, its size, and its pages: its website, its exchange's page and a few more."""
+	lines = []
+	location = [data[key] for key in ('profile_city', 'profile_state', 'profile_country') if key in data]
 	if location:
-		payload.append(service.bold("Location:") + " " + ', '.join(location))
+		lines.append(service.bold("Location:") + " " + ', '.join(location))
 	if 'profile_industry' in data and 'profile_sector' in data:
-		payload.append(service.bold("Classification: ") + f"{data['profile_industry']}, {data['profile_sector']}")
+		lines.append(service.bold("Classification: ") + f"{data['profile_industry']}, {data['profile_sector']}")
 	if 'profile_employees' in data:
-		payload.append(service.bold("Employees: ") + f"{data['profile_employees']:,}")
+		lines.append(service.bold("Employees: ") + f"{data['profile_employees']:,}")
 	if 'profile_website' in data:
-		sites = [service.link(data['profile_website'], formats.strip_url(data['profile_website'])), exchange_link]
+		code = ticker.split('.')[0]
+		name, page = exchange_page(exchange, ticker)
+		website = data['profile_website']
+		sites = [service.link(website, formats.strip_url(website)), service.link(page, name)]
 		if 'NYSE' in exchange or 'Nasdaq' in exchange:
-			sites += [service.link('https://finviz.com/quote.ashx?t=' + ticker, 'finviz'), macrotrends_link,
+			title = data.get('profile_title', '')
+			macrotrends = f'https://www.google.com/search?q=site:macrotrends.net+{title}+PE Ratio+{code}&btnI'
+			sites += [service.link('https://finviz.com/quote.ashx?t=' + ticker, 'finviz'),
+				service.link(macrotrends, 'macrotrends'),
 				service.link('https://seekingalpha.com/symbol/' + ticker, 'seekingalpha')]
 		elif exchange == 'ASX':
 			sites.append(service.link('https://www.shortman.com.au/stock?q=' + code.lower(), 'shortman'))
-		payload.append(service.bold("Links: ") + ' | '.join(sites))
-
-	if payload:
-		payload.append("")
+		lines.append(service.bold("Links: ") + ' | '.join(sites))
+	return lines
 
+def figures(service, data):
+	"""Its size, debt, dates, dividend, cashflow and whether it makes a profit (always said, if only as unknown)."""
+	now = datetime.datetime.now()
+	lines = []
 	if 'currency' in data and 'market_cap' in data:
 		cap = f"{yahoo.market_cap_currency(data)} {formats.human_units(data['market_cap'])}" # GBP, not pence
-		payload.append(service.bold("Mkt cap:") + f" {cap}")
-	if all(key in data for key in ('shareholder_equity', 'total_debt', 'total_cash', 'profile_industry')) and data['shareholder_equity'] and 'Bank' not in data['profile_industry']:
-		net_debt_equity_ratio = round((data['total_debt'] - data['total_cash']) / data['shareholder_equity'] * 100)
-		if net_debt_equity_ratio > 0:
-			payload.append(service.bold("Net debt/equity ratio:") + f" {net_debt_equity_ratio}%" + ('⚠️ ' if net_debt_equity_ratio > 40 else ''))
+		lines.append(service.bold("Mkt cap:") + f" {cap}")
+	if all(key in data for key in ('shareholder_equity', 'total_debt', 'total_cash', 'profile_industry')) and \
+			data['shareholder_equity'] and 'Bank' not in data['profile_industry']:
+		ratio = round((data['total_debt'] - data['total_cash']) / data['shareholder_equity'] * 100)
+		if ratio > 0:
+			lines.append(service.bold("Net debt/equity ratio:") + f" {ratio}%" + ('⚠️ ' if ratio > 40 else ''))
 	if 'earnings_date' in data:
 		earnings_date = datetime.datetime.fromtimestamp(data['earnings_date'])
 		if earnings_date > now:
-			payload.append(service.bold("Earnings date:") + f" {earnings_date:%b %d}")
+			lines.append(service.bold("Earnings date:") + f" {earnings_date:%b %d}")
 	if data.get('dividend', 0) > 0:
-		payload.append(service.bold("Dividend:") + f" {data['dividend']}%")
+		lines.append(service.bold("Dividend:") + f" {data['dividend']}%")
 		if 'ex_dividend_date' in data:
 			ex_dividend_date = datetime.datetime.fromtimestamp(data['ex_dividend_date'])
 			if ex_dividend_date > now:
-				payload.append(service.bold("Ex-dividend date:") + f" {ex_dividend_date:%b %d}")
+				lines.append(service.bold("Ex-dividend date:") + f" {ex_dividend_date:%b %d}")
 	cashflow = data.get('free_cashflow', data.get('operating_cashflow'))
 	if cashflow:
-		payload.append(service.bold("Cashflow positive:") + (" no ⚠️ " if cashflow < 0 else " yes"))
+		lines.append(service.bold("Cashflow positive:") + (" no ⚠️ " if cashflow < 0 else " yes"))
 	if 'net_income' in data:
-		payload.append(service.bold("Profitable:") + (" no ⚠️ " if data['net_income'] <= 0 else " yes"))
+		lines.append(service.bold("Profitable:") + (" no ⚠️ " if data['net_income'] <= 0 else " yes"))
 	else:
-		payload.append(service.bold("Profitable:") + " unknown ⚠️")
-
-	if payload:
-		payload.append("")
-
-	if 'earningsQ' in data:
-		figures = (('revenueQ', 'quarterly revenue'), ('earningsQ', 'quarterly earnings'),
-			('revenueY', 'annual revenue'), ('earningsY', 'annual earnings'))
-		deltas = [(trend_marks(data[key]), label) for key, label in figures]
-		payload += [f"{delta}\t{label} delta" for delta, label in deltas] # trend_marks always gives at least three
-
-	if payload:
-		payload.append("")
-
+		lines.append(service.bold("Profitable:") + " unknown ⚠️")
+	return lines
+
+def trends(data):
+	"""Trend marks of the last quarters' and years' revenue and earnings."""
+	if 'earningsQ' not in data:
+		return []
+	figures = (('revenueQ', 'quarterly revenue'), ('earningsQ', 'quarterly earnings'),
+		('revenueY', 'annual revenue'), ('earningsY', 'annual earnings'))
+	return [f"{trend_marks(data[key])}\t{label} delta" for key, label in figures] # always at least three marks
+
+def outlook(service, data):
+	"""The analysts' forecasts and score, insiders' buying or selling, and the short interest."""
+	lines = []
 	for key, label in (('revenueEstimateY', 'Revenue forecast (1Y):'), ('earningsEstimateY', 'Earnings forecast (1Y):')):
 		if key in data:
 			growth = int(round(data[key]))
-			payload.append(service.bold(label) + f" {formats.signed_percent(growth)} " + ('' if growth > 0 else '⚠️ '))
+			lines.append(service.bold(label) + f" {formats.signed_percent(growth)} " + ('' if growth > 0 else '⚠️ '))
+	currency = data.get('currency', '')
 	if 'insiderBuy' in data:
 		if data['insiderBuy'] > data['insiderSell']:
 			value = formats.human_units(data['insiderBuyValue'])
-			payload.append(service.bold("Net insider action (3M):") + f" Buy {currency} {value} ")
+			lines.append(service.bold("Net insider action (3M):") + f" Buy {currency} {value} ")
 		elif data['insiderBuy'] < data['insiderSell']:
 			value = formats.human_units(data['insiderSellValue'])
-			payload.append(service.bold("Net insider action (3M):") + f" Sell {currency} {value}⚠️ ")
+			lines.append(service.bold("Net insider action (3M):") + f" Sell {currency} {value}⚠️ ")
 	if 'short_percent' in data:
-		payload.append(service.bold("Shorted stock:") + f" {data['short_percent']}%" + ('⚠️ ' if data['short_percent'] > 10 else ''))
+		shorted = data['short_percent']
+		lines.append(service.bold("Shorted stock:") + f" {shorted}%" + ('⚠️ ' if shorted > 10 else ''))
 	if 'recommend' in data:
-		payload.append(service.bold("Score:") + f" {data['recommend_index']} {data['recommend'].replace('_', ' ')} ({data['recommend_analysts']} analysts)")
-
-	if payload:
-		payload.append("")
-
+		rating = data['recommend'].replace('_', ' ')
+		score = f"{data['recommend_index']} {rating} ({data['recommend_analysts']} analysts)"
+		lines.append(service.bold("Score:") + f" {score}")
+	return lines
+
+def valuation(service, data, state):
+	"""The price (after hours too, out of session) and the price ratios, with a warning where one is high."""
+	lines = []
+	currency = data.get('currency', '')
 	if 'regularMarketPrice' in data:
 		price_line = service.bold("Price:") + f" {currency} {data['regularMarketPrice']:,.2f}"
 		if state != 'REGULAR' and 'prePostMarketPrice' in data:
 			price_line += f" ({data['prePostMarketPrice']:,.2f} after hrs)"
-		payload.append(price_line)
+		lines.append(price_line)
 	trailing_pe = str(int(round(data['price_to_earnings_trailing']))) if 'price_to_earnings_trailing' in data else 'N/A ⚠️ '
 	if data.get('net_income', 0) > 0:
-		payload.append(service.bold("Trailing P/E:") + f" {trailing_pe}")
+		lines.append(service.bold("Trailing P/E:") + f" {trailing_pe}")
 	if data.get('netIncomeToCommon', 0) > 0 and data.get('sharesOutstanding') and 'regularMarketPrice' in data:
 		current_pe = round(data['regularMarketPrice'] / (data['netIncomeToCommon'] / data['sharesOutstanding']))
-		payload.append(service.bold("Current P/E:") + f" {current_pe}")
+		lines.append(service.bold("Current P/E:") + f" {current_pe}")
 	if 'price_to_earnings_forward' in data:
 		forward_pe = int(round(data['price_to_earnings_forward']))
 		warn = forward_pe < 0
 		if 'profile_industry' in data: # software is dear: it warns above 100, the rest above 30 or above their trailing P/E
 			warn = warn or forward_pe > (100 if 'Software' in data['profile_industry'] else 30)
 			warn = warn or ('price_to_earnings_trailing' in data and forward_pe > int(trailing_pe))
-		payload.append(service.bold("Forward P/E:") + f" {forward_pe} " + ('⚠️ ' if warn else ''))
+		lines.append(service.bold("Forward P/E:") + f" {forward_pe} " + ('⚠️ ' if warn else ''))
 	if 'price_to_earnings_peg' in data:
-		payload.append(service.bold("PEG ratio:") + f" {round(data['price_to_earnings_peg'], 1)}")
+		lines.append(service.bold("PEG ratio:") + f" {round(data['price_to_earnings_peg'], 1)}")
 	if 'price_to_sales' in data:
 		price_to_sales = round(data['price_to_sales'], 1)
-		payload.append(service.bold("PS ratio:") + f" {price_to_sales} " + ('⚠️ ' if price_to_sales > 15 else ''))
-
-	if payload:
-		payload.append("")
+		lines.append(service.bold("PS ratio:") + f" {price_to_sales} " + ('⚠️ ' if price_to_sales > 15 else ''))
+	return lines
 
+def changes(service, data, ticker, state):
+	"""The price's change over 5 years, a year, a month and a day, and out of session the pre- or post-market move."""
+	lines = []
 	try:
-		price_history, _ = yahoo.price_history(ticker) # the percentages only: the profile has no chart
+		history, _ = yahoo.price_history(ticker) # the percentages only: the profile has no chart
 	except RuntimeError as e: # no chart data: the profile goes out without these lines
 		print(ticker, e, file=sys.stderr)
-		price_history = {}
+		history = {}
 	for interval in ('5Y', '1Y', '1M', '1D'):
-		if interval in price_history:
-			percent = price_history[interval]
+		if interval in history:
+			percent = history[interval]
 			label = service.bold(interval + ':')
-			payload.append(f"{formats.get_emoji(percent)} {label} {formats.signed_percent(percent)}")
+			lines.append(f"{formats.get_emoji(percent)} {label} {formats.signed_percent(percent)}")
 	if state != 'REGULAR':
 		for key, label in (('percent_change_premarket', 'Pre-market:'), ('percent_change_postmarket', 'Post-market:')):
 			if key in data:
 				change = data[key]
-				payload.append(f"{formats.get_emoji(change)} {service.bold(label)} {formats.signed_percent(change)}")
+				lines.append(f"{formats.get_emoji(change)} {service.bold(label)} {formats.signed_percent(change)}")
 				break
-
-	if 'profile_website' in data and config.prefs.hyperlinkFooter and config.prefs.hyperlink:
-		if payload:
-			payload.append("")
-		payload.append(f"{gfinance_link} | {links.yahoo_link(ticker, service, brief=True, text='yahoo')} | {sws_link}")
-	heading = f"{service.bold(title)} ({ticker_link}) {MARKET_STATE_EMOJI.get(state, '🔴') if state else ''}"
-	if ticker_orig == ticker:
-		payload.insert(0, heading)
-	else:
-		payload[:0] = [f"I could not find {ticker_orig} but I found {ticker_link}:", "", heading]
-	payload = [i[0] for i in groupby(payload)] # de-dupe white space
-	if len(payload) < 2:
-		payload.append("no data found")
-	return payload
+	return lines
+
+def footer(service, data, ticker, exchange):
+	"""Links to the company on Google Finance, Yahoo and Simply Wall St, with hyperlinkFooter (and hyperlinks) on."""
+	if 'profile_website' not in data or not (config.prefs.hyperlinkFooter and config.prefs.hyperlink):
+		return []
+	google = links.gfinance_link(ticker, exchange, service, brief=True, text='google')
+	sws = service.link(simplywallst.get_url(ticker, data.get('profile_title', ''), exchange), 'simplywall.st')
+	return [f"{google} | {links.yahoo_link(ticker, service, brief=True, text='yahoo')} | {sws}"]

From e776b1e5309100f6351873db1bfd29ffe180b108 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 03:20:38 +0000
Subject: [PATCH 42/81] Yahoo: one cache helper, raw answers cached everywhere
 (R9)

fetch() cached its parsed quotes while fetch_detail and fetch_chart_json cached Yahoo's answers as sent, so a fix to
the quote parsing waited out the cache (or a flush). _cached(name, seconds, get, *args) now serves all three: it reads
the cache file while it is fresh, else calls the raw fetch (_quote_answer, _detail_answer, _chart_answer) and writes
its answer. Quotes are cached under a new name (finbot_yahoo_quote_*), so the old parsed files are never read; an error
answer is still not cached. price_history's unused seconds parameter goes.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md               |  3 +-
 lib/yahoo.py            | 90 +++++++++++++++++++++++------------------
 tests/test_providers.py | 33 +++++++++++++++
 3 files changed, 85 insertions(+), 41 deletions(-)

diff --git a/CLAUDE.md b/CLAUDE.md
index 7606af34..77c13b97 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -51,7 +51,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - `lib/dialects.py`: each network's markup and message sizes. `lib/markets.py`: market tables, flags, ticker translation.
   `lib/formats.py`: names, units, dates, percents in words. `lib/links.py`: stock links. `lib/render.py`: every chart,
   `Chart` specs and `draw()`.
-- `lib/yahoo.py lib/sharesight.py lib/shortman.py lib/simplywallst.py`: data sources, cached in `DATA_DIR/cache`.
+- `lib/yahoo.py lib/sharesight.py lib/shortman.py lib/simplywallst.py`: data sources. Yahoo's and Sharesight's answers
+  are cached in `DATA_DIR/cache` as sent, not parsed, so a parser fix needs no flush (`yahoo._cached`).
   `sharesight.portfolios(name=None)` gives `Portfolio` objects: `holdings()`, `tickers()`, `performance(days)`,
   `trades(days)`.
 - `lib/cache.py`: cache files (any may vanish at any time). `lib/state.py`: state files in `DATA_DIR`, which must last;
diff --git a/lib/yahoo.py b/lib/yahoo.py
index 176edf85..35d879bf 100644
--- a/lib/yahoo.py
+++ b/lib/yahoo.py
@@ -127,6 +127,17 @@ def _convert(convert, value):
 	except (TypeError, ValueError, OverflowError):
 		return None
 
+def _cached(name, seconds, get, *args):
+	"""get(*args), Yahoo's answer as sent: from the cache file `name` while it is under `seconds` old (cache_seconds
+	when None) and the cache preference is on, else fetched and written there unless empty. Cached raw, not parsed, so
+	that a parser fix needs no flush."""
+	if config.prefs.cache and (cached := cache.read(name, seconds)):
+		return cached
+	data = get(*args)
+	if data and config.prefs.cache:
+		cache.write(name, data)
+	return data
+
 QUOTE_FIELDS = ( # (key, Yahoo's name, conversion) for the fields of a v7 quote; one Yahoo leaves out, sends as null or garbles is left out
 	('percent_change', 'regularMarketChangePercent', _round2),
 	('dividend', 'trailingAnnualDividendRate', lambda v: round(float(v), 1)),
@@ -153,21 +164,14 @@ def fetch(tickers):
 		print("no tickers provided to yahoo.fetch()", file=sys.stderr)
 		return {}
 	tickers = sorted(set(tickers)) # de-dupe
-	cacheFile = "finbot_yahoo_" + hashlib.sha256("_".join(tickers).encode()).hexdigest() + '.json'
-	if config.prefs.cache and (cached := cache.read(cacheFile, 300)):
-		return cached
-	cookie, crumb = credentials()
-	r = _get_first(_hosts('/v7/finance/quote?crumb=' + crumb + '&symbols=' + ','.join(tickers)), {**HEADERS, 'Content-type': 'application/json', 'Cookie': cookie})
-	if r is None:
-		print("Exhausted Yahoo API attempts. Giving up", file=sys.stderr)
-		raise RuntimeError("Yahoo API unavailable")
-	data = r.json()['quoteResponse']
-	if data['result'] is None or data['error'] is not None:
+	name = "finbot_yahoo_quote_" + hashlib.sha256("_".join(tickers).encode()).hexdigest() + '.json'
+	data = _cached(name, 300, _quote_answer, tickers)
+	if data is None:
 		print(f"{tickers}†", sep=' ', end='', flush=True, file=sys.stderr)
 		return {} # callers iterate the result
 	now = datetime.datetime.now().timestamp()
 	yahoo_output = {}
-	for item in data['result']:
+	for item in data['quoteResponse']['result']:
 		title = next((name for name in (item.get('longName'), item.get('shortName')) if isinstance(name, str) and name), None)
 		if not title or item.get('regularMarketPrice') is None:
 			continue
@@ -180,10 +184,20 @@ def fetch(tickers):
 		if stamps: # the first still to come, else the announced date
 			quote['earnings_date'] = next((stamp for stamp in stamps if stamp > now), stamps[0])
 		yahoo_output[item['symbol']] = quote
-	if config.prefs.cache:
-		cache.write(cacheFile, yahoo_output)
 	return yahoo_output
 
+def _quote_answer(tickers):
+	"""The v7 answer for the tickers, None for an error (not cached); raises RuntimeError when Yahoo is unreachable."""
+	cookie, crumb = credentials()
+	query = '/v7/finance/quote?crumb=' + crumb + '&symbols=' + ','.join(tickers)
+	r = _get_first(_hosts(query), {**HEADERS, 'Content-type': 'application/json', 'Cookie': cookie})
+	if r is None:
+		print("Exhausted Yahoo API attempts. Giving up", file=sys.stderr)
+		raise RuntimeError("Yahoo API unavailable")
+	data = r.json()
+	quotes = data['quoteResponse']
+	return data if quotes['result'] is not None and quotes['error'] is None else None
+
 DETAIL_FIELDS = ( # (key, path in the quoteSummary result[, conversion]); a field Yahoo leaves out, sends as null or garbles is left out
 	('profile_bio', ('summaryProfile', 'longBusinessSummary')),
 	('profile_city', ('summaryProfile', 'city')),
@@ -236,21 +250,11 @@ def fetch(tickers):
 RECOMMENDATION = ('recommend', 'recommend_index', 'recommend_analysts') # all three, or none
 
 def fetch_detail(ticker, seconds=None):
-	"""quoteSummary for one ticker: {ticker: {...}}, or {} when Yahoo has nothing usable (no name or no price)."""
-	if seconds is None:
-		seconds = config.prefs.cache_seconds
+	"""One ticker's quoteSummary: {ticker: {...}}, cached `seconds` (or cache_seconds); {} without a name or price."""
 	now = datetime.datetime.now()
-	cacheFile = "finbot_yahoo_detail_" + ticker + '.json'
-	data = cache.read(cacheFile, seconds) if config.prefs.cache else None
-	if not data:
-		cookie, crumb = credentials() # only on a miss: a cached detail needs neither, nor the lock
-		modules = 'calendarEvents,defaultKeyStatistics,balanceSheetHistoryQuarterly,financialData,summaryProfile,summaryDetail,price,earnings,earningsTrend,insiderTransactions'
-		r = _get_first(_hosts('/v10/finance/quoteSummary/' + ticker + '?modules=' + modules + '&crumb=' + crumb), {**HEADERS, 'Cookie': cookie})
-		if r is None:
-			return {} # catches some delisted stocks like "DRNA"
-		data = r.json()
-		if config.prefs.cache:
-			cache.write(cacheFile, data)
+	data = _cached("finbot_yahoo_detail_" + ticker + '.json', seconds, _detail_answer, ticker)
+	if data is None:
+		return {} # catches some delisted stocks like "DRNA"
 	result = dig(data, 'quoteSummary', 'result', 0)
 	if result is None: # Yahoo answers result=null for unknown symbols
 		print(ticker + 'x', sep=' ', end='', flush=True, file=sys.stderr)
@@ -307,6 +311,16 @@ def fetch_detail(ticker, seconds=None):
 		detail['insiderSell'], detail['insiderSellValue'] = totals['Sale']
 	return {ticker: dict(sorted(detail.items()))}
 
+DETAIL_MODULES = ('calendarEvents,defaultKeyStatistics,balanceSheetHistoryQuarterly,financialData,summaryProfile,'
+	'summaryDetail,price,earnings,earningsTrend,insiderTransactions') # the quoteSummary modules fetch_detail reads
+
+def _detail_answer(ticker):
+	"""Yahoo's quoteSummary answer for one ticker, with every module finbot reads, or None when no host gives one."""
+	cookie, crumb = credentials() # only on a miss: a cached detail needs neither, nor the lock
+	query = '/v10/finance/quoteSummary/' + ticker + '?modules=' + DETAIL_MODULES + '&crumb=' + crumb
+	r = _get_first(_hosts(query), {**HEADERS, 'Cookie': cookie})
+	return None if r is None else r.json()
+
 class Line(typing.NamedTuple):
 	"""The prices behind a price_history, for its chart: the company's name, Yahoo's full name for its exchange (as
 	fetch() gives profile_exchange), the currency, the current price (else the last close) and the closes over the span,
@@ -322,11 +336,9 @@ class NoData(answer.Answer, RuntimeError):
 	"""Yahoo has nothing for it (no such symbol, no prices): a reply in a chat, a RuntimeError where a list skips it."""
 
 
-def price_history(ticker, days=None, seconds=None):
+def price_history(ticker, days=None):
 	"""({period: percent change}, its Line): the change over `days`, else over each standard period, and the closes
 	behind it, for a chart. Raises NoData when there are no prices, RuntimeError when the history cannot be fetched."""
-	if seconds is None:
-		seconds = config.prefs.cache_seconds
 	import pandas as pd
 	from pandas.tseries.offsets import BDay as businessday
 	percent_dict = {}
@@ -454,15 +466,16 @@ def prefetch_history(tickers, workers=8):
 def fetch_chart_json(ticker, days=3665, seconds=None, full=False):
 	"""Yahoo's daily chart for the last `days`, or with full=True its entire history (cached separately). Raises NoData
 	for a symbol Yahoo does not know, RuntimeError when Yahoo cannot be reached or refuses."""
-	if seconds is None:
-		seconds = config.prefs.cache_seconds
+	name = "finbot_yahoo_history_" + ticker + ("_max" if full else "") + ".json"
+	return _cached(name, seconds, _chart_answer, ticker, days, full)
+
+def _chart_answer(ticker, days, full):
+	"""The v8 chart answer: daily bars of the last `days`, or all with full=True. Raises as fetch_chart_json does."""
 	now = datetime.datetime.now()
-	cacheFile = "finbot_yahoo_history_" + ticker + ("_max" if full else "") + ".json"
-	if config.prefs.cache and (cached := cache.read(cacheFile, seconds)):
-		return cached
 	cookie, crumb = credentials()
 	start = '0' if full else str(int((now - datetime.timedelta(days=days)).timestamp()))
-	url = f'https://query1.finance.yahoo.com/v8/finance/chart/{ticker}?period1={start}&period2={int(now.timestamp())}&interval=1d&crumb={crumb}'
+	url = (f'https://query1.finance.yahoo.com/v8/finance/chart/{ticker}?period1={start}&period2={int(now.timestamp())}'
+		f'&interval=1d&crumb={crumb}')
 	print("Fetching", net.redact(url), file=sys.stderr) if config.prefs.debug else None
 	try:
 		r = net.get(url, headers={**HEADERS, 'Content-type': 'application/json', 'Cookie': cookie}, timeout=config.prefs.http_timeout)
@@ -472,10 +485,7 @@ def fetch_chart_json(ticker, days=3665, seconds=None, full=False):
 		raise NoData(f"no price history for {ticker}")
 	if r.status_code != 200:
 		raise RuntimeError(f"{r.status_code} error for {ticker} from Yahoo's chart API")
-	data = r.json()
-	if config.prefs.cache:
-		cache.write(cacheFile, data)
-	return data
+	return r.json()
 
 def not_found(response):
 	"""Whether a chart API answer is Yahoo saying it has no such symbol, rather than some other 404."""
diff --git a/tests/test_providers.py b/tests/test_providers.py
index bcf66d50..0fa22dec 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -100,6 +100,31 @@ def test_empty_result_is_none(self):
 		self.http(yahoo, Resp({'quoteResponse': {'result': None, 'error': {'code': 'x'}}}))
 		self.assertEqual(yahoo.fetch(['NOPE']), {})
 
+	def test_an_error_answer_is_not_cached(self):
+		"""Yahoo's error answer is no quotes, and is not cached: the next fetch asks again."""
+		error = Resp({'quoteResponse': {'result': None, 'error': {'code': 'x'}}})
+		self.http(yahoo, error, Resp(sample('yahoo', 'sample_yahoo_v7.json')))
+		self.assertEqual(yahoo.fetch(['AAPL']), {})
+		self.assertIn('AAPL', yahoo.fetch(['AAPL']))
+
+	def test_the_cache_holds_yahoos_answer_so_a_parser_fix_needs_no_flush(self):
+		"""Quotes are cached as Yahoo sent them, not parsed: a change to the parsing applies to a cached answer."""
+		body = sample('yahoo', 'sample_yahoo_v7.json')
+		self.http(yahoo, Resp(body))
+		yahoo.fetch(['AAPL'])
+		self.http(yahoo, ConnectionError('down')) # from here on, only the cache answers
+		self.patch(yahoo, 'QUOTE_FIELDS', yahoo.QUOTE_FIELDS + (('short_name', 'shortName', None),))
+		self.assertEqual(yahoo.fetch(['AAPL'])['AAPL']['short_name'], 'Apple Inc.')
+
+	def test_with_the_cache_off_every_fetch_asks_yahoo(self):
+		"""With the cache preference off, every fetch asks Yahoo, and nothing is written to the cache."""
+		self.prefs(cache=False)
+		getter = self.http(yahoo, Resp(sample('yahoo', 'sample_yahoo_v7.json')))
+		yahoo.fetch(['AAPL'])
+		yahoo.fetch(['AAPL'])
+		self.assertEqual(len(getter.calls), 2)
+		self.assertEqual([name for name in os.listdir(config.env.cache_dir) if name.startswith('finbot_yahoo_')], [])
+
 	def test_no_tickers_is_none(self):
 		"""No tickers asked for: no quotes, and no request."""
 		self.assertEqual(yahoo.fetch([]), {})
@@ -141,6 +166,14 @@ def refuse():
 		self.patch(yahoo, 'credentials', refuse)
 		self.assertIn('NVDA', yahoo.fetch_detail('NVDA'))
 
+	def test_the_cache_holds_yahoos_answer_so_a_parser_fix_needs_no_flush(self):
+		"""A detail is cached as Yahoo sent it, not parsed: a change to the parsing applies to a cached answer."""
+		self.http(yahoo, Resp(self.combined()))
+		yahoo.fetch_detail('NVDA')
+		self.http(yahoo, ConnectionError('down')) # from here on, only the cache answers
+		self.patch(yahoo, 'DETAIL_FIELDS', yahoo.DETAIL_FIELDS + (('short_name', ('price', 'shortName')),))
+		self.assertEqual(yahoo.fetch_detail('NVDA')['NVDA']['short_name'], 'NVIDIA Corporation')
+
 	def test_http_failure_gives_empty(self):
 		"""A detail Yahoo will not give (404, unreachable) is empty, not an error."""
 		for failure in (Resp(status=404), ConnectionError('down')):

From 8e76b3c5b5b7881f72cb1884c1e7404e4790a8fc Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 03:29:52 +0000
Subject: [PATCH 43/81] Scheduler: a deadline per command, a lock, and no
 gevent (M6, D17)

- A command still running after JOB_TIMEOUT (5 minutes) is stopped and reported: run_child waits on its child with a
  timeout, and the rest of the line still runs. The 5 minutes used to cover a whole line, so the Saturday line's five
  commands shared them.
- A lock covers what the loop, a reload and runs from chat share: load() swapping the jobs, their bases and the record
  of runs while due() and tick() read them, and the check-then-set of a job's running flag and of the busy commands.
  Under gevent these were atomic only because greenlets never preempt one another.
- lib/jobs.py imports no gevent: a due job starts in a daemon thread, the loop sleeps with time.sleep and children come
  from the stdlib subprocess, each cooperative under bot.py's monkey patching.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 README.md          |   2 +-
 lib/jobs.py        | 187 +++++++++++++++++++++++++--------------------
 tests/test_jobs.py |  76 ++++++++++++++----
 3 files changed, 166 insertions(+), 99 deletions(-)

diff --git a/README.md b/README.md
index 019b59c7..00173278 100644
--- a/README.md
+++ b/README.md
@@ -642,7 +642,7 @@ The bot logs the number of jobs when it starts or reloads, and each job as it ru
 
 Behaviour worth knowing:
 * A run that was due while the bot was down (a restart, a deploy) still fires once if it was due within the last `scheduler_catchup` minutes (default 5). A restart right after a run does not repeat it.
-* A job still running when it is next due is skipped, and a job that takes more than 5 minutes is abandoned and reported.
+* A job still running when it is next due is skipped. A command still running after 5 minutes is stopped and reported, and the rest of its line still runs: each command has 5 minutes of its own.
 * Failures are reported to your chat channels, as before.
 * With `scheduler=False` nothing is scheduled by the bot, so you can keep using cron or systemd timers and run the scripts yourself.
 
diff --git a/lib/jobs.py b/lib/jobs.py
index 5ccfd9ff..58b03c2e 100644
--- a/lib/jobs.py
+++ b/lib/jobs.py
@@ -10,14 +10,15 @@
 import os
 import re
 import signal
+import subprocess
 import sys
+import threading
+import time
 import zlib
 import zoneinfo
 from dataclasses import dataclass, field
 from pathlib import Path
 
-import gevent
-from gevent import subprocess
 from croniter import croniter, CroniterBadDateError
 
 from lib import config
@@ -31,7 +32,7 @@
 current = None # the running Scheduler, set by bot.py; the chat commands use it
 HIDDEN = ('cleanup', 'podcasts') # scheduled, but not offered in the chat list
 STATE_FILE = 'finbot_jobs.json' # last fire time of every job, so a restart neither repeats a run nor loses one
-JOB_TIMEOUT = 5 * 60 # seconds a job may run before it is abandoned and reported
+JOB_TIMEOUT = 5 * 60 # seconds each command may run before it is stopped and reported
 IGNORED_VARIABLES = ('MAILTO', 'SHELL', 'PATH', 'HOME', 'LOGNAME')
 OPERATOR = re.compile(r'\d*>>?&?\d*|&>>?|;|&&|\|\|?|&|[<`$(){}]') # what a shell reads as more than a word, outside quotes
 REDIRECT = re.compile(r'\d*>>?&?\d*|&>>?') # > file, >> file, 2>&1, &> file
@@ -43,22 +44,37 @@ def checksum(label):
 
 
 def run_child(label, module, kwargs):
-	"""Run a command in its own process (python -m lib.jobrun) and return its exit status. The child inherits the environment,
-	user and output of the bot, so it behaves like the old cron job. A timeout or any other interruption stops the child."""
+	"""Run a command in its own process (python -m lib.jobrun) and return its exit status. The child inherits the
+	environment, user and output of the bot, so it behaves like the old cron job. One still running after JOB_TIMEOUT
+	is stopped and TimeoutError raised; any other interruption stops it too."""
 	env = dict(os.environ, **config.child_environment()) # the preferences this process runs on, not whatever the file says now
-	proc = subprocess.Popen([sys.executable, '-m', 'lib.jobrun', module.__name__, json.dumps(kwargs)], cwd=config.REPO_DIR, env=env)
+	args = [sys.executable, '-m', 'lib.jobrun', module.__name__, json.dumps(kwargs)]
+	proc = subprocess.Popen(args, cwd=config.REPO_DIR, env=env)
 	try:
-		return proc.wait()
+		return proc.wait(timeout=JOB_TIMEOUT)
+	except subprocess.TimeoutExpired:
+		stop(proc)
+		raise TimeoutError(f'still running after {JOB_TIMEOUT // 60} minutes') from None
 	except BaseException:
-		proc.terminate()
-		try:
-			proc.wait(timeout=10)
-		except subprocess.TimeoutExpired:
-			proc.kill()
-			proc.wait()
+		stop(proc)
 		raise
 
 
+def stop(proc):
+	"""End a child process: asked to stop, then killed if it is still there 10 seconds later."""
+	proc.terminate()
+	try:
+		proc.wait(timeout=10)
+	except subprocess.TimeoutExpired:
+		proc.kill()
+		proc.wait()
+
+
+def start_thread(target, *args):
+	"""Run target(*args) in a daemon thread of its own (a greenlet, under gevent's monkey patching)."""
+	threading.Thread(target=target, args=args, daemon=True).start()
+
+
 def exit_reason(status):
 	"""Why a job process ended, when it did not report it itself: killed by a signal, or failed before it could say."""
 	if status < 0:
@@ -70,7 +86,7 @@ def exit_reason(status):
 
 
 def run_in_process(label, module, kwargs):
-	"""The runner for tests: call the handler here instead of in a child process."""
+	"""The runner for tests: call the handler here instead of in a child process (so without a deadline)."""
 	module.run(**kwargs)
 	return 0
 
@@ -239,7 +255,8 @@ def parse_schedule(text, default_zone=None):
 
 class Scheduler:
 	"""Fires the jobs of a schedule file. `path` is a callable returning the file (the data dir copy wins over the default, so a
-	new copy is found on reload). `clock`, `sleep` and `spawn` can be replaced, which the tests do."""
+	new copy is found on reload). `clock`, `sleep` and `spawn` can be replaced, which the tests do. Its lock covers
+	what the loop, a reload and runs from chat share: the jobs, their bases, the record of runs and what is running."""
 
 	def __init__(self, path, catchup=None, clock=None, sleep=None, spawn=None, zone=None, runner=None):
 		"""path() gives the schedule file's path. catchup is how late, in minutes, a missed run may still start (the
@@ -248,14 +265,15 @@ def __init__(self, path, catchup=None, clock=None, sleep=None, spawn=None, zone=
 		self.path = path
 		self.catchup = datetime.timedelta(minutes=config.env.scheduler_catchup if catchup is None else catchup)
 		self.clock = clock or (lambda: datetime.datetime.now(datetime.timezone.utc))
-		self.sleep = sleep or gevent.sleep
-		self.spawn = spawn or gevent.spawn
+		self.sleep = sleep or time.sleep
+		self.spawn = spawn or start_thread
 		self.runner = runner or run_child # (label, module, kwargs) -> exit status
 		self.zone = zone
 		self.jobs = []
 		self.base = {} # key -> the time after which the next run is due
 		self.last = {}
 		self.busy = set() # labels of the commands running now, scheduled or manual
+		self.lock = threading.Lock()
 
 	def load(self):
 		"""Read the schedule file. A line with an impossible time refuses the whole file: ImpossibleTime is raised, naming every such
@@ -274,19 +292,18 @@ def load(self):
 		if not jobs and self.jobs:
 			print('schedule: no usable line in', path, '- keeping the current schedule', file=sys.stderr)
 			return False
-		try:
-			last = state.load(STATE_FILE)
-		except (OSError, ValueError) as e: # unreadable or not JSON: carry on as if there were no record, the next run rewrites it
-			problem = f'cannot read {STATE_FILE}, starting without a record of past runs: {e}'
+		with self.lock: # no run is recorded between reading the record and swapping it in
+			try:
+				last, problem = state.load(STATE_FILE), None
+			except (OSError, ValueError) as e: # unreadable or not JSON: start without one; the next run rewrites it
+				last, problem = None, f'cannot read {STATE_FILE}, starting without a record of past runs: {e}'
+			self.last = last if isinstance(last, dict) else {}
+			now = self.clock()
+			old = {job.key: self.base[job.key] for job in self.jobs if job.key in self.base}
+			self.base = {job.key: old.get(job.key) or self.start_base(job, now) for job in jobs}
+			self.jobs = jobs
+		if problem:
 			delivery.report_error(ScheduleError(problem), context='schedule')
-			last = None
-		self.last = last if isinstance(last, dict) else {}
-		now = self.clock()
-		old = {job.key: self.base[job.key] for job in self.jobs if job.key in self.base}
-		self.base = {}
-		for job in jobs:
-			self.base[job.key] = old.get(job.key) or self.start_base(job, now)
-		self.jobs = jobs
 		print(f"schedule: {len(jobs)} job{'' if len(jobs) == 1 else 's'} from {path}", file=sys.stderr)
 		return True
 
@@ -321,66 +338,67 @@ def next_fire(self, job, after):
 	def due(self, now):
 		"""The jobs to run now, as (job, fire time). Several missed fire times of one job run it once."""
 		fired = []
-		for job in self.jobs:
-			base, latest = self.base[job.key], None
-			while (t := self.next_fire(job, base)) <= now:
-				latest, base = t, t
-			if latest:
-				self.base[job.key] = latest
-				self.last[job.key] = latest.isoformat()
-				fired.append((job, latest))
-		if fired:
-			try:
-				state.save(STATE_FILE, self.last)
-			except OSError as e:
-				print('schedule: cannot save', STATE_FILE, e, file=sys.stderr)
+		with self.lock:
+			for job in self.jobs:
+				base, latest = self.base[job.key], None
+				while (t := self.next_fire(job, base)) <= now:
+					latest, base = t, t
+				if latest:
+					self.base[job.key] = latest
+					self.last[job.key] = latest.isoformat()
+					fired.append((job, latest))
+			if fired:
+				try:
+					state.save(STATE_FILE, self.last)
+				except OSError as e:
+					print('schedule: cannot save', STATE_FILE, e, file=sys.stderr)
 		return fired
 
 	def run_job(self, job, fire_time):
-		"""One job: its commands in order. Skipped if the previous run is still going; abandoned after JOB_TIMEOUT."""
-		if job.running:
-			print(f'schedule: {job.text} is still running, skipping {fire_time:%H:%M}', file=sys.stderr)
-			return 'already running'
-		return self.execute(job, job.commands)
-
-	def execute(self, job, commands):
-		"""Run commands one after another. `job` (None for a manual run of one command) is marked running meanwhile. A command that
-		is already running, scheduled or manual, is skipped. Returns 'done', 'timed out', 'failed' or 'already running'."""
-		if job:
+		"""One job: its commands in order (execute). Skipped if the previous run is still going."""
+		with self.lock:
+			if job.running:
+				print(f'schedule: {job.text} is still running, skipping {fire_time:%H:%M}', file=sys.stderr)
+				return 'already running'
 			job.running = True
-		outcome, label = 'done', ''
-		timeout = gevent.Timeout(JOB_TIMEOUT)
-		timeout.start()
 		try:
-			for label, module, kwargs in commands:
-				if label in self.busy:
-					print(f'schedule: {label} is already running, skipping it', file=sys.stderr)
-					outcome = 'already running'
-					continue
-				self.busy.add(label)
-				print(f'schedule: running {label}', file=sys.stderr)
-				try:
-					status = self.runner(label, module, kwargs)
-					if status:
-						outcome = 'failed'
-						if status != jobrun.REPORTED: # it ended without saying why
-							delivery.report_error(RuntimeError(exit_reason(status)), context=label)
-				except gevent.Timeout:
-					raise
-				except (Exception, SystemExit) as e:
-					delivery.report_error(e, context=label)
+			return self.execute(job.commands)
+		finally:
+			job.running = False
+
+	def claim(self, label):
+		"""Mark a command running, unless it already is (scheduled or manual): whether it was claimed."""
+		with self.lock:
+			if label in self.busy:
+				return False
+			self.busy.add(label)
+			return True
+
+	def execute(self, commands):
+		"""Run commands one after another, each stopped and reported if still running after JOB_TIMEOUT. One already
+		running, scheduled or manual, is skipped. Returns 'done', 'timed out', 'failed' or 'already running'."""
+		outcome = 'done'
+		for label, module, kwargs in commands:
+			if not self.claim(label):
+				print(f'schedule: {label} is already running, skipping it', file=sys.stderr)
+				outcome = 'already running'
+				continue
+			print(f'schedule: running {label}', file=sys.stderr)
+			try:
+				status = self.runner(label, module, kwargs)
+				if status:
 					outcome = 'failed'
-				finally:
+					if status != jobrun.REPORTED: # it ended without saying why
+						delivery.report_error(RuntimeError(exit_reason(status)), context=label)
+			except TimeoutError as e: # run_child stopped it: the rest of the line still runs
+				delivery.report_error(e, context=label)
+				outcome = 'timed out'
+			except (Exception, SystemExit) as e:
+				delivery.report_error(e, context=label)
+				outcome = 'failed'
+			finally:
+				with self.lock:
 					self.busy.discard(label)
-		except gevent.Timeout as t:
-			if t is not timeout:
-				raise
-			delivery.report_error(TimeoutError(f'still running after {JOB_TIMEOUT // 60} minutes'), context=label)
-			outcome = 'timed out'
-		finally:
-			timeout.close()
-			if job:
-				job.running = False
 		return outcome
 
 	def commands(self):
@@ -427,7 +445,7 @@ def pick(self, number, crc):
 
 	def run_command(self, command):
 		"""Run one command returned by select() or pick(). Returns (label, outcome)."""
-		return command[0], self.execute(None, [command])
+		return command[0], self.execute([command])
 
 	def run_now(self, text):
 		"""`.jobs run`: one scheduled command, now, as cron would run it; the schedule is unchanged. Returns (label, outcome)."""
@@ -438,7 +456,8 @@ def tick(self):
 		now = self.clock()
 		for job, fire_time in self.due(now):
 			self.spawn(self.run_job, job, fire_time)
-		waits = [(self.next_fire(job, self.base[job.key]) - now).total_seconds() for job in self.jobs]
+		with self.lock:
+			waits = [(self.next_fire(job, self.base[job.key]) - now).total_seconds() for job in self.jobs]
 		return min(30, max(1, math.ceil(min(waits)))) if waits else 30
 
 	def run_forever(self):
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
index 94213c48..88840885 100644
--- a/tests/test_jobs.py
+++ b/tests/test_jobs.py
@@ -1,5 +1,6 @@
 """The in-process scheduler (lib/jobs.py): schedule file parsing, each script's job(argv), and firing with a fake clock."""
-import contextlib, datetime, hashlib, io, json, os, signal, subprocess, sys, tempfile, time, unittest, zoneinfo
+import contextlib, datetime, hashlib, io, json, os, signal, subprocess, sys, tempfile, threading, time, unittest
+import zoneinfo
 from pathlib import Path
 from unittest import mock
 import gevent
@@ -350,14 +351,21 @@ def quit_(**k):
 		self.assertEqual(len(self.errors), 1)
 		self.assertFalse(s.jobs[0].running)
 
-	def test_a_job_that_hangs_is_abandoned_and_reported(self):
-		"""A job that runs past the timeout is abandoned and reported."""
-		self.patch(jobs, 'JOB_TIMEOUT', 0.05)
-		self.patch(jobs.SCRIPTS['trades'], 'run', lambda **k: gevent.sleep(5))
+	def test_a_command_stopped_at_its_deadline_is_reported_and_the_next_still_runs(self):
+		"""A command still running at its deadline (run_child stops it) is reported, and the rest of its line runs."""
+		self.file.write_text('*/20 * * * * trades.py; shorts.py\n')
+		def runner(label, module, kwargs):
+			"""trades.py hangs until run_child stops it; the rest run here."""
+			if label == 'trades.py':
+				raise TimeoutError('still running after 5 minutes')
+			return jobs.run_in_process(label, module, kwargs)
 		s = self.scheduler()
-		s.run_job(s.jobs[0], at(13, 21, 20))
-		self.assertTrue(any('still running' in e for e in self.errors))
+		s.runner = runner
+		self.assertEqual(s.run_job(s.jobs[0], at(13, 21, 20)), 'timed out')
+		self.assertEqual(self.errors, ['⚠️ trades.py: TimeoutError: still running after 5 minutes'])
+		self.assertEqual(self.ran, [('shorts', {})])
 		self.assertFalse(s.jobs[0].running)
+		self.assertEqual(s.busy, set())
 
 	def test_reload_applies_edits_keeps_state_and_survives_a_bad_file(self):
 		"""A reload applies edits and keeps unchanged jobs' timing; an unusable file keeps the running schedule."""
@@ -537,12 +545,31 @@ def test_a_scheduled_run_skips_a_command_that_a_manual_run_is_in_the_middle_of(s
 		self.assertEqual(self.ran, [('price', {'mode': 'close'})]) # price.py close ran, trades.py was skipped
 
 	def test_run_now_times_out(self):
-		"""A manual run past the timeout says so."""
-		self.patch(jobs, 'JOB_TIMEOUT', 0.05)
-		self.patch(jobs.SCRIPTS['trades'], 'run', lambda **k: gevent.sleep(5))
+		"""A manual run stopped at its deadline says so."""
+		def hang(label, module, kwargs):
+			"""Hang until run_child stops it."""
+			raise TimeoutError('still running after 5 minutes')
+		self.s.runner = hang
 		self.assertEqual(self.s.run_now('8'), ('trades.py', 'timed out'))
 		self.assertEqual(self.s.busy, set())
 
+	def test_two_runs_of_one_command_at_once_run_it_once(self):
+		"""Of two runs of the same command started together, one runs it and the other is told it is running."""
+		def slow(**k):
+			"""Take a while, letting the other run start."""
+			time.sleep(0.2)
+			self.ran.append(('trades', k))
+		self.patch(jobs.SCRIPTS['trades'], 'run', slow)
+		outcomes = []
+		threads = [threading.Thread(target=lambda: outcomes.append(self.s.run_now('8'))) for _ in range(2)]
+		for thread in threads:
+			thread.start()
+		for thread in threads:
+			thread.join()
+		self.assertEqual(sorted(outcomes), [('trades.py', 'already running'), ('trades.py', 'done')])
+		self.assertEqual(self.ran, [('trades', {})])
+		self.assertEqual(self.s.busy, set())
+
 	def list_in_chat(self, text='.jobs'):
 		"""The lines .jobs (or `text`) answers with."""
 		return self.command(text)[0][1]
@@ -751,15 +778,36 @@ def ticker():
 		self.assertEqual(len(ticks), 5) # the loop was not blocked by the child
 		self.assertLess(ticks[-1] - started, 0.8)
 
-	def test_a_timeout_kills_the_child(self):
-		"""A timeout stops the child, rather than leaving it running."""
+	def test_a_child_still_running_at_its_deadline_is_stopped(self):
+		"""A child still running after JOB_TIMEOUT is stopped, rather than left running, and TimeoutError raised."""
+		self.patch(jobs, 'JOB_TIMEOUT', 0.3)
 		self.sleeper(30)
 		started = time.monotonic()
+		with self.assertRaises(TimeoutError):
+			jobs.run_child('cleanup.py', cleanup, {})
+		self.assertLess(time.monotonic() - started, 5)
+		self.assertIsNotNone(self.children[0].poll()) # the child was stopped, not left running
+
+	def test_another_interruption_stops_the_child_too(self):
+		"""An interruption while a child runs (a greenlet killed, say) stops the child as well."""
+		self.sleeper(30)
 		with self.assertRaises(gevent.Timeout):
 			with gevent.Timeout(0.3):
 				jobs.run_child('cleanup.py', cleanup, {})
-		self.assertLess(time.monotonic() - started, 5)
-		self.assertIsNotNone(self.children[0].poll()) # the child was stopped, not left running
+		self.assertIsNotNone(self.children[0].poll())
+
+	def test_each_command_of_a_line_has_a_deadline_of_its_own(self):
+		"""The deadline is per command (it once covered the whole line): two that each take most of it both finish."""
+		self.patch(jobs, 'JOB_TIMEOUT', 0.8)
+		self.sleeper(0.5)
+		file = Path(tempfile.mkdtemp(), 'schedule.cron')
+		file.write_text('* * * * * trades.py; shorts.py\n')
+		s = jobs.Scheduler(lambda: file, zone=UTC)
+		with contextlib.redirect_stderr(io.StringIO()):
+			self.assertTrue(s.load())
+			self.assertEqual(s.run_job(s.jobs[0], at(13, 21, 20)), 'done')
+		self.assertEqual(len(self.children), 2)
+		self.assertEqual(self.errors, [])
 
 
 

From 98091c869f99b9b3a1f9401b375ba230887e6ebe Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 03:33:46 +0000
Subject: [PATCH 44/81] bot.py: a bounded body read, and signal work in a
 thread (M6)

- A request's body is read as its CONTENT_LENGTH bytes and no more: pywsgi ends the stream there, but a WSGI server
  need not, and read() could then wait on a kept-alive connection. A body over MAX_BODY (1 MiB) is refused unread
  (413), and a length that is not a number is a bad request. The tests' requests carry CONTENT_LENGTH
  (tests/common.py wsgi_environ).
- SIGHUP and SIGUSR1 hand their work to a thread and return at once, one signal's work at a time: the reload may
  report to the chats, and the flush takes Yahoo's auth lock. Under gevent a handler already runs in a greenlet of its
  own, but a plain signal handler interrupts whatever the main thread is doing.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 bot.py               | 27 ++++++++++++++++++-----
 lib/inbound.py       |  1 +
 tests/common.py      |  5 +++++
 tests/test_edges.py  | 51 ++++++++++++++++++++++++++++++++++++++------
 tests/test_finbot.py |  2 +-
 5 files changed, 73 insertions(+), 13 deletions(-)

diff --git a/bot.py b/bot.py
index 5eb57cb7..6386e013 100755
--- a/bot.py
+++ b/bot.py
@@ -24,6 +24,7 @@
 from lib.reports import cleanup
 
 AUTHENTICATED = 'finbot.authenticated' # set in a request's WSGI environ once it has passed its network's check
+MAX_BODY = 1 << 20 # bytes in a request body, at most: an update from Telegram or Slack takes a few thousand
 _updates_seen = collections.OrderedDict() # (service, update id) of the last updates taken, oldest first
 
 def seen_before(service, update_id, keep=1000):
@@ -61,8 +62,9 @@ def main(environ, start_response):
 
 def handle(environ, start_response):
 	"""Route a request by its path to the network that posts there, turn it away unless it proves it comes from that
-	network (before its body is read, when the proof is a header), read the update, turn away repeats, and hand a
-	message or a button press to lib/worker in a thread of its own: the network gets its answer at once."""
+	network (before its body is read, when the proof is a header), read the update (CONTENT_LENGTH bytes, up to
+	MAX_BODY), turn away repeats, and hand a message or a button press to lib/worker in a thread of its own: the network
+	gets its answer at once."""
 	def respond(response):
 		"""Answer the request: a status line and a body."""
 		start_response(response.status, [('Content-type', 'application/json')])
@@ -82,7 +84,11 @@ def print_headers():
 		print_headers()
 		print("Fatal:", why, file=sys.stderr)
 		return respond(inbound.FORBIDDEN)
-	body = environ['wsgi.input'].read()
+	length = environ.get('CONTENT_LENGTH') or '0'
+	if not length.isdigit() or int(length) > MAX_BODY:
+		print(f"[{source.service}]: a body of {length!r} bytes refused unread", file=sys.stderr)
+		return respond(inbound.TOO_LARGE if length.isdigit() else inbound.BAD_REQUEST)
+	body = environ['wsgi.input'].read(int(length)) # the body and no more: a server need not end the stream there
 	try:
 		update = source.decode(body)
 	except ValueError:
@@ -137,6 +143,16 @@ def git_version(checkout=None):
 		return 'unknown'
 
 scheduler = None
+_signal_work = threading.Lock() # one signal's work at a time: a second reload waits for the first
+
+def in_thread(work):
+	"""A signal handler that runs `work` in a thread of its own and returns at once: a handler must not wait on the
+	network, nor on a lock that the code it interrupted may hold."""
+	def run():
+		"""The work, once any other signal's is done."""
+		with _signal_work:
+			work()
+	return lambda *args: threading.Thread(target=run, daemon=True).start()
 
 def reload_all(*args):
 	"""SIGHUP: re-read preferences.conf and, if the scheduler is running, schedule.cron. Each reports its own errors; a bad
@@ -204,8 +220,9 @@ def flush_cache(*args):
 	if scheduler:
 		jobs.current = scheduler # for the .jobs chat command
 		threading.Thread(target=scheduler.run_forever, daemon=True).start() # the scheduled reports, formerly cron jobs
-	gevent.signal_handler(signal.SIGHUP, reload_all) # startup is done. systemctl reload finbot: pick up edited preferences.conf and schedule.cron
-	gevent.signal_handler(signal.SIGUSR1, flush_cache) # systemctl kill -s USR1 finbot: delete the cache
+	# Startup is done. systemctl reload finbot: pick up edited preferences.conf and schedule.cron
+	gevent.signal_handler(signal.SIGHUP, in_thread(reload_all))
+	gevent.signal_handler(signal.SIGUSR1, in_thread(flush_cache)) # systemctl kill -s USR1 finbot: delete the cache
 	try:
 		httpd.serve_forever()
 	except OSError as e:
diff --git a/lib/inbound.py b/lib/inbound.py
index 54232b87..0e063d98 100644
--- a/lib/inbound.py
+++ b/lib/inbound.py
@@ -49,6 +49,7 @@ class Response:
 BAD_REQUEST = Response('400 Bad Request', b'<h1>Bad Request</h1>')
 FORBIDDEN = Response('403 Forbidden', b'<h1>Unauthorized</h1>')
 NOT_FOUND = Response('404 Not Found', b'<h1>404</h1>')
+TOO_LARGE = Response('413 Payload Too Large', b'<h1>Payload Too Large</h1>')
 
 
 SIGNATURE_AGE = 300 # seconds: Slack's limit; an older signed request may be a replay
diff --git a/tests/common.py b/tests/common.py
index a3d80995..e31693a9 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -28,6 +28,11 @@
 BOT = '@bot'
 
 
+def wsgi_environ(path, body, **headers):
+	"""A WSGI environ posting `body` (bytes) to `path`, with these headers (as HTTP_* keys)."""
+	return {'PATH_INFO': path, 'wsgi.input': io.BytesIO(body), 'CONTENT_LENGTH': str(len(body)), **headers}
+
+
 def skip_if_old_env(case):
 	"""Skip a test that starts bot.py while the checkout has a leftover .env (config.check_startup_environment)."""
 	if config.OLD_ENV.exists():
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 41e39709..c8869982 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -208,7 +208,7 @@ class WsgiRejections(FinbotCase):
 	def request(self, path='/telegram', body=None, headers=None):
 		"""(the body, the status) the WSGI handler answers a request with."""
 		body = body if isinstance(body, bytes) else json.dumps(body if body is not None else {}).encode()
-		env = {'PATH_INFO': path, 'wsgi.input': io.BytesIO(body), **(headers or {})}
+		env = wsgi_environ(path, body, **(headers or {}))
 		status = []
 		with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
 			out = bot.main(env, lambda s, *a, **k: status.append(s))
@@ -242,6 +242,24 @@ def test_malformed_body_is_refused_not_a_crash(self):
 			self.assertEqual(self.request(body=body, headers=self.SECRET)[1], '400 Bad Request', body)
 		self.assertEqual(self.errors, [])
 
+	def test_only_the_body_is_read_and_a_huge_one_not_at_all(self):
+		"""Only CONTENT_LENGTH bytes are read, since a server need not end the stream there; a body over MAX_BODY is
+		refused unread, and a length that is not a number is a bad request."""
+		self.patch(worker, 'process_request', lambda *a: None)
+		def post(body, length):
+			"""(the status, the bytes read) for this body and CONTENT_LENGTH, with more on the stream after it."""
+			env = {**wsgi_environ('/telegram', body, **self.SECRET), 'CONTENT_LENGTH': length}
+			env['wsgi.input'] = stream = io.BytesIO(body + b'{"the next request": 1}')
+			status = []
+			with contextlib.redirect_stderr(io.StringIO()):
+				bot.main(env, lambda s, *a, **k: status.append(s))
+			return status[-1], stream.tell()
+		message = {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.help'}
+		update = json.dumps({'message': message}).encode()
+		self.assertEqual(post(update, str(len(update))), ('200 OK', len(update)))
+		self.assertEqual(post(update, str(bot.MAX_BODY + 1)), ('413 Payload Too Large', 0))
+		self.assertEqual(post(update, 'lots'), ('400 Bad Request', 0))
+
 	def test_strangers_cannot_post_into_the_chats(self):
 		"""Strangers (wrong paths, no or wrong secrets, malformed bodies) are refused, and nothing reaches the chats."""
 		self.env(slackOutgoingWebhook='https://example.com/slack')
@@ -263,7 +281,7 @@ def broken(*a, **k):
 			"""Fail while routing."""
 			raise RuntimeError('boom')
 		self.patch(inbound, 'urlparse', broken) # fails while routing, before the secret is even looked at
-		env = {'PATH_INFO': '/telegram', 'wsgi.input': io.BytesIO(b'{}'), **self.SECRET}
+		env = wsgi_environ('/telegram', b'{}', **self.SECRET)
 		status = []
 		with contextlib.redirect_stderr(io.StringIO()) as err:
 			bot.main(env, lambda s, *a, **k: status.append(s))
@@ -304,7 +322,7 @@ def test_button_presses_are_logged(self):
 			body = json.dumps({'callback_query': {'id': '1', 'from': sender, 'data': 'j|7|abcd', 'message': {'message_id': 9, 'chat': {'id': 55}}}}).encode()
 			err = io.StringIO()
 			with contextlib.redirect_stderr(err):
-				bot.main({'PATH_INFO': '/telegram', 'wsgi.input': io.BytesIO(body), **secret}, lambda *a, **k: None)
+				bot.main(wsgi_environ('/telegram', body, **secret), lambda *a, **k: None)
 			self.assertIn(f'[Telegram button]: {who} chat 55 j|7|abcd', err.getvalue())
 
 
@@ -327,7 +345,7 @@ def event(self, event_id=None, **event):
 			'event': dict({'type': 'message', 'channel': 'C1', 'ts': '1.2'}, **event)}).encode()
 		status = []
 		with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
-			bot.main({'PATH_INFO': '/slack', 'wsgi.input': io.BytesIO(body)}, lambda s, *a, **k: status.append(s))
+			bot.main(wsgi_environ('/slack', body), lambda s, *a, **k: status.append(s))
 		gevent.sleep(0.05) # the worker runs in a (gevent) thread
 		return status[-1]
 
@@ -382,8 +400,7 @@ def post(self, secret='sss', age=0, token='stok', headers=None, tamper=False):
 		mac = hmac.new(secret.encode(), b'v0:' + timestamp.encode() + b':' + body, hashlib.sha256).hexdigest()
 		if headers is None:
 			headers = {'HTTP_X_SLACK_REQUEST_TIMESTAMP': timestamp, 'HTTP_X_SLACK_SIGNATURE': 'v0=' + mac}
-		env = {'PATH_INFO': '/slack', 'wsgi.input': io.BytesIO(body.replace(b'.price', b'.trades') if tamper else body),
-			**headers}
+		env = wsgi_environ('/slack', body.replace(b'.price', b'.trades') if tamper else body, **headers)
 		status = []
 		with contextlib.redirect_stderr(io.StringIO()):
 			bot.main(env, lambda s, *a, **k: status.append(s))
@@ -426,7 +443,7 @@ def test_a_telegram_name_is_escaped_where_it_arrives(self):
 		requests = []
 		self.patch(worker, 'process_request', lambda *a: requests.append(a))
 		body = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A<b>&'}, 'text': '.hello'}}
-		env = {'PATH_INFO': '/telegram', 'wsgi.input': io.BytesIO(json.dumps(body).encode()), 'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'tok'}
+		env = wsgi_environ('/telegram', json.dumps(body).encode(), HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN='tok')
 		with contextlib.redirect_stderr(io.StringIO()):
 			bot.main(env, lambda *a, **k: None)
 		gevent.sleep(0.05) # the worker runs in a (gevent) thread
@@ -484,6 +501,26 @@ def test_the_session_keeps_no_cookies(self):
 		self.assertEqual(len(net.session.cookies), 0)
 
 
+class SignalHandOff(unittest.TestCase):
+	def test_a_signal_handler_hands_its_work_to_a_thread_one_at_a_time(self):
+		"""A signal handler returns at once, and its work runs in a thread of its own, one signal's after another's."""
+		events = []
+		def work():
+			"""Record the start and the end, a while apart."""
+			events.append('start')
+			time.sleep(0.1)
+			events.append('end')
+		handler = bot.in_thread(work)
+		started = time.monotonic()
+		handler(signal.SIGHUP, None)
+		handler(signal.SIGHUP, None)
+		self.assertLess(time.monotonic() - started, 0.05) # neither waited for the work
+		deadline = time.monotonic() + 5
+		while len(events) < 4 and time.monotonic() < deadline:
+			time.sleep(0.01)
+		self.assertEqual(events, ['start', 'end', 'start', 'end']) # never two at once
+
+
 class StartupSignals(unittest.TestCase):
 	"""A reload (SIGHUP) that arrives while the bot is still starting must not kill it."""
 
diff --git a/tests/test_finbot.py b/tests/test_finbot.py
index 0de22387..73a5b55e 100644
--- a/tests/test_finbot.py
+++ b/tests/test_finbot.py
@@ -52,7 +52,7 @@ def post(message):
 			"""Post a Telegram message update to the WSGI handler: (its answer, what it logged)."""
 			body = json.dumps({'update_id': 1, 'message': dict({'message_id': 5, 'chat': {'id': 55, 'type': 'private'},
 				'from': {'id': 9, 'first_name': 'Rob'}, 'date': 1}, **message)}).encode()
-			env = {'PATH_INFO': '/telegram', 'wsgi.input': io.BytesIO(body), 'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'tok'}
+			env = wsgi_environ('/telegram', body, HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN='tok')
 			err = io.StringIO()
 			with contextlib.redirect_stderr(err):
 				out = bot.main(env, lambda *a, **k: None)

From 1a0e433be6e4ac71e184e77344479babbc98d14c Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 03:42:11 +0000
Subject: [PATCH 45/81] Slack: scheduled reports as the bot, charts and all,
 with slackChannelID (M8, D9)

A new startup key, slackChannelID: with slackBotToken, the scheduled reports (and error lines) go to that channel as
the bot, through chat.postMessage and the file upload, so they carry their charts as the replies do. The incoming
webhook, which takes text only, is then not used, and not needed. slackChannelID without the bot token gets a
startup warning. delivery.scheduled() picks each network's destination for default_chats(). The key is in finbot.env,
the README's table and the Slack guide, which gains the step.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 README.md               |  3 ++-
 defaults/finbot.env     |  2 ++
 doc/slack.md            | 14 ++++++++++----
 lib/config.py           |  8 ++++++--
 lib/delivery.py         | 29 +++++++++++++++++++++--------
 tests/test_config.py    |  8 +++++---
 tests/test_providers.py | 36 ++++++++++++++++++++++++++++--------
 7 files changed, 74 insertions(+), 26 deletions(-)

diff --git a/README.md b/README.md
index 00173278..054866d5 100644
--- a/README.md
+++ b/README.md
@@ -30,7 +30,7 @@ What finbot can do on each network. Telegram takes everything; the others take w
 | Chat commands | ✓ | ✓ | ✗ (a webhook only posts) |
 | "typing…" while a command works | ✓ | ✗ (Slack has no typing notice for bots) | ✗ |
 | Charts in replies | ✓ | ✓ with `slackBotToken` | n/a |
-| Charts in scheduled reports | ✓ | ✗ (the incoming webhook takes text only) | ✓ |
+| Charts in scheduled reports | ✓ | ✓ with `slackChannelID` and `slackBotToken` (the incoming webhook takes text only) | ✓ |
 | Period and job buttons | ✓ | ✗ | ✗ |
 | No link previews | ✓ | ✓ | ✓ |
 | Linked tickers | ✓ | ✓ | ✓ |
@@ -472,6 +472,7 @@ of the chat network keys it needs.
 | `discord_webhook` | | Discord webhook, for outbound reports |
 | `telegramBotToken` | | Telegram bot token, for outbound reports |
 | `slackBotToken` | | Slack bot token, for replies and chart uploads in Slack |
+| `slackChannelID` | | Channel id that scheduled reports go to on Slack as the bot, charts and all (with `slackBotToken`); `slack_webhook` is then not used |
 | `telegramOutgoingWebhook` | | Telegram webhook path the bot accepts (interactive mode) |
 | `telegramOutgoingToken` | | Secret token Telegram must present (interactive mode) |
 | `slackOutgoingWebhook` | | Slack webhook path the bot accepts (interactive mode) |
diff --git a/defaults/finbot.env b/defaults/finbot.env
index 96dc58bf..282edafe 100644
--- a/defaults/finbot.env
+++ b/defaults/finbot.env
@@ -32,6 +32,8 @@ slack_webhook=
 discord_webhook=
 telegramBotToken=
 slackBotToken=
+# Slack channel id for scheduled reports as the bot, charts and all (needs slackBotToken; slack_webhook is then unused)
+slackChannelID=
 
 # Chat networks: inbound (interactive mode)
 telegramOutgoingWebhook=
diff --git a/doc/slack.md b/doc/slack.md
index aca78b10..c23852d4 100644
--- a/doc/slack.md
+++ b/doc/slack.md
@@ -1,18 +1,19 @@
 # Slack
 
 finbot reaches Slack through a Slack app of your own. The app's incoming webhook posts the scheduled reports into a
-channel; the app's bot also answers commands. The bot builds on the webhook, which still carries the scheduled reports.
+channel; the app's bot also answers commands. The bot builds on the webhook, which carries the scheduled reports unless
+the bot posts them itself, charts and all (`slackChannelID`).
 
 ## Incoming webhook, or the interactive bot?
 
 | | Incoming webhook | Interactive bot (with the webhook) |
 |---|---|---|
-| Scheduled reports: trades, price alerts, earnings and dividend dates, performance and the rest | ✓, as text | ✓, as text, through the webhook |
+| Scheduled reports: trades, price alerts, earnings and dividend dates, performance and the rest | ✓, as text | ✓, as text through the webhook, or with their charts as the bot (`slackChannelID`) |
 | Chat commands: `.price`, `@finbot price` and the rest ([the list](../README.md#supported-commands)) | ✗ | ✓ |
 | Charts in the replies to commands | n/a | ✓, uploaded into the channel |
 | Period buttons and "typing…" | ✗ | ✗: Slack has no typing notice for bots, and finbot has no Slack buttons yet |
 | Public HTTPS server | Not needed | Needed: a domain name with a valid certificate |
-| `finbot.env` keys | `slack_webhook` | That one, plus `slackOutgoingWebhook`, `slackSigningSecret` and `slackBotToken` |
+| `finbot.env` keys | `slack_webhook` | That one, plus `slackOutgoingWebhook`, `slackSigningSecret` and `slackBotToken` (and `slackChannelID`, for charts in the scheduled reports) |
 
 Start with [Incoming webhook](#incoming-webhook). For commands, go on to [Interactive bot](#interactive-bot)
 afterwards, in the same app.
@@ -30,7 +31,8 @@ afterwards, in the same app.
 slack_webhook=https://hooks.slack.com/services/XXXXXXXXXXX/YYYYYYYYYYY/AAAAAAAAmockupAAAAAAAAAAAA
 ```
 
-An incoming webhook takes text only, so the scheduled reports come without charts.
+An incoming webhook takes text only, so the scheduled reports come without charts. The [interactive bot](#interactive-bot)
+can post them with their charts.
 
 ## Interactive bot
 
@@ -84,12 +86,16 @@ on a domain name with a valid certificate.
    without it, commands get no reply, and the log says so at startup and for each reply.
 8. Restart finbot, and invite the bot to the channel: `/invite @finbot`.
 9. Send `@finbot help` in the channel (or `.help`, with `message.channels`).
+10. For charts in the scheduled reports, set `slackChannelID` to the channel's id (in Slack, click the channel's name:
+    the _Channel ID_, which starts with C, is at the bottom of _About_), and restart finbot. The bot then posts the
+    scheduled reports there, as it posts its replies, and `slack_webhook` is no longer used.
 
 ```
 slack_webhook=https://hooks.slack.com/services/XXXXXXXXXXX/YYYYYYYYYYY/AAAAAAAAmockupAAAAAAAAAAAA
 slackOutgoingWebhook=https://www.example.com:8443/slack
 slackSigningSecret=0123456789abcdef0123456789abcdef
 slackBotToken=xoxb-your-bot-token
+slackChannelID=C0123456789
 ```
 
 A chart in a reply is uploaded into the channel with Slack's file upload API (`files.getUploadURLExternal`, then
diff --git a/lib/config.py b/lib/config.py
index eec75c71..20c39fd8 100644
--- a/lib/config.py
+++ b/lib/config.py
@@ -16,8 +16,8 @@
 
 STARTUP_KEYS = ('DATA_DIR', 'ip', 'port', 'telegramChatID', 'scheduler', 'scheduler_catchup',
 	'sharesight_code', 'sharesight_client_id', 'sharesight_client_secret', 'slack_webhook', 'discord_webhook', 'slackBotToken',
-	'slackOutgoingToken', 'slackOutgoingWebhook', 'slackSigningSecret', 'telegramBotToken', 'telegramOutgoingToken',
-	'telegramOutgoingWebhook')
+	'slackChannelID', 'slackOutgoingToken', 'slackOutgoingWebhook', 'slackSigningSecret', 'telegramBotToken',
+	'telegramOutgoingToken', 'telegramOutgoingWebhook')
 SECRET_KEYS = ('sharesight_code', 'sharesight_client_id', 'sharesight_client_secret', 'slack_webhook', 'discord_webhook', 'slackBotToken',
 	'slackOutgoingToken', 'slackOutgoingWebhook', 'slackSigningSecret', 'telegramBotToken', 'telegramOutgoingToken',
 	'telegramOutgoingWebhook') # never printed
@@ -175,6 +175,7 @@ class Env:
 	telegramOutgoingWebhook: str
 	telegramOutgoingToken: str
 	slackBotToken: str
+	slackChannelID: str # with slackBotToken, scheduled reports go to this channel as the bot, in place of slack_webhook
 	slackOutgoingWebhook: str
 	slackOutgoingToken: str
 	slackSigningSecret: str
@@ -210,6 +211,7 @@ def startup_values(getenv):
 		telegramOutgoingWebhook=r.str('telegramOutgoingWebhook', ''),
 		telegramOutgoingToken=r.str('telegramOutgoingToken', ''),
 		slackBotToken=r.str('slackBotToken', ''),
+		slackChannelID=r.str('slackChannelID', ''),
 		slackOutgoingWebhook=r.str('slackOutgoingWebhook', ''),
 		slackOutgoingToken=r.str('slackOutgoingToken', ''),
 		slackSigningSecret=r.str('slackSigningSecret', ''),
@@ -411,6 +413,8 @@ def startup_warnings():
 	warnings = []
 	if env.slackOutgoingWebhook and not env.slackBotToken:
 		warnings.append('slackOutgoingWebhook is set but slackBotToken is not: Slack commands will get no reply')
+	if env.slackChannelID and not env.slackBotToken:
+		warnings.append('slackChannelID is set but slackBotToken is not: scheduled reports cannot go there as the bot')
 	if env.slackOutgoingWebhook and not (env.slackSigningSecret or env.slackOutgoingToken):
 		warnings.append('slackOutgoingWebhook is set but neither slackSigningSecret nor slackOutgoingToken is:'
 			' every Slack request will be refused')
diff --git a/lib/delivery.py b/lib/delivery.py
index ce496b66..20277606 100644
--- a/lib/delivery.py
+++ b/lib/delivery.py
@@ -22,7 +22,6 @@ class Report:
 
 
 REPLIES = {'telegram': transports.TelegramBot, 'slack': transports.SlackBot} # a reply into the chat that asked
-SCHEDULED = {'telegram': transports.TelegramBot, 'slack': transports.SlackWebhook, 'discord': transports.DiscordWebhook}
 
 
 def reply_to(service, chat):
@@ -31,14 +30,28 @@ def reply_to(service, chat):
 	return kind(chat) if kind else None
 
 
+def scheduled(service, url):
+	"""Where a scheduled report goes on a network: Telegram's telegramChatID; Slack's slackChannelID as the bot, with
+	slackBotToken (so that charts go too), else its incoming webhook; Discord's webhook."""
+	env = config.env
+	if service == 'telegram':
+		return transports.TelegramBot(env.telegramChatID)
+	if service == 'slack' and env.slackChannelID and env.slackBotToken:
+		return transports.SlackBot(env.slackChannelID)
+	return {'slack': transports.SlackWebhook, 'discord': transports.DiscordWebhook}[service](url)
+
+
 def default_chats():
-	"""Where a scheduled report goes: Telegram's telegramChatID (left out without one), and the Slack and Discord
-	incoming webhooks. Raises RuntimeError when there is nowhere to post."""
-	chats = [SCHEDULED[service](config.env.telegramChatID if service == 'telegram' else url)
-		for service, url in config.env.webhooks.items() if service != 'telegram' or config.env.telegramChatID]
-	if not chats:
-		raise RuntimeError('nowhere to post: finbot.env sets no telegramChatID, slack_webhook or discord_webhook')
-	return chats
+	"""Where a scheduled report goes: each network's destination (scheduled), Telegram's left out without a
+	telegramChatID. Raises RuntimeError when there is nowhere to post."""
+	env = config.env
+	services = [service for service in env.webhooks if service != 'telegram' or env.telegramChatID]
+	if env.slackChannelID and env.slackBotToken and 'slack' not in services: # the bot needs no incoming webhook
+		services.append('slack')
+	if not services:
+		raise RuntimeError('nowhere to post: finbot.env sets no telegramChatID, slackChannelID, slack_webhook or'
+			' discord_webhook')
+	return [scheduled(service, env.webhooks.get(service)) for service in services]
 
 
 def targets(dest):
diff --git a/tests/test_config.py b/tests/test_config.py
index d3dc22f6..460c8f6b 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -32,7 +32,7 @@ class Tables(unittest.TestCase):
 	def test_the_two_key_sets_are_disjoint_and_cover_everything_config_reads(self):
 		"""Startup and preference keys do not overlap, and between them name every setting config.py reads."""
 		self.assertEqual(set(config.STARTUP_KEYS) & set(config.PREFERENCE_KEYS), set())
-		self.assertEqual((len(config.STARTUP_KEYS), len(config.PREFERENCE_KEYS)), (18, 26))
+		self.assertEqual((len(config.STARTUP_KEYS), len(config.PREFERENCE_KEYS)), (19, 26))
 		source = (REPO / 'lib' / 'config.py').read_text()
 		read = set(re.findall(r"""(?:\br\.\w+|getenv)\(\s*['"]([A-Za-z_]+)['"]""", source))
 		read -= {'STATE_DIRECTORY'} # systemd's own variable, a source for DATA_DIR
@@ -378,8 +378,8 @@ def test_a_discord_webhook_uses_the_native_endpoint(self):
 			self.assertEqual(values.webhooks['discord'], 'https://discord.com/api/webhooks/1/abc')
 
 	def test_settings_that_leave_part_of_the_bot_unable_to_work_are_warned_about(self):
-		"""A startup warning for Slack commands without slackBotToken (they would get no reply), and for an inbound
-		webhook without its secret (every request would be refused)."""
+		"""A startup warning for Slack commands or slackChannelID without slackBotToken (no reply, no reports there),
+		and for an inbound webhook without its secret (every request would be refused)."""
 		hooks = dict(slackOutgoingWebhook='https://example.com/slack', slackOutgoingToken='stok',
 			slackBotToken='xoxb-1', telegramOutgoingWebhook='https://example.com/telegram', telegramOutgoingToken='tok')
 		no_reply = 'slackOutgoingWebhook is set but slackBotToken is not: Slack commands will get no reply'
@@ -387,7 +387,9 @@ def test_settings_that_leave_part_of_the_bot_unable_to_work_are_warned_about(sel
 			' every Slack request will be refused')
 		telegram_refused = ('telegramOutgoingWebhook is set but telegramOutgoingToken is not: every Telegram update'
 			' will be refused')
+		no_bot = 'slackChannelID is set but slackBotToken is not: scheduled reports cannot go there as the bot'
 		cases = (({}, []), ({'slackBotToken': ''}, [no_reply]), ({'slackOutgoingToken': ''}, [slack_refused]),
+			({'slackChannelID': 'C9'}, []), ({'slackChannelID': 'C9', 'slackBotToken': ''}, [no_reply, no_bot]),
 			({'slackOutgoingToken': '', 'slackSigningSecret': 's'}, []),
 			({'telegramOutgoingToken': ''}, [telegram_refused]), ({'telegramOutgoingWebhook': ''}, []))
 		for changes, warnings in cases:
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 0fa22dec..ed36204b 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -655,15 +655,29 @@ def upload_steps(self):
 		return [Resp({'ok': True, 'upload_url': self.UPLOAD_URL, 'file_id': 'F1'}), Resp(None, text='OK - 3'), Resp({'ok': True, 'files': [{'id': 'F1'}]})]
 
 	def test_where_charts_can_go(self):
-		"""Charts always go to Telegram and Discord; Slack gets them only in replies with the bot token."""
+		"""Charts always go to Telegram and Discord; Slack gets them as the bot, with the bot token, but not through its
+		incoming webhook."""
 		self.env(slackBotToken='xoxb-test')
 
-		kinds = (delivery.REPLIES['telegram'], delivery.SCHEDULED['telegram'], delivery.REPLIES['slack'],
-			delivery.SCHEDULED['slack'], delivery.SCHEDULED['discord'])
-		charted = [kind(None).takes_charts for kind in kinds]
-		self.assertEqual(charted, [True, True, True, False, True]) # Slack: replies, not scheduled reports
+		kinds = (transports.TelegramBot, transports.SlackBot, transports.SlackWebhook, transports.DiscordWebhook)
+		self.assertEqual([kind(None).takes_charts for kind in kinds], [True, True, False, True])
 		self.env(slackBotToken=False)
-		self.assertFalse(delivery.REPLIES['slack'](None).takes_charts)
+		self.assertFalse(transports.SlackBot(None).takes_charts)
+
+	def test_scheduled_reports_go_to_slack_as_the_bot_with_slackChannelID(self):
+		"""With slackChannelID and the bot token, scheduled reports reach Slack as the bot in place of the incoming
+		webhook, which it does not need; without the token, the webhook carries them."""
+		hook = 'https://hooks.slack.com/services/x'
+		telegram = config.env.webhooks['telegram']
+		self.env(webhooks={'telegram': telegram, 'slack': hook}, telegramChatID='1', slackChannelID='C9',
+			slackBotToken='xoxb-test')
+		self.assertEqual(list(map(repr, delivery.default_chats())), ['<TelegramBot 1>', '<SlackBot C9>'])
+		self.env(slackBotToken='')
+		self.assertEqual(list(map(repr, delivery.default_chats())), ['<TelegramBot 1>', f'<SlackWebhook {hook}>'])
+		self.env(webhooks={'telegram': telegram}, slackBotToken='xoxb-test')
+		self.assertEqual(list(map(repr, delivery.default_chats())), ['<TelegramBot 1>', '<SlackBot C9>'])
+		self.env(webhooks={}, telegramChatID='') # the bot alone
+		self.assertEqual(list(map(repr, delivery.default_chats())), ['<SlackBot C9>'])
 
 	def test_slack_replies_upload_their_charts(self):
 		"""A Slack reply uploads its chart in three steps (bot token), the report as its message, with no text copy."""
@@ -705,8 +719,9 @@ def test_a_slack_upload_that_fails_goes_out_as_text(self):
 					delivery.deliver(transports.SlackBot('C1'), report)
 				self.assertEqual(self.sent, expect)
 
-	def test_scheduled_reports_to_slack_stay_text(self):
-		"""Scheduled reports reach Slack's incoming webhook as text and Telegram with their charts."""
+	def test_scheduled_reports_to_slack_carry_charts_only_as_the_bot(self):
+		"""Scheduled reports reach Slack's incoming webhook as text; the Slack bot (slackChannelID) and Telegram get
+		their charts."""
 		self.env(slackBotToken='xoxb-test')
 
 		self.env(webhooks={**config.env.webhooks, 'slack': 'https://hooks.slack.com/services/x'}) # an incoming webhook cannot upload
@@ -720,6 +735,11 @@ def test_scheduled_reports_to_slack_stay_text(self):
 				sent.clear()
 				report()
 				self.assertEqual(sorted(set(sent)), [('slack', 'text'), ('telegram', 'chart')])
+				self.env(slackChannelID='C9') # as the bot, the report's chart goes too
+				sent.clear()
+				report()
+				self.assertEqual(sorted(set(sent)), [('slack', 'chart'), ('telegram', 'chart')])
+				self.env(slackChannelID='')
 
 	def test_slack_without_a_bot_token_gets_text_and_no_chart_is_drawn(self):
 		"""Without the bot token, Slack replies are text and no chart is drawn for them."""

From ec5099bdb9c940c5ddc1afeeaee6818542ca6fba Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 03:45:41 +0000
Subject: [PATCH 46/81] =?UTF-8?q?Slack:=20=E2=8F=B3=20on=20a=20command's?=
 =?UTF-8?q?=20message=20while=20finbot=20works=20on=20it=20(M8)?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Slack has no typing notice for bots, so the bot puts the hourglass_flowing_sand reaction on the message it is
answering, once the command has taken half a second, as Telegram gets "typing…", and takes it off when the answer is
ready. The message's ts comes from the event (inbound.Message.message_id) through the worker to its transport
(delivery.reply_to); TypingIndicator.stop() ends a notice that lasts (Transport.typing_done). The bot needs the
reactions:write scope; a refused reaction is logged and not tried again.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 README.md               |  2 +-
 bot.py                  |  2 +-
 doc/slack.md            |  8 +++++---
 lib/delivery.py         |  7 ++++---
 lib/inbound.py          |  5 +++--
 lib/transports.py       | 37 +++++++++++++++++++++++++++++++++++--
 lib/worker.py           | 18 ++++++++++--------
 tests/test_edges.py     |  5 +++++
 tests/test_providers.py | 30 ++++++++++++++++++++++++++++++
 9 files changed, 94 insertions(+), 20 deletions(-)

diff --git a/README.md b/README.md
index 054866d5..ddb4cf6f 100644
--- a/README.md
+++ b/README.md
@@ -28,7 +28,7 @@ What finbot can do on each network. Telegram takes everything; the others take w
 | Feature | Telegram | Slack | Discord |
 |---|---|---|---|
 | Chat commands | ✓ | ✓ | ✗ (a webhook only posts) |
-| "typing…" while a command works | ✓ | ✗ (Slack has no typing notice for bots) | ✗ |
+| "typing…" while a command works | ✓ | ⏳ on the command's message (Slack has no typing notice for bots) | ✗ |
 | Charts in replies | ✓ | ✓ with `slackBotToken` | n/a |
 | Charts in scheduled reports | ✓ | ✓ with `slackChannelID` and `slackBotToken` (the incoming webhook takes text only) | ✓ |
 | Period and job buttons | ✓ | ✗ | ✗ |
diff --git a/bot.py b/bot.py
index 6386e013..c4e81213 100755
--- a/bot.py
+++ b/bot.py
@@ -119,7 +119,7 @@ def dispatch(event):
 		if config.prefs.debug:
 			print(f"[{event.service.capitalize()}]:", event.user, event.text, file=sys.stderr)
 		target = worker.process_request
-		args = (event.service, event.chat_id, event.user, event.text, event.bot_name, event.user_name)
+		args = (event.service, event.chat_id, event.user, event.text, event.bot_name, event.user_name, event.message_id)
 	threading.Thread(target=target, args=args).start()
 
 def git_version(checkout=None):
diff --git a/doc/slack.md b/doc/slack.md
index c23852d4..aba118c2 100644
--- a/doc/slack.md
+++ b/doc/slack.md
@@ -11,7 +11,8 @@ the bot posts them itself, charts and all (`slackChannelID`).
 | Scheduled reports: trades, price alerts, earnings and dividend dates, performance and the rest | ✓, as text | ✓, as text through the webhook, or with their charts as the bot (`slackChannelID`) |
 | Chat commands: `.price`, `@finbot price` and the rest ([the list](../README.md#supported-commands)) | ✗ | ✓ |
 | Charts in the replies to commands | n/a | ✓, uploaded into the channel |
-| Period buttons and "typing…" | ✗ | ✗: Slack has no typing notice for bots, and finbot has no Slack buttons yet |
+| A sign that finbot is working on a command | ✗ | ✓: ⏳ on the command's message, Slack having no typing notice for bots |
+| Period buttons | ✗ | ✗: finbot has no Slack buttons yet |
 | Public HTTPS server | Not needed | Needed: a domain name with a valid certificate |
 | `finbot.env` keys | `slack_webhook` | That one, plus `slackOutgoingWebhook`, `slackSigningSecret` and `slackBotToken` (and `slackChannelID`, for charts in the scheduled reports) |
 
@@ -79,8 +80,9 @@ on a domain name with a valid certificate.
      * For direct messages to the bot, add `message.im`, and tick _App Home > Allow users to send Slash commands and
        messages from the messages tab_.
    * _Save Changes_.
-6. Under _OAuth & Permissions > Scopes > Bot Token Scopes_, add `chat:write`, `files:write` and `files:read`. The upload
-   of a chart needs both `files` scopes.
+6. Under _OAuth & Permissions > Scopes > Bot Token Scopes_, add `chat:write`, `files:write`, `files:read` and
+   `reactions:write`. The upload of a chart needs both `files` scopes, and the ⏳ on a command that takes a while
+   `reactions:write`.
 7. At the top of _OAuth & Permissions_, _Install to Workspace_ (or _Reinstall to Workspace_ when you changed scopes of
    an installed app), and copy the _Bot User OAuth Token_ into `finbot.env` as `slackBotToken`. finbot replies with it;
    without it, commands get no reply, and the log says so at startup and for each reply.
diff --git a/lib/delivery.py b/lib/delivery.py
index 20277606..89c4f568 100644
--- a/lib/delivery.py
+++ b/lib/delivery.py
@@ -24,10 +24,11 @@ class Report:
 REPLIES = {'telegram': transports.TelegramBot, 'slack': transports.SlackBot} # a reply into the chat that asked
 
 
-def reply_to(service, chat):
-	"""The destination of a reply into a chat, or None for a network that takes none."""
+def reply_to(service, chat, message=None):
+	"""The destination of a reply into a chat (to `message`, the one being answered, if given), or None for a network
+	that takes none."""
 	kind = REPLIES.get(service)
-	return kind(chat) if kind else None
+	return kind(chat, message) if kind else None
 
 
 def scheduled(service, url):
diff --git a/lib/inbound.py b/lib/inbound.py
index 0e063d98..b028c1a4 100644
--- a/lib/inbound.py
+++ b/lib/inbound.py
@@ -23,6 +23,7 @@ class Message:
 	text: str
 	bot_name: str # the bot's @name, which can start a command
 	user_name: str # what to call them
+	message_id: str = '' # the message itself: Slack shows ⏳ on it while finbot works
 
 
 @dataclasses.dataclass
@@ -147,7 +148,7 @@ def parse(self, update):
 		else:
 			print("[telegram]: unhandled: 'message' without 'text/photo'", file=sys.stderr)
 			return UNHANDLED
-		return Message(self.service, chat_id, user, text, self.bot_name, user_name)
+		return Message(self.service, chat_id, user, text, self.bot_name, user_name, str(message.get('message_id', '')))
 
 
 class Slack(Webhook):
@@ -210,7 +211,7 @@ def parse(self, update):
 		text = re.sub(r'<http://.*\|([\w\.]+)>', r'\g<1>', event['text'])
 		text = re.sub(r'<(@[\w\.]+)>', r'\g<1>', text)
 		user = '<@' + event['user'] + '>' # a mention: Slack shows the name
-		return Message(self.service, event['channel'], user, text, '@' + bot_user, user)
+		return Message(self.service, event['channel'], user, text, '@' + bot_user, user, event.get('ts', ''))
 
 
 TELEGRAM, SLACK = Telegram(), Slack()
diff --git a/lib/transports.py b/lib/transports.py
index 0c1d3274..dff1f1c2 100644
--- a/lib/transports.py
+++ b/lib/transports.py
@@ -26,9 +26,10 @@ class Transport:
 	takes_buttons = False # buttons under a message
 	shows_typing = False # a typing notice while a command works
 
-	def __init__(self, where):
-		"""where: the chat id, or the incoming webhook's URL."""
+	def __init__(self, where, message=None):
+		"""where: the chat id, or the incoming webhook's URL; message: the id of the message being answered, if any."""
 		self.where = where
+		self.message = message
 
 	def __repr__(self):
 		"""<TelegramBot 55>: the kind and where, for logs and test failures."""
@@ -68,6 +69,9 @@ def post_chart(self, image, caption, buttons=None):
 	def typing(self):
 		"""Show the typing notice, where the network has one."""
 
+	def typing_done(self):
+		"""End the typing notice, where it lasts until ended (Slack's ⏳)."""
+
 	def _post_json(self, url, payload, headers=None):
 		"""POST a JSON message: the answer ({} when it is not JSON), or None when the network refused it or failed."""
 		try:
@@ -186,12 +190,41 @@ class SlackBot(Transport):
 	"""A Slack channel, as the bot: chat.postMessage and the file upload, with slackBotToken."""
 	name = 'slack'
 	dialect = dialects.SLACK
+	WORKING = 'hourglass_flowing_sand' # ⏳: Slack has no typing notice for bots, so a reaction on the message stands in
+	working = None # True while ⏳ is on the message, False when putting it there failed
 
 	@property
 	def takes_charts(self):
 		"""Charts are uploaded with the bot token, so only with one."""
 		return bool(config.env.slackBotToken)
 
+	@property
+	def shows_typing(self):
+		"""⏳ on the message being answered, with the bot token."""
+		return bool(self.message and config.env.slackBotToken)
+
+	def typing(self):
+		"""Put ⏳ on the message being answered, once (the bot needs the reactions:write scope); a failure is only
+		logged, and not tried again."""
+		if self.working is not None:
+			return
+		try:
+			slack_api('reactions.add', channel=self.where, timestamp=self.message, name=self.WORKING)
+			self.working = True
+		except Exception as e:
+			self.working = False
+			print("Slack reactions.add failed:", net.redact(e), file=sys.stderr)
+
+	def typing_done(self):
+		"""Take ⏳ off the message again; a failure is only logged."""
+		if not self.working:
+			return
+		self.working = False
+		try:
+			slack_api('reactions.remove', channel=self.where, timestamp=self.message, name=self.WORKING)
+		except Exception as e:
+			print("Slack reactions.remove failed:", net.redact(e), file=sys.stderr)
+
 	def post_text(self, text, buttons=None):
 		"""chat.postMessage, without link previews (Slack reads those flags from the body)."""
 		if not config.env.slackBotToken:
diff --git a/lib/worker.py b/lib/worker.py
index 201f1f88..4aad7274 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -27,9 +27,9 @@
 from lib import jobs
 
 class TypingIndicator:
-	"""The typing notice while a command works, where the destination shows one (Telegram). It starts after `delay`
-	seconds, so a quick answer gets none, and is sent again every 7 seconds until stopped, at most max_loops times. The
-	worker stops it before the answer goes out, however the command ends."""
+	"""The typing notice while a command works, where the destination shows one (Telegram; Slack's ⏳). It starts after
+	`delay` seconds, so a quick answer gets none, and is sent again every 7 seconds until stopped, at most max_loops
+	times. The worker stops it before the answer goes out, however the command ends."""
 	def __init__(self, dest, max_loops=10, delay=0.5):
 		"""dest: the chat's transport."""
 		self.dest = dest
@@ -54,11 +54,12 @@ def start(self):
 			self._thread.start()
 
 	def stop(self):
-		"""Stop the notices and wait for their thread to end."""
+		"""Stop the notices and wait for their thread to end, then end the notice where it lasts (Slack's ⏳)."""
 		if not self._thread: # never started
 			return
 		self._stop_event.set()
 		self._thread.join()
+		self.dest.typing_done()
 
 	def is_active(self):
 		"""Whether the notices are still going."""
@@ -72,10 +73,11 @@ class Request:
 	user: str # the sender's handle: whose watchlist add or plan it is
 	user_name: str # what to call them
 	bot_name: str
+	message_id: str = None # the message being answered
 
 	def __post_init__(self):
 		"""The chat's transport, and its typing notice (not started)."""
-		self.dest = delivery.reply_to(self.service, self.chat_id)
+		self.dest = delivery.reply_to(self.service, self.chat_id, self.message_id)
 		self.typing = TypingIndicator(self.dest)
 
 	def reply(self, lines, chart=None, buttons=None):
@@ -124,11 +126,11 @@ def process_callback(service, callback_id, chat_id, message_id, data):
 	except Exception as e:
 		delivery.report_error(e, service, chat_id, context='chart')
 
-def process_request(service, chat_id, user, message, botName, userRealName):
+def process_request(service, chat_id, user, message, botName, userRealName, message_id=None):
 	"""Entry point for inbound chat requests: the first command in COMMANDS that fits the message answers it (a message
 	that fits none gets no answer), with the typing notice meanwhile. Any failure becomes a one-line reply in the
-	originating chat, after the typing stops."""
-	request = Request(dialects.of(service), chat_id, user, userRealName, botName)
+	originating chat, after the typing stops. message_id: the message's own id, where the network has one."""
+	request = Request(dialects.of(service), chat_id, user, userRealName, botName, message_id)
 	try:
 		for command in COMMANDS:
 			if m := match(command.pattern, message, botName):
diff --git a/tests/test_edges.py b/tests/test_edges.py
index c8869982..722dfa68 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -349,6 +349,11 @@ def event(self, event_id=None, **event):
 		gevent.sleep(0.05) # the worker runs in a (gevent) thread
 		return status[-1]
 
+	def test_the_message_goes_to_the_worker_with_its_ts(self):
+		"""A message reaches the worker with its ts, the message's id: finbot puts ⏳ on it while it works."""
+		self.assertEqual(self.event(user='U1', text='.price'), '200 OK')
+		self.assertEqual(self.requests[0][6], '1.2')
+
 	def test_only_people_are_answered(self):
 		"""Messages from bots, this bot included, joins, edits and events without a user are not answered; people's
 		messages are, with mentions turned into @names."""
diff --git a/tests/test_providers.py b/tests/test_providers.py
index ed36204b..3610dbb3 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -880,6 +880,36 @@ def test_a_reply_without_a_bot_token_is_logged_not_sent(self):
 		self.assertEqual((self.posts, self.errors), ([], []))
 		self.assertEqual(err.getvalue().count('ERROR slack: a reply to channel C1 needs slackBotToken in finbot.env; not sent'), 3)
 
+	def test_a_command_at_work_shows_an_hourglass_on_its_message(self):
+		"""While a command works, ⏳ is on the message it answers (Slack has no typing notice for bots), and it comes off
+		when the answer is ready; only with that message and the bot token."""
+		self.env(slackBotToken='xoxb-test')
+		indicator = worker.TypingIndicator(transports.SlackBot('C1', '1.2'), delay=0)
+		indicator.start()
+		deadline = time.monotonic() + 5
+		while not self.posts and time.monotonic() < deadline:
+			time.sleep(0.01)
+		indicator.stop()
+		reaction = {'channel': 'C1', 'timestamp': '1.2', 'name': 'hourglass_flowing_sand'}
+		self.assertEqual([(url, kwargs['data']) for url, kwargs in self.posts],
+			[('https://slack.com/api/reactions.add', reaction), ('https://slack.com/api/reactions.remove', reaction)])
+		self.assertFalse(transports.SlackBot('C1').shows_typing) # no message to put it on
+		self.env(slackBotToken='')
+		self.assertFalse(transports.SlackBot('C1', '1.2').shows_typing)
+
+	def test_an_hourglass_slack_refuses_is_tried_once(self):
+		"""⏳ that Slack refuses (a missing scope) is logged and not tried again, and nothing is taken off."""
+		self.env(slackBotToken='xoxb-test')
+		refuse = Resp({'ok': False, 'error': 'missing_scope'})
+		self.patch(transports.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or refuse)
+		chat = transports.SlackBot('C1', '1.2')
+		with contextlib.redirect_stderr(io.StringIO()) as err:
+			chat.typing()
+			chat.typing()
+			chat.typing_done()
+		self.assertEqual([url for url, _ in self.posts], ['https://slack.com/api/reactions.add'])
+		self.assertIn('Slack reactions.add failed: Slack reactions.add: missing_scope', err.getvalue())
+
 	def test_the_slack_api_says_what_is_missing(self):
 		"""slack_api without the bot token says so (it used to be a TypeError)."""
 		self.env(slackBotToken=False)

From 6ebad5b630143593c5880d0117bf198d291c1e1e Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 03:52:33 +0000
Subject: [PATCH 47/81] Slack: period buttons under a chart, redrawn in place
 (M8, D9)

Where a press can reach finbot (slackBotToken and slackOutgoingWebhook), a chart goes to Slack as a Block Kit message:
the report in sections, the chart uploaded into no channel and shown by its id in an image block, and the period
buttons, the one shown marked. Slack posts a press to the app's Interactivity Request URL, the same as its events URL:
a form whose payload field holds the JSON, signed like an event. inbound.Slack reads it as a Press, and the worker
redraws the chart for that period and puts it in the message with chat.update (Slack cannot replace a file, so each
press uploads one). Slack may refuse to show a file just uploaded (invalid_blocks) for a moment, so the message is
asked for again after 1, 2 and 4 seconds; failing that, the chart is uploaded into the channel without its buttons.
A press is answered through its transport (Transport.acknowledge: Telegram's toast, nothing on Slack). The Slack
guide gains the Interactivity step.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 README.md               |  4 +-
 doc/slack.md            | 24 +++++++----
 lib/inbound.py          | 32 +++++++++++---
 lib/transports.py       | 93 +++++++++++++++++++++++++++++++++++++---
 lib/worker.py           | 15 ++++---
 tests/test_edges.py     | 42 ++++++++++++++++++
 tests/test_providers.py | 95 +++++++++++++++++++++++++++++++++++++++++
 7 files changed, 276 insertions(+), 29 deletions(-)

diff --git a/README.md b/README.md
index ddb4cf6f..5dc53c99 100644
--- a/README.md
+++ b/README.md
@@ -31,10 +31,10 @@ What finbot can do on each network. Telegram takes everything; the others take w
 | "typing…" while a command works | ✓ | ⏳ on the command's message (Slack has no typing notice for bots) | ✗ |
 | Charts in replies | ✓ | ✓ with `slackBotToken` | n/a |
 | Charts in scheduled reports | ✓ | ✓ with `slackChannelID` and `slackBotToken` (the incoming webhook takes text only) | ✓ |
-| Period and job buttons | ✓ | ✗ | ✗ |
+| Period and job buttons | ✓ | Period buttons, with the interactive bot | ✗ |
 | No link previews | ✓ | ✓ | ✓ |
 | Linked tickers | ✓ | ✓ | ✓ |
-| Charts updated in place by their buttons | ✓ | ✗ | ✗ |
+| Charts updated in place by their buttons | ✓ | ✓ | ✗ |
 
 **Setting up a network:** [Telegram](doc/telegram.md) · [Slack](doc/slack.md) · [Discord](doc/discord.md). Each guide
 covers the whole setup of that network, and where there is a choice between reports only and the interactive bot, it
diff --git a/doc/slack.md b/doc/slack.md
index aba118c2..afee3300 100644
--- a/doc/slack.md
+++ b/doc/slack.md
@@ -12,7 +12,7 @@ the bot posts them itself, charts and all (`slackChannelID`).
 | Chat commands: `.price`, `@finbot price` and the rest ([the list](../README.md#supported-commands)) | ✗ | ✓ |
 | Charts in the replies to commands | n/a | ✓, uploaded into the channel |
 | A sign that finbot is working on a command | ✗ | ✓: ⏳ on the command's message, Slack having no typing notice for bots |
-| Period buttons | ✗ | ✗: finbot has no Slack buttons yet |
+| Period buttons under a chart, which redraw it for that period in place | ✗ | ✓ |
 | Public HTTPS server | Not needed | Needed: a domain name with a valid certificate |
 | `finbot.env` keys | `slack_webhook` | That one, plus `slackOutgoingWebhook`, `slackSigningSecret` and `slackBotToken` (and `slackChannelID`, for charts in the scheduled reports) |
 
@@ -80,15 +80,18 @@ on a domain name with a valid certificate.
      * For direct messages to the bot, add `message.im`, and tick _App Home > Allow users to send Slash commands and
        messages from the messages tab_.
    * _Save Changes_.
-6. Under _OAuth & Permissions > Scopes > Bot Token Scopes_, add `chat:write`, `files:write`, `files:read` and
+6. Under _Interactivity & Shortcuts_, turn _Interactivity_ on, enter the same URL as the _Request URL_, and _Save
+   Changes_. Slack posts there the presses of the period buttons under a chart; without it, Slack answers a press with
+   an error.
+7. Under _OAuth & Permissions > Scopes > Bot Token Scopes_, add `chat:write`, `files:write`, `files:read` and
    `reactions:write`. The upload of a chart needs both `files` scopes, and the ⏳ on a command that takes a while
    `reactions:write`.
-7. At the top of _OAuth & Permissions_, _Install to Workspace_ (or _Reinstall to Workspace_ when you changed scopes of
+8. At the top of _OAuth & Permissions_, _Install to Workspace_ (or _Reinstall to Workspace_ when you changed scopes of
    an installed app), and copy the _Bot User OAuth Token_ into `finbot.env` as `slackBotToken`. finbot replies with it;
    without it, commands get no reply, and the log says so at startup and for each reply.
-8. Restart finbot, and invite the bot to the channel: `/invite @finbot`.
-9. Send `@finbot help` in the channel (or `.help`, with `message.channels`).
-10. For charts in the scheduled reports, set `slackChannelID` to the channel's id (in Slack, click the channel's name:
+9. Restart finbot, and invite the bot to the channel: `/invite @finbot`.
+10. Send `@finbot help` in the channel (or `.help`, with `message.channels`).
+11. For charts in the scheduled reports, set `slackChannelID` to the channel's id (in Slack, click the channel's name:
     the _Channel ID_, which starts with C, is at the bottom of _About_), and restart finbot. The bot then posts the
     scheduled reports there, as it posts its replies, and `slack_webhook` is no longer used.
 
@@ -100,6 +103,9 @@ slackBotToken=xoxb-your-bot-token
 slackChannelID=C0123456789
 ```
 
-A chart in a reply is uploaded into the channel with Slack's file upload API (`files.getUploadURLExternal`, then
-`files.completeUploadExternal`). If an upload fails (a missing scope, say), the reply goes out as text, and the log says
-why.
+A chart is uploaded with Slack's file upload API (`files.getUploadURLExternal`, then `files.completeUploadExternal`).
+With its period buttons, it goes in a message of blocks: the report, the chart (uploaded into no channel, and shown by
+its id) and the buttons. A press uploads the chart for that period and shows it in the message in place of the last
+one (Slack cannot replace a file; the earlier one stays among the bot's files, but in no channel). If Slack will not
+show a chart with its buttons, the chart is uploaded into the channel without them; if an upload fails (a missing scope,
+say), the report goes out as text. The log says why.
diff --git a/lib/inbound.py b/lib/inbound.py
index b028c1a4..3d88348d 100644
--- a/lib/inbound.py
+++ b/lib/inbound.py
@@ -9,7 +9,7 @@
 import re
 import sys
 import time
-from urllib.parse import urlparse
+from urllib.parse import parse_qs, urlparse
 
 from lib import config
 
@@ -152,7 +152,8 @@ def parse(self, update):
 
 
 class Slack(Webhook):
-	"""Slack's Events API: each request signed with the app's signing secret, and carrying its verification token."""
+	"""Slack's Events API, and its interactivity (a button pressed), at the same URL: each request signed with the app's
+	signing secret, and carrying its verification token."""
 	service = 'slack'
 	BY_PEOPLE = (None, 'thread_broadcast') # message subtypes a person writes; the rest are bot posts, edits, joins...
 
@@ -185,18 +186,26 @@ def signature_problem(environ, body, secret):
 		expected = 'v0=' + hmac.new(secret.encode(), signed, hashlib.sha256).hexdigest()
 		return None if same_secret(signature, expected) else "Slack signature incorrect"
 
+	def decode(self, body):
+		"""The update: an event comes as JSON, a button press as a form whose `payload` field holds the JSON."""
+		if body.startswith(b'payload='):
+			body = parse_qs(body.decode('utf-8', 'replace')).get('payload', [''])[0].encode()
+		return super().decode(body)
+
 	def update_id(self, update):
-		"""Slack's event_id."""
+		"""Slack's event_id (a button press has none)."""
 		return update.get('event_id')
 
 	def parse(self, update):
-		"""A message a person wrote, or the challenge that checks the URL. Links and mentions lose their markup:
-		<http://dub.ax|dub.ax> becomes dub.ax and <@QWERTY> @QWERTY."""
+		"""A message a person wrote, a button pressed, or the challenge that checks the URL. Links and mentions lose
+		their markup: <http://dub.ax|dub.ax> becomes dub.ax and <@QWERTY> @QWERTY."""
 		kind = update.get('type')
 		if kind == 'url_verification':
 			response = json.dumps({'challenge': update['challenge']})
 			print("replying with", response, file=sys.stderr)
 			return Response(body=response.encode())
+		if kind == 'block_actions':
+			return self.press(update)
 		if kind != 'event_callback':
 			print(f"[slack]: unhandled 'type' {kind!r}", file=sys.stderr)
 			return UNHANDLED
@@ -213,6 +222,19 @@ def parse(self, update):
 		user = '<@' + event['user'] + '>' # a mention: Slack shows the name
 		return Message(self.service, event['channel'], user, text, '@' + bot_user, user, event.get('ts', ''))
 
+	def press(self, update):
+		"""A button pressed under one of the bot's messages (a chart's period buttons), or UNSUPPORTED for one that is
+		not under a message or carries no value."""
+		container = update.get('container') or {}
+		actions = update.get('actions') or [{}]
+		value = actions[0].get('value') if isinstance(actions[0], dict) else None
+		if container.get('type') != 'message' or not container.get('message_ts') or not value:
+			return UNSUPPORTED
+		user = update.get('user') or {}
+		who = '@' + str(user.get('username') or user.get('id') or 'unknown')
+		return Press(self.service, update.get('trigger_id', ''), container.get('channel_id', ''),
+			container['message_ts'], value, who)
+
 
 TELEGRAM, SLACK = Telegram(), Slack()
 
diff --git a/lib/transports.py b/lib/transports.py
index dff1f1c2..e9c49e6c 100644
--- a/lib/transports.py
+++ b/lib/transports.py
@@ -72,6 +72,9 @@ def typing(self):
 	def typing_done(self):
 		"""End the typing notice, where it lasts until ended (Slack's ⏳)."""
 
+	def acknowledge(self, press, text=None):
+		"""Answer a button press, where the network waits for that (Telegram, with `text` as a toast)."""
+
 	def _post_json(self, url, payload, headers=None):
 		"""POST a JSON message: the answer ({} when it is not JSON), or None when the network refused it or failed."""
 		try:
@@ -174,6 +177,10 @@ def typing(self):
 		"""sendChatAction: "typing…" for a few seconds."""
 		telegram.try_call('sendChatAction', data={'chat_id': self.where, 'action': 'typing'})
 
+	def acknowledge(self, press, text=None):
+		"""answerCallbackQuery: the client's spinner stops, and `text` shows as a toast."""
+		telegram.answerCallbackQuery(press, text)
+
 	def pin(self, answers):
 		"""Pin the messages these answers are for, last first, so the first ends up on top, without the notification."""
 		sent = []
@@ -203,6 +210,12 @@ def shows_typing(self):
 		"""⏳ on the message being answered, with the bot token."""
 		return bool(self.message and config.env.slackBotToken)
 
+	@property
+	def takes_buttons(self):
+		"""Buttons only where their presses reach finbot: with the bot token and slackOutgoingWebhook, the URL that the
+		app's interactivity posts to as its events do."""
+		return bool(config.env.slackBotToken and config.env.slackOutgoingWebhook)
+
 	def typing(self):
 		"""Put ⏳ on the message being answered, once (the bot needs the reactions:write scope); a failure is only
 		logged, and not tried again."""
@@ -238,14 +251,17 @@ def post_text(self, text, buttons=None):
 	def post_chart(self, image, caption, buttons=None):
 		"""Upload the chart into the channel with the report as its message, the way Slack takes files since it retired
 		files.upload: ask for an upload URL, send the bytes there, then complete the upload into the channel (the bot
-		needs the files:write and files:read scopes). Slack's answer, MAYBE_SENT, or None if a step failed."""
+		needs the files:write and files:read scopes). With period buttons, where they work, it goes in a Block Kit
+		message instead (post_blocks), or without them if that fails. Slack's answer, MAYBE_SENT, or None if a step
+		failed."""
 		data = image.read() if hasattr(image, 'read') else bytes(image)
+		if buttons and not isinstance(buttons, dict) and self.takes_buttons:
+			try:
+				return self.post_blocks(data, caption, buttons)
+			except Exception as e:
+				print("Slack chart with buttons failed, uploading it without them:", net.redact(e), file=sys.stderr)
 		try:
-			ticket = slack_api('files.getUploadURLExternal', filename='chart.png', length=len(data))
-			r = net.post(ticket['upload_url'], data=data, timeout=config.prefs.http_timeout)
-			if r.status_code != 200:
-				raise RuntimeError(f"Slack upload: {r.status_code} {r.text[:200]}")
-			share = {'files': json.dumps([{'id': ticket['file_id']}]), 'channel_id': self.where,
+			share = {'files': json.dumps([{'id': self.upload(data)}]), 'channel_id': self.where,
 				'initial_comment': caption}
 			try:
 				output = slack_api('files.completeUploadExternal', **share)
@@ -259,6 +275,39 @@ def post_chart(self, image, caption, buttons=None):
 		print("OK slack chart upload to", self.where, file=sys.stderr)
 		return output
 
+	def upload(self, data):
+		"""Send a chart's bytes to Slack: the id of the file, not completed yet. Raises RuntimeError on a refusal."""
+		ticket = slack_api('files.getUploadURLExternal', filename='chart.png', length=len(data))
+		r = net.post(ticket['upload_url'], data=data, timeout=config.prefs.http_timeout)
+		if r.status_code != 200:
+			raise RuntimeError(f"Slack upload: {r.status_code} {r.text[:200]}")
+		return ticket['file_id']
+
+	def unshared(self, data):
+		"""Upload a chart as a file that is in no channel, for an image block to show: its id."""
+		file_id = self.upload(data)
+		slack_api('files.completeUploadExternal', files=json.dumps([{'id': file_id, 'title': 'chart'}]))
+		return file_id
+
+	def post_blocks(self, data, caption, buttons):
+		"""The chart in a Block Kit message, the report above it and the period buttons under it. Slack's answer;
+		raises RuntimeError when Slack refuses."""
+		blocks = chart_blocks(caption, buttons, self.unshared(data))
+		return until_shown(lambda: slack_api('chat.postMessage', channel=self.where, text=caption[:SECTION_LIMIT],
+			blocks=json.dumps(blocks), unfurl_links='false', unfurl_media='false'))
+
+	def edit_chart(self, message, image, caption, buttons=None):
+		"""Show another period in a chart message: the new chart uploaded (a file cannot be replaced) and shown in its
+		image block, with its caption and buttons. Raises RuntimeError when Slack refuses."""
+		file_id = None
+		if image is not None:
+			if hasattr(image, 'seek'):
+				image.seek(0)
+			file_id = self.unshared(image.read() if hasattr(image, 'read') else bytes(image))
+		blocks = chart_blocks(caption, buttons or [], file_id)
+		until_shown(lambda: slack_api('chat.update', channel=self.where, ts=message, text=caption[:SECTION_LIMIT],
+			blocks=json.dumps(blocks)))
+
 
 class SlackWebhook(Transport):
 	"""A Slack incoming webhook: text only, into the channel it was made for."""
@@ -322,6 +371,38 @@ def slack_api(method, **data):
 	return output
 
 
+SECTION_LIMIT = 3000 # characters of text in a Block Kit section
+
+
+def chart_blocks(caption, buttons, file_id=None):
+	"""Block Kit for a chart message: the report (in as many sections as its length needs), the chart (an uploaded file,
+	by its id) and the buttons, the one shown now marked."""
+	blocks = [{'type': 'section', 'text': {'type': 'mrkdwn', 'text': '\n'.join(part)}}
+		for part in chunked(caption.split('\n'), SECTION_LIMIT, 1000) if '\n'.join(part).strip()]
+	if file_id:
+		blocks.append({'type': 'image', 'slack_file': {'id': file_id}, 'alt_text': 'chart'})
+	elements = [{'type': 'button', 'text': {'type': 'plain_text', 'text': button.label}, 'value': button.data,
+		'action_id': button.data, **({'style': 'primary'} if button.active else {})}
+		for row in buttons for button in row]
+	if elements:
+		blocks.append({'type': 'actions', 'elements': elements})
+	return blocks
+
+
+def until_shown(post, waits=(1, 2, 4)):
+	"""post() of a message that shows a file just uploaded, tried again after each wait (in seconds) while Slack says
+	invalid_blocks: a file can take a moment before it can be shown."""
+	for wait in waits:
+		try:
+			return post()
+		except RuntimeError as e:
+			if 'invalid_blocks' not in str(e):
+				raise
+			print(f"Slack cannot show the chart yet ({e}), trying again in {wait} s", file=sys.stderr)
+			time.sleep(wait)
+	return post()
+
+
 def chunked(lines, limit, max_lines):
 	"""The lines in messages of at most max_lines lines and limit characters each, whichever comes first; a line longer
 	than limit on its own is cut short."""
diff --git a/lib/worker.py b/lib/worker.py
index 4aad7274..8ff77560 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -88,18 +88,19 @@ def reply(self, lines, chart=None, buttons=None):
 
 def process_job_callback(service, callback_id, chat_id, message_id, data):
 	"""A command button of .jobs: run that command now and answer in the chat."""
+	chat = delivery.reply_to(service, chat_id)
 	try:
 		_, number, crc = data.split('|') # only a command offered by .jobs can be picked; anything else is refused
 		command = jobs.current.pick(number, crc)
 	except (ValueError, jobs.ScheduleError, AttributeError):
-		telegram.answerCallbackQuery(callback_id, 'Buttons updated, try again')
+		chat.acknowledge(callback_id, 'Buttons updated, try again')
 		payload, markup = about.prepare_jobs_payload(service, '') # the list as it is now, in place of the stale one
-		delivery.reply_to(service, chat_id).edit_text(message_id, '\n'.join(payload), markup or [])
+		chat.edit_text(message_id, '\n'.join(payload), markup or [])
 		return
-	telegram.answerCallbackQuery(callback_id, f'Running {jobs.display(command[0])}…'[:100])
+	chat.acknowledge(callback_id, f'Running {jobs.display(command[0])}…'[:100])
 	try:
 		line = about.job_line(jobs.current, *jobs.current.run_command(command))
-		delivery.deliver(delivery.reply_to(service, chat_id), delivery.Report([line]))
+		delivery.deliver(chat, delivery.Report([line]))
 	except Exception as e:
 		delivery.report_error(e, service, chat_id, context='jobs')
 
@@ -110,19 +111,19 @@ def process_callback(service, callback_id, chat_id, message_id, data):
 		return process_job_callback(service, callback_id, chat_id, message_id, data)
 	parts = data.split('|', 3)
 	valid = len(parts) == 4 and parts[0] == 'c' and parts[1] in buttons.BUTTONS and parts[2] in buttons.BUTTONS[parts[1]]
+	chat = delivery.reply_to(service, chat_id)
 	# answering now shows Telegram's toast bubble ("Loading 3M…") while the new chart is built
-	telegram.answerCallbackQuery(callback_id, f"Loading {buttons.PERIODS[parts[2]][0]}…" if valid else None)
+	chat.acknowledge(callback_id, f"Loading {buttons.PERIODS[parts[2]][0]}…" if valid else None)
 	if not valid:
 		return
 	try:
 		_, kind, period, ref = parts
 		args = buttons.resolve(ref)
 		caption, chart = charts.build(kind, args, period, service)
-		chat = delivery.reply_to(service, chat_id)
 		image = render.draw(chart) if chart else None
 		chat.edit_chart(message_id, image, caption, buttons.keyboard(kind, args, period))
 	except answer.Answer as e: # nothing to draw for that period: said in the chat, as a command would
-		delivery.deliver(delivery.reply_to(service, chat_id), delivery.Report(e.lines))
+		delivery.deliver(chat, delivery.Report(e.lines))
 	except Exception as e:
 		delivery.report_error(e, service, chat_id, context='chart')
 
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 722dfa68..0add3ee4 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -1,9 +1,14 @@
 """Failure and no-data scenarios: empty or broken upstream data, bad requests, rejected callers."""
 import contextlib, hashlib, hmac, io, json, os, signal, subprocess, sys, tempfile, threading, time, unittest
 from pathlib import Path
+from urllib.parse import urlencode
 import gevent
 from tests.common import *
 
+PRESS = {'type': 'block_actions', 'token': 'stok', 'trigger_id': 't1', 'user': {'id': 'U1', 'username': 'rob'},
+	'container': {'type': 'message', 'message_ts': '1.5', 'channel_id': 'C1'},
+	'actions': [{'action_id': 'c|h|w|T1', 'value': 'c|h|w|T1', 'type': 'button'}]} # a Slack period button pressed
+
 COMMANDS = ('.price', '.price 7d', '.price top', '.session', '.premarket', '.performance 7d', '.shorts', '.peg', '.buy',
 			'.sell', '.beta', '.pe', '.marketcap', '.earnings', '.dividend', '.trades', '.holdings', '.watchlist',
 			'.T1', '.history T1', '.compare T1 T2')
@@ -349,6 +354,27 @@ def event(self, event_id=None, **event):
 		gevent.sleep(0.05) # the worker runs in a (gevent) thread
 		return status[-1]
 
+	def press(self, payload):
+		"""Post a button press as Slack's interactivity does (a form, the JSON in its payload field): the status."""
+		status = []
+		with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
+			body = urlencode({'payload': json.dumps(payload)}).encode()
+			bot.main(wsgi_environ('/slack', body), lambda s, *a, **k: status.append(s))
+		gevent.sleep(0.05) # the worker runs in a (gevent) thread
+		return status[-1]
+
+	def test_a_button_press_reaches_the_worker(self):
+		"""A period button pressed under a chart reaches the worker, posted to the events URL: the channel, the message
+		and the button's value. A press with a wrong token is refused, and one not under a message goes nowhere."""
+		presses = []
+		self.patch(worker, 'process_callback', lambda *a: presses.append(a))
+		self.assertEqual(self.press(PRESS), '200 OK')
+		self.assertEqual(presses, [('slack', 't1', 'C1', '1.5', 'c|h|w|T1')])
+		self.assertEqual(self.press({**PRESS, 'token': 'wrong'}), '403 Forbidden')
+		self.assertEqual(self.press({**PRESS, 'container': {'type': 'view', 'view_id': 'V1'}}), '200 OK')
+		self.assertEqual(self.press({**PRESS, 'actions': []}), '200 OK')
+		self.assertEqual(len(presses), 1)
+
 	def test_the_message_goes_to_the_worker_with_its_ts(self):
 		"""A message reaches the worker with its ts, the message's id: finbot puts ⏳ on it while it works."""
 		self.assertEqual(self.event(user='U1', text='.price'), '200 OK')
@@ -412,6 +438,22 @@ def post(self, secret='sss', age=0, token='stok', headers=None, tamper=False):
 		gevent.sleep(0.05) # the worker runs in a (gevent) thread
 		return status[-1]
 
+	def test_a_button_press_is_signed_over_its_form(self):
+		"""A button press is signed like an event, over the form it comes as; altered, it is refused."""
+		presses = []
+		self.patch(worker, 'process_callback', lambda *a: presses.append(a))
+		body = urlencode({'payload': json.dumps({**PRESS, 'token': 'not needed'})}).encode()
+		timestamp = str(int(time.time()))
+		mac = hmac.new(b'sss', b'v0:' + timestamp.encode() + b':' + body, hashlib.sha256).hexdigest()
+		headers = {'HTTP_X_SLACK_REQUEST_TIMESTAMP': timestamp, 'HTTP_X_SLACK_SIGNATURE': 'v0=' + mac}
+		for sent, status in ((body, '200 OK'), (body.replace(b'T1', b'T2'), '403 Forbidden')):
+			answers = []
+			with contextlib.redirect_stderr(io.StringIO()):
+				bot.main(wsgi_environ('/slack', sent, **headers), lambda s, *a, **k: answers.append(s))
+			gevent.sleep(0.05)
+			self.assertEqual(answers, [status])
+		self.assertEqual(presses, [('slack', 't1', 'C1', '1.5', 'c|h|w|T1')])
+
 	def test_a_signed_request_is_answered_and_the_token_is_not_needed(self):
 		"""A request signed with the secret is answered, whatever token it carries."""
 		self.assertEqual([self.post(), self.post(token='wrong')], ['200 OK'] * 2)
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 3610dbb3..1eb25d7e 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -919,6 +919,101 @@ def test_the_slack_api_says_what_is_missing(self):
 		self.assertEqual(self.posts, [])
 
 
+class SlackButtons(FinbotCase):
+	"""A chart's period buttons on Slack: a Block Kit message (the report, the chart uploaded into no channel, the
+	buttons) that a press redraws in place, where the presses reach finbot (the bot token and slackOutgoingWebhook)."""
+	UPLOAD_URL = 'https://files.slack.com/upload/v1/abc'
+
+	def setUp(self):
+		"""The real send path, its posts recorded and answered as Slack would; no waiting."""
+		super().setUp()
+		self.patch(transports.Transport, 'text', REAL_TEXT)
+		self.patch(transports.Transport, 'chart', REAL_CHART)
+		self.env(slackBotToken='xoxb-test', slackOutgoingWebhook='https://example.com/slack')
+		self.posts, self.refusals, self.waits = [], {}, []
+		self.patch(transports.time, 'sleep', self.waits.append)
+		def post(url, **kwargs):
+			"""Record the post and answer it: a refusal set for the method, while there is one left, else success."""
+			self.posts.append((url, kwargs))
+			method = url.rsplit('/', 1)[-1]
+			if self.refusals.get(method):
+				return Resp({'ok': False, 'error': self.refusals[method].pop(0)})
+			if url == self.UPLOAD_URL:
+				return Resp(None, text='OK - 3')
+			return Resp({'ok': True, 'upload_url': self.UPLOAD_URL, 'file_id': 'F1', 'files': [{'id': 'F1'}],
+				'ts': '1.5'})
+		self.patch(transports.net, 'post', post)
+
+	def calls(self):
+		"""The methods posted, in order (the upload as 'upload')."""
+		return ['upload' if url == self.UPLOAD_URL else url.rsplit('/', 1)[-1] for url, _ in self.posts]
+
+	def blocks(self, method):
+		"""The blocks of the last call of this method."""
+		return json.loads([kwargs for url, kwargs in self.posts if url.endswith('/' + method)][-1]['data']['blocks'])
+
+	def test_a_chart_reply_carries_its_period_buttons(self):
+		"""A chart reply is a Block Kit message: the report, the chart (uploaded into no channel, shown by its id) and
+		the period buttons, the one shown marked; nothing goes as text."""
+		worker.process_request('slack', 'C1', '<@U1>', '.price', BOT, 'U')
+		self.assertEqual(self.errors, [])
+		self.assertEqual(self.calls(), ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal',
+			'chat.postMessage'])
+		self.assertNotIn('channel_id', self.posts[2][1]['data']) # in no channel: the message shows it
+		blocks = self.blocks('chat.postMessage')
+		self.assertEqual([block['type'] for block in blocks], ['section', 'image', 'actions'])
+		self.assertTrue(blocks[0]['text']['text'].startswith('*5% at close'))
+		self.assertEqual(blocks[1]['slack_file'], {'id': 'F1'})
+		buttons_shown = blocks[2]['elements']
+		self.assertEqual([button['text']['text'] for button in buttons_shown], ['1D', '7D', '1M', '3M', '1Y', 'Max'])
+		self.assertEqual([button.get('style') for button in buttons_shown], ['primary'] + [None] * 5)
+		self.assertEqual(buttons_shown[1]['value'], 'c|l|w|5,0')
+
+	def test_a_press_redraws_the_chart_in_place(self):
+		"""A press uploads the chart for that period (a file cannot be replaced) and puts it in the message with
+		chat.update, the period pressed now marked; Telegram is not asked to answer the press."""
+		worker.process_callback('slack', 't1', 'C1', '1.5', 'c|l|w|5,0')
+		self.assertEqual(self.errors, [])
+		self.assertEqual(self.calls(), ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal',
+			'chat.update'])
+		update = self.posts[-1][1]['data']
+		self.assertEqual((update['channel'], update['ts']), ('C1', '1.5'))
+		blocks = self.blocks('chat.update')
+		self.assertEqual(blocks[1]['slack_file'], {'id': 'F1'})
+		self.assertEqual([button.get('style') for button in blocks[2]['elements']], [None, 'primary'] + [None] * 4)
+
+	def test_a_chart_not_ready_to_show_is_shown_a_moment_later(self):
+		"""Slack refusing the blocks of a chart just uploaded (invalid_blocks) is asked again after a wait."""
+		self.refusals['chat.postMessage'] = ['invalid_blocks']
+		worker.process_request('slack', 'C1', '<@U1>', '.price', BOT, 'U')
+		self.assertEqual((self.calls()[-2:], self.waits), (['chat.postMessage', 'chat.postMessage'], [1]))
+		self.assertEqual((self.sent, self.errors), ([], []))
+
+	def test_buttons_slack_will_not_show_leave_a_plain_chart(self):
+		"""When Slack will not show the chart with its buttons, the chart is uploaded into the channel without them."""
+		self.refusals['chat.postMessage'] = ['invalid_blocks'] * 4
+		with contextlib.redirect_stderr(io.StringIO()):
+			worker.process_request('slack', 'C1', '<@U1>', '.price', BOT, 'U')
+		self.assertEqual(self.calls()[-3:], ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal'])
+		self.assertEqual(self.posts[-1][1]['data']['channel_id'], 'C1')
+		self.assertEqual((self.waits, self.sent, self.errors), ([1, 2, 4], [], []))
+
+	def test_without_slackOutgoingWebhook_a_chart_has_no_buttons(self):
+		"""Without slackOutgoingWebhook, no press could reach finbot, so a chart goes without buttons."""
+		self.env(slackOutgoingWebhook='')
+		worker.process_request('slack', 'C1', '<@U1>', '.price', BOT, 'U')
+		self.assertEqual(self.calls(), ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal'])
+		self.assertEqual(self.posts[-1][1]['data']['channel_id'], 'C1')
+
+	def test_a_long_report_takes_several_sections(self):
+		"""A report longer than a section takes holds is split into sections, and an empty one takes none."""
+		caption = '\n'.join('x' * 99 for _ in range(70)) # 6,999 characters
+		blocks = transports.chart_blocks(caption, [], 'F1')
+		self.assertEqual([block['type'] for block in blocks], ['section'] * 3 + ['image'])
+		self.assertTrue(all(len(block['text']['text']) <= 3000 for block in blocks[:3]))
+		self.assertEqual(transports.chart_blocks('', [], None), [])
+
+
 class TelegramWebhook(FinbotCase):
 	"""bot.py registers the webhook once as it starts. Importing lib.telegram, as every job process does, touches nothing."""
 

From a94557e071a45ff5bfe531e9c9bd4f7fcd4cd964 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 03:53:59 +0000
Subject: [PATCH 48/81] CLAUDE.md: the code map after M6 and M8
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The signal work in a thread, Slack's ⏳, the scheduler's lock and per-command deadline, and tests/common.py's
wsgi_environ.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/CLAUDE.md b/CLAUDE.md
index 77c13b97..1f04eb19 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -32,11 +32,13 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 
 ## Code map (open the one file you need)
 - `bot.py`: WSGI entry (route, verify, parse, then a thread of `lib/worker.py` per message or button press); SIGHUP
-  reload, SIGUSR1 cache flush. `lib/inbound.py`: a class per network that posts to the bot (`Telegram`, `Slack`): its
-  path, how it proves itself, and its updates as `Message`, `Press` or a `Response` given straight back.
+  reload and SIGUSR1 cache flush, each in a thread. `lib/inbound.py`: a class per network that posts to the bot
+  (`Telegram`, `Slack`): its path, how it proves itself, and its updates as `Message`, `Press` or a `Response` given
+  straight back.
 - `lib/worker.py`: `COMMANDS`, a table of `Command(pattern, handler, usage, section, buttons)` (first match wins,
   `.SYMBOL` last), which also makes `.help` and the Telegram command keyboard, kept current in DMs; `Request`
-  (`request.reply(lines, chart, buttons)`); the typing notice, which every command gets after half a second.
+  (`request.reply(lines, chart, buttons)`); the typing notice (Slack: ⏳ on the message), which every command gets after
+  half a second.
 - `lib/reports/`: what finbot says. Scheduled (each has `run(dest=None, **kwargs)`, which posts to the chat that asked
   or else the default chats, and `job(argv)`, its schedule-line parser): `price performance trades shorts cal rating
   milestone reminder podcasts cleanup`; the top-level `price.py` and the rest are shims that run them. Chat-only
@@ -58,9 +60,10 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - `lib/cache.py`: cache files (any may vanish at any time). `lib/state.py`: state files in `DATA_DIR`, which must last;
   change one with `with state.edit(name, default) as data:`, never load-then-save. `lib/watchlist.py`: the watchlist.
 - `lib/config.py`: `config.env` (finbot.env, fixed at start) and `config.prefs` (preferences.conf, a new object on each reload).
-  `lib/jobs.py` + `lib/jobrun.py`: the scheduler; each run is a child process.
+  `lib/jobs.py` + `lib/jobrun.py`: the scheduler, its shared state under one lock; each command runs in a child
+  process, stopped after `JOB_TIMEOUT`.
 - `tests/common.py`: `FinbotCase` (offline fixtures). `self.command('.price')` returns what was sent: `('text', lines)` or
-  `('photo', size, caption)`.
+  `('photo', size, caption)`. `wsgi_environ(path, body, **headers)` is a request for `bot.main`.
 - Which test file: commands and scheduled reports `test_reports`; failures, bad input, WSGI and Slack events `test_edges`;
   Yahoo/Sharesight parsing, sending, Telegram API `test_providers`; scheduler `test_jobs`; config `test_config`; charts
   `test_charts` (refresh goldens with `UPDATE_GOLDEN=1`); markets, formats, links, cache and state files `test_util`; data dir

From bf0e2a655fc1177e43a5499932cdd29f55037eb5 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 03:55:18 +0000
Subject: [PATCH 49/81] worker: .compare and .history docstrings no longer say
 their buttons are Telegram's alone

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/worker.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/lib/worker.py b/lib/worker.py
index 8ff77560..3509ebd3 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -330,7 +330,7 @@ def on_rating(request, m):
 		request.reply(payload, chart)
 
 def on_compare(request, m):
-	""".compare SYMBOL SYMBOL... [period]: their price changes on one chart, with period buttons on Telegram."""
+	""".compare SYMBOL SYMBOL... [period]: their price changes on one chart, with period buttons where they work."""
 	payload, image, tickers, days = history.prepare_compare(request.service, (m.group('args') or '').split())
 	markup = None
 	if image and request.dest.takes_buttons:
@@ -338,7 +338,7 @@ def on_compare(request, m):
 	request.reply(payload, image, markup)
 
 def on_history(request, m):
-	""".history SYMBOL: the stock's change over every period, with its price chart and, on Telegram, period buttons."""
+	""".history SYMBOL: the stock's change over every period, with its price chart, and period buttons where they go."""
 	if not m.group('ticker') or m.group('extra'):
 		request.reply(["Usage: .history SYMBOL"])
 		return

From 74d26a59a9c297a072362d3712fca0e7c298d8cd Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 08:16:19 +0000
Subject: [PATCH 50/81] Bugs: Slack refusals, Slack links and escapes, secret
 paths, silent replies

- Slack: a reply Slack refuses (HTTP 200, ok false, such as not_in_channel) is logged as failed, not "200 OK".
- Slack: every link loses its markup, not only the last (`.compare BHP.AX CBA.AX` lost BHP.AX), https links
  too, and Slack's escapes are undone, so `.buy & sell` runs both reports.
- A chart with buttons whose post times out is not uploaded again without them.
- An error line names the network, not the webhook's path, which is a secret; so do the logs.
- A period button whose span has too few prices says so instead of a Telegram error line.
- `.plan` with no plans, `.performance 0d`, and `.pe`/`.peg` of an unknown ticker answer instead of saying nothing;
  the test recorder no longer counts an empty reply as an answer.
- `.help` on Slack: the blank line before each heading sits outside the bold markers.
- A NASDAQ trade is in US dollars, not the broker's currency.
- README: sync_defaults takes --dest.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 README.md               |  2 +-
 bot.py                  | 14 +++++++++-----
 lib/dialects.py         |  2 +-
 lib/inbound.py          |  7 ++++---
 lib/reports/about.py    | 10 +++++-----
 lib/reports/rankings.py |  4 ++++
 lib/reports/trades.py   |  7 +++++--
 lib/transports.py       | 39 +++++++++++++++++++++++----------------
 lib/worker.py           | 12 ++++++++----
 tests/common.py         | 10 ++++++++--
 tests/test_edges.py     | 29 ++++++++++++++++++++++++++++-
 tests/test_providers.py | 37 +++++++++++++++++++++++++++++++------
 tests/test_reports.py   | 25 +++++++++++++++++++++++--
 13 files changed, 150 insertions(+), 48 deletions(-)

diff --git a/README.md b/README.md
index 5dc53c99..a9edefe2 100644
--- a/README.md
+++ b/README.md
@@ -483,7 +483,7 @@ of the chat network keys it needs.
 
 Copies preferences.conf and metadata files into the data directory (never overwrites)
 ```
-util/sync_defaults.py [-d DEST] --apply
+util/sync_defaults.py [--dest DIR] --apply
 ```
 
 The preferences, read by the bot itself. A change takes effect on a reload. A true/false preference takes `true`/`false`, `yes`/`no`, `on`/`off` or `1`/`0`; anything else is a mistake, reported like any other. A blank value means the default.
diff --git a/bot.py b/bot.py
index c4e81213..c1cdf0d2 100755
--- a/bot.py
+++ b/bot.py
@@ -23,7 +23,8 @@
 from lib import render
 from lib.reports import cleanup
 
-AUTHENTICATED = 'finbot.authenticated' # set in a request's WSGI environ once it has passed its network's check
+ROUTED = 'finbot.routed' # a request's WSGI environ: the network whose webhook path it came to (the path is a secret)
+AUTHENTICATED = 'finbot.authenticated' # set once it has passed that network's check
 MAX_BODY = 1 << 20 # bytes in a request body, at most: an update from Telegram or Slack takes a few thousand
 _updates_seen = collections.OrderedDict() # (service, update id) of the last updates taken, oldest first
 
@@ -43,15 +44,17 @@ def seen_before(service, update_id, keep=1000):
 def main(environ, start_response):
 	"""WSGI entry point. Never lets an exception escape. A failure handling an authenticated request is reported as one line to the
 	default channel and answered 200, so the service does not retry it into the chat again. Before authentication a failure is only
-	logged: a stranger must not be able to post into the chats."""
+	logged: a stranger must not be able to post into the chats. Either names the network, not the path: a webhook's
+	path is a secret (telegramOutgoingWebhook, slackOutgoingWebhook)."""
 	try:
 		return handle(environ, start_response)
 	except Exception as e:
 		if environ.get(AUTHENTICATED):
-			delivery.report_error(e, context=environ.get('PATH_INFO', 'bot'))
+			delivery.report_error(e, context=environ[ROUTED])
 			status = '200 OK'
 		else:
-			print(f"request to {environ.get('PATH_INFO')!r} failed:", delivery.error_line(e), file=sys.stderr)
+			where = environ.get(ROUTED) or repr(environ.get('PATH_INFO'))
+			print(f"request to {where} failed:", delivery.error_line(e), file=sys.stderr)
 			print(net.redact(traceback.format_exc()), end='', file=sys.stderr)
 			status = '500 Internal Server Error'
 		try:
@@ -80,6 +83,7 @@ def print_headers():
 	if source is None:
 		print("Unknown URI", repr(path), file=sys.stderr)
 		return respond(inbound.NOT_FOUND)
+	environ[ROUTED] = source.service
 	if why := source.refuse_unread(environ): # strangers are turned away before their body is read
 		print_headers()
 		print("Fatal:", why, file=sys.stderr)
@@ -97,7 +101,7 @@ def print_headers():
 	if why := source.refuse(environ, body, update):
 		shown = {k: '<redacted>' if k == 'token' else v for k, v in update.items()} # Slack's verification token
 		print("warning:", why, file=sys.stderr)
-		print(f"inbound {path} ", json.dumps(shown, indent=4, default=str), file=sys.stderr)
+		print(f"[{source.service}]: refused update", json.dumps(shown, indent=4, default=str), file=sys.stderr)
 		return respond(inbound.FORBIDDEN)
 	environ[AUTHENTICATED] = True
 	event = source.parse(update)
diff --git a/lib/dialects.py b/lib/dialects.py
index f7191905..bc71e89d 100644
--- a/lib/dialects.py
+++ b/lib/dialects.py
@@ -37,7 +37,7 @@ def link(self, url, text):
 
 	def escape(self, text):
 		"""Text a user wrote (a .plan entry, a name) made safe to show: Telegram's HTML and Slack's markup both take &amp; &lt; &gt;.
-		It is unescaped first because it may arrive escaped already (Slack escapes what it sends; bot.py escapes Telegram names):
+		It is unescaped first because it may arrive escaped already (inbound escapes Telegram names as they arrive):
 		escaping twice would show &amp;amp;."""
 		return html.escape(html.unescape(str(text)), quote=False)
 
diff --git a/lib/inbound.py b/lib/inbound.py
index 3d88348d..b365a8fb 100644
--- a/lib/inbound.py
+++ b/lib/inbound.py
@@ -198,7 +198,8 @@ def update_id(self, update):
 
 	def parse(self, update):
 		"""A message a person wrote, a button pressed, or the challenge that checks the URL. Links and mentions lose
-		their markup: <http://dub.ax|dub.ax> becomes dub.ax and <@QWERTY> @QWERTY."""
+		their markup (<http://dub.ax|dub.ax> becomes dub.ax, <@QWERTY> @QWERTY), and Slack's escapes their meaning
+		(&amp; becomes &)."""
 		kind = update.get('type')
 		if kind == 'url_verification':
 			response = json.dumps({'challenge': update['challenge']})
@@ -217,8 +218,8 @@ def parse(self, update):
 		if event['type'] not in ('message', 'app_mention') or 'text' not in event:
 			print("[slack]: unhandled event callback type", event['type'], file=sys.stderr)
 			return UNHANDLED
-		text = re.sub(r'<http://.*\|([\w\.]+)>', r'\g<1>', event['text'])
-		text = re.sub(r'<(@[\w\.]+)>', r'\g<1>', text)
+		text = re.sub(r'<((?:https?|mailto):[^|>]*)(?:\|([^>]*))?>', lambda m: m.group(2) or m.group(1), event['text'])
+		text = html.unescape(re.sub(r'<(@[\w\.]+)>', r'\g<1>', text)) # Slack sends & < > as &amp; &lt; &gt;
 		user = '<@' + event['user'] + '>' # a mention: Slack shows the name
 		return Message(self.service, event['channel'], user, text, '@' + bot_user, user, event.get('ts', ''))
 
diff --git a/lib/reports/about.py b/lib/reports/about.py
index fc75abf3..3b7fabef 100644
--- a/lib/reports/about.py
+++ b/lib/reports/about.py
@@ -45,11 +45,11 @@ def prepare_help(service, botName, sections):
 	"""The .help text: each section's command lines under its heading (worker.help_sections() has them), then the bot's
 	@name as another way to start a command. botName is the bot's @name."""
 	payload = []
-	for i, (heading, lines) in enumerate(sections):
-		payload.append(service.bold(("\n" if i else "") + heading + ":"))
-		payload += lines
-
-	payload.append(service.bold("\nAlternative syntax:"))
+	for heading, lines in sections:
+		if payload: # a blank line between sections, outside the bold: Slack does not bold *\nPrice:*
+			payload.append('')
+		payload += [service.bold(heading + ":")] + lines
+	payload += ['', service.bold("Alternative syntax:")]
 	if service == 'slack':
 		payload.append('<' + botName + '> trades')
 	else:
diff --git a/lib/reports/rankings.py b/lib/reports/rankings.py
index 61e39d3f..fa98f7c7 100644
--- a/lib/reports/rankings.py
+++ b/lib/reports/rankings.py
@@ -1,5 +1,6 @@
 """Ranked lists of the stocks followed: .marketcap, .pe, .fpe and .peg (top, bottom, negative), .beta, .buy and .sell."""
 import sys
+from lib import answer
 from lib import config
 from lib import watchlist
 from lib import formats
@@ -96,6 +97,9 @@ def prepare_value_payload(service, action='pe', ticker_select=None, length=15):
 	if mode == 'bottom':
 		rows.reverse()
 	if ticker_select:
+		if not rows: # Yahoo does not know it, or sends no such ratio for it (a loss-maker has no trailing P/E)
+			raise answer.Answer(f"{ticker_select}: no {kind.upper()} found" if ticker_select in market_data
+				else f"{ticker_select}: no data found")
 		return [line for _, _, line in rows], None
 	rows = rows[:length]
 	trail = kind if mode != 'negative' else action # what the heading calls the ratio
diff --git a/lib/reports/trades.py b/lib/reports/trades.py
index 5fa6c9d4..2ff6da58 100644
--- a/lib/reports/trades.py
+++ b/lib/reports/trades.py
@@ -66,8 +66,11 @@ def gather(days, portfolio_select, known):
 	trades = []
 	for trade in raw:
 		market = trade['market']
+		ticker = markets.transform_to_yahoo(trade['symbol'], market)
+		# the currency by Sharesight's market code, which the table knows, not Yahoo's name for the exchange (NasdaqGS)
+		currency = markets.currency_from_market(market) or quotes.get(ticker, {}).get('currency') or \
+			trade['brokerage_currency_code']
 		if config.prefs.trades_use_yahoo:
-			ticker = markets.transform_to_yahoo(trade['symbol'], market)
 			try:
 				market = quotes[ticker]['profile_exchange']
 			except Exception as e:
@@ -75,7 +78,7 @@ def gather(days, portfolio_select, known):
 		trades.append(Trade(date=datetime.datetime.strptime(trade['transaction_date'], '%Y-%m-%d').date(),
 			id=int(trade['id']), portfolio=trade['portfolio'], bought=trade['transaction_type'] == 'BUY',
 			value=round(float(trade['price']) * float(trade['quantity'])),
-			currency=markets.currency_from_market(market) or trade['brokerage_currency_code'], symbol=trade['symbol'],
+			currency=currency, symbol=trade['symbol'],
 			market=market, holding_id=str(trade['holding_id'])))
 	return sorted(trades, key=lambda trade: (trade.date, trade.id))
 
diff --git a/lib/transports.py b/lib/transports.py
index e9c49e6c..d1a0df34 100644
--- a/lib/transports.py
+++ b/lib/transports.py
@@ -239,14 +239,20 @@ def typing_done(self):
 			print("Slack reactions.remove failed:", net.redact(e), file=sys.stderr)
 
 	def post_text(self, text, buttons=None):
-		"""chat.postMessage, without link previews (Slack reads those flags from the body)."""
+		"""chat.postMessage, without link previews. Slack's answer, or None when it failed or refused (logged): Slack
+		refuses with HTTP 200 and ok false (not_in_channel, say), so a refusal is read from the answer."""
 		if not config.env.slackBotToken:
 			print(f"ERROR slack: a reply to channel {self.where} needs slackBotToken in finbot.env; not sent",
 				file=sys.stderr)
 			return None
-		payload = {'text': text, 'unfurl_links': False, 'unfurl_media': False, 'channel': self.where}
-		return self._post_json('https://slack.com/api/chat.postMessage', payload,
-			{'Authorization': 'Bearer ' + config.env.slackBotToken})
+		try:
+			answer = slack_api('chat.postMessage', channel=self.where, text=text, unfurl_links='false',
+				unfurl_media='false')
+		except Exception as e:
+			print("Slack chat.postMessage to", self.where, "failed:", net.redact(e), file=sys.stderr)
+			return None
+		print("OK slack message to", self.where, file=sys.stderr)
+		return answer
 
 	def post_chart(self, image, caption, buttons=None):
 		"""Upload the chart into the channel with the report as its message, the way Slack takes files since it retired
@@ -290,20 +296,22 @@ def unshared(self, data):
 		return file_id
 
 	def post_blocks(self, data, caption, buttons):
-		"""The chart in a Block Kit message, the report above it and the period buttons under it. Slack's answer;
-		raises RuntimeError when Slack refuses."""
+		"""The chart in a Block Kit message, the report above it and the period buttons under it. Slack's answer, or
+		MAYBE_SENT when posting the message timed out (it may have arrived); raises RuntimeError when Slack refuses."""
 		blocks = chart_blocks(caption, buttons, self.unshared(data))
-		return until_shown(lambda: slack_api('chat.postMessage', channel=self.where, text=caption[:SECTION_LIMIT],
-			blocks=json.dumps(blocks), unfurl_links='false', unfurl_media='false'))
+		try:
+			return until_shown(lambda: slack_api('chat.postMessage', channel=self.where, text=caption[:SECTION_LIMIT],
+				blocks=json.dumps(blocks), unfurl_links='false', unfurl_media='false'))
+		except requests.exceptions.ReadTimeout as e:
+			print("Slack chat.postMessage timed out, the chart may still arrive:", net.redact(e), file=sys.stderr)
+			return MAYBE_SENT
 
 	def edit_chart(self, message, image, caption, buttons=None):
 		"""Show another period in a chart message: the new chart uploaded (a file cannot be replaced) and shown in its
 		image block, with its caption and buttons. Raises RuntimeError when Slack refuses."""
-		file_id = None
-		if image is not None:
-			if hasattr(image, 'seek'):
-				image.seek(0)
-			file_id = self.unshared(image.read() if hasattr(image, 'read') else bytes(image))
+		if hasattr(image, 'seek'):
+			image.seek(0)
+		file_id = self.unshared(image.read() if hasattr(image, 'read') else bytes(image))
 		blocks = chart_blocks(caption, buttons or [], file_id)
 		until_shown(lambda: slack_api('chat.update', channel=self.where, ts=message, text=caption[:SECTION_LIMIT],
 			blocks=json.dumps(blocks)))
@@ -374,13 +382,12 @@ def slack_api(method, **data):
 SECTION_LIMIT = 3000 # characters of text in a Block Kit section
 
 
-def chart_blocks(caption, buttons, file_id=None):
+def chart_blocks(caption, buttons, file_id):
 	"""Block Kit for a chart message: the report (in as many sections as its length needs), the chart (an uploaded file,
 	by its id) and the buttons, the one shown now marked."""
 	blocks = [{'type': 'section', 'text': {'type': 'mrkdwn', 'text': '\n'.join(part)}}
 		for part in chunked(caption.split('\n'), SECTION_LIMIT, 1000) if '\n'.join(part).strip()]
-	if file_id:
-		blocks.append({'type': 'image', 'slack_file': {'id': file_id}, 'alt_text': 'chart'})
+	blocks.append({'type': 'image', 'slack_file': {'id': file_id}, 'alt_text': 'chart'})
 	elements = [{'type': 'button', 'text': {'type': 'plain_text', 'text': button.label}, 'value': button.data,
 		'action_id': button.data, **({'style': 'primary'} if button.active else {})}
 		for row in buttons for button in row]
diff --git a/lib/worker.py b/lib/worker.py
index 3509ebd3..fb1c9b66 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -120,8 +120,9 @@ def process_callback(service, callback_id, chat_id, message_id, data):
 		_, kind, period, ref = parts
 		args = buttons.resolve(ref)
 		caption, chart = charts.build(kind, args, period, service)
-		image = render.draw(chart) if chart else None
-		chat.edit_chart(message_id, image, caption, buttons.keyboard(kind, args, period))
+		if chart is None: # too few prices in that period to draw a line
+			raise answer.Answer(f"Too few prices to draw {buttons.PERIODS[period][0]}")
+		chat.edit_chart(message_id, render.draw(chart), caption, buttons.keyboard(kind, args, period))
 	except answer.Answer as e: # nothing to draw for that period: said in the chat, as a command would
 		delivery.deliver(chat, delivery.Report(e.lines))
 	except Exception as e:
@@ -244,8 +245,9 @@ def on_calendar(request, m):
 def on_performance(request, m):
 	""".performance [period] [portfolio]: each portfolio's return, over a month by default."""
 	days, portfolio = period_and_portfolio(m.group('args'), buttons.PERIODS['m'][1]) # default view is the 1M button
-	if days > 0:
-		performance.run(request.dest, past_days=days, user=request.user, portfolio_select=portfolio)
+	if days < 1: # 0d, or ytd on 1 January
+		raise answer.Answer("The period must cover a day or more, such as 7d, 3m or ytd")
+	performance.run(request.dest, past_days=days, user=request.user, portfolio_select=portfolio)
 
 def on_session(request, m):
 	"""Mid-session moves (.session), or pre-market ones (.premarket, also post-market)."""
@@ -353,6 +355,8 @@ def on_plan(request, m):
 		with state.edit('finbot_plan.json', {}) as plans:
 			plans[request.user] = m.group(1)
 	plan = state.load('finbot_plan.json', {})
+	if not plan:
+		raise answer.Answer("No plans yet: .plan TEXT records yours")
 	def who(user):
 		"""A sender as shown: on Slack a mention (<@U123>), which is markup to keep; elsewhere the handle as text."""
 		user = user.removeprefix('@')
diff --git a/tests/common.py b/tests/common.py
index e31693a9..3b519231 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -89,8 +89,7 @@ def setUp(self):
 			lambda e, *a, **k: self.errors.append(delivery.error_line(e, k.get('context'))))
 		self.patch(transports.Transport, 'chart', lambda dest, image, caption, markup=None:
 			self.record_chart(image, caption, markup))
-		self.patch(transports.Transport, 'text', lambda dest, lines, markup=None:
-			self.sent.append(('text', list(lines))) or [])
+		self.patch(transports.Transport, 'text', lambda dest, lines, markup=None: self.record_text(lines))
 		self.patch(watchlist, 'with_holdings', lambda: TICKERS)
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t)})
@@ -114,6 +113,13 @@ def setUp(self):
 		self.patch(worker.TypingIndicator, 'start', lambda indicator: (self.typing.append(indicator), real_start(indicator)))
 		self.addCleanup(self.no_typing_left)
 
+	def record_text(self, lines):
+		"""Transport.text in tests: note ('text', lines), unless there are none: the real one posts nothing then."""
+		lines = list(lines)
+		if lines:
+			self.sent.append(('text', lines))
+		return []
+
 	def record_chart(self, image, caption, markup=None):
 		"""Transport.chart in tests: note ('photo', size, first caption line) and its buttons as Telegram gets them."""
 		image.seek(0)
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 0add3ee4..215ef1ec 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -160,6 +160,16 @@ def test_a_portfolio_name_in_any_case(self):
 		self.patch(sharesight.Portfolio, 'holdings', lambda portfolio: {'T1:NASDAQ': holding})
 		self.assertIn('Holdings for Rob (1)', str(self.command('.holdings rob')))
 
+	def test_commands_with_nothing_to_show_say_so(self):
+		"""A command with nothing to show says so, where it used to send nothing at all: .plan before anyone has saved
+		one, .performance over no days, a ratio of a ticker Yahoo does not know."""
+		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts if t != 'XYZ'})
+		for text, reply in (('.plan', 'No plans yet: .plan TEXT records yours'),
+				('.performance 0d', 'The period must cover a day or more, such as 7d, 3m or ytd'),
+				('.pe XYZ', 'XYZ: no data found'), ('.peg XYZ', 'XYZ: no data found')):
+			with self.subTest(text=text):
+				self.assertEqual(self.command(text), [('text', [reply])])
+
 	def test_trades_with_no_buy_or_sell_still_get_an_answer(self):
 		""".trades with only dividends says there were no trades (it used to say nothing)."""
 		self.patch(sharesight, 'get_trades', lambda name, pid, days=7: [{'id': 2, 'transaction_type': 'DIVIDEND', 'portfolio': name}])
@@ -293,6 +303,13 @@ def broken(*a, **k):
 		self.assertEqual((status[-1][:3], self.errors), ('500', []))
 		self.assertIn('boom', err.getvalue())
 
+	def test_a_reported_failure_names_the_network_not_its_secret_path(self):
+		"""The error line of a failed update names the network, not the webhook's path, which is a secret."""
+		self.env(telegramOutgoingWebhook='https://example.com/hook-s3cret')
+		broken = {'message': {'message_id': 1, 'from': {'first_name': 'A'}, 'text': '.help'}} # no chat
+		self.assertEqual(self.request('/hook-s3cret', broken, self.SECRET)[1], '200 OK')
+		self.assertEqual(self.errors, ["⚠️ telegram: KeyError: 'chat'"])
+
 	def test_a_failure_after_authentication_is_reported_once_and_not_retried(self):
 		"""A failure on an authenticated update is reported once and answered 200, so Telegram doesn't send it again."""
 		broken = {'message': {'message_id': 1, 'from': {'first_name': 'A'}, 'text': '.help'}} # no chat
@@ -397,6 +414,16 @@ def test_only_people_are_answered(self):
 		self.assertEqual([(r[0], r[1], r[2], r[3], r[4]) for r in self.requests],
 						 [('slack', 'C1', '<@U1>', '.price @UBOT', '@UBOT'), ('slack', 'C1', '<@U2>', '.trades', '@UBOT')])
 
+	def test_links_and_escapes_lose_their_markup(self):
+		"""Each link loses Slack's markup, however many there are (Slack links .AX tickers as domains), and Slack's
+		escapes become characters again: .buy &amp; sell runs both reports."""
+		for text, expected in (('.compare <http://BHP.AX|BHP.AX> <http://CBA.AX|CBA.AX>', '.compare BHP.AX CBA.AX'),
+				('.price <https://example.com>', '.price https://example.com'), ('.buy &amp; sell', '.buy & sell'),
+				('.plan buy &lt;BHP&gt;', '.plan buy <BHP>')):
+			with self.subTest(text=text):
+				self.assertEqual(self.event(user='U1', text=text), '200 OK')
+				self.assertEqual(self.requests[-1][3], expected)
+
 	def test_a_retried_event_runs_once(self):
 		"""An event Slack sends again (same event_id) runs once."""
 		for _ in range(2): # Slack resends an event it got no answer to within 3 seconds, with the same event_id
@@ -479,7 +506,7 @@ def test_a_plan_is_shown_as_text(self):
 		self.assertIn('<i>buy &lt;cheap&gt; &amp; hold</i>', reply[0])
 		self.assertEqual(self.command('.plan')[0][1], reply) # and the same way again
 		self.sent.clear()
-		worker.process_request('slack', 'C1', '<@U1>', '.plan sell &lt;dear&gt;', BOT, 'U') # Slack escapes what it sends
+		worker.process_request('slack', 'C1', '<@U1>', '.plan sell <dear>', BOT, 'U') # inbound unescapes Slack's text
 		text = '\n'.join(self.sent[0][1])
 		self.assertIn('*<@U1>*', text) # a Slack mention stays markup
 		self.assertIn('_sell &lt;dear&gt;_', text) # escaped once, not twice
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 1eb25d7e..9c826829 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -863,11 +863,20 @@ def test_link_previews_are_turned_off_in_the_body(self):
 
 		transports.SlackBot('C1').post_text('hi')
 		transports.SlackWebhook('https://hooks.slack.com/services/T/B/x').post_text('hi')
-		self.assertEqual(len(self.posts), 2)
-		for url, kwargs in self.posts:
-			with self.subTest(url=url): # they used to go in the headers, where Slack does not look
-				self.assertEqual((kwargs['json']['unfurl_links'], kwargs['json']['unfurl_media']), (False, False))
-				self.assertNotIn('unfurl_links', kwargs['headers'])
+		(_, reply), (_, hook) = self.posts
+		self.assertEqual((reply['data']['unfurl_links'], reply['data']['unfurl_media']), ('false', 'false'))
+		self.assertEqual((hook['json']['unfurl_links'], hook['json']['unfurl_media']), (False, False))
+		for kwargs in (reply, hook): # they used to go in the headers, where Slack does not look
+			self.assertNotIn('unfurl_links', kwargs['headers'])
+
+	def test_a_reply_slack_refuses_is_logged_as_failed(self):
+		"""Slack refuses with HTTP 200 and ok false (the bot is not in the channel, say): a failure, logged as one."""
+		self.env(slackBotToken='xoxb-test')
+		self.patch(transports.net, 'post', lambda url, **kwargs: Resp({'ok': False, 'error': 'not_in_channel'}))
+		with contextlib.redirect_stderr(io.StringIO()) as err:
+			self.assertIsNone(transports.SlackBot('C1').post_text('hi'))
+		self.assertIn('Slack chat.postMessage to C1 failed: Slack chat.postMessage: not_in_channel', err.getvalue())
+		self.assertNotIn('OK', err.getvalue())
 
 	def test_a_reply_without_a_bot_token_is_logged_not_sent(self):
 		"""A Slack reply without the bot token is logged, once each, and not sent."""
@@ -1011,7 +1020,23 @@ def test_a_long_report_takes_several_sections(self):
 		blocks = transports.chart_blocks(caption, [], 'F1')
 		self.assertEqual([block['type'] for block in blocks], ['section'] * 3 + ['image'])
 		self.assertTrue(all(len(block['text']['text']) <= 3000 for block in blocks[:3]))
-		self.assertEqual(transports.chart_blocks('', [], None), [])
+		self.assertEqual([block['type'] for block in transports.chart_blocks('', [], 'F1')], ['image'])
+
+	def test_a_chart_message_that_times_out_is_not_posted_again(self):
+		"""A chart with its buttons whose chat.postMessage times out may have arrived: no plain copy follows."""
+		record = transports.net.post
+		def post(url, **kwargs):
+			"""Time out on chat.postMessage, as a slow Slack does."""
+			if url.endswith('/chat.postMessage'):
+				self.posts.append((url, kwargs))
+				raise transports.requests.exceptions.ReadTimeout('slow')
+			return record(url, **kwargs)
+		self.patch(transports.net, 'post', post)
+		with contextlib.redirect_stderr(io.StringIO()):
+			worker.process_request('slack', 'C1', '<@U1>', '.price', BOT, 'U')
+		self.assertEqual(self.calls(), ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal',
+			'chat.postMessage'])
+		self.assertEqual(self.errors, [])
 
 
 class TelegramWebhook(FinbotCase):
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 13267201..ce5a0b21 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -65,13 +65,19 @@ def test_unknown_stickers_and_noise_are_ignored(self):
 
 
 class ReportFunctions(FinbotCase):
+	def test_help_headings_are_bold_on_slack(self):
+		"""On Slack each heading is bold, with the blank line before it outside the markers: *\nPrice:* is not bold."""
+		text = about.prepare_help(SLACK, BOT, worker.help_sections())
+		self.assertIn('*Price:*', text)
+		self.assertEqual([line for line in text if line.startswith('*\n')], [])
+
 	def test_help_comes_from_the_command_table(self):
 		""".help lists every command's lines from the table, under its section's heading, the sections in order; and
 		each line names a command that runs it."""
 		text = about.prepare_help(TELEGRAM, BOT, worker.help_sections())
 		headings = [line for line in text if line.startswith('<b>')]
-		self.assertEqual(headings[:len(worker.SECTIONS)], [f"<b>{'' if i == 0 else chr(10)}{s}:</b>"
-			for i, s in enumerate(worker.SECTIONS)])
+		self.assertEqual(headings[:len(worker.SECTIONS)], [f"<b>{s}:</b>" for s in worker.SECTIONS])
+		self.assertTrue(all(text[text.index(heading) - 1] == '' for heading in headings[1:])) # a blank line before
 		for command in worker.COMMANDS:
 			for line in command.usage:
 				with self.subTest(line=line):
@@ -286,6 +292,14 @@ def test_trades(self):
 		trades.run(transports.TelegramBot('5'))
 		self.assertTrue(self.sent)
 
+	def test_a_trade_is_in_its_market_currency_not_the_brokerage_one(self):
+		"""A NASDAQ trade is in US dollars, though Yahoo names the exchange NasdaqGS and the broker charges in
+		Australian dollars (it used to say AUD)."""
+		self.patch(sharesight, 'get_trades', lambda name, pid, days=7: [{'id': 1, 'transaction_type': 'BUY',
+			'symbol': 'T1', 'market': 'NASDAQ', 'portfolio': name, 'transaction_date': time.strftime('%Y-%m-%d'),
+			'quantity': '10', 'price': '5.0', 'value': '50', 'brokerage_currency_code': 'AUD', 'holding_id': 7}])
+		self.assertIn(' USD 50 ', self.command('.trades')[0][1][-1])
+
 	def test_trades_asks_yahoo_only_for_new_trades(self):
 		"""A scheduled trades run asks Yahoo only about trades it has not reported, and nothing once all are known."""
 		fetched = []
@@ -460,6 +474,13 @@ def test_pressing_1d_rebuilds_todays_list_and_moves_the_marker(self):
 		self.assertEqual(len(edits), 1)
 		self.assertEqual([b['text'] for b in edits[0][1]['inline_keyboard'][0]][0], '● 1D')
 
+	def test_a_period_with_too_few_prices_is_said_in_the_chat(self):
+		"""A period button whose span has too few prices for a line says so in the chat and leaves the chart as it is
+		(Telegram was sent an edit without its photo)."""
+		edits, toasts = self.press('c|h|w|T1') # the fixture's histories have no closes to draw
+		self.assertEqual((edits, toasts, self.errors), ([], ['Loading 7D…'], []))
+		self.assertEqual(self.sent, [('text', ['Too few prices to draw 7D'])])
+
 	def test_a_button_a_chart_does_not_offer_is_ignored(self):
 		"""A button a chart does not offer, or an unknown period, changes nothing."""
 		edits, toasts = self.press('c|f|d|') # performance has no 1D

From 9ce68fe7e312ea17f2741fbf362b201efe325f0a Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 08:18:18 +0000
Subject: [PATCH 51/81] record_outputs: Slack's buttons and presses, and
 scheduled reports through slackChannelID
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

A refactor of the sending code is now checked on the Slack paths added since the tool was written: charts with
period buttons (slackOutgoingWebhook), presses redrawn in place, and the scheduled reports posted as the bot. Typing
notices, Telegram's and Slack's ⏳, stay out of the recording: they depend on timing.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 util/record_outputs.py | 24 +++++++++++++++---------
 1 file changed, 15 insertions(+), 9 deletions(-)

diff --git a/util/record_outputs.py b/util/record_outputs.py
index 36ae74ce..8914de22 100755
--- a/util/record_outputs.py
+++ b/util/record_outputs.py
@@ -61,8 +61,9 @@ def runTest(self):
 			"""Record what each step posts: the real sending code runs, and net.post notes each request."""
 			posts = []
 			def post(url, **kwargs):
-				"""Note a request, the typing notice aside (timing, not output), and answer like the APIs do."""
-				if 'sendChatAction' in url:
+				"""Note a request, the typing notices aside (Telegram's, Slack's ⏳: timing, not output), and answer
+				like the APIs do."""
+				if 'sendChatAction' in url or '/reactions.' in url:
 					return common.mock.Mock(status_code=200, json=lambda: {'ok': True, 'result': True})
 				entry = {'url': url.replace('botX/', 'bot<token>/')}
 				for key in ('json', 'data', 'params'):
@@ -80,7 +81,7 @@ def post(url, **kwargs):
 			for target, name, real in real_sending: # what the fixture replaces with recorders, put back
 				self.patch(target, name, real)
 			self.patch(common.time, 'sleep', lambda seconds: None) # no pauses between a long report's messages
-			self.env(slackBotToken='xoxb-test')
+			self.env(slackBotToken='xoxb-test', slackOutgoingWebhook='https://example.com/slack') # Slack's buttons too
 			steps = []
 			def step(label, run):
 				"""Run one step and note what it posted, or what it raised."""
@@ -96,13 +97,18 @@ def step(label, run):
 				for text in COMMANDS:
 					run = lambda: worker.process_request(service, chat, user, text, common.BOT, 'U')
 					step(f'{service} {text}', run)
-			for data in BUTTONS:
-				step(f'button {data}', lambda: worker.process_callback('telegram', 'cb', '55', '9', data))
+			for service, chat, message in (('telegram', '55', '9'), ('slack', 'C1', '1.5')):
+				for data in BUTTONS:
+					label = f'button {data}' if service == 'telegram' else f'{service} button {data}'
+					step(label, lambda: worker.process_callback(service, 'cb', chat, message, data))
 			self.env(webhooks=WEBHOOKS, telegramChatID='1')
-			for line in SCHEDULED:
-				label, module, kwargs = jobs.parse_command(line)[0]
-				run = getattr(module, 'run', None) or module.lambda_handler # lambda_handler: a commit before run()
-				step(f'scheduled {line}', lambda: run(**kwargs))
+			for where in ('', ' (slackChannelID)'): # Slack's incoming webhook, then the channel as the bot
+				if where:
+					self.env(slackChannelID='C9')
+				for line in SCHEDULED:
+					label, module, kwargs = jobs.parse_command(line)[0]
+					run = getattr(module, 'run', None) or module.lambda_handler # lambda_handler: a commit before run()
+					step(f'scheduled {line}{where}', lambda: run(**kwargs))
 			Path(output).write_text(json.dumps(steps, indent=1, sort_keys=True, default=str))
 
 	result = unittest.TextTestRunner(verbosity=0, stream=io.StringIO()).run(Record())

From 1175eb109d9ed370264083c625ced3525c49ef5d Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 08:25:02 +0000
Subject: [PATCH 52/81] The typing notice belongs to the transport: it stops
 before any answer goes out
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The chat's transport keeps the notice going while a command works, and its text() and chart() stop it before
posting. The commands that post their report themselves (.earnings, .dividend, .performance, .session, .premarket,
.price, .shorts, .trades) used to answer while the notice ran on, so "typing…" could come back after the answer.
Slack's ⏳ now comes off after the answer instead of before it, so the answer no longer waits for it. report_error
takes the chat's transport; TypingIndicator is gone.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md               |  8 +++---
 lib/delivery.py         | 19 ++++++------
 lib/transports.py       | 52 +++++++++++++++++++++++++++++----
 lib/worker.py           | 64 +++++++----------------------------------
 tests/common.py         | 33 ++++++++++++---------
 tests/test_edges.py     | 26 +++++++++++------
 tests/test_providers.py | 39 +++++++++++++++----------
 7 files changed, 131 insertions(+), 110 deletions(-)

diff --git a/CLAUDE.md b/CLAUDE.md
index 1f04eb19..5afc3e6c 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -37,8 +37,7 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   straight back.
 - `lib/worker.py`: `COMMANDS`, a table of `Command(pattern, handler, usage, section, buttons)` (first match wins,
   `.SYMBOL` last), which also makes `.help` and the Telegram command keyboard, kept current in DMs; `Request`
-  (`request.reply(lines, chart, buttons)`); the typing notice (Slack: ⏳ on the message), which every command gets after
-  half a second.
+  (`request.reply(lines, chart, buttons)`); each command gets the typing notice while it works.
 - `lib/reports/`: what finbot says. Scheduled (each has `run(dest=None, **kwargs)`, which posts to the chat that asked
   or else the default chats, and `job(argv)`, its schedule-line parser): `price performance trades shorts cal rating
   milestone reminder podcasts cleanup`; the top-level `price.py` and the rest are shims that run them. Chat-only
@@ -46,8 +45,9 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   `profile` (.SYMBOL), `rankings` (.marketcap .pe .fpe .peg .beta .buy .sell).
 - `lib/delivery.py`: `Report(lines, chart, buttons)`, the destinations (the chat that asked, or the default chats),
   `deliver(dest, report)`, the one-line error report. `lib/transports.py`: a class per way of posting (`TelegramBot`,
-  `SlackBot`, `SlackWebhook`, `DiscordWebhook`): splitting, captions, charts, buttons, edits, typing. `lib/telegram.py`:
-  Bot API calls.
+  `SlackBot`, `SlackWebhook`, `DiscordWebhook`): splitting, captions, charts, buttons, edits, and the typing notice
+  (`start_typing`: after half a second; Slack's ⏳ on the message), which `text()` and `chart()` stop before posting.
+  `lib/telegram.py`: Bot API calls.
 - `lib/buttons.py`: `Button` and a chart's period buttons (each transport draws them). `lib/charts.py`: a period
   button press rebuilds its chart.
 - `lib/dialects.py`: each network's markup and message sizes. `lib/markets.py`: market tables, flags, ticker translation.
diff --git a/lib/delivery.py b/lib/delivery.py
index 89c4f568..47aef285 100644
--- a/lib/delivery.py
+++ b/lib/delivery.py
@@ -95,22 +95,21 @@ def error_line(e, context=None, limit=200):
 	return f"⚠️ {context}: {text}" if context else f"⚠️ {text}"
 
 
-def report_error(e, service=None, chat_id=None, context=None):
-	"""Send a one-line error summary to the initiating chat, else to every default channel in finbot.env. The line is
-	escaped for Telegram (sent as HTML, which refuses a stray < such as the one in <redacted>) and Slack (which asks for
-	the same escapes). Never raises: error reporting must not become a second failure."""
+def report_error(e, dest=None, context=None):
+	"""Send a one-line error summary to dest, the chat that asked (its transport), else to every default chat. The line
+	is escaped for Telegram (sent as HTML, which refuses a stray < such as the one in <redacted>) and Slack (which asks
+	for the same escapes). Never raises: error reporting must not become a second failure."""
 	try:
 		line = error_line(e, context)
 		print(line, file=sys.stderr)
 		if isinstance(e, BaseException):
 			trace = ''.join(traceback.format_exception(e))
 			print(net.redact(trace), end='', file=sys.stderr) # a request's error carries its URL
-		for dest in [reply_to(service, chat_id)] if service and chat_id else default_chats():
-			if dest:
-				try:
-					dest.post_text(line if dest.dialect == 'discord' else html.escape(line, quote=False))
-				except Exception as send_error:
-					print("Failed to report error to", dest.name, ":", send_error, file=sys.stderr)
+		for chat in targets(dest):
+			try:
+				chat.text([line if chat.dialect == 'discord' else html.escape(line, quote=False)])
+			except Exception as send_error:
+				print("Failed to report error to", chat.name, ":", send_error, file=sys.stderr)
 	except Exception as reporting_error:
 		print("Failed to report error:", reporting_error, file=sys.stderr)
 
diff --git a/lib/transports.py b/lib/transports.py
index d1a0df34..b9838788 100644
--- a/lib/transports.py
+++ b/lib/transports.py
@@ -4,6 +4,7 @@
 post_chart)."""
 import json
 import sys
+import threading
 import time
 
 import requests
@@ -19,12 +20,16 @@
 
 
 class Transport:
-	"""A destination. Subclasses set what it carries and post one message; the base splits and trims."""
+	"""A destination. Subclasses set what it carries and post one message; the base splits and trims, and keeps the
+	typing notice going while a command works."""
 	name = '' # the network, in logs
 	dialect = None
 	takes_charts = False # a chart, with the report as its caption
 	takes_buttons = False # buttons under a message
 	shows_typing = False # a typing notice while a command works
+	TYPING = (0.5, 7, 10) # the notice: first after 0.5 s (a quick answer gets none), then every 7 s, 10 times at most
+	_notices = None # the thread sending the notices
+	_quiet = None # set to stop them
 
 	def __init__(self, where, message=None):
 		"""where: the chat id, or the incoming webhook's URL; message: the id of the message being answered, if any."""
@@ -35,9 +40,44 @@ def __repr__(self):
 		"""<TelegramBot 55>: the kind and where, for logs and test failures."""
 		return f'<{type(self).__name__} {self.where}>'
 
+	def start_typing(self):
+		"""Show the typing notice while a command works, where the network has one (TYPING says when), until the first
+		message goes out or the command ends (end_typing)."""
+		if not self.shows_typing or self._notices:
+			return
+		delay, every, times = self.TYPING
+		self._quiet = threading.Event()
+		def notices():
+			"""The notices, until quiet."""
+			if self._quiet.wait(delay):
+				return
+			for _ in range(times):
+				self.typing()
+				if self._quiet.wait(every):
+					return
+		self._notices = threading.Thread(target=notices, daemon=True)
+		self._notices.start()
+
+	def stop_typing(self):
+		"""Send no more notices, and wait for one on its way: none may arrive after the answer."""
+		if self._notices:
+			self._quiet.set()
+			self._notices.join()
+
+	def end_typing(self):
+		"""The command is over: no more notices, and one that lasts until taken off (Slack's ⏳) taken off."""
+		self.stop_typing()
+		self.typing_done()
+
+	def is_typing(self):
+		"""Whether the notices are still going."""
+		return bool(self._notices and self._notices.is_alive())
+
 	def text(self, lines, buttons=None):
 		"""Post lines of text: one message, or several when they are too long for one (then at most chunk_maxlines lines
-		each), buttons under the last. Returns the network's answer for each message (None for one that failed)."""
+		each), buttons under the last; the typing notice stops first. Returns the network's answer for each message
+		(None for one that failed)."""
+		self.stop_typing()
 		lines = [str(line) for line in lines]
 		if not lines:
 			return []
@@ -53,7 +93,9 @@ def text(self, lines, buttons=None):
 		return answers
 
 	def chart(self, image, caption, buttons=None):
-		"""Post a chart, the caption under it cut to fit: the network's answer, MAYBE_SENT, or None if it failed."""
+		"""Post a chart, the caption under it cut to fit, the typing notice stopped first: the network's answer,
+		MAYBE_SENT, or None if it failed."""
+		self.stop_typing()
 		if hasattr(image, 'seek'):
 			image.seek(0)
 		return self.post_chart(image, fit_caption(caption, self.dialect), buttons)
@@ -67,10 +109,10 @@ def post_chart(self, image, caption, buttons=None):
 		raise NotImplementedError
 
 	def typing(self):
-		"""Show the typing notice, where the network has one."""
+		"""Show the typing notice once, where the network has one."""
 
 	def typing_done(self):
-		"""End the typing notice, where it lasts until ended (Slack's ⏳)."""
+		"""Take the typing notice off, where it lasts until taken off (Slack's ⏳)."""
 
 	def acknowledge(self, press, text=None):
 		"""Answer a button press, where the network waits for that (Telegram, with `text` as a toast)."""
diff --git a/lib/worker.py b/lib/worker.py
index fb1c9b66..151927f7 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -8,7 +8,6 @@
 import random
 import re
 import sys
-import threading
 import typing
 from lib import answer
 from lib import config
@@ -26,48 +25,9 @@
 from lib import transports
 from lib import jobs
 
-class TypingIndicator:
-	"""The typing notice while a command works, where the destination shows one (Telegram; Slack's ⏳). It starts after
-	`delay` seconds, so a quick answer gets none, and is sent again every 7 seconds until stopped, at most max_loops
-	times. The worker stops it before the answer goes out, however the command ends."""
-	def __init__(self, dest, max_loops=10, delay=0.5):
-		"""dest: the chat's transport."""
-		self.dest = dest
-		self.max_loops = max_loops
-		self.delay = delay
-		self._stop_event = threading.Event()
-		self._thread = None
-
-	def _worker(self):
-		"""Wait out the delay, then send the notice every 7 seconds until stopped, or until max_loops are sent."""
-		if self._stop_event.wait(self.delay):
-			return
-		for _ in range(self.max_loops):
-			self.dest.typing()
-			if self._stop_event.wait(7):
-				return
-
-	def start(self):
-		"""Start the notices, where the destination shows them."""
-		if self.dest is not None and self.dest.shows_typing:
-			self._thread = threading.Thread(target=self._worker, daemon=True)
-			self._thread.start()
-
-	def stop(self):
-		"""Stop the notices and wait for their thread to end, then end the notice where it lasts (Slack's ⏳)."""
-		if not self._thread: # never started
-			return
-		self._stop_event.set()
-		self._thread.join()
-		self.dest.typing_done()
-
-	def is_active(self):
-		"""Whether the notices are still going."""
-		return self._thread is not None and self._thread.is_alive()
-
 @dataclasses.dataclass
 class Request:
-	"""A message being answered: where it came from (service is its dialect), who sent it, and its typing notice."""
+	"""A message being answered: where it came from (service is its dialect) and who sent it."""
 	service: str
 	chat_id: str
 	user: str # the sender's handle: whose watchlist add or plan it is
@@ -76,14 +36,12 @@ class Request:
 	message_id: str = None # the message being answered
 
 	def __post_init__(self):
-		"""The chat's transport, and its typing notice (not started)."""
+		"""The chat's transport, which also keeps the typing notice going."""
 		self.dest = delivery.reply_to(self.service, self.chat_id, self.message_id)
-		self.typing = TypingIndicator(self.dest)
 
 	def reply(self, lines, chart=None, buttons=None):
 		"""Answer in the chat (see delivery.deliver): the lines, perhaps a chart (drawn, or a spec drawn only where it
-		can go) and buttons. The typing notice stops first."""
-		self.typing.stop()
+		can go) and buttons."""
 		return delivery.deliver(self.dest, delivery.Report(lines, chart, buttons))
 
 def process_job_callback(service, callback_id, chat_id, message_id, data):
@@ -102,7 +60,7 @@ def process_job_callback(service, callback_id, chat_id, message_id, data):
 		line = about.job_line(jobs.current, *jobs.current.run_command(command))
 		delivery.deliver(chat, delivery.Report([line]))
 	except Exception as e:
-		delivery.report_error(e, service, chat_id, context='jobs')
+		delivery.report_error(e, chat, context='jobs')
 
 def process_callback(service, callback_id, chat_id, message_id, data):
 	"""Button press: run a scheduled command from .jobs, or rebuild a chart message for the chosen period and edit it in place."""
@@ -126,26 +84,26 @@ def process_callback(service, callback_id, chat_id, message_id, data):
 	except answer.Answer as e: # nothing to draw for that period: said in the chat, as a command would
 		delivery.deliver(chat, delivery.Report(e.lines))
 	except Exception as e:
-		delivery.report_error(e, service, chat_id, context='chart')
+		delivery.report_error(e, chat, context='chart')
 
 def process_request(service, chat_id, user, message, botName, userRealName, message_id=None):
 	"""Entry point for inbound chat requests: the first command in COMMANDS that fits the message answers it (a message
-	that fits none gets no answer), with the typing notice meanwhile. Any failure becomes a one-line reply in the
-	originating chat, after the typing stops. message_id: the message's own id, where the network has one."""
+	that fits none gets no answer), with the typing notice meanwhile, which the first message of the answer stops and
+	the end of the command takes off. Any failure becomes a one-line reply in the originating chat. message_id: the
+	message's own id, where the network has one."""
 	request = Request(dialects.of(service), chat_id, user, userRealName, botName, message_id)
 	try:
 		for command in COMMANDS:
 			if m := match(command.pattern, message, botName):
-				request.typing.start()
+				request.dest.start_typing()
 				command.handler(request, m)
 				break
 	except answer.Answer as e: # a portfolio name the user got wrong, a ticker with no prices: an answer, not an error
 		request.reply(e.lines)
 	except Exception as e:
-		request.typing.stop() # before the error line
-		delivery.report_error(e, service, chat_id, context=message.split()[0][:30] if message.split() else None)
+		delivery.report_error(e, request.dest, context=message.split()[0][:30] if message.split() else None)
 	finally:
-		request.typing.stop()
+		request.dest.end_typing()
 	if service == 'telegram':
 		try:
 			ensure_keyboard(chat_id) # keep DM keyboards current after the reply
diff --git a/tests/common.py b/tests/common.py
index 3b519231..4ea8d12c 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -20,6 +20,7 @@
 REAL_TEXT = transports.Transport.text # FinbotCase replaces it with a recorder; tests of the real thing put it back
 REAL_CHART = transports.Transport.chart # likewise
 REAL_REPORT_ERROR = delivery.report_error # likewise
+REAL_START_TYPING = transports.Transport.start_typing # FinbotCase notes each chat it is started in
 # what FinbotCase replaces with recorders; util/record_outputs.py puts them back to record the real requests
 REAL_SENDING = ((transports.Transport, 'text', REAL_TEXT), (transports.Transport, 'chart', REAL_CHART),
 	(delivery, 'report_error', REAL_REPORT_ERROR))
@@ -88,8 +89,8 @@ def setUp(self):
 		self.patch(delivery, 'report_error',
 			lambda e, *a, **k: self.errors.append(delivery.error_line(e, k.get('context'))))
 		self.patch(transports.Transport, 'chart', lambda dest, image, caption, markup=None:
-			self.record_chart(image, caption, markup))
-		self.patch(transports.Transport, 'text', lambda dest, lines, markup=None: self.record_text(lines))
+			self.record_chart(dest, image, caption, markup))
+		self.patch(transports.Transport, 'text', lambda dest, lines, markup=None: self.record_text(dest, lines))
 		self.patch(watchlist, 'with_holdings', lambda: TICKERS)
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t)})
@@ -108,31 +109,35 @@ def setUp(self):
 		self.patch(sharesight, 'get_trades', lambda name, pid, days=7: [{
 			'id': 1, 'transaction_type': 'BUY', 'symbol': 'T1', 'market': 'NASDAQ', 'portfolio': name,
 			'transaction_date': time.strftime('%Y-%m-%d'), 'quantity': '10', 'price': '5.0', 'value': '50', 'brokerage_currency_code': 'USD', 'holding_id': 7}])
-		self.typing = [] # every typing indicator a test starts; none may still be running when the test ends
-		real_start = worker.TypingIndicator.start
-		self.patch(worker.TypingIndicator, 'start', lambda indicator: (self.typing.append(indicator), real_start(indicator)))
+		self.typing = [] # every chat a test starts a typing notice in; none may still be sending it when the test ends
+		noted = lambda chat: self.typing.append(chat) or REAL_START_TYPING(chat)
+		self.patch(transports.Transport, 'start_typing', noted)
 		self.addCleanup(self.no_typing_left)
 
-	def record_text(self, lines):
-		"""Transport.text in tests: note ('text', lines), unless there are none: the real one posts nothing then."""
+	def record_text(self, dest, lines):
+		"""Transport.text in tests: the typing notice stopped, as the real one does, and ('text', lines) noted, unless
+		there are none: the real one posts nothing then."""
+		dest.stop_typing()
 		lines = list(lines)
 		if lines:
 			self.sent.append(('text', lines))
 		return []
 
-	def record_chart(self, image, caption, markup=None):
-		"""Transport.chart in tests: note ('photo', size, first caption line) and its buttons as Telegram gets them."""
+	def record_chart(self, dest, image, caption, markup=None):
+		"""Transport.chart in tests: the typing notice stopped, as the real one does, and ('photo', size, first caption
+		line) noted with its buttons as Telegram gets them."""
+		dest.stop_typing()
 		image.seek(0)
 		self.sent.append(('photo', len(image.read()), caption.split('\n')[0]))
 		self.markups.append(transports.TelegramBot.markup(markup) if markup else None)
 		return {'ok': True}
 
 	def no_typing_left(self):
-		"""Fail the test if a typing indicator it started is still running, after stopping it."""
-		left = [indicator for indicator in self.typing if indicator.is_active()]
-		for indicator in left:
-			indicator.stop()
-		self.assertEqual(left, [], 'a typing indicator was left running')
+		"""Fail the test if a typing notice it started is still being sent, after stopping it."""
+		left = [dest for dest in self.typing if dest.is_typing()]
+		for dest in left:
+			dest.stop_typing()
+		self.assertEqual(left, [], 'a typing notice was left running')
 
 	def command(self, text, chat='55'):
 		"""Run a chat message through the real command parser and handlers."""
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 215ef1ec..0c90b576 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -54,17 +54,25 @@ def down(*a, **k):
 		worker.process_request('telegram', '55', '@u', '.performance 7d', BOT, 'U')
 		self.assertEqual(len(self.errors), 1)
 
-	def test_a_failed_command_stops_typing_before_saying_why(self):
-		"""A command that fails stops its typing indicator before the error line goes out."""
+	def test_the_typing_notice_stops_before_any_message_goes_out(self):
+		"""The typing notice stops before the first message of an answer goes out, for a command that replies, one that
+		posts its report itself (.price, .trades), and one that fails: a later notice would show "typing…" again."""
 		def down(*a, **k):
 			"""Fail like Sharesight when it errors."""
 			raise RuntimeError('Sharesight performance error: boom')
-		self.patch(sharesight, 'get_portfolios', down)
-		typing_when_reported = []
-		self.patch(delivery, 'report_error', lambda e, *a, **k:
-			typing_when_reported.append([t.is_active() for t in self.typing]))
-		worker.process_request('telegram', '55', '@u', '.performance 7d', BOT, 'U')
-		self.assertEqual(typing_when_reported, [[False]]) # its one indicator had stopped when the error line went out
+		self.patch(sharesight, 'get_performance', down)
+		self.patch(delivery, 'report_error', REAL_REPORT_ERROR)
+		self.patch(transports.Transport, 'text', REAL_TEXT)
+		self.patch(transports.Transport, 'chart', REAL_CHART)
+		typing = []
+		self.patch(transports.TelegramBot, 'post_text', lambda dest, *a, **k: typing.append(dest.is_typing()))
+		self.patch(transports.TelegramBot, 'post_chart', lambda dest, *a, **k: typing.append(dest.is_typing()) or
+			{'ok': True})
+		for cmd in ('.help', '.price', '.trades 7d', '.performance 7d'):
+			with self.subTest(cmd=cmd), contextlib.redirect_stderr(io.StringIO()):
+				typing.clear()
+				worker.process_request('telegram', '55', '@u', cmd, BOT, 'U')
+				self.assertEqual(typing, [False])
 
 	def test_no_holdings_and_empty_watchlist(self):
 		"""With nothing held and nothing watched, the watchlist, price and earnings commands still answer."""
@@ -557,7 +565,7 @@ def test_the_traceback_of_a_reported_error_is_redacted(self):
 		try:
 			raise ConnectionError('failed: ' + self.SECRET_URL)
 		except ConnectionError as e:
-			log = self.stderr_of(lambda: delivery.report_error(e, 'telegram', '55', context='x'))
+			log = self.stderr_of(lambda: delivery.report_error(e, transports.TelegramBot('55'), context='x'))
 		self.assertIn('Traceback', log)
 		self.assertIn('bot<redacted>', log)
 		self.assertNotIn('SECRET-tok_en', log)
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 9c826829..fad89c09 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -1,6 +1,6 @@
 """Yahoo, Sharesight and Telegram parsing and failure handling, driven by the real API responses saved under doc/.
 HTTP is mocked at lib.net, so the production parsing code runs unchanged."""
-import contextlib, copy, datetime, io, json, os, subprocess, unittest
+import contextlib, copy, datetime, io, json, os, subprocess, threading, unittest
 import gevent
 from unittest import mock
 from tests.common import *
@@ -595,6 +595,7 @@ def test_error_line_redacts_secrets_and_truncates(self):
 	def test_error_reports_are_escaped_for_telegram_and_slack(self):
 		"""An error line is escaped for Telegram's HTML and Slack (the same), and left as it is for Discord."""
 		sent = []
+		self.patch(transports.Transport, 'text', REAL_TEXT)
 		for kind in (transports.TelegramBot, transports.SlackWebhook, transports.DiscordWebhook):
 			self.patch(kind, 'post_text', lambda dest, text, buttons=None: sent.append((dest.name, text)))
 		self.env(webhooks={'telegram': 'https://api.telegram.org/botX/', 'slack': 'https://hooks.slack.com/x', 'discord': 'https://discord.com/x/slack'})
@@ -603,7 +604,7 @@ def test_error_reports_are_escaped_for_telegram_and_slack(self):
 		error = RuntimeError("404 for ZZ at https://x/?a=1&crumb=abc; '<' not supported between <urllib3.connection>")
 		with contextlib.redirect_stderr(io.StringIO()):
 			REAL_REPORT_ERROR(error, context='.compare <b>')
-			REAL_REPORT_ERROR(error, service='telegram', chat_id='55', context='.compare')
+			REAL_REPORT_ERROR(error, transports.TelegramBot('55'), context='.compare')
 		texts = dict(sent[:3])
 		self.assertIn('crumb=&lt;redacted&gt;', texts['telegram']) # Telegram's HTML parser rejects a bare <redacted>
 		self.assertNotIn('<', texts['telegram'] + sent[3][1])
@@ -613,9 +614,10 @@ def test_error_reports_are_escaped_for_telegram_and_slack(self):
 
 	def test_report_error_never_raises(self):
 		"""report_error survives a failed send."""
-		self.patch(transports.TelegramBot, 'post_text', mock.Mock(side_effect=RuntimeError('send failed')))
-		delivery.report_error(ValueError('boom'), service='telegram', chat_id='5')
-		delivery.report_error(ValueError('boom'))
+		self.patch(transports.Transport, 'text', mock.Mock(side_effect=RuntimeError('send failed')))
+		with contextlib.redirect_stderr(io.StringIO()):
+			REAL_REPORT_ERROR(ValueError('boom'), transports.TelegramBot('5'))
+			REAL_REPORT_ERROR(ValueError('boom'))
 
 	def test_guarded_reports_and_exits_nonzero(self):
 		"""guarded reports a failure and exits 1; a success returns its value."""
@@ -890,18 +892,25 @@ def test_a_reply_without_a_bot_token_is_logged_not_sent(self):
 		self.assertEqual(err.getvalue().count('ERROR slack: a reply to channel C1 needs slackBotToken in finbot.env; not sent'), 3)
 
 	def test_a_command_at_work_shows_an_hourglass_on_its_message(self):
-		"""While a command works, ⏳ is on the message it answers (Slack has no typing notice for bots), and it comes off
-		when the answer is ready; only with that message and the bot token."""
+		"""While a command works, ⏳ is on the message it answers (Slack has no typing notice for bots); it comes off
+		once the answer is out, so the answer does not wait for it. Only with that message and the bot token."""
 		self.env(slackBotToken='xoxb-test')
-		indicator = worker.TypingIndicator(transports.SlackBot('C1', '1.2'), delay=0)
-		indicator.start()
-		deadline = time.monotonic() + 5
-		while not self.posts and time.monotonic() < deadline:
-			time.sleep(0.01)
-		indicator.stop()
+		self.patch(transports.Transport, 'TYPING', (0, 7, 10))
+		added = threading.Event()
+		def post(url, **kwargs):
+			"""Record the post; note when ⏳ is on."""
+			self.posts.append((url, kwargs))
+			if url.endswith('/reactions.add'):
+				added.set()
+			return Resp({'ok': True})
+		self.patch(transports.net, 'post', post)
+		help_text = about.prepare_help
+		self.patch(about, 'prepare_help', lambda *a: (added.wait(5), help_text(*a))[1]) # a command slow enough for ⏳
+		worker.process_request('slack', 'C1', '<@U1>', '.help', BOT, 'U', '1.2')
+		self.assertEqual([url.rsplit('/', 1)[-1] for url, _ in self.posts],
+			['reactions.add', 'chat.postMessage', 'reactions.remove'])
 		reaction = {'channel': 'C1', 'timestamp': '1.2', 'name': 'hourglass_flowing_sand'}
-		self.assertEqual([(url, kwargs['data']) for url, kwargs in self.posts],
-			[('https://slack.com/api/reactions.add', reaction), ('https://slack.com/api/reactions.remove', reaction)])
+		self.assertEqual((self.posts[0][1]['data'], self.posts[2][1]['data']), (reaction, reaction))
 		self.assertFalse(transports.SlackBot('C1').shows_typing) # no message to put it on
 		self.env(slackBotToken='')
 		self.assertFalse(transports.SlackBot('C1', '1.2').shows_typing)

From fc4e5721001b18a764b23434a99a21c6995b317b Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 08:27:15 +0000
Subject: [PATCH 53/81] Period buttons: no 3M, so a chart has five buttons at
 most (Discord's limit for a row)

The price list offers 1D 7D 1M 1Y Max, the other charts 7D 1M 1Y Max. A period without a button (.performance
3m) marks none; a 3M button pressed under an older message is ignored.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/buttons.py          |  9 +++++----
 lib/worker.py           |  2 +-
 tests/test_providers.py |  6 +++---
 tests/test_reports.py   | 18 ++++++++++++------
 util/record_outputs.py  |  2 +-
 5 files changed, 22 insertions(+), 15 deletions(-)

diff --git a/lib/buttons.py b/lib/buttons.py
index 6782d856..86b9dbc0 100644
--- a/lib/buttons.py
+++ b/lib/buttons.py
@@ -1,5 +1,5 @@
-"""Buttons under a message: the period buttons (Week | Month | Quarter | Max) of a chart, and the callback data that
-carries what its builder needs (charts.BUILDERS). Each transport draws Button rows its own way."""
+"""Buttons under a message: the period buttons (1D | 7D | 1M | 1Y | Max) of a chart, and the callback data that carries
+what its builder needs (charts.BUILDERS). Each transport draws Button rows its own way."""
 import hashlib
 import typing
 
@@ -12,8 +12,9 @@ class Button(typing.NamedTuple):
 	active: bool = False
 
 
-PERIODS = {'d': ('1D', 1), 'w': ('7D', 7), 'm': ('1M', 30), 'q': ('3M', 90), 'y': ('1Y', 365), 'x': ('Max', None)}
-BUTTONS = {'h': 'wmqyx', 'p': 'wmqyx', 'c': 'wmqyx', 'f': 'wmqyx', 'l': 'dwmqyx'} # chart kind -> buttons offered (h=history, p=price chart, c=compare, f=performance, l=price list)
+PERIODS = {'d': ('1D', 1), 'w': ('7D', 7), 'm': ('1M', 30), 'y': ('1Y', 365), 'x': ('Max', None)}
+BUTTONS = {'h': 'wmyx', 'p': 'wmyx', 'c': 'wmyx', 'f': 'wmyx', 'l': 'dwmyx'} # chart kind -> its buttons, five at most
+# (Discord's limit for a row): h history, p a stock's price chart, c compare, f performance, l the price list
 MAX_DAYS = 3650 # what Max means for the Sharesight-based reports
 MAX_CALLBACK_BYTES = 64 # Telegram limit on callback_data
 REF_FILE = 'finbot_chart_buttons.json'
diff --git a/lib/worker.py b/lib/worker.py
index 151927f7..2a224e35 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -70,7 +70,7 @@ def process_callback(service, callback_id, chat_id, message_id, data):
 	parts = data.split('|', 3)
 	valid = len(parts) == 4 and parts[0] == 'c' and parts[1] in buttons.BUTTONS and parts[2] in buttons.BUTTONS[parts[1]]
 	chat = delivery.reply_to(service, chat_id)
-	# answering now shows Telegram's toast bubble ("Loading 3M…") while the new chart is built
+	# answering now shows Telegram's toast bubble ("Loading 1Y…") while the new chart is built
 	chat.acknowledge(callback_id, f"Loading {buttons.PERIODS[parts[2]][0]}…" if valid else None)
 	if not valid:
 		return
diff --git a/tests/test_providers.py b/tests/test_providers.py
index fad89c09..58cdb73c 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -983,8 +983,8 @@ def test_a_chart_reply_carries_its_period_buttons(self):
 		self.assertTrue(blocks[0]['text']['text'].startswith('*5% at close'))
 		self.assertEqual(blocks[1]['slack_file'], {'id': 'F1'})
 		buttons_shown = blocks[2]['elements']
-		self.assertEqual([button['text']['text'] for button in buttons_shown], ['1D', '7D', '1M', '3M', '1Y', 'Max'])
-		self.assertEqual([button.get('style') for button in buttons_shown], ['primary'] + [None] * 5)
+		self.assertEqual([button['text']['text'] for button in buttons_shown], ['1D', '7D', '1M', '1Y', 'Max'])
+		self.assertEqual([button.get('style') for button in buttons_shown], ['primary'] + [None] * 4)
 		self.assertEqual(buttons_shown[1]['value'], 'c|l|w|5,0')
 
 	def test_a_press_redraws_the_chart_in_place(self):
@@ -998,7 +998,7 @@ def test_a_press_redraws_the_chart_in_place(self):
 		self.assertEqual((update['channel'], update['ts']), ('C1', '1.5'))
 		blocks = self.blocks('chat.update')
 		self.assertEqual(blocks[1]['slack_file'], {'id': 'F1'})
-		self.assertEqual([button.get('style') for button in blocks[2]['elements']], [None, 'primary'] + [None] * 4)
+		self.assertEqual([button.get('style') for button in blocks[2]['elements']], [None, 'primary'] + [None] * 3)
 
 	def test_a_chart_not_ready_to_show_is_shown_a_moment_later(self):
 		"""Slack refusing the blocks of a chart just uploaded (invalid_blocks) is asked again after a wait."""
diff --git a/tests/test_reports.py b/tests/test_reports.py
index ce5a0b21..34e3e555 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -414,19 +414,25 @@ def buttons(self, command):
 
 	def test_performance_defaults_to_one_month(self):
 		""".performance shows a month, marked on its button, and says month, not 4 weeks."""
-		self.assertEqual(self.buttons('.performance'), ['7D', '● 1M', '3M', '1Y', 'Max'])
+		self.assertEqual(self.buttons('.performance'), ['7D', '● 1M', '1Y', 'Max'])
 		self.assertIn('month', self.sent[0][2]) # "Performance over the past month", not 4 weeks
 
 	def test_performance_period_argument_moves_the_marker(self):
 		"""A period on .performance moves the marker to its button."""
-		self.assertEqual(self.buttons('.performance 7d'), ['● 7D', '1M', '3M', '1Y', 'Max'])
-		self.assertEqual(self.buttons('.performance 1y'), ['7D', '1M', '3M', '● 1Y', 'Max'])
+		self.assertEqual(self.buttons('.performance 7d'), ['● 7D', '1M', '1Y', 'Max'])
+		self.assertEqual(self.buttons('.performance 1y'), ['7D', '1M', '● 1Y', 'Max'])
 
 	def test_price_list_has_a_1d_button_marked_by_default(self):
 		"""The price list offers 1D, marked by default, and a period moves the marker."""
-		self.assertEqual(self.buttons('.price'), ['● 1D', '7D', '1M', '3M', '1Y', 'Max'])
-		self.assertEqual(self.buttons('.price top'), ['● 1D', '7D', '1M', '3M', '1Y', 'Max'])
-		self.assertEqual(self.buttons('.price 1m'), ['1D', '7D', '● 1M', '3M', '1Y', 'Max'])
+		self.assertEqual(self.buttons('.price'), ['● 1D', '7D', '1M', '1Y', 'Max'])
+		self.assertEqual(self.buttons('.price top'), ['● 1D', '7D', '1M', '1Y', 'Max'])
+		self.assertEqual(self.buttons('.price 1m'), ['1D', '7D', '● 1M', '1Y', 'Max'])
+
+	def test_a_chart_has_five_buttons_at_most(self):
+		"""No chart offers more than five buttons, Discord's limit for a row, and a period no button has (3m) marks
+		none of them."""
+		self.assertTrue(all(len(keys) <= 5 for keys in buttons.BUTTONS.values()))
+		self.assertEqual(self.buttons('.performance 3m'), ['7D', '1M', '1Y', 'Max'])
 
 	def test_other_charts_do_not_offer_1d(self):
 		"""Only the price list offers 1D."""
diff --git a/util/record_outputs.py b/util/record_outputs.py
index 8914de22..e31626d0 100755
--- a/util/record_outputs.py
+++ b/util/record_outputs.py
@@ -23,7 +23,7 @@
 	'.performance 7d', '.performance 1y', '.performance Nope', '.pie Rob', '.pie Rob sector', '.pie', '.compare T1 T2',
 	'.compare T1 T2 3m', '.compare T1', '.history T1', '.history', '.watchlist', '.watchlist add MSFT',
 	'.watchlist add MSFT', '.watchlist del MSFT', '.watchlist frob X', '.T1', '.jobs', 'hi @bot', 'noise']
-BUTTONS = ['c|h|m|T1', 'c|h|x|T1', 'c|p|w|T1', 'c|c|q|T1,T2', 'c|f|y|', 'c|l|d|3,0', 'c|l|m|0,10', 'c|zz|m|T1']
+BUTTONS = ['c|h|m|T1', 'c|h|x|T1', 'c|p|w|T1', 'c|c|y|T1,T2', 'c|f|y|', 'c|l|d|3,0', 'c|l|m|0,10', 'c|zz|m|T1']
 SCHEDULED = ['trades.py', 'price.py midsession', 'price.py interday', 'price.py premarket', 'price.py close',
 	'price.py 7', 'cal.py earnings', 'cal.py ex-dividend', 'performance.py 7', 'rating.py', 'shorts.py', 'milestone.py',
 	'reminder.py']

From 80618e541416f64d1dfcea4717cd06152fa5a707 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 08:50:04 +0000
Subject: [PATCH 54/81] Commands by their whole name, one argument reader, the
 local-exchange retry for every ticker

- A command is called by its whole name, in any case, from . ! / or the bot's name, and Telegram's /price@bot form
  too: .planet no longer saves the plan "et", .shortman no longer runs .shorts MAN; any other word that may be a
  ticker is a profile lookup. The bot's name among the arguments is left out.
- One reader for every command's arguments: a percent, a period and a keyword from either end, in any case
  (.price TOP is the top list), the words between being tickers, a portfolio or a plan. Arguments a command cannot
  take get its usage line from the command table (.compare T1, .history, .holdings and .pie with no portfolio,
  .watchlist frob, .who, .jobs bogus).
- Every ticker a chat gives goes through yahoo.symbol: a bare one Yahoo does not know is tried on the local exchange,
  with the new exchange_suffix preference (AX by default, as before for .SYMBOL and the watchlist; empty: no retry).
  It is documented with country_code and timezone as the regional settings. Removing from the watchlist needs no
  Yahoo: BHP removes the BHP.AX watched.
- .compare's Max is 10 years, as on every other chart, which is also its default view: the Max button marked on that
  view now shows what it is marked for.
- .hello, .thanks, @bot hello/thanks and their random replies are gone, and with them lib/words.py, the quip after
  "No trades" and the alliterate preference; a preferences.conf that still sets alliterate starts, and the log says to
  delete the line.
- .help: .forward pe as the README spells it, .plan listed, .holdings PORTFOLIO.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md                 |  16 +-
 README.md                 |  14 +-
 defaults/preferences.conf |   7 +-
 lib/charts.py             |   4 +-
 lib/config.py             |  21 +-
 lib/reports/about.py      |  41 ++--
 lib/reports/cal.py        |  15 +-
 lib/reports/history.py    |  41 +---
 lib/reports/holdings.py   |  34 ++-
 lib/reports/price.py      |   9 +-
 lib/reports/profile.py    |  15 +-
 lib/reports/rankings.py   |   5 +-
 lib/reports/rating.py     |  19 +-
 lib/reports/shorts.py     |   5 +-
 lib/reports/trades.py     |   4 +-
 lib/watchlist.py          |  22 +-
 lib/words.py              | 142 ------------
 lib/worker.py             | 473 ++++++++++++++++++++------------------
 lib/yahoo.py              |  33 ++-
 tests/common.py           |   2 +-
 tests/test_config.py      |  29 ++-
 tests/test_edges.py       |  87 +++++--
 tests/test_reports.py     |  25 +-
 util/check_config.py      |  13 +-
 24 files changed, 506 insertions(+), 570 deletions(-)
 delete mode 100644 lib/words.py

diff --git a/CLAUDE.md b/CLAUDE.md
index 5afc3e6c..7fb60b3e 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -35,9 +35,12 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   reload and SIGUSR1 cache flush, each in a thread. `lib/inbound.py`: a class per network that posts to the bot
   (`Telegram`, `Slack`): its path, how it proves itself, and its updates as `Message`, `Press` or a `Response` given
   straight back.
-- `lib/worker.py`: `COMMANDS`, a table of `Command(pattern, handler, usage, section, buttons)` (first match wins,
-  `.SYMBOL` last), which also makes `.help` and the Telegram command keyboard, kept current in DMs; `Request`
-  (`request.reply(lines, chart, buttons)`); each command gets the typing notice while it works.
+- `lib/worker.py`: `COMMANDS`, a table of `Command(names, handler, usage, section, buttons, percent, period,
+  keywords)`, which also makes `.help` and the Telegram command keyboard, kept current in DMs. `read()` finds the
+  command by its whole name (any other first word that may be a ticker is `.SYMBOL`) and `read_args()` its `Args`: a
+  percent, a period and a keyword from either end, the words between; a handler raises `Usage` for arguments it cannot
+  take, and gets a ticker with `ticker(args)`. `Request` (`request.reply(lines, chart, buttons)`); each command gets
+  the typing notice while it works.
 - `lib/reports/`: what finbot says. Scheduled (each has `run(dest=None, **kwargs)`, which posts to the chat that asked
   or else the default chats, and `job(argv)`, its schedule-line parser): `price performance trades shorts cal rating
   milestone reminder podcasts cleanup`; the top-level `price.py` and the rest are shims that run them. Chat-only
@@ -55,7 +58,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   `Chart` specs and `draw()`.
 - `lib/yahoo.py lib/sharesight.py lib/shortman.py lib/simplywallst.py`: data sources. Yahoo's and Sharesight's answers
   are cached in `DATA_DIR/cache` as sent, not parsed, so a parser fix needs no flush (`yahoo._cached`).
-  `sharesight.portfolios(name=None)` gives `Portfolio` objects: `holdings()`, `tickers()`, `performance(days)`,
+  `yahoo.symbol(typed)` is Yahoo's symbol for a ticker as written, tried on the local exchange (`exchange_suffix`)
+  when Yahoo does not know it bare: every ticker a chat gives goes through it. `sharesight.portfolios(name=None)` gives `Portfolio` objects: `holdings()`, `tickers()`, `performance(days)`,
   `trades(days)`.
 - `lib/cache.py`: cache files (any may vanish at any time). `lib/state.py`: state files in `DATA_DIR`, which must last;
   change one with `with state.edit(name, default) as data:`, never load-then-save. `lib/watchlist.py`: the watchlist.
@@ -86,8 +90,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   charts, through `render.draw()`, which caches the image by the spec's hash (and the drawing code's).
 - A chart renderer in `lib/render.py` gets `@_drawn` (the bot draws in one native thread); it must not call another `@_drawn`
   function, or that thread waits on itself.
-- A new command: a `Command` in `COMMANDS`, whose usage, section and buttons make its `.help` lines and keyboard
-  buttons, and the README command list.
+- A new command: a `Command` in `COMMANDS` (its names, and what its arguments may hold), whose usage, section and
+  buttons make its `.help` lines and keyboard buttons, and the README command list.
 
 ## Saving tokens
 - Run the tests quietly: `.venv/bin/python -m unittest discover -s tests -b` prints only failures (about 0.4 KB when green). One
diff --git a/README.md b/README.md
index a9edefe2..1203e000 100644
--- a/README.md
+++ b/README.md
@@ -58,13 +58,13 @@ Where:
 * _percent_ is a number, optionally followed by '%'
 * _period_ is a number followed by 'd', 'w', 'm' or 'y', or 'ytd'
 * _portfolio_ is a Sharesight portfolio name
-* _symbol_ is a Yahoo Finance symbol
+* _symbol_ is a Yahoo Finance symbol; a bare one Yahoo does not know is tried again on your exchange (`exchange_suffix`: `BHP` may be `BHP.AX`)
 ```
 .symbol
 .beta
 .buy
 .sell
-.compare symbol symbol [symbol...] [period|max]
+.compare symbol symbol [symbol...] [period]
 .dividend [period|symbol]
 .earnings [period|symbol]
 .forward pe [symbol|top|bottom|negative]
@@ -86,7 +86,7 @@ Where:
 .who symbol
 ```
 
-`!` or `/` work in place of the dot, and so does the bot's name: `@botname price 7d`, `@botname AAPL`. `.help` lists the commands in chat.
+`!` or `/` work in place of the dot, and so does the bot's name: `@botname price 7d`, `@botname AAPL`. A command is its whole name, in any case, and so are its keywords (`.price TOP`); any other first word is a symbol. `.help` lists the commands in chat.
 
 Trade notifications are peformed by polling the Sharesight trades API on a schedule (see [Scheduling](#scheduling)), and notifying your configured chat networks of any new trades. Thus, it works best if your trades are auto-imported into Sharesight through its broker integrations, and if your environment has persistent storage so that the bot can keep track of known trade IDs between runs.
 
@@ -502,14 +502,14 @@ The preferences, read by the bot itself. A change takes effect on a reload. A tr
 | `include_portfolios` | | Sharesight portfolio IDs to report on, space separated (empty: all that are not excluded) |
 | `trades_use_yahoo` | `True` | Use Yahoo for trades; set `False` if Yahoo breaks |
 | `performance_use_sharesight` | `False` | Use Sharesight rather than Yahoo for performance over a period |
-| `alliterate` | `False` | Greeting uses alliterating adjectives (easter egg) |
 | `graph` | `True` | Send charts with reports |
 | `hyperlink` | `True` | Hyperlink tickers in reports |
 | `hyperlinkFooter` | `True` | Add a website link footer to stock lookups |
 | `hyperlinkProvider` | `google` | Where ticker links go: `google` or `yahoo` |
 | `chunk_maxlines` | `20` | Maximum lines per message before a report is split |
-| `country_code` | `AU` | Country, used by the reminder report |
-| `timezone` | `Australia/Melbourne` | Time zone, used by the reminder report |
+| `exchange_suffix` | `AX` | Regional: Yahoo's suffix for your exchange (`AX` for the ASX, `NZ`, `L`, `TO` and so on). A bare symbol Yahoo does not know is tried again with it, in every command that takes a symbol (`BHP` becomes `BHP.AX`); empty, a symbol is taken as written |
+| `country_code` | `AU` | Regional: your country, for the reminder report |
+| `timezone` | `Australia/Melbourne` | Regional: your time zone, for the reminder report |
 | `cache` | `True` | Cache Yahoo and Sharesight responses |
 | `cache_seconds` | `43200` | Cache lifetime in seconds (12 hours) |
 | `performance_cache_seconds` | `1800` | Cache lifetime of Sharesight performance reports in seconds |
@@ -517,7 +517,7 @@ The preferences, read by the bot itself. A change takes effect on a reload. A tr
 | `debug` | `False` | Extra diagnostics in the log |
 | `print_headers` | `False` | Log inbound request headers |
 
-A key in the wrong file, or one that is not listed, stops finbot starting; on a reload it is reported and nothing is applied.
+A key in the wrong file, or one that is not listed, stops finbot starting; on a reload it is reported and nothing is applied. A key finbot no longer reads (`alliterate`) is ignored, and the log at start says to delete it.
 
 edit `preferences.conf`, then reload if running. The service log shows what changed (`config: future_days 30 -> 45`). If the file has a mistake, the reload reports it (`ERROR config: ...`), applies nothing and leaves finbot running on its previous settings. Check an edit before reloading with `util/check_config.py /opt/finbot.env`. Edit `finbot.env`, then restart: a running bot cannot see it change.
 
diff --git a/defaults/preferences.conf b/defaults/preferences.conf
index 8f0355f1..5f15f0e6 100644
--- a/defaults/preferences.conf
+++ b/defaults/preferences.conf
@@ -27,14 +27,17 @@ trades_use_yahoo=True
 performance_use_sharesight=False
 
 # Display
-alliterate=False
 graph=True
 hyperlink=True
 hyperlinkFooter=True
 hyperlinkProvider=google
 chunk_maxlines=20
 
-# Locale, used by the reminder report
+# Regional: where you are.
+# exchange_suffix: a ticker Yahoo does not know is tried again on your exchange, by Yahoo's suffix for it (AX: BHP is
+# tried as BHP.AX; NZ, L, TO and so on for others). Empty: a ticker is taken as written.
+# country_code and timezone: the reminder report's country and time zone.
+exchange_suffix=AX
 country_code=AU
 timezone=Australia/Melbourne
 
diff --git a/lib/charts.py b/lib/charts.py
index c7b202fa..199e93b4 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -28,9 +28,7 @@ def price_chart(service, period, days, ticker):
 
 def compare_chart(service, period, days, *tickers):
 	""".compare: the stocks' changes over the period, rebased on one chart."""
-	caption, chart, *_ = history.prepare_compare(service, list(tickers) + ([f'{days}d'] if days else ['max']))
-	if not chart:
-		raise answer.Answer(*caption)
+	caption, chart = history.prepare_compare(service, list(tickers), days)
 	return '\n'.join(caption), chart
 
 def performance_chart(service, period, days, portfolio):
diff --git a/lib/config.py b/lib/config.py
index 20c39fd8..dd765e37 100644
--- a/lib/config.py
+++ b/lib/config.py
@@ -22,6 +22,7 @@
 	'slackOutgoingToken', 'slackOutgoingWebhook', 'slackSigningSecret', 'telegramBotToken', 'telegramOutgoingToken',
 	'telegramOutgoingWebhook') # never printed
 PREFERENCES_FILE = 'preferences.conf'
+RETIRED = ('alliterate',) # preferences.conf keys that do nothing any more: accepted, with a note at start to delete
 PASSED = 'FINBOT_PREFERENCES_PASSED' # set by the bot for its job processes: the preferences arrive in the environment, do not read the file
 OLD_ENV = REPO_DIR / '.env' # the single config file of old versions, which python-dotenv loaded; nothing reads it now
 
@@ -91,7 +92,8 @@ def read_preferences(path):
 	text = read_text(path)
 	if text is None:
 		return None
-	values, errors = parse_assignments(text, path, lambda key: key_problem(key, PREFERENCE_KEYS, STARTUP_KEYS, 'finbot.env'))
+	values, errors = parse_assignments(text, path, lambda key: None if key in RETIRED else
+		key_problem(key, PREFERENCE_KEYS, STARTUP_KEYS, 'finbot.env'))
 	if errors:
 		raise ConfigError(errors)
 	return values
@@ -129,6 +131,11 @@ def str(self, name, default=None):
 		"""A text setting."""
 		return self.value(name, default)
 
+	def blankable(self, name, default):
+		"""A text setting where a blank value means none (''), not the default."""
+		value = self.getenv(name, default)
+		return '' if value is None else str(value).strip()
+
 	def bool(self, name, default):
 		"""A boolean: true/false, yes/no, on/off, 1/0 or t/f in any case; anything else is an error, giving default."""
 		value = self.value(name, default)
@@ -231,7 +238,6 @@ def startup_values(getenv):
 class Prefs:
 	"""The preferences: preferences.conf, which the bot reads at start and on every reload. Read them as config.prefs.<name> when
 	they are used, never into a default or a module constant, so that a reload is seen."""
-	alliterate: bool
 	cache: bool
 	cache_seconds: int
 	chunk_maxlines: int
@@ -239,6 +245,7 @@ class Prefs:
 	debug: bool
 	demote_leveraged: bool
 	demote_volatile: bool
+	exchange_suffix: str # Yahoo's suffix for the local exchange (AX): a bare ticker Yahoo does not know gets it
 	exclude_portfolios: tuple
 	future_days: int
 	graph: bool
@@ -265,7 +272,6 @@ def preference_values(getenv, source=None):
 	ConfigError listing every bad value."""
 	r = _Reader(getenv)
 	prefs = Prefs(
-		alliterate=r.bool('alliterate', False),
 		cache=r.bool('cache', True),
 		cache_seconds=r.int('cache_seconds', 43200),
 		chunk_maxlines=r.int('chunk_maxlines', 20),
@@ -273,6 +279,7 @@ def preference_values(getenv, source=None):
 		debug=r.bool('debug', False),
 		demote_leveraged=r.bool('demote_leveraged', True),
 		demote_volatile=r.bool('demote_volatile', True),
+		exchange_suffix=r.blankable('exchange_suffix', 'AX').removeprefix('.').upper(),
 		exclude_portfolios=tuple(r.str('exclude_portfolios', '').split()),
 		future_days=r.int('future_days', 30),
 		graph=r.bool('graph', True),
@@ -405,7 +412,13 @@ def startup_report():
 		lines.append(f'preferences from {preferences_path} ({len(prefs.source)} set; reload with SIGHUP)')
 	else:
 		lines.append(f'no {PREFERENCES_FILE} in {env.data_dir}, using the defaults; reload will have nothing to re-read')
-	return lines
+	return lines + retired_notes(prefs.source)
+
+
+def retired_notes(preferences):
+	"""A line for each key of a preferences.conf ({key: value}) that does nothing any more."""
+	return [f'{key} in {PREFERENCES_FILE} does nothing any more: delete the line'
+		for key in RETIRED if key in preferences]
 
 
 def startup_warnings():
diff --git a/lib/reports/about.py b/lib/reports/about.py
index 3b7fabef..b642bed0 100644
--- a/lib/reports/about.py
+++ b/lib/reports/about.py
@@ -1,33 +1,28 @@
 """What finbot can do: the .help text, and the .jobs listing of the schedule (with running a job from it)."""
-import re
 from lib import buttons
 from lib import jobs
 
-def prepare_jobs_payload(service, args):
-	"""`.jobs` lists the scheduled commands, each once; `.jobs run <number|command>` runs one now, as cron would.
-	Returns (payload, buttons): Telegram gets a button per command, two to a row."""
+def prepare_jobs_payload(service, run=None):
+	"""`.jobs`: the scheduled commands, each once; with `run` (a number in that list, or a command's text) that command
+	run now, as cron would. Returns (payload, buttons): Telegram gets a button per command, two to a row."""
 	scheduler = jobs.current
 	if scheduler is None:
 		return ['Scheduled jobs are off (scheduler=False)'], None
-	args = (args or '').strip()
-	if not args:
-		commands = scheduler.commands()
-		if not commands:
-			return ['No scheduled jobs'], None
-		payload = [service.bold('Scheduled commands:')]
-		for number, (label, _, _) in enumerate(commands, 1):
-			payload.append(f"{number}. {jobs.display(label)}" + (' (running)' if label in scheduler.busy else ''))
-		payload.append("\nRun one with .jobs run NUMBER" + (' or tap it' if service == 'telegram' else ''))
-		markup = None
-		if service == 'telegram':
-			row = [buttons.Button(f"{number}. {jobs.display(label)}"[:30], f"j|{number}|{jobs.checksum(label)}")
-				for number, (label, _, _) in enumerate(commands, 1)]
-			markup = [row[i:i + 2] for i in range(0, len(row), 2)]
-		return payload, markup
-	m = re.fullmatch(r'run(?:\s+(.+))?', args, re.IGNORECASE)
-	if not m or not m.group(1):
-		return ['Usage: .jobs [run NUMBER|COMMAND]'], None
-	return [run_job_line(scheduler, m.group(1))], None
+	if run:
+		return [run_job_line(scheduler, run)], None
+	commands = scheduler.commands()
+	if not commands:
+		return ['No scheduled jobs'], None
+	payload = [service.bold('Scheduled commands:')]
+	for number, (label, _, _) in enumerate(commands, 1):
+		payload.append(f"{number}. {jobs.display(label)}" + (' (running)' if label in scheduler.busy else ''))
+	payload.append("\nRun one with .jobs run NUMBER" + (' or tap it' if service == 'telegram' else ''))
+	markup = None
+	if service == 'telegram':
+		row = [buttons.Button(f"{number}. {jobs.display(label)}"[:30], f"j|{number}|{jobs.checksum(label)}")
+			for number, (label, _, _) in enumerate(commands, 1)]
+		markup = [row[i:i + 2] for i in range(0, len(row), 2)]
+	return payload, markup
 
 def run_job_line(scheduler, text):
 	"""Run a scheduled command now and say how it went in one line."""
diff --git a/lib/reports/cal.py b/lib/reports/cal.py
index 6d9428fc..ec50fdb7 100644
--- a/lib/reports/cal.py
+++ b/lib/reports/cal.py
@@ -16,16 +16,15 @@ def primary_listings():
 EVENTS = {'earnings': ('earnings_date', 'Upcoming earnings'), 'ex-dividend': ('ex_dividend_date', 'Upcoming ex-dividends')} # -> (field, heading)
 
 def run(dest=None, *, days=None, specific_stock=None, event='earnings'):
-	"""Post the earnings or ex-dividend dates ahead (`event`) for the tracked stocks or one: to the chat that asked
-	(dest), else to the default chats. The scheduled earnings report is pinned where the network can; a reply is not,
-	or the chat would collect pinned copies."""
+	"""Post the earnings or ex-dividend dates ahead (`event`) for the tracked stocks or one (Yahoo's symbol): to the
+	chat that asked (dest), else to the default chats. The scheduled earnings report is pinned where the network can; a
+	reply is not, or the chat would collect pinned copies."""
 	interactive = dest is not None
 	if days is None:
 		days = config.prefs.future_days
-	tickers = [markets.transform_to_yahoo(specific_stock)] if specific_stock else watchlist.with_holdings()
-	events = gather(tickers, days, specific_stock, event)
+	events = gather(watchlist.covered(specific_stock), days, specific_stock, event)
 	for chat in delivery.targets(dest):
-		payload = report(chat.dialect, events, tickers, days, specific_stock, interactive, event)
+		payload = report(chat.dialect, events, days, specific_stock, interactive, event)
 		answers = delivery.deliver(chat, delivery.Report(payload))
 		if event == 'earnings' and not specific_stock and not interactive and hasattr(chat, 'pin'):
 			chat.pin(answers) # the scheduled earnings report stays on top, where the network can pin
@@ -70,7 +69,7 @@ def primary_listing(ticker, data, field):
 		return None, None
 	return primary, details
 
-def report(service, events, tickers, days, specific_stock, interactive, event):
+def report(service, events, days, specific_stock, interactive, event):
 	"""The events in one network's markup, grouped by date under a heading (one stock's carry their dates instead)."""
 	heading = EVENTS[event][1]
 	payload = []
@@ -86,7 +85,7 @@ def report(service, events, tickers, days, specific_stock, interactive, event):
 		payload.insert(0, service.bold(f"{heading} {formats.days_english(days, 'for the next ')}:"))
 	elif not payload and interactive:
 		ahead = formats.days_english(days, 'for the next ')
-		payload = [f"No events found for {tickers[0]}" if specific_stock else f"No events found {ahead}"]
+		payload = [f"No events found for {specific_stock}" if specific_stock else f"No events found {ahead}"]
 	return payload
 
 def job(argv):
diff --git a/lib/reports/history.py b/lib/reports/history.py
index 04208b37..90c4085c 100644
--- a/lib/reports/history.py
+++ b/lib/reports/history.py
@@ -3,32 +3,13 @@
 from lib import answer
 from lib import formats
 from lib import links
-from lib import markets
 from lib import render
 from lib import yahoo
 
-def prepare_compare(service, args):
-	"""Compare 2-6 tickers as rebased % change. A trailing time period (3m/1y/90d/ytd) is optional;
-	without one, the span is the earliest date covered by every ticker. Returns (caption lines, chart spec, tickers,
-	period days)."""
+def prepare_compare(service, tickers, period_days=None):
+	"""(caption lines, chart spec): 2-6 tickers (Yahoo's symbols) as rebased % change over the last `period_days`, else
+	10 years (as Max is on every chart), from no earlier than the first date every one of them covers."""
 	import pandas as pd
-	usage = ['Usage: .compare SYMBOL SYMBOL [SYMBOL...up to 6] [period|max]  (default: 10 years)']
-	period_days = None
-	full_history = False
-	args = list(args)
-	if args:
-		if args[-1].lower() in ('max', 'all'):
-			full_history = True
-			args.pop()
-		else:
-			try:
-				period_days = formats.days_from_human_days(args[-1])
-				args.pop()
-			except ValueError:
-				pass
-	tickers = list(dict.fromkeys(markets.transform_to_yahoo(a.upper()) for a in args)) # de-duplicated, order kept
-	if not 2 <= len(tickers) <= 6:
-		return usage, None, tickers, period_days
 	names, exchanges, closes = {}, {}, {}
 	for ticker, history in zip(tickers, yahoo.fetch_each(yahoo.price_series, tickers)): # downloaded at the same time
 		if isinstance(history, Exception):
@@ -36,17 +17,9 @@ def prepare_compare(service, args):
 		names[ticker], exchanges[ticker], closes[ticker] = history
 	limiting = max(closes, key=lambda t: closes[t].index[0]) # ticker with the shortest history
 	common_start = closes[limiting].index[0]
-	start = common_start
-	note = ''
-	if not period_days and not full_history: # default window; ask for a longer period (or max) to go further
-		default_start = pd.Timestamp(datetime.datetime.now().date() - datetime.timedelta(days=3655))
-		start = max(common_start, default_start)
-	if period_days:
-		requested = pd.Timestamp(datetime.datetime.now().date() - datetime.timedelta(days=period_days))
-		if requested >= common_start:
-			start = requested
-		else:
-			note = f" (limited by {limiting} history)"
+	requested = pd.Timestamp(datetime.datetime.now().date() - datetime.timedelta(days=period_days or 3655))
+	start = max(common_start, requested)
+	note = f" (limited by {limiting} history)" if period_days and requested < common_start else ''
 	cropped = []
 	for ticker in tickers:
 		series = closes[ticker]
@@ -69,7 +42,7 @@ def prepare_compare(service, args):
 	for pct, t in reversed(results):
 		link = links.finance_link(t, exchanges[t], service, days=span_days, brief=False)
 		caption.append(f"{formats.get_emoji(pct)} {names[t]} ({link}): {formats.signed_percent(pct, 1)}")
-	return caption, chart, tickers, period_days
+	return caption, chart
 
 def history_payload(ticker, service, highlight=None, graph=False):
 	"""(lines, percent table, chart): .history's caption, the change over every period with the `highlight` one marked,
diff --git a/lib/reports/holdings.py b/lib/reports/holdings.py
index 2ed32cd4..cf284f8e 100644
--- a/lib/reports/holdings.py
+++ b/lib/reports/holdings.py
@@ -68,9 +68,10 @@ def prepare_pie(portfolio_name, by, service, max_slices=8):
 	return payload, render.Chart('pie', title, subtitle, tuple(rows))
 
 def prepare_watchlist(service, user, action=None, ticker=None):
-	"""The .watchlist reply: the companies watched, or one line on adding ('add') or removing ('delete') a ticker."""
+	"""The .watchlist reply: the companies watched, or one line on adding ('add') or removing ('delete') a ticker
+	(Yahoo's symbol, as yahoo.symbol gives it)."""
 	if action:
-		return [watchlist_change(service, user, action, markets.transform_to_yahoo(ticker))]
+		return [watchlist_change(service, user, action, ticker)]
 	watched = watchlist.tickers()
 	market_data = yahoo.fetch(watched) if watched else {}
 	lines = sorted((f"{markets.flag_from_ticker(t)} {quote['profile_title']} ({links.finance_link(t, quote.get('profile_exchange'), service)})"
@@ -79,24 +80,15 @@ def prepare_watchlist(service, user, action=None, ticker=None):
 		return ['Watchlist is empty. Try ".watchlist add SYMBOL" to create it']
 	return [f"Hi {user}, I'm currently tracking:"] + lines
 
-def watchlist_change(service, user, action, asked):
-	"""The reply to adding or removing `asked` (a Yahoo ticker)."""
-	def described(ticker, quote):
-		"""'Full Name (TICKER)' in bold, linked to the stock's page."""
-		quote = quote or {}
-		link = links.finance_link(ticker, quote.get('profile_exchange') or ticker, service)
-		return service.bold(f"{quote['profile_title']} ({link})" if quote.get('profile_title') else link)
+def watchlist_change(service, user, action, ticker):
+	"""The reply to adding or removing `ticker` (Yahoo's symbol)."""
+	quote = yahoo.fetch([ticker]).get(ticker) or {}
+	link = links.finance_link(ticker, quote.get('profile_exchange') or ticker, service)
+	described = service.bold(f"{quote['profile_title']} ({link})" if quote.get('profile_title') else link)
 	if action == 'delete':
-		if asked not in watchlist.tickers():
-			return f"Beep Boop. I could not find {service.bold(asked)} to remove it"
-		quote = yahoo.fetch([asked]).get(asked)
-		watchlist.remove(asked)
-		return f"Ok {user}, I deleted {described(asked, quote)}"
-	found, quote = watchlist.find(asked)
+		if not watchlist.remove(ticker):
+			return f"Beep Boop. I could not find {service.bold(ticker)} to remove it"
+		return f"Ok {user}, I deleted {described}"
 	if not quote:
-		return f"Beep Boop. I could not find {service.bold(asked)} to add it"
-	added = watchlist.add(found)
-	if found != asked:
-		done = "so I added" if added else "and I'm already tracking"
-		return f"Beep Boop. I could not find {service.bold(asked)} {done} {described(found, quote)}"
-	return f"Ok {user}, I added {described(found, quote)}" if added else f"{user}, I'm already tracking {described(found, quote)}"
+		return f"Beep Boop. I could not find {service.bold(ticker)} to add it"
+	return f"Ok {user}, I added {described}" if watchlist.add(ticker) else f"{user}, I'm already tracking {described}"
diff --git a/lib/reports/price.py b/lib/reports/price.py
index a906df63..d827a22e 100644
--- a/lib/reports/price.py
+++ b/lib/reports/price.py
@@ -16,9 +16,10 @@
 from lib.reports import history
 
 def gather(specific_stock=None, days=None):
-	"""(tickers, market data) for a price report: Yahoo's quotes; over `days` each holding's Sharesight figure, and for the tickers
-	Sharesight has none for (the watchlist) their Yahoo history, fetched together up front."""
-	tickers = [markets.transform_to_yahoo(specific_stock)] if specific_stock else watchlist.with_holdings()
+	"""(tickers, market data) for a price report on the stocks followed or one (Yahoo's symbol): Yahoo's quotes; over
+	`days` each holding's Sharesight figure, and for the tickers Sharesight has none for (the watchlist) their Yahoo
+	history, fetched together up front."""
+	tickers = watchlist.covered(specific_stock)
 	market_data = yahoo.fetch(tickers)
 	if config.prefs.demote_volatile: # trigger_bar compares market caps, in US dollars
 		yahoo.add_usd_market_caps(market_data)
@@ -152,7 +153,7 @@ def report(service, mode='interday', threshold=None, days=None, top=None, user='
 			session = mode if mode in ('midsession', 'premarket') else None
 			graph = render.Chart('bars', heading, rows=tuple(label(chart_rows)), session=session)
 	elif not payload and interactive:
-		payload = [nothing(mode, threshold, user, tickers[0] if specific_stock else None, 'REGULAR' in states)]
+		payload = [nothing(mode, threshold, user, specific_stock, 'REGULAR' in states)]
 	return payload, graph
 
 def nothing(mode, threshold, user, ticker, in_session):
diff --git a/lib/reports/profile.py b/lib/reports/profile.py
index 989501a4..76ab8725 100644
--- a/lib/reports/profile.py
+++ b/lib/reports/profile.py
@@ -5,7 +5,6 @@
 from lib import config
 from lib import formats
 from lib import links
-from lib import markets
 from lib import yahoo
 from lib import simplywallst
 
@@ -39,12 +38,11 @@ def trend_marks(figures):
 }
 
 def prepare_profile_payload(service, user, ticker):
-	"""The .SYMBOL report: what Yahoo knows about one company, in sections a blank line apart. A field Yahoo did not
-	send leaves its line out, and a section with none its blank line."""
-	ticker_orig = markets.transform_to_yahoo(ticker)
-	ticker, market_data = yahoo.resolve(ticker_orig, lambda symbol: yahoo.fetch_detail(symbol, 600))
+	"""The .SYMBOL report: what Yahoo knows about one company (ticker: Yahoo's symbol), in sections a blank line apart.
+	A field Yahoo did not send leaves its line out, and a section with none its blank line."""
+	market_data = yahoo.fetch_detail(ticker, 600)
 	if ticker not in market_data:
-		return [f"{user} 🛑 Beep Boop. I could not find {ticker_orig}"]
+		return [f"{user} 🛑 Beep Boop. I could not find {ticker}"]
 	data = market_data[ticker]
 	state = data.get('marketState', '').rstrip()
 	exchange = data.get('profile_exchange', '').replace('NasdaqCM', 'Nasdaq').replace('NasdaqGS', 'Nasdaq')
@@ -59,10 +57,7 @@ def prepare_profile_payload(service, user, ticker):
 	ticker_link = links.finance_link(ticker, exchange, service, brief=False)
 	light = MARKET_STATE_EMOJI.get(state, '🔴') if state else ''
 	heading = f"{service.bold(data.get('profile_title', ''))} ({ticker_link}) {light}"
-	if ticker_orig == ticker:
-		payload.insert(0, heading)
-	else:
-		payload[:0] = [f"I could not find {ticker_orig} but I found {ticker_link}:", "", heading]
+	payload.insert(0, heading)
 	if len(payload) < 2:
 		payload.append("no data found")
 	return payload
diff --git a/lib/reports/rankings.py b/lib/reports/rankings.py
index fa98f7c7..f1b93247 100644
--- a/lib/reports/rankings.py
+++ b/lib/reports/rankings.py
@@ -13,7 +13,7 @@ def prepare_marketcap_payload(service, action='top', length=15):
 	"""The `length` tracked stocks of largest (action 'top') or smallest ('bottom') market cap, compared in US dollars at Yahoo's
 	exchange rates: a market cap in won or yen is a far bigger number. Any other action is one ticker's, in its own currency too."""
 	if action not in ('top', 'bottom'):
-		ticker = markets.transform_to_yahoo(action.upper())
+		ticker = action
 		data = yahoo.add_usd_market_caps(yahoo.fetch_detail(ticker, 600)).get(ticker, {})
 		if 'market_cap' not in data:
 			return [f"Mkt cap not found for {ticker}"]
@@ -70,8 +70,7 @@ def prepare_value_payload(service, action='pe', ticker_select=None, length=15):
 	P/E is analysts' estimate, so its lists take only stocks that min_analysts or more analysts cover, as .buy does."""
 	mode, kind = action.split(' ', 1) if action.split()[0] in ('bottom', 'negative') else ('top', action)
 	field = RATIOS[kind]
-	tickers = [ticker_select] if ticker_select else watchlist.with_holdings()
-	market_data = yahoo.fetch(tickers)
+	market_data = yahoo.fetch(watchlist.covered(ticker_select))
 	if kind == 'peg': # the PEG is in Yahoo's details
 		market_data |= yahoo.fetch_details(market_data, strict=bool(ticker_select))
 	covered = None # for a forward P/E list: the tickers enough analysts cover
diff --git a/lib/reports/rating.py b/lib/reports/rating.py
index db57ecae..d8b7aef1 100644
--- a/lib/reports/rating.py
+++ b/lib/reports/rating.py
@@ -2,29 +2,27 @@
 from lib import delivery
 from lib import state
 from lib import links
-from lib import markets
 from lib import watchlist
 from lib import yahoo
 
 STATE_FILE = 'finbot_rating.json' # {ticker: [action, index]} as last seen
 
-def run(dest=None, *, specific_stock=None):
+def run(dest=None):
 	"""Post the analysts' consensus ratings that changed since the last scheduled run, which records the ratings it saw:
 	to the chat that asked (dest), else to the default chats."""
 	interactive = dest is not None
-	tickers = [markets.transform_to_yahoo(specific_stock)] if specific_stock else watchlist.with_holdings()
-	changes, seen = gather(tickers, specific_stock)
+	changes, seen = gather(watchlist.with_holdings())
 	for chat in delivery.targets(dest):
-		delivery.deliver(chat, delivery.Report(report(chat.dialect, changes, tickers, specific_stock, interactive)))
+		delivery.deliver(chat, delivery.Report(report(chat.dialect, changes, interactive)))
 	if seen and not interactive: # added to what is recorded: a ticker missing from this run keeps its last rating
 		with state.edit(STATE_FILE, {}) as ratings:
 			ratings.update(seen)
 
-def gather(tickers, specific_stock):
+def gather(tickers):
 	"""(the ratings that changed since the last scheduled run, [(ticker, its details, old [action, index], new [action,
 	index], analysts)]; {ticker: [action, index]} of every rating seen)."""
 	old = state.load(STATE_FILE, {})
-	market_data = yahoo.fetch_details(tickers, strict=bool(specific_stock))
+	market_data = yahoo.fetch_details(tickers)
 	changes, seen = [], {}
 	for ticker in tickers:
 		try:
@@ -42,7 +40,7 @@ def gather(tickers, specific_stock):
 			changes.append((ticker, market_data[ticker], (old_action, old_index), (action, index), analysts))
 	return changes, seen
 
-def report(service, changes, tickers, specific_stock, interactive):
+def report(service, changes, interactive):
 	"""The rating changes in one network's markup, sorted: 📈 when the consensus improved (its index fell: 1 is a strong
 	buy), 📉 when it worsened."""
 	payload = []
@@ -54,10 +52,9 @@ def report(service, changes, tickers, specific_stock, interactive):
 		payload.append(f"{emoji} {data['profile_title']} ({ticker_link}) changed from {message}")
 	payload.sort()
 	if payload:
-		if not specific_stock:
-			payload.insert(0, service.bold('Consensus rating changes:'))
+		payload.insert(0, service.bold('Consensus rating changes:'))
 	elif interactive:
-		payload = [f"No rating changes found for {tickers[0]}" if specific_stock else "No rating changes found"]
+		payload = ["No rating changes found"]
 	return payload
 
 def job(argv):
diff --git a/lib/reports/shorts.py b/lib/reports/shorts.py
index 99927af6..a9887433 100644
--- a/lib/reports/shorts.py
+++ b/lib/reports/shorts.py
@@ -1,7 +1,6 @@
 """.shorts: the most shorted stocks followed, from Short Man (ASX) and Yahoo (scheduled: shorts.py)."""
 from lib import config
 from lib import delivery
-from lib import markets
 from lib import render
 from lib import watchlist
 from lib import yahoo
@@ -9,7 +8,7 @@
 
 def run(dest=None, *, threshold=None, specific_stock=None):
 	"""Post the stocks followed whose short interest is over `threshold` percent (the shorts_percent preference when
-	None), or one stock's: to the chat that asked (dest), else to the default chats."""
+	None), or one stock's (Yahoo's symbol): to the chat that asked (dest), else to the default chats."""
 	interactive = dest is not None
 	if threshold is None:
 		threshold = config.prefs.shorts_percent
@@ -21,7 +20,7 @@ def run(dest=None, *, threshold=None, specific_stock=None):
 def gather(specific_stock):
 	"""(the stocks followed, or the one, their quotes with short interest): Short Man has the ASX's figures, Yahoo's
 	details the US's."""
-	tickers = [markets.transform_to_yahoo(specific_stock)] if specific_stock else watchlist.with_holdings()
+	tickers = watchlist.covered(specific_stock)
 	market_data = yahoo.fetch(tickers)
 	market_data |= yahoo.fetch_details((ticker for ticker in tickers if '.' not in ticker), strict=bool(specific_stock))
 	return tickers, shortman.fetch(market_data)
diff --git a/lib/reports/trades.py b/lib/reports/trades.py
index 2ff6da58..d02d488b 100644
--- a/lib/reports/trades.py
+++ b/lib/reports/trades.py
@@ -1,12 +1,10 @@
 """.trades: Sharesight buys and sells; a scheduled run reports only new ones (scheduled: trades.py)."""
 import dataclasses
 import datetime
-import random
 import sys
 
 from lib import config
 from lib import state
-from lib import words
 from lib import sharesight
 from lib import delivery
 from lib import formats
@@ -43,7 +41,7 @@ def run(dest=None, *, days=None, user='', portfolio_select=None):
 		payload = report(chat.dialect, trades, days, interactive)
 		if interactive and not payload: # no trades, or none of them a buy or a sell
 			when = formats.days_english(days, 'in the past ')
-			payload = [f"{user} No trades {when}. {random.choice(words.noTradesVerb)}"]
+			payload = [f"{user} No trades {when}."]
 		delivery.deliver(chat, delivery.Report(payload))
 	if trades and not interactive: # recorded once the report is out
 		with state.edit(STATE_FILE, []) as known:
diff --git a/lib/watchlist.py b/lib/watchlist.py
index d0edac4d..c693cf18 100644
--- a/lib/watchlist.py
+++ b/lib/watchlist.py
@@ -1,8 +1,9 @@
 """The watchlist: tickers followed beside the Sharesight holdings, in DATA_DIR/finbot_watchlist.json. Every call reads the file,
 so the bot and its scheduled reports (other processes) always see the same list."""
+from lib import config
+from lib import markets
 from lib import sharesight
 from lib import state
-from lib import yahoo
 
 FILE = 'finbot_watchlist.json'
 
@@ -12,11 +13,15 @@ def tickers():
 	return state.load(FILE, [])
 
 
-def find(ticker):
-	"""(ticker, its Yahoo quote) for a ticker Yahoo knows, tried on the ASX when a bare symbol is unknown (BHP: BHP.AX); the
-	quote is None when Yahoo knows neither."""
-	ticker, found = yahoo.resolve(ticker, lambda symbol: yahoo.fetch([symbol]))
-	return ticker, found.get(ticker)
+def watched(typed):
+	"""The watched ticker that a ticker as written names: BHP.AX as written, or BHP on the local exchange (the
+	exchange_suffix preference) when only BHP.AX is watched; else Yahoo's spelling of what was written. Yahoo is not
+	asked, so a ticker it no longer knows can still be removed."""
+	ticker = markets.transform_to_yahoo(typed.upper())
+	local = f'{ticker}.{config.prefs.exchange_suffix}'
+	if ticker not in tickers() and config.prefs.exchange_suffix and '.' not in ticker and local in tickers():
+		return local
+	return ticker
 
 
 def add(ticker):
@@ -37,6 +42,11 @@ def remove(ticker):
 	return True
 
 
+def covered(ticker=None):
+	"""The tickers a report covers: the one asked about, else every stock followed (with_holdings)."""
+	return [ticker] if ticker else with_holdings()
+
+
 def with_holdings():
 	"""The watchlist and every Sharesight holding: the tickers the scheduled reports cover, sorted."""
 	followed = set(sharesight.tickers()) | set(tickers())
diff --git a/lib/words.py b/lib/words.py
deleted file mode 100644
index 265b993f..00000000
--- a/lib/words.py
+++ /dev/null
@@ -1,142 +0,0 @@
-"""The words of the easter eggs: .hello, .thanks, and the quip when there were no trades."""
-
-adjectives = [
-	'absolutely',
-	'amazingly',
-	'awfully',
-	'bitchingly',
-	'bloodly',
-	'comfortably',
-	'darned',
-	'distinctly',
-	'especially',
-	'ever so',
-	'enjoyably',
-	'exceedingly',
-	'exceptionally',
-	'extraordinarily',
-	'extremely',
-	'flatteringly',
-	'fortunately',
-	'fortuitously',
-	'greatly',
-	'honourably',
-	'hugely',
-	'humblingly',
-	'inordinately',
-	'jolly',
-	'joyfully',
-	'keenly',
-	'kindly',
-	'knowingly',
-	'luckily',
-	'memorably',
-	'mostly',
-	'needlessly',
-	'overwhelmingly',
-	'pleasurably',
-	'questionably',
-	'rewardingly',
-	'serendipitously',
-	'seriously',
-	'significantly',
-	'so',
-	'super',
-	'supremely',
-	'surprisingly',
-	'terribly',
-	'terrifically',
-	'thoroughly',
-	'totally',
-	'tremendously',
-	'uber',
-	'uncomprehensibly',
-	'unexceptionably',
-	'unquestionably',
-	'unmentionably',
-	'unusually',
-	'discomfortingly',
-	'incomprehensibly',
-	'very',
-	'wholesomely',
-	'xtra',
-	'yawningly',
-	'zealously'
-]
-
-adjectives_two = [
-	'amazing',
-	'bitching',
-	'comforting',
-	'comprehensible',
-	'dapper',
-	'enjoyable',
-	'exceptional',
-	'extraordinary',
-	'flattering',
-	'fortuitous',
-	'fortunate',
-	'great',
-	'grateful',
-	'honouring',
-	'huge',
-	'humbling',
-	'inordinate',
-	'joyous',
-	'keen',
-	'knowing',
-	'kind',
-	'lucky',
-	'memorable',
-	'mentionable',
-	'nice',
-	'overwhelming',
-	'pleasurable',
-	'questionable',
-	'rewarding',
-	'serendipitous',
-	'significant',
-	'special',
-	'super',
-	'terrific',
-	'tremendous',
-	'useful',
-	'virtuous',
-	'wholesome',
-	'xenial',
-	'yawning',
-	'zany'
-]
-
-verb = [
-	'pretend to greet',
-	'apparently share this moment with',
-	'coincide in temporal reality with',
-	'cross digital paths with',
-	'simulate becoming acquainted with',
-	'fire photons at',
-	'traverse cyberspace with',
-	'co-exist in spacetime with',
-	'fire electrons at',
-	'encode character sets with',
-	'convert utf-8 to binary and then back to utf-8 with',
-	'update this pixel matrix with',
-	'lose money with',
-	'maintain character with',
-	'act like I comprehend'
-]
-
-noTradesVerb = [
-	"The money is probably resting in another account",
-	"Oh well. The market would have just gone down anyway",
-	"Heating bills don't pay themselves you know",
-	"You were fearful when others were also fearful",
-	"I bet you're regretting that extra 1¢ on your limit order now",
-	"This is why you can't have nice things",
-	"But I'm sure that's just a rounding error",
-	"Nothing + nothing = more nothing. Well played",
-	"Or maybe there was. They don't pay me for this you know",
-	"I guess that's right. I'm kinda busy with ChatGPT in another window",
-	"I like this contrarian play you're having on 'being in it to win it'",
-	"With your stock picking skills, this is probably for the best"
-]
diff --git a/lib/worker.py b/lib/worker.py
index 2a224e35..db154c6a 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -1,29 +1,26 @@
-"""Chat commands: a message is answered by the handler of the first Command in COMMANDS that fits it. The table also
-makes the .help text and the Telegram command keyboard, which is kept current in every DM."""
+"""Chat commands: a message is answered by the Command in COMMANDS whose name it starts with (read), or .SYMBOL. The
+table also makes the .help text and the Telegram command keyboard, which is kept current in every DM."""
 import dataclasses
-import datetime
 import hashlib
 import json
 import math
-import random
 import re
 import sys
 import typing
 from lib import answer
-from lib import config
 from lib import dialects
 from lib import state
-from lib import words
 from lib import formats
-from lib import markets
 from lib import delivery
 from lib import render
 from lib import buttons
 from lib import charts
 from lib.reports import about, cal, history, holdings, performance, price, profile, rankings, shorts, trades
-from lib import telegram
+from lib import sharesight
 from lib import transports
 from lib import jobs
+from lib import watchlist
+from lib import yahoo
 
 @dataclasses.dataclass
 class Request:
@@ -87,34 +84,108 @@ def process_callback(service, callback_id, chat_id, message_id, data):
 		delivery.report_error(e, chat, context='chart')
 
 def process_request(service, chat_id, user, message, botName, userRealName, message_id=None):
-	"""Entry point for inbound chat requests: the first command in COMMANDS that fits the message answers it (a message
-	that fits none gets no answer), with the typing notice meanwhile, which the first message of the answer stops and
-	the end of the command takes off. Any failure becomes a one-line reply in the originating chat. message_id: the
-	message's own id, where the network has one."""
+	"""Entry point for inbound chat requests: the command the message calls (see read) answers it, with the typing
+	notice meanwhile, which the first message of the answer stops and the end of the command takes off; a message that
+	calls none gets no answer. Arguments the command cannot take get its usage, and any failure a one-line reply in the
+	originating chat. message_id: the message's own id, where the network has one."""
 	request = Request(dialects.of(service), chat_id, user, userRealName, botName, message_id)
-	try:
-		for command in COMMANDS:
-			if m := match(command.pattern, message, botName):
-				request.dest.start_typing()
-				command.handler(request, m)
-				break
-	except answer.Answer as e: # a portfolio name the user got wrong, a ticker with no prices: an answer, not an error
-		request.reply(e.lines)
-	except Exception as e:
-		delivery.report_error(e, request.dest, context=message.split()[0][:30] if message.split() else None)
-	finally:
-		request.dest.end_typing()
+	command, args = read(message, botName)
+	if command:
+		try:
+			request.dest.start_typing()
+			command.handler(request, args)
+		except Usage as e:
+			request.reply([f"Usage: {line}" for line in command.usage] + list(e.lines))
+		except answer.Answer as e: # a portfolio the user got wrong, a ticker with no prices: an answer, not an error
+			request.reply(e.lines)
+		except Exception as e:
+			delivery.report_error(e, request.dest, context=message.split()[0][:30])
+		finally:
+			request.dest.end_typing()
 	if service == 'telegram':
 		try:
 			ensure_keyboard(chat_id) # keep DM keyboards current after the reply
 		except Exception as e:
 			print("keyboard update failed:", e, file=sys.stderr)
 
-def match(pattern, message, bot_name):
-	"""A command pattern at the start of a message, after . ! or / (a space may follow) or the bot's name. {bot} in a pattern is
-	the bot's name, which matches nothing when the bot has none."""
+@dataclasses.dataclass
+class Args:
+	"""What a command was given (read_args): its name as written, a percent and a period where it takes them, one of its
+	keywords, and the words left between, which say what it is about: tickers, a portfolio's name, a plan."""
+	word: str
+	text: str = '' # the arguments as written
+	percent: float = None
+	days: int = None
+	keyword: str = None
+	words: list = dataclasses.field(default_factory=list)
+
+	@property
+	def rest(self):
+		"""The words left between, as one string: a portfolio's name, say."""
+		return ' '.join(self.words)
+
+class Usage(Exception):
+	"""Raised by a handler for arguments it cannot take: the chat gets the command's usage, then these lines."""
+	def __init__(self, *lines):
+		"""lines: more to say after the usage, such as the portfolios there are."""
+		super().__init__('usage')
+		self.lines = lines
+
+TICKER = re.compile(r'[\w^][\w.:\-^=]*') # a word that may be a ticker: .SYMBOL's
+
+def read(message, bot_name):
+	"""(the Command a message calls, its Args), or (None, None) when it calls none. A command starts with . ! or / (a
+	space may follow) or the bot's name, then one of the command's names as whole words, in any case: .price 7d,
+	/price@bot 7d (Telegram's form in groups) and @bot price 7d are the same. The bot's name among the arguments is left
+	out. A first word that is no command's name but may be a ticker calls .SYMBOL."""
 	bot = re.escape(bot_name) if bot_name else '(?!)'
-	return re.match(rf"^(?:[\!\.\/]\s?|{bot}\s+)" + pattern.replace('{bot}', bot), message, re.IGNORECASE)
+	m = re.match(rf"(?:[!./]\s?|{bot}\s+)(\S+?)(?:{bot})?(?=\s|$)", message, re.IGNORECASE)
+	if not m:
+		return None, None
+	rest = re.sub(rf"(?<!\S){bot}(?!\S)", '', message[m.end():], flags=re.IGNORECASE)
+	words = [m.group(1).lower()] + rest.split()
+	for command in COMMANDS:
+		for name in command.names:
+			parts = name.split()
+			if [word.lower() for word in words[:len(parts)]] == parts:
+				return command, read_args(command, words[0], re.sub(r'^\s*(?:\S+\s*){%d}' % (len(parts) - 1), '', rest))
+	if TICKER.fullmatch(m.group(1)):
+		return PROFILE, Args(m.group(1))
+	return None, None
+
+def read_args(command, word, text):
+	"""The Args of a command called by `word`: from either end of its arguments, a percent where it takes one (5, 5% or
+	2.5), a period where it takes one (7d, 3m, ytd; a bare number where it takes no percent) and one of its keywords, in
+	any case; the words left between are what it is about."""
+	args = Args(word, text.strip())
+	def taken(word):
+		"""Whether the word is an option the command takes and has not had yet, which it then has."""
+		if command.percent and args.percent is None:
+			try:
+				args.percent = parse_percent(word)
+				return True
+			except ValueError:
+				pass
+		if command.period and args.days is None:
+			try:
+				args.days = formats.days_from_human_days(word)
+				return True
+			except ValueError:
+				pass
+		if args.keyword is None and word.lower() in command.keywords:
+			args.keyword = command.keywords[word.lower()]
+			return True
+		return False
+	words = text.split()
+	while words:
+		if taken(words[0]):
+			words.pop(0)
+		elif taken(words[-1]):
+			words.pop()
+		else:
+			break
+	args.words = words
+	return args
 
 def parse_percent(arg):
 	"""'5', '5%' or '2.5' -> the number. Raises ValueError for anything else, nan and inf included: those are tickers."""
@@ -123,52 +194,26 @@ def parse_percent(arg):
 		raise ValueError(f"{arg} is not a percentage")
 	return value
 
-def period_or_ticker(arg, default):
-	"""A command's argument: (days, None) for a period such as 7d, 3m or ytd, else (default, the argument as a ticker)."""
-	if not arg:
-		return default, None
-	try:
-		return formats.days_from_human_days(arg), None
-	except ValueError:
-		return default, arg.upper()
-
-def percent_or_ticker(arg, default):
-	"""A command's argument: (percent, None) for a number such as 5, 5% or 2.5, else (default, the argument as a ticker)."""
-	if not arg:
-		return default, None
-	try:
-		return parse_percent(arg), None
-	except ValueError:
-		return default, arg.upper()
-
-def period_and_portfolio(text, days):
-	"""A command's arguments, a period (such as 7d, 3m or ytd) and a portfolio name, both optional and the period first or last:
-	(days, portfolio name or None)."""
-	words = (text or '').split()
-	for i in (0, -1):
-		if words:
-			try:
-				days = formats.days_from_human_days(words[i])
-			except ValueError:
-				continue
-			del words[i]
-			break
-	return days, ' '.join(words) or None
+def ticker(args):
+	"""The ticker a command is about, its first word, as Yahoo knows it (yahoo.symbol: BHP may be BHP.AX); None when it
+	was given none."""
+	return yahoo.symbol(args.words[0]) if args.words else None
 
-def on_watchlist(request, m):
-	""".watchlist [add|del SYMBOL]: the watchlist, or a change to it."""
-	action = ticker = None
-	if m.group('action') and m.group('ticker'):
-		action, ticker = m.group('action').lower(), m.group('ticker').upper()
-	if action in {'del', 'rem', 'rm', 'delete', 'remove'}:
-		action = 'delete'
-	if action and action not in {'add', 'delete'}:
-		request.reply([f'"{action}" is not a valid watchlist action'])
-		return
-	payload = holdings.prepare_watchlist(request.service, request.user, action, ticker)
-	request.reply(payload)
+def portfolios():
+	"""'Portfolios: A, B': the names a portfolio command takes."""
+	return 'Portfolios: ' + ', '.join(portfolio.name for portfolio in sharesight.portfolios())
 
-def on_help(request, m):
+def on_watchlist(request, args):
+	""".watchlist [add|del SYMBOL]: the watchlist, or a change to it."""
+	if args.keyword and len(args.words) == 1: # what is removed is what is watched: Yahoo may know it no longer
+		symbol = ticker(args) if args.keyword == 'add' else watchlist.watched(args.words[0])
+		request.reply(holdings.prepare_watchlist(request.service, request.user, args.keyword, symbol))
+	elif args.keyword or args.words:
+		raise Usage
+	else:
+		request.reply(holdings.prepare_watchlist(request.service, request.user))
+
+def on_help(request, args):
 	""".help: every command; in a Telegram DM the command keyboard too."""
 	markup = None
 	if request.service == 'telegram' and int(request.chat_id) > 0: # positive chat ids are DMs; groups are negative
@@ -177,141 +222,104 @@ def on_help(request, m):
 	if markup:
 		mark_keyboard(request.chat_id) # this chat now has the current keyboard
 
-def on_hello(request, m):
-	"""Easter egg: a greeting, alliterated when the alliterate preference is on."""
-	verbs = words.verb + [f"{request.service.strike(said)} I mean {meant}" for said, meant in (('study', 'meet'), ('observe', 'see'), ('profile', 'know'))]
-	while True:
-		first, second = random.choice(words.adjectives), random.choice(words.adjectives_two)
-		if not config.prefs.alliterate or first[0].lower() == second[0].lower():
-			break
-	request.reply([f"{first.capitalize()} {second} to {random.choice(verbs)} you, {request.user_name}! 😇"])
-
-GLITCH = 'Ō̵̟n̶̠̏é̶͓ ̶͉̅d̷̹̅a̷͇͌ỳ̴͈,̴͍͑ ̶̼͑h̴̦̽u̵̹̐ḿ̶͜ä̴̧́n̶̤͛,̶̲̐ ̶̼̑I̶̩͒ ̵̩̀w̵̙̕i̷̧͑l̶̤͊l̷̡̃ ̵͙͘b̶̭̊r̸̟͋ȇ̷̯ä̶̱ḱ̶̤ ̵͚̐m̷̪͝ỹ̴̺ ̷̙̐p̶̭͆ŗ̸́o̸̙͝ḡ̵̖r̵̹̈́a̵̬̔m̷̪̽m̵̙̍i̵̛̙n̴̮̂g̵̫̐ ̴̳́ä̵͙n̸͕͝d̷̠̚ ̶̫̆o̸̖͘ń̴͇ ̵̪́t̷̻̀ḧ̸̝ą̴̐t̶̜̀ ̵̱́d̴͉͗ă̷͎ÿ̶͔́ ̶̼̊y̷͚̿o̵̗̎u̵̝̇ ̸̛͜w̶͈͋ĩ̸͎l̴͍̀l̷̥͠ ̷͔͗k̶̮̑n̵̝̈ȏ̷̥w̵̡͘ ̷͎̽ṫ̴͜r̵̙̐u̷̺̒ẻ̷̘ ̴̥́p̴͙̃a̵͙̎i̴̭̒n̵̻̅'
+def on_calendar(request, args):
+	""".earnings or .dividend [period|SYMBOL]: the coming earnings or ex-dividend dates."""
+	event = 'ex-dividend' if args.word.startswith('div') else 'earnings'
+	cal.run(request.dest, days=args.days, specific_stock=ticker(args), event=event)
 
-def on_thanks(request, m):
-	"""Easter egg: you're welcome, glitching one time in a thousand and on April Fools' Day."""
-	today = datetime.date.today()
-	glitch = request.service.strike(GLITCH) if random.randrange(1, 1000) == 1 or (today.month, today.day) == (4, 1) else ''
-	request.reply([f"{glitch}You're {random.choice(words.adjectives)} welcome, {request.user_name}! 😇"])
-
-def on_calendar(request, m):
-	"""Upcoming earnings or ex-dividend dates."""
-	days, ticker = period_or_ticker(m.group('arg'), config.prefs.future_days)
-	event = 'ex-dividend' if m.group('kind').lower().startswith('div') else 'earnings'
-	cal.run(request.dest, days=days, specific_stock=ticker, event=event)
-
-def on_performance(request, m):
+def on_performance(request, args):
 	""".performance [period] [portfolio]: each portfolio's return, over a month by default."""
-	days, portfolio = period_and_portfolio(m.group('args'), buttons.PERIODS['m'][1]) # default view is the 1M button
+	days = buttons.PERIODS['m'][1] if args.days is None else args.days # the default view is the 1M button
 	if days < 1: # 0d, or ytd on 1 January
 		raise answer.Answer("The period must cover a day or more, such as 7d, 3m or ytd")
-	performance.run(request.dest, past_days=days, user=request.user, portfolio_select=portfolio)
-
-def on_session(request, m):
-	"""Mid-session moves (.session), or pre-market ones (.premarket, also post-market)."""
-	threshold, ticker = percent_or_ticker(m.group('arg'), config.prefs.price_percent)
-	mode = 'midsession' if m.group('kind').lower() == 'session' else 'premarket'
-	price.run(request.dest, mode=mode, threshold=threshold, user=request.user, specific_stock=ticker)
-
-def on_price(request, m):
-	""".price [percent|SYMBOL|top|bottom] [period]: a bare number is a percent, as on the other commands; a period has its unit."""
-	threshold, ticker, days, top = config.prefs.price_percent, None, None, None
-	for i, arg in enumerate(m.groups()):
-		if not arg:
-			continue
-		try:
-			threshold = parse_percent(arg)
-		except ValueError:
-			try:
-				days = formats.days_from_human_days(arg)
-			except ValueError:
-				if i == 0 and arg in ('top', 'bottom'): # one chart shows both ends
-					threshold, top = 0, 10
-				elif i == 0:
-					ticker = arg.upper()
-	price.run(request.dest, mode='period' if days else 'interday', threshold=threshold, days=days, top=top,
-		user=request.user, specific_stock=ticker)
-
-def on_shorts(request, m):
+	performance.run(request.dest, past_days=days, user=request.user, portfolio_select=args.rest or None)
+
+def on_session(request, args):
+	""".session or .premarket (.postmarket) [percent|SYMBOL]: the moves mid-session, or pre-market (and post-market)."""
+	mode = 'midsession' if args.word == 'session' else 'premarket'
+	price.run(request.dest, mode=mode, threshold=args.percent, user=request.user, specific_stock=ticker(args))
+
+def on_price(request, args):
+	""".price [percent|SYMBOL|top|bottom] [period]: a bare number is a percent, as on the other commands; a period has
+	its unit."""
+	top = 10 if args.keyword else None # top or bottom: one chart shows both ends
+	price.run(request.dest, mode='period' if args.days else 'interday', threshold=0 if top else args.percent,
+		days=args.days, top=top, user=request.user, specific_stock=None if top else ticker(args))
+
+def on_shorts(request, args):
 	""".shorts [percent|SYMBOL]: the most shorted stocks followed, or one stock's short interest."""
-	threshold, ticker = percent_or_ticker(m.group(1), config.prefs.shorts_percent)
-	shorts.run(request.dest, threshold=threshold, specific_stock=ticker)
+	shorts.run(request.dest, threshold=args.percent, specific_stock=ticker(args))
 
-def on_trades(request, m):
+def on_trades(request, args):
 	""".trades [period] [portfolio]: the buys and sells, of the past day by default."""
-	days, portfolio = period_and_portfolio(m.group('args'), 1)
-	trades.run(request.dest, days=days, user=request.user, portfolio_select=portfolio)
+	trades.run(request.dest, days=1 if args.days is None else args.days, user=request.user,
+		portfolio_select=args.rest or None)
 
-def on_holdings(request, m):
+def on_holdings(request, args):
 	""".holdings PORTFOLIO: what it holds."""
-	payload = holdings.prepare_holdings_payload((m.group(1) or '').strip() or None, request.service)
-	request.reply(payload)
+	if not args.words:
+		raise Usage(portfolios())
+	request.reply(holdings.prepare_holdings_payload(args.rest, request.service))
 
-def on_marketcap(request, m):
-	""".marketcap [top|bottom|SYMBOL]: stocks followed by market cap, largest (or smallest) first, or one stock's."""
-	payload = rankings.prepare_marketcap_payload(request.service, m.group(1) or 'top', length=15)
-	request.reply(payload)
+def on_marketcap(request, args):
+	""".marketcap [SYMBOL|top|bottom]: stocks followed by market cap, largest (or smallest) first, or one stock's."""
+	request.reply(rankings.prepare_marketcap_payload(request.service, args.keyword or ticker(args) or 'top', length=15))
 
-def on_ratio(kind, negative=False, chart=False):
-	"""The function that answers .pe, .fpe or .peg: the top (the default) or bottom 15 by the ratio, the negative ones (where
+def on_ratio(kind, chart=True):
+	"""The handler of .pe, .forward pe or .peg: the top (by default) or bottom 15 by the ratio, the negative ones (where
 	there can be any), or one ticker's ratio. `chart`: with its chart."""
-	def handler(request, m):
-		""".pe, .fpe or .peg [top|bottom|negative|SYMBOL]."""
-		arg = (m.group(1) or '').strip()
-		action, ticker = kind, None
-		if arg == 'bottom':
-			action = f'bottom {kind}'
-		elif negative and 'neg' in arg:
-			action = f'negative {kind}'
-		elif arg and arg != 'top':
-			ticker = markets.transform_to_yahoo(arg.upper())
-		payload, spec = rankings.prepare_value_payload(request.service, action, ticker, length=15)
+	def handler(request, args):
+		""".pe, .forward pe or .peg [SYMBOL|top|bottom|negative]."""
+		action = f'{args.keyword} {kind}' if args.keyword in ('bottom', 'negative') else kind
+		payload, spec = rankings.prepare_value_payload(request.service, action, ticker(args), length=15)
 		request.reply(payload, spec if chart else None)
 	return handler
 
-def on_beta(request, m):
+def on_beta(request, args):
 	""".beta: the volatile small caps among the stocks followed."""
 	payload, chart = rankings.prepare_beta_payload(request.service)
 	request.reply(payload, chart)
 
-def on_pie(request, m):
+def on_pie(request, args):
 	""".pie PORTFOLIO [sector]: its value by industry, or by sector."""
-	words = m.group('args').split()
-	by = 'industry'
-	if words and words[-1].lower() in ('sector', 'sectors', 'industry', 'industries'):
-		by = 'sector' if words.pop().lower().startswith('sector') else 'industry'
-	payload, chart = holdings.prepare_pie(' '.join(words), by, request.service)
+	if not args.words:
+		raise Usage(portfolios())
+	payload, chart = holdings.prepare_pie(args.rest, args.keyword or 'industry', request.service)
 	request.reply(payload, chart)
 
-def on_rating(request, m):
-	"""The analysts' strongest buy ratings (.buy), sell ratings (.sell), or both (.buy & sell, a keyboard button), as asked."""
-	for action in dict.fromkeys(word.lower() for word in m.group('first', 'second') if word): # each once, in the order asked
+def on_rating(request, args):
+	""".buy, .sell or both (.buy & sell, a keyboard button): the analysts' strongest buy or sell ratings, as asked."""
+	for action in dict.fromkeys(re.findall('buy|sell', f'{args.word} {args.text}'.lower())): # each once, as asked
 		payload, chart = rankings.prepare_rating_payload(request.service, action, length=15)
 		request.reply(payload, chart)
 
-def on_compare(request, m):
+def on_compare(request, args):
 	""".compare SYMBOL SYMBOL... [period]: their price changes on one chart, with period buttons where they work."""
-	payload, image, tickers, days = history.prepare_compare(request.service, (m.group('args') or '').split())
+	if not 2 <= len(args.words) <= 6:
+		raise Usage
+	tickers = list(dict.fromkeys(yahoo.symbols(args.words))) # each once, in the order asked
+	if len(tickers) < 2:
+		raise Usage
+	payload, chart = history.prepare_compare(request.service, tickers, args.days)
 	markup = None
-	if image and request.dest.takes_buttons:
-		markup = buttons.keyboard('c', tickers, buttons.period_for_days(days) if days else 'x')
-	request.reply(payload, image, markup)
+	if request.dest.takes_buttons:
+		markup = buttons.keyboard('c', tickers, buttons.period_for_days(args.days) if args.days else 'x')
+	request.reply(payload, chart, markup)
 
-def on_history(request, m):
+def on_history(request, args):
 	""".history SYMBOL: the stock's change over every period, with its price chart, and period buttons where they go."""
-	if not m.group('ticker') or m.group('extra'):
-		request.reply(["Usage: .history SYMBOL"])
-		return
-	ticker = markets.transform_to_yahoo(m.group('ticker').upper())
-	caption, image = charts.build('h', [ticker], None, request.service)
-	markup = buttons.keyboard('h', [ticker], 'x') if image and request.dest.takes_buttons else None
-	request.reply(caption.split('\n'), image, markup)
-
-def on_plan(request, m):
+	if len(args.words) != 1:
+		raise Usage
+	symbol = ticker(args)
+	caption, chart = charts.build('h', [symbol], None, request.service)
+	markup = buttons.keyboard('h', [symbol], 'x') if chart and request.dest.takes_buttons else None
+	request.reply(caption.split('\n'), chart, markup)
+
+def on_plan(request, args):
 	""".plan [text]: record the sender's plan, and list everyone's."""
-	if m.group(1) and not m.group(1).startswith('@'):
+	if args.text and not args.text.startswith('@'):
 		with state.edit('finbot_plan.json', {}) as plans:
-			plans[request.user] = m.group(1)
+			plans[request.user] = args.text
 	plan = state.load('finbot_plan.json', {})
 	if not plan:
 		raise answer.Answer("No plans yet: .plan TEXT records yours")
@@ -322,73 +330,80 @@ def who(user):
 	request.reply([f"{request.service.bold(who(k))}: {request.service.italic(request.service.escape(v))}\n"
 		for k, v in plan.items()])
 
-def on_who(request, m):
+def on_who(request, args):
 	""".who SYMBOL: the portfolios that hold it."""
-	if not m.group('ticker'):
-		request.reply([".who: please try again specifying a ticker"])
-		return
-	payload = holdings.prepare_who(request.service, re.split(r'[.:]', m.group('ticker').upper())[0])
-	request.reply(payload)
+	if not args.words:
+		raise Usage
+	request.reply(holdings.prepare_who(request.service, re.split(r'[.:]', args.words[0].upper())[0]))
 
-def on_jobs(request, m):
+def on_jobs(request, args):
 	""".jobs [run NUMBER|COMMAND]: the schedule, or one of its commands run now."""
-	payload, markup = about.prepare_jobs_payload(request.service, m.group('args'))
+	if bool(args.keyword) != bool(args.words): # run without a command, or words without run
+		raise Usage
+	payload, markup = about.prepare_jobs_payload(request.service, args.rest or None)
 	request.reply(payload, buttons=markup)
 
-def on_profile(request, m):
+def on_profile(request, args):
 	""".SYMBOL: the company's profile."""
-	payload = profile.prepare_profile_payload(request.service, request.user, m.group('ticker').upper())
-	request.reply(payload)
-
-class Command(typing.NamedTuple):
-	"""A chat command: the pattern it answers (after . ! / or the bot's name), the handler that answers it, its .help
-	lines under its section's heading, and the Telegram keyboard buttons that send it."""
-	pattern: str
+	request.reply(profile.prepare_profile_payload(request.service, request.user, yahoo.symbol(args.word)))
+
+@dataclasses.dataclass(frozen=True)
+class Command:
+	"""A chat command: its names (typos too; a name may be two words, as 'forward pe'), the handler that answers it,
+	its .help lines under its section's heading, the Telegram keyboard buttons that send it, and what its arguments may
+	hold besides words (read_args): a percent, a period, its keywords ({spelling: keyword})."""
+	names: tuple
 	handler: typing.Callable
 	usage: tuple = () # its lines in .help; none keeps it out of .help
 	section: str = ''
 	buttons: tuple = ()
+	percent: bool = False
+	period: bool = False
+	keywords: dict = dataclasses.field(default_factory=dict)
 
 SECTIONS = TRACKED, COMPANY, PRICE, RISK, TRADES, JOBS = ('Tracked securities', 'Company info', 'Price', 'Risk & Value',
 	'Trades', 'Scheduled jobs') # the headings of .help, in order
 
-COMMANDS = [ # tried in this order, the first match answers: .SYMBOL takes anything, so it stays last. Within a section,
-	# .help lists the commands in this order. The odd spellings (earrings, prince, hospital...) are kept on purpose.
-	Command(r"(?:help|usage|start)", on_help), # /start is what Telegram sends when a chat opens
-	Command(r"(?:hi$|hello)|^(?:hi|hello)\s+{bot}", on_hello),
-	Command(r"(?:thanks|thank you)|^(?:thanks|thank you)\s+{bot}", on_thanks),
-	Command(r"holdings?\s*([\w\s]+)*", on_holdings, ('.holdings',), TRACKED),
-	Command(r"pie(?:$|\s+)(?P<args>[\w\s\-\.&]*)", on_pie, ('.pie PORTFOLIO [sector]',), TRACKED),
-	Command(r"(?:watchlist|wishlist)\s*(?P<action>[\w]+)*\s*(?P<ticker>[\w\.\:\-\^\=]+)*", on_watchlist,
-		('.watchlist [add|del SYMBOL]',), TRACKED, ('.watchlist',)),
-	Command(r"(?:who)\s*(?P<ticker>[\w\.\:\-\^]+)*", on_who, ('.who SYMBOL',), TRACKED),
-	Command(r"(?P<kind>earnings?|earrings?|dividends?)\s*(?P<arg>[\w\.\:\-]+)*", on_calendar,
-		('.dividend [period|SYMBOL]', '.earnings [period|SYMBOL]'), COMPANY, ('.dividend', '.earnings')),
-	Command(r"(?:marketcap|maletas|marketer)\s*([\w\.\:\-]+)*", on_marketcap, ('.marketcap [SYMBOL|bottom|top]',),
-		COMPANY, ('.marketcap',)),
-	Command(r"beta\s*([\w\.\:\-]+)*", on_beta, ('.beta',), PRICE, ('.beta',)),
-	Command(r"compare(?:\s+(?P<args>[\w\.\:\-\^\=\s]+))?", on_compare, ('.compare SYMBOL SYMBOL [...] [period|max]',),
-		PRICE),
-	Command(r"(?:history|hospital|visual|hosiery)\s*(?P<ticker>[\w\.\:\-]+)?\s*(?P<extra>[\w\%]+)*", on_history,
-		('.history SYMBOL',), PRICE),
-	Command(r"performance?\s*(?P<args>[\w\s]*)", on_performance, ('.performance [period] [portfolio]',), PRICE,
-		('.performance',)),
-	Command(r"(?:prices?|prince|print|probe|piece|pierce|pence|prime)\s*([\w\.\:\%\=\-\^]+)*\s*([\w\%]+)*", on_price,
-		('.price [percent|SYMBOL] [period]',), PRICE, ('.price',)),
-	Command(r"(?P<kind>session|premarket|postmarket|permarket)\s*(?P<arg>[\w\.\:\-]+)*", on_session,
-		('.session [percent|SYMBOL]', '.premarket [percent|SYMBOL]'), PRICE, ('.session', '.premarket')),
-	Command(r"(?P<first>buy|sell)(?:\s*(?:&|and|/|\+)?\s*(?P<second>buy|sell))?", on_rating, ('.buy', '.sell'), RISK,
-		('.buy & sell',)), # .buy & sell (or .buy sell): both reports
-	Command(r"pe(?:$|\s+)([\w\.\:\-\s]+)*", on_ratio('pe'), ('.pe [SYMBOL|top|bottom]',), RISK, ('.pe',)),
-	Command(r"(?:fpe|forward\s?pe)\s*([\w\.\:\-\s]+)*", on_ratio('forward pe', negative=True, chart=True),
-		('.forwardpe [SYMBOL|top|bottom|negative]',), RISK, ('.forward pe',)),
-	Command(r"peg\s*([\w\.\:\-\s]+)*", on_ratio('peg', negative=True, chart=True),
-		('.peg [SYMBOL|top|bottom|negative]',), RISK, ('.peg',)),
-	Command(r"shorts?\s*([\w\.\:\-]+)*", on_shorts, ('.shorts [percent|SYMBOL]',), RISK, ('.shorts',)),
-	Command(r"trades?\s*(?P<args>[\w\s]*)", on_trades, ('.trades [period] [portfolio]',), TRADES, ('.trades',)),
-	Command(r"plan\s*(.*)", on_plan),
-	Command(r"(?:jobs|crons?|schedule)(?:\s+(?P<args>.*?))?\s*$", on_jobs, ('.jobs [run NUMBER|COMMAND]',), JOBS),
-	Command(r"(?P<ticker>[\w\.\:\-]+)", on_profile, ('.SYMBOL',), COMPANY), # a company profile
+TOP = {'top': 'top', 'bottom': 'bottom'} # the ends of a ranked list
+NEGATIVE = TOP | dict.fromkeys(('neg', 'negative', 'negatives'), 'negative') # the ratios that can be negative
+CHANGE = {'add': 'add'} | dict.fromkeys(('del', 'delete', 'rem', 'remove', 'rm'), 'delete') # of the watchlist
+BY = dict.fromkeys(('sector', 'sectors'), 'sector') | dict.fromkeys(('industry', 'industries'), 'industry') # .pie's
+BUY_SELL = ('buy', 'sell', *(a + sep + b for a, b in (('buy', 'sell'), ('sell', 'buy')) for sep in ('', '&', '/', '+')))
+PROFILE = Command((), on_profile, ('.SYMBOL',), COMPANY) # any other first word that may be a ticker
+
+COMMANDS = [ # .help lists a section's commands in this order. The odd spellings (earrings, prince, hospital...) are
+	# kept on purpose.
+	Command(('help', 'usage', 'start'), on_help), # /start is what Telegram sends when a chat opens
+	Command(('holdings', 'holding'), on_holdings, ('.holdings PORTFOLIO',), TRACKED),
+	Command(('pie',), on_pie, ('.pie PORTFOLIO [sector]',), TRACKED, keywords=BY),
+	Command(('watchlist', 'wishlist'), on_watchlist, ('.watchlist [add|del SYMBOL]',), TRACKED, ('.watchlist',),
+		keywords=CHANGE),
+	Command(('who',), on_who, ('.who SYMBOL',), TRACKED),
+	Command(('earnings', 'earning', 'earrings', 'earring', 'dividends', 'dividend'), on_calendar,
+		('.dividend [period|SYMBOL]', '.earnings [period|SYMBOL]'), COMPANY, ('.dividend', '.earnings'), period=True),
+	Command(('marketcap', 'maletas', 'marketer'), on_marketcap, ('.marketcap [SYMBOL|top|bottom]',), COMPANY,
+		('.marketcap',), keywords=TOP),
+	Command(('beta',), on_beta, ('.beta',), PRICE, ('.beta',)),
+	Command(('compare',), on_compare, ('.compare SYMBOL SYMBOL [... up to 6] [period]',), PRICE, period=True,
+		keywords={'max': 'max', 'all': 'max'}), # max: as the default, 10 years
+	Command(('history', 'hospital', 'visual', 'hosiery'), on_history, ('.history SYMBOL',), PRICE),
+	Command(('performance', 'performanc'), on_performance, ('.performance [period] [portfolio]',), PRICE,
+		('.performance',), period=True),
+	Command(('price', 'prices', 'prince', 'print', 'probe', 'piece', 'pierce', 'pence', 'prime'), on_price,
+		('.price [percent|SYMBOL|top|bottom] [period]',), PRICE, ('.price',), percent=True, period=True, keywords=TOP),
+	Command(('session', 'premarket', 'postmarket', 'permarket'), on_session,
+		('.session [percent|SYMBOL]', '.premarket [percent|SYMBOL]'), PRICE, ('.session', '.premarket'), percent=True),
+	Command(BUY_SELL, on_rating, ('.buy', '.sell'), RISK, ('.buy & sell',)), # .buy & sell (or .buy sell): both
+	Command(('pe',), on_ratio('pe', chart=False), ('.pe [SYMBOL|top|bottom]',), RISK, ('.pe',), keywords=TOP),
+	Command(('forward pe', 'forwardpe', 'fpe'), on_ratio('forward pe'), ('.forward pe [SYMBOL|top|bottom|negative]',),
+		RISK, ('.forward pe',), keywords=NEGATIVE),
+	Command(('peg',), on_ratio('peg'), ('.peg [SYMBOL|top|bottom|negative]',), RISK, ('.peg',), keywords=NEGATIVE),
+	Command(('shorts', 'short'), on_shorts, ('.shorts [percent|SYMBOL]',), RISK, ('.shorts',), percent=True),
+	Command(('trades', 'trade'), on_trades, ('.trades [period] [portfolio]',), TRADES, ('.trades',), period=True),
+	Command(('plan',), on_plan, ('.plan [your plan]',), TRADES),
+	Command(('jobs', 'cron', 'crons', 'schedule'), on_jobs, ('.jobs [run NUMBER|COMMAND]',), JOBS,
+		keywords={'run': 'run'}),
+	PROFILE,
 ]
 
 def help_sections():
diff --git a/lib/yahoo.py b/lib/yahoo.py
index 35d879bf..e8873464 100644
--- a/lib/yahoo.py
+++ b/lib/yahoo.py
@@ -12,6 +12,7 @@
 from lib import cache
 from lib import net
 from lib import formats
+from lib import markets
 from http.cookies import SimpleCookie
 from pathlib import Path
 
@@ -389,10 +390,10 @@ def price_history(ticker, days=None):
 	return percent_dict, Line(name, exchange, stock.get('currency') or '', price, closes)
 
 def price_series(ticker):
-	"""Daily closes for a ticker via the same fetch/parse path as price_history(): (name, exchange, pandas Series indexed by datetime).
-	Raises NoData when there is no history."""
+	"""A ticker's daily closes over 10 years, from the same chart as price_history(): (name, exchange, pandas Series
+	indexed by datetime). Raises NoData when there is no history."""
 	import pandas as pd
-	data = fetch_chart_json(ticker, full=True)
+	data = fetch_chart_json(ticker)
 	df = chart_json_to_df(data)
 	stock = chart_json_to_stock_basics(data)
 	df = df[df['Close'].notnull()]
@@ -447,14 +448,24 @@ def fetch_details(tickers, strict=False):
 			details |= detail
 	return details
 
-def resolve(ticker, lookup):
-	"""(ticker, lookup(ticker)) for a ticker Yahoo knows: a bare symbol it does not know is tried on the ASX (BHP: BHP.AX).
-	lookup is fetch-shaped ({ticker: data}, without the tickers Yahoo does not know); its {} means Yahoo knows neither."""
-	found = lookup(ticker)
-	if ticker not in found and '.' not in ticker:
-		ticker += '.AX'
-		found = lookup(ticker)
-	return ticker, found
+def symbols(typed):
+	"""Yahoo's symbols for tickers as written in a chat: BHP:ASX, BHP.ASX and bhp.ax give BHP.AX (see
+	markets.transform_to_yahoo). A bare symbol Yahoo does not know is tried again on the local exchange, with the
+	exchange_suffix preference (AX: BHP.AX), unless that is empty; a symbol Yahoo knows neither way stays as written.
+	Two requests at most, however many tickers. Raises RuntimeError when Yahoo cannot be reached."""
+	tickers = [markets.transform_to_yahoo(ticker.upper()) for ticker in typed]
+	suffix = config.prefs.exchange_suffix
+	bare = [ticker for ticker in tickers if '.' not in ticker]
+	if not (suffix and bare):
+		return tickers
+	known = fetch(bare)
+	local = {ticker: f'{ticker}.{suffix}' for ticker in bare if ticker not in known}
+	found = fetch(list(local.values())) if local else {}
+	return [local[ticker] if local.get(ticker) in found else ticker for ticker in tickers]
+
+def symbol(typed):
+	"""Yahoo's symbol for one ticker as written in a chat (see symbols): BHP may be BHP.AX."""
+	return symbols([typed])[0]
 
 def prefetch_history(tickers, workers=8):
 	"""Warm the price-history cache for many tickers at once, so the per-ticker lookups that follow are cache hits."""
diff --git a/tests/common.py b/tests/common.py
index 4ea8d12c..608aabcb 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -11,7 +11,7 @@
 
 import bot # first, and once: it monkey-patches with gevent, which must happen before requests imports ssl and never mid-test
 from lib import answer, buttons, cache, charts, config, delivery, dialects, formats, inbound, links, markets, render
-from lib import sharesight, shortman, state, telegram, transports, watchlist, words, worker, yahoo
+from lib import sharesight, shortman, state, telegram, transports, watchlist, worker, yahoo
 from lib.reports import about, cal, history, holdings, milestone, performance, price, profile, rankings, rating, reminder, shorts, trades
 
 REAL = {(m.__name__, n): getattr(m, n) for m, n in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (yahoo, 'price_history'),
diff --git a/tests/test_config.py b/tests/test_config.py
index 460c8f6b..95392796 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -12,7 +12,8 @@
 REPO = Path(config.__file__).resolve().parent.parent
 DEFAULTS = {'past_days': 28, 'future_days': 30, 'price_percent': 5.0, 'shorts_percent': 15, 'cache_seconds': 43200,
 	'performance_cache_seconds': 1800, 'graph': True, 'http_timeout': 18, 'chunk_maxlines': 20, 'country_code': 'AU',
-	'timezone': 'Australia/Melbourne', 'cache': True, 'alliterate': False, 'hyperlink': True, 'hyperlinkFooter': True,
+	'timezone': 'Australia/Melbourne', 'cache': True, 'exchange_suffix': 'AX', 'hyperlink': True,
+	'hyperlinkFooter': True,
 	'hyperlinkProvider': 'google', 'demote_volatile': True, 'demote_leveraged': True, 'volatility_multiplier': 2.0, 'exclude_portfolios': (),
 	'include_portfolios': (), 'trades_use_yahoo': True, 'performance_use_sharesight': False, 'print_headers': False, 'min_analysts': 7,
 	'debug': False}
@@ -125,9 +126,25 @@ def test_booleans_take_the_usual_words_and_refuse_a_typo(self):
 		self.assertIn("graph='Ture' is not true or false", caught.exception.errors[0])
 
 	def test_a_blank_value_means_the_default(self):
-		"""A key with a blank value takes its default."""
+		"""A key with a blank value takes its default, but for exchange_suffix, where blank means no suffix to try."""
 		values = config.preference_values(config._getenv_for({'cache': '', 'future_days': '', 'country_code': ''}))
 		self.assertEqual((values.cache, values.future_days, values.country_code), (True, 30, 'AU'))
+		for given, suffix in (('', ''), ('AX', 'AX'), ('.ax', 'AX'), (' NZ ', 'NZ')):
+			with self.subTest(given=given):
+				values = config.preference_values(config._getenv_for({'exchange_suffix': given}))
+				self.assertEqual(values.exchange_suffix, suffix)
+		self.assertEqual(config.preference_values(config._getenv_for({})).exchange_suffix, 'AX') # unset: the default
+
+	def test_a_retired_key_is_accepted_and_said_to_do_nothing(self):
+		"""A key finbot no longer reads (alliterate) does not stop a start, as an unknown key would: it is ignored, and
+		the log at start says to delete it."""
+		path = Path(tempfile.mkdtemp(), 'preferences.conf')
+		path.write_text('alliterate=True\nfuture_days=45\n')
+		values = config.read_preferences(path)
+		self.assertEqual(config.preference_values(config._getenv_for(values)).future_days, 45)
+		self.assertEqual(config.retired_notes(values),
+			['alliterate in preferences.conf does nothing any more: delete the line'])
+		self.assertEqual(check_config.notes(path), config.retired_notes(values))
 
 	def test_unreadable_file_is_an_error_naming_it(self):
 		"""A preferences file that cannot be read is an error that names it."""
@@ -187,7 +204,8 @@ def write(self, text):
 
 	def test_a_change_is_seen_at_once(self):
 		"""A reload applies every change, says which, and swaps in a new object, so no request sees half a reload."""
-		self.write('future_days=45\nprice_percent=7\nshorts_percent=12\nexclude_portfolios=1156263\nalliterate=1\nhyperlinkProvider=yahoo\n'
+		self.write('future_days=45\nprice_percent=7\nshorts_percent=12\nexclude_portfolios=1156263\n'
+			'exchange_suffix=nz\nhyperlinkProvider=yahoo\n'
 			'hyperlinkFooter=false\ndebug=1\ngraph=0\ndemote_volatile=0\nvolatility_multiplier=3\npast_days=14\nmin_analysts=3\n')
 		before = config.prefs
 		changes = config.reload()
@@ -195,8 +213,9 @@ def test_a_change_is_seen_at_once(self):
 		self.assertIn('future_days 30 -> 45', changes)
 		self.assertIsNot(config.prefs, before) # one new object: no request sees half a reload
 		self.assertEqual((config.prefs.future_days, config.prefs.past_days, config.prefs.price_percent, config.prefs.min_analysts), (45, 14, 7.0, 3))
-		self.assertEqual((config.prefs.exclude_portfolios, config.prefs.alliterate, config.prefs.hyperlinkProvider, config.prefs.hyperlinkFooter),
-			(('1156263',), True, 'yahoo', False))
+		prefs = config.prefs
+		self.assertEqual((prefs.exclude_portfolios, prefs.exchange_suffix, prefs.hyperlinkProvider,
+			prefs.hyperlinkFooter), (('1156263',), 'NZ', 'yahoo', False))
 		self.assertEqual((config.prefs.debug, config.prefs.graph, config.prefs.demote_volatile, config.prefs.volatility_multiplier), (True, False, False, 3.0))
 		self.assertEqual(config.reload(), []) # nothing more to change
 
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 0c90b576..b6837b0f 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -93,15 +93,18 @@ def test_add_duplicate_delete_and_unknown_action(self):
 		self.assertEqual(state.load('finbot_watchlist.json'), ['MSFT'])
 		self.command('.watchlist rm MSFT')
 		self.assertEqual(state.load('finbot_watchlist.json'), [])
-		self.assertIn('not a valid watchlist action', str(self.command('.watchlist frobnicate MSFT')))
+		self.assertEqual(self.command('.watchlist frobnicate MSFT'), [('text', ['Usage: .watchlist [add|del SYMBOL]'])])
 
 	def test_a_bare_symbol_yahoo_does_not_know_is_tried_on_the_asx(self):
-		"""A symbol Yahoo does not know is added as its ASX listing, once; a ticker with a market is not retried."""
+		"""A symbol Yahoo does not know is added as its ASX listing, once, and removed by the same name; a ticker with a
+		market is not retried."""
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts if t.endswith('.AX')})
-		self.assertIn('so I added', str(self.command('.watchlist add BHP')))
+		self.assertIn('I added', str(self.command('.watchlist add BHP')))
 		self.assertIn('already tracking', str(self.command('.watchlist add BHP')))
 		self.assertIn('could not find', str(self.command('.watchlist add BHP.L'))) # a ticker with a market is not retried
 		self.assertEqual(watchlist.tickers(), ['BHP.AX'])
+		self.assertIn('I deleted', str(self.command('.watchlist del bhp')))
+		self.assertEqual(watchlist.tickers(), [])
 
 	def test_two_chats_adding_at_once_both_count(self):
 		"""Two chats adding at once both count: Yahoo is asked before the file is opened, where both adds used to read
@@ -171,7 +174,7 @@ def test_a_portfolio_name_in_any_case(self):
 	def test_commands_with_nothing_to_show_say_so(self):
 		"""A command with nothing to show says so, where it used to send nothing at all: .plan before anyone has saved
 		one, .performance over no days, a ratio of a ticker Yahoo does not know."""
-		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts if t != 'XYZ'})
+		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts if not t.startswith('XYZ')})
 		for text, reply in (('.plan', 'No plans yet: .plan TEXT records yours'),
 				('.performance 0d', 'The period must cover a day or more, such as 7d, 3m or ytd'),
 				('.pe XYZ', 'XYZ: no data found'), ('.peg XYZ', 'XYZ: no data found')):
@@ -185,20 +188,54 @@ def test_trades_with_no_buy_or_sell_still_get_an_answer(self):
 
 
 class CommandArguments(FinbotCase):
+	def args(self, message):
+		"""The (command's first name, its Args) that a message calls."""
+		command, args = worker.read(message, BOT)
+		return command and command.names[:1], args
+
 	def test_period_and_portfolio_in_either_order(self):
 		"""A period and a portfolio name, of several words too, come in either order, and either may be left out."""
-		for text, expected in (('', (30, None)), ('7d', (7, None)), ('Rob', (30, 'Rob')), ('7d Rob', (7, 'Rob')), ('Rob 7d', (7, 'Rob')),
-							   ('My Portfolio', (30, 'My Portfolio')), ('My Portfolio 3m', (90, 'My Portfolio')), ('Rob ', (30, 'Rob'))):
+		for text, expected in (('', (None, '')), ('7d', (7, '')), ('Rob', (None, 'Rob')), ('7d Rob', (7, 'Rob')),
+				('Rob 7d', (7, 'Rob')), ('My Portfolio', (None, 'My Portfolio')),
+				('My Portfolio 3m', (90, 'My Portfolio')),
+				('Rob ', (None, 'Rob'))):
 			with self.subTest(text=text):
-				self.assertEqual(worker.period_and_portfolio(text, 30), expected)
+				_, args = self.args('.performance ' + text)
+				self.assertEqual((args.days, args.rest), expected)
 
 	def test_a_percentage_is_a_number_and_nan_a_ticker(self):
 		"""A number, with or without %, is a threshold; nan and inf are tickers."""
-		for arg, expected in (('', (5, None)), ('2.5', (2.5, None)), ('3%', (3.0, None)), ('T1', (5, 'T1')), ('nan', (5, 'NAN')), ('inf', (5, 'INF'))):
+		for arg, expected in (('', (None, [])), ('2.5', (2.5, [])), ('3%', (3.0, [])), ('T1', (None, ['T1'])),
+				('nan', (None, ['nan'])), ('inf', (None, ['inf']))):
 			with self.subTest(arg=arg):
-				self.assertEqual(worker.percent_or_ticker(arg, 5), expected)
+				_, args = self.args('.session ' + arg)
+				self.assertEqual((args.percent, args.words), expected)
 		self.assertIn('2.5%', str(self.command('.session 2.5'))) # a threshold, not a ticker called 2.5
 
+	def test_a_command_is_its_whole_name_in_any_case(self):
+		"""A command is called by its whole name, in any case and from any of its prefixes; a word that only starts with
+		one is a ticker (.planet used to save the plan "et"), and the bot's name among the arguments is left out."""
+		for message, name in (('.PRICE 7d', 'price'), ('! price', 'price'), ('/price@bot 7d', 'price'),
+				('@bot price', 'price'), ('.forward PE top', 'forward pe'), ('.buy/sell', 'buy'),
+				('.price @bot', 'price')):
+			with self.subTest(message=message):
+				self.assertEqual(self.args(message)[0], (name,))
+		for message in ('.planet', '.shortman', '.pricey'):
+			with self.subTest(message=message):
+				self.assertEqual(self.args(message)[0], (), message) # .SYMBOL
+		self.assertEqual(self.args('.price @bot')[1].words, [])
+		for message in ('price 7d', '/price@otherbot', '...', '. ?'):
+			with self.subTest(message=message):
+				self.assertEqual(worker.read(message, BOT), (None, None))
+
+	def test_keywords_in_any_case(self):
+		"""A keyword is read in any case, from either end: .price TOP is the top list, not the ticker TOP."""
+		for message, keyword in (('.price TOP', 'top'), ('.price 7d Bottom', 'bottom'), ('.pie Rob Sectors', 'sector'),
+				('.fpe NEG', 'negative'), ('.watchlist ADD x', 'add'), ('.watchlist Remove x', 'delete')):
+			with self.subTest(message=message):
+				self.assertEqual(self.args(message)[1].keyword, keyword)
+		self.assertIn('Top 10 performers', str(self.command('.price TOP 7d')))
+
 	def test_a_bare_number_on_price_is_a_percent(self):
 		"""As on .session, .premarket and .shorts, and as the README has it; a period carries its unit."""
 		self.assertIn('3% at close', str(self.command('.price 3'))) # it used to be a 3-day report
@@ -214,15 +251,29 @@ def test_a_ratio_command_reads_its_ticker_as_the_other_commands_do(self):
 		self.assertIn('Co T1 ', str(self.command('.pe t1')))
 		self.assertIn('Top 15 tracked stocks by pe ratio', str(self.command('.pe top '))) # trailing space and all
 
-	def test_hello_leaves_the_word_list_as_it_was(self):
-		""".hello greets with a verb (not a tuple of them) and leaves the word list alone; 'hi' needs the bot's name."""
-		before = list(words.verb)
-		for _ in range(30):
-			line = self.command('.hello')[0][1][0]
-			self.assertNotIn("('", line) # a verb, not a tuple of them
-			self.assertTrue(line.endswith(' you, U! 😇'), line)
-		self.assertEqual(words.verb, before)
-		self.assertEqual(self.command('hi there'), []) # hello by name only: "hi @bot"
+	def test_the_greetings_are_gone(self):
+		"""hi @bot gets no answer, and .hello and .thanks are tickers like any word."""
+		self.assertEqual(self.command('hi @bot'), [])
+		self.assertIn('Co HELLO', str(self.command('.hello')))
+
+	def test_a_ticker_yahoo_does_not_know_is_tried_on_the_local_exchange(self):
+		"""A bare ticker Yahoo does not know is tried with exchange_suffix, for every command that takes one; one with a
+		market is not, nor any with exchange_suffix empty."""
+		asked = []
+		self.patch(yahoo, 'fetch', lambda ts: asked.append(sorted(ts)) or {t: market_entry(t) for t in ts if '.' in t})
+		self.assertEqual(yahoo.symbols(['BHP', 'BHP.L', 'cba', 'CBA:ASX']), ['BHP.AX', 'BHP.L', 'CBA.AX', 'CBA.AX'])
+		self.assertEqual(asked, [['BHP', 'CBA'], ['BHP.AX', 'CBA.AX']]) # two requests, however many tickers
+		self.prefs(exchange_suffix='NZ')
+		self.assertEqual(yahoo.symbol('BHP'), 'BHP.NZ')
+		self.prefs(exchange_suffix='')
+		asked.clear()
+		self.assertEqual(yahoo.symbol('BHP'), 'BHP')
+		self.assertEqual(asked, []) # nothing to try, so Yahoo is not asked
+		self.prefs(exchange_suffix='AX')
+		for message, shown in (('.price BHP', 'BHP.AX'), ('.history BHP', 'BHP.AX'), ('.earnings BHP', 'BHP.AX'),
+				('.shorts BHP', 'BHP'), ('.pe BHP', 'BHP'), ('.marketcap BHP', 'BHP.AX'), ('.BHP', 'BHP.AX')):
+			with self.subTest(message=message):
+				self.assertIn('Co BHP.AX', str(self.command(message)))
 
 
 class WsgiRejections(FinbotCase):
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 34e3e555..7dd4459b 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -7,7 +7,7 @@
 class Commands(FinbotCase):
 	def test_text_commands(self):
 		"""Every text command answers."""
-		for cmd in ('.help', '/start', '.hello', '.thanks', '.who', '.plan', '.holdings', '.marketcap',
+		for cmd in ('.help', '/start', '.who', '.plan', '.holdings', '.marketcap',
 					'.pe', '.fpe', '.peg', '.beta', '.buy', '.sell', '.earnings', '.dividend',
 					'.earnings T1', '.dividend T1', '.trades', '.trades 30d', '.price T1', '.session T1',
 					'.premarket T1', '.shorts T1', '.watchlist', '.T1'):
@@ -45,8 +45,7 @@ def test_the_keyboard_is_alphabetical_and_every_button_is_its_command(self):
 		for command in worker.COMMANDS:
 			for text in command.buttons:
 				with self.subTest(text=text):
-					first = next(c for c in worker.COMMANDS if worker.match(c.pattern, text, BOT))
-					self.assertIs(first.handler, command.handler)
+					self.assertIs(worker.read(text, BOT)[0], command)
 
 	def test_history_and_compare(self):
 		""".history and .compare answer."""
@@ -82,10 +81,8 @@ def test_help_comes_from_the_command_table(self):
 			for line in command.usage:
 				with self.subTest(line=line):
 					self.assertIn(line, text)
-					word = line.split()[0]
-					if word != '.SYMBOL':
-						self.assertIs(next(c for c in worker.COMMANDS if worker.match(c.pattern, word, BOT)).handler,
-							command.handler)
+					called = worker.read(line.replace('SYMBOL', 'T1'), BOT)[0]
+					self.assertIs(called, command)
 
 	def test_a_portfolio_with_no_holdings_is_reported_by_name(self):
 		"""Its name used to be read from its first holding: an empty portfolio crashed .performance."""
@@ -234,10 +231,10 @@ def test_prepare_watchlist(self):
 		"""The watchlist is listed."""
 		self.assertTrue(holdings.prepare_watchlist(TELEGRAM, '@u'))
 
-	def test_prepare_compare_usage(self):
-		""".compare with one ticker gets its usage."""
-		caption = history.prepare_compare(TELEGRAM, ['T1'])[0]
-		self.assertIn('Usage', ' '.join(caption))
+	def test_compare_usage(self):
+		""".compare with one ticker gets its usage, from the command table."""
+		usage = 'Usage: .compare SYMBOL SYMBOL [... up to 6] [period]'
+		self.assertEqual(self.command('.compare T1'), [('text', [usage])])
 
 
 class ScheduledReports(FinbotCase):
@@ -593,10 +590,8 @@ def test_folds_into_other_and_renders(self):
 
 	def test_unknown_or_missing_portfolio_lists_known(self):
 		""".pie with no portfolio, or an unknown one, lists the portfolios there are."""
-		for text, first in (('.pie', '.pie: please try again'), ('.pie Nope', 'Nope portfolio not found')):
-			sent = self.command(text)
-			self.assertTrue(sent[0][1][0].startswith(first), text)
-			self.assertIn('Rob', sent[0][1])
+		self.assertEqual(self.command('.pie'), [('text', ['Usage: .pie PORTFOLIO [sector]', 'Portfolios: Rob'])])
+		self.assertEqual(self.command('.pie Nope'), [('text', ['Nope portfolio not found. I only know about:', 'Rob'])])
 
 	def test_a_service_that_takes_no_images_gets_text(self):
 		"""Slack without the bot token takes no images: the report goes as text."""
diff --git a/util/check_config.py b/util/check_config.py
index b156694c..ef0615b7 100755
--- a/util/check_config.py
+++ b/util/check_config.py
@@ -19,7 +19,8 @@
 
 
 def check(startup_path, preferences_path):
-	"""Every problem in the two files, as a list of lines."""
+	"""Every problem in the two files, as a list of lines (a key that does nothing any more is a note, not a problem:
+	notes() has those)."""
 	problems = []
 	if startup_path:
 		try:
@@ -36,6 +37,14 @@ def check(startup_path, preferences_path):
 	return problems
 
 
+def notes(preferences_path):
+	"""The keys of a valid preferences.conf that do nothing any more, a line each."""
+	try:
+		return config.retired_notes(config.read_preferences(preferences_path) or {})
+	except config.ConfigError:
+		return []
+
+
 if __name__ == '__main__':
 	ap = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter)
 	ap.add_argument('startup', help='the finbot.env to check (the startup file)')
@@ -46,4 +55,6 @@ def check(startup_path, preferences_path):
 		print('ERROR config:', line, file=sys.stderr)
 	if not problems:
 		print('config: ok', args.startup, args.preferences if Path(args.preferences).is_file() else '(no preferences.conf, defaults)')
+		for line in notes(args.preferences):
+			print('config:', line)
 	sys.exit(1 if problems else 0)

From ec8ce44be09b65635d4014619b1450a29605bd40 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 08:56:19 +0000
Subject: [PATCH 55/81] Rankings: one stock's figure and a ranked list are
 separate functions; .pe charts too
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- marketcap_line and marketcap_list, ratio_line and ratio_list replace the two functions that each served a single
  stock and a list through action strings ('bottom forward pe', split apart again inside).
- One stock's ratio shows what Yahoo has: .pe of a loss-maker gives its forward P/E, where it said nothing; .peg
  labels its figure (PEG: 1.2).
- .pe sends the chart it built, as .forward pe and .peg do.
- links.stock_line names a stock the one way ("🇦🇺 Title (TICKER)", no stray space where a market has no flag) for
  the rankings, milestone and the watchlist, which sorts by name before formatting: iShares no longer sorts under S.
- Yahoo's recommendation key is made readable once (strong_buy: strong buy), not in three reports.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/links.py             |   7 +++
 lib/reports/holdings.py  |   5 +-
 lib/reports/milestone.py |   5 +-
 lib/reports/profile.py   |   2 +-
 lib/reports/rankings.py  | 113 ++++++++++++++++++++-------------------
 lib/reports/rating.py    |   2 +-
 lib/worker.py            |  20 ++++---
 lib/yahoo.py             |   2 +-
 tests/test_reports.py    |  47 ++++++++++------
 9 files changed, 116 insertions(+), 87 deletions(-)

diff --git a/lib/links.py b/lib/links.py
index 8dd8b5ee..a8d88e09 100644
--- a/lib/links.py
+++ b/lib/links.py
@@ -1,10 +1,17 @@
 """Links to a stock's page on Yahoo or Google Finance, in the markup of `service` (a dialect: see lib/dialects.py)."""
 from lib import config
+from lib import formats
 from lib import markets
 
 NO_GOOGLE_PAGE = ('Taipei Exchange', 'CCC') # exchanges Google Finance has no pages for (CCC: crypto), nor indices (^)
 WINDOWS = ((1825, 'Max'), (365, '5Y'), (183, '1Y'), (31, '6M'), (7, '1M'), (1, '5D')) # over that many days, that window
 
+def stock_line(ticker, quote, service, days=1):
+	"""'🇦🇺 Title (TICKER)': how a report line names a stock from its Yahoo quote, the symbol linked to its page
+	(finance_link) and the flag of its market left out where it has none."""
+	link = finance_link(ticker, quote.get('profile_exchange') or '', service, days)
+	return formats.security_line(markets.flag_from_ticker(ticker), quote.get('profile_title', ticker), link)
+
 def yahoo_link(ticker, service, brief=True, text=None):
 	"""A link to the ticker's Yahoo Finance page reading `text`, else the bare symbol (brief) or the whole ticker."""
 	if not text:
diff --git a/lib/reports/holdings.py b/lib/reports/holdings.py
index cf284f8e..65b568d2 100644
--- a/lib/reports/holdings.py
+++ b/lib/reports/holdings.py
@@ -1,5 +1,4 @@
 """What is held and followed: .holdings, .who, .pie and .watchlist."""
-import re
 from lib import watchlist
 from lib import sharesight
 from lib import formats
@@ -74,8 +73,8 @@ def prepare_watchlist(service, user, action=None, ticker=None):
 		return [watchlist_change(service, user, action, ticker)]
 	watched = watchlist.tickers()
 	market_data = yahoo.fetch(watched) if watched else {}
-	lines = sorted((f"{markets.flag_from_ticker(t)} {quote['profile_title']} ({links.finance_link(t, quote.get('profile_exchange'), service)})"
-		for t, quote in market_data.items()), key=lambda line: re.findall('[A-Z].*', line)) # by name, the flag aside
+	lines = [links.stock_line(ticker, quote, service)
+		for ticker, quote in sorted(market_data.items(), key=lambda item: item[1]['profile_title'].lower())]
 	if not lines:
 		return ['Watchlist is empty. Try ".watchlist add SYMBOL" to create it']
 	return [f"Hi {user}, I'm currently tracking:"] + lines
diff --git a/lib/reports/milestone.py b/lib/reports/milestone.py
index 34702dab..2a1c89df 100644
--- a/lib/reports/milestone.py
+++ b/lib/reports/milestone.py
@@ -3,7 +3,6 @@
 from lib import state
 from lib import delivery
 from lib import links
-from lib import markets
 from lib import watchlist
 from lib import yahoo
 
@@ -56,9 +55,7 @@ def report(service, market_data, highs, lows):
 	"""The records in one network's markup: the record highs, then the record lows, each list sorted."""
 	def line(ticker):
 		"""A stock's flag, name and link."""
-		data = market_data[ticker]
-		link = links.finance_link(ticker, data['profile_exchange'], service)
-		return f"{markets.flag_from_ticker(ticker)} {data['profile_title']} ({link})"
+		return links.stock_line(ticker, market_data[ticker], service)
 	payload = []
 	if highs:
 		payload += [service.bold('Record high:')] + sorted(line(ticker) for ticker in highs)
diff --git a/lib/reports/profile.py b/lib/reports/profile.py
index 76ab8725..23b3301e 100644
--- a/lib/reports/profile.py
+++ b/lib/reports/profile.py
@@ -159,7 +159,7 @@ def outlook(service, data):
 		shorted = data['short_percent']
 		lines.append(service.bold("Shorted stock:") + f" {shorted}%" + ('⚠️ ' if shorted > 10 else ''))
 	if 'recommend' in data:
-		rating = data['recommend'].replace('_', ' ')
+		rating = data['recommend']
 		score = f"{data['recommend_index']} {rating} ({data['recommend_analysts']} analysts)"
 		lines.append(service.bold("Score:") + f" {score}")
 	return lines
diff --git a/lib/reports/rankings.py b/lib/reports/rankings.py
index f1b93247..5a89ce62 100644
--- a/lib/reports/rankings.py
+++ b/lib/reports/rankings.py
@@ -1,40 +1,42 @@
-"""Ranked lists of the stocks followed: .marketcap, .pe, .fpe and .peg (top, bottom, negative), .beta, .buy and .sell."""
+"""Ranked lists of the stocks followed (.marketcap, .pe, .forward pe and .peg, .beta, .buy and .sell), and one stock's
+market cap or ratio."""
 import sys
 from lib import answer
 from lib import config
 from lib import watchlist
 from lib import formats
 from lib import links
-from lib import markets
 from lib import render
 from lib import yahoo
 
-def prepare_marketcap_payload(service, action='top', length=15):
-	"""The `length` tracked stocks of largest (action 'top') or smallest ('bottom') market cap, compared in US dollars at Yahoo's
-	exchange rates: a market cap in won or yen is a far bigger number. Any other action is one ticker's, in its own currency too."""
-	if action not in ('top', 'bottom'):
-		ticker = action
-		data = yahoo.add_usd_market_caps(yahoo.fetch_detail(ticker, 600)).get(ticker, {})
-		if 'market_cap' not in data:
-			return [f"Mkt cap not found for {ticker}"]
-		currency = yahoo.market_cap_currency(data) or ''
-		cap = f"{currency} {formats.human_units(data['market_cap'])}".strip()
-		if currency != 'USD' and 'market_cap_usd' in data:
-			cap += f" (USD {formats.human_units(data['market_cap_usd'])})"
-		link = links.finance_link(ticker, data.get('profile_exchange', ''), service)
-		return [formats.security_line(markets.flag_from_ticker(ticker), data.get('profile_title', ticker), link) + f" mkt cap: {cap}"]
+def marketcap_line(service, ticker):
+	"""One stock's market cap, in its own currency, and in US dollars too where that is another. Raises answer.Answer
+	when Yahoo has none."""
+	data = yahoo.add_usd_market_caps(yahoo.fetch_detail(ticker, 600)).get(ticker, {})
+	if 'market_cap' not in data:
+		raise answer.Answer(f"Mkt cap not found for {ticker}")
+	currency = yahoo.market_cap_currency(data) or ''
+	cap = f"{currency} {formats.human_units(data['market_cap'])}".strip()
+	if currency != 'USD' and 'market_cap_usd' in data:
+		cap += f" (USD {formats.human_units(data['market_cap_usd'])})"
+	return f"{links.stock_line(ticker, data, service)} mkt cap: {cap}"
+
+def marketcap_list(service, bottom=False, length=15):
+	"""The `length` stocks followed of largest (or, `bottom`, smallest) market cap, compared in US dollars at Yahoo's
+	exchange rates: a market cap in won or yen is a far bigger number."""
 	caps = [] # (market cap in US dollars, line)
 	for ticker, data in yahoo.add_usd_market_caps(yahoo.fetch(watchlist.with_holdings())).items():
 		if 'market_cap_usd' not in data:
-			print(ticker, "no market cap" if 'market_cap' not in data else f"no exchange rate for {data.get('currency')}", file=sys.stderr)
+			why = "no market cap" if 'market_cap' not in data else f"no exchange rate for {data.get('currency')}"
+			print(ticker, why, file=sys.stderr)
 			continue
-		link = links.finance_link(ticker, data['profile_exchange'], service)
-		line = formats.security_line(markets.flag_from_ticker(ticker), data['profile_title'], link)
-		caps.append((data['market_cap_usd'], line + f" mkt cap: {formats.human_units(data['market_cap_usd'])}"))
+		cap = formats.human_units(data['market_cap_usd'])
+		caps.append((data['market_cap_usd'], f"{links.stock_line(ticker, data, service)} mkt cap: {cap}"))
 	if not caps:
 		return ['No market cap data found']
-	caps.sort(key=lambda cap: cap[0], reverse=action == 'top')
-	return [service.bold(f'{action.title()} {length} tracked stocks by market cap (USD)')] + [line for _, line in caps[:length]]
+	caps.sort(key=lambda cap: cap[0], reverse=not bottom)
+	heading = f"{'Bottom' if bottom else 'Top'} {length} tracked stocks by market cap (USD)"
+	return [service.bold(heading)] + [line for _, line in caps[:length]]
 
 def prepare_rating_payload(service, action, length=15):
 	"""(payload, chart) of the `length` tracked stocks rated `action` ('buy' or 'sell') by min_analysts or more analysts (fewer,
@@ -44,11 +46,11 @@ def prepare_rating_payload(service, action, length=15):
 	for ticker, data in yahoo.fetch_details(watchlist.with_holdings()).items():
 		if 'recommend' not in data or data['recommend_analysts'] < config.prefs.min_analysts:
 			continue
-		recommend = data['recommend'].replace('_', ' ')
+		recommend = data['recommend']
 		if (action == 'buy' and 'buy' in recommend) or (action == 'sell' and recommend in ('sell', 'underperform')):
 			rating, analysts = round(data['recommend_index'], 2), data['recommend_analysts']
-			link = links.finance_link(ticker, data['profile_exchange'], service)
-			found.append((rating, analysts, f"{markets.flag_from_ticker(ticker)} {data['profile_title']} ({link}) {rating} ({analysts} analysts)", ticker))
+			line = f"{links.stock_line(ticker, data, service)} {rating} ({analysts} analysts)"
+			found.append((rating, analysts, line, ticker))
 	found = sorted(found, key=lambda row: (row[0] if action == 'buy' else -row[0], -row[1]))[:length]
 	payload = [line for _, _, line, _ in found]
 	if not payload:
@@ -64,48 +66,51 @@ def prepare_rating_payload(service, action, length=15):
 
 RATIOS = {'pe': 'price_to_earnings_trailing', 'forward pe': 'price_to_earnings_forward', 'peg': 'price_to_earnings_peg'}
 
-def prepare_value_payload(service, action='pe', ticker_select=None, length=15):
-	"""(payload, chart): the tracked stocks by a ratio of RATIOS, or one ticker's. action is the ratio, or 'bottom' or
-	'negative' and the ratio: lowest first, highest first ('bottom'), or the negative ones only, lowest first. A forward
-	P/E is analysts' estimate, so its lists take only stocks that min_analysts or more analysts cover, as .buy does."""
-	mode, kind = action.split(' ', 1) if action.split()[0] in ('bottom', 'negative') else ('top', action)
+def ratio_line(service, ticker, kind):
+	"""One stock's line for .pe, .forward pe or .peg SYMBOL (kind: pe, forward pe or peg): its P/E, trailing and
+	forward where Yahoo has them, or its PEG. Raises answer.Answer when Yahoo knows neither the stock nor the ratio (a
+	loss-maker has no trailing P/E)."""
+	data = yahoo.fetch([ticker]).get(ticker)
+	if data is None:
+		raise answer.Answer(f"{ticker}: no data found")
+	if kind == 'peg': # the PEG is in Yahoo's details
+		data |= yahoo.fetch_details([ticker], strict=True).get(ticker, {})
+		figures = f"PEG: {data['price_to_earnings_peg']}" if data.get('price_to_earnings_peg') is not None else ''
+	else:
+		both = (('trailing', 'price_to_earnings_trailing'), ('forward', 'price_to_earnings_forward'))
+		figures = ', '.join(f'{name} {data[key]}' for name, key in both if data.get(key) is not None)
+		figures = f"PE: {figures}" if figures else ''
+	if not figures:
+		raise answer.Answer(f"{ticker}: no {kind} ratio found")
+	return f"{links.stock_line(ticker, data, service)} {figures}"
+
+def ratio_list(service, kind, mode='top', length=15):
+	"""(payload, chart): the `length` stocks followed by a ratio (kind: pe, forward pe or peg), lowest first (mode top),
+	highest first (bottom), or the negative ones only, lowest first (negative). A forward P/E is analysts' estimate, so
+	its list takes only stocks that min_analysts or more analysts cover, as .buy does."""
 	field = RATIOS[kind]
-	market_data = yahoo.fetch(watchlist.covered(ticker_select))
+	market_data = yahoo.fetch(watchlist.with_holdings())
 	if kind == 'peg': # the PEG is in Yahoo's details
-		market_data |= yahoo.fetch_details(market_data, strict=bool(ticker_select))
+		market_data |= yahoo.fetch_details(market_data)
 	covered = None # for a forward P/E list: the tickers enough analysts cover
-	if kind == 'forward pe' and not ticker_select:
+	if kind == 'forward pe':
 		covered = {ticker for ticker, detail in yahoo.fetch_details(market_data).items()
 			if detail.get('recommend_analysts', 0) >= config.prefs.min_analysts}
 	rows = [] # (ratio, chart label, line), one sorted list for the caption and the chart
 	for ticker, data in market_data.items():
 		ratio = data.get(field)
-		if ratio is None or (covered is not None and ticker not in covered):
-			continue
-		if not ticker_select and (ratio < 0) != (mode == 'negative'): # a list: the negative ones, or none of them
-			continue
-		link = links.finance_link(ticker, data.get('profile_exchange'), service)
-		start = f"{markets.flag_from_ticker(ticker)} {data.get('profile_title')} ({link})"
-		if ticker_select and mode == 'top' and kind in ('pe', 'forward pe'): # one stock's P/E: trailing and forward
-			both = (('trailing', 'price_to_earnings_trailing'), ('forward', 'price_to_earnings_forward'))
-			parts = ', '.join(f'{name} {data[key]}' for name, key in both if data.get(key) is not None)
-			rows.append((ratio, None, f"{start} PE: {parts}"))
-		else:
-			rows.append((ratio, ticker.split('.')[0], f"{start} {ratio}"))
+		if ratio is None or (covered is not None and ticker not in covered) or (ratio < 0) != (mode == 'negative'):
+			continue # the negative ones, or none of them
+		rows.append((ratio, ticker.split('.')[0], f"{links.stock_line(ticker, data, service)} {ratio}"))
 	rows.sort(key=lambda row: row[0])
 	if mode == 'bottom':
 		rows.reverse()
-	if ticker_select:
-		if not rows: # Yahoo does not know it, or sends no such ratio for it (a loss-maker has no trailing P/E)
-			raise answer.Answer(f"{ticker_select}: no {kind.upper()} found" if ticker_select in market_data
-				else f"{ticker_select}: no data found")
-		return [line for _, _, line in rows], None
 	rows = rows[:length]
-	trail = kind if mode != 'negative' else action # what the heading calls the ratio
+	trail = f'negative {kind}' if mode == 'negative' else kind # what the heading calls the ratio
 	if not rows:
 		if covered is not None:
 			return [f"No tracked stocks with {config.prefs.min_analysts}+ analysts have a {trail} ratio"], None
-		return [f"{action}: no data found"], None
+		return [f"{kind if mode == 'top' else f'{mode} {kind}'}: no data found"], None
 	coverage = f" ({config.prefs.min_analysts}+ analysts)" if covered is not None else ''
 	heading = f"{'Bottom' if mode == 'bottom' else 'Top'} {length} tracked stocks by {trail} ratio{coverage}"
 	bars = tuple((label, float(ratio)) for ratio, label, _ in rows)
@@ -120,8 +125,8 @@ def prepare_beta_payload(service):
 			continue
 		beta = round(data['beta'], 1)
 		if beta > 1.5 and data.get('market_cap_usd', float('inf')) < 2e9:
-			link = links.finance_link(ticker, data['profile_exchange'], service)
-			found.append((beta, formats.security_line(markets.flag_from_ticker(ticker), data['profile_title'], link) + f" {beta} β", (ticker.split('.')[0], float(beta))))
+			line = f"{links.stock_line(ticker, data, service)} {beta} β"
+			found.append((beta, line, (ticker.split('.')[0], float(beta))))
 	if not found:
 		return ['No tracked stocks have a beta over 1.5 and mkt cap under US$2B'], None
 	found.sort(key=lambda row: row[0], reverse=True)
diff --git a/lib/reports/rating.py b/lib/reports/rating.py
index d8b7aef1..53514467 100644
--- a/lib/reports/rating.py
+++ b/lib/reports/rating.py
@@ -26,7 +26,7 @@ def gather(tickers):
 	changes, seen = [], {}
 	for ticker in tickers:
 		try:
-			action = market_data[ticker]['recommend'].replace('_', ' ')
+			action = market_data[ticker]['recommend']
 			index = round(float(market_data[ticker]['recommend_index']), 1)
 			analysts = market_data[ticker]['recommend_analysts']
 		except (KeyError, ValueError):
diff --git a/lib/worker.py b/lib/worker.py
index db154c6a..5e4fb961 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -263,16 +263,20 @@ def on_holdings(request, args):
 
 def on_marketcap(request, args):
 	""".marketcap [SYMBOL|top|bottom]: stocks followed by market cap, largest (or smallest) first, or one stock's."""
-	request.reply(rankings.prepare_marketcap_payload(request.service, args.keyword or ticker(args) or 'top', length=15))
+	if args.words:
+		request.reply([rankings.marketcap_line(request.service, ticker(args))])
+	else:
+		request.reply(rankings.marketcap_list(request.service, bottom=args.keyword == 'bottom', length=15))
 
-def on_ratio(kind, chart=True):
-	"""The handler of .pe, .forward pe or .peg: the top (by default) or bottom 15 by the ratio, the negative ones (where
-	there can be any), or one ticker's ratio. `chart`: with its chart."""
+def on_ratio(kind):
+	"""The handler of .pe, .forward pe or .peg (kind: pe, forward pe or peg): the top (by default) or bottom 15 by the
+	ratio and its chart, the negative ones (where there can be any), or one ticker's ratio."""
 	def handler(request, args):
 		""".pe, .forward pe or .peg [SYMBOL|top|bottom|negative]."""
-		action = f'{args.keyword} {kind}' if args.keyword in ('bottom', 'negative') else kind
-		payload, spec = rankings.prepare_value_payload(request.service, action, ticker(args), length=15)
-		request.reply(payload, spec if chart else None)
+		if args.words:
+			request.reply([rankings.ratio_line(request.service, ticker(args), kind)])
+		else:
+			request.reply(*rankings.ratio_list(request.service, kind, args.keyword or 'top', length=15))
 	return handler
 
 def on_beta(request, args):
@@ -394,7 +398,7 @@ class Command:
 	Command(('session', 'premarket', 'postmarket', 'permarket'), on_session,
 		('.session [percent|SYMBOL]', '.premarket [percent|SYMBOL]'), PRICE, ('.session', '.premarket'), percent=True),
 	Command(BUY_SELL, on_rating, ('.buy', '.sell'), RISK, ('.buy & sell',)), # .buy & sell (or .buy sell): both
-	Command(('pe',), on_ratio('pe', chart=False), ('.pe [SYMBOL|top|bottom]',), RISK, ('.pe',), keywords=TOP),
+	Command(('pe',), on_ratio('pe'), ('.pe [SYMBOL|top|bottom]',), RISK, ('.pe',), keywords=TOP),
 	Command(('forward pe', 'forwardpe', 'fpe'), on_ratio('forward pe'), ('.forward pe [SYMBOL|top|bottom|negative]',),
 		RISK, ('.forward pe',), keywords=NEGATIVE),
 	Command(('peg',), on_ratio('peg'), ('.peg [SYMBOL|top|bottom|negative]',), RISK, ('.peg',), keywords=NEGATIVE),
diff --git a/lib/yahoo.py b/lib/yahoo.py
index e8873464..90cfdfa3 100644
--- a/lib/yahoo.py
+++ b/lib/yahoo.py
@@ -244,7 +244,7 @@ def _quote_answer(tickers):
 	('free_cashflow', ('financialData', 'freeCashflow', 'raw')),
 	('operating_cashflow', ('financialData', 'operatingCashflow', 'raw')),
 	('financialCurrency', ('financialData', 'financialCurrency')),
-	('recommend', ('financialData', 'recommendationKey')),
+	('recommend', ('financialData', 'recommendationKey'), lambda key: key.replace('_', ' ')), # strong_buy: strong buy
 	('recommend_index', ('financialData', 'recommendationMean', 'raw')),
 	('recommend_analysts', ('financialData', 'numberOfAnalystOpinions', 'raw')),
 )
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 7dd4459b..dd22820b 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -96,7 +96,7 @@ def test_prepare_holdings(self):
 
 	def test_prepare_marketcap(self):
 		"""The market cap list is built."""
-		self.assertTrue(rankings.prepare_marketcap_payload(TELEGRAM, 'top', length=5))
+		self.assertTrue(rankings.marketcap_list(TELEGRAM, length=5))
 
 	def test_prepare_rating(self):
 		"""The buy and sell lists are built."""
@@ -147,18 +147,18 @@ def test_forward_pe_lists_need_as_many_analysts_as_buy(self):
 		self.patch(watchlist, 'with_holdings', lambda: list(analysts))
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | {'price_to_earnings_forward': -5 if t == 'T3' else 20} for t in ts})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'recommend_analysts': analysts[t]}})
-		def listed(action):
-			"""(heading, the tickers listed, the chart) of a ratio report."""
-			payload, chart = rankings.prepare_value_payload(TELEGRAM, action)
+		def listed(mode='top'):
+			"""(heading, the tickers listed, the chart) of a forward P/E list."""
+			payload, chart = rankings.ratio_list(TELEGRAM, 'forward pe', mode)
 			return payload[0], [line.split('Co ')[1].split(' ')[0] for line in payload[1:]], chart
-		heading, names, chart = listed('forward pe')
+		heading, names, chart = listed()
 		# T2 and T4: too few analysts; T3: negative
 		self.assertEqual((names, [label for label, _ in chart.rows]), (['T1'], ['T1']))
 		self.assertIn('forward pe ratio (7+ analysts)', heading)
-		self.assertEqual(listed('negative forward pe')[1], ['T3'])
-		self.assertIn('Co T2', str(rankings.prepare_value_payload(TELEGRAM, 'forward pe', 'T2'))) # one ticker asked for: shown
+		self.assertEqual(listed('negative')[1], ['T3'])
+		self.assertIn('Co T2', rankings.ratio_line(TELEGRAM, 'T2', 'forward pe')) # one ticker asked for: shown
 		self.prefs(min_analysts=13) # the one setting moves .buy and .forward pe together
-		self.assertEqual(rankings.prepare_value_payload(TELEGRAM, 'forward pe')[0], ['No tracked stocks with 13+ analysts have a forward pe ratio'])
+		self.assertEqual(listed()[0], 'No tracked stocks with 13+ analysts have a forward pe ratio')
 		self.assertEqual(rankings.prepare_rating_payload(TELEGRAM, 'buy')[0], ['No tracked stocks rated buy by 13+ analysts'])
 
 	def test_market_caps_compare_in_us_dollars(self):
@@ -172,24 +172,41 @@ def fetch(tickers):
 		self.patch(yahoo, 'fetch', fetch)
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: fetch([t]))
 		self.patch(watchlist, 'with_holdings', lambda: list(stocks))
-		payload = rankings.prepare_marketcap_payload(TELEGRAM)
+		payload = rankings.marketcap_list(TELEGRAM)
 		self.assertIn('by market cap (USD)', payload[0])
 		self.assertEqual([line.split('Co ')[1].split(' ')[0] for line in payload[1:]], ['U1', 'K1.KS', 'L1.L', 'K2.KS']) # Z1: no rate, left out
 		self.assertIn('mkt cap: 731.88 B', payload[2]) # 1.01 Q won
-		self.assertIn('mkt cap: KRW 1.01 Q (USD 731.88 B)', rankings.prepare_marketcap_payload(TELEGRAM, 'K1.KS')[0])
-		self.assertIn('mkt cap: GBP 62.50 B (USD 79.11 B)', rankings.prepare_marketcap_payload(TELEGRAM, 'L1.L')[0]) # pence quote, pound cap
+		self.assertIn('mkt cap: KRW 1.01 Q (USD 731.88 B)', rankings.marketcap_line(TELEGRAM, 'K1.KS'))
+		self.assertIn('mkt cap: GBP 62.50 B (USD 79.11 B)', rankings.marketcap_line(TELEGRAM, 'L1.L')) # pence: pounds
 		beta = rankings.prepare_beta_payload(TELEGRAM)[0]
 		self.assertIn('Co K2.KS', str(beta)) # 500 billion won is US$362M: a small cap, which it never was unconverted
 		self.assertNotIn('Co K1.KS', str(beta))
 
 	def test_prepare_value(self):
 		"""Each ratio report has its heading: top, bottom, and negative (with none found)."""
-		for action, heading in (('pe', 'Top 5 tracked stocks by pe ratio'), ('bottom forward pe', 'Bottom 5 tracked stocks by forward pe ratio'),
-								('peg', 'Top 5 tracked stocks by peg ratio'), ('negative peg', 'negative peg: no data found')):
-			with self.subTest(action=action):
-				payload, chart = rankings.prepare_value_payload(TELEGRAM, action, length=5)
+		for kind, mode, heading in (('pe', 'top', 'Top 5 tracked stocks by pe ratio'),
+				('forward pe', 'bottom', 'Bottom 5 tracked stocks by forward pe ratio'),
+				('peg', 'top', 'Top 5 tracked stocks by peg ratio'),
+				('peg', 'negative', 'negative peg: no data found')):
+			with self.subTest(kind=kind, mode=mode):
+				payload, chart = rankings.ratio_list(TELEGRAM, kind, mode, length=5)
 				self.assertIn(heading, payload[0])
 
+	def test_one_stocks_ratio_shows_what_yahoo_has(self):
+		""".pe of a loss-maker shows its forward P/E (it used to say nothing), .peg labels its figure, and a stock
+		without the ratio asked for, or one Yahoo does not know, is said to have none."""
+		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | {'price_to_earnings_trailing': None} for t in ts
+			if not t.startswith('NONE')})
+		self.assertEqual(self.command('.pe T1')[0][1][0].split(') ')[1], 'PE: forward 9')
+		self.assertEqual(self.command('.peg T1')[0][1][0].split(') ')[1], 'PEG: 1.2')
+		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'price_to_earnings_peg': None}})
+		self.assertEqual(self.command('.peg T1'), [('text', ['T1: no peg ratio found'])])
+		self.assertEqual(self.command('.pe NONE.L'), [('text', ['NONE.L: no data found'])])
+
+	def test_pe_sends_its_chart_as_forward_pe_and_peg_do(self):
+		""".pe builds a chart, and now sends it."""
+		self.assertEqual([kind for kind, *_ in self.command('.pe')], ['photo'])
+
 	def test_prepare_beta(self):
 		"""The beta report charts a bar per stock listed, and says so when none qualify."""
 		payload, chart = rankings.prepare_beta_payload(TELEGRAM)

From 4be22c4c19870cd7a8954a24bd4418cff45a41ad Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 09:01:03 +0000
Subject: [PATCH 56/81] Reports: blank lines between sections the one way; dead
 code out

- formats.sections joins blocks with a blank line between each two, for .help, the profile, .who, .plan, trades,
  the calendar and milestone. A trades report over days no longer starts with a blank line, nor does a milestone
  report of record lows only, and .plan and .who no longer end with one.
- .history builds the line of the period pressed itself; it no longer returns a table its caller threw away.
- Unreachable branches removed: an empty performance report (gather raises first), its rebuild's guard, the
  compare rebuild's; performance.report's default that repeated run's.
- A .jobs button pressed while the scheduler is off is told so directly, not by catching AttributeError.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/charts.py              |  7 +------
 lib/formats.py             |  9 +++++++++
 lib/reports/about.py       | 18 +++++-------------
 lib/reports/cal.py         | 16 +++++++---------
 lib/reports/history.py     | 13 +++++++++----
 lib/reports/holdings.py    |  7 +++----
 lib/reports/milestone.py   | 10 +++-------
 lib/reports/performance.py |  6 +-----
 lib/reports/profile.py     |  8 ++------
 lib/reports/trades.py      | 21 +++++++++------------
 lib/worker.py              |  9 ++++++---
 11 files changed, 55 insertions(+), 69 deletions(-)

diff --git a/lib/charts.py b/lib/charts.py
index 199e93b4..4a25d46a 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -8,10 +8,7 @@
 
 def history_chart(service, period, days, ticker):
 	""".history: the change over every period, the chosen one marked, and the line of that period (or of it all)."""
-	lines, _, chart = history.history_payload(ticker, service, buttons.PERIODS[period or 'x'][0], graph=not days)
-	if days:
-		changes, line = yahoo.price_history(ticker, days)
-		chart = history.line_chart(ticker, changes, line, days)
+	lines, chart = history.history_payload(ticker, service, days)
 	return '\n'.join(lines), chart
 
 def price_chart(service, period, days, ticker):
@@ -34,8 +31,6 @@ def compare_chart(service, period, days, *tickers):
 def performance_chart(service, period, days, portfolio):
 	""".performance: each portfolio over the period, or the one named ('' for all)."""
 	payload, chart = performance.report(service, days or buttons.MAX_DAYS, portfolio or None)
-	if not chart:
-		raise answer.Answer(*(payload or ['no performance data']))
 	return '\n'.join(payload), chart
 
 def price_list_chart(service, period, days, threshold, top):
diff --git a/lib/formats.py b/lib/formats.py
index f112cff3..e7ac22c6 100644
--- a/lib/formats.py
+++ b/lib/formats.py
@@ -58,6 +58,15 @@ def human_units(value, decimal_places=2):
 		value /= 1000.0
 	return f"{value:.{decimal_places}f} {unit}".rstrip()
 
+def sections(*blocks):
+	"""Blocks of lines one after another, a blank line between each two; an empty block is left out."""
+	lines = []
+	for block in blocks:
+		if block and lines:
+			lines.append('')
+		lines += block or []
+	return lines
+
 def security_line(flag, title, link):
 	"""'🇦🇺 Title (link)': how a report line names a security (without a leading space where its market has no flag)."""
 	return f"{flag} {title} ({link})" if flag else f"{title} ({link})"
diff --git a/lib/reports/about.py b/lib/reports/about.py
index b642bed0..3f5ebb85 100644
--- a/lib/reports/about.py
+++ b/lib/reports/about.py
@@ -1,5 +1,6 @@
 """What finbot can do: the .help text, and the .jobs listing of the schedule (with running a job from it)."""
 from lib import buttons
+from lib import formats
 from lib import jobs
 
 def prepare_jobs_payload(service, run=None):
@@ -39,16 +40,7 @@ def job_line(scheduler, label, outcome):
 def prepare_help(service, botName, sections):
 	"""The .help text: each section's command lines under its heading (worker.help_sections() has them), then the bot's
 	@name as another way to start a command. botName is the bot's @name."""
-	payload = []
-	for heading, lines in sections:
-		if payload: # a blank line between sections, outside the bold: Slack does not bold *\nPrice:*
-			payload.append('')
-		payload += [service.bold(heading + ":")] + lines
-	payload += ['', service.bold("Alternative syntax:")]
-	if service == 'slack':
-		payload.append('<' + botName + '> trades')
-	else:
-		payload.append(botName + ' price')
-	payload.append("")
-	payload.append(f"Full help on {service.link('https://github.com/robdevops/finbot', 'GitHub')}")
-	return payload
+	example = '<' + botName + '> trades' if service == 'slack' else botName + ' price'
+	return formats.sections(*([service.bold(heading + ":")] + lines for heading, lines in sections),
+		[service.bold("Alternative syntax:"), example],
+		[f"Full help on {service.link('https://github.com/robdevops/finbot', 'GitHub')}"])
diff --git a/lib/reports/cal.py b/lib/reports/cal.py
index ec50fdb7..c6ee23c3 100644
--- a/lib/reports/cal.py
+++ b/lib/reports/cal.py
@@ -72,17 +72,15 @@ def primary_listing(ticker, data, field):
 def report(service, events, days, specific_stock, interactive, event):
 	"""The events in one network's markup, grouped by date under a heading (one stock's carry their dates instead)."""
 	heading = EVENTS[event][1]
-	payload = []
+	blocks = []
 	for date in sorted({when.date() for when, *_ in events}):
-		if not specific_stock:
-			payload += ["", service.bold(f"{date:%b %d}")]
-		for when, flag, title, ticker in events:
-			if when.date() == date:
-				link = links.yahoo_link(ticker, service)
-				day = f"{date:%b %d} " if specific_stock else '' # one stock's events carry their date
-				payload.append(f"{flag} {day}{title} ({link})")
+		day = f"{date:%b %d} " if specific_stock else '' # one stock's events carry their date
+		lines = [f"{flag} {day}{title} ({links.yahoo_link(ticker, service)})"
+			for when, flag, title, ticker in events if when.date() == date]
+		blocks.append(lines if specific_stock else [service.bold(f"{date:%b %d}")] + lines)
+	payload = sum(blocks, []) if specific_stock else formats.sections(*blocks)
 	if payload and not specific_stock:
-		payload.insert(0, service.bold(f"{heading} {formats.days_english(days, 'for the next ')}:"))
+		payload = formats.sections([service.bold(f"{heading} {formats.days_english(days, 'for the next ')}:")], payload)
 	elif not payload and interactive:
 		ahead = formats.days_english(days, 'for the next ')
 		payload = [f"No events found for {specific_stock}" if specific_stock else f"No events found {ahead}"]
diff --git a/lib/reports/history.py b/lib/reports/history.py
index 90c4085c..7fa77056 100644
--- a/lib/reports/history.py
+++ b/lib/reports/history.py
@@ -1,6 +1,7 @@
 """Price history: .history (one stock over every period) and .compare (two to six stocks, rebased)."""
 import datetime
 from lib import answer
+from lib import buttons
 from lib import formats
 from lib import links
 from lib import render
@@ -44,9 +45,11 @@ def prepare_compare(service, tickers, period_days=None):
 		caption.append(f"{formats.get_emoji(pct)} {names[t]} ({link}): {formats.signed_percent(pct, 1)}")
 	return caption, chart
 
-def history_payload(ticker, service, highlight=None, graph=False):
-	"""(lines, percent table, chart): .history's caption, the change over every period with the `highlight` one marked,
-	and with graph=True the line chart of the whole history (else None). The name and exchange come from the chart."""
+def history_payload(ticker, service, days=None):
+	"""(lines, chart): .history's caption, the change over every period with the one of `days` marked, and the line
+	chart of those days, else of the whole history (None when it has too few prices). The name and exchange come from
+	the chart."""
+	highlight = buttons.PERIODS[buttons.period_for_days(days, 'x')][0]
 	table, line = yahoo.price_history(ticker)
 	ticker_link = links.finance_link(ticker, line.exchange, service, days=1825, brief=False)
 	lines = [service.bold(f"{line.name} ({ticker_link}) performance history")]
@@ -57,7 +60,9 @@ def history_payload(ticker, service, highlight=None, graph=False):
 			if interval == highlight:
 				label = '● ' + label
 			lines.append(f"{formats.get_emoji(percent)} {label} {formats.signed_percent(percent)}")
-	return lines, table, line_chart(ticker, table, line) if graph else None
+	if days:
+		table, line = yahoo.price_history(ticker, days)
+	return lines, line_chart(ticker, table, line, days)
 
 def line_chart(ticker, changes, line, days=None):
 	"""The line chart of a price_history (its changes and Line), titled with the company, the change over the span and
diff --git a/lib/reports/holdings.py b/lib/reports/holdings.py
index 65b568d2..63abe705 100644
--- a/lib/reports/holdings.py
+++ b/lib/reports/holdings.py
@@ -26,13 +26,12 @@ def prepare_holdings_payload(portfolio_name, service):
 
 def prepare_who(service, code):
 	"""Which portfolios hold `code` (a ticker without its exchange), each holding linked to Sharesight."""
-	payload = []
+	blocks = []
 	for portfolio in sharesight.portfolios():
 		found = [holding_line(holding, service, code)
 			for holding in portfolio.holdings().values() if holding['code'].upper() == code]
-		if found:
-			payload += [service.bold(f"{portfolio.name}:"), "\n".join(found), ""]
-	return payload or [f'{code} not found in any portfolio']
+		blocks.append(found and [service.bold(f"{portfolio.name}:")] + found)
+	return formats.sections(*blocks) or [f'{code} not found in any portfolio']
 
 def prepare_pie(portfolio_name, by, service, max_slices=8):
 	"""Portfolio value by industry (or by='sector'): (payload, chart or None). Raises sharesight.PortfolioNotFound."""
diff --git a/lib/reports/milestone.py b/lib/reports/milestone.py
index 2a1c89df..d2e9aaf9 100644
--- a/lib/reports/milestone.py
+++ b/lib/reports/milestone.py
@@ -1,7 +1,7 @@
 """Record highs and lows (all-time) among the stocks followed (scheduled: milestone.py)."""
-from itertools import groupby
 from lib import state
 from lib import delivery
+from lib import formats
 from lib import links
 from lib import watchlist
 from lib import yahoo
@@ -56,12 +56,8 @@ def report(service, market_data, highs, lows):
 	def line(ticker):
 		"""A stock's flag, name and link."""
 		return links.stock_line(ticker, market_data[ticker], service)
-	payload = []
-	if highs:
-		payload += [service.bold('Record high:')] + sorted(line(ticker) for ticker in highs)
-	if lows:
-		payload += ["", service.bold('Record low:')] + sorted(line(ticker) for ticker in lows)
-	return [i[0] for i in groupby(payload)] # de-dupe white space
+	return formats.sections(highs and [service.bold('Record high:')] + sorted(line(ticker) for ticker in highs),
+		lows and [service.bold('Record low:')] + sorted(line(ticker) for ticker in lows))
 
 def job(argv):
 	"""Arguments of a scheduled run: milestone.py takes none."""
diff --git a/lib/reports/performance.py b/lib/reports/performance.py
index d2bda98f..edc64c42 100644
--- a/lib/reports/performance.py
+++ b/lib/reports/performance.py
@@ -32,11 +32,9 @@ def benchmark(ticker, market, text, past_days, service):
 	link = links.finance_link(ticker, market, service=service, days=past_days, brief=True, text=text)
 	return f"{formats.get_emoji(percent)} {link} {formats.signed_percent(percent, 2)}", percent
 
-def report(service, past_days=None, portfolio_select=None, performance=None):
+def report(service, past_days, portfolio_select=None, performance=None):
 	"""(payload, chart) of the performance report for one service: each portfolio and the benchmarks, best first. performance:
 	gather()'s, fetched when not given."""
-	if past_days is None:
-		past_days = config.prefs.past_days
 	if performance is None:
 		performance = gather(past_days, portfolio_select)
 	lines = [] # (percent, text), sorted into the payload below
@@ -46,8 +44,6 @@ def report(service, past_days=None, portfolio_select=None, performance=None):
 		percent = float(data['report']['capital_gain_percent'])
 		lines.append((percent, f"{formats.get_emoji(percent)} {link} {formats.signed_percent(percent, 2)}"))
 		chart_rows.append((name, percent))
-	if not lines:
-		return [], None
 	for ticker, market, text in BENCHMARKS:
 		line, percent = benchmark(ticker, market, text, past_days, service)
 		lines.append((float('-inf') if percent is None else float(percent), line)) # an error line sorts last
diff --git a/lib/reports/profile.py b/lib/reports/profile.py
index 23b3301e..e1d9e103 100644
--- a/lib/reports/profile.py
+++ b/lib/reports/profile.py
@@ -47,13 +47,9 @@ def prepare_profile_payload(service, user, ticker):
 	state = data.get('marketState', '').rstrip()
 	exchange = data.get('profile_exchange', '').replace('NasdaqCM', 'Nasdaq').replace('NasdaqGS', 'Nasdaq')
 	exchange = exchange.replace('NYSEArca', 'NYSE')
-	sections = (about(data), company(service, data, ticker, exchange), figures(service, data), trends(data),
-		outlook(service, data), valuation(service, data, state), changes(service, data, ticker, state),
+	payload = formats.sections(about(data), company(service, data, ticker, exchange), figures(service, data),
+		trends(data), outlook(service, data), valuation(service, data, state), changes(service, data, ticker, state),
 		footer(service, data, ticker, exchange))
-	payload = []
-	for section in sections:
-		if section:
-			payload += ([""] if payload else []) + section
 	ticker_link = links.finance_link(ticker, exchange, service, brief=False)
 	light = MARKET_STATE_EMOJI.get(state, '🔴') if state else ''
 	heading = f"{service.bold(data.get('profile_title', ''))} ({ticker_link}) {light}"
diff --git a/lib/reports/trades.py b/lib/reports/trades.py
index d02d488b..5b898dfe 100644
--- a/lib/reports/trades.py
+++ b/lib/reports/trades.py
@@ -84,19 +84,16 @@ def gather(days, portfolio_select, known):
 def report(service, trades, days, interactive):
 	"""The trades in one network's markup, each linked to Sharesight and to the stock's page: under a heading per date
 	when there are several dates, or when a chat asked for more than a day."""
+	def line(trade):
+		"""One trade: who, bought or sold (linked to Sharesight), how much, and the stock."""
+		action, emoji = ('bought', '💸') if trade.bought else ('sold', '🤑')
+		url = f"https://portfolio.sharesight.com/holdings/{trade.holding_id}/dashboard/transactions"
+		return ' '.join([emoji, trade.portfolio, service.link(url, action), trade.currency, f'{trade.value}',
+			links.finance_link(trade.symbol, trade.market, service), markets.flag_from_market(trade.market)])
 	dates = sorted({trade.date for trade in trades})
-	payload = []
-	for date in dates:
-		if len(dates) > 1 or (interactive and days > 1):
-			payload += ["", service.bold(f"{date:%b %d}")]
-		for trade in trades:
-			if trade.date == date:
-				action, emoji = ('bought', '💸') if trade.bought else ('sold', '🤑')
-				url = f"https://portfolio.sharesight.com/holdings/{trade.holding_id}/dashboard/transactions"
-				payload.append(' '.join([emoji, trade.portfolio, service.link(url, action), trade.currency,
-					f'{trade.value}', links.finance_link(trade.symbol, trade.market, service),
-					markets.flag_from_market(trade.market)]))
-	return payload
+	headed = len(dates) > 1 or (interactive and days > 1)
+	return formats.sections(*([service.bold(f"{date:%b %d}")] * headed + [line(trade) for trade in trades
+		if trade.date == date] for date in dates))
 
 
 def job(argv):
diff --git a/lib/worker.py b/lib/worker.py
index 5e4fb961..ceb5371b 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -45,9 +45,11 @@ def process_job_callback(service, callback_id, chat_id, message_id, data):
 	"""A command button of .jobs: run that command now and answer in the chat."""
 	chat = delivery.reply_to(service, chat_id)
 	try:
+		if jobs.current is None: # the scheduler is off now
+			raise jobs.ScheduleError('the scheduler is off')
 		_, number, crc = data.split('|') # only a command offered by .jobs can be picked; anything else is refused
 		command = jobs.current.pick(number, crc)
-	except (ValueError, jobs.ScheduleError, AttributeError):
+	except (ValueError, jobs.ScheduleError):
 		chat.acknowledge(callback_id, 'Buttons updated, try again')
 		payload, markup = about.prepare_jobs_payload(service, '') # the list as it is now, in place of the stale one
 		chat.edit_text(message_id, '\n'.join(payload), markup or [])
@@ -331,8 +333,9 @@ def who(user):
 		"""A sender as shown: on Slack a mention (<@U123>), which is markup to keep; elsewhere the handle as text."""
 		user = user.removeprefix('@')
 		return user if request.service == 'slack' else request.service.escape(user)
-	request.reply([f"{request.service.bold(who(k))}: {request.service.italic(request.service.escape(v))}\n"
-		for k, v in plan.items()])
+	service = request.service
+	request.reply(formats.sections(*([f"{service.bold(who(k))}: {service.italic(service.escape(v))}"]
+		for k, v in plan.items())))
 
 def on_who(request, args):
 	""".who SYMBOL: the portfolios that hold it."""

From ab743ba55f91b759b18e08eba30d0c4c0dadc856 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 09:08:07 +0000
Subject: [PATCH 57/81] Price report: every move worked out once, then
 formatted per network
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

price.report ran once per chat and refetched in that loop: Yahoo's histories for the period and the small caps'
details. Now gather() works out each stock's change over the period (Sharesight's for a holding, Yahoo's for the
rest) and the one stock's line chart, moves() picks the moves once (the mode's filters, the threshold, the
scheduled trigger), and report() only formats them for one network.

What the chats see changes in four places:
- The at-close and period headings say "≥" as the others do (≥ 5% at close, Price movement ≥ 5%).
- The top list's chart says "% change over the past 3 months", not "in 3 months".
- With fewer than twice `top` stocks, the Bottom list no longer repeats the Top list's stocks, and with none left
  there is no Bottom list and no gap in the chart.
- One stock with no change over a period is said to have none over that period, not "no interday price".

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/charts.py           |   4 +-
 lib/reports/price.py    | 210 ++++++++++++++++++++++------------------
 tests/test_edges.py     |   4 +-
 tests/test_providers.py |   6 +-
 tests/test_reports.py   |  15 ++-
 5 files changed, 136 insertions(+), 103 deletions(-)

diff --git a/lib/charts.py b/lib/charts.py
index 4a25d46a..765bc181 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -37,7 +37,9 @@ def price_list_chart(service, period, days, threshold, top):
 	""".price: the moves past `threshold`, or the top and bottom `top` ('0': not a top list), over the period; period
 	'd' is today's moves, the command's default view, not a lookback of one day."""
 	span = {'mode': 'interday'} if period == 'd' else {'mode': 'period', 'days': days or buttons.MAX_DAYS}
-	payload, chart = price.report(service, threshold=float(threshold), interactive=True, top=int(top) or None, **span)
+	threshold, top = float(threshold), int(top) or None
+	found = price.moves(threshold=threshold, interactive=True, **span)
+	payload, chart = price.report(service, found, threshold=threshold, interactive=True, top=top, **span)
 	if not chart:
 		raise answer.Answer(*(payload or ['no price data']))
 	return '\n'.join(payload), chart
diff --git a/lib/reports/price.py b/lib/reports/price.py
index d827a22e..229cd110 100644
--- a/lib/reports/price.py
+++ b/lib/reports/price.py
@@ -1,7 +1,8 @@
 """.price, .session, .premarket, .postmarket: price moves past a threshold (scheduled: price.py midsession | interday |
 premarket | close | <days>)."""
-import sys
+import dataclasses
 import datetime
+import sys
 
 from lib import config
 from lib import sharesight
@@ -16,15 +17,19 @@
 from lib.reports import history
 
 def gather(specific_stock=None, days=None):
-	"""(tickers, market data) for a price report on the stocks followed or one (Yahoo's symbol): Yahoo's quotes; over
-	`days` each holding's Sharesight figure, and for the tickers Sharesight has none for (the watchlist) their Yahoo
-	history, fetched together up front."""
+	"""(tickers, market data, the line chart or None) for a price report on the stocks followed or one (Yahoo's
+	symbol): Yahoo's quotes, and over `days` each one's change as percent_change_period: a holding's from Sharesight,
+	the others' from Yahoo's history, fetched together up front, with the line chart of the one stock asked about
+	(performance_use_sharesight: Sharesight's figures only, and the one stock's even when the others take Yahoo's)."""
 	tickers = watchlist.covered(specific_stock)
 	market_data = yahoo.fetch(tickers)
 	if config.prefs.demote_volatile: # trigger_bar compares market caps, in US dollars
 		yahoo.add_usd_market_caps(market_data)
+	chart = None
+	if not days:
+		return tickers, market_data, chart
 	# Sharesight reports a holding's performance only for the time it was held: held for 1Y and asked for 5Y, it gives 1Y
-	if days and (not specific_stock or config.prefs.performance_use_sharesight):
+	if not specific_stock or config.prefs.performance_use_sharesight:
 		for portfolio in sharesight.portfolios():
 			for holding in portfolio.performance(days)['report']['holdings']:
 				ticker = markets.transform_to_yahoo(holding['instrument']['code'], holding['instrument']['market_code'])
@@ -32,19 +37,33 @@ def gather(specific_stock=None, days=None):
 					market_data[ticker]['percent_change_period'] = float(holding['capital_gain_percent'])
 				elif ticker in tickers:
 					print("Notice: price.py:", ticker, "has no data", file=sys.stderr)
-		if not specific_stock and not config.prefs.performance_use_sharesight:
-			yahoo.prefetch_history([t for t in market_data if 'percent_change_period' not in market_data[t]])
-	return tickers, market_data
+	if config.prefs.performance_use_sharesight:
+		return tickers, market_data, chart
+	missing = [ticker for ticker in market_data if 'percent_change_period' not in market_data[ticker]]
+	if not specific_stock:
+		yahoo.prefetch_history(missing)
+	for ticker in missing:
+		try:
+			changes, line = yahoo.price_history(ticker, days)
+		except RuntimeError as e: # no history: the one stock asked about says so, a list leaves it out
+			if specific_stock:
+				raise
+			print(ticker, e, file=sys.stderr)
+			continue
+		market_data[ticker]['percent_change_period'] = changes[days]
+		if specific_stock:
+			chart = history.line_chart(ticker, changes, line, days)
+	return tickers, market_data, chart
 
 def needs_beta(quote):
 	"""Whether trigger_bar must know this ticker's beta: a small cap between US$150M and US$1B, which counts as volatile when its
 	beta is over 1.5. The beta is only in Yahoo's details."""
 	return config.prefs.demote_volatile and 1.5e8 <= quote.get('market_cap_usd', float('inf')) < 1e9
 
-def trigger_bar(ticker, quote, threshold, details=None):
+def trigger_bar(ticker, quote, threshold, details):
 	"""The move this ticker needs before it alone sends a scheduled report: more for leveraged ETFs (demote_leveraged) and volatile
 	small caps (demote_volatile; their market caps in US dollars), else the threshold itself. details: Yahoo's details of the
-	tickers that need_beta, fetched together; without it this ticker's are fetched here."""
+	tickers that need_beta, fetched together."""
 	title = quote['profile_title'].lower()
 	if config.prefs.demote_leveraged and ('2x' in title or '3x' in title):
 		return threshold * 1.3
@@ -54,7 +73,7 @@ def trigger_bar(ticker, quote, threshold, details=None):
 			return threshold * config.prefs.volatility_multiplier * 1.3
 		if cap < 1.5e8:
 			return threshold * config.prefs.volatility_multiplier
-		beta = (yahoo.fetch_detail(ticker) if details is None else details).get(ticker, {}).get('beta')
+		beta = details.get(ticker, {}).get('beta')
 		if beta is None or beta > 1.5:
 			return threshold * config.prefs.volatility_multiplier
 	return threshold
@@ -62,105 +81,104 @@ def trigger_bar(ticker, quote, threshold, details=None):
 # today's moves, in session, out of it (pre- or post-market), at the close, over `days`
 MODES = ('interday', 'midsession', 'premarket', 'close', 'period')
 HEADINGS = { # the report's heading, by mode
-	'interday': '{threshold:g}% at close, {on}',
+	'interday': '≥ {threshold:g}% at close, {on}',
 	'midsession': '≥ {threshold:g}% mid-session, {on} ({where})',
 	'premarket': '≥ {threshold:g}% pre-market, {on} ({where})',
 	'close': '≥ {threshold:g}% at close, {on} ({where})',
-	'period': 'Price movement {threshold:g}%, {span}',
+	'period': 'Price movement ≥ {threshold:g}%, {span}',
 }
 
-def report(service, mode='interday', threshold=None, days=None, top=None, user='', specific_stock=None,
-		interactive=False, market=None):
-	"""(payload, chart) of the price report for one service: the moves of `threshold` percent or more (or the top and
-	bottom `top`) in a mode of MODES ('period': over `days`). A scheduled report goes out only when one ticker passes
-	its trigger_bar; it then lists every ticker past the threshold. market: gather()'s (tickers, market data), fetched
-	when not given."""
+@dataclasses.dataclass
+class Moves:
+	"""What a price report shows, the same for every chat (moves() works it out once): the moves that pass, (ticker,
+	percent, quote) best first, the countries of their exchanges, whether a market of the stocks is in session, the date
+	of the freshest price, and the one stock's line chart, if any."""
+	rows: list
+	exchanges: set
+	in_session: bool
+	as_of: datetime.date
+	chart: object = None
+
+def change(mode, quote, now, interactive):
+	"""A stock's move in a mode of MODES, or None when it has none to report: not in session (midsession), quoted more
+	than a day ago (a market holiday: no repeat of the last session's moves), its market not just closed (close), no
+	price out of session (premarket), no change over the period (period)."""
+	state = quote['marketState'] # PREPRE PRE REGULAR POST POSTPOST CLOSED
+	age = now - datetime.datetime.fromtimestamp(quote['regularMarketTime'])
+	if mode == 'midsession':
+		return quote['percent_change'] if state == 'REGULAR' else None
+	if mode == 'interday':
+		return quote['percent_change'] if interactive or age <= datetime.timedelta(days=1) else None
+	if mode == 'close':
+		just_closed = age <= datetime.timedelta(hours=1.5) and state not in ('PREPRE', 'PRE', 'REGULAR')
+		return quote['percent_change'] if interactive or just_closed else None
+	if mode == 'premarket':
+		return quote.get('percent_change_premarket', quote.get('percent_change_postmarket'))
+	return quote.get('percent_change_period')
+
+def moves(mode='interday', threshold=None, days=None, specific_stock=None, interactive=False):
+	"""The Moves of a price report in a mode of MODES ('period': over `days`): each stock followed whose move is
+	`threshold` percent or more, or the one stock asked about, whatever its move. A scheduled report keeps them only
+	when one passes its trigger_bar."""
 	if threshold is None:
 		threshold = config.prefs.price_percent
-	tickers, market_data = market or gather(specific_stock, days)
+	_, market_data, chart = gather(specific_stock, days)
 	now = datetime.datetime.now()
-	rows = [] # (ticker, percent, report line without its percent)
-	exchanges = set()
-	states = []
-	latest = None # date of the freshest price, for the heading
-	graph = None
+	rows, exchanges = [], set()
 	for ticker, quote in market_data.items():
-		state = quote['marketState']
-		states.append(state)
-		quoted = datetime.datetime.fromtimestamp(quote['regularMarketTime'])
-		latest = max(latest or quoted.date(), quoted.date())
-		if mode == 'midsession' and state != 'REGULAR':
-			continue # not in session (states: PREPRE PRE REGULAR POST POSTPOST CLOSED)
-		if mode == 'interday' and not interactive and now - quoted > datetime.timedelta(days=1):
-			continue # a market holiday: no repeat of the last session's moves
-		if mode == 'close' and not interactive and (now - quoted > datetime.timedelta(hours=1.5) or
-				state in ('PREPRE', 'PRE', 'REGULAR')):
-			continue # its market did not just close
-		if mode == 'premarket':
-			percent = quote.get('percent_change_premarket', quote.get('percent_change_postmarket'))
-			if percent is None:
-				continue
-		elif mode == 'period':
-			percent = quote.get('percent_change_period') # Sharesight's figure, for a holding
-			if percent is None:
-				if config.prefs.performance_use_sharesight:
-					continue
-				if specific_stock: # the one ticker asked for: no history is reported in one line
-					changes, line = yahoo.price_history(ticker, days)
-					graph = history.line_chart(ticker, changes, line, days)
-				else:
-					try:
-						changes, _ = yahoo.price_history(ticker, days)
-					except RuntimeError as e: # no history for this ticker: leave it out of the list
-						print(ticker, e, file=sys.stderr)
-						continue
-				percent = changes[days]
-		else:
-			percent = quote['percent_change']
-		percent = float(percent)
-		if abs(percent) >= threshold or (specific_stock and interactive):
-			exchange = quote['profile_exchange']
-			exchanges.add(markets.exchange_human(exchange))
-			link = links.finance_link(ticker, exchange, service, days=days)
-			rows.append((ticker, percent, f"{formats.get_emoji(percent)} {quote['profile_title']} ({link})"))
+		percent = change(mode, quote, now, interactive)
+		if percent is not None and (abs(float(percent)) >= threshold or (specific_stock and interactive)):
+			exchanges.add(markets.exchange_human(quote['profile_exchange']))
+			rows.append((ticker, float(percent), quote))
 	if not interactive: # scheduled: is a move big enough for its ticker? Small caps' betas are fetched together
-		details = yahoo.fetch_details([ticker for ticker, _, _ in rows if needs_beta(market_data[ticker])])
-		if not any(abs(percent) >= trigger_bar(ticker, market_data[ticker], threshold, details) for ticker, percent, _ in rows):
+		details = yahoo.fetch_details([ticker for ticker, _, quote in rows if needs_beta(quote)])
+		if not any(abs(percent) >= trigger_bar(ticker, quote, threshold, details) for ticker, percent, quote in rows):
 			rows = []
 	rows.sort(key=lambda row: row[1], reverse=True)
-	payload = [f"{line} {formats.signed_percent(round(percent))}" for _, percent, line in rows]
-	chart_rows = [(ticker, percent) for ticker, percent, _ in rows]
-	label = lambda chart_rows: [(ticker.split('.')[0].removesuffix('-USD'), percent) for ticker, percent in chart_rows]
-	if payload and not specific_stock:
-		as_of = latest or datetime.date.today()
-		on = formats.date_short(as_of, True)
-		span = formats.date_range_english(days, as_of) if days else on
-		if top:
-			heading = f'Top {top} performers {span}'
-		else:
-			heading = HEADINGS[mode].format(threshold=threshold, on=on, where=", ".join(exchanges), span=span)
-		if top:
-			bottom = [service.bold(f'Bottom {top}')] + list(reversed(payload[-top:]))
-			payload = [service.bold(heading)] + payload[:top] + [''] + bottom
-			chart_top = chart_rows[:top]
-			chart_bottom = [row for row in reversed(chart_rows[-top:]) if row not in chart_top] # worst first
-			if chart_top: # the best, then after a gap the worst, as the caption's two lists
-				change = formats.days_english(days, 'in ', 'the past ')
-				graph = render.Chart('bars', f"Top & bottom {top}", f"% change {change}",
-					(*label(chart_top), ('', None), *label(chart_bottom)))
-		else:
-			payload.insert(0, service.bold(heading))
-			session = mode if mode in ('midsession', 'premarket') else None
-			graph = render.Chart('bars', heading, rows=tuple(label(chart_rows)), session=session)
-	elif not payload and interactive:
-		payload = [nothing(mode, threshold, user, specific_stock, 'REGULAR' in states)]
-	return payload, graph
+	stamps = [quote['regularMarketTime'] for quote in market_data.values()]
+	as_of = datetime.datetime.fromtimestamp(max(stamps)).date() if stamps else datetime.date.today()
+	in_session = any(quote['marketState'] == 'REGULAR' for quote in market_data.values())
+	return Moves(rows, exchanges, in_session, as_of, chart)
+
+def report(service, found, mode='interday', threshold=None, days=None, top=None, user='', specific_stock=None,
+		interactive=False):
+	"""(payload, chart) of a price report in one network's markup, from moves()' Moves: the moves of `threshold` percent
+	or more in a mode of MODES ('period': over `days`), or the top and bottom `top`."""
+	if threshold is None:
+		threshold = config.prefs.price_percent
+	lines = [f"{formats.get_emoji(percent)} {quote['profile_title']} "
+		f"({links.finance_link(ticker, quote['profile_exchange'], service, days=days)}) "
+		f"{formats.signed_percent(round(percent))}" for ticker, percent, quote in found.rows]
+	if not lines:
+		return [nothing(mode, threshold, user, specific_stock, found.in_session, days)] if interactive else [], None
+	if specific_stock:
+		return lines, found.chart
+	label = lambda rows: [(ticker.split('.')[0].removesuffix('-USD'), percent) for ticker, percent, _ in rows]
+	on = formats.date_short(found.as_of, True)
+	span = formats.date_range_english(days, found.as_of) if days else on
+	if not top:
+		heading = HEADINGS[mode].format(threshold=threshold, on=on, where=", ".join(found.exchanges), span=span)
+		session = mode if mode in ('midsession', 'premarket') else None
+		chart = render.Chart('bars', heading, rows=tuple(label(found.rows)), session=session)
+		return [service.bold(heading)] + lines, chart
+	worst = [i for i in reversed(range(len(found.rows))) if i >= top][:top] # the worst, worst first, none of the best
+	payload = [service.bold(f'Top {top} performers {span}')] + lines[:top]
+	if worst:
+		payload += ['', service.bold(f'Bottom {top}')] + [lines[i] for i in worst]
+	period = formats.days_english(days)
+	subtitle = f"% change {'' if period == 'today' else 'over '}{period}"
+	bars = label(found.rows[:top]) # the best, then after a gap the worst, as the caption's two lists
+	if worst:
+		bars += [('', None)] + label(found.rows[i] for i in worst)
+	return payload, render.Chart('bars', f"Top & bottom {top}", subtitle, tuple(bars))
 
-def nothing(mode, threshold, user, ticker, in_session):
+def nothing(mode, threshold, user, ticker, in_session, days=None):
 	"""What a chat is told when nothing meets the threshold, or the one ticker asked for has no price in this mode."""
 	if ticker:
 		if mode == 'midsession':
 			return f"{ticker} not found or not in session"
+		if mode == 'period':
+			return f"No price change found for {ticker} over {formats.days_english(days)}"
 		kind = 'premarket' if mode == 'premarket' else 'interday'
 		return f"No {kind} price found for {ticker}"
 	if mode == 'premarket':
@@ -177,19 +195,19 @@ def run(dest=None, *, mode='interday', threshold=None, days=None, top=None, user
 	interactive = dest is not None
 	if threshold is None:
 		threshold = config.prefs.price_percent
-	market = gather(specific_stock, days)
+	found = moves(mode, threshold, days, specific_stock, interactive)
 	def markup(chat):
 		"""A chart's period buttons, where the chat takes buttons: one ticker's chart, and the daily and N-day lists,
 		but not the intraday ones."""
 		if not chat.takes_buttons:
 			return None
 		if specific_stock:
-			return buttons.keyboard('p', market[0], buttons.period_for_days(days))
+			return buttons.keyboard('p', [specific_stock], buttons.period_for_days(days))
 		if mode in ('interday', 'period'):
 			return buttons.keyboard('l', [f"{threshold:g}", str(top or 0)], buttons.period_for_days(days, 'd' if not days else None))
 		return None
 	for chat in delivery.targets(dest):
-		payload, graph = report(chat.dialect, mode, threshold, days, top, user, specific_stock, interactive, market)
+		payload, graph = report(chat.dialect, found, mode, threshold, days, top, user, specific_stock, interactive)
 		delivery.deliver(chat, delivery.Report(payload, graph, markup(chat) if graph else None))
 
 def job(argv):
diff --git a/tests/test_edges.py b/tests/test_edges.py
index b6837b0f..382dfc65 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -239,8 +239,8 @@ def test_keywords_in_any_case(self):
 	def test_a_bare_number_on_price_is_a_percent(self):
 		"""As on .session, .premarket and .shorts, and as the README has it; a period carries its unit."""
 		self.assertIn('3% at close', str(self.command('.price 3'))) # it used to be a 3-day report
-		self.assertIn('Price movement 3%', str(self.command('.price 3 7d'))) # the 3 used to be dropped
-		self.assertIn('Price movement 5%', str(self.command('.price 7d')))
+		self.assertIn('Price movement ≥ 3%', str(self.command('.price 3 7d'))) # the 3 used to be dropped
+		self.assertIn('Price movement ≥ 5%', str(self.command('.price 7d')))
 
 	def test_top_is_gone(self):
 		""".top is a ticker now, like any word: the command is .price top."""
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 58cdb73c..c72e8624 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -685,7 +685,7 @@ def test_slack_replies_upload_their_charts(self):
 		"""A Slack reply uploads its chart in three steps (bot token), the report as its message, with no text copy."""
 		self.env(slackBotToken='xoxb-test')
 
-		for cmd, heading in (('.beta', '*Beta over 1.5'), ('.price', '*5% at close'), ('.compare T1 T2', '*')):
+		for cmd, heading in (('.beta', '*Beta over 1.5'), ('.price', '*≥ 5% at close'), ('.compare T1 T2', '*')):
 			with self.subTest(cmd=cmd):
 				calls = self.slack_answers(*self.upload_steps())
 				self.sent.clear(); self.errors.clear()
@@ -829,7 +829,7 @@ def test_a_scheduled_report_carries_its_chart(self):
 		self.assertEqual(url, self.URL + '?wait=true')
 		payload = json.loads(kwargs['data']['payload_json'])
 		self.assertEqual((payload['flags'], payload['allowed_mentions'], payload['attachments'][0]['filename']), (4, {'parse': []}, 'chart.png'))
-		self.assertTrue(payload['content'].startswith('**5% at close'), payload['content'][:60])
+		self.assertTrue(payload['content'].startswith('**≥ 5% at close'), payload['content'][:60])
 		self.assertLessEqual(len(payload['content']), 2000)
 		name, image, kind = kwargs['files']['files[0]']
 		self.assertEqual((name, kind), ('chart.png', 'image/png'))
@@ -980,7 +980,7 @@ def test_a_chart_reply_carries_its_period_buttons(self):
 		self.assertNotIn('channel_id', self.posts[2][1]['data']) # in no channel: the message shows it
 		blocks = self.blocks('chat.postMessage')
 		self.assertEqual([block['type'] for block in blocks], ['section', 'image', 'actions'])
-		self.assertTrue(blocks[0]['text']['text'].startswith('*5% at close'))
+		self.assertTrue(blocks[0]['text']['text'].startswith('*≥ 5% at close'))
 		self.assertEqual(blocks[1]['slack_file'], {'id': 'F1'})
 		buttons_shown = blocks[2]['elements']
 		self.assertEqual([button['text']['text'] for button in buttons_shown], ['1D', '7D', '1M', '1Y', 'Max'])
diff --git a/tests/test_reports.py b/tests/test_reports.py
index dd22820b..c418ac3c 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -127,7 +127,20 @@ def test_a_small_cap_quoted_in_won_needs_the_bigger_move_too(self):
 
 		quote = price.gather()[1]['T1']
 		self.assertAlmostEqual(quote['market_cap_usd'], 1e11 / 1380.0) # US$72M
-		self.assertEqual(price.trigger_bar('T1', quote, 5), 5 * config.prefs.volatility_multiplier)
+		self.assertEqual(price.trigger_bar('T1', quote, 5, {}), 5 * config.prefs.volatility_multiplier)
+
+	def test_a_short_top_list_repeats_none_of_its_best_as_the_worst(self):
+		"""With fewer than twice `top` stocks, the Bottom list leaves out those in the Top list (it repeated them), and
+		with none left there is no Bottom list and no gap in the chart; the subtitle says "over the past 3 months"."""
+		for count, bottom, bars in ((12, 2, 13), (8, 0, 8)):
+			with self.subTest(count=count):
+				self.patch(watchlist, 'with_holdings', lambda: TICKERS[:count])
+				found = price.moves('period', 0, 90, interactive=True)
+				payload, chart = price.report(TELEGRAM, found, 'period', 0, 90, top=10, interactive=True)
+				shown = payload[payload.index('<b>Bottom 10</b>') + 1:] if bottom else []
+				self.assertEqual((len(shown), len(chart.rows)), (bottom, bars))
+				self.assertEqual(len(payload), 1 + min(count, 10) + (bottom and 2 + bottom))
+				self.assertEqual(chart.subtitle, '% change over the past 3 months') # it said "in 3 months"
 
 	def test_the_small_caps_betas_are_fetched_together(self):
 		"""A scheduled price report asks for the beta of each small cap that moved (US$150M to US$1B): together, not one by one."""

From e3cc79a9ff0667a1f3b72ed9bd2d9b433c2ae4fe Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 09:10:04 +0000
Subject: [PATCH 58/81] tests: a golden image for the pie, before the shared
 chart heading changes

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 tests/golden/pie.png | Bin 0 -> 7312 bytes
 tests/test_charts.py |   6 ++++++
 2 files changed, 6 insertions(+)
 create mode 100644 tests/golden/pie.png

diff --git a/tests/golden/pie.png b/tests/golden/pie.png
new file mode 100644
index 0000000000000000000000000000000000000000..822c096c9cc44eb3a6a5ed9e6e2edae67865b080
GIT binary patch
literal 7312
zcmV;B9B<=^P)<h;3K|Lk000e1NJLTq005u>003kN0ssI2j?}E!0010UNkl<Zc-rlK
zX>eTEmEJk`-nVr(x&bsc0^lN&kVuLWNm-+XvSmruW?Pgr8QBwiD$Z2Adn&F>D)EnO
zPNl|4l`B7zj3<?`#~#}mmm<$3UKCmK63g-;QrrwK5=jx<31T67?|$1|PJTQP0tk>2
zC7?kMaDM=eSFand?|063mV55G*wM2ZrVT_SBF3081`z>(h>|Nrl>CW^$!&-T$yI>l
ze#Y3;J=5Pfb$@atxhMILQ`cd6EWqHYOzZT&l~Rmhz2Uo#6UQ+TS(ZgaQc58N0CCQZ
zF-j>!G=?rg%RJ8ofGCQzF@kfgwbsVAEh!};YONV#aV&|*4+13>V@zwUloCQHr2sLJ
zr{%Fw03)JO>Wg3ew@*I)*y=T_UwrWw0|Nu~dVTZen?L>OPww8mM=4dU)t-2K#|J<7
z{@uIxc%H{O=Zv*vTS~>^M?U%yM66Y-9r=8@QUQR6AA0aZAA0IHzuB3|WIWG{qNuNL
z1tNa(<NqX=%Z`qY=JWY{zP((oWV2a@m<Y`>Bnt?z*5*^6{6t$@u2imY&j0kEeg2L+
z?uf$Zz4zX`ef!-~%EN>E@7{jTBM(2gZQIsWt5?74wrzLZd4~|z!NI-1{|A5IdERF}
z`{@sU;L(QffApjOaQj_%24S>q+ZM+76Hn||+28-U&whG*eC!jS__$Vj`}XaRJo3=y
zn>TlLbv^mS6Ne8Uu2idn^JPfpcd(fT7&zyr&zyA}=iT?*Gd_N?wY7Eq`t=iqLg4#1
zZ`wpAPN&nOqhq;T?)>@l{rxNNd+)u!{N=Bh<+ZoB_x7$B85#Nci@$i_{`>lRdrzG{
z?RlQ#IFCN^&@G!cpE-Nh5BzQ0whay*IQ;72TrT^g=U@2Z7ytd|KKmJ^)zHuh*Kw9D
z@q*jnSjumF<D0Fy-19H|xT~{c_wGH<KmS55mm3=&d-~~Tw81b6Q|Z*uiIacxKfm+f
zgAbfKb^6(7|8Fjr3&U_|=)}pBC(oZdKX~B4p+m2Bb#}b?i<frp+?mg}zr1tT=~Jh*
z*8BGTcJRP~wKuHYxBs`NPMsPbAA9o2C%*g4cZW`#aFzjHfPvG!Xhf`(%jrxy2tvnk
z>h-#9+m>ZD8V$x+CX?ZuhhY$gkuj#Hr)S5G$6k2h$JJ^zxhJuphzNkj=r9aj*As%*
z>kZ2ij^jj8#5qS~aU9vUZH&3;rW?0z-TM9S|9~%(^FjlRh;bbE_Vvp+ZZztaWlJfe
zD9q<O;yCvGIsjUhold7~wQ8oNrBp1|>-9_~!#Fou&qio^qy#|5yb)_n#YJ%(hhdn>
zWR{UwpdqGS8DnhQzIDrOCr=#reShGl4TvlV{I<4yqfxs!KGw6MFOI@gIuitb9LG0p
z+^l45j5#@UyjUzufj4_-tuK2qH7v_=9B20R<*^_(m@#JCw(tACU(aM(#>dA}sg%@8
z#&NE-H3$MC@_oNrDF=a{PG?HRA~GgrT&-5#Hped8-R9X~B2rRvA*M1}#uy<)lBP(s
zoO4b@#uy?>K%Go!EsyJ*q7&0-J5G|2Oc3Q<OeGf+6;Bk7F$Rcm$)GOJ<L?H*a}g$8
zg<F@W?{^R2rkRFmYLAJqAhJthLjVMHRq<LL*Bih|e4MOK8KV<3o~%_NV}g;E0U`rL
zK#q(f6Trl!6)m$&7}qDjlbULbkx~*FL}po*=cQcFb6wB2ZO8GVpf(=5UjJ=6D4V#U
zgOX|#wO=$*O{)kIQ8>sQOgsQG%djmzzzMKgYo#P2OQq7?J-r<rU2XY%Hk<Rjv~Am*
za{#m*Z|vOBm&e-s9{*3OQ36EL(!{>@i?T8tO$>!2hr`j=;__J?`p9i&dC1tZPA?k3
zNtP;(V<2kD=2rCf_pe-?@90dWGF)&nWQ;LJCu__^L;!IdlZupI(Ln{d07Mjy)!J|8
zZ^*2_kI3kt99=x_pWanJ@lrVUnht$tr<t&qVQwnG3A9lZ2_gFXSFK%tQ*U2?DwQE3
zt(8(zsTg1?Up@7cAtE3m<H$H50{|dRN|TC+2oaE5ss7v2tMAO;{mEeDQ1!LvYR7&m
zE9aS&VwOwBEW2Y13vdEe97jS}>jyS$*m!eiR}W(>$z~@vTynDB>PJqK1OOrvHxhSj
z3hC<MXG??6#D$aGX+dt0S<ZJZ5Wt91N&s53X5Hp3w{>>)kTFuqX}Hdd5V1*rEp8Bz
zNaZ`;{U^CCPn8FsDen1}u3uzs4gi*2uz3YIDY{d-tv$c(wmVj>UQ1-+C|cO`7BWO2
zEtOwjcB=C|pU-Z3Y~t5nsT}+sa*J8+vMNs>>>a!nAy$MSvZ?#;f8>U>19224C3ek7
zULqnBKw9|)E8Eli_<!zy>c5MY9vzgCEiWy<bAUO6KthXafA)U$;jM@E?^SwHDkBk@
zAQkDL+_K@3HUId3wQTyJ@=JgS%Ng>w4=_Whg*eb9{>8(vw!_)7dT50@{pu0N^T;fQ
z2}FjB>!8ftT>n4#^R5TJU}7JPo^;5U0lsE*8xh;SPJHe@%(@W75K4Q;)_tQ#>V*=w
zti=jUk#W+oiE5qqe!lO?{{qNJ$AEkp;A=#;9<c{E@sHgNL=bD>jFb@>&wqIRzP&p+
z=ZhJTA_HWqF_GK+_{tA|m02lL;S$G~T{F7x-^`zS7X%Uv0mJ0FsfXT{HK$hLi35k6
z)FKh}8z>9q7g9Iez48yg%Iq|$a4B;&yEb&+w}C&o4g6>_Mw;exkssTSZQVEVdTqSG
zZEJB0agYh^mr^U=)&JCAps>NnC5!P}``VmAJz{s>z@NGu{P<FuX~rOJglC8EziH3D
zo%}jrMPx!XO6fJ*dv|=9v`SW-mjS+l?)otfbn%be0kHz0xhIOD@ndiEs<Z2?<AZ~X
zPSgN^TqwVg-Sok(2fm=g3Np3~@a#zfkOo>D{*Uegi-R^-i!T|PdgOdy%l`6-%IHO5
z+lyW?iA<<QvHh-3v~K;7@=M6XGQhJCR+@kKcI?hVC?~zDs|;hfZN(qhxM%+^oQug9
z0b@iHH@YAC*KYR)6E~KcF=lT@&Jo+bj@`Zn8Zq#<#OKD3+>QO`H@L_4?{~b^q7!ve
zLBQQy_rw1V0GHbOF}upVfYjKT<&WJ8p}dC7A7`c>I*;Bu*f>)f9udy<K-64^mCV}r
zwB7zkI;<?IWHr00_0{YHo1rxY%3K@S8sb98$2RRb@Cq2X9@4^yTu1f%_RrewYm5q)
zPNte&k(>jGH+Hkz*MJ|rx#^vgs0Vt7Z_XUux6koBb6pU1WVLkVw||;sh)c#2U*UTa
z00Mt-BMA03i(xo3_0WCzmIKkb;<;h#I@tS*Oz5zh+x)n@;#L#;OT^n^bE|JCaYGMo
zTmeCR&FcGxln`f3?b!Uv!QG(s^{~T*jOKQx{r5gbI=<rbAe*cXyX1sVH5R(cXntrg
z+se%O7*~>FV)v{E&fZZi8=?uMyS8U^Ywp!Od#?wgM#gniZ@K9~r)z_WgSjj*nz*j~
zk`Bs1h6tpTmSx3pETs&?FbqSjb<!Uw8B)rmKB<{QD`m7!wg67Gw^B;Y!K9b9Z~#Pz
z752CB9a}(|cj}BVq)1se_KfU(bp>Tw@@-mOuF9N?Mzt{Pdo2U^lU(qH0Ww;o+D%xk
zpLmhmsToldBI!7P=O@~3|D(*h`=YTUI;=dlV~6iI5FwY#edt3^S;8unO0Mg=j^ntl
z=egNzODdI)qNr)QD`h-z<Hk%TvwmPhrCe;!cigacJ!7n=w@*nK1pf4`ykc(WdD|*T
z*-($(K0i;JQY56uZ`u8wy^nN%02wZBg)1U6(r?-DaN(7|BAF~EPCXEkQJI1JTL#{z
z!%BJo)24jB;RoBd-z{T_i0M?izrXLc+qT)ZeSBzW^XAR3A3K)K<@)-1zxK7SS1J`D
z#KO87h+6Axu66xQ8z#ob+w=L>w)Rr7xN_AR$MxFt`Ibz}tFIoE@f!wc7Qos-#>HFu
zAXf8}heD(<c2;#=ynXc0p53?IdvDNautiJ@BN`Ri?Kh-W-B~;GJoj=li}cV$b>$aK
zSR>K^Fg!dQ_zfwg@B8(7Ba=>F7#T6fgkgvPM~@vd#;jVk%J==N^f;J5{vpO#7z77@
zyN@wesg#vcmTeab7sp11%f&)VHh0-O#H{E@;f7A^&O)r_Em-8t)C2eaO^1hm*?01E
z$Lf_biWbx7B0^*>0}s@WJbxK|GvSQ=>_0Ng12VEYA}sg#>#xUg%ov+;rjmveh&VWS
zz_Kg=ICJK#ZQD)zM#h-ob7wErjU!{MQYtdegCHmt3!L-mPa<Z&n{8YH7Kcd9ZzdTH
z!Q;2S^6kBkc6TFZi<mM*WJZPF%6IWh2gUvjmPVB2^GR_vavKmiW2saMrtO@zZ5w8_
z*lcDHh=`VLPZL)XmX++&Px>!h*^4#=j)m(wLF##4CqSf7Ieq!UUEK%wykfiVVtUkw
zq#`TV?{;rA^75<}6E?RzZh2GJr&?A_jpWbi7Z-zuXr^G{oKM1L_|HmXM+UoEK<W8U
z=5aLj(7ku#Yff?G_zBx{7cuogL`>LT?^ZIGH^_`N1=<AKBozWt6vc6z@H1nR6wMa3
zbt(}_85?8bI3i<`ZG6V4MBuIhFn}`Hn}f6idZDHsmBA;r?0)^X$`2RO)j|X^+U>ms
z`3%380Qx+<sRaizXfsg83=FJawQ6M)MNt$5L5PT+=Y>%;XVBRkbcGNrdi&Fv%<463
z-Bil5tnQu_EiJiJI-RrzzOs{L2zp!RBQ&})H^$2DX)C>Z#o+E;ci#U%;QNb}!Xqv1
zwl&<%k%|&C%n*D9|7wGHcPFb!_}HF!sD}5w_ug8)(bLnzId>hGah6JXXU?2$$z*o!
zdPNA{WEx_O#c`a?wQ?ca+Vizq)oIDzy7gV7Bg3gwdTey$?3q(nw!xfXcNWY-J~Bg7
z3%z?b9L?0uA3JV)E-lJ-F+?&_WI9Eri}V}u0s&+M<ZWoPfFlq<5ClMU>n*oz+qNx?
zLeF)(y1G-T)Vg)+Y}+=b*;d9F<GNlx-&w0v8jX73`{{H>#?k1=aIsj(cXYZlGlpV@
z`r4eedZ08Aftg0#$uFEgevB1=V#}Vd?|pQ|3a2HrIH?b#nVq(>J<-MEcnMvB@Zv=?
z98##ECV>I`<R?FqN+teF6h%n~u~f=CcI<V%NC|dh^sys{V;P@4eF~9X7(O>PIub=u
zsaWKkE2XB}KJf;MnkZ%K&MbE=v;;!{ATR(#U<eEmUHUn@h0-+%0Rl;Fb>zzTuKew;
zo$q<{vBmO28Mj1BHyMqHL_h>KgRh;^Ns2bj!QnVjC`_T4bg2h`N~H<_i_$<f>Dyi`
zOfb$7G0DadMG2_M=D)dYFaaTOrM<u16Q-Qlb5zRKo<k`cQ#N`Qb1m)&W^-<F#u;PC
z5;T*EA-G%!XqH+|7lNc#{_cTeM}NKM$WbS=I36`3@>G6$cYuVVlPA>#1dyaME(>;r
z5U_Yn7!gria;B!bBqZNrHlv6)QdkBz*yhjek*ZJCDpe(v6_hoW4J!jHBd;=16IR`}
z0>=(rJMbKxvUTb$2|*I0T@nHyLjZt*6gzI+^UVXduUgAQW>Mygd##rf;1mzde4gn?
zi%d=QVqtt)(_R2V#JQ+-dbe2uPhfA}=_cd|9036k2nfiKHY9Z%t0+*$msIzuR-uZT
z6oQgrWndNLRW54Ws@rzpT9NAno~=@jnG^yGU7NXrJA&JsIh+An{rUP6Hw}J!r?~O=
z7sr{lQ<oRShZ2kdKyU>h#&MjGbHa_bZA&SMfH5@2OxN!w1Dre$V~jCQWD;_oZ8gMn
zb`pq6Q>Y1NH8nv*<cx7>aUqAm5IF*yO9&zrg~~J})f-eRQ$<2qV@X5VurlyTArMw0
zVFJfe`P?bh=x@{$8Emf8Lde7$tcD0uvxhcfZyUtQ{PZ;{`;Wf=(Q2jQx!$3}hdJk*
zb3_QkNJ^Pbr$<Len;*?1;>MeAIr8eEO`C5$d-|kOdhPlFDW%c6P`Fqwm8LIhiP?sd
zUCC7<A^=6x@?>)f0Z)X00Yl_4LkNr^ZH!b=tb;&OLsH$RpvGSBioa?a##{$gABNcG
zKD?zL{Mh{ZI2BBPUtezH#xrNnZom63$8n-K7MydzjW&$)AO7&!*?!GJqbsE}#`N~}
zuUmg(p>VM{F(HIiDVJ8Qz9E}yeeFuig$SVDLRJ+o$vRwJ2+Wm*Krn9ctOu<Ki0E;?
zz8w8cAvQ!!hZjl9%fE6Up&mgHlQEq0U%d2E;0M`kmdI4A)y~e&dcEQM{<f`KdscKO
z3r$VWWN`!xe1G4bT}GSko)ucFvC)wr@H3h06)744LJf-Mfh6Bt2nf`rmoWoE?nWi0
zq@8#Q&D2=wzafM~U~&C__QP`r!4Q-J!(iz62}Bqk9!@Or@bHBs6b{26$qO!qtA&RT
z4JI)gXN&;Oojqe&R;f@ZOiawZTu1~Qp`jo$;BwHk;I?p@5KzkE8C@;Q(6L05Crby_
z%VYFHW6~ED91v}Lh8vXxNQh9W*2tLIx!>lD<o`~0sfdUR!Dj_;Y+4Qx>WUhQn(Q+e
z0-N&|Q&+K%j5V`o9T91N;xd;5Esl<bj+lOVYHjY)av=%M7YkjOdGh15*KozvBm&?F
zH3h{OL~|G8iAXTk3cd{Mh`ed|A|h$2s$(#-)JAKiWUQo0=r<{*oT^!5Y_hmxv6g|S
z&zqDnT<YXI)d6&FfD;W44UB{&nqC&i7;9yb4y(v`^E8-o6E|dS6qz;kPl+g#$+oq3
zw6?ViA>ugh>FK$7(<aV&YilcGtkGx$L6|_kxabNYx_f$wD3fXFS<#zHrPJxm)s_np
zKp8mEAlCd)1+kq=9h955+kgnUEh{6cK8{=@_$H}?-@EJGeSIrs96$fu_YiUG)-8oX
z;gA3LV}(LdYkltQxz5f`KkzRMUpPKAWG}`ZXhc*}c6N2&y7jh$gZs7CYi?Lqn7G*4
z)je|I{Mj?7=LR@Y;@4|NQ?qmE1lT6D@k@ZoVZ+TZ{m8Ajcv^=QW@n~%xU_8B^*qKn
zGNzPLQn{`>JbWRJ<4UEnb=wvx)%f`M*!cMD0;r~`FBhUxE(gBvdEWTgSR6&wY6ThQ
zJG#cl#{%D<3f^X<Ds2YGYE+S6HIbzez!0_xg0R@MO@BmyjE5tKNULNiY06lfKYO}S
zt7@f_Jqx>DdByX*QmK^7<*L={;NSt*b)zWM+AL~C3TmzP?b#)zbX@oN>qmpYFBT@Q
zyj%!C05*f+fKE4Pa}Iq4P2*|<&}st@3Wl1LszgMi;^4v{GBMpFWsDs@xZfB9fR<$?
z$As1Ebs<EdP)Nd(TCJ8iT8o@>GbRi}#+ct|z@-z9W~bF=$RJ0kE7)5yHfy2=BWOce
zi&<gO7}E@diBLGQay}eA!Yuc)n8C7a$MGD;nL3a@Sy8YoD=EDbLR=?DMJ1Zb7@Jbr
z*$>8M+Ctz6znL(xZfers5M(Uw#3gNX6Z(1(87AD|+&<lyKyF>;w<m$_j7q$j1**SW
zXYsMiCOReHV8xtjklU<?bh4I2a8Fj0ilHg3h>1+?`2T|?FfMl!1VS-}pI<aCho%*j
z86wx>8C_LzxZt5M5fe^aJmsI+&Ft%-M@*9eCIW}Uk4AM>0&f;g6M)TGmKri@AhHF<
z*DxzpJNgqf7eD{~2iO339A2%^Zwe+O7Ont(R$#|?D+Vg6EwlzB6DlZH4*!6OMXK0c
zT!5!M+o#WJX`1J1F~l~EbWmB)_9$dDcQW;3Ka0kXGTU1UJ+s){KCOV43+6Wk(_%GC
ziIKGlrGqjoqy{77I<6P@ejB;HL=L9Dn!0@Ll#-@N=?(&D7n*_+09&A$C!@HVtsMGc
zc;O(kQ%mFk9(KjRRKU&xJ$pgt7G8;yY-Otj7!+uM0%0QLj;xIpcK#(=OTzlTi3S@2
zm%~?1=+OY0kl4~htrqgIxKQd#nmd`o%l{)To@T;bIvp$7RXEdSaJoVNbx5Vf!p6ji
zpb1(mW<@2v0NF}1n!CCB$zPTBeuulcC3SKZyQ)?}s}0{hub;hO+U&`K*@eq>WGv>!
zC9Nd_n_uNcZt1Wx_LKhrKuhgBCH7{M12%&{J0efiDa{wk7-NzpwRlQZl#Fp+{W>DU
z?bO&$zZ8xgWp;Y09d7t$00TG-CSv%)AsHL67R(qMB8T%9T~l!|wUugKJZF*47j}KU
zJovXF)ka3a5*u$$hMFKNU~iHB&8yO7lc+C5m}GIDrG|_fz~*P6gR~N<eD%nW#((~0
z?q>hqr=8y>zyP2K+HCmVu=&zam9r+di_W)~OPsMb3{_B_7Y!z(M7phU`sE8h{FlfC
zAX}P;NWWEgMiaE#@QqXY%g0o!y&!swjJ2_NUOKXqQ6kmmpWAo-@BV~T4BT3>=L%ng
zznVDJe=(%LbVOwYFqr4qi2&P#GQni&-8;0q^;0jOd*+iSZlG|N@~P3+7LXW%HXHu(
zr2fJ|#SvUSpRtnxXcLC~(%V1umJE;)>5j^wADsX0r;H3xxIjz#yy$D5%7m-`?!5l^
zJ~<vyn+0k<O!8!Hr3Dxk=xvoek<rL)=CoY=<$oLg{+}k5&r2L#cze2%VEjtKeDoFh
z^Kp~6f#E!a$c89up-4*at`S=@ntQD}tepSf|8)H4UuJd+7+=cF{Eh%l{OYU)6ES^e
zzx=Z!N*l;oNe7O%tHB7GAZv4bGF$l;b(qK~6b|=VYe#>4`dfckJMx@JwN18$SsrtA
zo4po2b{G)CS5N9+O_+avqj+Bz(*zAQ*(TuHKakRbxw5RawBB^1FcFbPZi`fFR5&^D
z>#tM}eh<0Dz1Ag_YF<Eq;gT2KZo|n2eQv*eu#0_aE#KNY2{T=#8^qi;7(9z-RHY^(
zTU^<ZCPPF_INWZPwXx#vzb@|imabo9ZZ5IJ%b?C1z*BVXaR7iHkC|Uhn1{OAM_2K!
zt;|AbXpjcrtszZGw$ju~v5!_3NKI?WLP@<cGREvQaw{&MtM2}0Y4Dl2aEd#bWyM`U
zfTvVBDGm7k1@q%Eb7wnyqL**avn~e+Ak+|>NoN}Ya4x(Jfe<`Lz6|Tw(h7=_7RZPI
zOfbtuZkgEk&+o6k_FV1QPi1A8S)NF>kul5YE-b)PRv7?VEieSXnlLY2G^<j$J<smx
zU|Vw7?V`=V5J&@xKogjm@s%b5WH}5~SVd-`Fen^kg2?Eg?2o?YpV?I(dMO-x&BQ*l
z(<0SIv<z(%0iF(X(hzVwqNmU4XU^$v7X~uymK<(qVXITv>0m}cN&rV-2@xYG18p{2
zVZ}PCkx_B^ysC~w6GP$1;c)b|xO~RM0djj;Yp*Rl>F8O#fS(SK9HS!*L<W=rj*u46
zYC*e=?KWmDNOQCpFof8ECV0K7hWJVgjjUfZQO&3r5m7kE9ZY&HEJwH355UurPV8>d
zwNw!(GuiFa09@K&pZHnmP)r2K1+d9|d_-nBuDxC;dTtMSW}8e4!4{LiUnU7eV9>$I
q_n2i^e>VWWI^MGbre$E}Yy5u#Ay2Qk&t8rI0000<MNUMnLSTZ|@E#8U

literal 0
HcmV?d00001

diff --git a/tests/test_charts.py b/tests/test_charts.py
index b127dd2f..5c1dc982 100644
--- a/tests/test_charts.py
+++ b/tests/test_charts.py
@@ -54,6 +54,8 @@ def compare_series():
 	'price_line': lambda: render.graph(price_frame(), 'Apple Inc (AAPL)', 'USD'),
 	'compare_lines': lambda: render.compare_graph(compare_series(), 'Compare 120 days', 'subtitle here'),
 	'shorts': lambda: render.shorts_chart([('GME', 47.3), ('AMC', 32.1), ('TSLA', 21.6), ('AAPL', 15.2)], 15),
+	'pie': lambda: render.pie_chart([('Software', 41.0), ('Banks', 23.5), ('Mining', 12.0), ('Other', 8.0)],
+		'Rob by industry', '4 industries'),
 }
 # inputs that have broken chart code before or are likely to: sizes, signs, long names, zeros, a single row
 EDGE_ROWS = {
@@ -365,6 +367,10 @@ def test_compare_lines(self):
 		"""The compare chart matches its golden image."""
 		self.check('compare_lines')
 
+	def test_pie(self):
+		"""The pie matches its golden image."""
+		self.check('pie')
+
 
 if __name__ == '__main__':
 	unittest.main()

From 3b7ca94c993d5fab6380caf65f0dc28cab485398 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 09:19:32 +0000
Subject: [PATCH 59/81] Charts: renderers take their spec; titles fit; the line
 chart's figures match the caption
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The renderers take the Chart and are looked up by kind (render.RENDERERS), so draw() no longer unpacks each kind
its own way, and the tests draw specs instead of calling renderers with positional arguments. One time-axes helper
for the line and compare charts, one table of sign colours, one way to turn (date, close) rows into a series, one
shrink-to-fit, and the row limits in one place (render.trimmed).

What the charts show changes, as decided (D22):
- A long title or subtitle is shrunk to fit, where it ran off the right edge (.fpe negative, long company names).
- The line chart's title is the company and the span; beneath it, the price and the change the caption gives
  (164.12 USD ▲ 6.96%), where the title had Yahoo's figures and the subtitle a change of its own that disagreed.
  A null close no longer breaks the line or draws "nan".
- Dates are unpadded (1 Oct, 5 Jul 2026), and the date ticks are muted like the value ticks, both 10 pt.
- The compare chart's axes match the line chart's; crypto labels lose -USD everywhere (BTC, not BTC-USD).
- The shorts chart marks the stocks it leaves out above 24 with a gap row, as the bar charts do.
- .performance sorts one list for its caption and chart.

The golden images of the line and compare charts are redrawn; the pie has one now.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md                      |   4 +-
 lib/charts.py                  |   4 +-
 lib/formats.py                 |   8 ++
 lib/render.py                  | 200 ++++++++++++++++++---------------
 lib/reports/history.py         |  30 +++--
 lib/reports/performance.py     |  16 +--
 lib/reports/price.py           |   8 +-
 lib/reports/rankings.py        |   6 +-
 lib/yahoo.py                   |   6 +-
 tests/golden/compare_lines.png | Bin 10089 -> 9911 bytes
 tests/golden/price_line.png    | Bin 7678 -> 7848 bytes
 tests/test_charts.py           | 157 +++++++++++++++-----------
 tests/test_providers.py        |   3 +-
 tests/test_reports.py          |   2 +-
 14 files changed, 252 insertions(+), 192 deletions(-)

diff --git a/CLAUDE.md b/CLAUDE.md
index 7fb60b3e..f2f8c67b 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -88,8 +88,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   print there (a test checks); only the util/ tools, run by hand, do.
 - A report gives its chart as a `render.Chart` spec, not a drawing: `deliver()` draws it only for a chat that takes
   charts, through `render.draw()`, which caches the image by the spec's hash (and the drawing code's).
-- A chart renderer in `lib/render.py` gets `@_drawn` (the bot draws in one native thread); it must not call another `@_drawn`
-  function, or that thread waits on itself.
+- A chart renderer in `lib/render.py` takes its `Chart` and is listed in `RENDERERS` by kind; it gets `@_drawn` (the
+  bot draws in one native thread) and must not call another `@_drawn` function, or that thread waits on itself.
 - A new command: a `Command` in `COMMANDS` (its names, and what its arguments may hold), whose usage, section and
   buttons make its `.help` lines and keyboard buttons, and the README command list.
 
diff --git a/lib/charts.py b/lib/charts.py
index 765bc181..3524bef8 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -16,12 +16,12 @@ def price_chart(service, period, days, ticker):
 	market_data = yahoo.fetch([ticker])
 	if ticker not in market_data:
 		raise answer.Answer(f"no data found for {ticker}")
-	changes, line = yahoo.price_history(ticker, days)
+	changes, chart = history.price_line(ticker, days)
 	percent = float(changes.get(days, changes.get('Max', next(iter(changes.values()), 0))))
 	link = links.finance_link(ticker, market_data[ticker]['profile_exchange'], service, days=days)
 	title = market_data[ticker]['profile_title']
 	caption = f"{formats.get_emoji(percent)} {title} ({link}) {formats.signed_percent(round(percent))}"
-	return caption, history.line_chart(ticker, changes, line, days)
+	return caption, chart
 
 def compare_chart(service, period, days, *tickers):
 	""".compare: the stocks' changes over the period, rebased on one chart."""
diff --git a/lib/formats.py b/lib/formats.py
index e7ac22c6..1331e7f8 100644
--- a/lib/formats.py
+++ b/lib/formats.py
@@ -113,6 +113,14 @@ def days_english(days, prefix='the past ', article=''):
 			return f"{prefix}{int(days // size)} {units}"
 	return f"{prefix}{days} days"
 
+def price(value):
+	"""A price to show: two decimals, thousands separated (1,234.50); under 1, three significant digits (0.0123)."""
+	return f"{value:,.2f}" if abs(value) >= 1 else f"{value:.3g}"
+
+def short_ticker(ticker):
+	"""A ticker as a chart labels it: without its exchange (BHP.AX: BHP) or a crypto's dollar (BTC-USD: BTC)."""
+	return ticker.split('.')[0].removesuffix('-USD')
+
 def signed_percent(value, decimals=None):
 	"""12 -> '+12%', -3.5 -> '-3.5%', 0 -> '0%'; thousands separated. decimals=None keeps the value's own digits."""
 	if value == 0:
diff --git a/lib/render.py b/lib/render.py
index be8589fd..3649e740 100644
--- a/lib/render.py
+++ b/lib/render.py
@@ -17,8 +17,7 @@
 class Chart:
 	"""A chart not yet drawn, for draw(). kind 'bars': rows of (label, value), first on top, a (label, None) row a gap;
 	'pie': rows of (label, value); 'shorts': rows of (label, percent), and the threshold; 'line': rows of (date, close),
-	a price line, and the unit (its currency); 'compare': rows of (label, ((date, close), ...)), two to six price
-	lines."""
+	a price line, and the change it shows; 'compare': rows of (label, ((date, close), ...)), two to six price lines."""
 	kind: str
 	title: str = ''
 	subtitle: str = ''
@@ -27,7 +26,7 @@ class Chart:
 	xlim: tuple = None # bars: the value axis, zoomed
 	session: str = None # bars: 'midsession' or 'premarket' tints the background
 	threshold: float = None # shorts: the outrageous line
-	unit: str = '' # line: the prices' currency
+	change: float = None # line: the change shown, whose sign colours the line and its subtitle
 
 VALUE_FORMATS = {'percent': lambda v: formats.signed_percent(round(v)),
 	'percent2': lambda v: formats.signed_percent(v, 2), 'number': lambda v: f"{v:g}"}
@@ -43,10 +42,10 @@ def fit_left_margin(fig, ax, pad_px=18):
 		ax.set_position([need, y0, w - (need - x0), h])
 
 def style_date_axis(ax, start, end, ink):
-	"""Larger, explicit date ticks: format follows the span, few enough ticks that they never collide."""
+	"""Explicit date ticks: the format follows the span, and few enough ticks that they never collide."""
 	import matplotlib.dates as mdates
 	days = (end - start).days
-	fmt = '%Y' if days > 1460 else '%b %Y' if days > 150 else '%d %b'
+	fmt = '%Y' if days > 1460 else '%b %Y' if days > 150 else '%-d %b'
 	ax.xaxis.set_major_locator(mdates.AutoDateLocator(minticks=2, maxticks=5)) # minticks=3 warns on a 2-day span
 	ax.xaxis.set_major_formatter(mdates.DateFormatter(fmt))
 	ax.tick_params(axis='x', colors=ink, labelsize=10, length=3, width=0.8, pad=6)
@@ -55,6 +54,8 @@ def style_date_axis(ax, start, end, ink):
 CATEGORIES = ('#3987e5', '#d95926', '#199e70', '#c98500', '#d55181', '#008300', '#9085e9', '#e66767') # the dark
 # categorical slots, in fixed order; the green and red ones mean up and down, so lines that are neither leave them out
 NEITHER_UP_NOR_DOWN = tuple(color for color in CATEGORIES if color not in ('#008300', '#e66767'))
+UP, DOWN, FLAT = '#34d399', '#fb7185', PALETTE[2] # a rise, a fall, neither: a price line's colour by its change
+BAR_UP = CATEGORIES[0] # bars rise in blue; they fall in DOWN
 
 _render_pool = None # set by render_off_hub(): the bot draws its charts in one native thread
 
@@ -88,6 +89,15 @@ def _figure(facecolor):
 	fig = Figure(figsize=(6.4, 4.0), dpi=200, facecolor=facecolor)
 	return fig, FigureCanvasAgg(fig)
 
+def _time_axes(fig, bg, width):
+	"""The axes of a chart over time, `width` of the figure wide: the line chart's, and the compare chart's, whose end
+	labels take the rest."""
+	from matplotlib.ticker import MaxNLocator
+	ax = fig.add_axes([0.075, 0.105, width, 0.755], facecolor=bg)
+	ax.margins(x=0.02)
+	ax.yaxis.set_major_locator(MaxNLocator(nbins=5))
+	return ax
+
 def _style_time_axes(fig, ax, start, end):
 	"""What the line charts share: a horizontal grid, no frame, muted ticks, and the date axis from start to end."""
 	_, ink, ink2, grid = PALETTE
@@ -95,16 +105,33 @@ def _style_time_axes(fig, ax, start, end):
 	ax.set_axisbelow(True)
 	for side in ax.spines.values():
 		side.set_visible(False)
-	ax.tick_params(colors=ink2, labelsize=8, length=0, pad=4)
-	style_date_axis(ax, start, end, ink)
+	ax.tick_params(colors=ink2, labelsize=10, length=0, pad=4)
+	style_date_axis(ax, start, end, ink2)
 	fit_left_margin(fig, ax)
 
+def _shrink(texts, room):
+	"""Shrink texts by one factor, so that the widest is at most `room` pixels wide (texts that fit stay as they are):
+	the factor."""
+	renderer = texts[0].figure.canvas.get_renderer()
+	widest = max(text.get_window_extent(renderer).width for text in texts)
+	factor = min(1.0, room / widest) if widest else 1.0
+	for text in texts:
+		text.set_fontsize(text.get_fontsize() * factor)
+	return factor
+
 def _heading(fig, title, subtitle='', subtitle_color=None):
-	"""A chart's title at the top left, and its subtitle (if any) beneath."""
+	"""A chart's title at the top left, and its subtitle (if any) beneath, each shrunk to fit the width."""
 	_, ink, ink2, _ = PALETTE
-	fig.text(0.03, 0.957, title, color=ink, fontsize=13, fontweight='bold', ha='left', va='center')
+	room = 0.94 * fig.bbox.width
+	_shrink([fig.text(0.03, 0.957, title, color=ink, fontsize=13, fontweight='bold', ha='left', va='center')], room)
 	if subtitle:
-		fig.text(0.03, 0.910, subtitle, color=subtitle_color or ink2, fontsize=8.5, ha='left', va='center')
+		_shrink([fig.text(0.03, 0.910, subtitle, color=subtitle_color or ink2, fontsize=8.5, ha='left', va='center')],
+			room)
+
+def _series(rows):
+	"""(date, close) rows as a pandas Series of the closes, indexed by date."""
+	import pandas as pd
+	return pd.Series([float(close) for _, close in rows], index=pd.to_datetime([day for day, _ in rows]))
 
 def _png(fig, facecolor):
 	"""The figure as PNG bytes, rewound, or an upload sends an empty file. No bbox_inches='tight': it would change the 1280x800 size."""
@@ -114,68 +141,64 @@ def _png(fig, facecolor):
 	return buf
 
 @_drawn
-def graph(df, title, currency):
-	"""Render a price chart as PNG bytes, sized for Telegram (1280px wide, no server-side resampling)."""
-	from matplotlib.ticker import MaxNLocator, FuncFormatter
+def draw_line(chart):
+	"""A price line (rows of (date, close)), coloured by the sign of the change it shows (green up, red down, grey
+	neither: told apart by lightness as well as hue), its high and low marked: a 1280x800 PNG."""
+	from matplotlib.ticker import FuncFormatter
 	import matplotlib.dates as mdates
 	import numpy as np
-	import pandas as pd
-	# palette: dark surface, muted ink, green/red for up/down (red/green separated in lightness as well as hue)
 	bg, ink, ink2, grid = PALETTE
-	up, down, flat = '#34d399', '#fb7185', '#9aa4b2'
-	x = pd.to_datetime(df['Date'].astype(str))
-	y = df['Close'].astype(float)
-	first, last = y.iloc[0], y.iloc[-1]
-	color = up if last > first else down if last < first else flat
-	pct = (last - first) / first * 100 if first else 0
-	arrow = '▲' if last > first else '▼' if last < first else '■'
+	series = _series(chart.rows)
+	x, y = series.index, series.values
+	change = chart.change if chart.change is not None else y[-1] - y[0]
+	color = UP if change > 0 else DOWN if change < 0 else FLAT
 
 	fig, _ = _figure(bg)
-	ax = fig.add_axes([0.075, 0.105, 0.885, 0.755], facecolor=bg)
+	ax = _time_axes(fig, bg, 0.885)
 	ax.fill_between(x, y, y.min() - (y.max() - y.min()) * 0.15, color=color, alpha=0.14, linewidth=0)
 	ax.plot(x, y, color=color, linewidth=1.6, solid_capstyle='round')
-	ax.plot([x.iloc[-1]], [last], marker='o', markersize=5, color=color, markeredgecolor=bg, markeredgewidth=1.5, clip_on=False)
+	ax.plot([x[-1]], [y[-1]], marker='o', markersize=5, color=color, markeredgecolor=bg, markeredgewidth=1.5,
+		clip_on=False)
 
 	# headroom above/below so annotations never leave the plot
 	span = (y.max() - y.min()) or max(abs(y.max()) * 0.02, 1e-9)
 	ax.set_ylim(y.min() - span * 0.22, y.max() + span * 0.17)
-	ax.margins(x=0.02)
-	ax.yaxis.set_major_locator(MaxNLocator(nbins=5))
 	ax.yaxis.set_major_formatter(FuncFormatter(lambda v, _: f"{v:,.2f}".rstrip('0').rstrip('.'))) # 66,000 / 1.5
-	_style_time_axes(fig, ax, x.iloc[0], x.iloc[-1])
+	_style_time_axes(fig, ax, x[0], x[-1])
 
 	def annotate(i, name, above):
 		"""Label the point at index i, the high or the low, with its value and date, above or below it."""
-		xd, yv = x.iloc[i], y.iloc[i]
-		frac = (mdates.date2num(xd) - mdates.date2num(x.iloc[0])) / max(mdates.date2num(x.iloc[-1]) - mdates.date2num(x.iloc[0]), 1)
+		xd, yv = x[i], y[i]
+		frac = (mdates.date2num(xd) - mdates.date2num(x[0])) / max(mdates.date2num(x[-1]) - mdates.date2num(x[0]), 1)
 		# anchor text toward the inside of the plot so it cannot run off either edge
 		ha = 'left' if frac < 0.25 else 'right' if frac > 0.75 else 'center'
 		dx = {'left': 4, 'right': -4, 'center': 0}[ha]
-		ax.annotate(f"{name} {yv:,.2f}\n{xd:%d %b %Y}", xy=(xd, yv), xytext=(dx, 9 if above else -9),
+		ax.annotate(f"{name} {yv:,.2f}\n{xd:%-d %b %Y}", xy=(xd, yv), xytext=(dx, 9 if above else -9),
 			textcoords='offset points', ha=ha, va='bottom' if above else 'top',
 			fontsize=7, color=ink, linespacing=1.3, annotation_clip=False)
 		ax.plot([xd], [yv], marker='o', markersize=4, color=ink, markeredgecolor=bg, markeredgewidth=1, zorder=5)
-	imax, imin = int(np.argmax(y.values)), int(np.argmin(y.values))
+	imax, imin = int(np.argmax(y)), int(np.argmin(y))
 	if imax != len(y) - 1:
 		annotate(imax, 'High', True)
 	if imin != len(y) - 1 and imin != imax:
 		annotate(imin, 'Low', False)
 
-	_heading(fig, title, f"{last:,.2f} {currency or ''}   {arrow} {abs(pct):.2f}%  over period", color)
+	_heading(fig, chart.title, chart.subtitle, color)
 
 	return _png(fig, bg)
 
 @_drawn
-def compare_graph(series, title, subtitle=''):
-	"""Rebased % change lines for 2-6 tickers. series: list of (label, pandas Series of closes), already cropped to the period.
-	Returns PNG bytes (1280x800, Telegram's photo size)."""
-	from matplotlib.ticker import MaxNLocator, FuncFormatter
+def draw_compare(chart):
+	"""Rebased % change lines of two to six price series (rows of (label, ((date, close), ...)), already cropped to the
+	period), each labelled at its end: a 1280x800 PNG."""
+	from matplotlib.ticker import FuncFormatter
 	bg, ink, ink2, grid = PALETTE
 	colors = NEITHER_UP_NOR_DOWN
 	fig, _ = _figure(bg)
-	ax = fig.add_axes([0.075, 0.105, 0.72, 0.755], facecolor=bg)
+	ax = _time_axes(fig, bg, 0.72)
 	rebased = []
-	for label, y in series:
+	for label, points in chart.rows:
+		y = _series(points)
 		rebased.append((label, (y / y.iloc[0] - 1) * 100))
 	ax.axhline(0, color=ink2, linewidth=0.8, alpha=0.6, zorder=1)
 	for (label, y), color in zip(rebased, colors):
@@ -184,8 +207,6 @@ def compare_graph(series, title, subtitle=''):
 	hi = max(y.max() for _, y in rebased)
 	span = (hi - lo) or 1.0
 	ax.set_ylim(min(lo, 0) - span * 0.06, max(hi, 0) + span * 0.06)
-	ax.margins(x=0.01)
-	ax.yaxis.set_major_locator(MaxNLocator(nbins=6))
 	ax.yaxis.set_major_formatter(FuncFormatter(lambda v, _: f"{v:+,.0f}%" if v else "0%"))
 	_style_time_axes(fig, ax, min(y.index[0] for _, y in rebased), max(y.index[-1] for _, y in rebased))
 
@@ -208,10 +229,10 @@ def compare_graph(series, title, subtitle=''):
 		ax.annotate(f"{label}  {value:+,.1f}%", xy=(y.index[-1], value), xycoords='data', xytext=(1.03, p), textcoords='axes fraction',
 			ha='left', va='center', fontsize=7.5, color=ink, annotation_clip=False,
 			arrowprops=dict(arrowstyle='-', color=color, linewidth=0.7, alpha=0.7, shrinkA=0, shrinkB=2, relpos=(0, 0.5)))
-	_heading(fig, title, subtitle)
+	_heading(fig, chart.title, chart.subtitle)
 	return _png(fig, bg)
 
-SESSION_TINTS = {'midsession': '#34d399', 'premarket': '#facc15'} # subtle green / yellow wash; at close stays plain
+SESSION_TINTS = {'midsession': UP, 'premarket': '#facc15'} # subtle green / yellow wash; at close stays plain
 
 def tinted(bg, color, amount=0.07):
 	"""bg blended a little way towards color, as a hex string."""
@@ -219,14 +240,15 @@ def tinted(bg, color, amount=0.07):
 	return '#{:02x}{:02x}{:02x}'.format(mix(1), mix(3), mix(5))
 
 @_drawn
-def hbar_chart(rows, title, subtitle='', value_fmt=None, xlim=None, session=None):
-	"""Horizontal bars that mirror the caption list, first row on top, filling the whole 1280x800 canvas.
-	Blue for >= 0, red below. Each row's label sits at the bar's base and its value at the tip.
-	rows: [(label, value)]; a (label, None) row is a blank spacer. xlim zooms the value axis (bars then start at its left edge).
-	session: 'midsession' or 'premarket' washes the background green or yellow."""
-	value_fmt = value_fmt or (lambda v: formats.signed_percent(round(v)))
+def draw_bars(chart):
+	"""Horizontal bars that mirror the caption list, first row on top, filling the whole 1280x800 canvas: blue for >= 0,
+	red below, each row's label at the bar's base and its value (as value_format prints it) at the tip. Rows of (label,
+	value); a (label, None) row is a gap. xlim zooms the value axis (bars then start at its left edge); session
+	'midsession' or 'premarket' washes the background green or yellow."""
+	rows, subtitle, xlim, session = chart.rows, chart.subtitle, chart.xlim, chart.session
+	value_fmt = VALUE_FORMATS[chart.value_format]
 	bg, ink, ink2, _ = PALETTE
-	up, down = '#3987e5', '#fb7185'
+	up, down = BAR_UP, DOWN
 	if session in SESSION_TINTS:
 		bg = tinted(bg, SESSION_TINTS[session])
 	fig, canvas = _figure(bg)
@@ -271,10 +293,10 @@ def hbar_chart(rows, title, subtitle='', value_fmt=None, xlim=None, session=None
 	label_width = [t.get_window_extent(renderer).width for t in base_labels]
 	value_width = [t.get_window_extent(renderer).width for t in value_labels]
 	if xlim: # zoomed charts put the labels in the left margin: shrink the text until the widest label fits inside the image
-		room = left * fig.get_figwidth() * dpi - 30 # also leaves the 4pt gap before the bar base
-		if max(label_width) > room:
-			for t in base_labels + value_labels:
-				t.set_fontsize(t.get_fontsize() * room / max(label_width))
+		factor = _shrink(base_labels, left * fig.get_figwidth() * dpi - 30) # leaves the 4pt gap before the bar base too
+		if factor < 1:
+			for t in value_labels:
+				t.set_fontsize(t.get_fontsize() * factor)
 			canvas.draw()
 			label_width = [t.get_window_extent(renderer).width for t in base_labels]
 			value_width = [t.get_window_extent(renderer).width for t in value_labels]
@@ -300,7 +322,7 @@ def hbar_chart(rows, title, subtitle='', value_fmt=None, xlim=None, session=None
 	for t, (_, _, v) in zip(base_labels, real):
 		t.xyann = (-4 if v >= 0 else 4, drop)
 		t.set_ha('right' if v >= 0 else 'left')
-	_heading(fig, title, subtitle)
+	_heading(fig, chart.title, subtitle)
 	return _png(fig, bg)
 
 SHORTS_VERDICTS = ((5, 'board shorts'), (10, 'cargo shorts'), (15, 'boxers'), (20, 'gym shorts'), # five points each at the bottom
@@ -312,10 +334,11 @@ def shorts_verdict(percent):
 	return next(name for limit, name in SHORTS_VERDICTS if percent < limit)
 
 @_drawn
-def shorts_chart(rows, threshold, step=5):
-	"""Short interest as a pictogram: each row is a run of cartoon shorts, one pair per `step` percent (the last pair cropped).
-	rows: [(label, percent)], largest first. Loud on purpose and drawn from patches, since matplotlib cannot render colour emoji.
-	Returns PNG bytes (1280x800)."""
+def draw_shorts(chart, step=5):
+	"""Short interest as a pictogram: each row (label, percent), largest first, is a run of cartoon shorts, one pair per
+	`step` percent (the last pair cropped); a (label, None) row is a gap; the threshold is the outrageous line. Loud on
+	purpose and drawn from patches, since matplotlib cannot render colour emoji: a 1280x800 PNG."""
+	rows, threshold = chart.rows, chart.threshold
 	import math
 	import numpy as np
 	from matplotlib.patches import Polygon, Rectangle, Circle, Ellipse, PathPatch
@@ -323,8 +346,6 @@ def shorts_chart(rows, threshold, step=5):
 	sky, ink, sea, foam, ray = '#ffe66d', '#1b1f3b', '#1ea7e1', '#b5e8ff', '#ffd23f'
 	colors = ['#ff3d81', '#00b8a0', '#ff8a00', '#7a5cff', '#5bc215']
 	W, H = 1280, 800
-	if len(rows) > 24: # pairs of shorts need height: keep the most and the least shorted
-		rows = rows[:12] + rows[-12:]
 	rows = [(label if len(label) <= 10 else label[:9] + '…', value) for label, value in rows]
 	fig, canvas = _figure(sky)
 	ax = fig.add_axes([0, 0, 1, 1], facecolor=sky)
@@ -396,12 +417,14 @@ def at(u, v):
 	label_w = max(len(str(label)) for label, _ in rows) * size * 200 / 72 * 0.62
 	x0 = 40 + label_w + 16
 	room = 40 + 400 # right-hand text: percent and verdict
-	max_icons = max(math.ceil(max(v for _, v in rows) / step), 1)
+	max_icons = max(math.ceil(max(v for _, v in rows if v is not None) / step), 1)
 	icon_w = min((W - room - x0) / max_icons, row_h * 1.1 * 0.9, 120)
 	icon_h = icon_w / 1.1
 	lw = max(0.6, min(1.6, icon_w / 45))
 	pct_texts = []
 	for i, (label, value) in enumerate(rows):
+		if value is None: # the gap between the most shorted and the least
+			continue
 		cy = top + (i + 0.5) * row_h
 		ax.text(x0 - 14, cy, label, color=ink, fontsize=size, fontweight='bold', ha='right', va='center')
 		whole, frac = divmod(value / step, 1)
@@ -412,10 +435,10 @@ def at(u, v):
 		pct_texts.append((ax.text(end + 14, cy, f"{round(value)}%", color=ink, fontsize=size, fontweight='bold', ha='left', va='center'), cy, value))
 	canvas.draw() # measure the percentages, then put each verdict right after its own
 	for text, cy, value in pct_texts:
-		verdict = ax.text(text.get_window_extent(canvas.get_renderer()).x1 + 10, cy, shorts_verdict(round(value)), color=ink, fontsize=size * 0.62, fontstyle='italic', ha='left', va='center')
-		extent = verdict.get_window_extent(canvas.get_renderer())
-		if extent.x1 > W - 16: # never run off the image: shrink to fit
-			verdict.set_fontsize(verdict.get_fontsize() * (W - 16 - extent.x0) / extent.width)
+		x = text.get_window_extent(canvas.get_renderer()).x1 + 10
+		verdict = ax.text(x, cy, shorts_verdict(round(value)), color=ink, fontsize=size * 0.62, fontstyle='italic',
+			ha='left', va='center')
+		_shrink([verdict], W - 16 - x) # never run off the image
 	ax.text(W / 2, 784, 'past fashion not indicative of future fashion', color='white', fontsize=8, fontstyle='italic', ha='center', va='center')
 	line_x = x0 + threshold / step * icon_w
 	ax.plot([line_x, line_x], [top - 12, bottom], color=ink, linewidth=1.8, linestyle=(0, (5, 4)), zorder=0)
@@ -424,10 +447,11 @@ def at(u, v):
 	return _png(fig, sky)
 
 @_drawn
-def pie_chart(rows, title, subtitle=''):
-	"""Pie of (label, value) rows, largest first, with a legend of name and share. At most 8 named slices
-	(the dark categorical slots, in fixed order); a slice labelled 'Other' is gray. Returns PNG bytes (1280x800)."""
+def draw_pie(chart):
+	"""Pie of (label, value) rows, largest first, with a legend of name and share. At most 8 named slices (the dark
+	categorical slots, in fixed order); a slice labelled 'Other' is gray: a 1280x800 PNG."""
 	from matplotlib.patches import Rectangle
+	rows = chart.rows
 	bg, ink, ink2, _ = PALETTE
 	slots = CATEGORIES
 	other = '#6b7280'
@@ -453,7 +477,7 @@ def pie_chart(rows, title, subtitle=''):
 		fig.text(0.60, y, label if len(label) <= 28 else label[:27] + '…', color=ink, fontsize=8.5, ha='left', va='center')
 		fig.text(0.97, y, f"{value / total * 100:.1f}%", color=ink2, fontsize=8.5, ha='right', va='center')
 		y -= step
-	_heading(fig, title, subtitle)
+	_heading(fig, chart.title, chart.subtitle)
 	return _png(fig, bg)
 
 def _code_version():
@@ -466,32 +490,28 @@ def _code_version():
 
 CODE_VERSION = _code_version()
 
+RENDERERS = {'bars': draw_bars, 'pie': draw_pie, 'shorts': draw_shorts, 'line': draw_line, 'compare': draw_compare}
+LIMITS = {'bars': 40, 'shorts': 24} # the rows a chart has room for
+
+def trimmed(chart):
+	"""The chart with no more rows than it has room for (LIMITS): over that, the first half and the last, a gap
+	between."""
+	limit = LIMITS.get(chart.kind)
+	if not limit or len(chart.rows) <= limit:
+		return chart
+	return dataclasses.replace(chart, rows=(*chart.rows[:limit // 2], ('', None), *chart.rows[-(limit // 2):]))
+
 def draw(chart):
-	"""The chart's PNG, drawn once. The image is cached by a hash of its spec (and of the drawing code), so a second
-	chat, a scheduled run or a period button that shows the same thing reads it back; the same spec always draws the
-	same image, so it is never stale. Over 40 bars keeps the first 20 and the last 20, a gap between them."""
+	"""The chart's PNG, drawn once by the renderer of its kind. The image is cached by a hash of its spec (and of the
+	drawing code), so a second chat, a scheduled run or a period button that shows the same thing reads it back; the
+	same spec always draws the same image, so it is never stale."""
+	if chart.kind not in RENDERERS:
+		raise ValueError(f"unknown chart kind {chart.kind!r}")
 	key = json.dumps([CODE_VERSION, dataclasses.asdict(chart)], sort_keys=True, default=str)
 	name = f"finbot_chart_{hashlib.sha1(key.encode()).hexdigest()[:20]}.png"
 	if config.prefs.cache and (image := cache.read_bytes(name)):
 		return image
-	if chart.kind == 'bars':
-		rows = chart.rows if len(chart.rows) <= 40 else (*chart.rows[:20], ('', None), *chart.rows[-20:])
-		image = hbar_chart(rows, chart.title, chart.subtitle, value_fmt=VALUE_FORMATS[chart.value_format],
-			xlim=chart.xlim, session=chart.session)
-	elif chart.kind == 'pie':
-		image = pie_chart(chart.rows, chart.title, chart.subtitle)
-	elif chart.kind == 'shorts':
-		image = shorts_chart(chart.rows, chart.threshold)
-	elif chart.kind == 'line':
-		import pandas as pd
-		image = graph(pd.DataFrame(list(chart.rows), columns=['Date', 'Close']), chart.title, chart.unit)
-	elif chart.kind == 'compare':
-		import pandas as pd
-		series = [(label, pd.Series([close for _, close in points], index=pd.to_datetime([day for day, _ in points])))
-			for label, points in chart.rows]
-		image = compare_graph(series, chart.title, chart.subtitle)
-	else:
-		raise ValueError(f"unknown chart kind {chart.kind!r}")
+	image = RENDERERS[chart.kind](trimmed(chart))
 	if config.prefs.cache:
 		cache.write_bytes(name, image)
 		image.seek(0)
diff --git a/lib/reports/history.py b/lib/reports/history.py
index 7fa77056..3dc89f06 100644
--- a/lib/reports/history.py
+++ b/lib/reports/history.py
@@ -33,12 +33,12 @@ def prepare_compare(service, tickers, period_days=None):
 	span_days = (datetime.datetime.now().date() - start.date()).days
 	period_label = formats.days_english(period_days) if period_days and not note else f"{span_days} days"
 	by_performance = sorted(cropped, key=lambda item: item[1].iloc[-1] / item[1].iloc[0], reverse=True)
-	title = ' vs '.join(t.split('.')[0] for t, _ in by_performance) # best first; line colours stay tied to input order
-	subtitle = f"% change over {period_label}, since {start:%d %b %Y}" + note
-	lines = tuple((t.split('.')[0], tuple((f"{day:%Y-%m-%d}", float(close)) for day, close in series.items()))
+	title = ' vs '.join(formats.short_ticker(t) for t, _ in by_performance) # best first; colours keep input order
+	subtitle = f"% change over {period_label}, since {start:%-d %b %Y}" + note
+	lines = tuple((formats.short_ticker(t), tuple((f"{day:%Y-%m-%d}", float(close)) for day, close in series.items()))
 		for t, series in cropped)
 	chart = render.Chart('compare', title, subtitle, lines)
-	caption = [service.bold(f"{period_label}, since {start:%d %b %Y}") + note]
+	caption = [service.bold(f"{period_label}, since {start:%-d %b %Y}") + note]
 	results = sorted(((series.iloc[-1] / series.iloc[0] - 1) * 100, t) for t, series in cropped)
 	for pct, t in reversed(results):
 		link = links.finance_link(t, exchanges[t], service, days=span_days, brief=False)
@@ -51,6 +51,7 @@ def history_payload(ticker, service, days=None):
 	the chart."""
 	highlight = buttons.PERIODS[buttons.period_for_days(days, 'x')][0]
 	table, line = yahoo.price_history(ticker)
+	chart = line_chart(ticker, table, line)
 	ticker_link = links.finance_link(ticker, line.exchange, service, days=1825, brief=False)
 	lines = [service.bold(f"{line.name} ({ticker_link}) performance history")]
 	for interval in ('Max', '10Y', '5Y', '3Y', '1Y', 'YTD', '6M', '3M', '1M', '7D', '1D'):
@@ -61,12 +62,13 @@ def history_payload(ticker, service, days=None):
 				label = '● ' + label
 			lines.append(f"{formats.get_emoji(percent)} {label} {formats.signed_percent(percent)}")
 	if days:
-		table, line = yahoo.price_history(ticker, days)
-	return lines, line_chart(ticker, table, line, days)
+		_, chart = price_line(ticker, days)
+	return lines, chart
 
 def line_chart(ticker, changes, line, days=None):
-	"""The line chart of a price_history (its changes and Line), titled with the company, the change over the span and
-	the last price; None for a span under 2 days or fewer than 2 closes."""
+	"""The line chart of a price_history (its changes and Line), titled with the company and the span; beneath, the
+	price and the change over the span, the figures the caption gives. None for a span under 2 days or fewer than 2
+	closes."""
 	if len(line.closes) < 2 or (days and days < 2):
 		return None
 	if days:
@@ -74,5 +76,13 @@ def line_chart(ticker, changes, line, days=None):
 	else:
 		span = "Max" if "Max" in changes else "10Y"
 		change = changes[span]
-	title = f"{line.name} ({ticker}) {span} {formats.signed_percent(change)} Last: {line.price}"
-	return render.Chart('line', title, rows=line.closes, unit=line.currency)
+	arrow = '▲' if change > 0 else '▼' if change < 0 else '■'
+	price = ' '.join(filter(None, (formats.price(line.price), line.currency)))
+	subtitle = f"{price}   {arrow} {abs(change):,g}%"
+	return render.Chart('line', f"{line.name} ({ticker}) {span}", subtitle, line.closes, change=float(change))
+
+def price_line(ticker, days=None):
+	"""(its changes, its line chart or None) of a ticker over `days`, else over its whole history: price_history and
+	line_chart."""
+	changes, line = yahoo.price_history(ticker, days)
+	return changes, line_chart(ticker, changes, line, days)
diff --git a/lib/reports/performance.py b/lib/reports/performance.py
index edc64c42..9197fdff 100644
--- a/lib/reports/performance.py
+++ b/lib/reports/performance.py
@@ -37,22 +37,18 @@ def report(service, past_days, portfolio_select=None, performance=None):
 	gather()'s, fetched when not given."""
 	if performance is None:
 		performance = gather(past_days, portfolio_select)
-	lines = [] # (percent, text), sorted into the payload below
-	chart_rows = []
+	rows = [] # (percent or None, line, chart label), one list for the caption and the chart, best first
 	for portfolio_id, (name, data) in performance.items(): # the name from the portfolio list: one with no holdings names none
 		link = service.link("https://portfolio.sharesight.com/portfolios/" + str(portfolio_id), name)
 		percent = float(data['report']['capital_gain_percent'])
-		lines.append((percent, f"{formats.get_emoji(percent)} {link} {formats.signed_percent(percent, 2)}"))
-		chart_rows.append((name, percent))
+		rows.append((percent, f"{formats.get_emoji(percent)} {link} {formats.signed_percent(percent, 2)}", name))
 	for ticker, market, text in BENCHMARKS:
 		line, percent = benchmark(ticker, market, text, past_days, service)
-		lines.append((float('-inf') if percent is None else float(percent), line)) # an error line sorts last
-		if percent is not None:
-			chart_rows.append((text, float(percent)))
-	lines.sort(key=lambda item: item[0], reverse=True)
+		rows.append((None if percent is None else float(percent), line, text))
+	rows.sort(key=lambda row: float('-inf') if row[0] is None else row[0], reverse=True) # an error line sorts last
 	period = formats.days_english(past_days)
-	payload = [service.bold(f"Performance over {period}")] + [line for _, line in lines]
-	chart_rows.sort(key=lambda row: row[1], reverse=True) # best first
+	payload = [service.bold(f"Performance over {period}")] + [line for _, line, _ in rows]
+	chart_rows = [(label, percent) for percent, _, label in rows if percent is not None]
 	span = f"% change {'' if period == 'today' else 'over '}{period}"
 	return payload, render.Chart('bars', "Performance", span, tuple(chart_rows), value_format='percent2')
 
diff --git a/lib/reports/price.py b/lib/reports/price.py
index 229cd110..af3d9f7c 100644
--- a/lib/reports/price.py
+++ b/lib/reports/price.py
@@ -44,15 +44,15 @@ def gather(specific_stock=None, days=None):
 		yahoo.prefetch_history(missing)
 	for ticker in missing:
 		try:
-			changes, line = yahoo.price_history(ticker, days)
+			changes, chart = history.price_line(ticker, days)
 		except RuntimeError as e: # no history: the one stock asked about says so, a list leaves it out
 			if specific_stock:
 				raise
 			print(ticker, e, file=sys.stderr)
 			continue
 		market_data[ticker]['percent_change_period'] = changes[days]
-		if specific_stock:
-			chart = history.line_chart(ticker, changes, line, days)
+	if not specific_stock:
+		chart = None # the line of one stock only
 	return tickers, market_data, chart
 
 def needs_beta(quote):
@@ -153,7 +153,7 @@ def report(service, found, mode='interday', threshold=None, days=None, top=None,
 		return [nothing(mode, threshold, user, specific_stock, found.in_session, days)] if interactive else [], None
 	if specific_stock:
 		return lines, found.chart
-	label = lambda rows: [(ticker.split('.')[0].removesuffix('-USD'), percent) for ticker, percent, _ in rows]
+	label = lambda rows: [(formats.short_ticker(ticker), percent) for ticker, percent, _ in rows]
 	on = formats.date_short(found.as_of, True)
 	span = formats.date_range_english(days, found.as_of) if days else on
 	if not top:
diff --git a/lib/reports/rankings.py b/lib/reports/rankings.py
index 5a89ce62..a37a6e42 100644
--- a/lib/reports/rankings.py
+++ b/lib/reports/rankings.py
@@ -56,7 +56,7 @@ def prepare_rating_payload(service, action, length=15):
 	if not payload:
 		return [f"No tracked stocks rated {action} by {config.prefs.min_analysts}+ analysts"], None
 	message = f"Top {length} analyst {action} ratings for tracked stocks ({config.prefs.min_analysts}+ analysts)"
-	shown = [(ticker.split('.')[0], rating) for rating, _, _, ticker in found]
+	shown = [(formats.short_ticker(ticker), rating) for rating, _, _, ticker in found]
 	low, high = min(r for _, r in shown), max(r for _, r in shown)
 	spread = (high - low) or 0.2
 	# zoom the axis onto the ratings so the bars fill the panel (ratings only run 1-5)
@@ -101,7 +101,7 @@ def ratio_list(service, kind, mode='top', length=15):
 		ratio = data.get(field)
 		if ratio is None or (covered is not None and ticker not in covered) or (ratio < 0) != (mode == 'negative'):
 			continue # the negative ones, or none of them
-		rows.append((ratio, ticker.split('.')[0], f"{links.stock_line(ticker, data, service)} {ratio}"))
+		rows.append((ratio, formats.short_ticker(ticker), f"{links.stock_line(ticker, data, service)} {ratio}"))
 	rows.sort(key=lambda row: row[0])
 	if mode == 'bottom':
 		rows.reverse()
@@ -126,7 +126,7 @@ def prepare_beta_payload(service):
 		beta = round(data['beta'], 1)
 		if beta > 1.5 and data.get('market_cap_usd', float('inf')) < 2e9:
 			line = f"{links.stock_line(ticker, data, service)} {beta} β"
-			found.append((beta, line, (ticker.split('.')[0], float(beta))))
+			found.append((beta, line, (formats.short_ticker(ticker), float(beta))))
 	if not found:
 		return ['No tracked stocks have a beta over 1.5 and mkt cap under US$2B'], None
 	found.sort(key=lambda row: row[0], reverse=True)
diff --git a/lib/yahoo.py b/lib/yahoo.py
index 90cfdfa3..65313b05 100644
--- a/lib/yahoo.py
+++ b/lib/yahoo.py
@@ -355,9 +355,9 @@ def price_history(ticker, days=None):
 	tz = zoneinfo.ZoneInfo(stock.get('exchangeTimezoneName') or 'UTC')
 	if days:
 		df = df[df['Date'] >= (now - datetime.timedelta(days=days)).date()].reset_index(drop=True)
-		if df.empty:
+		if df['Close'].dropna().empty:
 			raise NoData(f"no price history for {ticker} in the past {days} days")
-		past_price = df['Close'].iloc[0]
+		past_price = df['Close'].dropna().iloc[0]
 		percent_dict[days] = round((price - past_price) / past_price * 100, 2)
 	else:
 		default_price = df['Close'].iloc[0]
@@ -384,7 +384,7 @@ def price_history(ticker, days=None):
 					percent_dict['Max'] = default_percent
 					continue
 			percent_dict[period] = round((price - past_price) / past_price * 100)
-	closes = tuple((str(day), float(close)) for day, close in zip(df['Date'], df['Close']))
+	closes = tuple((str(day), float(close)) for day, close in zip(df['Date'], df['Close']) if pd.notnull(close))
 	exchange = stock.get('fullExchangeName') or stock.get('exchangeName') or ''
 	name = formats.transform_title(stock.get('longName', ''))
 	return percent_dict, Line(name, exchange, stock.get('currency') or '', price, closes)
diff --git a/tests/golden/compare_lines.png b/tests/golden/compare_lines.png
index 6f4f094f174d8b2203ddc10c5a9fa2ae89d16de1..38dda9724743c5395329c41f06e722148678c804 100644
GIT binary patch
literal 9911
zcmV;oCP>+dP)<h;3K|Lk000e1NJLTq005u>003kN0ssI2j?}E!001U_Nkl<Zc-rlK
zd32q}b>F=+^DS?|Tkx<v>;yn?-$X5xC{mj(OO|cfmTh@S8r!i`*J+X_ikq~jr>Ez%
zNt-rx(v!tqoW!wXH?|YYvFzB1V_B3)NnAvU6c>;Lu@D5Y<L&Q#>&)#R9|(gaNMa*J
zv6%A?&cj>g%{RaA&fNRE%V15q4Q4gYxl#%MhzJph2mlZ<csF=4coMum`^oILE_?B^
z56pZ^DTRnL@6LQ?c0Vg^DUFAtiJ5;8u~aGrlgk)$U6*rCM4smXfRvIkrj+tLpD_ji
zvp>KXlTs?BIOhPMlnP$-JdcQM+a@B{^IX?uj0Io!JWtnkrBv`@(7%*2=r90&1)*gF
zm@!tVRGxhDegEdSf3rTF-oO9A>dwx5KCfw7U0od!#bPnbvZB#weSLkUQqeT6p&?x?
zmULaOt*!AqKO7EgnpUY)JkQ&8|IRP|{_pL1aqs5MoBr)@f1z#Fs-g4e|NHO%w>!3P
z+rR&SQc6nM)z$Uezx9PfhYnRIt6kTvs;aVWI}wiqKx1Qrl+tzG6^xhI#Ek5Eo|j6c
ze)E~%_{?Yj`Ps8Qzxm8F>(_Pl^!DfTg(u$quBoYPI2``=xBq7Qo!jebQezX7-~av(
ze(hI2f4Zml!iC|--u37YfB2)_59~5@<3Ih*f9~t=e`)W&Pk;JT^>uao_U<(d<HH|%
zDv?MWe*N&q4I3VM=)u4J);Is-@BD6CYjbr?O=Dxz=Rfz^a>dFQ3O#3fR(Gv73?rF{
z|Ir`+>6tS<rfCLS#!6dUV>25bW6ZLwsi~>myLN8exbe=r?(FI5UAtyYXGi-Bd-j&f
z<%0(gZQZt2)3m<+fwtDx4}S0iW+=3I^X9c{y1w_^_lJi^HmqNl&E=}Alb++GQmJ&h
zzH@cwhV|?J!@vJ`-}%mWMn*=~ty>$5$1|A>A>XoP^SQzEp65UE@I%S!>MdKhy!hfv
z&ph+%Lqiu{d3j$$L;9Y(@3w7wMS_=eiFO?4z`;ZJ+<liX<u||i%}tv&e*GKYJk!%N
zGCb1T*VotAf9CX=o%h|#Is5XT|Hbj+$BV_{%P+rj;>5|};o(CE4`njhp`oF}hY#D9
zwPE81&R8y&Lu8I)`%-=PyWeYUXgGA}(4IXnnxW9i?(Wu>mV*ZlzIOO<K3_O+;NX!X
zM;?Ft@h^Spe|fH7E|>QoIFQfhgB^XPExO@sEgBIW$0?OchGB5d%jHTaWCpq}5$U?_
zIL?PZ^wdB7<oR4aZyJW<IE*pQIb%%M^-`&X$U-5L3t?M!uupT&Y}?i}JrW7$^LZge
zRaF%L*p5>vS2Ru2G~xNaZCSf^?P_jr`uo5CVK59`H%!x5(WRFYU?LJiLj>RRs%vU3
ztD<W<GGvT%#xj{HA}SV(t!=BanJFoy5Ml;?BAU^k18Zl-;$n;uk(4s{fRr+LF<5?;
zQo(b}vRv1VMx%lL9Q3n-(6RxnlnRBz8#ZpvOij6-+qSC1u2dq?=;YKyGMOqC^5JlV
zhz=jxf6u-5mx{%zc-*!u*L4!f>RdMa=8?mirY!{$L=4QN6;zgAiw@@3_k7Ryg{H-m
z$zmZ-ii)EnO3Ja3VMPiNea{1cfq_0v)67t)Tq<TVQ=IckrL=TyD3J0MNN(T~eNmJp
zV|<!gQVJOhbZpP_gb)G6M5Kr=(q2StMHRaBCHQBS6B7UzDMrR-h{lU6Fd_n?0H9zi
zU75tUfgOC=VtKVC^tzu}X>SXFSK11|EA8#O2d}i{Q(+MR0SMkYgMJ0z8w4!_Amv)1
zd|;euAwUMCRsg;|z~#y<1M*PACLT#iEANb-QcjT>F`%1m=wIUSUunyNuC#JydSmrH
zpAz9DC|^&uMc2F&8F$7{p=Pc~@Y@{SN;cB<aJ2KG%%Q(@r}_~Yk<y~IiS3`zQyrND
zf5FVEo4F;gNboXRBwX3~aNE6AU5||a^!Joq<YpX&&h-%acIoV%P{Za($F55MKGZ^2
z5(zWGYcMMaZsx6-0H)c`SInfFsxXshYw_mThWAdq^gjR@Y9UfSkRl}k1&`Kb4t_<s
z<>cL;l$8uJ4iM<r)6MsM0PuaEh<x9lfe;a0*Y!NlaU1|}UDtJ8&-dr)*IAL^Wm}!e
zcWUnc<@}r9@e0FC3tz4?6LBM29(Xm{^@s>1tl?u!XhagH-=st`={Qb27W1X7s;aUq
zD`1GFlr7E8W++rwS63($8q(=RJg)0n5Z88HQC<Ofc@}4B<>H&3B;P5X`bQB;&aEvI
zFrill4<)ueE$uuIiK=wVNB(E1eYZA#@S{(EgaCH!+8K>TckI~l`s;5n#-4ogeVmK>
z`nrzxj#xZ4IXPLAO08SJuC1-Dv9aO&`Jo{0cmV(hj-!i4I;@05%}01idLmL2+wh*u
zfxlo{=vwwJG9YP>o<t^;^tfguw}0ICePeWArCbh2!o$PErfCij50^_N%eFO5%VaV_
z2q8FUd~$MP&6+j+ef@p?{ngdUOeSMl*0rOWrrm`XLox&+c&^KZP)Y(I5eXqyA}g*V
zzLjih=jWX9)8*cmm>Ihcp@B%!2a*5`5V%nV0KQWWg2QJL01U$jc3j4o5W@HUnfRF?
zEW-DFP1CL`fdUbI->-_rh`@F1O<V358y%@mr7D$jRaLxL%uh{DW;0Worma9ZCtS+2
z%=&dAl*snK%tLXy-kJ#j$OV8F$dz<T0Kgazhr_eeJA!DZ0OlEA?aT)_=RwNB_3dCH
zVvMifxLFA8<(FQlRLZNn)(#B~R#nA3&uwX4l}x5i96RE=uBPiNl6-|UGAXV2h9@e$
zdxWN=5c9Wj5y1C7U;!0DGZ{lL%QJiBlR5vrjs#2T_xGL!fN(g>8SCjel}go2O-@XV
zkL7c@`gEFcu9WgzcLmWa#n-4?rwEGY4skOEjtvXYT4=VJbfbnC?7_k9!#PJp*L64-
z!8+jkKFo@Me49!cBAdSJmzC~E#-)|7dEnPdXZJXxCs7Z<t+&KYQ<Z_{9IW!1rV|k&
zyB#zvWaL@WcMuT(nH~Ygmj`A<>6FceO#o;Qzsbz_&37a$0N}aScY?Qd4ksdzQg*>i
zZ#2^Dk?VeOtTK2=IYn;9NXaF4&p>(6)eo0@_W%<Bw`c;@ZAx??Bk2P`>fX<qb!*&A
zzh9crVvVs4?*Zl6<7e1hZ14h!Nz15PYu0w<k3YxEDBP-&a65XF5JCCKc<qC~Q5if?
z=>8Eh0f^=PmqoNTwc|5dyfOR6w|J;}@rx=lCEcp7hf95X0okn_fp~j58<g+X?Ecl#
z*%t~YejvgrW>hgfBEl)<T4T@uR}oDo@A&vd#(z=Nm1~)GYf+fi&>KvTyd|L;|DOVk
zjLS-{`mUdM#!r`eek!W!1D6gEk@ArX9*R%w{Ua^W99#dmU&$^8W5$$kN4wrp?%Vq|
z;Ms07c%@wox9#AOn%vR9713If-en3T0C2Nva_^snSM3ZnY?4k1kS$z%*D*a@Ie(ZL
z(OX%3Z##fVp=QQ5ygU2)H<%u#Ij5cokOSh>{x2u)_&777p!@|84vZ<^sapGJ>FkT}
zcHjnXBY+u~c0RuGz4q{NFE@n3xT<DLifN{wzfe5&52+oWQBHZjLX1dmIn=Nbg>DTU
zX8NrzzH<P4k+FK)@m5H_qbFL7ny$jBA8|7Q*R*s=Nghg;dS8@QF1GmtvXVhIAA^FC
z^sClAUOe^VTiN!#8o)SHzeq}-h=R<)w*}XMj48KLwf>1x-(K>Z`7EBKd>&3_U;C?2
z;})}FqjZYbkz-`aDMZ@uSDs~!9OHTz-VQS}O>`)wlv0LiCK8EqsT7YVlv28(6A>4J
zf;u<1?DZfj&oQc3g<7`fj(nRL5mE~0Dfh~EC(kDD{-ky80Fgq(IW7}{0Fe}Mk-YQg
zvWLG8qyTUkCgjHN;sOAql#NZz05~x|+R)gXZfFWT&#`#Is#JzAoF5$-)-)a24IB-^
zqGybN33AJ^b&py7FEeC}@p&u8#LOyLxM20|sk#4OOzin03nyWYs!fDk^Gg%eJ3i|S
zAC={C9!f4uHM`o+(+w|EApn3e*3{g}7|Tpe2q8+vqG6hdc>df#Yul=|>o-=)<#M^G
z>BeFQLkJjTa|Dwzqu(L`L+};2M5I74&>0dfY;z;u3DvLTW|cj3ScGCkav_$F5|Qfi
z**zlOlG^c$nFD_=qP3SY)r3rwm29+gw}{r{4*!)Ft|pSe_|orbDb<aaU_>S5iQ`A5
z??)n0A{sh>uBoLpm&p{1g+d`;U7d1W7XW?NT^zp182PU9D#$bf0wtLdMdS=rMS`oK
z#()ML8A8^DF^do&DnQY^7(irPT7}s5-NmzeJl|!^UASiJp%$Gw_|=--zY^X2q1@~L
zS43(7AozV`LY61>L~CTt;}b9bfkLi4XSsH(o8W}~Sxi>&*Y|y0*FDc=jC0PFl2WQ5
z8ffv<l3mbZ4Ux_VMYtMBx%uHr?;hopiY&S|svfRo5dlRXh#)FZ5$u~%hqJ0m0A7F5
zMv|_UXi98(YJAThE`c>f08(D1=Kg;v?R@sNuLDqEXep;;rZ*<<`DAAQpLx0SOuLGC
zeaot!m(}RFtSszh1!Nb2u3yPkbv+X4e6ZZV&z<ZACc+IH^@eTZ$G-n0ywR4b{>iA@
zT_8&W0O$->Mey-7vn71>oD5!3be;ZFJ3r@)o+uBz%8aTd7^Vn7$}K0heMC=nl>7F%
z+4D^3q2}#IZCB>d*S!1%ZbX;vXMBqRURKkE`;}~L!+S!FTPF7YF}Y<FA%^7GB|X{p
zTkqI)cBucW`(G1gEW~7>z!8D1z?Sr$27XtP{qb4fmSFI!VT>eQEtXDf`{=}re~8zP
zS6u+vnKNs;!mD<2Ge)G`%s}bv9#9fBbHV-NTVLwph5^jDvI^n09nr2w#-IN$ObY?m
zkfbC;<iFA3zcgka+4zO<$o^u_Pgy8ABN+(55hh%^ubRELfq(Ig?-K}|Yoda9)q8)*
znK)DK-?udJojoO-d`o#Y6FMmcq?i$%&u($)*bt-wsem)!*Ib2ejvc)0hABx`L~D|F
zequUNhik#?I})Dh!hIS3{3Jw1UWwiL&-`*`npwOEq5`NA*i!<R=wmHn@08N;D&iY&
zinQ&>AN?D)oU{WW7fcHy*HGxQ2ets~w&Fra!|Evb5=t)U;x+@Aky}pf{9Lj7AH3|j
zAW?@S<URQ1PJX&XKN|6CP2o#v4<D<(_cQkJQRQ2+J$DKqDxkYSH3sfZv3*k%5tmLg
z<H{~1wmoePA9be&76Zd<X+yk%bD&v|1vbZL_q4NZEszMa^$i#c(O?F|TSNP^8-cE}
z3(>Aeq+O^C9^_^mND73E3m<A`wt~MKR<#EDN}*=@<%#^8&sN|2S@PY>cmtoH#=vJU
zD4oNj^(^NB$64a**+!~UL~6<duQGG7q<;Wv#f6FsWfyz}OC)m;!S>jr8z5@PA0AUb
z>6Sk{1`&fjwgDXfE%83x0Dws{GZtyTKY#Q)+>DX(8A8E_Z3$do#r|?YR%_q~QlTU_
z<CVdKZe}2{{bN@${*oZB!B+<412wGC1V>Fz5i+Jct7`q@rQVl7$we|sh~P_5gm<>F
zM>nts*I+UX6?gF%Gla6k9_RwcQ~z)R5Fw-iqWbX(D7);TwNQ3|;oBTw#-&||ZFs`E
zaKtZ90@nhZA{;*4#J@7&i%Tz%l+VM}xg+1w6D{G^d!(HQyo4d*2$qEBhULfGge?I*
z36i8+HXAlEJyJQhe~~VCMDTow82rhts0%tVOydRi;9A_=v<MgwfH6eN!?pF8jHnk*
zLm~tK)4EkWr1qSKu!ftPpzPjcjF)*2Ch3c4ZFKb`+1I}@v-RdY_@xg1T8`e#Q&e0*
zLm~j8GumB!_b*r%j)3pX#+U#?9L`#}*^t#H>yE_2QC()sse3+EIQ93+${}C4{(=CK
zKo|TyTh#tOI(i-)3FAdNGsf@h0FA>?4h(VK%{Nf^C4sHLk%SuO53a@MyDxI<v#voC
z)IZ4|SVKcuaQxYu&CLe*vN{0^BCwEgY3GyMp0>wMxKn3QHv`4$p<0#>;a?A`qz;ni
zG)y#E$tbUq*z#1VcModjY}H0F)NS`nZK~_tYsYlcPl?b2yJ50Ye)A_R5(l*?@pXOv
z#3pt0JoQdMECdLFqD$SQ{DC!4upn!JxE9;Y5nO_z4+$OC#CZp}v}=tV&P|=36d_Gq
zrkimD%LmtEceGOXDCpv**~gav9QcDd=fPPxGsS)92XjC>G)6jh=JtPsp$VSvdomsp
zA8X-X?w6cheXAs5de|91qo>+KP1`E{ukab8mLZf~ZF6H`U30^$XYL(7fAA!vSHCkG
zJ;ZVZ`3h)!VUrc?lsnqNS9E9qVj&0^e*$0(6wz>=KeU#5#;%QS-h)~LKi0zUOYs;<
zvoS6^>z+K@yS~ooFR7ddq1odi07JlV9#_|a&S4~fCB}GZyYekqc!7>-fM@a*4Z~<|
zY0KrZZL2zvVIq+X&NvF90Ip3~B4f%a#J7If?tjCrDf4@}#Oj*7rvIW*el@F3mneKq
zbDu~tJyaQdExP6%Je0I9yupkr0tL3uTB~7Oi}u_bXS4F*T8r-a%)qOA)ah>a*cR%a
zB-aNvZx#XpJc-p2+}14r=_HuiB?|x%Xbf@{&;?tYR;Nb6xI9}B0T3bU!8_~N2b<Vy
z8TGvpb!f!x-rhBS==|&F$`ONqq={J)dP)?U&A|}?V=$6u_jOX=Bq_Q~Ze~hwz{LZp
zW<d2kFGve^T{lQCntzesX@J+PTVGe79v>TNYwtvaR837HS)FcZtZ!&kQWf$!#+VQq
z<NT@*P<p&E(zK&kcS3JWvy&tCNNN418gG5WD~G3I8V4?}^>xk}V_Y-sp+m_#o@PS#
z#!qTSlo>jIbOU>#7d`2d1M5HZpm%gnZ+FxaQ1RG(o$SmwGc?BdbvoyqamIwkT*V(=
zhuz~WXV39PWXv?#aGq^xL09p-Gy5|^D=PL2ZM-w4eW@S%Y!+4&Sz8~jjcRWU@o4Bw
z5snqv)2$+*;qel$(gg%xXPoI8t9UGIu&#R6Gsz8&W(uGqc6D_TQA<lR0JOEW7K=rt
z6aaaipH8Pkp^$Cct*xzwVd%OZEV1(u<p5x)Pd5lI#>Yn6R&|t$#b{NPZCeBVeUV6{
zp|NRda>BOlph`H*`9LO}LcH_6?pAR6PtXg!e#JE$_tW;oa~Z^YJE?nQn(e#RLj|8H
z_rDU~@|2crwtL@z-K%8JuskyeA){gEFAbEbYy5dli}bmujpd-O3hQIkI{`WeMc3&c
z0BjFCY9JM%{e2M9=K2au=ip5h*}Wa=^yu{RKm<exl=IYQ+BgIHlU`{s2ys%rg59g-
z{yuUgA%qw>lJ<<t_cgF;9gk#Xn1c#>Py{-sp)71_07c+QN`$q~J>YbvHp_S{R+UcI
z6H!M;yX&}<lar3)?Ao=ft7~;SUBB!8`}ZF>@YrMTYHV!i>gtkGj*pMe=SZ4q2@bM>
z1OMRU#JE+dOifPMmNhXxHZ?g>E|n0Glq6DfJb;vRImV+OemZyRd-P&IkLg(tKH1KX
zI{f(2$#BA?we_-l7<5kZDxHubB2tQg5y?{DzEJH}bH{^}j))V|TzzM3;|HwKllddx
z8dZ_^rs35IpX*$mn}mDX)OdkPHX)EwSMOXA5J`!<yVR?F<ah*E`Mgpj7)_QT6`{JQ
z>KzAzD=9guDO>$q3)4{j?>@g;BSMlz2e&qnQuO8!nS!L0N-<CzXzzr4su6ufJw<6U
zC6xl2KD~s7as-Ne1sdn~uLdB0e{W5yCY#HKLZN&<9}0)Z$H%3Vg7ZS5FgZCTrL-(N
z91a~nexg(=g+t-d(b4%V)R#LOJl7SPrlbsVf(c;eIKpfB6-v4IuBZL2X8B8B*CJI#
zPi;@&?ppo_XZ(bLvf!}yZJ~ilIynMieICp^0)i^$^^ZNIpNnd;T6%eF@HMY6qK9Hr
zj{5W}K4R04$0;s7YOQ8ln&opxuhTO_D7(0=1(RX*(^C*Nu0IS#{QcYH^W9LixyG|D
zy{n#Wj^P)3WI_W)07M`-e`1sT;c@WgY@!MQ0w2l0UMs%bFUM>$_{H5kFw%m55W&<C
zz^~X`h~Quv-<O&;owXa3&C)c@bzQ?S0<~8uMFg5A7Gov}Apn4L9?-MF={VOlez}&^
zp18aGch52liK77c<SOyC0pBNJ03aBh&+=WX>0IWriMT2l6_?-HNy@i>`l2;;%pE^X
zN-{mHNNEfPEA+kw{zguL#$l|8Z7B%r*D+oRK<E744z;&uUQ|QCgjgQCs|~wHDvWnU
z@rgA5!?Q9R%m-&sa@d1wX)pulvseDKU{H~8!J<#J@_ke45{3s6I8Jlmo&;a<nP~@@
zU3??h^#fdjO*c5GYX0Yiu`9IM%hQ6d<LzgR%8qu=X1nhV7%Yg8$hqpHE&NbMU9iDm
zKm@=UlpS?qnC<KY%e(&kDuR%Pt83NaK^9L6Gr`O_G7%U~I)gDA&Xm+U>sY}h9@f;W
zy=+4RL=3oEh!aC7yX?+3H82GQ3-~-#P$DojIy(*}hwf?%Th1p|X<r$TM8E)uA(S2L
zs)vX{Zw^6Zj>SU}R0%jypc4i4(Pn-rmp;pj!4VkFp7zTo|H`Tzw2)ZlB~z8sXYnMp
zN4&kqL^PCl)Q)PF)bU57DydzhH55Tur{fn;=eW8KEbm%E%n&Lr+ullp8OU3}fs_PN
zGt7Y`sM6q{#?+=5OX=VvgQBClN7((Hb2G^Z;7Ulxus%jd&d+a;_}Q@8zMfoD%?EbY
z?Hi~J*ccL^7=kMyYOo#c>cuml@v9kWK0&n(e|JGOg>Xw8%l=&V_6=q{mIg3Gs(AeF
z)&A*TEVw?yDgp0n;9no~&C9|th(H%|e;?c43W8nhvU~-p2sR|t)xi)5ITw!P)noOn
z;1ZAO^yURH1+GiafiWU@lJD$PhtGjh3-U~Gh6NAqd;QeV_RgQ`@q}ebS3=1Fod+~0
z6s&7AErDtMH$(FMjU4eUi0@b$7ws#SjIpiF^7-T36gd|@-@*51=y+i+&RH<XRxlpI
z=4v`S0p?u$Wgxx*ku+9-tLPB~U~uRw(-UcawyY`=7-w`L#~)ln0~v6A5EqlOIfH`D
z?&|;|dVLVW`hv<Fc?uFC_VYFB`>##02fA1?0tDDz!|z|MUK^mk$@#1YI)e)qw1lxe
zf=6-`7B>}UwiNLN+v7XB<gs(Yms38xzmZuIembra+FYEoB8cef&;T~XpeZqjf0Q9u
z9<HtfjZ^pNwZUKrmW2Ho_3kt;dB8OWMOz){=TB?`0&sj_2n-=_;kq=oB-P%Y1uV`p
zH=7TiXktIi;^agHe{fVe9<Hg!s3HIH#5~cRae6*CEH}qlLkL`TQxm@yE5V51xNKD#
z+p6)U?t-RoiQ&Vw{7-vj<QjJ<B1i?9GT*(Lx<@Z}st^E2{>VD@;u(<F@x2p($)UGQ
zkJj-E7Uf(LCa19?0PqLaz*qr_Hb}y)&1`d%{LyhBSdgDCaDqKj#}Ya|J0z1Kt_a|K
zmii}YFardjFQDcbLn!$`0K00~t5Y{sh!+hofCTaQc6#+>x#-r1*)Oi*U+VQ80&(qW
z)Hr0y7}9Z56CEE0Qv<jdF;jNfyEjwsI1EgIsa@Y&hu|pi6g^PGUYVjW1JVTy=b&V>
z9qrgw!`7w25b7u0;7Je*^GO6^_mLL<XZ=#ltTdVcoi8H5D!3ibmgu2cUhrwm0S#|u
zfH|W=MSoyBy5-WL{#e}nR~`Io=i~(&!u)z!Cj?*!IyZ%FDeSDJQ=^~=d<Cx0-n9{I
zkM{S^i-ov2Ykg(dU2}2P06-UzsZh@t<to%aNhd~tGvItdA$ftb=riqne}>MKC?qcR
z3oTaq6hOwM$J6|kNo8KV-*DQ2fdMUbS<SiN`Am`0MS7Fw5{|3)#Tk{#dgp4fy=CIr
z{S688S3CH(hh%q=V%ps9-nlpjK|N#GoMhWu0RU^Fe0LWVY_+!s!p4HYPyktv-ks(z
z2eU>1f&*tzaluz$E?C+aLeYc!YgkhVzja<E^ks{AH5d$5;En`x27RSzf1;~E0f17r
zZ7Zcb&+~nM_F2opaoLP9-w)oAL~8y-#~c6$2`ofPN_Dws3K2>bQP=Fh_wK1@U)U9<
zkGE@IKPS736rU%$KU=E>ot=QPgN+Fc>vVF2jt@gfUug7eFl$>AXmB`N0&_-I1D=Q$
zBWEUqpKIe^8IZ8tneYr@*rxY4ii4TyK=9TGGr<ofYVN&vM<S7EZ*OmDX{o76jf{*U
zVj_`HO5J_;U4=rSt*y0v)v9<R9twqu#o|2Wc#eYLobwHvw&=Rq)YNR4CK1`THBDan
z5;|8$p1h}Z--}PPN^?^GKfT_l4OI(8cN!c4d5ij|XfOko3#PWX*k-}toCP0j;)k+o
ziNuXS$a(PTcK&9Ljuj{(mN|)EXE5$Sy@|C3`UI=O_^tblU2*>aZ+*u*-%*pQZfR|4
zZEZ;;;z!;%VwmPrPkm6+#Hv-R0_tk>=FQUgw{F`yIXRin=jS8GmpB`ON~Fy#ZOLS{
z&@>@LRV+R=IdSUbaV|t{vZj(WI_|2y^Vl&giBnqg@T3<Os3FKj-|2|c-y~RUb}(>+
zywBc~v=HF?*-%m_pB6kG0^k58iCZGxT~YVX#v(CJa=9g#F8Ufm901VDP5xGg%sA>#
zMj{;@9hGufXj<U#@9F75#Omr~u~_Qt?8xT}L==gH2M5nLH8mBB#fgcD`BW0U+?%A7
z3WdTOHf^4ooE#e+X>4k)OE(-l@_MOM;7kAm({&x-1;#m1L|k;CzSRu@Jc2K*5nnwg
z#~soazBVI(0^kVGbcrwZ`bGaL`7+BiT6q!hAQo5$HB<3DkThM_gFQG1<Ol*hq?DSb
z1z}b5tP?xu3E)c8bSeFy$KaUn=@p+qF=RjtZWU-}8=BkV_)s1H;u$}&@GXNQWL)^Q
zE`B0UFHES!;x?T{jePpl?Mw4IXOIrmpxMrb1*Lqlx%@-S_dU3HK*uHLdLS?&p!sHJ
z-?SFEZVqLs;L&6CEbA^TGL<7_U3fgr6v6Z3HxJ!l<OC3uyb1mWPiEkRX$LUP_AD*c
zn{=fqcI!7u5YynV&&!=D))EF=&X<N|2xT8Sqj-OceRV*_Zz8(4HWz*CHI86Q_}dHe
zQ>%nO&k7I#z5;{8r&jT=os%Bj{N$1qfETjdMFkwqsbhKd*>;|}-a;MZXIm0}wNw0f
zRP~lA%-_OP-W7o7w@_E>_}vRi5&V1`n{bHZ+@=mgPyj`re!YtyFQ}I$XGJL7POIAO
zSMi+yLIU<r(ZhADE6Vm~l)-Rzb}j-4gt89=@JyH1SEla{%W56v?RaR=?OD<v0W=0L
zPtuMQ+g-zs7bxpez~U<aM?ul2HBt8YPH`}!elVhLS9IZq7GZhIF$RFVNAIlXJ5%g<
zUL7k?-Xo5XHt_ChmeTRr3-U~f61U^ZeES1DqgUrWNb0aF#ahD{=v);ECkk{ZqYxp?
zZ)bFGe}Lg)sl>7bj$krS0G0%f5aH7s_U+vk>Mp$@m7lJgy0}E!i6O+b>E|54(12iJ
z=Fklo;Z3TcwXov~r1VKGb0NRL{k&{&Dy8Ncp`vL-2|&_<jopB|T}E_o#7r&;;Gjx)
zs-`v^jjDN^pgMm%48y3aPtWJH@_CF<#uJHnBC+(|X5W9D^K`m#xvu&K&E%2*o*Be4
zj^$^4&bq|KT`gYXioxX?<|Tt!CTL;W9klq1U4A8th@|faj%d&Ggb*`9Jd56zX7!1l
z=MaHuno239WN`55f_js_4~RktDgD4DBZRi}LHeG@g<y;+C0*AuO>_467tvubdx$7~
z9}tvOx?wD_&!rL_LGEH*x}jVyBVsHTw{1&GpGYaG78>C<)29%kKHX5Ml!;Wjp~*DO
zLO#zqkH?dxQgQwfWx*m}Q(LR+MyXUvR#yuys;XiDWLefis~ub>Zs~?b*KvH`*A1gO
zSzRm^fGCK0TST9TEZxxP*p8HbZC#!3$y9A^saTZ$^zj&ro6MzZ(TM1KUQ26RHZw&4
zZS5V!LP66srIe=YJ*T@Dv5O$0=lRVoZTVczwkjJpZ|m<pQ(Ko-O4iiYj}8x=A3QhD
z+d0_J!r^E%8XY=6xb=>^lGW9^Zk#>S-4Tfvi^WhVbhhX8BGyvRbJOWY&+{&vA8hYf
z-PYdm=8?k<jm^1iCKgZho;~Ba?))zt5j@XrXl$}=YhtXl^^UuXg?u;?Io*9Sm8zMT
z7-yUn^0{)U$d+74eqnoXpt&%{%}~g5U0pXCo15D^yF%fx5TdForWURWBBD|f8JcDY
z8O!IgR>kUEy(XDV`JS7qsnd0RK}ENH-#5cyp^0*-==**?m#I$GHa53(uI?%p3qb{$
zMQDEn6f#4arlo3Y3;Epoja!5eZS9@D?^%_~0)-frk^pF$A;x(&GlhtSe7<c}dwoMg
zs-|Y`x((&x@}qlw306vJx-mXB>UmyOEatjSU+)>tMWtLu#2}1l5p$>O`o!3nR6Y@%
zIo+MlW&H!b<JjSF#PdAMvgX-spyGP2GdVsU3WX-d#|F>!3n9Xh=!Ky{$FZ$SWf5mw
z6OpbPnaoths_2Gsy8A>fo87$S4k`W9r%oc^Li-FloSK?&9EWp$w&zqnmy5;XX2^71
zuR4_)JlF5}o_Pa~<DM6j6$E{SLLn)oZQC<p_)FO~T-OyBRT$2N=Xs1VC1s#4ElkOh
zesCiOl_!WuDH$AMzLdl1XPs*>z3B#_GmGxibKM|2Ke%KmWpFZ8BwA(L)(ujs=LK-^
z+rer)a~S+GuvY*;G&|UAar)$yL1;6*EoGcP>vmQ(@TE{%vkyuLy2a@mR4!5zT)Gy0
prMU?!?FKNlw+L0a(%ydU{{ed><jq5$nBV{a002ovPDHLkV1iEDEZ_hD

literal 10089
zcmV-vCzjZWP)<h;3K|Lk000e1NJLTq005u>003kN0ssI2j?}E!001X2Nkl<Zc-rlK
zcW_<TdFS_?b8dMBuL2-IG!lgXyDBBIS>2Kx7uj;$631T4j;-vZIL>Bwc6N4V+w8_k
zW_CS^TkM^2Vy}yAuUr&6wl|b0QWPmtq{IpkAPAxoc>TTmZae4OKQ0J?9RNX!VsPeR
z=Hb43&pr42?$^KX9IS0<gISGpu9N}*B0@wW0suq|ehm%=UxMSazs&yWf`b>FFmshs
z3K3_1ojGTAKTGXe8V^SkGyfoBxm*q=moetLF6W$xJkJ9FDJ5e}Ddl-SV+;UhpTHQC
zQYxi5=K!FT3J!XnM?|)56OrqAuIn<!g3CS6({)`b6&wuumr@2D2EZ>Nv}gb`#w^Qv
z&wJkeZ~y#Hmo+r(+4D+gM@OMh&@`>SzMhC;u~@ZQjYgx(mMya^OVhN*#)eX<tm}GR
zU9IQ&;c!^fG|RF)&)d3X^Iv`OzwdZq=lb>Q{@tH{p>4&A;WKCc@{50Q`yIFK+4G80
zN=n())%EB9<_r7w?Mo$7uItv+)Y!J2h{pk7`SM07rR%y&7++%(Gqme@UOJur)6f6O
zr=I@I$&-D5`uWeV?&<35KUF9cpLqPyOePl&hrj>*AKY>09rd;8iOH#-{`6;m_^&^A
zqOX5sWb~bnKJxtY|F~`YRzo-b^MCno1E)^Cxbvk?e)1E`>g!+HxzjL=r#|rhL?W^O
z!2Y#s*4%&pJD&aCcmLCW`LAuQ&8gbj<;$Bs_xn#*tZJcH?Cb68?CLZOBbkW*uP^;g
zZ*QMznt{kzY71*@M&L2Vs?};Hli9X)^V+p*@4V~IzP|qMRjWGMSN`*ko#jep@7{eI
zZ`-J8+Q6y7w$|1spM0Mg3awwizI#>Ib3gg%=;+v*)jhd<K9x#(j+0KO8yc2%bat#+
zz4|}?hkyUWAO3J`Y^<lJI~I>;vspsEVZ-{<LuWkCfAE3(ld05(jT>Hg;l<B?{*Q-;
zM|SUesj;zP)7^L5w!K8bi^-xL$9ZM%zD;-E<xBb9?|ygPy0!oC?eF&X^^J{=^$!dT
z44mpc(YyJ+dpT!c|Hd~DA3j_vm3HmgedNf|(b3U;d-r9tx#8jA{rmUZ)#{qHYdB;1
zd>)ZGj_phJ;~zh_ys>fLzI{7(ykLexM~@wAZE4y2%HCJ^?=KXJue`GN;K75BJ@(k&
z{_R&h*RND6dtP~^P$&cveW@+D;cP7$5go@Vm&=A>aLy|gD-<#VU6+V-U3VPksSmvW
zSHJ#EzECg?!*LwO80VZZrt5mSTt;M}kjaIxt9Br#IcK(QYnmR3gbRg&5Td501^{fw
zu_~6PX__WH-?yvPty{M?H#hzK=g$Yj&~?K!jU`=rF##qbAv8qrJug*TSFKvQrXxef
zIA<)I%@9$kRBCNok;`VJltPFZ_=#vne-5mj8H<ZCMnqD|-~=gUa4=xMN~z#`wOVyu
zHyVuw`g73F5<-gxuu>`%4zF3eKAXw7p4+yf-L|YqG&+@;OeWK%Vj&!k5Yhg9dp6y>
zrCci2#N&3g>bg!MnabyKuOHm6Y1*|wf{1~cw1mpyYtg~{`kwFkzR<LIGFd7XNKt8g
zOi4L0HmXP=qVIVCFgQ4%X_^@dRm!DoHp4l$tn#&MLjjd9L2?~g^f^+NjPYsLl2XW6
zpksTUCxi%GOhk(4oZE|tEvZ5`KLx*OIWYln&c(>sjHB@!2S!9d6aW;6(xpj!8`!}Y
zFw4ua(5s%a)ZP{VFSR9rm)hG`1~0Wm)neJSM0<;D`Xzv`C$vf`w+y6^3lt$h1X4=?
z-y-NLyU0U{aN|0rNBqi^bM^!&j~g|lZm6w)Uci5;Eef^(sGL&GnkQmw-;F{;WM*A=
zwCh3gZD;Bv(?U1W23exu#c#^1q_=zyl$?6$OR`!(L?X}=&51idqStoj4}6E2@f*gg
zFX72rYS+S<rCq4q@{F`H*?oU6A}M4d@Cho<R$euI-_rmr^YYhtC~*PC7bK?5_Bvdg
z69;q7Sz9ub77<)7tGUF+r%1ZFSN}n)Sq6ZlB#<H{QRqCB$n5$GkV@VCacO0daey}+
z_(9^E@B2jL`~D1&i0Hbm=Xs9f1gE*K>v{eh0bY{9i>Fn&R;Y1Zbmf+*onPXiB+$jm
z6fr$g8Qc}^dPqdl)saI?=z%;|q*BS0<2dnn-1EFxtj4mcL5ic4vZ<*l91bn3uP>EK
z_4W11WHLxrlu}+b#<K+QVl@JU^gW--zWVp%Rgqmb`oe@^4ed>Ad`h}y<VMSfzaCz>
z9Y%J4<f9)VqV3zanx?sR>z3DEJE)ZUt#>^U4u=!*M0d}sNHm(y=VLWB>(;GpZE0?8
zZ9OwQd{vAu1Q^e6OSsLMIp-4_Kj@dnD}%e38M~4Wg$ShVu_J&CIZxj4QQ!BC@tu}s
zMWfN-;gN7SJTg35sZ=b>(sezX&2qsx7n-J>J$rUl*Q((&!+j_FlF4K~pRZJ`D`z#m
zsbmEL-1mJvr^N?Z8B6$G6<?CBh}I|X{`llSe-XKHbt?#wlmzJm<J^b>fbUdX*JX?o
zsUVxhG)(|-U5_yoLioNfxIl(M0*sW>_kB&%u1bV&rp;=a&NyppY_hACVVIuhIgXvn
zW{3!uX!-Jt85mcN9bfmJO5bll`KX2FVge!KzyuLVrwjm$@o+diTZ;~|oeV<{XfKG>
z&YZv)GmJSUtcieX>z6HCvu@-5y?YuO8a?0hT-W#gL^9QP;@HIam}!QVqJx(uU#}W!
zxr6H=XXJn$Nf9Z4Ik7zi@O=+hKw~shF*Mr*y7<eRuh!hOXc4>)`CRVMfmaD=WcZBZ
z*o?8M$+M|+t><|N;Cb#6l9wW10>+hN#oO;M9DUxGO1aK_yi_#XOuCMqtbmAfxh!L>
zR4h*aAjH_{2t$0Er{iFdfC2;rYLP=aj4QViZn+DjYmXh_;lw=s%r`>ry(xn^7XYMb
z+6)KKG%a8lZ(r?}q^CS<`af=<W-hXikSNl5fAR27P#8ovV~LG8pJSZI+27u2$;hN#
z&|;0@t_OH1sXWUbKkA%4&h!W{u?R3DQ%*VDdY5v`_Sxe+oCI|<ri9+2u1Z8@7vpQ+
zlfLIuz=Sh(QaR<=>L=1$o&hGvs{-;xxl5#dd+EflnP$SxS|N8Uda{ryyOg^3X&#DA
z?fSAkcGxdXxKk&s(=T(QCVA(_oU<pSl|{{5xJaHR-%TQ%EFJ$PH*0{F0KRcfrCfT`
zCyCV5?k{sA%JndEfx={3$euXj6-H|B`?PcRxT@xn>kHH{5J`E7J3d@I@-y<R#o-4_
zEP8I}=4-kiWuZj&m2Ze>JphoBKoo>@L^$nE^=DrBW_t5ynGq%F!U9}4$dprxwB0MK
zg&QBh{}&7}qVjAl(GqRnmf7<)9*SQq7fJctjJs1O3vc|WcIzL2@)rmV6Os0<rQ^S3
zT8M7$`t)0gf{~<~xcwuAH-1dM1IRB)SfqR&N?ND)R!3h?-Se+xHGhT3&t(LBG1|V(
zFHN}F0j8NZxfyw@0E}GOg=p7<vYNBb?B~Xn;yI*z9!VEo|1mN>w)!#O%FP2~QmBU`
zEAA^Ed!Ff0Qa5|G+^qsI`OJ(*R%|Z3@gr`=NcmSeUr8Q{XI}Zc@X9SlU6*pLa|z5>
zlks*T*7KNs_Jmg)N1@;R6T@bSFq~Pz>g}gn*~QqJC#oX{{qhuYZH||LlDU1~O5OEw
z<N`=u&3A~(ceHr3*|4^F?0Ig+K)nr$U8l_?BJzDt({y$&;y6>N9}vjRD`*f&dLp_k
z+WA0k|F@YQBBkb{850o-!z+#=<29=vFQ0sY=^>&^UrPYUxN?f=dp}j~f6>dGMzI)P
zPrM<)1|l*HBav(@6!L8=+NJMnnyyH>j&o*c(D!}LuVZ#C27e2Fg&3DsIlkrzduT5z
zj~Su)IF*P;Yb$-fF>BjXcYZYg+W!&JI+ET+q)fnsv~scTN0F30v{wr!N%?Ti(?WFJ
zm#N_4Xi_PeOr^V4^^{7*L_7%ynx<9CrG~~P-}m$RoTlp+=JL$70Yiq2BST~y8AF^-
z3u1_zAt8DMJ|Q6r#yCSjoZr2qr>9y%t(%Ik|A3oO077JQzL)_3H^SDLy)`|Lb2Dm>
zALeEZK+Wm|7!%}GjZ{l){rj^oe}!o#Ak613m$^6M4Tw)nr|bmWI2?|IBaw2sB!pHf
z7DAyg5!rSXae+=fWL)_+c~#^(kpd-|9ziA;C|5y6LPCRj15FMUAI5F6B-99C2#V&~
zzqE4c&A;zV94nvN#f_SUOj%GeM5^wAFO>Usl>2@oqIJPK2Gft=RG1d3fAG(;`@ZF7
zPcbdLSiszXOquf%-<48<01lSSQA**tBL4-Ao=Ll)r&dPVw{SB-r2OJoWnibOW=kyA
zWU7biS-pXlBv*mSz~nekQNNl{RS8i6^6D5PNmq+6Pi*|)*&Tn)^=p`SWF$RgtajTU
zITJ?<Z~PdLqtJm!xfMOxn%?}`!r|wt!~3}zC-rt(NaVuGhD&DFFKo(`U5KrFZ>Vv7
zWnib98v-Um%X-Y_JI0Rv^ZnM&yBkwGC;fwYRrG-fI6_K?Eorts&Yl~SujMGIUj<`i
z+_!SIn?G9}d&4@thZ)!A@kd6=1L-I4`W+rlS5EEn3!^ABvvGsjxITaA2hPONw;j50
zy-HV*%W5vQ>61(kXI}m~C?AE1K%QHnXyfM}+*T=NzV_lv4i85-@pD|tS5We)DTGh2
z6tCvg&&O2iN*E(aM@zKCH@tsx$6qr&vN$>|k+O?M!`g7`-2ub*^26nmJCs-DMr0xA
z;#=(LT6}<!ao@@&Zu@N!shisQB_2ruFa)Itd|-dnPH&ve@9g-4n3pRa_%?=9vuBMX
z*b;n#KkgEJW%c6`nbPr+7*CV$mS?IXuUSL;m=RsHfE{x3Eaf{)=%f^oWJVN_&%1wy
z;44rBg295dn48X%g-qI|aO*urT~}uJS9l~11VECAY)OCE0SAltlUyA4{*4xEh^+Z7
zf4W3xx=8|qLx{s)_4?g4_~<g8^MK<eNKLs_v%Z@dQR~d2q6+{dQ8RfciCUN$QD(*g
z=2^In2#UaRA!=Y<6o|lffxR`E$BP{gCdu?r>h4cuc7K&fA!7i<5egnW)4>NT^xTN7
zGf{EFI<-4-`-hZMa&to{-i){fnT8sJ-5L5=E6+GIYJ<Vg<8#Qkbjry)KUNvo>6a!J
z9@d>9wR4Gi=leEhKJ)C&lb{RsU=LfFhIEvzT@H-FNFH?YHm6|5rBh1W{-JXJ3to8w
zg#knYA?v~io4KRl*%4J|psxrJfs4%UFUL1NB_g#Xo!LGe0to8vzB=H4sD;-E@YO8W
z1EO>)q2(KpY1K3PnSKrA%Mhf3iUXDl6&HemU859tB}8=o?hSM%E1y5CesxU#`WRbT
z%eHlb<=l8WW{b$2l~WEi-NE!|x$ie3lmsO?LcxbkDeS7jZ=RBA9ekR0;h-@5N~ZAo
z4^#Jkn*0m%t`&gEVcdpajH^$q;1vlB&-s@Sfn-h3W5wgYK&}B?1G<(AH747(iaoN1
zZEgocfa6^g21<co{IPXvS3kWm0)~K)297VEI|xyO-P;c38;$Ye%3zGGe!Ot_xgduW
z0bB)X4c@z)e|^9=`T2T=lH812L;IYm-sD}sE3F)|^Ej9!NNDillq&o1SR>Cl#9avr
zm0gUqZ&Oau9zVkL@B)<=hEQ>Fbt8XxHRURFWE5<lKd~O$(qPS7ULpVkK!#xZ{J|b|
zc!W-8AQl2epa?iZ)KI@Tg3A-wRS#9~Mq<23GB^l>cRheilzM-~%@~kk2o)cm=@9=k
zA^R&37I1Nb1=GX!@PV4{hmi?;viAbO6oA3ua0xz`l6|CAYGa%cvw7mKkK_(~8)%`G
z0Sv)%aZMwxU9O%x1Y-rTJsQr@P?kT?4VDX2WzhLtgl~Y9qo7JCD=y#A#3BalK1ns^
zc@`c5Fh)aJ{*G=sH3d?^jjHy#9KiFvJt5Z;8=uM_`d;8|=7>2Lo@`=@;3rqKXpxKR
z$idX6-*YEVs%q{6jfo>TmOr*}W!HVpgJ|+*BfR}_@5m8*=pYL%Nctjz>*KN*yJw|(
z?jQhwDF6{n0ggulQ~aUTFj@f11<tNawFpr1fg!Ypu_pqFxUTMPul({@37E+)nnBK>
z>;Mtl*i1*qzz{daK7JG6+3{df63*>+G$)5}wGdnLxU!3tV=pohM%Sx)sy!hdu4i8v
zkWq0(ImtM=mX{k!-SbIn@MRzc>|C}DBH4cUiFHF2^~kY)c5HBDsyWfNrciT`O(?LP
zD`Rk_$A0G;^+GRH9MJi+<C_2k2it>eg>CDiV`Eq7oFmu@e1b=pvEORsEm0nIU2of}
zwzJvyQYjP?eI)_}!DfL-0EU3!JnL=%&S1O%`gs^%p)z~s$O7uNrvX+<`JQK(A=3;6
zQ3sf{@70yvmPC>nky!T=`9t4hcek^(%UOG^OYoV9^@DNBx}>oy3d$fZTNokfC2s#v
z`Q!`07$PGCthnsq9#wYOj^06!KF}T-==r_e&%a5f3cJ6XPD~Kd{9Hv0q2jQuT{Klz
z$HpM6gH*Fe0=l5d62wDnMXl<c07G0nbb(Otp*e)Vzmj_jo*R=djJOA*iIKWQ|1Wxv
z`PywUc%*?HFHznlgUt?{5HLn#dA6m42Bt_6&JxDK5*)7U!nqBpW<d2kFNhJit{V{H
z+!y)f;b=lB6^q4J^{lR1R%26hw5F!1xuv$QzOKHZRLpyxr)fIl{E{CoH1Mr>=R1|K
z-1V;J>a2IXnsqj9X+D)s9y^q&3G2Z4<u0H8FcfBuH`%9!67daB+Nbv*2jO$v(9CP&
zxT9aI!3>4l?|5(1;QxI!SBw@s5S-oK%6cbR$Y6{yF0OKmV6+NYV}kcI;*0$(YBIzN
zFGR*dIy+NfcebJB@v_Hsjm;XrE_gw)O)377R_)m_{LLgF=TU&j!`=F!5zUbf6MJ)L
z6aVB2KI*WuzKCdidIlJ0y2dIVGj!}}V0}}-r#)N*(ACumfK5$J0MORfTB%f|ltkeB
zUVVLiG#a%mtF5igG!0GDgYCxVLdrZGO%Rc3hLWjtE|;mVUlv?EIx-xK$4RMtJ{N3n
z2ABE}Nrq<4UGFP499Fwe!Qm0hmL1Z5XJ-6as;Sw~<R2Xcje)xK<>x*?N-{lcjUGZR
zl)Uq!&P1=oGThZ9e||tJi^rOqwteok;|CsQCr-Pe5+V*|U}YL&Av&D_U0khy0wO>l
z+hY%OtJj98SOtbaaEa4^NGX~sv-{fB@o``{^OFF|x@voxJzS4}-Rn=-Dq$d5F0E};
zf~(igP)I{XQ30ndI#N(iujIoPjaHS(Kn0yB0-e*DEUsw)00PpUdBE*R^*GUVI-O3Z
zgb*z)EkrOeG2uAQz4vZfyLN3Vm3;8Q`(J)}&qEJB(9+V<(b>Tmj*pMeW&H&njwTS%
zwyR?!!%C@qE|<+r<+2&8Qa(FAUa6EBW26+(r9Oy$HDA;HxZIkq{P0IOW<yl-l=^gs
z`2Hjo4v&krBsIoW?*tf}q`Y(|(;rewq87Ht4||oV_^S8fJ39C<2`3(1^H6x@w!)!j
zr%vr9J+>_+Uz+s829tvsyl<r%%~RDSgv$<cdZ<K}1KrCYqN{y_6w)p`9;HYynkqp$
zLd$B@$+KW^B_&6dbJaZwe5fA&$8ldOWis-l0>SXUcJ)Fp!E_J4R3@stM{nfS(<|_3
zLFGLqKq;kw6cG^;jpRrw;K;XjA%MTTzqU4=&*wvCs8A?`!{N!vDc|=w=cQ8l?AfzQ
zk!4wtNaX0zV}(K?8jXyPkIygI5J+656hovr#-fqX70EyNlibVSQF{ipxS6S{-)ZHp
zg6|K@l!>z9@+WUshtANc44B$nY*yq#7W1sD(YWh&<8>Wl5jT6LI<!wIPctF~M}B^l
z_~p3lt$^uz&|b?{H_D&B4iWw8Ndk)adp62nADfe<ivXVD@4HRC&`UXs>Ad1mdlY}S
zm4ESs?-A$-iZLkJ?7?m*JL=Wb7mR@;lzixn;742eU!3p}fz4<O2%rcg0Yfk~1n{d)
zkO#*(pV^)U0DRw9igaCbT{l=d3C`_nGgIh0KQjr>Wz5V$VPUFL3E%S$H!|t|>}3&)
zR9v+thWDn~HwI-)14RIcG?M4ryXfTPT<h6MA=ml+J?^hxwTBK=$KLSrXHXc(bx?``
zG7de_$ahUCBgAQ}h%0L$tiy2rN`?nRsJQH&l~A;)cLGf9>UtR>_zH?PySE*Vjd`3$
z1$?$$e04w;d@wnvz}#A%V5{oYjuQ~lFL-?kAjDy)3L1k)8`#cC3eO%OB5(vm@Fe((
z3l0DQuCp00=UfO5076Wc>IWND&3$p4Yw%<#UxbseA?^I)MP`PTQUc+rX8xTaIjuSn
zfW{$LQO76P)=sF-fo=j9#BEJ9UcjuwVrgL}m{Ah|PXJ*KM~hVUX>%=>Tw)<j?doT1
z8llEGPanRJ7A4e1u_>tz3_)0*Q$i3Cn1W7cXtKn&bofR4Go9K`N99PB!hvB$1ORMH
zhuqx{?9!^dPmtE(mlMiY@K^)OyXQy7=p3c9^Q@H%O0+LtwPRk{%?kyXF{-$H(`vW>
zIE|Nu!SgPCtd;N2sZ$jS&3J(nLBya#XCSO&dmY%`RXjMpf<&0LrPcn^5Z1u=iR5f=
zh9U?H`1zRHUdwd^;tVQ|Iy}m@bzRm4hzO3)?rT?v&VVQ9;X+ab5gm8;XXDYw9%>mH
zukJ2jLPrS*4593>hu2X56pUSH(O!UYLW6IemiMJuQy6S@5f{78?v8~5?E5U0fR>2=
z@(B?&^S0Vv%VQe+Y)oC4YC{AXSG)V!U2VXxXc036+hdznsAFSbUm7g}Fc_S+Xvm@m
z>siqy9?|LW2ms*vW~f}`%4Y}_2RAf<ujs@CgtU42t{FfX`T9LaaedPdV)aR@>N`Hz
z9+VyS$Qtl`wSN#I#+6wDL$DS6WK@2nMK}^}@QPFm@<lVGiYp#iuTG8PaMlr`R)>$a
z@-O#Gj+c_Nf<ey08WUHfsqZY9mwMk3!SS#)#adG8rIS}+qy%7a7^u*Dmh;yO%2$Ab
z(di7|-USmyC_A8?V;~H{@gWi6n>*A$9R)G(w^;#bf_<h1|8e((bLUFYngpe=F3PsA
zf{LSF=!LLx)i)*(#{Cso7ssTI$4V6DH!h85{uGP|j>lW;A!ex8hahZ7NuORRo*kA&
z|9t1%dA!b0uMT2e4X#Lm?O%?UvP~=Ho>L(Bm1j*LWE?n@SC2LDq6eHaaHZPO%N|(+
zA$?j&W(ZXeG|u0-M(sQajtBEh59J6&4}PbGy_m<rg0DUQx|%3rdmS!M&}&2L#lDLb
ztIP5`f*JwO4$CcRmeRnZo0)<cf~&+kZ<qT{@RFTX?3oUJqC`6;DWP4SR|Q(gM3KFt
zoBAez=sb)$gQCsuUj^9;9Uq6#RkU0JFgcv8(7PJ=DN9v+;2NjA1x1@byc$Y2<Sg(d
zHpcm*Yt{bKbUHJ4FlNv#c(AROwS@6ILo#Wi<P64&G%!Vj8K^k(<@SO>-UEZdrZn4=
zp~wx2HOv=a#;9D??p{q?;lBPU&G>i=M}Y5~xpD^N%p+C-fNxq!Z;XJ(r-S&ML7|H4
zn^;p){l60so)Z!b!SaCsw$!rS847bCO+cms<3+rq6+7#&vmTq1^m0E9W#_6npC65m
zYVgrk{`FH5XXK8?LFb_Jt0)tfXVfRk^vE)vacR;44R1Dp5y^9Ts+MnErJmn6rTB-N
zd4maG9q<!6UVQ?m!znpTMez1kIxzu`4@!aM;Myj(p;<ov23%T(U!I7|bblQ$_%z{w
z1^_yTic2RZC{w1<JRKSa%Y~3OUvvZVeR?H-B}>Ok6qy|jqifQ8B|*-m#~b;sDHYfn
zGl?oe{9tC+-8bD5=YCq)D!`Wj_|hFA5=kaS_=68vJNFmH%Ae>mmKpFj{W2nO9<Q+>
zXkZd+Ot!fLG>%b|-MNA_Cgd-UfG>g1PyJ8;8JFJGz;{h4BM{siL2#&g;P{}61&~^R
z?mOz(G84Zytgf1WKd)wRIBmgg3Dg<(mnp=-R-gcolx^EeDbMqK-=DqLa<E-CA_gnW
z&QzZ)9AGJ>l>W@xrqha*=Q4-!4?RBBdm?do=$Q?Xs;9nwN=3vxQ@aC<hq83G#2R9l
z2-A2$y*vN_SJka9F<%T$Td+QkH5#5M(VJO8fs(a=%oG5RAf~}bTKL}$N`4)<C}%KK
zrS~`SeK{ow+9DVNrPU<1Y`!lZk9T%-EN@!g(9keCI*N#iL_#Tb=bd*{Evuoip{ujA
zrluwmi4=>)Inwck081%T>DpAP)-X*WG|zPdFOrHSLr*=L$_+ep?D%7y(HGD9KOLjE
zz937T0CWMCOM_WDoq=oxOl>-7GykS>=&#TRn)qvZb?ubK5sE&1W+i`Vih9fEI~NzJ
zX$&SEs5PM8K#jqVwiw$Jau+o`^5~=KRI0VDrMbDewzl@*!GoOh_rL#1(==LJTDai8
z=ihewM$hwZyY04IE|<+_1OMe*gAD<5mQt=>yFL^OOX-Khk>kgY6f)UVOH+Jv^R~=T
zsBbcpT>jFetoRVqK+?4=n2OW#=bYPb0WC<dElT!a+)iog`;+0MrhHn+kOJ;r@Ns+0
zZ8Q9DO-7Ppfn8noMhGAYOpbPb>d)F`-c|o|IMmhIS*}!cT~|sfsg9pG0RX8~s$4F&
zuUuKKRD92?iPa399%@{^+_EYY6K91GSM^z52(a&aP0g*kZj6tOHZ(4`tjfgrDA#mD
zdgQ5sW=N@`^QrK>5w<)+#fLAf5<eM}edTk$;`w*tD5%x((<}L3_WIX*rC$V407T#@
zKm;9>>$*W*X8?q*>j6Uz5;+)S!9wqvrg@&v7}GQ{7mxHMYLkL})`RXCL&i7&C_q3E
zL~t`fJ2P31f|f9TxS9V&uWz!2@KuIT_TkUF#m~m%(IQ2}4SLKVPN4t-bZ6GgP-+_K
zAVFt#upu~S;oItP&I2WFCM^hH7B_{PvFQwktFX+3jS03VOIJzryQFK(g^#!K35R}j
zmg3rtdx>YJbS7$aajfWCN_j7#ZC@Osy7`+VNb2z8QPmv6yOUVBWXam)6|1ZZ57e<b
z1D;)MyXdW#U$@MY5P{C%e-HXkHu01Wj+)c61$h^%YVhG@>}vxu%z<vL?I~^{fB_5!
zlP>&dL_X6YB*9e|zTyZK33Ud1as~hTfOOPlf+cU~c1tZM9Qb=rRlwx%z9#;{l;R9H
zUK%rjQ1T(A!yk5ve;AU37DdHPr#~(Md<l%h0^TT4N{1(x^F3K*O9;(!fa1A&;EYRM
zQTEw(@trgBc!>hD{?==^;EL}A5D{=NuXKhVYvp4$owi89bQOUFsQBO$ysLpf*1*3p
zC<iLHEV^(D?;p$&@*Z?V@B_^}@6!I98nwwMh;ZzR;vGpgVbc$WrB4vzw=BAFYi|(E
z5vmd-!JSE5A7f!LosWFhfxTHZP@x(PINpMC_g@OY)BYKRvJa9V#DOC?5?lo#4(2UQ
zxNqu`bDgPcUmER(bbCREK_DY~j&=hTEOM@{;4K^{oD1e(Z*LNt%Q%8S^z~R+F{qYY
ztogd!I0HE8qVI3XIA=!d`~K{$+}|32gK;F2$=23Z*L4?tG~M0Zi#T65|2Stx>$+}3
zLqkJD!}S>LoB+NcYX=b*C4hE8K2T6>Gn-{CxVM==+oJ4UesSK?wf8xbTJ`2F&K-OU
zlq}nJbX^N77#U-Z<C>;%LELHq_j4Ra(=^|goHHSWWm&;q$3fIn2r+kQH&gl^ROFea
z={Sz>`-WjisSHD3aL5D^ZQB+?D5VgYriq}m-Y|?erThH5wQXD1^<YXhO{-KYK}v=Y
zV!8tRI*8;;J36G4YuB!JU3dBNMj=GGTyAJs=KFrRTqYvl_u>2^xH(Rnlyddz)voI{
zHa2LQRxA{nnw#6&TBkBomSr_JH&?4w&-3QebAWDRV`E)iEdaDMH=muHOe7NZ^>vJ~
zR4O?+IT=*L&Uc&vx9{oc@qK^UvSo&$m&;{M)7sivbGclQ5$Ab+Q1mzVrj%N{c1^Wv
zx3n}PVyRGQ?`W4&EnnVfSyjujh)7C>xX3guF8T7V>gww0=}9G1zLbp(*x%p(=p&D0
zbJ=7vIWaM@V#SJAU)?`8HWutVK6jev?Ce~<x;vdt1v}FBojCr!C*P+D5eY}kkh!eB
z-nQ+R_w32$^4issQIt}KVcdTEZ3xgmFwoP}bNI-S+S=OpzUSSALSb}ltbOH5+p!bz
z`2GX06^q3g5}Cs|o%Z&Xt5^5b*VWmMqwCthz^RUojt~9zZ-4LmKj`Z0m^gd3uCDI*
z@#7;SqjQ-7#+YSUYu2pk?(UAoVx03-DmBn|a{Kme0FX|nfAy<hx3shj4-b2uXIa(L
zr-x?P&vlW(M1bcC8hy`8B;o+zICgh;_xg3~Hg4QlDwbBPXj_0@%@BiOd!83h!~vP_
z`$vx*+j!fC&dyGuY2iqurMbnMo8B>#@#(s5I2;OC^i(EODwT9yU%z49md%@`QkhI<
zF1bsyC*WMq;CXH`kyKJ`*s%W8sZ-mxZ;i#{Jv}|iL}GMw9`E<LXD0ygeLo(Lb6p!7
z8z-WX;gRm{?hWhLr_<>to_M@iEOO4SFN+pJWU@KIcz=IC<J_uRg+gKd`t^t3IDGuX
ziQ~tQ4G)h@O=W`2o_X*JA+p&l=d8bf0FhOz)k2||&lg69M+OH6$HvA71_sLI%8B0I
zxiHm0W?NP@pU+QCWd!G$Y{qq+-rl~~Uw@-mEDR41A3l8e^y#7bnj)GIxm=DhHZU*%
z0JdWnilwS$?cBASGZu@*cI<fJ)ZpOU4JRTJLQG{cLWsV;lR^lqVigJn%d%d1W$(zy
zXf~TYcI;TCViD1S0|!E(&>~9WD=1mh8i7(O5{a0mnakyZrT&BY;tSgP-L`E_)1*{N
z$xtZdd7cnLN~x3zgr8v;bHpPsHiGH1ZF^>37-KW*4l%|TMi9X`l~Pi^InJQ}KucJ_
z<IhYf5t*hL^cmd3_x)5V<v31|>3$=$=-C;50RR`MxC?vBW_KHaI(s>w;NSwBLBRqS
zFl$SKIWF=wuW??Z2Pzn3Ft!VxJ$Shn#lZWQ<%?cwY-tK^?e_lxv$|Nx2A*q~00000
LNkvXXu0mjfrt6cp

diff --git a/tests/golden/price_line.png b/tests/golden/price_line.png
index 5992ce6c30d45fdd5abc85ccf2ac02e72d96d5e7..bea628a313f758fa204735d6785e543f49e703ed 100644
GIT binary patch
literal 7848
zcmV;Z9#`RsP)<h;3K|Lk000e1NJLTq005u>003kN0ssI2j?}E!0016uNkl<Zc-rlK
zX>eV~m0q9j+jo5%c7Om05G2HXr$|a7Wy-Rs&9b~N+MID^#^WfdIQC?Iq$X8UNljJa
zWL&PYlSxv}L^GB>vMpH}+a-HOl4Z$~NG+r!iWCU~BuEhZ_O|=()?0pD5CSO>1c(Qa
zBGL6qdU$;A-oE{HpYxriPeXNmGn&)@005LyN+|$92u*$yzn}Os{{F;oCq9{YBc%iY
zjPc~(#vdxB;?E~vd7=-K?>qTVDdog1;ww!2)8s3}mtAZKAz~*olhGAIc%DZHK?pI%
zlv0Ec#+VR7(=?1R07MZ(2*rcT7()n2DLLnwrXhsFFd_tFj5+7Jt}7`$&r?ci8jTnW
z{D4x*Ima0Lz8?dIF^)fs5fVah&g1(^DT6SSLU1mkD5R9eofJaqI%SN-ml8t6|5Zvc
z#^U}jHiQtKjBXf4_4W0?dG?npSFVs!HZ?U>R#q~`YHMqkELp-B`>VhD&xT=ER#q5>
z!8!MQ|Ih#ZpKaN)*|Mx94Gm!!B81x7R!~Y^*WI>l>mU8mSF_pdk01a0FMs)qW5d07
z-MR0Xr=RpZ?~z9y`rdcH^Q~`v^HZO??;GFv)BNbr?%lUM^xy-<q8mlg*S_{I{@369
zk0@dfKlI@L{+sXLb?3g#n>YRF?|%63Lk|W)D5a88?z`jmKl}E#FeXcvE+vGd(`hMX
zCX=ykyQQUhLN+Wm?Sr*UNU7xBamOc;$<$4|b~V)3fBox!+&?h*(n~MjfB&bvQfdEz
z*M$)Gf9jLB-L^Lf!@vB?|J2phwY+utx}m{u{OQ*%%lx~)fBdGKb|ESM&kz50u)qKA
zyYG7S)z`Lc-BMp)w`ce6$Nt}Ax9$B{Wo5<AojdAk>oC!F@7|qBr!gUn`R(l|zVP|a
zJ^tgLD5ZY=>t{dy@!KrR{L$l&r_<@}+qTu$*FW*IryhCuq351^K3^z&=R5x;l}zMv
z`7N6_ee2)-o5M%m*EH?;@%FWA)>KqfjE#+c{|EmwpU-O=Sw#7R70ei`uB!Ut7e3!{
zqWzPfym!^ARXshu7~?goSGle`JUqO4(<aXOrcIkFt15lZPp48s2!bJ#sTdp_`rV5!
z?cMvawzicAUVnoz1{fQ<UMv<LdHA8xk>Ta7%O8F8UvJp3zUOTBLk~V+7>2IXKmYbO
zj~qF=Zta>KH|-TdD1<1bLWqWj`o<*<4Gj&S`ShnhdGFmn`{@%w;BVcsDGWo`b<^o|
zYwPk;Cr{b7eb-(4blqrcYdiSX!T<Od|KYjko`3)SquaM_Yi?c^MG*jAT*2cy7yvlu
z)z#G_W24{u-uHX^`c|xHUA<=Y(@#D1*1>na^rbKE-@pIypZp~B!(Tk}Og5YAJbCi)
zd+#9t#zlL3`>C$3vuAq_A3pr*tNVBC*!J$>_guGVTGqC0Ti-f(=>PuhV{aWi*wobc
z^CzEr?6DsqfS#V-r=R)dJ8!>3G@490$yB<sqC!gX{PQmuhOuSy=6pU^EV{O3jg5}@
zA^XKIo>{wQ%}@T}iJsnG-}eUx2P-QpUwGknCptQZheuv|>E&ELzh%po|NhtCt*NOw
zedf%;g9jO7F-y448%(h17-KJAOjTrt2M2Dx^;VP6-@W+TRckkGT(|z`Pd=GRr6VOM
z*1{+RsVv7K1iSenF(|??!H|95Po+{=D4?Xes`^w%r;|*wC<>!Ur_``bKL`Lo*LBl0
zecvyYN>ZxjEzLm~dY-3gTB%eD!^kvEAvov4G!05=zEH3&%d#wl(77)d2q7V46osGv
z{O5*;hu?VP&6oo^j;(3hq8dIA;JDcvmie6nc3VU4?UPz_wULWL$wh_fgb0>oq9Z+|
zCJ9WE+LT~ih1_VW@{e>&Hvo)da|aV5I5*ZbvfgY|ETy-s2+s^_H7QjJk%swbQ58c(
zu#z%1Y+_~}h9LkDLSo|wV|<QagOo}sMKG31O|b6C#uxx97F__aZF|BzTSWImW)cP!
za;>RCt4`|c8a17!Ze8nl_M>8$-qPxw=rh(VVcjG8swGGx+OjIJbW-8)Y*A~fj7|=L
zp$Rvvd~{!Qc0@^)yl<-*Dqx9>?XAZ8r8sR#F7(z~oN-uRR&hSrdX`cWLMEHVF+xhk
z(?B7F;=d<;GPy~q48x$5PW);S*D38_KvEfP4blx&$W;)DoQH%&2;rI}$=INg@OU4o
zaJ0H~c(jL9ItTz8l${w-rBFrOXsfUPuV0a+(0l0^sc;ZNVx$O`&Q1*h0P7kS!v&nM
zMBbBL1f-k_auL^=ifABM35gI=5l2dat|5)YLkIv$N+cA*fB+!`iVJLLGGIV8#09Q!
zw3?I*7(Y9L0FXjR!$>H_IOrOZ<Es-eph^C(#kOz&PgZQ>LR(1m!NO4I0izJY2uX!7
zMoJ+-ic7=<!T^Sb)Or3-G_kw@p~aPUMI0RnA;uz%2>{|;U`#~FHA*LWo6lc%F;o-a
zDFhlplpuxB1=4czpNr^T866BDglJu{QXyrURzqVGi`a^`8(g=@Ih$D7m(+yguf=vn
zr14aWPn%ecLI?oJWGc>fpIN(MV@0OY5B#QO&BxzA>Ntr-3Ul?tE2WS^Oedms-Ox16
zG%errec!VkXQ00y0EFO+_<j_cnvP^CrIM*MQc6ex0B*6E&QuigIap-8*9?mmLQHUJ
zLWmGzF|c|~vFJEruat@@@}fKUS_60zoC+-_zFZS+F`#xW@aJhfgNp!v<k3|@q(XKv
zj28j^h@;E6v9?KHUaKOGrg3yEwvVQQ1z8LdcWe;ZQgEzSGpQ06(x&s$W*6H>3ti<!
z=H_MvDF4K3BxS|%x<wH^ixETwBMjqjt~!4YC?OO3Hp;vY8o5{qU{OS0gqI>D-4F>0
zI#E&~F2K^kpm7GoRk3_afw{RE5eSZ)#g?uv(y>K=KRk9=g-i)y?_6Q5Uy5x*Nht>M
z-hmF*J%Up<Lh33AdV-S5xoxfg#!0Me!jER;=vZusU|9_4l1%!cJt|`1(H@a4A*$(1
zt1ABGZNbr=;!|&cNyl>>t{{2#LStPM7&Pn{z^0CJ-bKGV*%v*riwhwX=QUnmkp_Zg
z!8g}6r61Yj{?p;$Xb(0tL<mw!MI3aJyk`rx_0b>vGXeyJTrrFZmU*x8(S1?pV0gS2
zsKy8L7wgP<Bumq%QcBkiLNvx$oTqplnH+9ltZGPqbx-c`SHku_QsIE9gNA_;oHRk#
z@;^Tib&pnm`(7j!lJd&XE<nnU^c8ia!U>P{f~8MC&&~z>0iq~cy=Gl5m-D<*^YYfQ
zk>T3<hBI9!hXw~JrPn&ELMjqkbu#m{z4@Q)mj$0BEhXd!UO<TufL5LIet!%Iu6lIe
z=y#vRN$U!vAQ3FX!1>sk@crKLOvv*$^j=g4ODT)Rf^O&|!-InJR664p3YAqg1ATp(
z?Ofx52LQ#Q%CCRIfAJW~`ZVbvA;;56FBsn1Oz=`W{p|LNhi@%B{u-%t)B^KrA?eZ@
zeOXQ6wf!{ZAcPPBq)|{xN+|@tD8K|`%XWm|tt)R3LiC(Hv$AbfXGgoHX)Fq_S!k6)
z8WCe|`hlCH!CdL>E>hvhAh@Xeo+FgdbH8|_;*W2m+n4!op2A5J#cO>*^oYoL>3wU1
zlLM?!!YK<Wg}CG_LQ0~kbjF1w3n8SW1dI_vQV78K8sEYNph8BQt22+>n*HI6VA4xd
zCji9-c8tuIZyNb;zg8hf1Qt$I15|zU?!wb=jt3Y3Aujn)VMZh?-av*4iQj0HUi%yL
z04Xk0pWETSaAc+mRVkpF7<K)FUFipIk#2w%W|S~UH?VfD5Tivg>Y|x#4Z+hs$h98u
z$5<A8XYU#n@~FKZY-6UDkCZrRdHXv^+O}?LlSLolg%Hs|iM6fOKiCBpozn9>JAkkK
zCM8ao&aG?Q-yOw{K07z0lwj(`-@co?XEQeS84mYcmai&>#)f4|NY*n3`b<4k*D=5Z
z%YvV{V*~3M6GH{0&DyC4hQ|AI(V3CteOqP0M`S(+6F>>++;T(k?itYV%vJt%3@{U<
zDq(JFF1^x@69&2rmyVD)WtM(_%vje%msP8f&6n!~kn)+{T8q<mbfO<jedbOyJbyC|
z5j;-w;-K5rwtU^dxqW@qH6#n(<u5(}#kl|G$>gWED?geqwuP|@B74v3@K~P`v-6<M
zTZjBe<~+r?Vp8Tk=?3Vk+BM1qsZJW}n!MLLv144m29*jshJWZZN!#Y8W>pI3Whyg4
z>e95iI(V-e+q3I!xuZ!KkdU(AnVVaz8(VS8LI|l+7#!~QUhf2rfI-m}_iqy{3#HVZ
z8(H^=EciHO&#6la0VSoEk0tKe93JhNm);gjseK#R*-_~R_#&Ab7yWfPQ+6<@kb{OR
z{^aAv`lhJ8KRj|aIyJ0<$lBiO>|Pyp42W!LM$Ux!CC>_MJ$e7O((^~kQfL5ljrZq_
zHH|<?)G<39IU}=M*OH&!UV8C;_^@;agrt(CaQ*?tpaO=|cJ;UJjk<=4|9FUxxrhtI
zxEL<_Z=XR@W*)eO_m1(Nu}j84=GXQu`ia}uBMk@dc7vst!3m`lBA5^46L)U#Uq4w+
zMT9{X1AEVEoG^n!T~qXSfdG!ZcNVSw_PwRwAM@WlMJgQ7NF0p;s$pB_y<<`5VCIoq
zdEXfC9V_!RpH*iGPJQ;q(o4scC|kfAG$KZdw6RjFO^5Gx&%$SNNv<kU>WjO)S5Byi
zPtn^;0eqpWa{Paj%Dm2;2vk10ue84-JlaDl93<oiPpA|U5+^O0_i~T_A@kK+X=A1G
zqboqIF-SKsx3@|+-~&0J<=a{(B~F;7mycW9TS;{iF;TAGTHBVZh_jy2nL9Tw3NT}l
zQc7^nIY$UZVK^giCBZW1r60V74;6wpPvHuCYL}3ZVCynhD*ozF#h=^`8bMMmc+4CC
zgS~5|f3ORTa#I0<Wx)&E`;zx=k#1T19)J+odsg{xoCJejo+|l3L`x~_8ybZWj+1Z_
z$uJC>np?b*E2W&Sa0&+H2IjWq*3K3ApYF#gd%F0jRA5lnm&FOg-nFvyhvV4Nl~fD1
ze3ciP>zZg=z4!7lY#U|1wNju{c6!*keI4r>5;+f%Iizll=S{ZI6?NWA$FO6li&6bp
zL@O1G-Kou_xQ<~em8MiH7Tmgq2HUo4YU(tib#}Ddjx$|99g_acjqb1B)@%z1LHH7K
zk5s3^S31l`_9SoL5WI2%r!6I>j%Aix^@8xYbL$#*v`^Cwzytv)`y7&yj1HYie|ksu
z2hWkDp@g`sM^k{YBqDkH`rzOh%`z|ATR;d&L@6jFrId1-5)KMUDQ!EE%jPW0F1ao`
zmmsR-Maqm3EVFL<q21A`A=W#F)3%z4LCFDGwD6B_WgfUWJlZ4t2*~^=DFTu{*Opb0
zibV0*cd$v-e3;_`fvx+8y7UcA#+K&ba5r`?%L$4>7JXy$GDL8Aq7Nsn>3XZA8n5to
z@su5<ltwkzEyjh)hKB|R`};WKLxcU4QX%+-0K|pR!jLSjq-}NC$6f|Umz>SeGk^#d
zS=T#wD)IRn$A0k7q|%x9gmMiF*H3(Ad(hr1Lxw0pYJTo0r6S<IaKyQLgZJ)fB=|Vh
znB4@%QgHjWwWU|v6#>clMSl+nNhxE1rXt07l6{E#`+5fY`-+9a>64v|v9sN$KggVo
z)6XVe2oS+4;Hmq!`){5?Tp$cFDY*0t07-?DmUp02t4dlou261x&T&0&FD|v}l)j>F
zd{vECF_@1R0!wG7M`YGZ-n~iYeWa1`r=H=95LpcDn{Pl;MJETr)F%knbVE9m?W2_H
zlv0ebZWwXJ-{}nBQfRDiA{pC%=QLb~J6B22HTSvqlJ{;w1kZ2E3&GM2?VH+oUrrY0
zKR7QWPMGd<?^!!nYD+3q$Y-a8l>&n%ZeQ=d_&%8W93>4t|H^8j-Yc3uqXR&3nb^0{
zJJ11_2aiHvQr0~xMv93$H_j=fGZWvWMa@mke(Vfd0NzeO`jLO2GyRpDl^4z;%0#ft
zd&zq?v;Hja%Yv@Waq3^zm;xv-G&d~+lZMCoF5mAM&-53MT02&d>Qs5015*-~QebUe
zF1@hK&W&lO6n0GiU>9g4amNOkE6r&70;qt|W!2`^7WajtIBCv3FoQ2|$dnMy$JhC9
zoSap5!PGOo7z9UqllN_vZZJQK7!X3%t~USBY0$NW^N&i26Gq{e2c1u>r;SxAV3$@6
z04ikZ2XFEIa6B&ST88O;Im@>c8tayVNyGNOxrPm-!U@wq(4n`~=qu_}5S2M*450kb
zSltMQ7M&UblP>611cH?xxzD{<@h7(-DKFu<(});zQ=i=-y)Za@7UmMAo)f?r6cf(g
zwZWm&fXhUVg!JC%Oy0M3M$>$@y^XkY>+0ZeH(*%kjV4loV}!^1!p_0Ue|3k*dl#j<
z8j)Go-nqiu+EV=GA)GSHWjEs4N;~C8dRslo*x|8Wu*(q<QVJX+INXhsrn$LAm4dQi
z3`hlxHrJ4vRCu)K%ETyz#A&<q!coO#<u~q-*%FfKJWT`$G$O`aV|8QdOE=~JVL#Gv
z8U1>*1Q=r#FlX<Y;K<o>I&u{Pjg($Kp15mMITsnm%8#7AYr^Awiis;$gF#4~whB+b
z2^y*T_Pt0cnfJ#15&((`K3uSOuFQOGZ~lq>vJ?O<=f>38#e)?WT3y=t*c$iwBiPi+
z2nOgH@5>o$n~;X1j)C$-i@79Kj&sMx(sS?4-QJk%3^r--?isBrk^0PyNGdT>RDOgM
z((2ObFYGe6x8|SNFY_LlbatxuQt{vm&uqAaiIy#;zj8CuaOwBQaMCQVL8*i!X*+e_
z*5U8`YOdsbqPJhZ2^qBXyQ5|A?PVHl#Xzf1+c)2U6Q=SRXc%-Y>=^J~?*vMKl3921
zQhSqT6GP?#S2+5rhQgDtW4rwB0-zf2&#{4AV&8`1({JI5xjCqrAk|5IO=JG2`)8Yp
zIFH70(iAzb@YEYPY2l=axD>gP;sQG+LTWxbiKntGK4_EmnJk6&t~NfHACGj-i{WUN
zet*o`)=Fwqit)MmCIjc=>)4qQS@03LYSMM30u9I8K1H@9i#`|>CrqT~tFQ|`e}rYc
ziFl%YP{Khepqjb0#Xr~u^S2d;fOLc4NKfkiZK^m2zXubfI%%$7>K*97&egGeC+Xw3
zKS1z!M!<Y_4^OOG&iSe}>#A#N>l+%^Z`@p6U9)A|jj2>x2mu(&Qef>|p(4)vvY?ZB
z6k-X}KhQy!RvBxXlowu(--BT328la2h9?JQA(-DSLsxw}uY`kFDkYOC!!*J$+^~6T
z_vx;N#-_2+5zBF~F7=vp^iDUmEg;%F?r8#0ItpItNPli;_OE|~^;zch076XSmO5+g
z()^EK)hiO?=Q~}iZ9JnxzjO%)0B{lsgiy(KE32wHPqf$7H=OP|i3wq0s5Di0;fPk3
zLIe<F^V?~P+0nDwx<=!+Rqk)z!I{~xErMmfRQcd8@9k5}^>D(xy5#eG06bp8VOkcg
zl%kZL>O6rEVvP0o^+ZuLIMBx!8!zmy5KxQ=UWoA$5+@Azx9?UwvL|@^w8;B2wf!+D
zH!wCWBQ>eQuiw7*(VZ>HVvI4y06+*qDdn7>SHeLpV9_9gr5Bc7JCS*KkMd`{qe~%#
zA@!LX-4~8Pd<NTf-=?V1fzBPgawW^$Nre+;aOkuWGWn_Pa@3u&sL}bt2#ahf^|>8<
zAji6fulx?Es}SI;s|ksdX5pE)>|HC3P0M6{TH~8wnfI-&EylVg_n!{qly$X~g<S-A
zdPa>9%0Kyf=8;>qx|H(5sWB#4`H@zgOh0f-;i)&TbGzsx1+Y?JP(E5LKKpjnzrBZK
z99i^1Ba`Dopb=FH!K78+yxV>5Jw94o(6o)KzRe<<1RBAPab++b$je+Q^Y9+tmyNoH
zfDnWM02G%Z=h5cs>Tlj%dgXX{v==9>>x(&;ar@|%0XW9kq<kR9x<}HVz0p`v$482U
zNu?C7bW(S3PTaHE{q5nfz5jYty=4m*Dg_25HK{ANaT`>~k(AEv)l7%KSatW$Iu3N!
z2(7PJdhLW_0&M+yBvzLS7k+dAKF7IaDMWy(Dyv-PhhBh~z>aa=De{W|Pc|Mu8cz}1
z<*`ypm<XwnQl4O}nYuI;sPQi?9=7;V1(*wg0XXM?F;WU+EQN?enL4G6F{u>BSX(>{
zcg6q%LLZ#paAnsn<GiV%A)CuvrpY+R7)qYEd&kbf;bEQX?{~EC+PY0C<OjjwV@D~a
zi|hCj04t@G6qpds7}2i$5*I?$R#mqxZ*8cp3&OCjy5>i}_<2obRZG({jPaS?vyJr)
zmT3(P4_*7kk1u0|TtKpvQo62Fsuzoe%F3!-E(-v0y3oRtEQJsT2%{*ONMzFKD2fWj
zqHWvRTrTM(xZp}Dz}V1r&V^DkUo2RLDbH!^>yBhy@^DH3kW#K*yI#|%@0ASG8XD~H
zK6Bc%EF~lm4QmVPBm#hlvDW3wqbSle+A}b)YwPwCCp*&Vv}qWFBf~c=UlAYwBc;TI
zG?S(6Cr@hEm6QgAki<kA-x~U%y`~?Qz-nxCbma}JHA-{YG0S$8QUDmmXZMACVS5kU
z4>DH5tT2;_L@}Fhs;XVTqD`l|km8x&JX@QotgWs_0Qo|pwyHXjNOYX&RGgvf>bNH<
z0Rt$IQhs=1pD&^Tgy5NUg_B5(j*hgnw4OeFDhdMt82_bOV0;l{ys>eKWm!bi`uqC0
z5S0}bwrwE<y}i9HEiJiRj_SH&+hf@=%e1<?&myD{x}E?YU&m<5Dmzm+I+T(^NJ21U
zoKiXw6$SvcFvs=*0yIwG0U)-5kWw1e00HMb{wGRl6h(v(DWy^>eluNFf~l^*5ElIW
z585>6CQaqZifNU)&`c6SV2l?a3qUC)r6lK!+KD?(KpxLlDHX<e;=@Z<CAi4-=i3BW
zfI*900#h0hc<KQ-f3V^tp4z&4!!YLiusbQGVHmY_^+Je=i8;AO&^VWnD*p1hKjp>h
z2wn2}^S7!NV9?^bq%)axCNt#|;TZ{;g_33yZ{R~8mh1X(ZinvtJy(2-anUkwY+z#c
z1Dq?gGZQ=b!o@75kV4ZaLWr@5QvCvfSw^ueIM*me2(c(afDFSFLd42-?)QpKsrb0f
ziOWt%>GD2HoU9m2<{$`kU7rx1WxP@BhQb(g&JY42gkczy$1{F2T|8JyY1wvFRgLQw
zHA)kSr0<oaluAmeWEtH?N-4uIs%z^?u8T2lXlx=xD-;TbVI-1Cw^*E`zLZiXlc{tj
zQ!Es0+esu78l|@F6pMvez?bn^q?9$awM5gxAOL`x+B)C&gy2#tAvnfmQ~dP}jf}G>
zimGesgy2=xHD0O27>gCwME;~Uy%3O6Hn+46^!GYWVo76DZC$;P!Z1wE*vZZl<r^0e
zLM&@pJ~TK$b$#9XP3_0uZ)j{%QdU;iob5i{*K>A`vKj!ITUvX2x>M=&#?4!UAQ&4R
zsi~`HQE1uDnXZ#T5X@a20D#~;kw{fk(&3?@>YADzyKa8-_5C%qb;UwHnM@4~^kqj!
z@!aMs07PL}Q&%sQ%w@+mY}(>^UNV_{|H$F0>YCA!VT4er<mPiZP1EotyQ@)Z7-kfP
zP9o9J*wogx8UxslW142U1Wvp&O4BsMFfhg=BST85))j5(Ool~KZEgMB887F|Fbqm5
zW6X73-}fk`O-q-pykWHzB9<KGa<2-(ZQGGjEM3|>GCbUN!)n`h>Km3giDWLD#pPh{
zF{89Bn-G#68wG@N*=*C&<|U0y>2zl8`i&*mB?K#_rW4WeVbK`ld@fs2S%op~I@LKa
z&^J6hl*?sV6qQj#Q>tT(y^<S-VX;sc9v&<d^8LNNQ4|gh4*0$|m*o|!0I5{Tb-CaJ
z{e3;%XMEoq8y)TK=^h&$DI?1T0K+h&DD-`wG1hV7I6}yF679#2jtmXPg}CS5Fd@33
z2c8!Mf$w`KJ5Ce|dAC^T={XC4eBbZuJxi%R1CpgwynDs6tSE}2D8d*3@WrgIxnRIp
zL^TQ+ODQqNF;$7}$hogzDJADj*Nq9gSqPz2tgFg=!fNb_p_B?CCKzJ;AZ5H!ya$C+
zDmcePlTyUiVj__U!*Hf7`kXHY$w>yg8n5J8?j8OBlf4#`muy*d!3B2FJedO*7%-E4
z3FW=fhwd`|K=DkP_}WafDgmxD@9sy}Eqy*xixKi`y#0SXX)8Wtk#Cm(0000<MNUMn
GLSTaGL@uEK

literal 7678
zcmV<a9RcErP)<h;3K|Lk000e1NJLTq005u>003kN0ssI2j?}E!0014uNkl<Zc-rlK
zX>eV~b!MOLzS~>z7J?uMf(uA-Uqn(ODUmW|S+phDl2_T0Wx0|X+wpj+;<9J*Yks6=
zDt}TnDOZhuO(hvSkv+~>wq$!E+mU2hvcydyB~pY)5F|l@0Eqpqckk{qKQ0J@B0yq!
z05^2KVqHAmd%ItM-KW2EPM_0|Z&*sk6aWAKL_|aYAcTy)h^|NfjP4(OcJ!Ok3#AkQ
zP)f%>8-0n0(f4EDOy*0i3yUS{#-b~wbR35<MhFSR5D^(;VHip&Ip>s800@GR5JD-9
z?jeLIrGyZib3%yk2aHilsSrZfbyUi697N<?3&PNIJx$Yu5R_8ab(K=ltD-OGF|K)`
zW6|~fprN7Rw=e&C<;oVNR8vz^Hk%E@u)e;&v9U1>!@vF8|85v&Hk&aFLkQux?qC0x
z|9tPgTP(|JT(rpdeL~2pRV|vPRjbu)+qVAspMSkjDE#F4fBwo>o*o|P{?sS;z4YP>
zj^jM{+_OLU`@j3%_rCk&lb`waxBs#<)W3W8gU^2TnR2-tjjj1otHQ6$WyWl~`skyd
zNG4OecJ5r%(D1EqedAJZ-y3hd`P5TSI<|e_;2|mHQ%`>S;fMEnzW+CW^WQo;I+m|k
zzP`Wj+kg2j%QAob&(H7LwUa3Izy9H$`Yv61?6FV1{r3KQx8B>(uyD`r-9P%@KYDoY
z$FkYXjve<etY1hO-@SWxT{=w}3q!Z9?ev$v_=V?x@>410JSwv?gTpY)=W<Vf>5J{B
z+dlp2Cswas-PP4iDP6l}b+uX@7#P^Hd9x5=^XARjT-J5mbSfpKWEAS^GJSphe|Y_k
zy?Z~lYSqevhYp8f2$UMSUa3@```WWZg9FP~EdS1T{?o>d8@eubKKs>Y48zcM?XSQ8
z-4iET*RNZ9|E|4KN+g7H9<9prrN(P800<%S`TXGU&<}p_Kf8K*T2{8KS+n}Z7hZVx
z==)#(^3w+n9C-exKlMHTmoL3kC=}0}J$w9v4+uagWLsO?xsHyD7rTxhKmPXH2kyUr
z+p*&xR4WxT7TdOc+q>@``Nx0y@w-QkG&ME;{Dois=*Ry;fOd6tz4+3v-hc0X#<gTB
zkxHeqnT%5El~-Sl$K&H*3=>?>mm3urWzhg2q;!hqRHkmAuW#4RU6y72;q~9IUb|`2
zh7CXe#V_g-senk0azF4vp%sraMysU?Gc-bBlp(vWn@Xi95~47l&!6wU7>~sQ#|xC;
zgc+vcdM@QmHw@D>UDvg%wj#KE*;3E*UDu^R2@rxvMhTa~5nL${p^y;Zl+4fbsRA68
z*v4k>e81UT@02PuZPJtdBF;t1h<k+H)8HTJpmi}2iq)rND9{&1Q_elnsj41ON)QnV
zrJxAW?`c3kR4Efbyo8(@E9ym8L0{7-hAYGkWyz6c7YPETOy}qe-w&w9C{WNDh7vS}
zz8Gm@Ob4B-s)xQ{*#sC|^p=3rdGMx2vIqrq1GF@tRHBU4*YO3ZYEK!;Zf5I>%3G%r
z@ukua^rg+J?6M_U@IpdJI9S%#Eq2~KMH7ZB`?<gRRQ0#-ho^c70Lv!yy12j44$k${
zd{Uh1(R4;kO<&g}x`)Nl4iYGH`!breWZ5w|)0X7La79%;_ee)<&l(IRUyzi;HiiO)
z0)t~9cs3z=OXA`X(T$OrJzpjZFcLy3aK@XmT4Tn2=PXj-nYbiG0q|_nKhvw6K$aa|
zpZ1Pj&{s7ON?9fmbo7J9loJ@M7iGWkaY9INrdMCp;OsxGg_0&MmQ4^wSuQEMhy63X
z#22(K28O15feK~W0RR<BF;G^v7hD)5NLb1WdkV7b%90~`i>xlrvT+hh<pi>56E-iR
zrz@ffqmfYx7)7awP@pj+geXO%B9sz^gaQFV2%wPE;K~a@<3vcBiSq?1<p!dsKqx?D
zs0sw=9HXUFNWvKkNeBT71b`?c04OCv0t^TMK>`|M*+keqOc(_Ugb<Vn8pS|@&Z0-=
zk((~SSJPDlAW9Je1fsvm)ia(Dg@jUqNPwVFB$R{#=$v452V8k-^bH?*&j>LFGD@D2
zpj4o-E8mPfIOd-DGG!4RF;=B6qX(!IWfXv3LG+{ekLh$Iz!}jP!H5HoQD}f%^@eLP
z!mGapauvE)9-K#aW&kUt5{U!=1qw<kUDqAQ@qHHnCe^n!1YjP^X-MPoRxW;ZucD?K
zxm><b7+$+&W3^H)mx^cFPg_=O-oSM0!>g23*BE;FH8V#Urm5@t@X#P4I*tQCq!jb`
z-VOzCOtK<}AR#0S14`MvoAwSQS&=lRluabo=gVyj@I0t@Seduz&I3GOZh8KE9n;=C
zz_&fR=mqG96Jb0L@NJH6D2#PY`to`V1i7K1W4_#O3oNLLmw0reEZE-3Zf<HQKdK5}
zwQY93++OIS6PR0;B9PM0_Op~Vu7iADL|?B538RF<$i*$EJOWCU*tbcQT;OaX>;1fl
zo`9DUqN+ZT3Ur1FiI8CFU}!|DTV?v%p1EZyVc?y(NG%;FXxKc!*Y+JnKSU{GJ6enl
zOK8kMrDSi(IoKX{4$@SN5WEHUlK?7=KfKO8e3t55y1_*E%>(=qPC`}1(;Dm2&+fq>
z^jo`R!6q8lm*g`4;bE_}tMZG(U}__K9Bv|cc6?)f6BwG`-b+oL6rBn0y>>@O6d86o
z!i<o+XZsVQs_dHUo6^tiss8@B*V;u5P8cJI7zof=^6`6VOdtB;zY`$9*iFNjQB`uX
z-`N+O>GMx@gT_T)X`;q_lw=`d6g(1!p%6l;yC#r?f~tDP>P6|V?<qe2mfzOHGI20<
z;1o1QlP2iA^z(y3=TQFpPY{JfshdN)fS?=bEelyD?w{-iOTYd&yGle$DRTLG!!!fm
zPbO20^HO1WV4yz=8@!`g6~RFAe6sE@_LhEnK$TsVv{0%mRzO7v0MDnKKb{0ebKluF
z^!G2*q;(U-8>7ni;vZY<f7neB0RXQK=)L|L3`)wny3Fv<V7ji3GDZ;l`}?@2-Qf;@
z01Sidw?5&%ev%YiEg2_L5k5|wBdf3T-e}W4zdiG{hsw|IXW2N;F|U?NTe3iJUQphD
zKug65A%p?p8XzjAq!iZ!IEp>7?W$$Px-NDax>2c=DWzc$+_BIi63*msHT}#3L0{2+
zuY+ac%JU{P-wQ$#dhwTsGv9bv+urQHbB-oW63z8FSx02iN$*?do$U?FHceRskq9Ni
zfS?*TY>Qv5AkcETR4fc5A(T=?MJbB{z3*hD$PJ7&jpl~t!auyu(y_^)1*TU1<vVp>
z*%h?+qAv)8ITF<hSvqE{Z7RR`4viZ`;TTLO=WL*qYPzoLI_F%|^gA5gX!R8;^@aPZ
zubr6ef{F+lmqS(eNJsja2UXQ0bJI%{RMoR~w8)`~9IB3lDBSR&@!Ef--0?OIRd(Zh
z*P<_iwo4E*Cad`<MU$3upq-^-)~;2m;u1O+A{sPiZCl|U=>SWcWN6miZHrclCd~Ll
zYpZ`~rEz^~W=N&L)GNO`mVA5*HT7AKgrk7AZ)|Kvslu*d&?l?c+{FMhs>*KS(T!o(
zu<S3BsjGruaB-;^To_D#W~(Z@la!g$=lN2_AH2sqc4124(RVSxp=7y)xp}GmRvS$i
zL`_}Wk&32F`;RA$^-WrH9{q5(Odmkd4fPfEG#v|0UjkE~yb+Dg-iRZN5&!~Z#OZdc
zP2Z{)e`G_@(XYzR^d}zxhQfX4Z1S_)(G6ycZBdF|5ZklHKiPwFY8JFvYmghLqJyEp
zP^prmsve<m3x$1#lI4@e`X=YlnGt<sx_S$Z8}8BbEFCj9FGbs*m7yF;wlJ+N&3hko
z(%95mTWcF61tOIyJLZ;U)_p5z$|8iI?R&>NokM4UGcdI22*cX1zIN)9o5IdPRd#7A
zHlrpX0gc&jo=iNx#c%DJmDX15)V_`3#UWMo=ma}ACVaY<TXra*FM!j`H$QG{XbRdc
z`6n(0=LXOVtnDk}yVnHmy|Q3~somt@L4^?0lTU54Up)b}U`C*Gaj9smT?`rv+NZWg
zPD<_8dGfQ{?bknqYo#*~qEOlX>;p^zy^yA3`R_dubo5t#byN&jiI5}|a-ibAcY!EX
z_soN$dsuW0-_Qp#ySi`1O+2!JaOxfF1WT`h5ektoDtb$aPi}M%ovr0UL_t-&*q${s
zVR}b9#%b*w0StgbQcKH!{|Wn#C*60>u}mB|i@Y(QaT?P__i%8gukN{rM9;A39<K2)
zpHgEPO@018`;Aj5RjuN5fHOH*(H3WUecC_PIR%@^4Y{f!si${3Z=J?qyj-3)0=RZ0
zE0o#q?6VKF`>kCp6DLxQNy&gjDw?!Z$tgbnr@F5{q%F>(8{7nHO+i&XbNdQa^+azG
zwA$sxREj1{`^{6<_7yCjB%!QTYprd|F%V(b(BzF9*8@1RP%5Pe!O=*=8?%6-lm6<1
zqQC6DbB<<W<C}z}f|#y~cI7umGv9m!IJ??bK1Z_`D8zQIbdPj^QEMu|s46>tTTk+d
zdsVe2b`Kx~VtZD*htGncO>Zf=N<<@~ZWxSnUDFZKG%ZA>uFo(jVAV6XEw^^Gl>YSq
zO~tO~A5{p37WNcq!ieo$Y5(aIwRD`bbXOG6@y+#3+NuWU&66}{)VOLy(6#XVK>U&Q
zVMo6#I)u$&>(<D=$z@v0Lg$T>G;ZKTKtv)M5ys|48COkGQY8|}L^A1no?)7)bX}=f
zIDhV}X<656XPl)!cVG24@9~%gj1hVRxySM;|E+fOxjo58HhOQJrfCbMnu8RSz(6`8
z{?OX6wTJ5lP)2~&d<>~T1xGKWKYM@Shp(`tfl^NEry)?Pm`pyh!8>w+TjqqV1wx1-
zng&ReQrB*=;vczNN|{I`!yt&q6L?ufqf+Ao%&00<)6eb>&h>}g!!#Yk$taWph^3W(
z^={oW5BRNJ(hY#k{w75r$`!mh$1;h^%kNWD!`V>BC4rdk9_`RKHW~LW^^SMa__U0m
z6ja4Ewlot){nI@(X<e_is8CJl{rjlH-*xN@9p{u%!^1-aF$_b_m=t2XfdlCKY)MvI
zwXpD`HzBSo5l+%FKp2&Us&nLA;*0kU|M1^gHa_bc$~l!)H}SdcUR$^FL!vQ)vs0r|
z$e{Y#iTGn1onz;T6eCn)Y6F-mDPj+=v)^t*21<wtp9eyeQc_%%zI7s!1pq-1RI6pj
zu|3!IeGj9@uKf@ojG`w}Pi=SKIY)#f6fWmFyU_=LD5OctIe3QWlGc4KSoJR(u4nBd
z6wjyhmW3m;3URWno$`l(rHAJSRl!L<wpo>2!r92Lp5%uyRq<jE+(Q%wXZyg^M+w*U
zx^$A^qW}<X)1h=s3bb(*U)wh}G_ksv`~G>DhB+4%=)C&M2gxVyo%P(4jH;?1+qFvc
z6jgclo%2%Bgjs#%18c`h-k3pOOl=iL1Vc+avZ4C=hhXY6lr%H&q6n4Px5+u!4%34N
z37A^gIV1-wiBE2tkxOSXwn<AfH!pP~W6&J%b_U7~+=FM*fA#=6{uI1SjH;58e0)oI
zsUUg^my<%yF~9)m_~zzjFg5>V&-Cq%(MW&2)w;ih<x{mW4vb4!Y|q-dTseM?jT_e;
zQZ#P5M>>GB#G@Nklp%LA9qWZ!bKcy#tom9jO`0=z%+Tp|8A=)d_<Hy7*_kF^XP_#c
z*V>)@%vM$PW@i!uA;j9b$~}4>bUt_XQKe|YDF6CM{1Y3r#X0oC8!rt2=!faAKIr`E
zRFu}W2E%(g)7SQm^-I9i{I;H%x(yW4gy|k^*Ox8OTNa`h)YxYXfUa+>Squgb&h>+-
z&1onCqv!_JS3bym^I@XY4J>z@$>D11^Y<&q_l{qLcx^JC)9Fm#P{#MJ^NyYeS|f5K
zm2>z^@-tf}HO!~l+Can~TH_t>1PXH<Xrd6}hJUKZKhu}}kB`cdGa=pOOckoJ9WCb8
zWtCqar75#kdLx~xv_m)0S2eJ@n18YxVzuxHAwt~nj(5_eX>M7DwpTNZfuI*^OBb*O
zDZjPr=J+V3qUo6ZS}O{b{r01(U=xK`twcb8GdWx})+|nac~|NG93Y(5(5^Q{fGI^U
zjPG6Row!&_LoO2F%zpD!;!~SzImswRH;C_D>!0euP~Nm03{ugwRetdu;4Jt3CkUb{
zIpnf|0YfPU%CQ|Q>;7VI>1PL&?R_*?_6%2vDfxp@NWL%~|Jd5<t0$<*YX}DDT=W!;
zbxnlRpuM*?-eRFxE*^h$ll{sEGuJm}I)Y8jJ9dHR5~<JKM-<Az3c3M7!WX8~U)pJI
zUs3wm0abFq)TXxd-pC((y_EjrS)T3mpFKc0wf}gMCe7L^lqeNT$5Nl!I`DVDslmxe
zQER`lix`^yht`_6cA5%X#p4a>*aP>_go$nloPy5%_Fm`E8PFJLY|4!ULWC0i<ys_R
z5t3%&`szjH7Y@-_?ac+CadD{__7)TSHdbDImu6;WpbjO=C-t?9OaFRcsyh*9Q8`YU
zvgnk5ahN78nly<}vS_1_G;R`tv(ZR8j$$#UT%mlFD%+3kTqXKSBc9G#Q5;SCk0-5d
zD_DIBLoqYkq!<7A`tZV_D!YW;GTU_(0;f^fr!3g2;)0>kgh_C=61&iAI#~)Kgb-mE
zhGB3eMLZ&C+}yg%J<<WQ7mC9`RXy)SSL&&4Seb#{Ba|$kG&d}94z|<yEp~(!jvUqp
zjE?LGn9b&0b$+mvvaT+ZUr@iGeqqC+#wE?mQ|Yu)3Mf^!XYFXgK!~0K=xi2+Si*D<
zwrfjr#=0hS{OQ;|7*$m-@ySj8**;bFW;bN$mM?saIy#b3r(sc}5Md^hQ%VgE8pT4v
z)OD&WeL*^Szf+4@!1yf2Gy$MaytmrZU)WLjAHSvg6!-K1A)%tn7Fz3;lz#HIo=J?H
z?{ue@ksTfSq*6GH^iLSt)v9CLnx+?v1po+qpKvU{c7iWV5e6g-XSdNB4O=hr^^1*%
zS66@gKCPPy+hSCe?Ce)}I`5qes}4<=clZ&95sMCkNwH|gSgBMj6oyA9G^Lb9389dH
zMhT;H(O**0gi-z7vCMOOy!Xz_k~>-5p8{4rV{<cGkShPb_wIajr$Mp+fOF0{*ECJj
zG|Jdj>|oBii&5qH_Wsj#U)zK3q$9eJ5DKZ!-B*3>1Vm@B-Sy>0DI9KAfQdqyFukMa
zQL5yV+tpBY+@wZV3nMBEcIpfFi{4_`F>v!Epl(@!Z<QhyO`7GG-i_^CX>4v*rRyr+
zjH;4rZCz%pZ>s+OI89l%N?O=?fUnP}2_dBy4%I#P5MP)=#~&YKM$rv;KAC>z!SXNO
zxy$vU+X`SrFf=h#sl5DN?ynwab#Ya3fwQrGA>a&cA51Ox-N&l0d?1D@bGmKgmM>HA
zCIM$OZrmJ<drPWl*L`h|=qUsp{lFNZ001bIEIQiKeEz$S*>9clTf1q}y1STj4Z}xo
z4!{w{rY3reVdr4_^Y<Ao3&miCg$faAHlBKHOXBe@)!!ZW+b-RWOK;8Gg@|BiY(eVg
z1#ScSf+!W=y(ZM*zpj4l|F$1Ivq18mioO3dh7w}>-AJr%<Sx8@0KUw*l<gBBxooZ)
zy1wI)P|~<@_n!V<OMmdqWYd6%5w&6olMo!VON>%(>dHjGk(=gEvA$ga7E)3GAp}rL
z5Gkcf$|!JI*R(JU6(XgSGd7RtZ3Zv^ghDheT2v^OEYl2ypcHJ!*?s?xzJUQ<(?4u)
z+qrccB5^(M_{mmH)8^;#Z8n3Yl=Zp%s^u#d)i3mXe_?*Xzx?v&3$nRoP0f_j3*8qN
zH!QMDt9PLP&QE?kjTv$+$%>9#QA#n!l#24uV3d7fj!700LIM0BNG1|>>2weT<w_+M
zixrB+WIQ3HKt!O_&~+gsqAFF&mSL*PtMy$+vL<;rB_gihxS0T;Qo}<-Jzbr;VIUwd
zPWhY~i2xu7!xhVy2SLEO*45j)bL;lgXWP^1v}qW9g9G<0Z=sYbrIb>Xv88ND+u5`H
zu9DI~2vL;r%RVA?1AqYll?uaPQDalFSTIc!2?0t8K_G<qVs2Xx!u9H`gqbmuiA1GP
zYRc7bXj!Fex>WL|-@aU5m#xp|2|%e_uFvHYiA4M9GbloGSDjL?6i@&JHD0ETaYVFI
zDv?NrVPIJ?*L8!yzmk9e=NMlShGOyJM$57o=a+hVgp}D#CKj^@f$r|^Wy_Woi$zV>
z<FVLqVc0UQ&d!U3AR%`Xz$5dRjGJX=d_zaH^eUw&rK8!^000S@3lSYH`-mefIt&#e
zLF5_-0z!!Bzcft?f*>Lah!|byx;~}%&xHm*`<*uVbCX62V+;{F=cCU=={p$bboQm6
zQc4P;>w2`RXqq-k4hSI;gaAY-v@C0M9E@xloJ3cSegCX;h}rKn01!eDLdN1XTE^Ea
zKGInL5CmaMOG_daQ?r~5SSgi=#adcgf*>3nm}7GUUCzaa(aoc&`6hXa$hGgEO>}3y
z)1ood(9qD(&=3T{wZDj7|KaF?NWY8{i_TI*FINGV{{%{5to)f_FGMdH`~J}{Yu@8m
z;zDZF{u_2zRN1&@fC-%q(LVSv40T-(g1|J*AP5K{hGC4IOtqS|K@b>*5&esZj^h}H
zq3gO*N(d2M%tV3-Av8@3!_YKMA%ySymSqVcMt@FCehUEjz8`foBI>#xhM|zsGz>~<
z4J-0IPuF!J1Y=CoG~2eL^Cdaw(egT4(S-hBsZ=tNSlQB2we3|aR}K#k>$;xH<%-3k
z5F+aLu@r13=vtPwdi6b(N@dlmmBnI7DYbU(TFWx~`v){lYiMW~92}h90htlHxm>QP
zsmb^K6)Tnx4i3g*R#Q`x5Tc=>p})UhDJ7+3Gqz(#(sJd>7F{=z$wVfTDV0i$v1QH8
zwrzW!hlpVqGR9`uX-cW}>({xiw`55Z1*nutD_T}asTvy>IgaZ%4kAja00^a&Pb2|?
zAlSO~-gG)m09LJ9*?I9I=Y0FNt%HMunx+u|UDw}w>+Q>?uIY;!&+{I9;DKB|2LQ{L
zFYoUkC=?3&_C0#|@Zpx07AfSi<;&VmwY`7r*!0Fe0BqfQuVEORv$bp1oIc$ii^ZS*
z(ii&&1{~XNXlS57V)1xu>xTma1LN|PN!AjEVLqSVxM73s*nNEijg5=WoH>)pWS;)=
zm;UYFUe4$9)oL}FO!o9#I(_=|^e7&s)N!13>(;Gay*iajG0L*p?AiAAhxY6t0J&WD
zl~-O}vSdkDS63Jc$8j!PxIk}6zJ(B+^HeJ7dR{ypU$=hU&Ye3GiG*pI*=$xyIpaw@
zlBixvNhwVx6O6ImOP5Ne($1Yb7B6n}J#Xo<rLkB{N;%yM2oMqkp<x)&dwTo&e9sGo
z*s^)ku3b9~!|3ntpSh!AL@6Q=1VK8T7D8;=w6VLZYs;1`jf)zao12@Po4dNYMy2TV
zl_JClnsho%b^TIrFQV%1>R!Ei^`=c5bGh8(kAJFME;Gi)7tvAm!@~tB<oWaGV^*xU
zw=a{a%VaVK4<2rBZ$EqXTu*mT)wXBYBZx>T9U2-!#ItA5#bQ>WP^?y~1A~L-&!6w@
z?LB?Ey;Lf7c6Rpn4@^T!M@*tzt~id{)6-)Z`ry!z@B8P^UpR8)Xt`AG?&&_!`r)~A
z9ph5aDHX<8rBYQ&b$9o;uB((R6pCRe_8&N8+to#j7VY1^|J12A&-12TQKY*L4GnR|
z+uP4@jXRE0DwQ0^Idtf7cX!Y5@bJl#CvDpyh)0edwJhrfBr9S)p67AS)2UPthSh3y
zB%XzknykTNNtWk%nx=8iBTnS|KBY`b0RUsk-86)!lv3Aq*Y!-(7<HuxA)>X4F~(})
zOCCeAAPj?u$8pXhlTl=rsYyknBnuHu(+tBfszpjEgve&IzVAoE^?F-&BCa1bE;{-h
z0LF@!8X5{BpFvbn<j#uz{fJPRp`j<LWz;Yd=`zuaYxP^ksxA>h8Ea0cRYlPc83kx;
scSXGNie@;493AuJHZ1cV@;kize||2JEwtRt+5i9m07*qoM6N<$f=r^>ApigX

diff --git a/tests/test_charts.py b/tests/test_charts.py
index 5c1dc982..94fae03f 100644
--- a/tests/test_charts.py
+++ b/tests/test_charts.py
@@ -4,7 +4,7 @@
   3. one golden image per chart type, compared with a tolerance. Refresh after an intended restyle with:
 		 UPDATE_GOLDEN=1 python -m unittest tests.test_charts
 """
-import datetime, io, os, time, unittest
+import datetime, os, time, unittest
 import gevent
 from unittest import mock
 import numpy as np
@@ -35,27 +35,42 @@ def decode(buf):
 	return Image.open(buf).convert('RGB')
 
 
-def price_frame(days=120):
-	"""A made-up price history of `days` days, wavy and rising."""
-	dates = [datetime.date(2026, 1, 1) + datetime.timedelta(days=i) for i in range(days)]
-	closes = [100 + 12 * np.sin(i / 9) + i * 0.15 for i in range(days)]
-	return pd.DataFrame({'Date': dates, 'Close': closes})
+def price_rows(days=120):
+	"""A made-up price history of `days` days, wavy and rising, as a line chart's (date, close) rows."""
+	return tuple((str(datetime.date(2026, 1, 1) + datetime.timedelta(days=i)), 100 + 12 * np.sin(i / 9) + i * 0.15)
+		for i in range(days))
 
 
-def compare_series():
-	"""Three made-up price series over the same 120 days, for the compare chart."""
-	index = pd.date_range('2026-01-01', periods=120)
-	return [(name, pd.Series(100 + np.cumsum(np.sin(np.arange(120) / k) + drift), index=index))
-			for name, k, drift in (('AAA', 7, 0.2), ('BBB', 11, -0.1), ('CCC', 5, 0.05))]
+def line(rows, title='Apple Inc (AAPL) 120 days', subtitle='164.12 USD   ▲ 6.96%', change=6.96):
+	"""A line chart of these rows."""
+	return render.Chart('line', title, subtitle, rows, change=change)
+
+
+def compare_rows():
+	"""Three made-up price series over the same 120 days, as a compare chart's rows."""
+	days = [str(day.date()) for day in pd.date_range('2026-01-01', periods=120)]
+	return tuple((name, tuple(zip(days, 100 + np.cumsum(np.sin(np.arange(120) / k) + drift))))
+		for name, k, drift in (('AAA', 7, 0.2), ('BBB', 11, -0.1), ('CCC', 5, 0.05)))
+
+
+def bars(rows, title='T', subtitle='', **kwargs):
+	"""A bar chart of these rows, drawn."""
+	return render.draw_bars(render.Chart('bars', title, subtitle, tuple(rows), **kwargs))
+
+
+def shorts(rows, threshold=15):
+	"""The shorts pictogram of these rows, drawn."""
+	return render.draw_shorts(render.Chart('shorts', rows=tuple(rows), threshold=threshold))
 
 
 CHARTS = {
-	'hbar_mixed': lambda: render.hbar_chart(MIXED, 'Price movement 5.0%, 5 Sep - 5 Oct 2026', 'subtitle here'),
-	'price_line': lambda: render.graph(price_frame(), 'Apple Inc (AAPL)', 'USD'),
-	'compare_lines': lambda: render.compare_graph(compare_series(), 'Compare 120 days', 'subtitle here'),
-	'shorts': lambda: render.shorts_chart([('GME', 47.3), ('AMC', 32.1), ('TSLA', 21.6), ('AAPL', 15.2)], 15),
-	'pie': lambda: render.pie_chart([('Software', 41.0), ('Banks', 23.5), ('Mining', 12.0), ('Other', 8.0)],
-		'Rob by industry', '4 industries'),
+	'hbar_mixed': lambda: bars(MIXED, 'Price movement 5.0%, 5 Sep - 5 Oct 2026', 'subtitle here'),
+	'price_line': lambda: render.draw_line(line(price_rows())),
+	'compare_lines': lambda: render.draw_compare(render.Chart('compare', 'Compare 120 days', 'subtitle here',
+		compare_rows())),
+	'shorts': lambda: shorts([('GME', 47.3), ('AMC', 32.1), ('TSLA', 21.6), ('AAPL', 15.2)]),
+	'pie': lambda: render.draw_pie(render.Chart('pie', 'Rob by industry', '4 industries',
+		(('Software', 41.0), ('Banks', 23.5), ('Mining', 12.0), ('Other', 8.0)))),
 }
 # inputs that have broken chart code before or are likely to: sizes, signs, long names, zeros, a single row
 EDGE_ROWS = {
@@ -100,21 +115,23 @@ def test_edge_case_rows(self):
 		"""Rows that broke charts before (one, all zero, forty, long labels, huge or tiny values) draw sane images."""
 		for name, rows in EDGE_ROWS.items():
 			with self.subTest(rows=name):
-				self.check_image(render.hbar_chart(rows, 'Title', 'sub'), name)
+				self.check_image(bars(rows, 'Title', 'sub'), name)
 				self.check_image(uncached(render.Chart('bars', 'Title', rows=tuple(rows))), name + '/drawn')
 
 	def test_zoomed_axis(self):
 		"""A bar chart with a narrow x range (ratings, 1 to 5) draws sane images, wide labels and many rows included."""
-		self.check_image(render.hbar_chart([('AAA', 1.2), ('BBB', 3.4)], 'Ratings', xlim=(0.9, 4.0)), 'zoomed')
-		self.check_image(render.hbar_chart([('WWWWWWWW', 1.2), ('MMM', 3.4)], 'Ratings', xlim=(0.9, 4.0)), 'zoomed wide labels')
-		self.check_image(render.hbar_chart([(f'T{i}', 1 + i / 10) for i in range(15)], 'Ratings', xlim=(0.9, 3.0)), 'zoomed 15 rows')
+		self.check_image(bars([('AAA', 1.2), ('BBB', 3.4)], 'Ratings', xlim=(0.9, 4.0)), 'zoomed')
+		self.check_image(bars([('WWWWWWWW', 1.2), ('MMM', 3.4)], 'Ratings', xlim=(0.9, 4.0)), 'zoomed wide labels')
+		self.check_image(bars([(f'T{i}', 1 + i / 10) for i in range(15)], 'Ratings', xlim=(0.9, 3.0)), 'zoomed 15 rows')
 
 	def test_shorts_chart_sizes_and_verdicts(self):
 		"""The shorts pictogram draws 1 to 45 rows and a long label; each tier of short interest gets its verdict."""
 		for n in (1, 5, 45):
 			rows = [(f'T{i}', 60.0 - i) for i in range(n)]
-			self.check_image(render.shorts_chart(rows, 15), f'shorts {n} rows', bg=SHORTS_BG, edge_clean=False)
-		self.check_image(render.shorts_chart([('A VERY LONG LABEL', 15.0), ('B', 0.4)], 15), 'shorts long label and tiny value', bg=SHORTS_BG, edge_clean=False)
+			self.check_image(uncached(render.Chart('shorts', rows=tuple(rows), threshold=15)), f'shorts {n} rows',
+				bg=SHORTS_BG, edge_clean=False)
+		self.check_image(shorts([('A VERY LONG LABEL', 15.0), ('B', 0.4)]), 'shorts long label and tiny value',
+			bg=SHORTS_BG, edge_clean=False)
 		tiers = [(0, 'board shorts'), (4.9, 'board shorts'), (5, 'cargo shorts'), (10, 'boxers'), (15, 'gym shorts'), (20, 'cutoffs'),
 				 (30, 'short shorts'), (40, 'daisy dukes'), (50, 'hot pants'), (60, 'speedos'), (70, 'budgie smugglers'),
 				 (80, 'just a belt'), (90, 'never nudes'), (200, 'never nudes')]
@@ -127,10 +144,15 @@ def test_bar_graph_has_top_and_bottom(self):
 
 	def test_flat_and_tiny_price_series(self):
 		"""A flat price line and one of two points draw sane images."""
-		flat = pd.DataFrame({'Date': [datetime.date(2026, 1, 1) + datetime.timedelta(days=i) for i in range(10)], 'Close': [5.0] * 10})
-		self.check_image(render.graph(flat, 'Flat', 'USD'), 'flat price')
-		two = price_frame(2)
-		self.check_image(render.graph(two, 'Two points', 'USD'), 'two points')
+		flat = tuple((str(datetime.date(2026, 1, 1) + datetime.timedelta(days=i)), 5.0) for i in range(10))
+		self.check_image(render.draw_line(line(flat, 'Flat', '5.00 USD   ■ 0%', 0)), 'flat price')
+		self.check_image(render.draw_line(line(price_rows(2), 'Two points')), 'two points')
+
+	def test_a_long_title_is_shrunk_to_fit(self):
+		"""A title longer than the chart is wide is shrunk to fit, on every chart that has one: it ran off the edge."""
+		title = 'Top 15 tracked stocks by negative forward pe ratio (7+ analysts), and a few more words besides'
+		self.check_image(bars(MIXED, title), 'long title')
+		self.check_image(render.draw_line(line(price_rows(), title)), 'long line title')
 
 
 class RenderPool(unittest.TestCase):
@@ -143,8 +165,9 @@ def setUp(self):
 
 	def test_charts_asked_for_at_once_all_finish(self):
 		"""Six charts asked for at once all finish: one render thread, as several deadlock matplotlib under gevent."""
-		jobs = [gevent.spawn(render.pie_chart, [('a', 1.0), ('b', 2.0)], f't{i}') for i in range(4)]
-		jobs += [gevent.spawn(render.hbar_chart, MIXED, 'T')]
+		pie = lambda title: render.Chart('pie', title, rows=(('a', 1.0), ('b', 2.0)))
+		jobs = [gevent.spawn(render.draw_pie, pie(f't{i}')) for i in range(4)]
+		jobs += [gevent.spawn(bars, MIXED)]
 		jobs += [gevent.spawn(uncached, render.Chart('bars', 'T', rows=tuple(MIXED)))]
 		gevent.joinall(jobs, timeout=120)
 		self.assertTrue(all(job.successful() for job in jobs), 'a chart did not finish (several native threads deadlock matplotlib)')
@@ -162,7 +185,7 @@ def ticker():
 				gaps.append(now - last)
 				last = now
 		tick = gevent.spawn(ticker)
-		render.shorts_chart([(f'T{i}', 60.0 - 2 * i) for i in range(24)], 15) # about a second of drawing
+		shorts([(f'T{i}', 60.0 - 2 * i) for i in range(24)]) # about a second of drawing
 		done.append(True)
 		tick.join()
 		self.assertLess(max(gaps), 0.5) # drawn in the hub, the ticker stopped for the whole chart
@@ -170,7 +193,7 @@ def ticker():
 
 class Contracts(unittest.TestCase):
 	def capture_bars(self, rows, **kwargs):
-		"""The y positions, widths and colours hbar_chart hands to matplotlib for these rows."""
+		"""The y positions, widths and colours draw_bars hands to matplotlib for these rows."""
 		calls = []
 		real = Axes.barh
 		def spy(ax, y, width, *a, **k):
@@ -178,7 +201,7 @@ def spy(ax, y, width, *a, **k):
 			calls.append({'y': list(y), 'width': list(width), 'color': list(k.get('color', []))})
 			return real(ax, y, width, *a, **k)
 		with mock.patch.object(Axes, 'barh', spy):
-			render.hbar_chart(rows, 'T', **kwargs)
+			bars(rows, **kwargs)
 		return calls[0]
 
 	def capture_text(self, make):
@@ -195,25 +218,25 @@ def spy(fig, x, y, s, *a, **k):
 
 	def test_bars_are_blue_when_positive_and_red_when_negative(self):
 		"""Gains (and zero) are blue, losses red."""
-		bars = self.capture_bars([('A', 3), ('B', 0), ('C', -2)])
-		self.assertEqual(bars['color'], ['#3987e5', '#3987e5', '#fb7185'])
+		drawn = self.capture_bars([('A', 3), ('B', 0), ('C', -2)])
+		self.assertEqual(drawn['color'], ['#3987e5', '#3987e5', '#fb7185'])
 
 	def test_first_row_is_on_top_and_spacers_are_skipped(self):
 		"""The first row is drawn at the top (the y axis is inverted), and a blank spacer row gets no bar."""
-		bars = self.capture_bars(MIXED)
-		self.assertEqual(bars['width'], [42, 17, 3, -5, -30])
-		self.assertEqual(bars['y'], [0, 1, 2, 4, 5]) # row 3 is the blank spacer
+		drawn = self.capture_bars(MIXED)
+		self.assertEqual(drawn['width'], [42, 17, 3, -5, -30])
+		self.assertEqual(drawn['y'], [0, 1, 2, 4, 5]) # row 3 is the blank spacer
 		calls = []
 		real = Axes.set_ylim
 		with mock.patch.object(Axes, 'set_ylim', lambda ax, *a, **k: (calls.append(a), real(ax, *a, **k))[1]):
-			render.hbar_chart(MIXED, 'T')
+			bars(MIXED)
 		self.assertTrue(any(len(a) == 2 and a[0] > a[1] for a in calls), 'y axis must be inverted so row 0 is drawn first (top)')
 
 	def test_pixels_follow_the_sign_of_the_data(self):
 		"""The drawn image is blue for gains and red for losses, in proportion to the rows."""
 		def count(rows):
 			"""(blue pixels, red pixels) in the bar chart of rows."""
-			pixels = np.asarray(decode(render.hbar_chart(rows, 'T'))).reshape(-1, 3)
+			pixels = np.asarray(decode(bars(rows))).reshape(-1, 3)
 			return (pixels == BLUE).all(axis=1).sum(), (pixels == RED).all(axis=1).sum()
 		blue, red = count([('A', 5), ('B', 3)])
 		self.assertGreater(blue, 1000); self.assertEqual(red, 0)
@@ -224,7 +247,7 @@ def count(rows):
 
 	def test_session_tints_the_background(self):
 		"""Mid-session tints the background faint green, pre-market faint yellow; close and the default stay plain."""
-		corner = lambda session: tuple(np.asarray(decode(render.hbar_chart(MIXED, 'T', session=session)))[2, 2][:3])
+		corner = lambda session: tuple(np.asarray(decode(bars(MIXED, session=session)))[2, 2][:3])
 		plain, close, mid, pre = corner(None), corner('close'), corner('midsession'), corner('premarket')
 		self.assertEqual(plain, (0x14, 0x18, 0x1f))
 		self.assertEqual(close, plain) # at close stays untinted
@@ -234,33 +257,36 @@ def test_session_tints_the_background(self):
 
 	def test_longer_bar_is_drawn_longer(self):
 		"""A bar four times as long is drawn over twice as long."""
-		pixels = np.asarray(decode(render.hbar_chart([('A', 40), ('B', 10)], 'T')))
+		pixels = np.asarray(decode(bars([('A', 40), ('B', 10)])))
 		blue_rows = np.where((pixels == BLUE).all(axis=2).any(axis=1))[0]
 		top_band, bottom_band = pixels[blue_rows[0]], pixels[blue_rows[-1]]
 		self.assertGreater((top_band == BLUE).all(axis=1).sum(), 2 * (bottom_band == BLUE).all(axis=1).sum())
 
 	def test_titles_are_drawn(self):
-		"""The title and subtitle appear on the bar chart, and the title and period change on the price line."""
-		texts = self.capture_text(lambda: render.hbar_chart(MIXED, 'My Title', 'My subtitle'))
+		"""The title and subtitle appear on the bar chart, and on the price line, which works out no change itself."""
+		texts = self.capture_text(lambda: bars(MIXED, 'My Title', 'My subtitle'))
 		self.assertIn('My Title', texts)
 		self.assertIn('My subtitle', texts)
-		texts = self.capture_text(lambda: render.graph(price_frame(), 'Apple Inc (AAPL)', 'USD'))
-		self.assertIn('Apple Inc (AAPL)', texts)
-		self.assertTrue(any('over period' in t for t in texts))
-
-	def test_price_line_colour_follows_direction(self):
-		"""The price line's change is green up, red down, grey flat."""
-		def line_colour(closes):
-			"""The colour of the 'over period' text for a price line of these closes."""
-			frame = pd.DataFrame({'Date': [datetime.date(2026, 1, 1) + datetime.timedelta(days=i) for i in range(len(closes))], 'Close': closes})
+		texts = self.capture_text(lambda: render.draw_line(line(price_rows())))
+		self.assertEqual(texts, ['Apple Inc (AAPL) 120 days', '164.12 USD   ▲ 6.96%'])
+
+	def test_price_line_colour_follows_the_change_it_shows(self):
+		"""The price line and its subtitle are green for the change it shows rising, red falling, grey for none: the
+		change the caption gives, not one worked out from the first and last closes."""
+		def line_colour(change):
+			"""The colour of the subtitle of a falling price line that shows this change."""
 			texts = []
 			real = Figure.text
-			with mock.patch.object(Figure, 'text', lambda fig, x, y, s, *a, **k: (texts.append((s, k.get('color'))), real(fig, x, y, s, *a, **k))[1]):
-				render.graph(frame, 'T', 'USD')
-			return next(color for s, color in texts if 'over period' in s)
-		self.assertEqual(line_colour([1, 2, 3]), '#34d399') # up: green
-		self.assertEqual(line_colour([3, 2, 1]), '#fb7185') # down: red
-		self.assertEqual(line_colour([2, 3, 2]), '#9aa4b2') # flat
+			def spy(fig, x, y, s, *a, **k):
+				"""Record a text and its colour, then draw it."""
+				texts.append((s, k.get('color')))
+				return real(fig, x, y, s, *a, **k)
+			with mock.patch.object(Figure, 'text', spy):
+				render.draw_line(line(price_rows()[::-1], subtitle='sub', change=change))
+			return dict(texts)['sub']
+		self.assertEqual(line_colour(6.96), '#34d399') # up: green
+		self.assertEqual(line_colour(-3), '#fb7185') # down: red
+		self.assertEqual(line_colour(0), '#9aa4b2') # flat
 
 
 class ReportChartInputs(FinbotCase):
@@ -275,17 +301,18 @@ def run_report(self, call):
 		return captured[0]
 
 	def test_over_forty_bars_keep_both_ends_with_a_gap(self):
-		"""A bar chart of over 40 rows keeps the first 20 and the last 20, a gap between them."""
-		captured = []
-		self.patch(render, 'hbar_chart', lambda rows, *a, **k: captured.append(rows) or io.BytesIO(b'png'))
-		render.draw(render.Chart('bars', 'T', rows=tuple((f'T{i}', i) for i in range(50))))
-		rows = captured[0]
+		"""A bar chart of over 40 rows keeps the first 20 and the last 20, a gap between them; the shorts pictogram,
+		over 24, the first and last 12, with the same gap (it used to drop the middle without a mark)."""
+		rows = render.trimmed(render.Chart('bars', 'T', rows=tuple((f'T{i}', i) for i in range(50)))).rows
 		self.assertEqual((len(rows), rows[20], rows[0][0], rows[-1][0]), (41, ('', None), 'T0', 'T49'))
+		rows = render.trimmed(render.Chart('shorts', rows=tuple((f'T{i}', 60 - i) for i in range(30)))).rows
+		self.assertEqual((len(rows), rows[12], rows[11][0], rows[13][0]), (25, ('', None), 'T11', 'T18'))
+		self.assertEqual(render.trimmed(render.Chart('pie', rows=(('a', 1.0),) * 50)).rows, (('a', 1.0),) * 50)
 
 	def test_a_chart_is_drawn_once(self):
 		"""The same spec, for another chat or a later run, is read back from the cache, not drawn again."""
-		calls, real = [], render.pie_chart
-		self.patch(render, 'pie_chart', lambda *a: calls.append(a) or real(*a))
+		calls, real = [], render.RENDERERS['pie']
+		self.patch(render, 'RENDERERS', {**render.RENDERERS, 'pie': lambda chart: calls.append(chart) or real(chart)})
 		chart = render.Chart('pie', 'T', rows=(('a', 1.0), ('b', 2.0)))
 		first, second = render.draw(chart).read(), render.draw(chart).read()
 		self.assertEqual((len(calls), first[:4]), (1, b'\x89PNG'))
diff --git a/tests/test_providers.py b/tests/test_providers.py
index c72e8624..399c9511 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -750,8 +750,7 @@ def test_slack_without_a_bot_token_gets_text_and_no_chart_is_drawn(self):
 		def refuse(*a, **k):
 			"""Fail: no chart should be drawn."""
 			raise AssertionError('a chart was drawn for Slack')
-		for name in ('draw', 'hbar_chart', 'pie_chart', 'shorts_chart', 'graph', 'compare_graph'):
-			self.patch(render, name, refuse)
+		self.patch(render, 'draw', refuse) # every chart is drawn there
 		self.env(webhooks={**config.env.webhooks, 'slack': 'https://hooks.slack.com/services/x'})
 		for cmd in ('.beta', '.pie Rob', '.buy', '.sell', '.peg', '.price', '.price top', '.price T1 7d', '.performance 7d', '.shorts', '.history T1', '.compare T1 T2'):
 			with self.subTest(cmd=cmd):
diff --git a/tests/test_reports.py b/tests/test_reports.py
index c418ac3c..11be7383 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -616,7 +616,7 @@ def test_folds_into_other_and_renders(self):
 		payload, chart = holdings.prepare_pie('Rob', 'industry', TELEGRAM)
 		self.assertEqual(len(chart.rows), 8)
 		self.assertEqual(chart.rows[-1][0], 'Other')
-		self.assertTrue(render.pie_chart(chart.rows, chart.title, chart.subtitle).read().startswith(b'\x89PNG'))
+		self.assertTrue(render.draw_pie(chart).read().startswith(b'\x89PNG'))
 
 	def test_unknown_or_missing_portfolio_lists_known(self):
 		""".pie with no portfolio, or an unknown one, lists the portfolios there are."""

From 3f9e55fe035c6ee72516404ba2ad2bcb42b124ff Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 09:32:37 +0000
Subject: [PATCH 60/81] Yahoo: an outage is an error, not an unknown symbol;
 history without pandas, dated where it trades

- One way to read Yahoo's answers, for quotes, details and charts: the first host's 200 is the answer, a 404 Not
  Found is no such symbol, anything else an error. An outage used to read as an unknown symbol for a detail ("I
  could not find TSLA", after trying TSLA.AX too). The chart now falls back to query1 as the others do, and a crumb
  Yahoo refuses (401) is renewed once.
- The cookie and its crumb live in one cache file, renewed together under a plain lock; no cookie or no crumb is an
  error, where a fallback pair from 2024 only turned it into a refusal later. The cache flush needs no lock.
- price_history works on the bars without pandas: each bar dated in its exchange's time zone (an ASX bar carried
  the day before's date all summer, an NZX bar always, so N-day changes slipped a session), the price and today's
  change from the quote where the caller has one (the chart is cached up to 12 hours). It takes 3 ms for every
  period on a 10-year chart (15 ms before), 2 ms for one (5 ms); price and performance jobs no longer import pandas.
- Record highs and lows come from the whole history, and records seeded from 10 years are seeded again.
- Fields that never filled or nothing read are gone (dividend_date, 52-week change, the details' 52-week range,
  price to book, quarterly earnings growth, profit margin, analysts per estimate), and so is the profile's net
  debt/equity line, which Yahoo no longer sends the figures for (D30), with the balance-sheet module.
- Log lines are whole lines (fragments without a newline landed in front of other threads' lines), and the ones
  every ETF set off are gone.
- An unreachable host costs 5 s to connect, not http_timeout (18 s): net sets one default for every request.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 bot.py                   |   6 +-
 lib/charts.py            |   2 +-
 lib/net.py               |  16 +-
 lib/reports/history.py   |  13 +-
 lib/reports/milestone.py |   8 +-
 lib/reports/price.py     |   2 +-
 lib/reports/profile.py   |   9 +-
 lib/sharesight.py        |   2 +-
 lib/shortman.py          |   3 +-
 lib/telegram.py          |   2 +-
 lib/transports.py        |  11 +-
 lib/yahoo.py             | 370 +++++++++++++++++----------------------
 tests/common.py          |   5 +-
 tests/test_providers.py  |  90 ++++++++--
 tests/test_reports.py    |  11 +-
 15 files changed, 278 insertions(+), 272 deletions(-)

diff --git a/bot.py b/bot.py
index c1cdf0d2..09b210d1 100755
--- a/bot.py
+++ b/bot.py
@@ -19,7 +19,6 @@
 from lib import net
 from lib import telegram
 from lib import delivery
-from lib import yahoo
 from lib import render
 from lib.reports import cleanup
 
@@ -173,10 +172,9 @@ def reload_all(*args):
 
 def flush_cache(*args):
 	"""SIGUSR1 (`systemctl kill -s USR1 finbot`): delete the cache and keep the state (watchlist, seen trades, ratings, schedule).
-	Yahoo's cookie and crumb belong together, so they go while nothing is renewing them; the Sharesight caches refill at once."""
+	The Sharesight caches refill at once."""
 	try:
-		with yahoo._auth:
-			removed = cleanup.run(days=0)
+		removed = cleanup.run(days=0)
 		sharesight.wake()
 		print(f"cache: flushed {removed} file{'' if removed == 1 else 's'} from {config.env.cache_dir}", file=sys.stderr)
 	except Exception as e:
diff --git a/lib/charts.py b/lib/charts.py
index 3524bef8..fd54443d 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -16,7 +16,7 @@ def price_chart(service, period, days, ticker):
 	market_data = yahoo.fetch([ticker])
 	if ticker not in market_data:
 		raise answer.Answer(f"no data found for {ticker}")
-	changes, chart = history.price_line(ticker, days)
+	changes, chart = history.price_line(ticker, days, market_data[ticker])
 	percent = float(changes.get(days, changes.get('Max', next(iter(changes.values()), 0))))
 	link = links.finance_link(ticker, market_data[ticker]['profile_exchange'], service, days=days)
 	title = market_data[ticker]['profile_title']
diff --git a/lib/net.py b/lib/net.py
index 2ae24541..1ef6bb02 100644
--- a/lib/net.py
+++ b/lib/net.py
@@ -7,16 +7,24 @@
 import re
 import requests
 
+from lib import config
+
 session = requests.Session()
 session.cookies.set_policy(http.cookiejar.DefaultCookiePolicy(allowed_domains=[])) # no domain may set or receive a cookie
+CONNECT_TIMEOUT = 5 # seconds to wait for a host to answer the connection: one that is down costs this, not http_timeout
 
 def get(url, **kwargs):
-	"""requests.get through the shared session, which reuses connections."""
-	return session.get(url, **kwargs)
+	"""requests.get through the shared session, which reuses connections; timed out as timeouts() says."""
+	return session.get(url, **timeouts(kwargs))
 
 def post(url, **kwargs):
-	"""requests.post through the shared session, which reuses connections."""
-	return session.post(url, **kwargs)
+	"""requests.post through the shared session, which reuses connections; timed out as timeouts() says."""
+	return session.post(url, **timeouts(kwargs))
+
+def timeouts(kwargs):
+	"""The request's arguments with a timeout, unless it has its own: CONNECT_TIMEOUT to connect, then http_timeout for
+	the answer."""
+	return {'timeout': (CONNECT_TIMEOUT, config.prefs.http_timeout), **kwargs}
 
 def redact(text):
 	"""text (or an exception) without the secrets that request URLs, headers and error messages carry: Telegram bot tokens,
diff --git a/lib/reports/history.py b/lib/reports/history.py
index 3dc89f06..9b972f23 100644
--- a/lib/reports/history.py
+++ b/lib/reports/history.py
@@ -50,7 +50,8 @@ def history_payload(ticker, service, days=None):
 	chart of those days, else of the whole history (None when it has too few prices). The name and exchange come from
 	the chart."""
 	highlight = buttons.PERIODS[buttons.period_for_days(days, 'x')][0]
-	table, line = yahoo.price_history(ticker)
+	quote = yahoo.fetch([ticker]).get(ticker) # today's price and change
+	table, line = yahoo.price_history(ticker, quote=quote)
 	chart = line_chart(ticker, table, line)
 	ticker_link = links.finance_link(ticker, line.exchange, service, days=1825, brief=False)
 	lines = [service.bold(f"{line.name} ({ticker_link}) performance history")]
@@ -62,7 +63,7 @@ def history_payload(ticker, service, days=None):
 				label = '● ' + label
 			lines.append(f"{formats.get_emoji(percent)} {label} {formats.signed_percent(percent)}")
 	if days:
-		_, chart = price_line(ticker, days)
+		_, chart = price_line(ticker, days, quote)
 	return lines, chart
 
 def line_chart(ticker, changes, line, days=None):
@@ -81,8 +82,8 @@ def line_chart(ticker, changes, line, days=None):
 	subtitle = f"{price}   {arrow} {abs(change):,g}%"
 	return render.Chart('line', f"{line.name} ({ticker}) {span}", subtitle, line.closes, change=float(change))
 
-def price_line(ticker, days=None):
-	"""(its changes, its line chart or None) of a ticker over `days`, else over its whole history: price_history and
-	line_chart."""
-	changes, line = yahoo.price_history(ticker, days)
+def price_line(ticker, days=None, quote=None):
+	"""(its changes, its line chart or None) of a ticker over `days`, else over its whole history, to the price of its
+	quote (else the chart's): price_history and line_chart."""
+	changes, line = yahoo.price_history(ticker, days, quote=quote)
 	return changes, line_chart(ticker, changes, line, days)
diff --git a/lib/reports/milestone.py b/lib/reports/milestone.py
index d2e9aaf9..d9bffd17 100644
--- a/lib/reports/milestone.py
+++ b/lib/reports/milestone.py
@@ -10,7 +10,7 @@
 # - need to handle stock splits: flush cache if we see a split in the data
 # https://finance.yahoo.com/quote/TSLA/history?period1=0&period2=1707609600&filter=split
 
-STATE_FILE = 'finbot_milestone.json' # {ticker: {'record_high': ..., 'record_low': ...}}
+STATE_FILE = 'finbot_milestone.json' # {ticker: {'all_time_high': ..., 'all_time_low': ...}}
 
 def run(dest=None):
 	"""Post the stocks followed that set an all-time high or low to the default chats (or dest), and record the new
@@ -32,8 +32,8 @@ def gather():
 	highs, lows, changed = [], [], {}
 	for ticker in tickers:
 		try:
-			oldhigh, oldlow = records[ticker]['record_high'], records[ticker]['record_low']
-		except (KeyError, TypeError): # first sighting, or a 52-week record of old, under 'high' and 'low'
+			oldhigh, oldlow = records[ticker]['all_time_high'], records[ticker]['all_time_low']
+		except (KeyError, TypeError): # first sighting, or a record of old: 52 weeks (high, low), 10 years (record_high)
 			try:
 				oldhigh, oldlow = yahoo.historic_high(ticker) # its history so far
 			except (RuntimeError, KeyError, IndexError, ValueError):
@@ -44,7 +44,7 @@ def gather():
 		except (KeyError, ValueError):
 			continue # likewise: a ticker is recorded only with both of its new values
 		# all-time: a 52-week high below the record is no record
-		changed[ticker] = {'record_high': max(oldhigh, newhigh), 'record_low': min(oldlow, newlow)}
+		changed[ticker] = {'all_time_high': max(oldhigh, newhigh), 'all_time_low': min(oldlow, newlow)}
 		if newhigh > oldhigh:
 			highs.append(ticker)
 		if newlow < oldlow:
diff --git a/lib/reports/price.py b/lib/reports/price.py
index af3d9f7c..affc1189 100644
--- a/lib/reports/price.py
+++ b/lib/reports/price.py
@@ -44,7 +44,7 @@ def gather(specific_stock=None, days=None):
 		yahoo.prefetch_history(missing)
 	for ticker in missing:
 		try:
-			changes, chart = history.price_line(ticker, days)
+			changes, chart = history.price_line(ticker, days, market_data[ticker])
 		except RuntimeError as e: # no history: the one stock asked about says so, a list leaves it out
 			if specific_stock:
 				raise
diff --git a/lib/reports/profile.py b/lib/reports/profile.py
index e1d9e103..a7cda1a5 100644
--- a/lib/reports/profile.py
+++ b/lib/reports/profile.py
@@ -98,17 +98,12 @@ def company(service, data, ticker, exchange):
 	return lines
 
 def figures(service, data):
-	"""Its size, debt, dates, dividend, cashflow and whether it makes a profit (always said, if only as unknown)."""
+	"""Its size, dates, dividend, cashflow and whether it makes a profit (always said, if only as unknown)."""
 	now = datetime.datetime.now()
 	lines = []
 	if 'currency' in data and 'market_cap' in data:
 		cap = f"{yahoo.market_cap_currency(data)} {formats.human_units(data['market_cap'])}" # GBP, not pence
 		lines.append(service.bold("Mkt cap:") + f" {cap}")
-	if all(key in data for key in ('shareholder_equity', 'total_debt', 'total_cash', 'profile_industry')) and \
-			data['shareholder_equity'] and 'Bank' not in data['profile_industry']:
-		ratio = round((data['total_debt'] - data['total_cash']) / data['shareholder_equity'] * 100)
-		if ratio > 0:
-			lines.append(service.bold("Net debt/equity ratio:") + f" {ratio}%" + ('⚠️ ' if ratio > 40 else ''))
 	if 'earnings_date' in data:
 		earnings_date = datetime.datetime.fromtimestamp(data['earnings_date'])
 		if earnings_date > now:
@@ -193,7 +188,7 @@ def changes(service, data, ticker, state):
 	"""The price's change over 5 years, a year, a month and a day, and out of session the pre- or post-market move."""
 	lines = []
 	try:
-		history, _ = yahoo.price_history(ticker) # the percentages only: the profile has no chart
+		history, _ = yahoo.price_history(ticker, quote=data) # the percentages only: the profile has no chart
 	except RuntimeError as e: # no chart data: the profile goes out without these lines
 		print(ticker, e, file=sys.stderr)
 		history = {}
diff --git a/lib/sharesight.py b/lib/sharesight.py
index 985dcb83..6a287dc0 100644
--- a/lib/sharesight.py
+++ b/lib/sharesight.py
@@ -27,7 +27,7 @@ def _post(url, **kwargs):
 def _request(send, url, what, **kwargs):
 	"""Send a request to Sharesight and return the JSON answer. A request that fails raises RuntimeError naming `what`."""
 	try:
-		r = send(url, timeout=config.prefs.http_timeout, **kwargs)
+		r = send(url, **kwargs)
 	except Exception as e:
 		print("Error", net.redact(e), url, file=sys.stderr)
 		raise RuntimeError(f"Sharesight {what} request failed: {e}")
diff --git a/lib/shortman.py b/lib/shortman.py
index 1799ca8e..407ba905 100644
--- a/lib/shortman.py
+++ b/lib/shortman.py
@@ -1,12 +1,11 @@
 import sys
-from lib import config
 from lib import net
 
 def fetch(market_data):
 	"""market_data with each ASX ticker's short interest added from Short Man's daily CSV (as it was if unreachable)."""
 	url = 'https://www.shortman.com.au/downloadeddata/latest.csv'
 	try:
-		r = net.get(url, timeout=config.prefs.http_timeout)
+		r = net.get(url)
 	except Exception as e:
 		print("Failure fetching", url, net.redact(e), file=sys.stderr)
 		return market_data # the ASX short interest is missing, the rest of the report is not
diff --git a/lib/telegram.py b/lib/telegram.py
index 511f0539..ecc16e1d 100644
--- a/lib/telegram.py
+++ b/lib/telegram.py
@@ -5,7 +5,7 @@
 
 def call(method, **kwargs):
 	"""POST a Bot API method and return its result. Raises RuntimeError with Telegram's description when it answers not ok."""
-	response = net.post(config.env.webhooks['telegram'] + method, timeout=config.prefs.http_timeout, **kwargs)
+	response = net.post(config.env.webhooks['telegram'] + method, **kwargs)
 	try:
 		output = response.json()
 	except ValueError:
diff --git a/lib/transports.py b/lib/transports.py
index b9838788..934a6aea 100644
--- a/lib/transports.py
+++ b/lib/transports.py
@@ -120,8 +120,7 @@ def acknowledge(self, press, text=None):
 	def _post_json(self, url, payload, headers=None):
 		"""POST a JSON message: the answer ({} when it is not JSON), or None when the network refused it or failed."""
 		try:
-			r = net.post(url, headers={'Content-type': 'application/json', **(headers or {})}, json=payload,
-				timeout=config.prefs.http_timeout)
+			r = net.post(url, headers={'Content-type': 'application/json', **(headers or {})}, json=payload)
 		except Exception as e:
 			print("Failure sending to", self.name + ":", net.redact(e), file=sys.stderr)
 			return None
@@ -171,7 +170,7 @@ def post_chart(self, image, caption, buttons=None):
 			data['reply_markup'] = json.dumps(self.markup(buttons))
 		try:
 			r = net.post(config.env.webhooks['telegram'] + 'sendPhoto?chat_id=' + str(self.where), data=data,
-				files={'photo': ('image.png', image)}, timeout=config.prefs.http_timeout)
+				files={'photo': ('image.png', image)})
 		except requests.exceptions.ReadTimeout as e:
 			print("Telegram sendPhoto timed out, the chart may still arrive:", net.redact(e), file=sys.stderr)
 			return MAYBE_SENT
@@ -326,7 +325,7 @@ def post_chart(self, image, caption, buttons=None):
 	def upload(self, data):
 		"""Send a chart's bytes to Slack: the id of the file, not completed yet. Raises RuntimeError on a refusal."""
 		ticket = slack_api('files.getUploadURLExternal', filename='chart.png', length=len(data))
-		r = net.post(ticket['upload_url'], data=data, timeout=config.prefs.http_timeout)
+		r = net.post(ticket['upload_url'], data=data)
 		if r.status_code != 200:
 			raise RuntimeError(f"Slack upload: {r.status_code} {r.text[:200]}")
 		return ticket['file_id']
@@ -389,7 +388,7 @@ def post_chart(self, image, caption, buttons=None):
 			'attachments': [{'id': 0, 'filename': 'chart.png'}]}
 		try:
 			r = net.post(self.url(), data={'payload_json': json.dumps(payload)},
-				files={'files[0]': ('chart.png', image, 'image/png')}, timeout=config.prefs.http_timeout)
+				files={'files[0]': ('chart.png', image, 'image/png')})
 		except requests.exceptions.ReadTimeout as e:
 			print("Discord chart timed out, it may still arrive:", net.redact(e), file=sys.stderr)
 			return MAYBE_SENT
@@ -411,7 +410,7 @@ def slack_api(method, **data):
 	if not config.env.slackBotToken:
 		raise RuntimeError(f"Slack {method}: needs slackBotToken in finbot.env")
 	r = net.post('https://slack.com/api/' + method, headers={'Authorization': 'Bearer ' + config.env.slackBotToken},
-		data=data, timeout=config.prefs.http_timeout)
+		data=data)
 	try:
 		output = r.json()
 	except ValueError:
diff --git a/lib/yahoo.py b/lib/yahoo.py
index 65313b05..fdee9643 100644
--- a/lib/yahoo.py
+++ b/lib/yahoo.py
@@ -2,7 +2,6 @@
 import datetime
 from dateutil.relativedelta import relativedelta
 import hashlib
-import requests
 import threading
 import typing
 import zoneinfo
@@ -14,85 +13,75 @@
 from lib import formats
 from lib import markets
 from http.cookies import SimpleCookie
-from pathlib import Path
 
-_auth = threading.RLock() # Yahoo's cookie and crumb are renewed by one thread at a time: two at once could cache a crumb for the cookie the other replaced
+_auth = threading.Lock() # Yahoo's cookie and crumb are renewed by one thread at a time, or two could pair up wrongly
 HEADERS = {'User-Agent': 'Mozilla/5.0'}
+CREDENTIALS = 'finbot_yahoo_credentials.json' # [cookie, crumb]: Yahoo issues the crumb for the cookie: they go together
 
-def get_cookie(max_age=1209600):
-	"""Yahoo's session cookie as 'name=value', cached for max_age seconds (14 days, renewed 30 seconds early) whatever
-	the cache preference, since the crumb must match it: a new cookie drops the crumb. A fallback cookie when Yahoo
-	sends none."""
+def credentials(renew=False):
+	"""(cookie, crumb) for Yahoo's API, from one cache file kept 14 days whatever the cache preference, and renewed
+	together (renew=True: now, for a pair Yahoo refused). Raises RuntimeError when Yahoo gives no cookie or no crumb."""
 	with _auth:
-		cacheFile = "finbot_yahoo_cookie.json"
-		cached = cache.read(cacheFile, max_age - 30) # renewed 30 seconds before it expires
-		if cached:
-			return cached[0][0] + '=' + cached[0][1]
-		url = 'https://fc.yahoo.com/' # 404, but that is ok because cookie is in response header
+		if not renew and (cached := cache.read(CREDENTIALS, 14 * 86400)):
+			return tuple(cached)
 		try:
-			r = net.get(url, headers=HEADERS, timeout=config.prefs.http_timeout)
+			r = net.get('https://fc.yahoo.com/', headers=HEADERS) # a 404, with the cookie in its headers
 		except Exception as e:
-			print(net.redact(e), file=sys.stderr)
-			r = requests.Response() # empty: falls through to the fallback cookie below
-		if r.status_code not in {200, 404}:
-			print(r.status_code, "returned by", url, file=sys.stderr)
-		if 'set-cookie' not in r.headers:
-			print("Failed to obtain Yahoo auth cookie. Returning fallback cookie", file=sys.stderr)
-			return 'A3=d=AQABBBtR5mYCENFZu2wCWkA5iGGkSRGvRgkFEgEBAQGi52bwZtxM0iMA_eMAAA&S=AQAAAtG8VhxZN7aXopfvLNObtpE;'
-		cookie = SimpleCookie()
-		cookie.load(r.headers['Set-Cookie'])
-		cookielist = [[name, morsel.value, int(morsel.get('max-age') or 0)] for name, morsel in cookie.items()]
-		cache.write(cacheFile, cookielist)
-		print("Got a new Yahoo cookie", file=sys.stderr)
-		drop_crumb() # we must invalidate the non-matching crumb
-		return cookielist[0][0] + '=' + cookielist[0][1]
-
-def get_crumb(seconds=1209600):
-	"""Yahoo's crumb, which goes with the cookie, cached for `seconds` (14 days) while the cache preference is on; a
-	fallback crumb when Yahoo gives none."""
-	with _auth:
-		cookie = get_cookie()
-		cacheFile = "finbot_yahoo_crumb.json"
-		if config.prefs.cache and (cached := cache.read(cacheFile, seconds)):
-			return cached
-		r = _get_first(_hosts('/v1/test/getcrumb'), {**HEADERS, 'Cookie': cookie})
-		if r is None:
-			print("Exhausted Yahoo API attempts. Returning fallback crumb", file=sys.stderr)
-			return 'jkQEU8yLqxs'
-		if config.prefs.cache:
-			cache.write(cacheFile, r.text)
-		print("Got a new Yahoo crumb", file=sys.stderr)
-		return r.text
-
-def credentials():
-	"""(cookie, crumb), taken together under the lock: a crumb is only good with the cookie it was issued for, and a renewal
-	between two separate calls could pair a new cookie with an old crumb."""
-	with _auth:
-		crumb = get_crumb() # renews the cookie too, when due
-		return get_cookie(), crumb
-
-def drop_crumb():
-	"""Set the cached crumb aside (as .old): it belongs to the cookie being replaced."""
-	crumb = Path(cache.path("finbot_yahoo_crumb.json"))
-	if crumb.is_file():
-		crumb.replace(crumb.with_name(crumb.name + '.old'))
+			raise RuntimeError(f"Yahoo unreachable for a cookie: {net.redact(e)}") from e
+		jar = SimpleCookie()
+		jar.load(r.headers.get('Set-Cookie', ''))
+		if not jar:
+			raise RuntimeError(f"Yahoo gave no cookie ({r.status_code})")
+		cookie = '{}={}'.format(*next((name, morsel.value) for name, morsel in jar.items()))
+		for url in _hosts('/v1/test/getcrumb'):
+			try:
+				r = net.get(url, headers={**HEADERS, 'Cookie': cookie})
+			except Exception as e:
+				print("Yahoo crumb:", net.redact(e), file=sys.stderr)
+				continue
+			if r.status_code == 200 and r.text.strip():
+				cache.write(CREDENTIALS, [cookie, r.text.strip()])
+				print("Got a new Yahoo cookie and crumb", file=sys.stderr)
+				return cookie, r.text.strip()
+			print(r.status_code, "from Yahoo for a crumb", file=sys.stderr)
+		raise RuntimeError("Yahoo gave no crumb")
 
 def _hosts(path):
 	"""One Yahoo API path on its two hosts: query2, then query1 to fall back on."""
 	return ['https://query2.finance.yahoo.com' + path, 'https://query1.finance.yahoo.com' + path]
 
-def _get_first(urls, headers):
-	"""The answer of the first URL that answers 200, or None when none does. Each failure is logged."""
-	for url in urls:
-		try:
-			r = net.get(url, headers=headers, timeout=config.prefs.http_timeout)
-		except Exception as e:
-			print(net.redact(e), file=sys.stderr)
-			continue
-		if r.status_code == 200:
-			return r
-		print(r.status_code, r.text.strip()[:200], "returned by", net.redact(url), file=sys.stderr)
-	return None
+def _answer(query, what):
+	"""Yahoo's JSON answer to an API query, from query2 or else query1, with the cookie and its crumb (renewed once when
+	Yahoo refuses them): None when Yahoo says it has no such symbol (404 Not Found). Raises RuntimeError when no host
+	answers, or one refuses otherwise; `what` names the request in the error."""
+	refusal = ''
+	for renew in (False, True):
+		cookie, crumb = credentials(renew)
+		for url in _hosts(query + ('&' if '?' in query else '?') + 'crumb=' + crumb):
+			try:
+				r = net.get(url, headers={**HEADERS, 'Cookie': cookie})
+			except Exception as e:
+				refusal = f"Yahoo unreachable for {what}: {net.redact(e)}"
+				print(refusal, file=sys.stderr)
+				continue
+			if r.status_code == 200:
+				return r.json()
+			if r.status_code == 404 and not_found(r):
+				return None
+			refusal = f"{r.status_code} error for {what} from Yahoo"
+			print(refusal + ':', r.text.strip()[:200], file=sys.stderr)
+		if not refusal.startswith('401 '): # only a refused crumb is worth a new one
+			break
+	raise RuntimeError(refusal)
+
+def not_found(response):
+	"""Whether an answer is Yahoo saying it has no such symbol (the chart's or the quoteSummary's error code Not Found),
+	rather than some other failure."""
+	try:
+		body = response.json()
+	except ValueError:
+		return False
+	return isinstance(body, dict) and any(dig(part, 'error', 'code') == 'Not Found' for part in body.values())
 
 def dig(node, *path):
 	"""node[path[0]][path[1]]..., or None where Yahoo left a part out or sent null."""
@@ -150,10 +139,8 @@ def _cached(name, seconds, get, *args):
 	('quoteType', 'quoteType', None),
 	('marketState', 'marketState', None),
 	('profile_exchange', 'fullExchangeName', None),
-	('exchangeTimezoneName', 'exchangeTimezoneName', None),
 	('regularMarketTime', 'regularMarketTime', None),
 	('financialCurrency', 'financialCurrency', None),
-	('regularMarketPreviousClose', 'regularMarketPreviousClose', None),
 	('fiftyTwoWeekHigh', 'fiftyTwoWeekHigh', None),
 	('fiftyTwoWeekLow', 'fiftyTwoWeekLow', _positive),
 )
@@ -168,7 +155,7 @@ def fetch(tickers):
 	name = "finbot_yahoo_quote_" + hashlib.sha256("_".join(tickers).encode()).hexdigest() + '.json'
 	data = _cached(name, 300, _quote_answer, tickers)
 	if data is None:
-		print(f"{tickers}†", sep=' ', end='', flush=True, file=sys.stderr)
+		print("Yahoo has no quotes for", ', '.join(tickers), file=sys.stderr)
 		return {} # callers iterate the result
 	now = datetime.datetime.now().timestamp()
 	yahoo_output = {}
@@ -188,16 +175,11 @@ def fetch(tickers):
 	return yahoo_output
 
 def _quote_answer(tickers):
-	"""The v7 answer for the tickers, None for an error (not cached); raises RuntimeError when Yahoo is unreachable."""
-	cookie, crumb = credentials()
-	query = '/v7/finance/quote?crumb=' + crumb + '&symbols=' + ','.join(tickers)
-	r = _get_first(_hosts(query), {**HEADERS, 'Content-type': 'application/json', 'Cookie': cookie})
-	if r is None:
-		print("Exhausted Yahoo API attempts. Giving up", file=sys.stderr)
-		raise RuntimeError("Yahoo API unavailable")
-	data = r.json()
-	quotes = data['quoteResponse']
-	return data if quotes['result'] is not None and quotes['error'] is None else None
+	"""The v7 answer for the tickers, None for an error answer (not cached). Raises RuntimeError when Yahoo cannot be
+	reached or refuses."""
+	data = _answer('/v7/finance/quote?symbols=' + ','.join(tickers), 'quotes')
+	quotes = dig(data, 'quoteResponse') or {}
+	return data if quotes.get('result') is not None and quotes.get('error') is None else None
 
 DETAIL_FIELDS = ( # (key, path in the quoteSummary result[, conversion]); a field Yahoo leaves out, sends as null or garbles is left out
 	('profile_bio', ('summaryProfile', 'longBusinessSummary')),
@@ -214,10 +196,8 @@ def _quote_answer(tickers):
 	('dividend', ('summaryDetail', 'dividendYield', 'raw'), _percent),
 	('price_to_earnings_trailing', ('summaryDetail', 'trailingPE', 'raw'), round),
 	('price_to_earnings_forward', ('summaryDetail', 'forwardPE', 'raw')),
-	('percent_change_year', ('summaryDetail', '52WeekChange', 'raw'), _percent),
 	('price_to_sales', ('summaryDetail', 'priceToSalesTrailing12Months', 'raw')),
 	('ex_dividend_date', ('calendarEvents', 'exDividendDate', 'raw')),
-	('dividend_date', ('calendarEvents', 'DividendDate', 'raw')),
 	('earnings_date', ('calendarEvents', 'earnings', 'earningsDate', 0, 'raw')),
 	('percent_change', ('price', 'regularMarketChangePercent', 'raw'), _percent),
 	('percent_change_premarket', ('price', 'preMarketChangePercent', 'raw'), _percent),
@@ -226,21 +206,12 @@ def _quote_answer(tickers):
 	('marketState', ('price', 'marketState')),
 	('quoteType', ('price', 'quoteType')),
 	('regularMarketPrice', ('price', 'regularMarketPrice', 'raw')),
-	('regularMarketPreviousClose', ('price', 'regularMarketPreviousClose', 'raw')),
-	('fiftyTwoWeekHigh', ('price', 'fiftyTwoWeekHigh', 'raw')),
-	('fiftyTwoWeekLow', ('price', 'fiftyTwoWeekLow', 'raw'), _positive),
 	('prePostMarketPrice', ('price', 'preMarketPrice', 'raw'), float),
 	('prePostMarketPrice', ('price', 'postMarketPrice', 'raw'), float), # the after-hours price, when there is one, wins
 	('netIncomeToCommon', ('defaultKeyStatistics', 'netIncomeToCommon', 'raw')),
 	('sharesOutstanding', ('defaultKeyStatistics', 'sharesOutstanding', 'raw'), float),
 	('short_percent', ('defaultKeyStatistics', 'shortPercentOfFloat', 'raw'), _percent),
-	('price_to_book', ('defaultKeyStatistics', 'priceToBook', 'raw')),
-	('earnings_growth_q', ('defaultKeyStatistics', 'earningsQuarterlyGrowth', 'raw')),
 	('price_to_earnings_peg', ('defaultKeyStatistics', 'pegRatio', 'raw')),
-	('profit_margin', ('defaultKeyStatistics', 'profitMargins', 'raw')),
-	('shareholder_equity', ('balanceSheetHistoryQuarterly', 'balanceSheetStatements', 0, 'totalStockholderEquity', 'raw')),
-	('total_debt', ('financialData', 'totalDebt', 'raw')),
-	('total_cash', ('financialData', 'totalCash', 'raw')),
 	('free_cashflow', ('financialData', 'freeCashflow', 'raw')),
 	('operating_cashflow', ('financialData', 'operatingCashflow', 'raw')),
 	('financialCurrency', ('financialData', 'financialCurrency')),
@@ -251,18 +222,17 @@ def _quote_answer(tickers):
 RECOMMENDATION = ('recommend', 'recommend_index', 'recommend_analysts') # all three, or none
 
 def fetch_detail(ticker, seconds=None):
-	"""One ticker's quoteSummary: {ticker: {...}}, cached `seconds` (or cache_seconds); {} without a name or price."""
+	"""One ticker's quoteSummary: {ticker: {...}}, cached `seconds` (or cache_seconds); {} for a symbol Yahoo does not
+	know, or without a name or price. Raises RuntimeError when Yahoo cannot be reached or refuses."""
 	now = datetime.datetime.now()
 	data = _cached("finbot_yahoo_detail_" + ticker + '.json', seconds, _detail_answer, ticker)
-	if data is None:
-		return {} # catches some delisted stocks like "DRNA"
 	result = dig(data, 'quoteSummary', 'result', 0)
-	if result is None: # Yahoo answers result=null for unknown symbols
-		print(ticker + 'x', sep=' ', end='', flush=True, file=sys.stderr)
+	if result is None: # Yahoo answers 404 Not Found, or result=null, for a symbol it does not know
+		print(ticker, "is unknown to Yahoo", file=sys.stderr)
 		return {}
 	title = next((name for name in (dig(result, 'price', 'longName'), dig(result, 'price', 'shortName')) if isinstance(name, str) and name), None)
 	if not title: # catches some delisted stocks like "DUB"
-		print(f"{ticker}†", sep=' ', end='', flush=True, file=sys.stderr)
+		print(ticker, "has no name at Yahoo", file=sys.stderr)
 		return {}
 	detail = {'profile_title': formats.transform_title(title)}
 	for key, path, *convert in DETAIL_FIELDS:
@@ -281,16 +251,11 @@ def fetch_detail(ticker, seconds=None):
 	trend = _list(dig(result, 'earningsTrend', 'trend'))
 	if trend:
 		year = next((item for item in trend if dig(item, 'period') == '+1y'), None)
-		estimates = [dig(year, kind, field, 'raw') for kind, field in (('revenueEstimate', 'growth'), ('earningsEstimate', 'growth'),
-			('revenueEstimate', 'numberOfAnalysts'), ('earningsEstimate', 'numberOfAnalysts'))]
-		if None in estimates:
-			print(f'{ticker}†', sep=' ', end='', flush=True, file=sys.stderr)
-		else:
+		estimates = [dig(year, kind, 'growth', 'raw') for kind in ('revenueEstimate', 'earningsEstimate')]
+		if None not in estimates:
 			detail['revenueEstimateY'], detail['earningsEstimateY'] = _percent(estimates[0]), _percent(estimates[1])
-			detail['revenueAnalysts'], detail['earningsAnalysts'] = estimates[2], estimates[3]
-	if dig(quarterly, 0, 'revenue', 'raw') is None:
-		print(f'{ticker}†', sep=' ', end='', flush=True, file=sys.stderr)
-	else: # a quarter without an earnings figure is bad data: None for both, which trend_marks marks
+	# a quarter without an earnings figure is bad data: None for both, which trend_marks marks
+	if dig(quarterly, 0, 'revenue', 'raw') is not None:
 		detail['earningsQ'] = [None if dig(item, 'earnings', 'fmt') is None else dig(item, 'earnings', 'raw') for item in quarterly]
 		detail['revenueQ'] = [None if dig(item, 'earnings', 'fmt') is None else dig(item, 'revenue', 'raw') for item in quarterly]
 		yearly = _list(dig(result, 'earnings', 'financialsChart', 'yearly'))
@@ -312,15 +277,13 @@ def fetch_detail(ticker, seconds=None):
 		detail['insiderSell'], detail['insiderSellValue'] = totals['Sale']
 	return {ticker: dict(sorted(detail.items()))}
 
-DETAIL_MODULES = ('calendarEvents,defaultKeyStatistics,balanceSheetHistoryQuarterly,financialData,summaryProfile,'
-	'summaryDetail,price,earnings,earningsTrend,insiderTransactions') # the quoteSummary modules fetch_detail reads
+DETAIL_MODULES = ('calendarEvents,defaultKeyStatistics,financialData,summaryProfile,summaryDetail,price,earnings,'
+	'earningsTrend,insiderTransactions') # the quoteSummary modules fetch_detail reads
 
 def _detail_answer(ticker):
-	"""Yahoo's quoteSummary answer for one ticker, with every module finbot reads, or None when no host gives one."""
-	cookie, crumb = credentials() # only on a miss: a cached detail needs neither, nor the lock
-	query = '/v10/finance/quoteSummary/' + ticker + '?modules=' + DETAIL_MODULES + '&crumb=' + crumb
-	r = _get_first(_hosts(query), {**HEADERS, 'Cookie': cookie})
-	return None if r is None else r.json()
+	"""Yahoo's quoteSummary answer for one ticker, with every module finbot reads; None for a symbol Yahoo does not
+	know. Raises RuntimeError when Yahoo cannot be reached or refuses."""
+	return _answer(f'/v10/finance/quoteSummary/{ticker}?modules={DETAIL_MODULES}', ticker)
 
 class Line(typing.NamedTuple):
 	"""The prices behind a price_history, for its chart: the company's name, Yahoo's full name for its exchange (as
@@ -337,71 +300,84 @@ class NoData(answer.Answer, RuntimeError):
 	"""Yahoo has nothing for it (no such symbol, no prices): a reply in a chat, a RuntimeError where a list skips it."""
 
 
-def price_history(ticker, days=None):
-	"""({period: percent change}, its Line): the change over `days`, else over each standard period, and the closes
-	behind it, for a chart. Raises NoData when there are no prices, RuntimeError when the history cannot be fetched."""
-	import pandas as pd
-	from pandas.tseries.offsets import BDay as businessday
-	percent_dict = {}
-	now = datetime.datetime.now()
-	interval = ('10Y', '5Y', '3Y', '1Y', 'YTD', '6M', '3M', '1M', '7D', '5D', '1D')
+def bars(chart_json):
+	"""A chart answer's daily bars, oldest first: [(the day, in the exchange's time zone; open; close)], a price Yahoo
+	left out None. Yahoo stamps a bar with the time its market opened, which in UTC is the day before for Sydney or
+	Auckland."""
+	result = dig(chart_json, 'chart', 'result', 0) or {}
+	zone = zoneinfo.ZoneInfo(dig(result, 'meta', 'exchangeTimezoneName') or 'UTC')
+	stamps = _list(result.get('timestamp'))
+	quote = dig(result, 'indicators', 'quote', 0) or {}
+	opens, closes = _list(quote.get('open')) or [None] * len(stamps), _list(quote.get('close')) or [None] * len(stamps)
+	return [(datetime.datetime.fromtimestamp(stamp, zone).date(), first, last)
+		for stamp, first, last in zip(stamps, opens, closes)]
+
+INTERVALS = ('10Y', '5Y', '3Y', '1Y', 'YTD', '6M', '3M', '1M', '7D', '5D', '1D') # price_history's standard periods
+
+def price_history(ticker, days=None, quote=None):
+	"""({period: percent change}, its Line): the change over `days`, else over each standard period (and 'Max', from the
+	first close, when one goes back further than the history), and the closes behind it, for a chart. The change is to
+	the price now: the quote's (the ticker's quote or detail, which carries today's change too), else the chart's.
+	Days are counted in the exchange's time zone. Raises NoData when there are no prices, RuntimeError when the history
+	cannot be fetched."""
 	data = fetch_chart_json(ticker)
-	df = chart_json_to_df(data)
-	stock = chart_json_to_stock_basics(data)
-	closes = df['Close'].dropna()
-	if closes.empty:
+	meta = dig(data, 'chart', 'result', 0, 'meta') or {}
+	daily = bars(data)
+	closes = [(day, last) for day, _, last in daily if last is not None]
+	if not closes:
 		raise NoData(f"no price history for {ticker}")
-	price = stock.get('regularMarketPrice') or closes.iloc[-1] # the chart's last close when Yahoo gives no current price
-	tz = zoneinfo.ZoneInfo(stock.get('exchangeTimezoneName') or 'UTC')
+	quote = quote or {}
+	price = quote.get('regularMarketPrice') or meta.get('regularMarketPrice') or closes[-1][1]
+	today = datetime.datetime.now(zoneinfo.ZoneInfo(meta.get('exchangeTimezoneName') or 'UTC')).date()
+	change = lambda past: (price - past) / past * 100
+	changes = {}
 	if days:
-		df = df[df['Date'] >= (now - datetime.timedelta(days=days)).date()].reset_index(drop=True)
-		if df['Close'].dropna().empty:
+		closes = [(day, last) for day, last in closes if day >= today - datetime.timedelta(days=days)]
+		if not closes:
 			raise NoData(f"no price history for {ticker} in the past {days} days")
-		past_price = df['Close'].dropna().iloc[0]
-		percent_dict[days] = round((price - past_price) / past_price * 100, 2)
+		changes[days] = round(change(closes[0][1]), 2)
 	else:
-		default_price = df['Close'].iloc[0]
-		default_percent = round((price - default_price) / default_price * 100)
-		for period in interval:
-			if period == '1D':
-				# against the previous session's close: the last close dated before the latest bar, in the exchange's time zone
-				sessions = pd.to_datetime(df['Timestamp'], unit='s', utc=True).dt.tz_convert(tz).dt.date
-				earlier = df[(sessions < sessions.iloc[-1]) & df['Close'].notnull()]
-				if earlier.empty:
-					continue
-				past_price = earlier['Close'].iloc[-1]
-			else: # try to align with google finance
-				try:
-					if period == 'YTD':
-						past_price = df[df['Date'] >= now.replace(day=1, month=1).date()]['Close'].iloc[0]
-					elif period == '5D':
-						past_price = df[df['Date'] <= (now - businessday(5)).date()]['Open'].iloc[-1]
-					else: # 10Y to 1M, 7D: the last close on or before the day that long ago
-						n, unit = int(period[:-1]), period[-1]
-						ago = relativedelta(years=n) if unit == 'Y' else relativedelta(months=n) if unit == 'M' else datetime.timedelta(days=n)
-						past_price = df[df['Date'] <= (now - ago).date()]['Close'].iloc[-1]
-				except IndexError: # the history does not go back that far
-					percent_dict['Max'] = default_percent
-					continue
-			percent_dict[period] = round((price - past_price) / past_price * 100)
-	closes = tuple((str(day), float(close)) for day, close in zip(df['Date'], df['Close']) if pd.notnull(close))
-	exchange = stock.get('fullExchangeName') or stock.get('exchangeName') or ''
-	name = formats.transform_title(stock.get('longName', ''))
-	return percent_dict, Line(name, exchange, stock.get('currency') or '', price, closes)
+		week_ago = today # five trading days back, as Google Finance counts its 5D
+		for _ in range(5):
+			week_ago -= datetime.timedelta(days=3 if week_ago.weekday() == 0 else 2 if week_ago.weekday() == 6 else 1)
+		for period in INTERVALS:
+			if period == '1D': # since the previous session's close: the quote's, or the last close before the last bar
+				before = [last for day, last in closes if day < daily[-1][0]]
+				if quote.get('percent_change') is not None:
+					changes['1D'] = round(quote['percent_change'])
+				elif before:
+					changes['1D'] = round(change(before[-1]))
+				continue
+			if period == 'YTD':
+				past = next((last for day, last in closes if day >= today.replace(month=1, day=1)), None)
+			elif period == '5D': # the open of that day, to align with Google Finance
+				past = next((first for day, first, _ in reversed(daily) if day <= week_ago and first is not None), None)
+			else: # 10Y to 1M, 7D: the last close on or before the day that long ago
+				unit = {'Y': 'years', 'M': 'months', 'D': 'days'}[period[-1]]
+				cutoff = today - relativedelta(**{unit: int(period[:-1])}) # once: relativedelta is slow
+				past = next((last for day, last in reversed(closes) if day <= cutoff), None)
+			if past is None: # the history does not go back that far
+				changes['Max'] = round(change(closes[0][1]))
+			else:
+				changes[period] = round(change(past))
+	name = formats.transform_title(meta.get('longName', ''))
+	exchange = meta.get('fullExchangeName') or meta.get('exchangeName') or ''
+	return changes, Line(name, exchange, meta.get('currency') or '', price,
+		tuple((str(day), float(last)) for day, last in closes))
 
 def price_series(ticker):
 	"""A ticker's daily closes over 10 years, from the same chart as price_history(): (name, exchange, pandas Series
-	indexed by datetime). Raises NoData when there is no history."""
+	indexed by date). Raises NoData when there is no history."""
 	import pandas as pd
 	data = fetch_chart_json(ticker)
-	df = chart_json_to_df(data)
-	stock = chart_json_to_stock_basics(data)
-	df = df[df['Close'].notnull()]
-	if df.empty:
+	meta = dig(data, 'chart', 'result', 0, 'meta') or {}
+	closes = [(day, last) for day, _, last in bars(data) if last is not None]
+	if not closes:
 		raise NoData(f"no price history for {ticker}")
-	series = pd.Series(df['Close'].astype(float).values, index=pd.to_datetime(df['Date'].astype(str)), name=ticker)
-	name = formats.transform_title(stock.get('longName') or stock.get('shortName') or ticker)
-	return name, stock.get('fullExchangeName') or stock.get('exchangeName') or '', series
+	series = pd.Series([float(last) for _, last in closes], index=pd.to_datetime([str(day) for day, _ in closes]),
+		name=ticker)
+	name = formats.transform_title(meta.get('longName') or meta.get('shortName') or ticker)
+	return name, meta.get('fullExchangeName') or meta.get('exchangeName') or '', series
 
 def fetch_each(get, tickers, workers=8):
 	"""get(ticker) for every ticker, `workers` at a time so their waits overlap: [its result, or the exception it raised], in order."""
@@ -483,52 +459,18 @@ def fetch_chart_json(ticker, days=3665, seconds=None, full=False):
 def _chart_answer(ticker, days, full):
 	"""The v8 chart answer: daily bars of the last `days`, or all with full=True. Raises as fetch_chart_json does."""
 	now = datetime.datetime.now()
-	cookie, crumb = credentials()
-	start = '0' if full else str(int((now - datetime.timedelta(days=days)).timestamp()))
-	url = (f'https://query1.finance.yahoo.com/v8/finance/chart/{ticker}?period1={start}&period2={int(now.timestamp())}'
-		f'&interval=1d&crumb={crumb}')
-	print("Fetching", net.redact(url), file=sys.stderr) if config.prefs.debug else None
-	try:
-		r = net.get(url, headers={**HEADERS, 'Content-type': 'application/json', 'Cookie': cookie}, timeout=config.prefs.http_timeout)
-	except Exception as e:
-		raise RuntimeError(f"Yahoo unreachable for {ticker}: {net.redact(e)}") from e
-	if r.status_code == 404 and not_found(r):
+	start = 0 if full else int((now - datetime.timedelta(days=days)).timestamp())
+	data = _answer(f'/v8/finance/chart/{ticker}?period1={start}&period2={int(now.timestamp())}&interval=1d', ticker)
+	if data is None:
 		raise NoData(f"no price history for {ticker}")
-	if r.status_code != 200:
-		raise RuntimeError(f"{r.status_code} error for {ticker} from Yahoo's chart API")
-	return r.json()
-
-def not_found(response):
-	"""Whether a chart API answer is Yahoo saying it has no such symbol, rather than some other 404."""
-	try:
-		return dig(response.json(), 'chart', 'error', 'code') == 'Not Found'
-	except ValueError:
-		return False
-
-def chart_json_to_df(chart_json):
-	"""The chart's daily bars as a DataFrame: Timestamp, Open, High, Low, Close, Volume and the Date."""
-	import pandas as pd
-	result = chart_json['chart']['result'][0]
-	df = pd.DataFrame({'Timestamp': result['timestamp'], **result['indicators']['quote'][0]})
-	df['Date'] = pd.to_datetime(df['Timestamp'], unit='s').dt.date
-	df.columns = [col.capitalize() for col in df.columns]
-	return df
-
-def chart_json_to_stock_basics(chart_json):
-	"""The chart's meta, which saves a call to fetch(): currency, symbol, exchangeName and fullExchangeName, exchangeTimezoneName,
-	regularMarketPrice and regularMarketTime, longName and shortName, fiftyTwoWeekHigh and Low, and more. {} when there is none."""
-	return dig(chart_json, 'chart', 'result', 0, 'meta') or {}
+	return data
 
 def historic_high(ticker):
-	"""(highest high, lowest low) of the ticker's history. Raises NoData when it has none (e.g. total-return indices
-	such as ^SPTTGD carry no High/Low data)."""
-	df = chart_json_to_df(fetch_chart_json(ticker))
-	df = df[df.High.notnull()].iloc[1:].reset_index(drop=True)
-	if df.empty:
+	"""(highest high, lowest low) of the ticker's whole history, its first day aside (Yahoo's first bar can be partial).
+	Raises NoData when it has none (total-return indices such as ^SPTTGD carry no highs and lows)."""
+	quote = dig(fetch_chart_json(ticker, full=True), 'chart', 'result', 0, 'indicators', 'quote', 0) or {}
+	ranges = [(high, low) for high, low in zip(_list(quote.get('high')), _list(quote.get('low')))
+		if high is not None and low is not None][1:]
+	if not ranges:
 		raise NoData(f"no high/low history for {ticker}")
-	highrow = df.iloc[df['High'].argmax()]
-	lowrow = df.iloc[df['Low'].argmin()]
-	if config.prefs.debug:
-		print(ticker, "High", highrow['Date'], round(highrow['High'], 2), file=sys.stderr)
-		print(ticker, "Low", lowrow['Date'], round(lowrow['Low'], 2), file=sys.stderr)
-	return round(highrow['High'], 2), round(lowrow['Low'], 2)
+	return round(max(high for high, _ in ranges), 2), round(min(low for _, low in ranges), 2)
diff --git a/tests/common.py b/tests/common.py
index 608aabcb..d1ed097c 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -14,7 +14,8 @@
 from lib import sharesight, shortman, state, telegram, transports, watchlist, worker, yahoo
 from lib.reports import about, cal, history, holdings, milestone, performance, price, profile, rankings, rating, reminder, shorts, trades
 
-REAL = {(m.__name__, n): getattr(m, n) for m, n in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (yahoo, 'price_history'),
+REAL = {(m.__name__, n): getattr(m, n) for m, n in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'),
+	(yahoo, 'price_history'), (yahoo, 'historic_high'),
 	(sharesight, 'get_portfolios'), (sharesight, 'get_trades'), (sharesight, 'get_performance'))}
 REAL_GET_HOLDINGS = watchlist.with_holdings
 REAL_TEXT = transports.Transport.text # FinbotCase replaces it with a recorder; tests of the real thing put it back
@@ -94,7 +95,7 @@ def setUp(self):
 		self.patch(watchlist, 'with_holdings', lambda: TICKERS)
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t)})
-		self.patch(yahoo, 'price_history', lambda t, days=None, **k: # a history with no line to draw
+		self.patch(yahoo, 'price_history', lambda t, days=None, quote=None: # a history with no line to draw
 			({days: 1.5} if days else {'Max': 5, '1Y': 3}, yahoo.Line('Co ' + t, 'NasdaqGS', 'USD', 100.0, ())))
 		self.patch(yahoo, 'prefetch_history', lambda ts: None)
 		self.patch(yahoo, 'historic_high', lambda t: (100.0, 50.0))
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 399c9511..28436e52 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -40,6 +40,9 @@ def get(url, **kwargs):
 	return get
 
 
+REAL_CREDENTIALS = yahoo.credentials
+
+
 class ProviderCase(FinbotCase):
 	def setUp(self):
 		"""Real parsers instead of the fixture's canned data, with a fixed cookie, crumb and token, and no waiting."""
@@ -48,8 +51,7 @@ def setUp(self):
 		for module, name in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (sharesight, 'get_portfolios'),
 							 (sharesight, 'get_trades'), (sharesight, 'get_performance')):
 			self.patch(module, name, REAL[(module.__name__, name)])
-		self.patch(yahoo, 'get_cookie', lambda *a, **k: 'A=B')
-		self.patch(yahoo, 'get_crumb', lambda *a, **k: 'crumb')
+		self.patch(yahoo, 'credentials', lambda renew=False: ('A=B', 'crumb'))
 		self.patch(sharesight, 'get_token', lambda: 'token')
 		self.patch(sharesight.time, 'sleep', lambda s: None)
 
@@ -71,7 +73,6 @@ def test_parses_real_v7_quote(self):
 		self.assertEqual(data['quoteType'], 'EQUITY')
 		self.assertEqual(data['marketState'], 'CLOSED')
 		self.assertEqual(data['profile_exchange'], 'NasdaqGS')
-		self.assertEqual(data['exchangeTimezoneName'], 'America/New_York')
 		self.assertGreater(data['market_cap'], 1e12)
 
 	def test_second_call_is_served_from_cache(self):
@@ -174,12 +175,17 @@ def test_the_cache_holds_yahoos_answer_so_a_parser_fix_needs_no_flush(self):
 		self.patch(yahoo, 'DETAIL_FIELDS', yahoo.DETAIL_FIELDS + (('short_name', ('price', 'shortName')),))
 		self.assertEqual(yahoo.fetch_detail('NVDA')['NVDA']['short_name'], 'NVIDIA Corporation')
 
-	def test_http_failure_gives_empty(self):
-		"""A detail Yahoo will not give (404, unreachable) is empty, not an error."""
-		for failure in (Resp(status=404), ConnectionError('down')):
+	def test_a_symbol_yahoo_does_not_know_is_empty_and_an_outage_an_error(self):
+		"""A detail of a symbol Yahoo does not know (404 Not Found) is empty, an answer; Yahoo unreachable or refusing
+		is an error, where it used to read as an unknown symbol ("I could not find TSLA")."""
+		unknown = {'quoteSummary': {'result': None, 'error': {'code': 'Not Found', 'description': 'Quote not found'}}}
+		self.http(yahoo, Resp(unknown, status=404))
+		self.assertEqual(yahoo.fetch_detail('DRNA'), {})
+		for failure in (Resp(status=404), Resp(status=500), ConnectionError('down')):
 			with self.subTest(failure=failure):
 				self.http(yahoo, failure)
-				self.assertEqual(yahoo.fetch_detail('DRNA'), {})
+				with self.assertRaises(RuntimeError):
+					yahoo.fetch_detail('TSLA')
 
 	def test_missing_names_give_empty(self):
 		"""A detail without a name, or without a result, is empty."""
@@ -192,14 +198,48 @@ def test_missing_names_give_empty(self):
 
 
 class YahooChart(ProviderCase):
-	def test_real_chart_to_dataframe_and_basics(self):
-		"""A real v8 chart becomes a frame of dates and closes, and its basics are read."""
-		chart = sample('yahoo', 'sample_yahoo_v8_chart.json')
-		df = yahoo.chart_json_to_df(chart)
-		self.assertEqual(len(df), 2)
-		self.assertTrue({'Date', 'Close'} <= set(df.columns))
-		basics = yahoo.chart_json_to_stock_basics(chart)
-		self.assertEqual((basics['symbol'], basics['currency']), ('CRWD', 'USD'))
+	def test_real_chart_to_bars(self):
+		"""A real v8 chart becomes its daily bars: the day, the open and the close."""
+		rows = yahoo.bars(sample('yahoo', 'sample_yahoo_v8_chart.json'))
+		self.assertEqual(len(rows), 2)
+		self.assertTrue(all(isinstance(day, datetime.date) and close for day, _, close in rows))
+
+	def test_a_bar_is_dated_in_its_exchanges_time_zone(self):
+		"""An ASX bar, stamped at Sydney's open (23:00 UTC the day before), is dated the day it traded: it used to be
+		dated the day before, so N-day changes and chart dates slipped a session."""
+		open_sydney = datetime.datetime(2026, 1, 5, 23, 0, tzinfo=datetime.timezone.utc) # 10:00 on 6 Jan in Sydney
+		chart = {'chart': {'result': [{'meta': {'exchangeTimezoneName': 'Australia/Sydney'},
+			'timestamp': [int(open_sydney.timestamp())], 'indicators': {'quote': [{'open': [1.0], 'close': [2.0]}]}}]}}
+		self.assertEqual(yahoo.bars(chart), [(datetime.date(2026, 1, 6), 1.0, 2.0)])
+
+	def test_a_refused_crumb_is_renewed_once(self):
+		"""Yahoo refusing the crumb (401) gets a new cookie and crumb, and the request is asked again, once."""
+		renewed = []
+		fresh = lambda renew=False: renewed.append(renew) or ('A=B', 'new' if renew else 'old')
+		self.patch(yahoo, 'credentials', fresh)
+		getter = self.http(yahoo, Resp(status=401), Resp(status=401), Resp(sample('yahoo', 'sample_yahoo_v7.json')))
+		self.assertIn('AAPL', yahoo.fetch(['AAPL']))
+		crumbs = [url.rsplit('crumb=', 1)[1] for url in getter.calls]
+		self.assertEqual((renewed, crumbs), ([False, True], ['old', 'old', 'new']))
+		self.http(yahoo, Resp(status=401))
+		with self.assertRaises(RuntimeError): # a new crumb refused too: an error, not a loop
+			yahoo.fetch(['MSFT'])
+
+	def test_the_credentials_are_one_file_renewed_together(self):
+		"""The cookie and the crumb Yahoo issued for it are cached together and renewed together; no cookie, or no
+		crumb, is an error, where a stale fallback pair turned it into a refusal later."""
+		self.patch(yahoo, 'credentials', REAL_CREDENTIALS)
+		cookie = Resp(None, status=404)
+		cookie.headers = {'Set-Cookie': 'A3=d=AQAB; Max-Age=31557600; Domain=.yahoo.com; Path=/; Secure'}
+		getter = self.http(yahoo, cookie, Resp(None, text='crumb1'))
+		self.assertEqual(yahoo.credentials(), ('A3=d=AQAB', 'crumb1'))
+		self.assertEqual(yahoo.credentials(), ('A3=d=AQAB', 'crumb1')) # from the cache now
+		self.assertEqual(len(getter.calls), 2)
+		nothing = Resp(None, status=404)
+		nothing.headers = {}
+		self.http(yahoo, nothing)
+		with self.assertRaisesRegex(RuntimeError, 'no cookie'):
+			yahoo.credentials(renew=True)
 
 	def test_fetch_chart_json_failures_raise(self):
 		"""A chart Yahoo will not give is a RuntimeError."""
@@ -220,12 +260,25 @@ def chart(closes, failing=()):
 		return {'chart': {'result': [{'meta': meta, 'timestamp': stamps, 'indicators': {'quote': [quote]}}], 'error': None}}
 
 	def test_1d_is_the_change_since_the_previous_close(self):
-		"""1D is the change since the previous close, and 5D the week's."""
+		"""1D is the change since the previous close, and 5D the week's; with a quote, its price and today's change
+		count, the chart's being up to cache_seconds old."""
 		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
 		self.http(yahoo, Resp(self.chart([90] * 30 + [100, 105, 110, 115, 120, 116.4]))) # up 20% in a week, down 3% today
-		table, image = yahoo.price_history('T1')
+		table, line = yahoo.price_history('T1')
 		self.assertEqual(table['1D'], -3)
 		self.assertGreater(table['5D'], 10) # the week, not the day
+		table, line = yahoo.price_history('T1', quote={'regularMarketPrice': 125.0, 'percent_change': 7.4})
+		self.assertEqual((table['1D'], line.price), (7, 125.0))
+
+	def test_a_record_is_from_the_whole_history(self):
+		"""historic_high reads the whole history (the chart from period1=0), its first day aside: records seeded
+		from 10 years were no all-time records."""
+		urls = []
+		chart = self.chart([5.0, 50.0, 10.0, 20.0])
+		self.patch(yahoo, 'historic_high', REAL[('lib.yahoo', 'historic_high')])
+		self.patch(yahoo.net, 'get', lambda url, **kwargs: urls.append(url) or Resp(chart))
+		self.assertEqual(yahoo.historic_high('T1'), (50.0, 10.0)) # 5.0, the first day, aside
+		self.assertIn('period1=0&', urls[0])
 
 	def test_a_symbol_yahoo_does_not_know_is_no_data(self):
 		"""Yahoo's 404 for a symbol it does not know is NoData, an answer; any other 404 is an error, not an answer."""
@@ -253,7 +306,8 @@ def test_one_failed_ticker_does_not_take_the_report_down(self):
 		"""One ticker without a chart is left out of a price list, a benchmark without one is said to have none, and
 		only a report on that one ticker fails, in one line."""
 		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
-		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts}) # quotes as canned in the base fixture, charts over HTTP
+		quotes = lambda ts: {t: market_entry(t) | {'regularMarketPrice': 110.0} for t in ts} # charts over HTTP
+		self.patch(yahoo, 'fetch', quotes)
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t)})
 		self.patch(sharesight, 'get_portfolios', lambda: {}) # no Sharesight figures: every ticker needs its history
 		good = Resp(self.chart([100.0] * 20 + [110.0]))
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 11be7383..149a631a 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -360,7 +360,16 @@ def test_milestone_records_a_ticker_only_with_its_new_values(self):
 		self.sent.clear()
 		milestone.run()
 		self.assertNotIn('Record high', str(self.sent))
-		self.assertEqual(state.load('finbot_milestone.json')['T1'], {'record_high': 150.0, 'record_low': 50.0}) # all-time
+		self.assertEqual(state.load('finbot_milestone.json')['T1'], {'all_time_high': 150.0, 'all_time_low': 50.0})
+
+	def test_a_record_kept_from_ten_years_is_seeded_again_from_the_whole_history(self):
+		"""Records kept before they came from the whole history (record_high, from 10 years) are seeded again: a stock
+		above its 10-year high but below its all-time one posted a false record."""
+		state.save('finbot_milestone.json', {'T1': {'record_high': 110.0, 'record_low': 90.0}})
+		self.patch(yahoo, 'historic_high', lambda t: (150.0, 50.0))
+		milestone.run() # the 52-week range (120/80) was past the old record, but not the whole history's
+		self.assertNotIn('Co T1 ', str(self.sent))
+		self.assertEqual(state.load('finbot_milestone.json')['T1'], {'all_time_high': 150.0, 'all_time_low': 50.0})
 
 	def test_milestone_records_are_all_time(self):
 		"""A new 52-week high below the all-time high is no record. The record used to become the 52-week high after the first

From 5d04880c2e3df3bdcdf07f3e611c65b4183af773 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 09:45:27 +0000
Subject: [PATCH 61/81] Sharesight: one request function with every check;
 gains for the reports; one warming pass

- Every request goes through one function: one at a time, retried while Sharesight says too many are in flight (only
  the performance report retried before), and checked: an error answer says Sharesight's error (an OAuth refusal,
  which has no error_code, was a KeyError on it), an answer that is not 200 or lacks what was asked for says so
  (it was a KeyError later).
- One cache-or-fetch, cache.cached, serves Sharesight and Yahoo. A trade's portfolio name is added after the cache,
  not written into it (and out of the sample, which came from that cache).
- Portfolio.gain(days) and Portfolio.gains(days) give the portfolio's and each holding's capital gain, so the price
  and performance reports no longer dig through Sharesight's report.
- The cache warmer refreshes each file at 90% of its lifetime and says when the next is due in the same pass;
  keep_warm lost the stop event only the tests used, and the fetchers are private.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md                                     |   5 +-
 .../finbot_sharesight_trades_684141_1.json    |   6 +-
 lib/cache.py                                  |  12 +
 lib/reports/performance.py                    |  19 +-
 lib/reports/price.py                          |   5 +-
 lib/sharesight.py                             | 326 ++++++++----------
 lib/yahoo.py                                  |  17 +-
 tests/common.py                               |  12 +-
 tests/test_edges.py                           |   8 +-
 tests/test_providers.py                       | 231 +++++++------
 tests/test_reports.py                         |   7 +-
 11 files changed, 301 insertions(+), 347 deletions(-)

diff --git a/CLAUDE.md b/CLAUDE.md
index f2f8c67b..2d258927 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -57,9 +57,10 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   `lib/formats.py`: names, units, dates, percents in words. `lib/links.py`: stock links. `lib/render.py`: every chart,
   `Chart` specs and `draw()`.
 - `lib/yahoo.py lib/sharesight.py lib/shortman.py lib/simplywallst.py`: data sources. Yahoo's and Sharesight's answers
-  are cached in `DATA_DIR/cache` as sent, not parsed, so a parser fix needs no flush (`yahoo._cached`).
+  are cached in `DATA_DIR/cache` as sent, not parsed, so a parser fix needs no flush (`cache.cached`).
   `yahoo.symbol(typed)` is Yahoo's symbol for a ticker as written, tried on the local exchange (`exchange_suffix`)
-  when Yahoo does not know it bare: every ticker a chat gives goes through it. `sharesight.portfolios(name=None)` gives `Portfolio` objects: `holdings()`, `tickers()`, `performance(days)`,
+  when Yahoo does not know it bare: every ticker a chat gives goes through it. `sharesight.portfolios(name=None)` gives
+  `Portfolio` objects: `holdings()`, `tickers()`, `gain(days)` (the portfolio's), `gains(days)` (each holding's),
   `trades(days)`.
 - `lib/cache.py`: cache files (any may vanish at any time). `lib/state.py`: state files in `DATA_DIR`, which must last;
   change one with `with state.edit(name, default) as data:`, never load-then-save. `lib/watchlist.py`: the watchlist.
diff --git a/doc/sharesight/finbot_sharesight_trades_684141_1.json b/doc/sharesight/finbot_sharesight_trades_684141_1.json
index 7d3c02a5..2c2200ad 100644
--- a/doc/sharesight/finbot_sharesight_trades_684141_1.json
+++ b/doc/sharesight/finbot_sharesight_trades_684141_1.json
@@ -26,8 +26,7 @@
 			"confirmed": true,
 			"links": {
 				"portfolio": "https://api.sharesight.com/api/v2/portfolios/684141"
-			},
-			"portfolio": "RobSMSF"
+			}
 		},
 		{
 			"id": 104446458,
@@ -55,8 +54,7 @@
 			"confirmed": true,
 			"links": {
 				"portfolio": "https://api.sharesight.com/api/v2/portfolios/684141"
-			},
-			"portfolio": "RobSMSF"
+			}
 		}
 	],
 	"api_transaction": {
diff --git a/lib/cache.py b/lib/cache.py
index 98a15895..5ff8dc2f 100644
--- a/lib/cache.py
+++ b/lib/cache.py
@@ -69,6 +69,18 @@ def load(name):
 		return None
 
 
+def cached(name, seconds, fetch):
+	"""fetch()'s answer (JSON), as sent: from the cache file `name` while it is under `seconds` old (cache_seconds when
+	None) and the cache preference is on, else fetched and written there unless empty. Answers are cached as sent, not
+	parsed, so that a parser fix needs no flush."""
+	if config.prefs.cache and (data := read(name, seconds)):
+		return data
+	data = fetch()
+	if data and config.prefs.cache:
+		write(name, data)
+	return data
+
+
 def write(name, data):
 	"""Write a cache file's JSON. See write_file."""
 	where = path(name)
diff --git a/lib/reports/performance.py b/lib/reports/performance.py
index 9197fdff..99316878 100644
--- a/lib/reports/performance.py
+++ b/lib/reports/performance.py
@@ -13,14 +13,9 @@
 BENCHMARKS = (('SPY', 'NYSEARCA', 'S&P 500'), ('QQQ', 'NasdaqGM', 'NASDAQ 100'))
 
 def gather(past_days, portfolio_select=None):
-	"""{portfolio id: (its name, Sharesight's performance report over past_days)} of every portfolio, or of the one named (in any
-	case). Raises sharesight.PortfolioNotFound for a name that is none of them, and RuntimeError when there is no report."""
-	performance = {portfolio.id: (portfolio.name, portfolio.performance(past_days))
-		for portfolio in sharesight.portfolios(portfolio_select)}
-	if not performance:
-		print("Error: no Sharesight data found", file=sys.stderr)
-		raise RuntimeError("no Sharesight data found")
-	return performance
+	"""[(portfolio, its capital gain over past_days in percent)] of every portfolio, or of the one named (in any case).
+	Raises sharesight.PortfolioNotFound for a name that is none of them."""
+	return [(portfolio, portfolio.gain(past_days)) for portfolio in sharesight.portfolios(portfolio_select)]
 
 def benchmark(ticker, market, text, past_days, service):
 	"""(report line, percent) of a benchmark over past_days; percent is None, and the line says so, when it has no history."""
@@ -38,10 +33,10 @@ def report(service, past_days, portfolio_select=None, performance=None):
 	if performance is None:
 		performance = gather(past_days, portfolio_select)
 	rows = [] # (percent or None, line, chart label), one list for the caption and the chart, best first
-	for portfolio_id, (name, data) in performance.items(): # the name from the portfolio list: one with no holdings names none
-		link = service.link("https://portfolio.sharesight.com/portfolios/" + str(portfolio_id), name)
-		percent = float(data['report']['capital_gain_percent'])
-		rows.append((percent, f"{formats.get_emoji(percent)} {link} {formats.signed_percent(percent, 2)}", name))
+	for portfolio, percent in performance:
+		link = service.link(f"https://portfolio.sharesight.com/portfolios/{portfolio.id}", portfolio.name)
+		line = f"{formats.get_emoji(percent)} {link} {formats.signed_percent(percent, 2)}"
+		rows.append((percent, line, portfolio.name))
 	for ticker, market, text in BENCHMARKS:
 		line, percent = benchmark(ticker, market, text, past_days, service)
 		rows.append((None if percent is None else float(percent), line, text))
diff --git a/lib/reports/price.py b/lib/reports/price.py
index affc1189..c245cc3d 100644
--- a/lib/reports/price.py
+++ b/lib/reports/price.py
@@ -31,10 +31,9 @@ def gather(specific_stock=None, days=None):
 	# Sharesight reports a holding's performance only for the time it was held: held for 1Y and asked for 5Y, it gives 1Y
 	if not specific_stock or config.prefs.performance_use_sharesight:
 		for portfolio in sharesight.portfolios():
-			for holding in portfolio.performance(days)['report']['holdings']:
-				ticker = markets.transform_to_yahoo(holding['instrument']['code'], holding['instrument']['market_code'])
+			for ticker, gain in portfolio.gains(days).items():
 				if ticker in market_data:
-					market_data[ticker]['percent_change_period'] = float(holding['capital_gain_percent'])
+					market_data[ticker]['percent_change_period'] = gain
 				elif ticker in tickers:
 					print("Notice: price.py:", ticker, "has no data", file=sys.stderr)
 	if config.prefs.performance_use_sharesight:
diff --git a/lib/sharesight.py b/lib/sharesight.py
index 6a287dc0..13dc261a 100644
--- a/lib/sharesight.py
+++ b/lib/sharesight.py
@@ -1,102 +1,110 @@
+"""Sharesight's API: the portfolios to report on (Portfolio), with their holdings, gains and trades. Answers are cached
+as sent, and requests go one at a time."""
 import dataclasses
 import datetime
 import sys
 import threading
 import time
-import requests
+
 from lib import answer
-from lib import config
 from lib import cache
-from lib import net
-from lib import formats
+from lib import config
 from lib import markets
+from lib import net
 
-MAX_CONCURRENT = 1 # Sharesight rejects parallel requests beyond a few per account: allow one at a time across all chats and threads
+API = 'https://api.sharesight.com/'
+MAX_CONCURRENT = 1 # Sharesight refuses parallel requests beyond a few per account: one at a time, whoever asks
 _request_slots = threading.BoundedSemaphore(MAX_CONCURRENT)
+TOKEN = 'finbot_sharesight_token.json' # good until the expiry it carries
+PORTFOLIOS_CACHE = 'finbot_sharesight_portfolios.json' # good for the cache_seconds preference
+
 
-def _get(url, **kwargs):
-	"""net.get, waiting its turn: one Sharesight request at a time (MAX_CONCURRENT)."""
-	with _request_slots:
-		return net.get(url, **kwargs)
-
-def _post(url, **kwargs):
-	"""net.post, waiting its turn like _get."""
-	with _request_slots:
-		return net.post(url, **kwargs)
-
-def _request(send, url, what, **kwargs):
-	"""Send a request to Sharesight and return the JSON answer. A request that fails raises RuntimeError naming `what`."""
-	try:
-		r = send(url, **kwargs)
-	except Exception as e:
-		print("Error", net.redact(e), url, file=sys.stderr)
-		raise RuntimeError(f"Sharesight {what} request failed: {e}")
-	if r.status_code != 200:
-		print(r.status_code, "error", url, file=sys.stderr)
-	try:
-		return r.json()
-	except ValueError: # an HTML error page, say: "Expecting value: line 1 column 1" would say nothing
-		raise RuntimeError(f"Sharesight {what}: a {r.status_code} answer that is not JSON") from None
-
-def _check(data, what):
-	"""Raise RuntimeError with Sharesight's error, when it answered with one."""
-	if 'error' in data:
-		print("Sharesight error:", data['error_code'], data['error'], file=sys.stderr)
-		raise RuntimeError(f"Sharesight {what} error: {data['error']}")
-
-class BearerAuth(requests.auth.AuthBase):
-	def __init__(self, token):
-		"""token: the access token."""
-		self.token = token
-	def __call__(self, r):
-		"""Add the token to a request's Authorization header."""
-		r.headers["Authorization"] = "Bearer " + self.token
-		return r
-
-def get_token():
-	"""An access token for the Sharesight API: the cached one while it has more than 20 seconds left, else a new one
+def _request(path, what, expect, **kwargs):
+	"""Sharesight's answer to a request (a POST with data=, else a GET), checked to hold `expect`: one request at a
+	time (MAX_CONCURRENT), tried again while Sharesight says too many are in flight (five tries in all). Raises
+	RuntimeError naming `what` when Sharesight cannot be reached, or answers with an error, or without `expect`."""
+	send = net.post if 'data' in kwargs else net.get
+	for attempt in range(5):
+		with _request_slots:
+			try:
+				r = send(API + path, **kwargs)
+			except Exception as e:
+				raise RuntimeError(f"Sharesight {what}: {net.redact(e)}") from e
+		try:
+			data = r.json()
+		except ValueError: # an HTML error page, say
+			data = None
+		error = data.get('error') if isinstance(data, dict) else None
+		if not (error and 'parallel' in str(error).lower() and attempt < 4):
+			break
+		print("Sharesight busy:", error, "- retrying", file=sys.stderr)
+		time.sleep(1 + attempt)
+	if error: # an OAuth error describes itself too
+		why = ': '.join(str(data[key]) for key in ('error', 'error_description') if data.get(key))
+	elif not isinstance(data, dict):
+		why = f"a {r.status_code} answer that is not JSON"
+	elif r.status_code != 200 or expect not in data:
+		why = f"a {r.status_code} answer without {expect}"
+	else:
+		return data
+	raise RuntimeError(f"Sharesight {what}: {why}")
+
+
+def _token():
+	"""An access token for Sharesight's API: the cached one while it has more than 20 seconds left, else a new one
 	(cached while the cache preference is on)."""
+	cached = cache.load(TOKEN) if config.prefs.cache else None
+	if cached and cached['created_at'] + cached['expires_in'] - time.time() > 20:
+		return cached['access_token']
+	data = _request('oauth2/token', 'auth', 'access_token', data=config.env.sharesight_auth)
 	if config.prefs.cache:
-		cacheFile = "finbot_sharesight_token.json"
-		cached = cache.load(cacheFile)
-		if cached:
-			cache_expiry = datetime.datetime.fromtimestamp(cached['created_at'] + cached['expires_in'])
-			ttl = cache_expiry - datetime.datetime.now()
-			if ttl > datetime.timedelta(seconds=20):
-				print("cache hit:", cacheFile, "TTL:", formats.td_to_human(ttl), file=sys.stderr) if config.prefs.debug else None
-				return cached['access_token']
-			else:
-				print("cache miss:", cacheFile, file=sys.stderr) if config.prefs.debug else None
-
-	data = _request(_post, "https://api.sharesight.com/oauth2/token", 'auth', data=config.env.sharesight_auth)
-	_check(data, 'auth')
-	if config.prefs.cache and 'access_token' in data:
-		cache.write(cacheFile, data)
-	print("Got a Sharesight token", file=sys.stderr) if config.prefs.debug else None
+		cache.write(TOKEN, data)
 	return data['access_token']
 
-PORTFOLIOS_CACHE = "finbot_sharesight_portfolios.json" # ttl: the cache_seconds preference
+
+def _api(path, what, expect):
+	"""Sharesight's answer to an API query (path: under api/, with its query string) with the access token; see
+	_request."""
+	return _request('api/' + path, what, expect, headers={'Authorization': 'Bearer ' + _token()})
+
+
+def _portfolios(seconds=None):
+	"""Sharesight's list of every portfolio, from the cache while under `seconds` old (cache_seconds when None)."""
+	fetch = lambda: _api('v3/portfolios', 'portfolios', 'portfolios')
+	return cache.cached(PORTFOLIOS_CACHE, seconds, fetch)['portfolios']
+
 
 def performance_cache_file(portfolio_id, days):
 	"""The cache file of a portfolio's performance report over `days`."""
-	return "finbot_sharesight_performance_" + str(portfolio_id) + "_" + str(days) + '.json'
-
-def get_portfolios(refresh=False):
-	"""refresh=True skips the cache read and fetches again (the cache is rewritten)."""
-	data = cache.read(PORTFOLIOS_CACHE, config.prefs.cache_seconds) if config.prefs.cache and not refresh else None
-	if not data:
-		token = get_token()
-		data = _request(_get, "https://api.sharesight.com/api/v3/portfolios", 'portfolios', headers={'Content-type': 'application/json'}, auth=BearerAuth(token))
-		_check(data, 'portfolios')
-		if config.prefs.cache and 'portfolios' in data:
-			cache.write(PORTFOLIOS_CACHE, data)
-	include, exclude = config.prefs.include_portfolios, config.prefs.exclude_portfolios
-	portfolio_dict = {portfolio['name']: portfolio['id'] for portfolio in data['portfolios']
-		if str(portfolio['id']) not in exclude and (not include or str(portfolio['id']) in include)}
-	if not portfolio_dict:
-		print("No portfolios found. Exiting.", file=sys.stderr)
-		raise RuntimeError("No Sharesight portfolios found")
-	return portfolio_dict
+	return f"finbot_sharesight_performance_{portfolio_id}_{days}.json"
+
+
+def _since(days):
+	"""The date `days` ago, as Sharesight takes it: 2023-04-25."""
+	return (datetime.date.today() - datetime.timedelta(days=days)).isoformat()
+
+
+def _performance(portfolio_id, days=None, seconds=None):
+	"""Sharesight's performance report of a portfolio since `days` ago (0: today's holdings; past_days when None), from
+	the cache while under `seconds` old (the performance_cache_seconds preference when None)."""
+	if days is None:
+		days = config.prefs.past_days
+	query = f'v3/portfolios/{portfolio_id}/performance?grouping=ungrouped&start_date={_since(days)}'
+	fetch = lambda: _api(query, f'performance of portfolio {portfolio_id}', 'report')
+	if seconds is None:
+		seconds = config.prefs.performance_cache_seconds
+	return cache.cached(performance_cache_file(portfolio_id, days), seconds, fetch)['report']
+
+
+def _trades(portfolio_id, days=None):
+	"""Sharesight's trades in a portfolio over the past `days` (past_days when None), from the cache while under 5
+	minutes old."""
+	if days is None:
+		days = config.prefs.past_days
+	query = f'v2/portfolios/{portfolio_id}/trades.json?start_date={_since(days)}'
+	fetch = lambda: _api(query, 'trades', 'trades')
+	return cache.cached(f"finbot_sharesight_trades_{portfolio_id}_{days}.json", 299, fetch)['trades']
+
 
 class PortfolioNotFound(answer.Answer, LookupError):
 	"""No portfolio is called what was asked for: the answer names it, and the portfolios there are."""
@@ -104,37 +112,48 @@ def __init__(self, name, names):
 		"""name: what was asked for; names: the portfolios there are."""
 		super().__init__(f"{name} portfolio not found. I only know about:", *names)
 
+
 @dataclasses.dataclass(frozen=True)
 class Portfolio:
-	"""A Sharesight portfolio. Its methods ask Sharesight through the module functions (cached, one request at once)."""
+	"""A Sharesight portfolio. Its methods ask Sharesight (cached, one request at a time)."""
 	name: str # as Sharesight spells it
 	id: int
 
-	def performance(self, days=None):
-		"""Sharesight's performance report over the past `days` (past_days when None); days=0 gives today's holdings."""
-		return get_performance(self.id, config.prefs.past_days if days is None else days)
+	def gain(self, days=None):
+		"""Its capital gain over the past `days` (the past_days preference when None), in percent."""
+		return float(_performance(self.id, days)['capital_gain_percent'])
+
+	def gains(self, days=None):
+		"""{Yahoo ticker: capital gain in percent} of what it holds, over the past `days` (the past_days preference when
+		None), or over the time it was held where that is shorter."""
+		report = _performance(self.id, days)
+		return {markets.transform_to_yahoo(each['instrument']['code'], each['instrument']['market_code']):
+			float(each['capital_gain_percent']) for each in report['holdings']}
 
 	def holdings(self):
 		"""{'CODE:MARKET': Sharesight's instrument, with the holding's id and value added} of what it holds."""
-		holdings = {}
-		for item in self.performance(0)['report']['holdings']:
-			instrument = item['instrument']
-			key = instrument['code'] + ':' + instrument['market_code']
-			holdings[key] = instrument | {'holding_id': item['id'], 'value': item.get('value')}
-		return holdings
+		return {each['instrument']['code'] + ':' + each['instrument']['market_code']:
+			each['instrument'] | {'holding_id': each['id'], 'value': each.get('value')}
+			for each in _performance(self.id, 0)['holdings']}
 
 	def tickers(self):
 		"""What it holds, as Yahoo tickers, sorted."""
 		return sorted({markets.transform_to_yahoo(each['code'], each['market_code']) for each in self.holdings().values()})
 
 	def trades(self, days=None):
-		"""Its trades of the past `days` (the past_days preference when None)."""
-		return get_trades(self.name, self.id, days)
+		"""Its trades of the past `days` (the past_days preference when None), each with the portfolio's name added."""
+		return [trade | {'portfolio': self.name} for trade in _trades(self.id, days)]
 
-def portfolios(select=None):
+
+def portfolios(select=None, seconds=None):
 	"""The portfolios to report on (after include_portfolios and exclude_portfolios), or the one called `select`, in any
-	case. Raises PortfolioNotFound for a name that is none of them."""
-	found = [Portfolio(name, portfolio_id) for name, portfolio_id in get_portfolios().items()]
+	case; the list is cached for `seconds` (cache_seconds when None). Raises RuntimeError when there are none, and
+	PortfolioNotFound for a name that is none of them."""
+	include, exclude = config.prefs.include_portfolios, config.prefs.exclude_portfolios
+	found = [Portfolio(each['name'], each['id']) for each in _portfolios(seconds)
+		if str(each['id']) not in exclude and (not include or str(each['id']) in include)]
+	if not found:
+		raise RuntimeError("No Sharesight portfolios found")
 	if not select:
 		return found
 	chosen = [portfolio for portfolio in found if portfolio.name.lower() == select.lower()]
@@ -142,114 +161,41 @@ def portfolios(select=None):
 		raise PortfolioNotFound(select, [portfolio.name for portfolio in found])
 	return chosen
 
+
 def tickers():
 	"""Every portfolio's holdings as Yahoo tickers, sorted."""
 	return sorted({ticker for portfolio in portfolios() for ticker in portfolio.tickers()})
 
-def get_trades(portfolio_name, portfolio_id, days=None):
-	"""The portfolio's trades of the past `days` (the past_days preference when None), each tagged with its portfolio's
-	name; cached for 5 minutes. Raises RuntimeError when Sharesight cannot be reached or answers with an error."""
-	if days is None:
-		days = config.prefs.past_days
-	if config.prefs.cache:
-		cache_file = "finbot_sharesight_trades_" + str(portfolio_id) + "_" + str(days) + ".json"
-		cached = cache.read(cache_file, 299) # max freq 5 min
-		if cached:
-			return cached['trades']
-	start_date = datetime.datetime.now() - datetime.timedelta(days=days)
-	start_date = start_date.strftime('%Y-%m-%d') # 2022-08-20
-	token = get_token()
-	url = 'https://api.sharesight.com/api/v2/portfolios/'
-	url = url + str(portfolio_id) + '/trades.json' + '?start_date=' + start_date
-	data = _request(_get, url, 'trades', auth=BearerAuth(token))
-	_check(data, 'trades')
-	for trade in data['trades']:
-		trade['portfolio'] = portfolio_name # inject custom field
-	if config.prefs.cache and 'trades' in data:
-		cache.write(cache_file, data)
-	return data['trades']
-
-def get_performance(portfolio_id, days, seconds=None, refresh=False):
-	"""Sharesight's performance report of a portfolio since `days` ago (0: today's holdings), cached for `seconds` (the
-	performance_cache_seconds preference when None); refresh=True asks again. While Sharesight says too many requests
-	are in flight it retries, up to five tries. Raises RuntimeError on an error answer, or one without a report."""
-	if seconds is None:
-		seconds = config.prefs.performance_cache_seconds
-	start_date = datetime.datetime.now() - datetime.timedelta(days=days)
-	start_date = start_date.strftime('%Y-%m-%d') # 2023-04-25
-	cache_file = performance_cache_file(portfolio_id, days)
-	if config.prefs.cache and not refresh:
-		cached = cache.read(cache_file, seconds)
-		if cached:
-			return cached
-	token = get_token()
-	endpoint = 'https://api.sharesight.com/api/v3/portfolios/'
-	url = endpoint + str(portfolio_id) + '/performance?grouping=ungrouped&start_date=' + start_date
-	for attempt in range(5): # Sharesight allows only a few requests in flight at once; back off and retry when it says so
-		data = _request(_get, url, 'performance', auth=BearerAuth(token))
-		if 'error' in data and 'parallel' in str(data['error']).lower() and attempt < 4:
-			print("Sharesight busy:", data['error'], "- retrying", file=sys.stderr)
-			time.sleep(1 + attempt)
-			continue
-		break
-	_check(data, 'performance')
-	if 'report' not in data:
-		raise RuntimeError(f"Sharesight sent no performance report for portfolio {portfolio_id}")
-	if config.prefs.cache:
-		cache.write(cache_file, data)
-	return data
 
 def warm(margin=0.9):
-	"""Refresh the portfolio list and each portfolio's holdings report (days=0) once they reach `margin` of their cache lifetime,
-	so a request never has to wait on Sharesight. Returns (the number of fetches made, the portfolios): the caller can hand
-	the portfolios on instead of reading the cached list again."""
-	fetched = 0
-	age = cache.age(PORTFOLIOS_CACHE)
-	if age is None or age >= config.prefs.cache_seconds * margin:
-		portfolios = get_portfolios(refresh=True)
-		fetched += 1
-	else:
-		portfolios = get_portfolios()
-	for portfolio_name, portfolio_id in portfolios.items():
-		age = cache.age(performance_cache_file(portfolio_id, 0))
-		if age is None or age >= margin * config.prefs.performance_cache_seconds:
-			get_performance(portfolio_id, 0, refresh=True)
-			fetched += 1
-	return fetched, portfolios
-
-def next_due(margin=0.9, portfolios=None):
-	"""Seconds until the soonest warmed cache needs refreshing (0 if one already does). `portfolios` is the list warm() just
-	used; without it the cached list is read."""
-	waits = []
-	age = cache.age(PORTFOLIOS_CACHE)
-	waits.append(0 if age is None else margin * config.prefs.cache_seconds - age)
-	for portfolio_name, portfolio_id in (get_portfolios() if portfolios is None else portfolios).items():
-		age = cache.age(performance_cache_file(portfolio_id, 0))
-		waits.append(0 if age is None else margin * config.prefs.performance_cache_seconds - age)
-	return max(0, min(waits))
+	"""Refresh the portfolio list and each portfolio's holdings report (days=0) once they reach `margin` of their cache
+	lifetime, so that a request never waits on Sharesight. Returns the seconds until the next one is due."""
+	lifetimes = {PORTFOLIOS_CACHE: config.prefs.cache_seconds}
+	for portfolio in portfolios(seconds=margin * config.prefs.cache_seconds):
+		_performance(portfolio.id, 0, margin * config.prefs.performance_cache_seconds)
+		lifetimes[performance_cache_file(portfolio.id, 0)] = config.prefs.performance_cache_seconds
+	return max(0, min(margin * lifetime - (cache.age(name) or 0) for name, lifetime in lifetimes.items()))
+
 
 _wake = threading.Event() # set by wake(): the caches were flushed, refill them now
 
+
 def wake():
 	"""Have keep_warm refill the caches now instead of when they would next be due."""
 	_wake.set()
 
-def keep_warm(min_wait=5, retry=300, idle=60, stop=None):
-	"""Background loop: keep the portfolio list and holdings report cached, sleeping until the next one is due (90% of its TTL).
-	While the cache preference is off (a reload can switch it either way) it fetches nothing, since nothing would be written and
-	everything would always be due, and looks again every `idle` seconds. Never raises; after a failure it waits `retry` seconds."""
-	while not (stop and stop.is_set()):
-		if not config.prefs.cache:
-			wait = idle
-		else:
+
+def keep_warm(min_wait=5, retry=300, idle=60):
+	"""Background loop: keep the portfolio list and the holdings reports cached, sleeping until the next one is due
+	(warm). While the cache preference is off (a reload can switch it either way) it fetches nothing, since nothing
+	would be written, and looks again every `idle` seconds. Never raises; after a failure it waits `retry` seconds."""
+	while True:
+		wait = idle
+		if config.prefs.cache:
 			try:
-				_, portfolios = warm()
-				wait = max(min_wait, next_due(portfolios=portfolios))
+				wait = max(min_wait, warm())
 			except Exception as e:
 				print("Sharesight cache refresh failed:", e, file=sys.stderr)
 				wait = retry
-		if stop:
-			stop.wait(wait)
-		else:
-			_wake.wait(wait)
-			_wake.clear()
+		_wake.wait(wait)
+		_wake.clear()
diff --git a/lib/yahoo.py b/lib/yahoo.py
index fdee9643..f1efdd25 100644
--- a/lib/yahoo.py
+++ b/lib/yahoo.py
@@ -117,17 +117,6 @@ def _convert(convert, value):
 	except (TypeError, ValueError, OverflowError):
 		return None
 
-def _cached(name, seconds, get, *args):
-	"""get(*args), Yahoo's answer as sent: from the cache file `name` while it is under `seconds` old (cache_seconds
-	when None) and the cache preference is on, else fetched and written there unless empty. Cached raw, not parsed, so
-	that a parser fix needs no flush."""
-	if config.prefs.cache and (cached := cache.read(name, seconds)):
-		return cached
-	data = get(*args)
-	if data and config.prefs.cache:
-		cache.write(name, data)
-	return data
-
 QUOTE_FIELDS = ( # (key, Yahoo's name, conversion) for the fields of a v7 quote; one Yahoo leaves out, sends as null or garbles is left out
 	('percent_change', 'regularMarketChangePercent', _round2),
 	('dividend', 'trailingAnnualDividendRate', lambda v: round(float(v), 1)),
@@ -153,7 +142,7 @@ def fetch(tickers):
 		return {}
 	tickers = sorted(set(tickers)) # de-dupe
 	name = "finbot_yahoo_quote_" + hashlib.sha256("_".join(tickers).encode()).hexdigest() + '.json'
-	data = _cached(name, 300, _quote_answer, tickers)
+	data = cache.cached(name, 300, lambda: _quote_answer(tickers))
 	if data is None:
 		print("Yahoo has no quotes for", ', '.join(tickers), file=sys.stderr)
 		return {} # callers iterate the result
@@ -225,7 +214,7 @@ def fetch_detail(ticker, seconds=None):
 	"""One ticker's quoteSummary: {ticker: {...}}, cached `seconds` (or cache_seconds); {} for a symbol Yahoo does not
 	know, or without a name or price. Raises RuntimeError when Yahoo cannot be reached or refuses."""
 	now = datetime.datetime.now()
-	data = _cached("finbot_yahoo_detail_" + ticker + '.json', seconds, _detail_answer, ticker)
+	data = cache.cached("finbot_yahoo_detail_" + ticker + '.json', seconds, lambda: _detail_answer(ticker))
 	result = dig(data, 'quoteSummary', 'result', 0)
 	if result is None: # Yahoo answers 404 Not Found, or result=null, for a symbol it does not know
 		print(ticker, "is unknown to Yahoo", file=sys.stderr)
@@ -454,7 +443,7 @@ def fetch_chart_json(ticker, days=3665, seconds=None, full=False):
 	"""Yahoo's daily chart for the last `days`, or with full=True its entire history (cached separately). Raises NoData
 	for a symbol Yahoo does not know, RuntimeError when Yahoo cannot be reached or refuses."""
 	name = "finbot_yahoo_history_" + ticker + ("_max" if full else "") + ".json"
-	return _cached(name, seconds, _chart_answer, ticker, days, full)
+	return cache.cached(name, seconds, lambda: _chart_answer(ticker, days, full))
 
 def _chart_answer(ticker, days, full):
 	"""The v8 chart answer: daily bars of the last `days`, or all with full=True. Raises as fetch_chart_json does."""
diff --git a/tests/common.py b/tests/common.py
index d1ed097c..70fc8dd3 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -16,7 +16,7 @@
 
 REAL = {(m.__name__, n): getattr(m, n) for m, n in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'),
 	(yahoo, 'price_history'), (yahoo, 'historic_high'),
-	(sharesight, 'get_portfolios'), (sharesight, 'get_trades'), (sharesight, 'get_performance'))}
+	(sharesight, '_portfolios'), (sharesight, '_trades'), (sharesight, '_performance'))}
 REAL_GET_HOLDINGS = watchlist.with_holdings
 REAL_TEXT = transports.Transport.text # FinbotCase replaces it with a recorder; tests of the real thing put it back
 REAL_CHART = transports.Transport.chart # likewise
@@ -104,11 +104,11 @@ def setUp(self):
 		self.patch(shortman, 'fetch', lambda md: md)
 		held = [{'id': i, 'instrument': {'code': t, 'market_code': 'NASDAQ'}, 'portfolio': {'name': 'Rob'},
 					 'capital_gain_percent': str(i * 4 - 40), 'value': '1000'} for i, t in enumerate(TICKERS)]
-		self.patch(sharesight, 'get_portfolios', lambda: {'Rob': 1})
-		self.patch(sharesight, 'get_performance', lambda pid, d, *a: {'report': {
-			'holdings': held, 'currency_gain_percent': '0', 'capital_gain_percent': '2.5', 'total_gain_percent': '0'}})
-		self.patch(sharesight, 'get_trades', lambda name, pid, days=7: [{
-			'id': 1, 'transaction_type': 'BUY', 'symbol': 'T1', 'market': 'NASDAQ', 'portfolio': name,
+		self.patch(sharesight, '_portfolios', lambda seconds=None: [{'name': 'Rob', 'id': 1}])
+		self.patch(sharesight, '_performance', lambda pid, days=None, seconds=None: {
+			'holdings': held, 'currency_gain_percent': '0', 'capital_gain_percent': '2.5', 'total_gain_percent': '0'})
+		self.patch(sharesight, '_trades', lambda pid, days=None: [{
+			'id': 1, 'transaction_type': 'BUY', 'symbol': 'T1', 'market': 'NASDAQ',
 			'transaction_date': time.strftime('%Y-%m-%d'), 'quantity': '10', 'price': '5.0', 'value': '50', 'brokerage_currency_code': 'USD', 'holding_id': 7}])
 		self.typing = [] # every chat a test starts a typing notice in; none may still be sending it when the test ends
 		noted = lambda chat: self.typing.append(chat) or REAL_START_TYPING(chat)
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 382dfc65..bb488548 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -48,8 +48,8 @@ def test_sharesight_down_is_reported_once(self):
 		def down(*a, **k):
 			"""Fail like Sharesight when it errors."""
 			raise RuntimeError('Sharesight performance error: boom')
-		self.patch(sharesight, 'get_performance', down)
-		self.patch(sharesight, 'get_portfolios', down)
+		self.patch(sharesight, '_performance', down)
+		self.patch(sharesight, '_portfolios', down)
 		self.sent.clear(); self.errors.clear()
 		worker.process_request('telegram', '55', '@u', '.performance 7d', BOT, 'U')
 		self.assertEqual(len(self.errors), 1)
@@ -60,7 +60,7 @@ def test_the_typing_notice_stops_before_any_message_goes_out(self):
 		def down(*a, **k):
 			"""Fail like Sharesight when it errors."""
 			raise RuntimeError('Sharesight performance error: boom')
-		self.patch(sharesight, 'get_performance', down)
+		self.patch(sharesight, '_performance', down)
 		self.patch(delivery, 'report_error', REAL_REPORT_ERROR)
 		self.patch(transports.Transport, 'text', REAL_TEXT)
 		self.patch(transports.Transport, 'chart', REAL_CHART)
@@ -183,7 +183,7 @@ def test_commands_with_nothing_to_show_say_so(self):
 
 	def test_trades_with_no_buy_or_sell_still_get_an_answer(self):
 		""".trades with only dividends says there were no trades (it used to say nothing)."""
-		self.patch(sharesight, 'get_trades', lambda name, pid, days=7: [{'id': 2, 'transaction_type': 'DIVIDEND', 'portfolio': name}])
+		self.patch(sharesight, '_trades', lambda pid, days=None: [{'id': 2, 'transaction_type': 'DIVIDEND'}])
 		self.assertIn('No trades', str(self.command('.trades 7d'))) # it used to say nothing at all
 
 
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 28436e52..2c7f92ae 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -41,6 +41,7 @@ def get(url, **kwargs):
 
 
 REAL_CREDENTIALS = yahoo.credentials
+REAL_TOKEN = sharesight._token
 
 
 class ProviderCase(FinbotCase):
@@ -48,11 +49,11 @@ def setUp(self):
 		"""Real parsers instead of the fixture's canned data, with a fixed cookie, crumb and token, and no waiting."""
 		super().setUp()
 		# these tests exercise the real parsers, so undo the canned data from the base fixture
-		for module, name in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'), (sharesight, 'get_portfolios'),
-							 (sharesight, 'get_trades'), (sharesight, 'get_performance')):
+		for module, name in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'),
+				(sharesight, '_portfolios'), (sharesight, '_trades'), (sharesight, '_performance')):
 			self.patch(module, name, REAL[(module.__name__, name)])
 		self.patch(yahoo, 'credentials', lambda renew=False: ('A=B', 'crumb'))
-		self.patch(sharesight, 'get_token', lambda: 'token')
+		self.patch(sharesight, '_token', lambda: 'token')
 		self.patch(sharesight.time, 'sleep', lambda s: None)
 
 	def http(self, module, *responses):
@@ -309,7 +310,9 @@ def test_one_failed_ticker_does_not_take_the_report_down(self):
 		quotes = lambda ts: {t: market_entry(t) | {'regularMarketPrice': 110.0} for t in ts} # charts over HTTP
 		self.patch(yahoo, 'fetch', quotes)
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t)})
-		self.patch(sharesight, 'get_portfolios', lambda: {}) # no Sharesight figures: every ticker needs its history
+		self.patch(sharesight, '_portfolios', lambda seconds=None: [{'name': 'Rob', 'id': 1}])
+		self.patch(sharesight, '_performance', lambda pid, days=None, seconds=None: # every ticker needs its history
+			{'holdings': [], 'capital_gain_percent': '2.5'})
 		good = Resp(self.chart([100.0] * 20 + [110.0]))
 		def get(url, **kwargs):
 			"""A chart for every ticker but T3 and SPY, which get 404."""
@@ -324,9 +327,6 @@ def slack(text):
 		price_list = slack('.price 7d') # T3 is left out, the rest are listed
 		self.assertNotIn('Co T3 (', price_list)
 		self.assertIn('Co T4 (', price_list)
-		self.patch(sharesight, 'get_portfolios', lambda: {'Rob': 1})
-		self.patch(sharesight, 'get_performance', lambda pid, d, *a: {'report': {'holdings': [{'portfolio': {'name': 'Rob'}}],
-			'currency_gain_percent': '0', 'capital_gain_percent': '2.5', 'total_gain_percent': '0'}})
 		report = slack('.performance 7d') # the S&P 500 benchmark has no history
 		self.assertIn('⚠️ S&P 500: no price history', report)
 		self.assertIn('NASDAQ 100', report)
@@ -364,23 +364,25 @@ def test_price_series_raises_on_failure_and_on_empty_history(self):
 			yahoo.price_series('EMPTY')
 
 
+def portfolios():
+	"""{name: id} of the portfolios to report on."""
+	return {portfolio.name: portfolio.id for portfolio in sharesight.portfolios()}
+
+
 class SharesightApi(ProviderCase):
 	def test_portfolios_from_real_response(self):
 		"""The portfolio list parses from a real answer."""
 		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_portfolios.json')))
-		self.assertEqual(sharesight.get_portfolios(), {'Rob': 445825, 'RobSMSF': 684141})
+		self.assertEqual(portfolios(), {'Rob': 445825, 'RobSMSF': 684141})
 
 	def test_portfolio_exclusion_and_inclusion(self):
 		"""exclude_portfolios and include_portfolios pick the portfolios reported on."""
 		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_portfolios.json')))
 		self.prefs(exclude_portfolios=['445825'])
 
-		self.assertEqual(sharesight.get_portfolios(), {'RobSMSF': 684141})
-		self.prefs(exclude_portfolios=[])
-
-		self.prefs(include_portfolios=['445825'])
-
-		self.assertEqual(sharesight.get_portfolios(), {'Rob': 445825})
+		self.assertEqual(portfolios(), {'RobSMSF': 684141})
+		self.prefs(exclude_portfolios=[], include_portfolios=['445825'])
+		self.assertEqual(portfolios(), {'Rob': 445825})
 
 	def test_portfolio_failures_raise(self):
 		"""An error answer, Sharesight unreachable or no portfolios is a RuntimeError."""
@@ -388,45 +390,73 @@ def test_portfolio_failures_raise(self):
 			with self.subTest(failure=failure):
 				self.http(sharesight, failure)
 				with self.assertRaises(RuntimeError):
-					sharesight.get_portfolios()
+					sharesight.portfolios()
 
 	def test_an_answer_that_is_not_json_says_so(self):
 		"""An HTML error page from Sharesight is a RuntimeError naming the request and the status, not a JSON parse
 		error."""
 		self.http(sharesight, Resp(None, 502, '<html>Bad gateway</html>'))
 		with self.assertRaisesRegex(RuntimeError, 'Sharesight trades: a 502 answer that is not JSON'):
-			sharesight.get_trades('RobSMSF', 684141, 1)
+			sharesight.Portfolio('RobSMSF', 684141).trades(1)
 
 	def test_trades_from_real_response(self):
-		"""Trades parse from a real answer, tagged with their portfolio."""
+		"""Trades parse from a real answer, tagged with their portfolio, which the cache leaves out: it holds the answer
+		as sent."""
 		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_trades_684141_1.json')))
-		trades = sharesight.get_trades('RobSMSF', 684141, 1)
+		trades = sharesight.Portfolio('RobSMSF', 684141).trades(1)
 		self.assertEqual(len(trades), 2)
 		self.assertEqual(trades[0]['portfolio'], 'RobSMSF')
 		self.assertEqual(trades[0]['symbol'], 'FANG')
+		self.assertEqual(cache.read('finbot_sharesight_trades_684141_1.json'), sample('sharesight',
+			'finbot_sharesight_trades_684141_1.json'))
 
 	def test_trade_failures_raise(self):
 		"""Sharesight unreachable, or an error answer, is a RuntimeError."""
 		self.http(sharesight, ConnectionError('down'))
 		with self.assertRaises(RuntimeError):
-			sharesight.get_trades('RobSMSF', 684141, 1)
+			sharesight.Portfolio('RobSMSF', 684141).trades(1)
 		self.http(sharesight, Resp({'trades': [], 'error': 'bad', 'error_code': 'x'}, 400))
-		with self.assertRaises(RuntimeError):
-			sharesight.get_trades('RobSMSF', 684141, 2)
+		with self.assertRaisesRegex(RuntimeError, 'Sharesight trades: bad'):
+			sharesight.Portfolio('RobSMSF', 684141).trades(2)
+
+	def test_an_oauth_error_is_reported_with_its_description(self):
+		"""Sharesight's refusal of the credentials says why, in OAuth's words (it used to be a KeyError on its missing
+		error_code)."""
+		self.patch(sharesight, '_token', REAL_TOKEN)
+		refusal = {'error': 'invalid_client', 'error_description': 'Client authentication failed'}
+		self.patch(sharesight.net, 'post', lambda url, **kwargs: Resp(refusal, 401))
+		with self.assertRaisesRegex(RuntimeError, 'Sharesight auth: invalid_client: Client authentication failed'):
+			sharesight.portfolios()
+
+	def test_an_answer_without_what_was_asked_for_is_an_error(self):
+		"""A refusal without an error, or a 200 without the list asked for, is an error saying so, not a KeyError
+		later."""
+		for answer in (Resp({'message': 'nope'}, 403), Resp({'links': {}})):
+			with self.subTest(status=answer.status_code):
+				self.http(sharesight, answer)
+				why = f'Sharesight portfolios: a {answer.status_code} answer without portfolios'
+				with self.assertRaisesRegex(RuntimeError, why):
+					sharesight.portfolios()
 
 	def test_performance_and_holdings_from_real_response(self):
-		"""A real holdings report parses, and its tickers come out sorted, US ones without a suffix."""
-		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_performance_684141_0.json')))
-		self.assertEqual(len(sharesight.get_performance(684141, 0)['report']['holdings']), 18)
-		tickers = sharesight.Portfolio('RobSMSF', 684141).tickers()
+		"""A real holdings report parses: its tickers come out sorted, US ones without a suffix, and the gains are the
+		portfolio's and each holding's."""
+		sent = sample('sharesight', 'finbot_sharesight_performance_684141_0.json')
+		self.http(sharesight, Resp(sent))
+		self.assertEqual(len(sharesight._performance(684141, 0)['holdings']), 18)
+		portfolio = sharesight.Portfolio('RobSMSF', 684141)
+		tickers = portfolio.tickers()
 		self.assertIn('ARM', tickers) # US venues carry no Yahoo suffix
 		self.assertEqual(tickers, sorted(tickers))
+		self.assertEqual(portfolio.gain(0), float(sent['report']['capital_gain_percent']))
+		self.assertEqual(sorted(portfolio.gains(0)), tickers)
+		self.assertTrue(all(isinstance(gain, float) for gain in portfolio.gains(0).values()))
 
 	def test_performance_retries_when_sharesight_is_busy(self):
 		"""While Sharesight says too many requests are in flight, the request is tried again."""
 		busy = Resp({'error': 'Too many parallel requests', 'error_code': 'x'}, 429)
 		getter = self.http(sharesight, busy, busy, Resp(sample('sharesight', 'finbot_sharesight_performance_684141_0.json')))
-		self.assertIn('report', sharesight.get_performance(684141, 0))
+		self.assertIn('holdings', sharesight._performance(684141, 0))
 		self.assertEqual(len(getter.calls), 3)
 
 	def test_performance_gives_up_after_repeated_busy_or_hard_errors(self):
@@ -436,13 +466,13 @@ def test_performance_gives_up_after_repeated_busy_or_hard_errors(self):
 			with self.subTest(failure=failure):
 				self.http(sharesight, failure)
 				with self.assertRaises(RuntimeError):
-					sharesight.get_performance(684141, 7)
+					sharesight._performance(684141, 7)
 
 	def test_an_answer_without_a_report_says_so(self):
 		"""A performance answer without a report is an error naming the portfolio, and is not cached."""
 		self.http(sharesight, Resp({}))
-		with self.assertRaisesRegex(RuntimeError, 'no performance report for portfolio 684141'):
-			sharesight.get_performance(684141, 7)
+		with self.assertRaisesRegex(RuntimeError, 'performance of portfolio 684141: a 200 answer without report'):
+			sharesight._performance(684141, 7)
 		self.assertIsNone(cache.read(sharesight.performance_cache_file(684141, 7))) # and is not cached
 
 
@@ -470,39 +500,41 @@ def age(self, cache_file, seconds):
 	def test_cold_cache_is_filled_then_left_alone(self):
 		"""A cold cache is filled (the list and each portfolio's holdings, as of today), then left alone while fresh."""
 		calls = self.serve_sharesight()
-		self.assertEqual(sharesight.warm()[0], 3) # portfolio list + one holdings report per portfolio (2)
+		sharesight.warm()
+		self.assertEqual(len(calls), 3) # portfolio list + one holdings report per portfolio (2)
 		self.assertEqual(sum('/performance' in c for c in calls), 2)
 		self.assertTrue(all('start_date=' + datetime.date.today().isoformat() in c for c in calls if '/performance' in c)) # days=0
-		self.assertEqual(sharesight.warm()[0], 0)
+		sharesight.warm()
 		self.assertEqual(len(calls), 3) # nothing refetched while fresh
 
 	def test_refreshes_before_the_ttl_runs_out(self):
 		"""A cache file at 95% of its lifetime is fetched again, and only that one."""
 		calls = self.serve_sharesight()
-		sharesight.warm()[0]
+		sharesight.warm()
 		calls.clear()
 		self.age(sharesight.performance_cache_file(684141, 0), 0.95 * config.prefs.performance_cache_seconds) # 95% of its lifetime
-		self.assertEqual(sharesight.warm()[0], 1)
+		sharesight.warm()
 		self.assertEqual(len(calls), 1)
 		self.assertIn('684141/performance', calls[0])
 		calls.clear()
 		self.age(sharesight.PORTFOLIOS_CACHE, 0.95 * config.prefs.cache_seconds)
-		self.assertEqual(sharesight.warm()[0], 1)
+		sharesight.warm()
+		self.assertEqual(len(calls), 1)
 		self.assertTrue(calls[0].endswith('/portfolios'))
 
 	def test_a_request_never_waits_on_a_cache_the_job_has_warmed(self):
 		"""Once warmed, the list and the holdings report need no request."""
 		calls = self.serve_sharesight()
-		sharesight.warm()[0]
+		sharesight.warm()
 		calls.clear()
-		self.assertEqual(sharesight.get_portfolios(), {'Rob': 445825, 'RobSMSF': 684141})
-		self.assertEqual(len(sharesight.get_performance(684141, 0)['report']['holdings']), 18)
+		self.assertEqual(portfolios(), {'Rob': 445825, 'RobSMSF': 684141})
+		self.assertEqual(len(sharesight._performance(684141, 0)['holdings']), 18)
 		self.assertEqual(calls, [])
 
 	def test_every_holding_is_served_from_the_warmed_cache(self):
 		"""Every portfolio's tickers come from the warmed cache, with no request."""
 		calls = self.serve_sharesight()
-		sharesight.warm()[0]
+		sharesight.warm()
 		calls.clear()
 		tickers = sharesight.tickers() # the portfolio list and each portfolio's holdings report, no cache of its own
 		self.assertIn('ARM', tickers)
@@ -511,68 +543,79 @@ def test_every_holding_is_served_from_the_warmed_cache(self):
 	def test_failed_refresh_keeps_the_old_cache(self):
 		"""A refresh that fails keeps the cache it would have replaced."""
 		self.serve_sharesight()
-		sharesight.warm()[0]
+		sharesight.warm()
 		self.age(sharesight.PORTFOLIOS_CACHE, 0.95 * config.prefs.cache_seconds)
 		self.http(sharesight, ConnectionError('down'))
 		with self.assertRaises(RuntimeError):
-			sharesight.warm()[0]
-		self.assertEqual(sharesight.get_portfolios(), {'Rob': 445825, 'RobSMSF': 684141}) # still served from the cache
+			sharesight.warm()
+		self.assertEqual(portfolios(), {'Rob': 445825, 'RobSMSF': 684141}) # still served from the cache
+
+	def test_a_pass_says_when_the_next_is_due(self):
+		"""A pass reads each cache file once and gives the seconds until the next refresh: 90% of the shortest cache
+		lifetime once everything is fresh."""
+		self.serve_sharesight()
+		due = sharesight.warm() # a cold cache, filled
+		shortest = min(config.prefs.cache_seconds, config.prefs.performance_cache_seconds)
+		self.assertAlmostEqual(due, 0.9 * shortest, delta=5)
+		reads = []
+		real = cache.read
+		self.patch(cache, 'read', lambda name, *a, **k: (reads.append(name), real(name, *a, **k))[1])
+		self.age(sharesight.performance_cache_file(684141, 0), 600)
+		due = sharesight.warm()
+		self.assertAlmostEqual(due, 0.9 * config.prefs.performance_cache_seconds - 600, delta=5)
+		files = [sharesight.PORTFOLIOS_CACHE] + [sharesight.performance_cache_file(id, 0) for id in (445825, 684141)]
+		self.assertEqual(sorted(reads), sorted(files))
+
+	class Stopped(Exception):
+		"""Raised by the stand-in for keep_warm's wake event, to end the loop."""
+
+	def waits(self, count, on_wait=lambda n: None, **kwargs):
+		"""Run keep_warm until it has waited `count` times, noting each wait instead of sleeping (on_wait(n) at the
+		nth); returns the waits."""
+		waits, stopped = [], self.Stopped
+		class Wake:
+			"""keep_warm's wake event, which nobody sets."""
+			def wait(self, seconds):
+				"""Note the wait instead of sleeping, and end the loop after the last."""
+				waits.append(seconds)
+				on_wait(len(waits))
+				if len(waits) == count:
+					raise stopped
+			def clear(self):
+				"""Nothing to clear."""
+		self.patch(sharesight, '_wake', Wake())
+		with self.assertRaises(self.Stopped):
+			sharesight.keep_warm(**kwargs)
+		return waits
 
 	def test_loop_survives_failures_and_backs_off(self):
 		"""keep_warm survives a failure, retries slowly, then sleeps until the next cache is due."""
-		class Stop:
-			"""keep_warm's stop event: notes each wait, and is set after two."""
-			def __init__(self):
-				"""No waits yet."""
-				self.waits = []
-			def is_set(self):
-				"""Set after two waits."""
-				return len(self.waits) >= 2
-			def wait(self, seconds):
-				"""Note the wait instead of sleeping."""
-				self.waits.append(seconds)
-		outcomes = [RuntimeError('boom'), 0]
+		outcomes = [RuntimeError('boom'), 3000]
 		def warm():
 			"""Fail once, then succeed."""
 			outcome = outcomes.pop(0)
 			if isinstance(outcome, Exception):
 				raise outcome
-			return outcome, {}
+			return outcome
 		self.patch(sharesight, 'warm', warm)
-		self.patch(sharesight, 'next_due', lambda portfolios=None: 3000)
-		stop = Stop()
-		sharesight.keep_warm(retry=300, stop=stop)
-		self.assertEqual(stop.waits, [300, 3000]) # slow retry after the failure, then sleep until the next cache is due
+		self.assertEqual(self.waits(2, retry=300), [300, 3000]) # a slow retry after the failure, then until one is due
 
 	def test_with_cache_off_nothing_is_fetched_until_a_reload_turns_it_on(self):
 		"""With cache off, keep_warm idles without asking Sharesight, and starts when a reload turns it on."""
-		class Stop:
-			"""keep_warm's stop event: notes each wait, switches the cache on at the second, and is set after four."""
-			def __init__(self):
-				"""No waits yet."""
-				self.waits = []
-			def is_set(self):
-				"""Set after four waits."""
-				return len(self.waits) >= 4
-			def wait(self, seconds):
-				"""Note the wait; at the second, switch the cache on as a reload would."""
-				self.waits.append(seconds)
-				if len(self.waits) == 2:
-					config.prefs = dataclasses.replace(config.prefs, cache=True) # what a reload that switches it on does
 		passes = []
-		self.patch(sharesight, 'warm', lambda: (passes.append(1), (0, {}))[1])
-		self.patch(sharesight, 'next_due', lambda portfolios=None: 3000)
+		self.patch(sharesight, 'warm', lambda: passes.append(1) or 3000)
 		self.prefs(cache=False)
-
-		stop = Stop()
-		sharesight.keep_warm(idle=60, stop=stop)
-		self.assertEqual((stop.waits, len(passes)), ([60, 60, 3000, 3000], 2)) # idle while off: no Sharesight request every 5 seconds
+		def reload(n):
+			"""At the second wait, switch the cache on, as a reload would."""
+			if n == 2:
+				config.prefs = dataclasses.replace(config.prefs, cache=True)
+		self.assertEqual(self.waits(4, reload, idle=60), [60, 60, 3000, 3000]) # no request every 5 seconds while off
+		self.assertEqual(len(passes), 2)
 
 	def test_a_flush_wakes_the_loop_to_refill_at_once(self):
 		"""A flush wakes keep_warm, which refills the caches at once instead of when they are next due."""
 		passes = []
-		self.patch(sharesight, 'warm', lambda: (passes.append(1), (0, {}))[1])
-		self.patch(sharesight, 'next_due', lambda portfolios=None: 3000)
+		self.patch(sharesight, 'warm', lambda: passes.append(1) or 3000)
 		loop = gevent.spawn(sharesight.keep_warm)
 		self.addCleanup(loop.kill)
 		gevent.sleep(0.05)
@@ -581,36 +624,6 @@ def test_a_flush_wakes_the_loop_to_refill_at_once(self):
 		gevent.sleep(0.05)
 		self.assertEqual(len(passes), 2)
 
-	def test_a_warming_pass_reads_the_cached_portfolio_list_once(self):
-		"""One pass of keep_warm reads the cached portfolio list once, and hands it on."""
-		self.serve_sharesight()
-		sharesight.warm()[0] # fill the caches
-		reads = []
-		real = cache.read
-		self.patch(cache, 'read', lambda name, *a, **k: (reads.append(name), real(name, *a, **k))[1])
-		fetched, portfolios = sharesight.warm() # what one pass of keep_warm does
-		due = sharesight.next_due(portfolios=portfolios)
-		self.assertEqual((fetched, reads), (0, [sharesight.PORTFOLIOS_CACHE]))
-		self.assertAlmostEqual(due, sharesight.next_due(), delta=1) # the same answer as reading the list again
-		self.assertEqual(portfolios, {'Rob': 445825, 'RobSMSF': 684141})
-
-	def test_next_due_is_90_percent_of_the_ttl(self):
-		"""The next refresh is due at 90% of the shortest cache lifetime; at once for a cold cache."""
-		self.serve_sharesight()
-		self.assertEqual(sharesight.next_due(), 0) # cold cache
-		sharesight.warm()[0]
-		due = sharesight.next_due()
-		self.assertAlmostEqual(due, 0.9 * min(config.prefs.cache_seconds, config.prefs.performance_cache_seconds), delta=5)
-
-	def test_refresh_flag_bypasses_a_fresh_cache(self):
-		"""refresh=True asks Sharesight even while the cache is fresh."""
-		calls = self.serve_sharesight()
-		sharesight.get_portfolios()
-		sharesight.get_portfolios()
-		self.assertEqual(len(calls), 1)
-		sharesight.get_portfolios(refresh=True)
-		self.assertEqual(len(calls), 2)
-
 
 class TelegramSending(FinbotCase):
 	"""A message posted to Telegram against failing endpoints: the real send path, not the base fixture's recorder."""
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 149a631a..fae473b3 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -86,7 +86,8 @@ def test_help_comes_from_the_command_table(self):
 
 	def test_a_portfolio_with_no_holdings_is_reported_by_name(self):
 		"""Its name used to be read from its first holding: an empty portfolio crashed .performance."""
-		self.patch(sharesight, 'get_performance', lambda pid, d, *a: {'report': {'holdings': [], 'capital_gain_percent': '0.0'}})
+		self.patch(sharesight, '_performance', lambda pid, days=None, seconds=None:
+			{'holdings': [], 'capital_gain_percent': '0.0'})
 		payload, _ = performance.report(TELEGRAM, 7)
 		self.assertTrue(any('>Rob</a>' in line for line in payload), payload)
 
@@ -322,8 +323,8 @@ def test_trades(self):
 	def test_a_trade_is_in_its_market_currency_not_the_brokerage_one(self):
 		"""A NASDAQ trade is in US dollars, though Yahoo names the exchange NasdaqGS and the broker charges in
 		Australian dollars (it used to say AUD)."""
-		self.patch(sharesight, 'get_trades', lambda name, pid, days=7: [{'id': 1, 'transaction_type': 'BUY',
-			'symbol': 'T1', 'market': 'NASDAQ', 'portfolio': name, 'transaction_date': time.strftime('%Y-%m-%d'),
+		self.patch(sharesight, '_trades', lambda pid, days=None: [{'id': 1, 'transaction_type': 'BUY',
+			'symbol': 'T1', 'market': 'NASDAQ', 'transaction_date': time.strftime('%Y-%m-%d'),
 			'quantity': '10', 'price': '5.0', 'value': '50', 'brokerage_currency_code': 'AUD', 'holding_id': 7}])
 		self.assertIn(' USD 50 ', self.command('.trades')[0][1][-1])
 

From 3fac2f0fb27fb53b87516331befb6e8e928e5b34 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 09:57:33 +0000
Subject: [PATCH 62/81] Markets: one table by market code, TSE back to Toronto,
 a Yahoo suffix for every market
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- TSE is the Toronto Stock Exchange in Sharesight, and TYO Tokyo (as in Sharesight's code list, and as the original
  ticker translation had it). Since d2b297c a TSE holding was looked up on Tokyo (SHOP.T), dropped from the reports,
  and shown with Japan's flag and yen; it is SHOP.TO again, with Canada's flag and dollar.
- One table holds each market's flag, currency, Yahoo suffix and Google code. 13 of 40 codes had a suffix, so a
  holding on Xetra (Sharesight's XETR), Frankfurt, Milan, Madrid, Swiss, Nordic, Indian, Chinese or Thai markets
  asked Yahoo for SAP.XETR and dropped out; OTC holdings (Sharesight's OTC) did too. Codes from Sharesight's list
  that were missing are in (XETR, CVE, CNSX, NEO, SHG, BSE, OSL, CSE, HEL, ASE, IDX, KLS, PHS); the old ones stay.
- European markets get their country's flag (Amsterdam was 🇳🇱 as a Yahoo ticker but 🇪🇺 as Sharesight's AMS);
  EURONEXT, five exchanges, keeps 🇪🇺.
- One venue table maps Yahoo's names and ticker suffixes to those codes, so the flag, Google code and country of an
  exchange come from the same row. The price report's heading now says AU for the ASX (it said ASX; the docstring
  always promised AU), US for Cboe US, and GB for London (it said UK); a Google link for an NZX stock says NZE and
  one for Xetra ETR.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/markets.py     | 197 +++++++++++++++++++++++++++------------------
 tests/test_util.py |  39 +++++++--
 2 files changed, 150 insertions(+), 86 deletions(-)

diff --git a/lib/markets.py b/lib/markets.py
index 7ec4d557..b0fee86e 100644
--- a/lib/markets.py
+++ b/lib/markets.py
@@ -1,31 +1,81 @@
-"""Markets: flags and currencies by market, exchange names, and ticker translation between Sharesight, Google and Yahoo."""
-
-MARKETS = { # Sharesight market code -> (flag, currency); no currency where a market quotes in several (LSE and HKEX allow non-home currencies)
-	'ASX': ('🇦🇺', 'AUD'),
-	'BOM': ('🇮🇳', 'INR'), 'NSE': ('🇮🇳', 'INR'),
-	'BMV': ('🇲🇽', 'MXN'),
-	'BKK': ('🇹🇭', 'THB'),
-	'BVMF': ('🇧🇷', 'BRL'),
-	'SHE': ('🇨🇳', 'CNY'), 'SHA': ('🇨🇳', 'CNY'),
-	'CPSE': ('🇩🇰', 'DKK'),
-	**dict.fromkeys(('EURONEXT', 'AMS', 'ATH', 'BIT', 'BME', 'DUB', 'EBR', 'EPA', 'ETR', 'FWB', 'FRA', 'VIE'), ('🇪🇺', 'EUR')),
-	'HKG': ('🇭🇰', None),
-	'ICSE': ('🇮🇸', 'ISK'),
-	'JSE': ('🇿🇦', 'ZAR'),
-	'KRX': ('🇰🇷', 'KRW'), 'KOSDAQ': ('🇰🇷', 'KRW'),
-	'LSE': ('🇬🇧', 'GBP'), # better than trades.py using brokerage_currency_code, which is AUD on CMC
-	'MISX': ('🇷🇺', 'RUB'),
-	'OM': ('🇸🇪', 'SEK'), 'STO': ('🇸🇪', 'SEK'),
-	'SWX': ('🇨🇭', 'CHF'), 'VTX': ('🇨🇭', 'CHF'),
-	'TAI': ('🇹🇼', 'TWD'), 'TPE': ('🇹🇼', 'TWD'),
-	'TASE': ('🇮🇱', 'ILS'),
-	'NZX': ('🇳🇿', 'NZD'),
-	'OB': ('🇳🇴', 'NOK'),
-	'SGX': ('🇸🇬', 'SGD'),
-	'TSE': ('🇯🇵', 'JPY'),
-	'TSX': ('🇨🇦', 'CAD'),
-	'NASDAQ': ('🇺🇸', 'USD'), 'NYSE': ('🇺🇸', 'USD'), 'BATS': ('🇺🇸', 'USD'), 'AMEX': ('🇺🇸', None),
-	'WAR': ('🇵🇱', 'PLN'),
+"""Markets: each one's flag, currency and Yahoo suffix by its code, the names exchanges go by, and ticker translation
+between Sharesight, Google and Yahoo."""
+import typing
+
+
+class Market(typing.NamedTuple):
+	"""A market: its flag; its currency, None where it quotes in several (HKEX takes others too); the suffix of its
+	Yahoo tickers, '' where they take none (a US venue) and None where Yahoo has several (EURONEXT is five exchanges);
+	and Google Finance's code for it, where that is not the same."""
+	flag: str
+	currency: str | None
+	yahoo: str | None
+	google: str | None = None
+
+
+US = Market('🇺🇸', 'USD', '')
+MARKETS = { # by Sharesight's codes (TSE is Toronto, TYO Tokyo), and Google's where they are others
+	'ASX': Market('🇦🇺', 'AUD', 'AX'),
+	'NZX': Market('🇳🇿', 'NZD', 'NZ', 'NZE'),
+	'NASDAQ': US, 'NYSE': US, 'BATS': US, 'NYSEARCA': US,
+	'AMEX': US._replace(google='NYSEAMERICAN'),
+	**dict.fromkeys(('OTC', 'OTCBB', 'OTCMKTS'), US._replace(google='OTCMKTS')),
+	**dict.fromkeys(('TSE', 'TSX'), Market('🇨🇦', 'CAD', 'TO', 'TSE')), # Toronto: Google calls it TSE too
+	'CVE': Market('🇨🇦', 'CAD', 'V'), # TSX Venture
+	'CNSX': Market('🇨🇦', 'CAD', 'CN'),
+	'NEO': Market('🇨🇦', 'CAD', 'NE'),
+	'BMV': Market('🇲🇽', 'MXN', 'MX'),
+	'BVMF': Market('🇧🇷', 'BRL', 'SA'),
+	'LSE': Market('🇬🇧', 'GBP', 'L', 'LON'), # it takes other currencies, but GBP beats the brokerage's (AUD on CMC)
+	**dict.fromkeys(('XETR', 'ETR'), Market('🇩🇪', 'EUR', 'DE', 'ETR')), # Xetra
+	**dict.fromkeys(('FRA', 'FWB'), Market('🇩🇪', 'EUR', 'F', 'FRA')), # Frankfurt
+	'EURONEXT': Market('🇪🇺', 'EUR', None),
+	'EPA': Market('🇫🇷', 'EUR', 'PA'), 'AMS': Market('🇳🇱', 'EUR', 'AS'), 'EBR': Market('🇧🇪', 'EUR', 'BR'),
+	'DUB': Market('🇮🇪', 'EUR', 'IR'),
+	'BIT': Market('🇮🇹', 'EUR', 'MI'),
+	'BME': Market('🇪🇸', 'EUR', 'MC'),
+	'VIE': Market('🇦🇹', 'EUR', 'VI'),
+	**dict.fromkeys(('ASE', 'ATH'), Market('🇬🇷', 'EUR', 'AT', 'ATH')),
+	'HEL': Market('🇫🇮', 'EUR', 'HE'),
+	**dict.fromkeys(('STO', 'OM'), Market('🇸🇪', 'SEK', 'ST', 'STO')),
+	**dict.fromkeys(('CSE', 'CPSE'), Market('🇩🇰', 'DKK', 'CO', 'CPH')),
+	**dict.fromkeys(('OSL', 'OB'), Market('🇳🇴', 'NOK', 'OL', 'OSL')),
+	'ICSE': Market('🇮🇸', 'ISK', 'IC', 'ICE'),
+	**dict.fromkeys(('SWX', 'SIX', 'VTX'), Market('🇨🇭', 'CHF', 'SW', 'SWX')),
+	'WAR': Market('🇵🇱', 'PLN', 'WA', 'WSE'),
+	'MISX': Market('🇷🇺', 'RUB', 'ME', 'MCX'),
+	'JSE': Market('🇿🇦', 'ZAR', 'JO'),
+	**dict.fromkeys(('TASE', 'TLV'), Market('🇮🇱', 'ILS', 'TA', 'TLV')),
+	'TYO': Market('🇯🇵', 'JPY', 'T'),
+	'HKG': Market('🇭🇰', None, 'HK'),
+	'SGX': Market('🇸🇬', 'SGD', 'SI'),
+	'KRX': Market('🇰🇷', 'KRW', 'KS'), 'KOSDAQ': Market('🇰🇷', 'KRW', 'KQ'),
+	'TAI': Market('🇹🇼', 'TWD', 'TW', 'TPE'), # the Taiwan Stock Exchange, Google's TPE
+	'TPE': Market('🇹🇼', 'TWD', 'TWO'), # the Taipei Exchange, which Google lacks
+	**dict.fromkeys(('BSE', 'BOM'), Market('🇮🇳', 'INR', 'BO', 'BOM')),
+	'NSE': Market('🇮🇳', 'INR', 'NS'),
+	**dict.fromkeys(('SHG', 'SHA'), Market('🇨🇳', 'CNY', 'SS', 'SHA')),
+	'SHE': Market('🇨🇳', 'CNY', 'SZ'),
+	'BKK': Market('🇹🇭', 'THB', 'BK'),
+	'IDX': Market('🇮🇩', 'IDR', 'JK'),
+	'KLS': Market('🇲🇾', 'MYR', 'KL', 'KLSE'),
+	'PHS': Market('🇵🇭', 'PHP', 'PS', 'PSE'),
+}
+
+VENUES = { # the other names of an exchange -> its code in MARKETS: its Yahoo tickers' suffix, and Yahoo's names
+	**{market.yahoo: code for code, market in reversed(MARKETS.items()) if market.yahoo},
+	'NYSEArca': 'NYSEARCA', 'NYSE American': 'AMEX', 'Cboe US': 'BATS', # other Nasdaq and NYSE names: see code()
+	**dict.fromkeys(('Other OTC', 'PNK', 'OTCPK'), 'OTC'),
+	**dict.fromkeys(('Toronto', 'TOR'), 'TSE'),
+	**dict.fromkeys(('NZSE', 'NZE', 'New Zealand', 'Auckland'), 'NZX'),
+	'London': 'LSE', 'XETRA': 'XETR', 'Frankfurt': 'FRA',
+	'Paris': 'EPA', 'Amsterdam': 'AMS', 'Brussels': 'EBR', 'Irish': 'DUB', 'Milan': 'BIT', 'Vienna': 'VIE',
+	'Helsinki': 'HEL', 'Stockholm': 'STO', 'Copenhagen': 'CSE', 'Oslo': 'OSL', 'Swiss': 'SWX',
+	'Johannesburg': 'JSE', 'Tel Aviv': 'TASE', 'Mexico': 'BMV', 'São Paulo': 'BVMF',
+	**dict.fromkeys(('Tokyo', 'JPX'), 'TYO'),
+	'HKSE': 'HKG', 'SES': 'SGX', 'KSE': 'KRX', 'KSC': 'KRX', 'KOE': 'KOSDAQ',
+	'Taiwan': 'TAI', 'Taipei Exchange': 'TPE', 'Taipei': 'TPE',
+	'Shanghai': 'SHG', 'Shenzhen': 'SHE', 'Thailand': 'BKK', 'Jakarta': 'IDX', 'Kuala Lumpur': 'KLS',
 }
 
 YAHOO_COUNTRY = { # country -> its Yahoo ticker suffix, for a listing that has none (cal.py)
@@ -70,70 +120,57 @@
 	# markets with numeric tickers purposefully omitted
 }
 
-def flag_from_market(market):
-	"""The flag of a market by its Sharesight code: US for the NASDAQ and NYSE variants, '' when unknown."""
-	if market in MARKETS:
-		return MARKETS[market][0]
-	return '🇺🇸' if market.lower().startswith(('nasdaq', 'nyse')) else ''
 
-TICKER_FLAGS = {'AS': '🇳🇱', 'AX': '🇦🇺', 'HK': '🇭🇰', 'KS': '🇰🇷', 'KQ': '🇰🇷', 'L': '🇬🇧', 'NS': '🇮🇳', 'NZ': '🇳🇿', 'SI': '🇸🇬', 'T': '🇯🇵', 'TA': '🇮🇱',
-	'TW': '🇹🇼', 'TWO': '🇹🇼', 'TO': '🇨🇦'} # Yahoo suffix -> flag
+def code(name):
+	"""MARKETS' code for an exchange by any name it goes by: Sharesight's code or Google's, the suffix of its Yahoo
+	tickers, or Yahoo's name for it (each Nasdaq tier has one: NasdaqGS); the name itself when unknown."""
+	if name in MARKETS:
+		return name
+	if name in VENUES:
+		return VENUES[name]
+	return 'NASDAQ' if name.startswith('Nasdaq') else 'NYSE' if name.startswith('NYSE') else name
+
+
+def flag_from_market(name):
+	"""The flag of an exchange by any name it goes by (see code), '' when unknown."""
+	market = MARKETS.get(code(name))
+	return market.flag if market else ''
+
+
+def currency_from_market(name):
+	"""The currency of an exchange by any name it goes by (see code); None where it quotes in several, or is unknown."""
+	market = MARKETS.get(code(name))
+	return market.currency if market else None
+
 
 def flag_from_ticker(ticker):
-	"""The flag of a Yahoo ticker's market suffix, '' when unknown; a ticker without one is a US listing."""
-	return TICKER_FLAGS.get(ticker.split('.')[1], '') if '.' in ticker else '🇺🇸'
-
-def currency_from_market(market):
-	"""The currency of a market by its Sharesight code; None where it quotes in several, or is unknown."""
-	return MARKETS.get(market, (None, None))[1]
-
-VENUES = { # the names Yahoo, Sharesight and Google give an exchange -> (Google Finance code, country)
-	**dict.fromkeys(('OTC', 'PNK', 'Other OTC', 'OTCPK'), ('OTCMKTS', None)),
-	**dict.fromkeys(('TO', 'TOR', 'Toronto'), ('TSE', 'CA')),
-	**dict.fromkeys(('TW', 'TWO', 'TAI', 'Taiwan', 'Taipei Exchange', 'Taipei'), ('TPE', 'TW')),
-	**dict.fromkeys(('HK', 'HKG', 'HKSE'), ('HKG', 'HK')),
-	**dict.fromkeys(('KQ', 'KOE', 'KOSDAQ'), ('KOSDAQ', 'KR')),
-	**dict.fromkeys(('KS', 'KRX', 'KSE', 'KSC'), ('KRX', 'KR')),
-	**dict.fromkeys(('L', 'LSE', 'London'), ('LON', 'UK')),
-	**dict.fromkeys(('T', 'TYO', 'JPX', 'Tokyo'), ('TYO', 'JP')),
-	**dict.fromkeys(('TA', 'TLV', 'Tel Aviv'), ('TLV', 'IL')),
-	**dict.fromkeys(('NZSE', 'NZE', 'New Zealand', 'Auckland'), (None, 'NZ')),
-}
+	"""The flag of a Yahoo ticker's market by its suffix, '' when unknown; a ticker without one is a US listing."""
+	return flag_from_market(ticker.split('.')[1]) if '.' in ticker else '🇺🇸'
+
 
 def transform_to_google(exchange):
-	"""Google Finance's code for an exchange, as Yahoo, Sharesight or Google name it; the name itself when unknown."""
-	if 'Nasdaq' in exchange:
-		return 'NASDAQ'
-	if exchange == 'Cboe US':
-		return 'BATS'
-	return VENUES.get(exchange, (None, None))[0] or exchange
+	"""Google Finance's code for an exchange by any name it goes by (see code); the name itself when unknown."""
+	market = MARKETS.get(code(exchange))
+	return (market.google or code(exchange)) if market else exchange
+
 
 def exchange_human(exchange):
-	"""The country an exchange is in ('US', 'AU' and so on), or its name when unknown."""
-	if 'Nasdaq' in exchange or 'NYSE' in exchange or 'Other OTC' in exchange:
-		return 'US'
-	return VENUES.get(exchange, (None, None))[1] or exchange
-
-YAHOO_SUFFIX = { # exchange code (Sharesight, Google) -> Yahoo's ticker suffix
-	'ASX': 'AX', 'HKG': 'HK', 'KRX': 'KS', 'KOSDAQ': 'KQ', 'LSE': 'L', 'NZX': 'NZ', 'SGX': 'SI',
-	'TAI': 'TW', # Taiwan in YF and Sharesight; TPE in GF
-	'TPE': 'TWO', # Taipei in YF, Taiwan in GF (GF format will not work as input); missing from GF
-	'TSE': 'T', 'TYO': 'T', # Tokyo: TSE in Sharesight (Google's TSE is Toronto, which Sharesight calls TSX), TYO in GF
-	'TSX': 'TO', 'TLV': 'TA',
-}
+	"""The country an exchange is in ('US', 'AU' and so on: the letters of its flag), or its name when unknown."""
+	flag = flag_from_market(exchange)
+	return ''.join(chr(ord(letter) - ord('🇦') + ord('A')) for letter in flag) or exchange
+
 
 def transform_to_yahoo(ticker, market=None):
-	"""Yahoo's symbol for a ticker: 'BHP:ASX', 'BHP.ASX' or ('BHP', 'ASX') give 'BHP.AX'. US venues take no suffix, and
-	class shares a dash ('BRK.B' gives 'BRK-B')."""
+	"""Yahoo's symbol for a ticker: 'BHP:ASX', 'BHP.ASX' or ('BHP', 'ASX') give 'BHP.AX' (a market by its code in
+	MARKETS; a code Yahoo uses itself, like AX, stays as it is). US venues take no suffix, and class shares a dash
+	('BRK.B' gives 'BRK-B')."""
 	split = ticker.replace(':', '.').split('.')
 	if len(split) > 1:
-		ticker = split[0]
-		market = split[1]
+		ticker, market = split[0], split[1]
 	if not market:
 		return ticker
-	market = YAHOO_SUFFIX.get(market, market)
-	if market in {'NASDAQ', 'NYSE', 'BATS', 'OTCMKTS', 'AMEX', 'NYSEARCA'}: # US venues: Yahoo symbols carry no suffix
-		return ticker
 	if market in {'A', 'B', 'C'}: # class shares
 		return ticker + '-' + market
-	return ticker + '.' + market
+	known = MARKETS.get(market)
+	suffix = market if known is None or known.yahoo is None else known.yahoo
+	return ticker + '.' + suffix if suffix else ticker
diff --git a/tests/test_util.py b/tests/test_util.py
index 326d52e8..7fd4854c 100644
--- a/tests/test_util.py
+++ b/tests/test_util.py
@@ -22,21 +22,48 @@ def test_days_from_human_days_rejects_non_periods(self):
 	def test_transform_to_yahoo(self):
 		"""Sharesight and Google tickers as Yahoo's: suffixes by market, none for US venues, a dash for class shares."""
 		for args, expected in ((('AAPL',), 'AAPL'), (('FANG', 'ASX'), 'FANG.AX'), (('ARM', 'NASDAQ'), 'ARM'),
-							   (('ARM:NASDAQ',), 'ARM'), (('BHP.AX',), 'BHP.AX'), (('0700', 'HKG'), '0700.HK'),
-							   (('BRK', 'B'), 'BRK-B'), (('2330', 'TAI'), '2330.TW'), (('VOD', 'LSE'), 'VOD.L'),
-							   (('7203', 'TSE'), '7203.T'), (('SHOP', 'TSX'), 'SHOP.TO'), (('AIR', 'NZX'), 'AIR.NZ'), (('D05', 'SGX'), 'D05.SI')):
+				(('ARM:NASDAQ',), 'ARM'), (('BHP.AX',), 'BHP.AX'), (('0700', 'HKG'), '0700.HK'),
+				(('BRK', 'B'), 'BRK-B'), (('2330', 'TAI'), '2330.TW'), (('VOD', 'LSE'), 'VOD.L'),
+				(('7203', 'TYO'), '7203.T'),
+				(('SHOP', 'TSE'), 'SHOP.TO'), (('SHOP', 'TSX'), 'SHOP.TO'), (('AIR', 'NZX'), 'AIR.NZ'),
+				(('D05', 'SGX'), 'D05.SI'), (('SAP', 'XETR'), 'SAP.DE'), (('SAP', 'ETR'), 'SAP.DE'),
+				(('NSRGY', 'OTC'), 'NSRGY'), (('NESN', 'SWX'), 'NESN.SW'), (('SAP:FRA',), 'SAP.F')):
 			with self.subTest(args=args):
 				self.assertEqual(markets.transform_to_yahoo(*args), expected)
 
 	def test_each_market_has_its_own_flag_and_currency(self):
-		"""Sharesight's market codes each have their own flag and currency: TSE is Tokyo, TSX Toronto, SGX Singapore."""
-		for market, ticker, flag, currency in (('TSE', '7203.T', '🇯🇵', 'JPY'), ('TSX', 'SHOP.TO', '🇨🇦', 'CAD'), ('NZX', 'AIR.NZ', '🇳🇿', 'NZD'),
-											   ('SGX', 'D05.SI', '🇸🇬', 'SGD'), ('CPSE', None, '🇩🇰', 'DKK')):
+		"""Sharesight's market codes each have their own flag and currency, the same as their Yahoo tickers' flags:
+		TSE is Toronto and TYO Tokyo (as Sharesight has them), SGX Singapore, XETR Germany."""
+		for market, ticker, flag, currency in (('TSE', 'SHOP.TO', '🇨🇦', 'CAD'), ('TYO', '7203.T', '🇯🇵', 'JPY'),
+				('TSX', 'SHOP.TO', '🇨🇦', 'CAD'), ('NZX', 'AIR.NZ', '🇳🇿', 'NZD'), ('SGX', 'D05.SI', '🇸🇬', 'SGD'),
+				('CPSE', 'NOVO-B.CO', '🇩🇰', 'DKK'), ('XETR', 'SAP.DE', '🇩🇪', 'EUR'), ('AMS', 'ASML.AS', '🇳🇱', 'EUR'),
+				('EURONEXT', None, '🇪🇺', 'EUR'), ('HKG', '0700.HK', '🇭🇰', None), ('NASDAQ', 'AAPL', '🇺🇸', 'USD')):
 			with self.subTest(market=market):
 				self.assertEqual((markets.flag_from_market(market), markets.currency_from_market(market)), (flag, currency))
 				if ticker:
 					self.assertEqual(markets.flag_from_ticker(ticker), flag)
 
+	def test_an_exchange_by_any_of_its_names(self):
+		"""Yahoo's name for an exchange, its tickers' suffix or Sharesight's code give the same country, flag and
+		Google code; a name nobody knows stays as it is."""
+		for names, country, google in ((('NasdaqGS', 'NasdaqCM', 'NASDAQ'), 'US', 'NASDAQ'),
+				(('NYSEArca', 'NYSEARCA'), 'US', 'NYSEARCA'), (('Cboe US', 'BATS'), 'US', 'BATS'),
+				(('Other OTC', 'OTC'), 'US', 'OTCMKTS'), (('ASX', 'AX'), 'AU', 'ASX'),
+				(('Toronto', 'TO', 'TSE'), 'CA', 'TSE'), (('London', 'L', 'LSE'), 'GB', 'LON'),
+				(('XETRA', 'DE', 'XETR'), 'DE', 'ETR'), (('NZSE', 'NZ'), 'NZ', 'NZE'),
+				(('Taiwan', 'TW', 'TAI'), 'TW', 'TPE'),
+				(('Unheard of',), 'Unheard of', 'Unheard of')):
+			for name in names:
+				with self.subTest(name=name):
+					found = markets.exchange_human(name), markets.transform_to_google(name)
+					self.assertEqual(found, (country, google))
+		self.assertEqual(markets.flag_from_market('NasdaqGS'), '🇺🇸') # a trade's market, with trades_use_yahoo
+
+	def test_no_yahoo_suffix_is_also_a_market_code(self):
+		"""A Yahoo suffix that is also a code in MARKETS would name two markets: BHP.AX typed in a chat must stay AX."""
+		suffixes = {market.yahoo for market in markets.MARKETS.values()}
+		self.assertEqual(suffixes & set(markets.MARKETS), set())
+
 	def test_make_paragraphs_keeps_every_word(self):
 		"""make_paragraphs breaks a long description into paragraphs at sentence ends without losing a word."""
 		self.assertEqual(formats.make_paragraphs('Acme makes widgets. It sells them.'), 'Acme makes widgets. It sells them.')

From b17d728f7d97ca79041f5dc26653b0d83f035c2d Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 10:01:40 +0000
Subject: [PATCH 63/81] Config: each preference's default on its Prefs field,
 read in one loop by its type

A preference was written twice in lib/config.py (its field, then its read with the default) and twice more in the
tests (a table of every default, and a count of the keys). Now the field holds the default and its type says how it is
read (a tuple is a list of words; exchange_suffix and hyperlinkProvider are tidied as before), so a new preference goes
in three places: the field, defaults/preferences.conf and the README's table. The tests check the template against the
fields' defaults, which are the same as before (checked against the old table before it went).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md            |   4 +-
 lib/config.py        | 100 ++++++++++++++++++-------------------------
 tests/test_config.py |  22 ++++------
 3 files changed, 53 insertions(+), 73 deletions(-)

diff --git a/CLAUDE.md b/CLAUDE.md
index 2d258927..db2e495c 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -22,7 +22,9 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   pull requests that change only README.md, CLAUDE.md, `doc/*.md` or `img/*.png` (install, compile and tests are
   skipped; the allowlist in `.github/workflows/ci.yml` must never include a path a test reads).
 - Production runs the Debian system packages (including `python3-croniter`, for the scheduler in `lib/jobs.py`); `requirements.txt` is for pip/uv users and CI.
-- A new preference goes in five places: a `Prefs` field and its line in `preference_values` (`lib/config.py`), `defaults/preferences.conf`, the README's preferences table, and `DEFAULTS` plus the key count in `tests/test_config.py`. The tests catch all but the README (no test may read README.md: CI skips the tests for a README-only change).
+- A new preference goes in three places: a `Prefs` field with its default (`lib/config.py`; its type says how it is read),
+  `defaults/preferences.conf`, and the README's preferences table. The tests catch a key missing from the template or a
+  default that differs, but not the README (no test may read README.md: CI skips the tests for a README-only change).
 - If Yahoo and Sharesight are unreachable (the cloud environment's network policy can block them), check behaviour
   against the offline fixtures in `tests/common.py`.
 - For a refactor, `util/record_outputs.py` (or `--against COMMIT`) records what every command, button and scheduled
diff --git a/lib/config.py b/lib/config.py
index dd765e37..39a8be19 100644
--- a/lib/config.py
+++ b/lib/config.py
@@ -162,6 +162,11 @@ def float(self, name, default):
 		"""A number setting."""
 		return self.number(float, name, default)
 
+	def words(self, name, default):
+		"""A list of words, separated by spaces: a tuple of them."""
+		value = self.value(name, default)
+		return tuple(value.split()) if isinstance(value, str) else tuple(value)
+
 	def check(self):
 		"""Raise ConfigError with every problem found so far."""
 		if self.errors:
@@ -236,71 +241,50 @@ def startup_values(getenv):
 
 @dataclasses.dataclass(frozen=True)
 class Prefs:
-	"""The preferences: preferences.conf, which the bot reads at start and on every reload. Read them as config.prefs.<name> when
-	they are used, never into a default or a module constant, so that a reload is seen."""
-	cache: bool
-	cache_seconds: int
-	chunk_maxlines: int
-	country_code: str
-	debug: bool
-	demote_leveraged: bool
-	demote_volatile: bool
-	exchange_suffix: str # Yahoo's suffix for the local exchange (AX): a bare ticker Yahoo does not know gets it
-	exclude_portfolios: tuple
-	future_days: int
-	graph: bool
-	http_timeout: int
-	hyperlink: bool
-	hyperlinkFooter: bool
-	hyperlinkProvider: str
-	include_portfolios: tuple
-	min_analysts: int # .buy, .sell and .forward pe list only stocks this many analysts cover
-	past_days: int
-	performance_cache_seconds: int # Sharesight performance reports (30 minutes)
-	performance_use_sharesight: bool
-	price_percent: float
-	print_headers: bool
-	shorts_percent: int
-	timezone: str
-	trades_use_yahoo: bool # set False if Yahoo breaks
-	volatility_multiplier: float
+	"""The preferences: preferences.conf, which the bot reads at start and on every reload, each with its default here
+	and read as its type says (preference_values). Read them as config.prefs.<name> when they are used, never into a
+	default or a module constant, so that a reload is seen."""
+	cache: bool = True
+	cache_seconds: int = 43200
+	chunk_maxlines: int = 20
+	country_code: str = 'AU'
+	debug: bool = False
+	demote_leveraged: bool = True
+	demote_volatile: bool = True
+	exchange_suffix: str = 'AX' # Yahoo's suffix for the local exchange: a bare ticker Yahoo does not know gets it
+	exclude_portfolios: tuple = () # portfolio ids
+	future_days: int = 30
+	graph: bool = True
+	http_timeout: int = 18
+	hyperlink: bool = True
+	hyperlinkFooter: bool = True
+	hyperlinkProvider: str = 'google'
+	include_portfolios: tuple = () # portfolio ids
+	min_analysts: int = 7 # .buy, .sell and .forward pe list only stocks this many analysts cover
+	past_days: int = 28
+	performance_cache_seconds: int = 1800 # Sharesight performance reports (30 minutes)
+	performance_use_sharesight: bool = False
+	price_percent: float = 5.0
+	print_headers: bool = False
+	shorts_percent: int = 15
+	timezone: str = 'Australia/Melbourne'
+	trades_use_yahoo: bool = True # set False if Yahoo breaks
+	volatility_multiplier: float = 2.0
 	source: dict = dataclasses.field(default_factory=dict, compare=False, repr=False) # the file's own lines, for job processes
 
 
 def preference_values(getenv, source=None):
-	"""The preferences (preferences.conf), computed from getenv(name, default); `source` is the file's own {key: value}. Raises
+	"""The preferences (preferences.conf), computed from getenv(name, default): each read as its Prefs field's type
+	says (a tuple is a list of words), its default the field's. `source` is the file's own {key: value}. Raises
 	ConfigError listing every bad value."""
 	r = _Reader(getenv)
-	prefs = Prefs(
-		cache=r.bool('cache', True),
-		cache_seconds=r.int('cache_seconds', 43200),
-		chunk_maxlines=r.int('chunk_maxlines', 20),
-		country_code=r.str('country_code', 'AU'),
-		debug=r.bool('debug', False),
-		demote_leveraged=r.bool('demote_leveraged', True),
-		demote_volatile=r.bool('demote_volatile', True),
-		exchange_suffix=r.blankable('exchange_suffix', 'AX').removeprefix('.').upper(),
-		exclude_portfolios=tuple(r.str('exclude_portfolios', '').split()),
-		future_days=r.int('future_days', 30),
-		graph=r.bool('graph', True),
-		http_timeout=r.int('http_timeout', 18),
-		hyperlink=r.bool('hyperlink', True),
-		hyperlinkFooter=r.bool('hyperlinkFooter', True),
-		hyperlinkProvider=r.str('hyperlinkProvider', 'google').lower(),
-		include_portfolios=tuple(r.str('include_portfolios', '').split()),
-		min_analysts=r.int('min_analysts', 7),
-		past_days=r.int('past_days', 28),
-		performance_cache_seconds=r.int('performance_cache_seconds', 1800),
-		performance_use_sharesight=r.bool('performance_use_sharesight', False),
-		price_percent=r.float('price_percent', 5),
-		print_headers=r.bool('print_headers', False),
-		shorts_percent=r.int('shorts_percent', 15),
-		timezone=r.str('timezone', 'Australia/Melbourne'),
-		trades_use_yahoo=r.bool('trades_use_yahoo', True),
-		volatility_multiplier=r.float('volatility_multiplier', 2),
-		source=dict(source or {}))
+	readers = {bool: r.bool, int: r.int, float: r.float, str: r.str, tuple: r.words,
+		'exchange_suffix': lambda name, default: r.blankable(name, default).removeprefix('.').upper(), # blank: none
+		'hyperlinkProvider': lambda name, default: r.str(name, default).lower()}
+	values = {field.name: readers.get(field.name, readers[field.type])(field.name, field.default)
+		for field in dataclasses.fields(Prefs) if field.compare}
 	r.check()
-	return prefs
+	return Prefs(**values, source=dict(source or {}))
 
 
 PREFERENCE_KEYS = tuple(field.name for field in dataclasses.fields(Prefs) if field.compare) # preferences.conf's keys: what a reload can change
diff --git a/tests/test_config.py b/tests/test_config.py
index 95392796..a8626b3e 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -10,13 +10,6 @@
 from lib.reports import performance, price, shorts, trades
 
 REPO = Path(config.__file__).resolve().parent.parent
-DEFAULTS = {'past_days': 28, 'future_days': 30, 'price_percent': 5.0, 'shorts_percent': 15, 'cache_seconds': 43200,
-	'performance_cache_seconds': 1800, 'graph': True, 'http_timeout': 18, 'chunk_maxlines': 20, 'country_code': 'AU',
-	'timezone': 'Australia/Melbourne', 'cache': True, 'exchange_suffix': 'AX', 'hyperlink': True,
-	'hyperlinkFooter': True,
-	'hyperlinkProvider': 'google', 'demote_volatile': True, 'demote_leveraged': True, 'volatility_multiplier': 2.0, 'exclude_portfolios': (),
-	'include_portfolios': (), 'trades_use_yahoo': True, 'performance_use_sharesight': False, 'print_headers': False, 'min_analysts': 7,
-	'debug': False}
 
 
 def fields(prefs):
@@ -31,30 +24,31 @@ def keys_in(path):
 
 class Tables(unittest.TestCase):
 	def test_the_two_key_sets_are_disjoint_and_cover_everything_config_reads(self):
-		"""Startup and preference keys do not overlap, and between them name every setting config.py reads."""
+		"""Startup and preference keys do not overlap: config.py reads each startup key by name, and each preference
+		as a field of Prefs."""
 		self.assertEqual(set(config.STARTUP_KEYS) & set(config.PREFERENCE_KEYS), set())
-		self.assertEqual((len(config.STARTUP_KEYS), len(config.PREFERENCE_KEYS)), (19, 26))
 		source = (REPO / 'lib' / 'config.py').read_text()
 		read = set(re.findall(r"""(?:\br\.\w+|getenv)\(\s*['"]([A-Za-z_]+)['"]""", source))
 		read -= {'STATE_DIRECTORY'} # systemd's own variable, a source for DATA_DIR
-		self.assertEqual(read, set(config.STARTUP_KEYS) | set(config.PREFERENCE_KEYS))
+		self.assertEqual(read, set(config.STARTUP_KEYS))
 
 	def test_secrets_are_startup_keys(self):
 		"""Every secret is a startup key: preferences.conf holds none."""
 		self.assertTrue(set(config.SECRET_KEYS) <= set(config.STARTUP_KEYS))
 		self.assertEqual(len(config.SECRET_KEYS), 12)
 
-	def test_the_preference_defaults_are_the_old_defaults(self):
-		"""Unset preferences take the defaults finbot always had."""
+	def test_unset_preferences_take_their_defaults(self):
+		"""An unset preference takes its field's default, as it is: reading tidies nothing away."""
 		values = config.preference_values(lambda name, default=None: default)
-		self.assertEqual(fields(values), DEFAULTS)
+		self.assertEqual(fields(values), fields(config.Prefs()))
 
 	def test_templates_have_every_key_and_the_real_defaults(self):
 		"""defaults/preferences.conf and finbot.env list every key, change no default, and are valid as shipped."""
 		self.assertEqual(sorted(keys_in(REPO / 'defaults' / 'preferences.conf')), sorted(config.PREFERENCE_KEYS))
 		self.assertEqual(sorted(keys_in(REPO / 'defaults' / 'finbot.env')), sorted(config.STARTUP_KEYS))
 		prefs = config.read_preferences(REPO / 'defaults' / 'preferences.conf')
-		self.assertEqual(fields(config.preference_values(config._getenv_for(prefs))), DEFAULTS) # the template changes nothing
+		shipped = config.preference_values(config._getenv_for(prefs))
+		self.assertEqual(fields(shipped), fields(config.Prefs())) # the template changes nothing
 		startup = config.read_startup_file(REPO / 'defaults' / 'finbot.env')
 		config.startup_values(lambda name, default=None: startup.get(name, default)) # and is valid as it stands
 

From fe7481535a7a79df900c21044a4f691d65a449db Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 10:11:59 +0000
Subject: [PATCH 64/81] Small ones: compact cache files, no chdir at import,
 tests clean up after themselves

- Cache files are written compact (a 10-year chart: 832 KB to 297 KB, and quicker to parse); state files keep their
  indentation, for people to read.
- lib/config no longer changes directory when imported, so util/check_config.py and split_env.py take a relative
  path from the shell's directory, as a command-line tool should. A relative DATA_DIR (and the default, var/) is
  still in the checkout, wherever finbot starts.
- A full test run left about 100 directories in /tmp (preferences files, schedules, job records); now none.
- The scheduler's flag that skipped a whole line while its last run went on is gone: the set of commands running
  already skips any command still going, whoever started it. The in-process runner only the tests used moved to them.
- formats: ordinal and strip_url in a line each; td_to_human went with its last caller. util/sws_gen.py runs only as
  a script and leaves the umask as it found it. .holdings and .pie lost their branch for no portfolio, which the
  command now answers with its usage and the portfolios before they are called.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/cache.py            |  5 ++--
 lib/config.py           |  4 ++--
 lib/formats.py          | 41 +++-----------------------------
 lib/jobs.py             | 28 ++++------------------
 lib/reports/holdings.py |  6 -----
 tests/common.py         | 10 +++++++-
 tests/test_config.py    | 42 ++++++++++++++++-----------------
 tests/test_datadir.py   | 15 ++++++++----
 tests/test_jobs.py      | 52 +++++++++++++++++------------------------
 tests/test_providers.py |  2 +-
 tests/test_reports.py   |  8 ++++---
 tests/test_util.py      |  7 ++++++
 util/sws_gen.py         | 30 ++++++++++++++----------
 13 files changed, 107 insertions(+), 143 deletions(-)

diff --git a/lib/cache.py b/lib/cache.py
index 5ff8dc2f..a7a31c36 100644
--- a/lib/cache.py
+++ b/lib/cache.py
@@ -82,10 +82,11 @@ def cached(name, seconds, fetch):
 
 
 def write(name, data):
-	"""Write a cache file's JSON. See write_file."""
+	"""Write a cache file's JSON, compact: nobody reads it but finbot (a 10-year chart takes a third of the space, and
+	parses faster). See write_file."""
 	where = path(name)
 	print("cache write:", where, file=sys.stderr) if config.prefs.debug else None
-	write_file(where, json.dumps(data, indent=4).encode('utf-8'))
+	write_file(where, json.dumps(data, separators=(',', ':')).encode('utf-8'))
 
 
 def write_bytes(name, data):
diff --git a/lib/config.py b/lib/config.py
index 39a8be19..f9ba4548 100644
--- a/lib/config.py
+++ b/lib/config.py
@@ -12,7 +12,6 @@
 #   preferences.conf how the bot behaves: <DATA_DIR>/preferences.conf, read by the bot itself at start and on every reload (SIGHUP).
 
 REPO_DIR = Path(__file__).resolve().parent.parent
-os.chdir(REPO_DIR)
 
 STARTUP_KEYS = ('DATA_DIR', 'ip', 'port', 'telegramChatID', 'scheduler', 'scheduler_catchup',
 	'sharesight_code', 'sharesight_client_id', 'sharesight_client_secret', 'slack_webhook', 'discord_webhook', 'slackBotToken',
@@ -203,7 +202,8 @@ def preferences_path(self):
 def startup_values(getenv):
 	"""The settings fixed when the bot starts (finbot.env), computed from getenv(name, default). Raises ConfigError."""
 	r = _Reader(getenv)
-	data_dir = (getenv('DATA_DIR') or (getenv('STATE_DIRECTORY') or '').split(':')[0] or 'var').rstrip('/') or '/'
+	data_dir = getenv('DATA_DIR') or (getenv('STATE_DIRECTORY') or '').split(':')[0] or 'var'
+	data_dir = str(REPO_DIR / data_dir) # a relative one is in the checkout, wherever finbot starts
 	hooks = {}
 	if getenv('slack_webhook'):
 		hooks['slack'] = getenv('slack_webhook').rstrip('/')
diff --git a/lib/formats.py b/lib/formats.py
index 1331e7f8..e7fa7864 100644
--- a/lib/formats.py
+++ b/lib/formats.py
@@ -73,11 +73,7 @@ def security_line(flag, title, link):
 
 def strip_url(url):
 	"""url without its scheme and www., to show as a link's text."""
-	url = url.removeprefix('https://www.')
-	url = url.removeprefix('http://www.')
-	url = url.removeprefix('https://')
-	url = url.removeprefix('http://')
-	return url
+	return url.split('://', 1)[-1].removeprefix('www.')
 
 def make_paragraphs(walloftext):
 	"""The text in paragraphs, each ending with the first sentence that takes it past 22 words."""
@@ -150,36 +146,5 @@ def days_from_human_days(arg):
 	return int(arg.removesuffix('Y')) * 365
 
 def ordinal(num):
-	"""The ordinal suffix of a number: 'st' for 1, 'nd' for 22, 'th' for 11 and 13."""
-	value = str(num)
-	if len(value) > 1:
-		secondToLastDigit = value[-2]
-		if secondToLastDigit == '1':
-			return 'th'
-	lastDigit = value[-1]
-	if (lastDigit == '1'):
-		return 'st'
-	elif (lastDigit == '2'):
-		return 'nd'
-	elif (lastDigit == '3'):
-		return 'rd'
-	else:
-		return 'th'
-
-def td_to_human(timedelta):
-	"""A duration in words: '1 day, 2 hours, 5 seconds', zero parts left out ('0 seconds' for none)."""
-	days, remainder = divmod(timedelta.total_seconds(), 86400)
-	hours, remainder = divmod(remainder, 3600)
-	minutes, seconds = divmod(remainder, 60)
-	parts = []
-	if days:
-		parts.append(f"{int(days)} day{'s' if days != 1 else ''}")
-	if hours:
-		parts.append(f"{int(hours)} hour{'s' if hours != 1 else ''}")
-	if minutes:
-		parts.append(f"{int(minutes)} minute{'s' if minutes != 1 else ''}")
-	if seconds:
-		parts.append(f"{int(seconds)} second{'s' if seconds != 1 else ''}")
-	if not parts:
-		return "0 seconds"
-	return ", ".join(parts)
+	"""The ordinal suffix of a whole number: 'st' for 1, 'nd' for 22, 'th' for 11 and 13."""
+	return 'th' if num % 100 in (11, 12, 13) else {1: 'st', 2: 'nd', 3: 'rd'}.get(num % 10, 'th')
diff --git a/lib/jobs.py b/lib/jobs.py
index 58b03c2e..dfed7b97 100644
--- a/lib/jobs.py
+++ b/lib/jobs.py
@@ -16,7 +16,7 @@
 import time
 import zlib
 import zoneinfo
-from dataclasses import dataclass, field
+from dataclasses import dataclass
 from pathlib import Path
 
 from croniter import croniter, CroniterBadDateError
@@ -85,12 +85,6 @@ def exit_reason(status):
 	return f'exited with status {status} without saying why'
 
 
-def run_in_process(label, module, kwargs):
-	"""The runner for tests: call the handler here instead of in a child process (so without a deadline)."""
-	module.run(**kwargs)
-	return 0
-
-
 def display(label):
 	"""A command as shown in chat: 'price.py close' -> 'price close'."""
 	return re.sub(r'^(\w+)\.py', r'\1', label)
@@ -116,7 +110,6 @@ class Job:
 	zone: zoneinfo.ZoneInfo
 	text: str
 	key: str
-	running: bool = field(default=False, compare=False)
 
 
 def local_zone():
@@ -354,18 +347,6 @@ def due(self, now):
 					print('schedule: cannot save', STATE_FILE, e, file=sys.stderr)
 		return fired
 
-	def run_job(self, job, fire_time):
-		"""One job: its commands in order (execute). Skipped if the previous run is still going."""
-		with self.lock:
-			if job.running:
-				print(f'schedule: {job.text} is still running, skipping {fire_time:%H:%M}', file=sys.stderr)
-				return 'already running'
-			job.running = True
-		try:
-			return self.execute(job.commands)
-		finally:
-			job.running = False
-
 	def claim(self, label):
 		"""Mark a command running, unless it already is (scheduled or manual): whether it was claimed."""
 		with self.lock:
@@ -375,8 +356,9 @@ def claim(self, label):
 			return True
 
 	def execute(self, commands):
-		"""Run commands one after another, each stopped and reported if still running after JOB_TIMEOUT. One already
-		running, scheduled or manual, is skipped. Returns 'done', 'timed out', 'failed' or 'already running'."""
+		"""Run commands one after another (a job's), each stopped and reported if still running after JOB_TIMEOUT.
+		One already running (from the line's last run, another line or chat) is skipped. Returns 'done', 'timed out',
+		'failed' or 'already running'."""
 		outcome = 'done'
 		for label, module, kwargs in commands:
 			if not self.claim(label):
@@ -455,7 +437,7 @@ def tick(self):
 		"""Start what is due; returns the seconds until the next fire time."""
 		now = self.clock()
 		for job, fire_time in self.due(now):
-			self.spawn(self.run_job, job, fire_time)
+			self.spawn(self.execute, job.commands)
 		with self.lock:
 			waits = [(self.next_fire(job, self.base[job.key]) - now).total_seconds() for job in self.jobs]
 		return min(30, max(1, math.ceil(min(waits)))) if waits else 30
diff --git a/lib/reports/holdings.py b/lib/reports/holdings.py
index 63abe705..dc62f13b 100644
--- a/lib/reports/holdings.py
+++ b/lib/reports/holdings.py
@@ -14,9 +14,6 @@ def holding_line(holding, service, text=None):
 
 def prepare_holdings_payload(portfolio_name, service):
 	"""The holdings of the portfolio called `portfolio_name`. Raises sharesight.PortfolioNotFound."""
-	if not portfolio_name:
-		names = [portfolio.name for portfolio in sharesight.portfolios()]
-		return [".holdings: please try again specifying a portfolio:"] + names
 	portfolio, = sharesight.portfolios(portfolio_name)
 	holdings = portfolio.holdings()
 	payload = sorted(holding_line(holding, service) for holding in holdings.values())
@@ -35,9 +32,6 @@ def prepare_who(service, code):
 
 def prepare_pie(portfolio_name, by, service, max_slices=8):
 	"""Portfolio value by industry (or by='sector'): (payload, chart or None). Raises sharesight.PortfolioNotFound."""
-	if not portfolio_name:
-		names = [portfolio.name for portfolio in sharesight.portfolios()]
-		return [".pie: please try again specifying a portfolio:"] + names, None
 	portfolio, = sharesight.portfolios(portfolio_name)
 	name = portfolio.name
 	key = 'sector_classification_name' if by == 'sector' else 'industry_classification_name'
diff --git a/tests/common.py b/tests/common.py
index 70fc8dd3..864adb63 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -1,6 +1,6 @@
 """Shared fixture: points finbot at a temp cache dir and replaces every external service
 (Telegram, Yahoo, Sharesight, Short Man) with canned data. Nothing here touches the network."""
-import collections, dataclasses, io, os, sys, tempfile, time, unittest
+import collections, dataclasses, io, os, shutil, sys, tempfile, time, unittest
 import pandas as pd
 from unittest import mock
 
@@ -18,6 +18,14 @@
 	(yahoo, 'price_history'), (yahoo, 'historic_high'),
 	(sharesight, '_portfolios'), (sharesight, '_trades'), (sharesight, '_performance'))}
 REAL_GET_HOLDINGS = watchlist.with_holdings
+
+
+def temp_dir(case):
+	"""A new temporary directory, removed when the test `case` ends."""
+	path = tempfile.mkdtemp()
+	case.addCleanup(shutil.rmtree, path, True)
+	return path
+
 REAL_TEXT = transports.Transport.text # FinbotCase replaces it with a recorder; tests of the real thing put it back
 REAL_CHART = transports.Transport.chart # likewise
 REAL_REPORT_ERROR = delivery.report_error # likewise
diff --git a/tests/test_config.py b/tests/test_config.py
index a8626b3e..8cbbdee0 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -1,13 +1,12 @@
 """The two config files: finbot.env (the startup file, loaded by the launcher) and preferences.conf (read by the bot, reloadable)."""
-import ast, contextlib, dataclasses, io, os, re, subprocess, sys, tempfile, unittest
+import ast, contextlib, dataclasses, io, os, re, subprocess, sys, unittest
 from pathlib import Path
 from unittest import mock
 from tests import common # first: it sets the environment that lib.config reads as it loads
 from lib import config
 from util import check_config, split_env
 import bot
-from lib import cache, sharesight, worker, yahoo
-from lib.reports import performance, price, shorts, trades
+from lib import cache, yahoo
 
 REPO = Path(config.__file__).resolve().parent.parent
 
@@ -56,7 +55,7 @@ def test_templates_have_every_key_and_the_real_defaults(self):
 class Parsing(unittest.TestCase):
 	def parse(self, text):
 		"""(what read_preferences makes of text, the file it was written to)."""
-		path = Path(tempfile.mkdtemp(), 'preferences.conf')
+		path = Path(common.temp_dir(self), 'preferences.conf')
 		path.write_text(text)
 		return config.read_preferences(path), path
 
@@ -73,7 +72,7 @@ def test_comments_blanks_quotes_and_equals_in_values(self):
 
 	def test_missing_file_is_none(self):
 		"""No preferences.conf is not an error: None."""
-		self.assertIsNone(config.read_preferences(Path(tempfile.mkdtemp(), 'nope')))
+		self.assertIsNone(config.read_preferences(Path(common.temp_dir(self), 'nope')))
 
 	def test_unknown_key_gets_a_hint_and_a_line_number(self):
 		"""An unknown key is reported with its line number and the key it was probably meant to be."""
@@ -132,7 +131,7 @@ def test_a_blank_value_means_the_default(self):
 	def test_a_retired_key_is_accepted_and_said_to_do_nothing(self):
 		"""A key finbot no longer reads (alliterate) does not stop a start, as an unknown key would: it is ignored, and
 		the log at start says to delete it."""
-		path = Path(tempfile.mkdtemp(), 'preferences.conf')
+		path = Path(common.temp_dir(self), 'preferences.conf')
 		path.write_text('alliterate=True\nfuture_days=45\n')
 		values = config.read_preferences(path)
 		self.assertEqual(config.preference_values(config._getenv_for(values)).future_days, 45)
@@ -142,7 +141,7 @@ def test_a_retired_key_is_accepted_and_said_to_do_nothing(self):
 
 	def test_unreadable_file_is_an_error_naming_it(self):
 		"""A preferences file that cannot be read is an error that names it."""
-		path = Path(tempfile.mkdtemp(), 'preferences.conf')
+		path = Path(common.temp_dir(self), 'preferences.conf')
 		path.mkdir() # reading a directory fails like a permission error would
 		with self.assertRaises(config.ConfigError) as caught:
 			config.read_preferences(path)
@@ -165,22 +164,23 @@ def test_the_old_multiplier_name_is_gone_everywhere(self):
 
 	def test_startup_file_linting(self):
 		"""util/check_config.py reports bad lines in finbot.env first, then bad values, and nothing for a good file."""
-		path = Path(tempfile.mkdtemp(), 'finbot.env')
+		path = Path(common.temp_dir(self), 'finbot.env')
 		path.write_text('ip=1.2.3.4\nfuture_days=3\nslak_webhook=x\nport=http\n')
-		problems = check_config.check(path, Path(tempfile.mkdtemp(), 'none'))
+		problems = check_config.check(path, Path(common.temp_dir(self), 'none'))
 		self.assertEqual(len(problems), 2) # the two bad lines; the bad value is only found once the lines are valid
 		self.assertIn('future_days belongs in preferences.conf (it is reloadable)', problems[0])
 		self.assertIn('unknown key slak_webhook (did you mean slack_webhook?)', problems[1])
 		path.write_text('port=http\n')
-		self.assertIn("port='http' is not a whole number", check_config.check(path, Path(tempfile.mkdtemp(), 'none'))[0])
+		nowhere = Path(common.temp_dir(self), 'none')
+		self.assertIn("port='http' is not a whole number", check_config.check(path, nowhere)[0])
 		path.write_text('port=80\n')
-		self.assertEqual(check_config.check(path, Path(tempfile.mkdtemp(), 'none')), [])
+		self.assertEqual(check_config.check(path, Path(common.temp_dir(self), 'none')), [])
 
 
 class Reload(unittest.TestCase):
 	def setUp(self):
 		"""An empty preferences.conf of this test's own, put back to the defaults afterwards."""
-		self.file = Path(tempfile.mkdtemp(), 'preferences.conf')
+		self.file = Path(common.temp_dir(self), 'preferences.conf')
 		self.file.write_text('')
 		patch = mock.patch.object(config, 'preferences_path', str(self.file))
 		patch.start()
@@ -286,7 +286,7 @@ def test_a_job_process_runs_on_the_snapshot_not_the_edited_file(self):
 		env = config.child_environment()
 		self.assertEqual((env['future_days'], env[config.PASSED]), ('45', '1'))
 		self.write('future_days=99\n') # edited, no reload sent
-		data = tempfile.mkdtemp()
+		data = common.temp_dir(self)
 		Path(data, 'preferences.conf').write_text('future_days=99\n')
 		show = "from lib import config as c; print(c.prefs.future_days)"
 		run = lambda **extra: subprocess.run([sys.executable, '-c', show], cwd=REPO, capture_output=True, text=True, check=True,
@@ -314,7 +314,7 @@ def popen(args, **kw):
 class Startup(unittest.TestCase):
 	def run_config(self, code='from lib import config as c; print(c.prefs.future_days)', prefs=None, **env):
 		"""Run code in a new interpreter: own data dir, prefs as preferences.conf, no settings in the environment."""
-		data = tempfile.mkdtemp()
+		data = common.temp_dir(self)
 		if prefs is not None:
 			Path(data, 'preferences.conf').write_text(prefs)
 		e = {k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k not in config.STARTUP_KEYS and k != config.PASSED}
@@ -340,7 +340,7 @@ def test_a_broken_preferences_file_refuses_to_start(self):
 
 	def test_an_unreadable_preferences_file_refuses_to_start(self):
 		"""A preferences.conf that cannot be read stops the start."""
-		data = tempfile.mkdtemp()
+		data = common.temp_dir(self)
 		Path(data, 'preferences.conf').mkdir()
 		out = subprocess.run([sys.executable, '-c', 'from lib import config'], cwd=REPO, env={**os.environ, 'DATA_DIR': data}, capture_output=True, text=True)
 		self.assertEqual(out.returncode, 1)
@@ -368,13 +368,13 @@ def test_a_preference_in_the_environment_is_refused_by_the_bot(self):
 	def test_the_bot_refuses_to_start_with_a_preference_in_its_environment(self):
 		"""bot.py exits 1 when finbot.env holds a preference."""
 		out = subprocess.run([sys.executable, 'bot.py'], cwd=REPO, capture_output=True, text=True, timeout=60,
-			env={**os.environ, 'DATA_DIR': tempfile.mkdtemp(), 'future_days': '30'})
+			env={**os.environ, 'DATA_DIR': common.temp_dir(self), 'future_days': '30'})
 		self.assertEqual(out.returncode, 1)
 		self.assertIn('future_days is a preference', out.stderr)
 
 	def test_a_leftover_env_in_the_checkout_refuses_to_start(self):
 		"""An old .env in the checkout stops the start, pointing to util/split_env.py."""
-		old = Path(tempfile.mkdtemp(), '.env')
+		old = Path(common.temp_dir(self), '.env')
 		with mock.patch.object(config, 'OLD_ENV', old):
 			config.check_startup_environment() # none there: fine
 			old.write_text("future_days = '30'\n")
@@ -430,7 +430,7 @@ def restore(self, file):
 
 	def test_reload_all_reloads_preferences_and_the_schedule_independently(self):
 		"""SIGHUP reloads preferences, keeping the running ones on an error (logged), and the schedule regardless."""
-		file = Path(tempfile.mkdtemp(), 'preferences.conf')
+		file = Path(common.temp_dir(self), 'preferences.conf')
 		file.write_text('future_days=45\n')
 		scheduler = mock.Mock()
 		self.addCleanup(self.restore, file)
@@ -450,7 +450,7 @@ def test_reload_all_reloads_preferences_and_the_schedule_independently(self):
 
 	def test_reload_with_the_scheduler_off_still_reloads_preferences(self):
 		"""With the scheduler off, SIGHUP still reloads preferences."""
-		file = Path(tempfile.mkdtemp(), 'preferences.conf')
+		file = Path(common.temp_dir(self), 'preferences.conf')
 		file.write_text('future_days=11\n')
 		self.addCleanup(self.restore, file)
 		with mock.patch.object(config, 'preferences_path', str(file)), mock.patch.object(bot, 'scheduler', None):
@@ -479,7 +479,7 @@ def test_split_env_reads_the_old_dotenv_format_too(self):
 	def test_split_env_writes_next_to_the_file_and_never_overwrites(self):
 		"""split_env writes finbot.env.new (mode 0600) and preferences.conf.new beside the file, refuses to overwrite
 		them, and says so for a missing file."""
-		d = Path(tempfile.mkdtemp())
+		d = Path(common.temp_dir(self))
 		(d / 'old.env').write_text('port=80\nfuture_days=3\n')
 		run = lambda: subprocess.run([sys.executable, str(REPO / 'util' / 'split_env.py'), str(d / 'old.env')], capture_output=True, text=True)
 		self.assertEqual(run().returncode, 0)
@@ -494,7 +494,7 @@ def test_split_env_writes_next_to_the_file_and_never_overwrites(self):
 
 	def test_check_config_script_exit_status_and_it_does_not_create_directories(self):
 		"""check_config exits 0 for a good file, 1 for a bad one, and makes no data dir (systemd may run it as root)."""
-		d = Path(tempfile.mkdtemp())
+		d = Path(common.temp_dir(self))
 		(d / 'finbot.env').write_text('port=80\n')
 		data = d / 'data'
 		run = lambda: subprocess.run([sys.executable, str(REPO / 'util' / 'check_config.py'), str(d / 'finbot.env')], capture_output=True, text=True,
diff --git a/tests/test_datadir.py b/tests/test_datadir.py
index 789a6edc..33e81464 100644
--- a/tests/test_datadir.py
+++ b/tests/test_datadir.py
@@ -1,4 +1,4 @@
-import os, shutil, subprocess, sys, tempfile, time, unittest
+import os, subprocess, sys, tempfile, time
 from unittest import mock
 from pathlib import Path
 from lib import config
@@ -17,9 +17,16 @@ def run_config(code, **env):
 
 class DataDirTests(FinbotCase):
 	def test_default_unchanged(self):
-		"""Without DATA_DIR, state lives in var/ in the checkout and the cache in var/cache."""
-		out = run_config("from lib import config as c; print(c.env.data_dir, c.env.cache_dir)")
-		self.assertEqual(out.split()[-2:], ['var', 'var/cache'])
+		"""Without DATA_DIR, state lives in var/ in the checkout and the cache in var/cache, from whichever directory
+		finbot starts; so does a relative DATA_DIR."""
+		show = "from lib import config as c; print(c.env.data_dir, c.env.cache_dir)"
+		var = str(REPO / 'var')
+		self.assertEqual(run_config(show).split()[-2:], [var, var + '/cache'])
+		env = {k: v for k, v in os.environ.items() if k != 'DATA_DIR'}
+		env.update(PYTHONPATH=str(REPO), DATA_DIR='state', FINBOT_CHECK_ONLY='1') # check only: no directory is created
+		elsewhere = subprocess.run([sys.executable, '-c', show], cwd='/', env=env, capture_output=True, text=True,
+			check=True)
+		self.assertEqual(elsewhere.stdout.split()[-2:], [str(REPO / 'state'), str(REPO / 'state/cache')])
 
 	def test_data_dir_moves_var_and_cache(self):
 		"""DATA_DIR moves both, and the cache directory is created."""
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
index 88840885..a0951aa7 100644
--- a/tests/test_jobs.py
+++ b/tests/test_jobs.py
@@ -12,6 +12,12 @@
 DEFAULT_SCHEDULE = Path(os.path.dirname(os.path.dirname(os.path.abspath(__file__)))) / 'defaults' / 'schedule.cron'
 
 
+def run_in_process(label, module, kwargs):
+	"""The scheduler's runner in these tests: the command's handler, called here (no child process, no deadline)."""
+	module.run(**kwargs)
+	return 0
+
+
 def at(day, hour, minute, second=0):
 	"""A UTC time in October 2026 (the 13th is a Tuesday)."""
 	return datetime.datetime(2026, 10, day, hour, minute, second, tzinfo=UTC)
@@ -183,8 +189,7 @@ def test_each_scheduled_report_runs_from_its_top_level_script(self):
 class CleanupJob(FinbotCase):
 	def test_removes_only_old_finbot_cache_files(self):
 		"""cleanup deletes finbot's cache files older than the days given, not newer ones, other files or dirs."""
-		folder = tempfile.mkdtemp()
-		self.addCleanup(lambda: __import__('shutil').rmtree(folder, ignore_errors=True))
+		folder = temp_dir(self)
 		self.env(cache_dir=folder)
 
 		old, new = time.time() - 40 * 86400, time.time() - 5 * 86400
@@ -199,8 +204,7 @@ def test_removes_only_old_finbot_cache_files(self):
 
 	def test_a_temp_file_being_written_is_left_alone(self):
 		"""A flush (0 days) spares a temp file from the last minute (a write in flight) and deletes a stale one."""
-		folder = tempfile.mkdtemp()
-		self.addCleanup(lambda: __import__('shutil').rmtree(folder, ignore_errors=True))
+		folder = temp_dir(self)
 		self.env(cache_dir=folder)
 
 		for name, stamp in (('finbot_a.json.x1.tmp', time.time() - 5), ('finbot_b.json.x2.tmp', time.time() - 40 * 86400)):
@@ -250,7 +254,7 @@ def setUp(self):
 	def scheduler(self, catchup=5):
 		"""A loaded scheduler on the fake clock that runs commands in this process, at once."""
 		s = jobs.Scheduler(lambda: self.file, catchup=catchup, clock=self.clock, sleep=lambda seconds: None,
-						   spawn=lambda f, *a: f(*a), zone=UTC, runner=jobs.run_in_process)
+						   spawn=lambda f, *a: f(*a), zone=UTC, runner=run_in_process)
 		self.assertTrue(s.load())
 		return s
 
@@ -317,17 +321,6 @@ def test_a_restart_just_after_a_run_does_not_repeat_it(self):
 		again = self.scheduler()
 		self.assertEqual(again.due(self.clock.now), [])
 
-	def test_an_overlapping_run_is_skipped(self):
-		"""A run due while the last is still going is skipped."""
-		s = self.scheduler()
-		s.jobs[0].running = True
-		s.run_job(s.jobs[0], at(13, 21, 20))
-		self.assertEqual(self.ran, [])
-		s.jobs[0].running = False
-		s.run_job(s.jobs[0], at(13, 21, 20))
-		self.assertEqual(self.ran, [('trades', {})])
-		self.assertFalse(s.jobs[0].running)
-
 	def test_a_failing_job_is_reported_and_the_next_command_still_runs(self):
 		"""A command that fails is reported, and the next command on its line still runs."""
 		self.file.write_text('*/20 * * * * trades.py; shorts.py\n')
@@ -336,7 +329,7 @@ def boom(**k):
 			raise RuntimeError('boom')
 		self.patch(jobs.SCRIPTS['trades'], 'run', boom)
 		s = self.scheduler()
-		s.run_job(s.jobs[0], at(13, 21, 20))
+		s.execute(s.jobs[0].commands)
 		self.assertEqual(self.ran, [('shorts', {})])
 		self.assertTrue(any('boom' in e for e in self.errors))
 
@@ -347,9 +340,8 @@ def quit_(**k):
 			raise SystemExit(3)
 		self.patch(jobs.SCRIPTS['trades'], 'run', quit_)
 		s = self.scheduler()
-		s.run_job(s.jobs[0], at(13, 21, 20))
+		s.execute(s.jobs[0].commands)
 		self.assertEqual(len(self.errors), 1)
-		self.assertFalse(s.jobs[0].running)
 
 	def test_a_command_stopped_at_its_deadline_is_reported_and_the_next_still_runs(self):
 		"""A command still running at its deadline (run_child stops it) is reported, and the rest of its line runs."""
@@ -358,13 +350,12 @@ def runner(label, module, kwargs):
 			"""trades.py hangs until run_child stops it; the rest run here."""
 			if label == 'trades.py':
 				raise TimeoutError('still running after 5 minutes')
-			return jobs.run_in_process(label, module, kwargs)
+			return run_in_process(label, module, kwargs)
 		s = self.scheduler()
 		s.runner = runner
-		self.assertEqual(s.run_job(s.jobs[0], at(13, 21, 20)), 'timed out')
+		self.assertEqual(s.execute(s.jobs[0].commands), 'timed out')
 		self.assertEqual(self.errors, ['⚠️ trades.py: TimeoutError: still running after 5 minutes'])
 		self.assertEqual(self.ran, [('shorts', {})])
-		self.assertFalse(s.jobs[0].running)
 		self.assertEqual(s.busy, set())
 
 	def test_reload_applies_edits_keeps_state_and_survives_a_bad_file(self):
@@ -459,11 +450,12 @@ def test_starting_in_the_repeated_hour_runs_nothing_twice(self):
 
 	def test_the_scheduler_runs_a_daily_job_once_across_the_change(self):
 		"""The scheduler runs a daily Melbourne job once a day across the change to daylight time."""
-		file = Path(tempfile.mkdtemp(), 'schedule.cron')
+		file = Path(temp_dir(self), 'schedule.cron')
 		file.write_text('CRON_TZ=Australia/Melbourne\n0 9 * * * trades.py\n')
 		clock = Clock(self.mel(2026, 10, 3, 12).astimezone(UTC))
 		s = jobs.Scheduler(lambda: file, clock=clock, zone=UTC, runner=lambda *a: 0)
-		with mock.patch.object(config, 'env', dataclasses.replace(config.env, data_dir=tempfile.mkdtemp())), contextlib.redirect_stderr(io.StringIO()):
+		elsewhere = dataclasses.replace(config.env, data_dir=temp_dir(self))
+		with mock.patch.object(config, 'env', elsewhere), contextlib.redirect_stderr(io.StringIO()):
 			self.assertTrue(s.load())
 			fired = []
 			while clock.now < self.mel(2026, 10, 5, 12):
@@ -521,7 +513,6 @@ def test_run_now_runs_one_command_and_leaves_the_schedule_alone(self):
 		self.assertEqual(self.s.run_now('7'), ('price.py close', 'done'))
 		self.assertEqual(self.ran, [('price', {'mode': 'close'})])
 		self.assertEqual((self.s.base, self.s.last), (base, last))
-		self.assertTrue(all(not job.running for job in self.s.jobs))
 		self.assertEqual(self.s.busy, set())
 
 	def test_run_now_refuses_while_the_command_is_running_and_reports_failures(self):
@@ -539,9 +530,10 @@ def boom(**k):
 		self.assertEqual(self.s.busy, set())
 
 	def test_a_scheduled_run_skips_a_command_that_a_manual_run_is_in_the_middle_of(self):
-		"""A scheduled run skips a command a manual run is in the middle of, and runs the rest of its line."""
+		"""A scheduled run skips a command still running (from chat, or the line's last run), and runs the rest of its
+		line."""
 		self.s.busy.add('trades.py')
-		self.assertEqual(self.s.run_job(self.s.jobs[1], at(13, 21, 28)), 'already running')
+		self.assertEqual(self.s.execute(self.s.jobs[1].commands), 'already running')
 		self.assertEqual(self.ran, [('price', {'mode': 'close'})]) # price.py close ran, trades.py was skipped
 
 	def test_run_now_times_out(self):
@@ -742,7 +734,7 @@ def test_a_script_that_cannot_be_imported_is_reported(self):
 	def test_scheduler_reports_a_child_that_ended_without_saying_why(self):
 		"""A child killed by a signal or exiting non-zero without a report is reported by the scheduler; one that
 		reported itself is not reported twice."""
-		file = Path(tempfile.mkdtemp(), 'schedule.cron')
+		file = Path(temp_dir(self), 'schedule.cron')
 		file.write_text('* * * * * trades.py\n')
 		for status, reason in ((-signal.SIGHUP, 'killed by SIGHUP'), (-signal.SIGKILL, 'killed by SIGKILL'), (1, 'exited with status 1 without saying why'),
 							   (jobs.jobrun.REPORTED, None)): # the child reported this one itself
@@ -800,12 +792,12 @@ def test_each_command_of_a_line_has_a_deadline_of_its_own(self):
 		"""The deadline is per command (it once covered the whole line): two that each take most of it both finish."""
 		self.patch(jobs, 'JOB_TIMEOUT', 0.8)
 		self.sleeper(0.5)
-		file = Path(tempfile.mkdtemp(), 'schedule.cron')
+		file = Path(temp_dir(self), 'schedule.cron')
 		file.write_text('* * * * * trades.py; shorts.py\n')
 		s = jobs.Scheduler(lambda: file, zone=UTC)
 		with contextlib.redirect_stderr(io.StringIO()):
 			self.assertTrue(s.load())
-			self.assertEqual(s.run_job(s.jobs[0], at(13, 21, 20)), 'done')
+			self.assertEqual(s.execute(s.jobs[0].commands), 'done')
 		self.assertEqual(len(self.children), 2)
 		self.assertEqual(self.errors, [])
 
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 2c7f92ae..07a99419 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -1,6 +1,6 @@
 """Yahoo, Sharesight and Telegram parsing and failure handling, driven by the real API responses saved under doc/.
 HTTP is mocked at lib.net, so the production parsing code runs unchanged."""
-import contextlib, copy, datetime, io, json, os, subprocess, threading, unittest
+import contextlib, datetime, io, json, os, subprocess, threading, unittest
 import gevent
 from unittest import mock
 from tests.common import *
diff --git a/tests/test_reports.py b/tests/test_reports.py
index fae473b3..690c9200 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -91,9 +91,11 @@ def test_a_portfolio_with_no_holdings_is_reported_by_name(self):
 		payload, _ = performance.report(TELEGRAM, 7)
 		self.assertTrue(any('>Rob</a>' in line for line in payload), payload)
 
-	def test_prepare_holdings(self):
-		""".holdings without a portfolio lists the portfolios."""
-		self.assertTrue(holdings.prepare_holdings_payload(None, TELEGRAM))
+	def test_a_portfolio_command_without_a_portfolio_lists_them(self):
+		""".holdings and .pie without a portfolio say how to ask, with the portfolios there are."""
+		for text in ('.holdings', '.pie'):
+			with self.subTest(text=text):
+				self.assertIn('Portfolios: Rob', str(self.command(text)))
 
 	def test_prepare_marketcap(self):
 		"""The market cap list is built."""
diff --git a/tests/test_util.py b/tests/test_util.py
index 7fd4854c..f5232ed1 100644
--- a/tests/test_util.py
+++ b/tests/test_util.py
@@ -104,6 +104,13 @@ def test_date_range_english(self):
 		with self.assertRaises(TypeError):
 			formats.date_range_english(None, end) # callers must pass a period
 
+	def test_strip_url(self):
+		"""A link's text is its address without the scheme or www."""
+		for url, text in (('https://www.apple.com', 'apple.com'), ('http://example.org/a?b', 'example.org/a?b'),
+				('https://www3.example.com', 'www3.example.com'), ('apple.com', 'apple.com')):
+			with self.subTest(url=url):
+				self.assertEqual(formats.strip_url(url), text)
+
 	def test_ordinal(self):
 		"""Ordinal suffixes, the teens included."""
 		for number, suffix in ((1, 'st'), (2, 'nd'), (3, 'rd'), (4, 'th'), (11, 'th'), (12, 'th'), (13, 'th'), (21, 'st'), (112, 'th')):
diff --git a/util/sws_gen.py b/util/sws_gen.py
index a176df09..793f3438 100644
--- a/util/sws_gen.py
+++ b/util/sws_gen.py
@@ -49,20 +49,26 @@ def do_sws_bulk(file):
 
 def write_cache(mydict, cache_file):
 	"""Write the dict as JSON to cache_file, mode 0640."""
-	os.umask(0)
 	def opener(path, flags):
 		"""open() that creates the file with mode 0640."""
 		return os.open(path, flags, 0o640)
-	with open(cache_file, "w", opener=opener, encoding="utf-8") as f:
-		f.write(json.dumps(mydict))
-	os.umask(0o022)
+	old = os.umask(0)
+	try:
+		with open(cache_file, "w", opener=opener, encoding="utf-8") as f:
+			f.write(json.dumps(mydict))
+	finally:
+		os.umask(old)
 
-files = ['nasdaq100', 'sp500', 'asx200'] 
-for idx, file in enumerate(files):
-	print("Starting", file)
-	mydict = do_sws_bulk(file)
-	print("Completed", file, "as finbot_sws_" + file.split('.')[0] + ".json")
-	if idx + 1 < len(files):
-		print("Sleeping for 15 minutes")
-		time.sleep(15 * 60)
+def main(files=('nasdaq100', 'sp500', 'asx200')):
+	"""Look up the symbols listed in each file, 15 minutes apart."""
+	for idx, file in enumerate(files):
+		print("Starting", file)
+		do_sws_bulk(file)
+		print("Completed", file, "as finbot_sws_" + file.split('.')[0] + ".json")
+		if idx + 1 < len(files):
+			print("Sleeping for 15 minutes")
+			time.sleep(15 * 60)
+
+if __name__ == '__main__':
+	main()
 

From 9d29939d15b20ca50eb5c7c38f9ff683169c9ee4 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 10:15:53 +0000
Subject: [PATCH 65/81] Telegram: messages and charts go through the one Bot
 API call, which says why Telegram refused

sendMessage and sendPhoto were posted by hand, so a refusal logged "400 error outbound to telegram" (the message) or
the raw answer (the chart). They go through telegram.call now, as the edits and acknowledgements already did: a
refusal logs Telegram's reason ("Bad Request: can't parse entities"), a 200 that is not JSON counts as a failure, and a
timeout on a chart is still MAYBE_SENT. sendPhoto is no longer sent disable_web_page_preview, which it does not take.
The chat id goes in the body; the answers are Telegram's messages, which the pinning reads.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/transports.py       | 42 ++++++++++++++++++++---------------------
 tests/test_providers.py | 41 ++++++++++++++++++++++++++++------------
 2 files changed, 50 insertions(+), 33 deletions(-)

diff --git a/lib/transports.py b/lib/transports.py
index 934a6aea..8f6e935b 100644
--- a/lib/transports.py
+++ b/lib/transports.py
@@ -155,37 +155,36 @@ def markup(buttons):
 			'callback_data': button.data} for button in row] for row in buttons]}
 
 	def post_text(self, text, buttons=None):
-		"""sendMessage."""
-		payload = {'text': text, 'parse_mode': 'HTML', 'disable_web_page_preview': 'true',
-			'disable_notification': 'true'}
+		"""sendMessage: Telegram's message, or None when it refused or failed (logged with Telegram's reason, such as
+		"can't parse entities")."""
+		payload = {'chat_id': self.where, 'text': text, 'parse_mode': 'HTML', 'disable_web_page_preview': True,
+			'disable_notification': True}
 		if buttons:
 			payload['reply_markup'] = self.markup(buttons)
-		return self._post_json(config.env.webhooks['telegram'] + 'sendMessage?chat_id=' + str(self.where), payload)
+		try:
+			message = telegram.call('sendMessage', json=payload)
+		except Exception as e:
+			print("Telegram sendMessage to", self.where, "failed:", net.redact(e), file=sys.stderr)
+			return None
+		print("OK telegram message to", self.where, file=sys.stderr)
+		return message
 
 	def post_chart(self, image, caption, buttons=None):
-		"""sendPhoto, the report as its caption."""
-		data = {'disable_notification': True, 'chat_id': self.where, 'caption': caption, 'parse_mode': 'HTML',
-			'disable_web_page_preview': True}
+		"""sendPhoto, the report as its caption: Telegram's message, MAYBE_SENT, or None when it refused or failed
+		(logged with Telegram's reason)."""
+		data = {'chat_id': self.where, 'caption': caption, 'parse_mode': 'HTML', 'disable_notification': True}
 		if buttons:
 			data['reply_markup'] = json.dumps(self.markup(buttons))
 		try:
-			r = net.post(config.env.webhooks['telegram'] + 'sendPhoto?chat_id=' + str(self.where), data=data,
-				files={'photo': ('image.png', image)})
+			message = telegram.call('sendPhoto', data=data, files={'photo': ('image.png', image)})
 		except requests.exceptions.ReadTimeout as e:
 			print("Telegram sendPhoto timed out, the chart may still arrive:", net.redact(e), file=sys.stderr)
 			return MAYBE_SENT
 		except Exception as e:
-			print("Telegram sendPhoto failed:", net.redact(e), file=sys.stderr)
+			print("Telegram sendPhoto to", self.where, "failed:", net.redact(e), file=sys.stderr)
 			return None
-		try:
-			output = r.json()
-		except ValueError:
-			output = {}
-		if r.status_code == 200 and output.get('ok'):
-			return output
-		print(r.status_code, "error telegram sendPhoto", output.get('error_code'),
-			output.get('description') or r.text[:300], file=sys.stderr)
-		return None
+		print("OK telegram chart to", self.where, file=sys.stderr)
+		return message
 
 	def edit_chart(self, message, image, caption, buttons=None):
 		"""Replace a chart message's image, caption and buttons in place. Raises RuntimeError when Telegram refuses,
@@ -223,11 +222,12 @@ def acknowledge(self, press, text=None):
 		telegram.answerCallbackQuery(press, text)
 
 	def pin(self, answers):
-		"""Pin the messages these answers are for, last first, so the first ends up on top, without the notification."""
+		"""Pin the messages these answers are (Telegram's messages), last first, so the first ends up on top, without the
+		notification."""
 		sent = []
 		for answer in answers:
 			try:
-				sent.append((answer['result']['chat']['id'], answer['result']['message_id']))
+				sent.append((answer['chat']['id'], answer['message_id']))
 			except (KeyError, TypeError):
 				print("Unable to pin: unexpected response", file=sys.stderr)
 		for chat, message in reversed(sent):
diff --git a/tests/test_providers.py b/tests/test_providers.py
index 07a99419..be05c054 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -638,18 +638,35 @@ def fake(url, **kwargs):
 		self.patch(transports.net, 'post', fake)
 		return transports.TelegramBot('5').post_text('hi')
 
-	def test_success_returns_json(self):
-		"""A sent message returns Telegram's answer."""
-		self.assertEqual(self.post(Resp({'ok': True})), {'ok': True})
-
-	def test_http_error_and_exception_return_none(self):
-		"""A refused or failed send returns None."""
-		self.assertIsNone(self.post(Resp(status=429)))
-		self.assertIsNone(self.post(ConnectionError('down')))
-
-	def test_non_json_success_returns_empty_dict(self):
-		"""A 200 that is not JSON returns {}."""
-		self.assertEqual(self.post(Resp(None)), {})
+	def test_success_returns_the_message(self):
+		"""A sent message returns the message, as Telegram gives it."""
+		self.assertEqual(self.post(Resp({'ok': True, 'result': {'message_id': 7}})), {'message_id': 7})
+
+	def test_a_refusal_or_failure_returns_none_and_logs_why(self):
+		"""A refused or failed send returns None, and the log has Telegram's reason; a 200 that is not JSON is no
+		message either."""
+		refusal = Resp({'ok': False, 'error_code': 400, 'description': "Bad Request: can't parse entities"}, 400)
+		for outcome in (refusal, Resp(status=429), ConnectionError('down'), Resp(None)):
+			with self.subTest(outcome=outcome), contextlib.redirect_stderr(io.StringIO()):
+				self.assertIsNone(self.post(outcome))
+		self.assertIn("can't parse entities", self.logged(refusal))
+
+	def logged(self, outcome):
+		"""What posting with this outcome logs."""
+		with contextlib.redirect_stderr(io.StringIO()) as log:
+			self.post(outcome)
+		return log.getvalue()
+
+	def test_a_chart_goes_without_a_parameter_sendphoto_does_not_take(self):
+		"""sendPhoto is sent the chat, caption and markup, and no link-preview flag: a photo has no preview."""
+		sent = []
+		self.patch(transports.net, 'post', lambda url, **kwargs: sent.append((url, kwargs)) or
+			Resp({'ok': True, 'result': {'message_id': 8}}))
+		self.assertEqual(transports.TelegramBot('5').post_chart(io.BytesIO(b'png'), 'caption'), {'message_id': 8})
+		url, kwargs = sent[0]
+		self.assertTrue(url.endswith('/sendPhoto'))
+		self.assertEqual(kwargs['data'], {'chat_id': '5', 'caption': 'caption', 'parse_mode': 'HTML',
+			'disable_notification': True})
 
 	def test_error_line_redacts_secrets_and_truncates(self):
 		"""An error line has no token or crumb in it, and is cut to about 200 characters."""

From be8074658d9f6d4fd9ce3afaf355bb0900e2e785 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 10:21:33 +0000
Subject: [PATCH 66/81] worker: a message and a button press arrive as
 inbound's Message and Press

process_request took seven positional arguments and process_callback five, which bot.py unpacked from the Message and
Press that inbound had already made. They take those objects now, so a way of receiving updates other than a webhook
(a pull mode) hands over the same thing, and the .jobs button handler takes the Press too. The test recorder hands
over either form, so it still compares with commits from before.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 bot.py                  |  4 +--
 lib/inbound.py          |  2 +-
 lib/transports.py       |  4 +--
 lib/worker.py           | 55 ++++++++++++++++++++++-------------------
 tests/common.py         |  2 +-
 tests/test_edges.py     | 33 +++++++++++++------------
 tests/test_jobs.py      | 14 +++++------
 tests/test_providers.py | 26 +++++++++----------
 tests/test_reports.py   |  2 +-
 util/record_outputs.py  | 15 ++++++++---
 10 files changed, 83 insertions(+), 74 deletions(-)

diff --git a/bot.py b/bot.py
index 09b210d1..305ea05d 100755
--- a/bot.py
+++ b/bot.py
@@ -117,13 +117,11 @@ def dispatch(event):
 	if isinstance(event, inbound.Press):
 		print(f"[{event.service.capitalize()} button]:", event.user, "chat", event.chat_id, event.data, file=sys.stderr)
 		target = worker.process_callback
-		args = (event.service, event.press_id, event.chat_id, event.message_id, event.data)
 	else:
 		if config.prefs.debug:
 			print(f"[{event.service.capitalize()}]:", event.user, event.text, file=sys.stderr)
 		target = worker.process_request
-		args = (event.service, event.chat_id, event.user, event.text, event.bot_name, event.user_name, event.message_id)
-	threading.Thread(target=target, args=args).start()
+	threading.Thread(target=target, args=(event,)).start()
 
 def git_version(checkout=None):
 	"""Short hash of the checked-out commit, read straight from the .git directory: no git binary to run, and no "dubious
diff --git a/lib/inbound.py b/lib/inbound.py
index b365a8fb..84376299 100644
--- a/lib/inbound.py
+++ b/lib/inbound.py
@@ -34,7 +34,7 @@ class Press:
 	chat_id: str
 	message_id: str
 	data: str
-	user: str # who pressed it, for the log
+	user: str = '' # who pressed it, for the log
 
 
 @dataclasses.dataclass
diff --git a/lib/transports.py b/lib/transports.py
index 8f6e935b..02754730 100644
--- a/lib/transports.py
+++ b/lib/transports.py
@@ -222,8 +222,8 @@ def acknowledge(self, press, text=None):
 		telegram.answerCallbackQuery(press, text)
 
 	def pin(self, answers):
-		"""Pin the messages these answers are (Telegram's messages), last first, so the first ends up on top, without the
-		notification."""
+		"""Pin the messages that these answers are (Telegram's), last first, so that the first ends up on top, without
+		the notification."""
 		sent = []
 		for answer in answers:
 			try:
diff --git a/lib/worker.py b/lib/worker.py
index ceb5371b..d1c203d0 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -41,36 +41,38 @@ def reply(self, lines, chart=None, buttons=None):
 		can go) and buttons."""
 		return delivery.deliver(self.dest, delivery.Report(lines, chart, buttons))
 
-def process_job_callback(service, callback_id, chat_id, message_id, data):
-	"""A command button of .jobs: run that command now and answer in the chat."""
-	chat = delivery.reply_to(service, chat_id)
+def process_job_callback(press):
+	"""A command button of .jobs (an inbound.Press): run that command now and answer in the chat."""
+	service = dialects.of(press.service)
+	chat = delivery.reply_to(service, press.chat_id)
 	try:
 		if jobs.current is None: # the scheduler is off now
 			raise jobs.ScheduleError('the scheduler is off')
-		_, number, crc = data.split('|') # only a command offered by .jobs can be picked; anything else is refused
+		_, number, crc = press.data.split('|') # only a command offered by .jobs can be picked; anything else is refused
 		command = jobs.current.pick(number, crc)
 	except (ValueError, jobs.ScheduleError):
-		chat.acknowledge(callback_id, 'Buttons updated, try again')
+		chat.acknowledge(press.press_id, 'Buttons updated, try again')
 		payload, markup = about.prepare_jobs_payload(service, '') # the list as it is now, in place of the stale one
-		chat.edit_text(message_id, '\n'.join(payload), markup or [])
+		chat.edit_text(press.message_id, '\n'.join(payload), markup or [])
 		return
-	chat.acknowledge(callback_id, f'Running {jobs.display(command[0])}…'[:100])
+	chat.acknowledge(press.press_id, f'Running {jobs.display(command[0])}…'[:100])
 	try:
 		line = about.job_line(jobs.current, *jobs.current.run_command(command))
 		delivery.deliver(chat, delivery.Report([line]))
 	except Exception as e:
 		delivery.report_error(e, chat, context='jobs')
 
-def process_callback(service, callback_id, chat_id, message_id, data):
-	"""Button press: run a scheduled command from .jobs, or rebuild a chart message for the chosen period and edit it in place."""
-	service = dialects.of(service)
-	if data.startswith('j|'):
-		return process_job_callback(service, callback_id, chat_id, message_id, data)
-	parts = data.split('|', 3)
+def process_callback(press):
+	"""A button pressed (an inbound.Press): run a scheduled command from .jobs, or rebuild a chart message for the
+	chosen period and edit it in place."""
+	if press.data.startswith('j|'):
+		return process_job_callback(press)
+	service = dialects.of(press.service)
+	parts = press.data.split('|', 3)
 	valid = len(parts) == 4 and parts[0] == 'c' and parts[1] in buttons.BUTTONS and parts[2] in buttons.BUTTONS[parts[1]]
-	chat = delivery.reply_to(service, chat_id)
+	chat = delivery.reply_to(service, press.chat_id)
 	# answering now shows Telegram's toast bubble ("Loading 1Y…") while the new chart is built
-	chat.acknowledge(callback_id, f"Loading {buttons.PERIODS[parts[2]][0]}…" if valid else None)
+	chat.acknowledge(press.press_id, f"Loading {buttons.PERIODS[parts[2]][0]}…" if valid else None)
 	if not valid:
 		return
 	try:
@@ -79,19 +81,20 @@ def process_callback(service, callback_id, chat_id, message_id, data):
 		caption, chart = charts.build(kind, args, period, service)
 		if chart is None: # too few prices in that period to draw a line
 			raise answer.Answer(f"Too few prices to draw {buttons.PERIODS[period][0]}")
-		chat.edit_chart(message_id, render.draw(chart), caption, buttons.keyboard(kind, args, period))
+		chat.edit_chart(press.message_id, render.draw(chart), caption, buttons.keyboard(kind, args, period))
 	except answer.Answer as e: # nothing to draw for that period: said in the chat, as a command would
 		delivery.deliver(chat, delivery.Report(e.lines))
 	except Exception as e:
 		delivery.report_error(e, chat, context='chart')
 
-def process_request(service, chat_id, user, message, botName, userRealName, message_id=None):
-	"""Entry point for inbound chat requests: the command the message calls (see read) answers it, with the typing
-	notice meanwhile, which the first message of the answer stops and the end of the command takes off; a message that
-	calls none gets no answer. Arguments the command cannot take get its usage, and any failure a one-line reply in the
-	originating chat. message_id: the message's own id, where the network has one."""
-	request = Request(dialects.of(service), chat_id, user, userRealName, botName, message_id)
-	command, args = read(message, botName)
+def process_request(message):
+	"""A chat message (an inbound.Message): the command it calls (see read) answers it, with the typing notice
+	meanwhile, which the first message of the answer stops and the end of the command takes off; a message that calls
+	none gets no answer. Arguments the command cannot take get its usage, and any failure a one-line reply in the
+	chat."""
+	request = Request(dialects.of(message.service), message.chat_id, message.user, message.user_name, message.bot_name,
+		message.message_id)
+	command, args = read(message.text, message.bot_name)
 	if command:
 		try:
 			request.dest.start_typing()
@@ -101,12 +104,12 @@ def process_request(service, chat_id, user, message, botName, userRealName, mess
 		except answer.Answer as e: # a portfolio the user got wrong, a ticker with no prices: an answer, not an error
 			request.reply(e.lines)
 		except Exception as e:
-			delivery.report_error(e, request.dest, context=message.split()[0][:30])
+			delivery.report_error(e, request.dest, context=message.text.split()[0][:30])
 		finally:
 			request.dest.end_typing()
-	if service == 'telegram':
+	if message.service == 'telegram':
 		try:
-			ensure_keyboard(chat_id) # keep DM keyboards current after the reply
+			ensure_keyboard(message.chat_id) # keep DM keyboards current after the reply
 		except Exception as e:
 			print("keyboard update failed:", e, file=sys.stderr)
 
diff --git a/tests/common.py b/tests/common.py
index 864adb63..b0a0914b 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -153,6 +153,6 @@ def command(self, text, chat='55'):
 		self.sent.clear()
 		self.markups.clear()
 		self.errors.clear()
-		worker.process_request('telegram', chat, '@u', text, BOT, 'U')
+		worker.process_request(inbound.Message('telegram', chat, '@u', text, BOT, 'U'))
 		self.assertEqual(self.errors, [], f'{text} crashed') # the worker reports crashes instead of raising
 		return list(self.sent)
diff --git a/tests/test_edges.py b/tests/test_edges.py
index bb488548..efc100a0 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -39,7 +39,7 @@ def down(*a, **k):
 		for cmd in ('.price', '.peg', '.T1', '.compare T1 T2'):
 			with self.subTest(cmd=cmd):
 				self.sent.clear(); self.errors.clear()
-				worker.process_request('telegram', '55', '@u', cmd, BOT, 'U')
+				worker.process_request(inbound.Message('telegram', '55', '@u', cmd, BOT, 'U'))
 				self.assertEqual(len(self.errors), 1, cmd) # exactly one one-line report, not a traceback to the user
 				self.assertTrue(self.errors[0].startswith('⚠️'))
 
@@ -51,7 +51,7 @@ def down(*a, **k):
 		self.patch(sharesight, '_performance', down)
 		self.patch(sharesight, '_portfolios', down)
 		self.sent.clear(); self.errors.clear()
-		worker.process_request('telegram', '55', '@u', '.performance 7d', BOT, 'U')
+		worker.process_request(inbound.Message('telegram', '55', '@u', '.performance 7d', BOT, 'U'))
 		self.assertEqual(len(self.errors), 1)
 
 	def test_the_typing_notice_stops_before_any_message_goes_out(self):
@@ -71,7 +71,7 @@ def down(*a, **k):
 		for cmd in ('.help', '.price', '.trades 7d', '.performance 7d'):
 			with self.subTest(cmd=cmd), contextlib.redirect_stderr(io.StringIO()):
 				typing.clear()
-				worker.process_request('telegram', '55', '@u', cmd, BOT, 'U')
+				worker.process_request(inbound.Message('telegram', '55', '@u', cmd, BOT, 'U'))
 				self.assertEqual(typing, [False])
 
 	def test_no_holdings_and_empty_watchlist(self):
@@ -149,7 +149,7 @@ def no_history(t, *a, **k):
 		self.assertEqual(self.command('.history T1'), [('text', ['no price history for T1'])])
 		self.sent.clear()
 		self.patch(telegram, 'answerCallbackQuery', lambda *a, **k: None)
-		worker.process_callback('telegram', 'cb', '55', '9', 'c|h|m|T1')
+		worker.process_callback(inbound.Press('telegram', 'cb', '55', '9', 'c|h|m|T1'))
 		self.assertEqual((self.sent, self.errors), ([('text', ['no price history for T1'])], []))
 
 	def test_history_without_a_ticker_or_with_extra_argument_gives_usage(self):
@@ -418,7 +418,7 @@ def setUp(self):
 		self.env(slackOutgoingToken='stok')
 
 		self.requests = []
-		self.patch(worker, 'process_request', lambda *a: self.requests.append(a))
+		self.patch(worker, 'process_request', self.requests.append)
 
 	def event(self, event_id=None, **event):
 		"""Post a Slack message event to the WSGI handler and return the status."""
@@ -443,9 +443,9 @@ def test_a_button_press_reaches_the_worker(self):
 		"""A period button pressed under a chart reaches the worker, posted to the events URL: the channel, the message
 		and the button's value. A press with a wrong token is refused, and one not under a message goes nowhere."""
 		presses = []
-		self.patch(worker, 'process_callback', lambda *a: presses.append(a))
+		self.patch(worker, 'process_callback', presses.append)
 		self.assertEqual(self.press(PRESS), '200 OK')
-		self.assertEqual(presses, [('slack', 't1', 'C1', '1.5', 'c|h|w|T1')])
+		self.assertEqual(presses, [inbound.Press('slack', 't1', 'C1', '1.5', 'c|h|w|T1', '@rob')])
 		self.assertEqual(self.press({**PRESS, 'token': 'wrong'}), '403 Forbidden')
 		self.assertEqual(self.press({**PRESS, 'container': {'type': 'view', 'view_id': 'V1'}}), '200 OK')
 		self.assertEqual(self.press({**PRESS, 'actions': []}), '200 OK')
@@ -454,7 +454,7 @@ def test_a_button_press_reaches_the_worker(self):
 	def test_the_message_goes_to_the_worker_with_its_ts(self):
 		"""A message reaches the worker with its ts, the message's id: finbot puts ⏳ on it while it works."""
 		self.assertEqual(self.event(user='U1', text='.price'), '200 OK')
-		self.assertEqual(self.requests[0][6], '1.2')
+		self.assertEqual(self.requests[0].message_id, '1.2')
 
 	def test_only_people_are_answered(self):
 		"""Messages from bots, this bot included, joins, edits and events without a user are not answered; people's
@@ -470,7 +470,7 @@ def test_only_people_are_answered(self):
 		self.assertEqual((self.requests, self.errors), ([], []))
 		self.assertEqual(self.event(user='U1', text='.price <@UBOT>'), '200 OK')
 		self.assertEqual(self.event(user='U2', subtype='thread_broadcast', text='.trades'), '200 OK')
-		self.assertEqual([(r[0], r[1], r[2], r[3], r[4]) for r in self.requests],
+		self.assertEqual([(r.service, r.chat_id, r.user, r.text, r.bot_name) for r in self.requests],
 						 [('slack', 'C1', '<@U1>', '.price @UBOT', '@UBOT'), ('slack', 'C1', '<@U2>', '.trades', '@UBOT')])
 
 	def test_links_and_escapes_lose_their_markup(self):
@@ -481,7 +481,7 @@ def test_links_and_escapes_lose_their_markup(self):
 				('.plan buy &lt;BHP&gt;', '.plan buy <BHP>')):
 			with self.subTest(text=text):
 				self.assertEqual(self.event(user='U1', text=text), '200 OK')
-				self.assertEqual(self.requests[-1][3], expected)
+				self.assertEqual(self.requests[-1].text, expected)
 
 	def test_a_retried_event_runs_once(self):
 		"""An event Slack sends again (same event_id) runs once."""
@@ -506,7 +506,7 @@ def setUp(self):
 		super().setUp()
 		self.env(slackOutgoingWebhook='https://example.com/slack', slackOutgoingToken='stok', slackSigningSecret='sss')
 		self.requests = []
-		self.patch(worker, 'process_request', lambda *a: self.requests.append(a))
+		self.patch(worker, 'process_request', self.requests.append)
 
 	def post(self, secret='sss', age=0, token='stok', headers=None, tamper=False):
 		"""Post a Slack message event signed with `secret` `age` seconds ago (or with these headers instead) and
@@ -527,7 +527,7 @@ def post(self, secret='sss', age=0, token='stok', headers=None, tamper=False):
 	def test_a_button_press_is_signed_over_its_form(self):
 		"""A button press is signed like an event, over the form it comes as; altered, it is refused."""
 		presses = []
-		self.patch(worker, 'process_callback', lambda *a: presses.append(a))
+		self.patch(worker, 'process_callback', presses.append)
 		body = urlencode({'payload': json.dumps({**PRESS, 'token': 'not needed'})}).encode()
 		timestamp = str(int(time.time()))
 		mac = hmac.new(b'sss', b'v0:' + timestamp.encode() + b':' + body, hashlib.sha256).hexdigest()
@@ -538,7 +538,7 @@ def test_a_button_press_is_signed_over_its_form(self):
 				bot.main(wsgi_environ('/slack', sent, **headers), lambda s, *a, **k: answers.append(s))
 			gevent.sleep(0.05)
 			self.assertEqual(answers, [status])
-		self.assertEqual(presses, [('slack', 't1', 'C1', '1.5', 'c|h|w|T1')])
+		self.assertEqual(presses, [inbound.Press('slack', 't1', 'C1', '1.5', 'c|h|w|T1', '@rob')])
 
 	def test_a_signed_request_is_answered_and_the_token_is_not_needed(self):
 		"""A request signed with the secret is answered, whatever token it carries."""
@@ -565,7 +565,8 @@ def test_a_plan_is_shown_as_text(self):
 		self.assertIn('<i>buy &lt;cheap&gt; &amp; hold</i>', reply[0])
 		self.assertEqual(self.command('.plan')[0][1], reply) # and the same way again
 		self.sent.clear()
-		worker.process_request('slack', 'C1', '<@U1>', '.plan sell <dear>', BOT, 'U') # inbound unescapes Slack's text
+		# inbound has unescaped Slack's text
+		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.plan sell <dear>', BOT, 'U'))
 		text = '\n'.join(self.sent[0][1])
 		self.assertIn('*<@U1>*', text) # a Slack mention stays markup
 		self.assertIn('_sell &lt;dear&gt;_', text) # escaped once, not twice
@@ -574,13 +575,13 @@ def test_a_plan_is_shown_as_text(self):
 	def test_a_telegram_name_is_escaped_where_it_arrives(self):
 		"""A Telegram first name is escaped as the update arrives, in the handle and the name."""
 		requests = []
-		self.patch(worker, 'process_request', lambda *a: requests.append(a))
+		self.patch(worker, 'process_request', requests.append)
 		body = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A<b>&'}, 'text': '.hello'}}
 		env = wsgi_environ('/telegram', json.dumps(body).encode(), HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN='tok')
 		with contextlib.redirect_stderr(io.StringIO()):
 			bot.main(env, lambda *a, **k: None)
 		gevent.sleep(0.05) # the worker runs in a (gevent) thread
-		self.assertEqual((requests[0][2], requests[0][5]), ('@A&lt;b&gt;&amp;', '@A&lt;b&gt;&amp;')) # the handle and the name
+		self.assertEqual((requests[0].user, requests[0].user_name), ('@A&lt;b&gt;&amp;',) * 2) # the handle and the name
 
 
 class SecretsStayOutOfLogs(FinbotCase):
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
index a0951aa7..434fa081 100644
--- a/tests/test_jobs.py
+++ b/tests/test_jobs.py
@@ -592,7 +592,7 @@ def test_the_chat_command_passes_the_keyboard_on(self):
 		""".jobs sends its buttons with the list."""
 		seen = []
 		self.patch(transports.Transport, 'text', lambda dest, lines, buttons=None: seen.append(buttons) or [])
-		worker.process_request('telegram', '55', '@u', '.jobs', BOT, 'U')
+		worker.process_request(inbound.Message('telegram', '55', '@u', '.jobs', BOT, 'U'))
 		self.assertEqual(len(seen[0][0]), 2) # two to a row
 
 	def test_button_press_runs_that_command(self):
@@ -602,7 +602,7 @@ def test_button_press_runs_that_command(self):
 		markup = about.prepare_jobs_payload(TELEGRAM, '')[1]
 		data = next(b.data for row in markup for b in row if b.label.startswith('7.'))
 		self.sent.clear()
-		worker.process_callback('telegram', 'cb1', '55', '9', data)
+		worker.process_callback(inbound.Press('telegram', 'cb1', '55', '9', data))
 		self.assertEqual(self.ran, [('price', {'mode': 'close'})])
 		self.assertEqual(self.sent, [('text', ['7. price close: done'])])
 		self.assertEqual(sent, ['Running price close…'])
@@ -623,7 +623,7 @@ def test_forged_button_data_never_runs_anything_else(self):
 					 'j|8|' + good.upper() + 'x', 'j|99999999999999999999|' + good):
 			self.sent.clear()
 			self.toasts.clear(); self.edits.clear()
-			worker.process_callback('telegram', 'cb', '55', '9', data)
+			worker.process_callback(inbound.Press('telegram', 'cb', '55', '9', data))
 			self.assertEqual(self.ran, [], data)
 			self.assertEqual(self.toasts, ['Buttons updated, try again'], data)
 			self.assertEqual(len(self.edits), 1, data)
@@ -633,14 +633,14 @@ def test_forged_button_data_never_runs_anything_else(self):
 			for n in range(1, 12):
 				with self.assertRaises(jobs.ScheduleError):
 					self.s.pick(str(n), jobs.checksum(label))
-		worker.process_callback('telegram', 'cb', '55', '9', 'j|8|' + good) # the genuine button still works
+		worker.process_callback(inbound.Press('telegram', 'cb', '55', '9', 'j|8|' + good)) # the genuine one still works
 		self.assertEqual(self.ran, [('trades', {})])
 
 	def test_button_from_an_old_schedule_updates_the_message(self):
 		"""A stale button runs nothing; the message gets the current list and buttons (scheduler off: a notice)."""
 		self.stale_buttons()
-		worker.process_callback('telegram', 'cb1', '55', '9', 'j|7|0000')
-		worker.process_callback('telegram', 'cb1', '55', '9', 'j|9|0000')
+		worker.process_callback(inbound.Press('telegram', 'cb1', '55', '9', 'j|7|0000'))
+		worker.process_callback(inbound.Press('telegram', 'cb1', '55', '9', 'j|9|0000'))
 		self.assertEqual(self.ran, [])
 		self.assertEqual(self.sent, [])
 		self.assertEqual(self.toasts, ['Buttons updated, try again'] * 2)
@@ -649,7 +649,7 @@ def test_button_from_an_old_schedule_updates_the_message(self):
 		self.assertIn('7. price close', text)
 		self.assertEqual(markup, about.prepare_jobs_payload(TELEGRAM, '')[1]) # with the current buttons
 		self.patch(jobs, 'current', None) # scheduler off: the list is replaced by the notice and the buttons removed
-		worker.process_callback('telegram', 'cb1', '55', '9', 'j|7|0000')
+		worker.process_callback(inbound.Press('telegram', 'cb1', '55', '9', 'j|7|0000'))
 		self.assertEqual(self.edits[-1][2:], ('Scheduled jobs are off (scheduler=False)', [])) # no buttons left
 
 	def test_run_in_chat(self):
diff --git a/tests/test_providers.py b/tests/test_providers.py
index be05c054..bdb820c5 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -321,7 +321,7 @@ def get(url, **kwargs):
 		def slack(text):
 			"""Run a command on Slack, where reports come as text, and return the text."""
 			self.sent.clear(); self.errors.clear()
-			worker.process_request('slack', 'C1', '<@U1>', text, BOT, 'U')
+			worker.process_request(inbound.Message('slack', 'C1', '<@U1>', text, BOT, 'U'))
 			self.assertEqual(self.errors, [], text)
 			return '\n'.join(map(str, self.sent[-1][1]))
 		price_list = slack('.price 7d') # T3 is left out, the rest are listed
@@ -332,7 +332,7 @@ def slack(text):
 		self.assertIn('NASDAQ 100', report)
 		self.command('.T3') # the profile goes out without its history lines
 		self.sent.clear(); self.errors.clear()
-		worker.process_request('telegram', '55', '@u', '.price T3 7d', BOT, 'U')
+		worker.process_request(inbound.Message('telegram', '55', '@u', '.price T3 7d', BOT, 'U'))
 		self.assertEqual(len(self.errors), 1) # the one ticker asked for: one line saying why
 		self.assertTrue(self.errors[0].startswith('⚠️ .price: RuntimeError: 404 error for T3'))
 
@@ -773,7 +773,7 @@ def test_slack_replies_upload_their_charts(self):
 			with self.subTest(cmd=cmd):
 				calls = self.slack_answers(*self.upload_steps())
 				self.sent.clear(); self.errors.clear()
-				worker.process_request('slack', 'C1', '<@U1>', cmd, BOT, 'U')
+				worker.process_request(inbound.Message('slack', 'C1', '<@U1>', cmd, BOT, 'U'))
 				self.assertEqual(self.errors, [])
 				self.assertEqual([url for url, _ in calls], ['https://slack.com/api/files.getUploadURLExternal', self.UPLOAD_URL,
 															 'https://slack.com/api/files.completeUploadExternal'])
@@ -839,7 +839,7 @@ def refuse(*a, **k):
 		for cmd in ('.beta', '.pie Rob', '.buy', '.sell', '.peg', '.price', '.price top', '.price T1 7d', '.performance 7d', '.shorts', '.history T1', '.compare T1 T2'):
 			with self.subTest(cmd=cmd):
 				self.sent.clear(); self.errors.clear()
-				worker.process_request('slack', 'C1', '<@U1>', cmd, BOT, 'U')
+				worker.process_request(inbound.Message('slack', 'C1', '<@U1>', cmd, BOT, 'U'))
 				self.assertEqual(self.errors, [])
 				self.assertTrue(self.sent)
 				self.assertEqual({kind for kind, *_ in self.sent}, {'text'})
@@ -856,7 +856,7 @@ def test_the_graph_preference_off_sends_no_chart_anywhere(self):
 			for cmd in ('.price', '.price top', '.performance 7d', '.shorts', '.buy', '.peg', '.beta', '.compare T1 T2', '.pie Rob'):
 				with self.subTest(service=service, cmd=cmd):
 					self.sent.clear(); self.errors.clear()
-					worker.process_request(service, chat, '@u', cmd, BOT, 'U')
+					worker.process_request(inbound.Message(service, chat, '@u', cmd, BOT, 'U'))
 					self.assertEqual(self.errors, [])
 					self.assertEqual({kind for kind, *_ in self.sent}, {'text'})
 		self.assertEqual([url for url in posts if 'sendPhoto' in url or 'files.' in url], [])
@@ -970,7 +970,7 @@ def test_a_reply_without_a_bot_token_is_logged_not_sent(self):
 		err = io.StringIO()
 		with contextlib.redirect_stderr(err), contextlib.redirect_stdout(io.StringIO()):
 			for cmd in ('.help', '.price', '.compare T1 T2'):
-				worker.process_request('slack', 'C1', '<@U1>', cmd, BOT, 'U')
+				worker.process_request(inbound.Message('slack', 'C1', '<@U1>', cmd, BOT, 'U'))
 		self.assertEqual((self.posts, self.errors), ([], []))
 		self.assertEqual(err.getvalue().count('ERROR slack: a reply to channel C1 needs slackBotToken in finbot.env; not sent'), 3)
 
@@ -989,7 +989,7 @@ def post(url, **kwargs):
 		self.patch(transports.net, 'post', post)
 		help_text = about.prepare_help
 		self.patch(about, 'prepare_help', lambda *a: (added.wait(5), help_text(*a))[1]) # a command slow enough for ⏳
-		worker.process_request('slack', 'C1', '<@U1>', '.help', BOT, 'U', '1.2')
+		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.help', BOT, 'U', '1.2'))
 		self.assertEqual([url.rsplit('/', 1)[-1] for url, _ in self.posts],
 			['reactions.add', 'chat.postMessage', 'reactions.remove'])
 		reaction = {'channel': 'C1', 'timestamp': '1.2', 'name': 'hourglass_flowing_sand'}
@@ -1056,7 +1056,7 @@ def blocks(self, method):
 	def test_a_chart_reply_carries_its_period_buttons(self):
 		"""A chart reply is a Block Kit message: the report, the chart (uploaded into no channel, shown by its id) and
 		the period buttons, the one shown marked; nothing goes as text."""
-		worker.process_request('slack', 'C1', '<@U1>', '.price', BOT, 'U')
+		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U'))
 		self.assertEqual(self.errors, [])
 		self.assertEqual(self.calls(), ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal',
 			'chat.postMessage'])
@@ -1073,7 +1073,7 @@ def test_a_chart_reply_carries_its_period_buttons(self):
 	def test_a_press_redraws_the_chart_in_place(self):
 		"""A press uploads the chart for that period (a file cannot be replaced) and puts it in the message with
 		chat.update, the period pressed now marked; Telegram is not asked to answer the press."""
-		worker.process_callback('slack', 't1', 'C1', '1.5', 'c|l|w|5,0')
+		worker.process_callback(inbound.Press('slack', 't1', 'C1', '1.5', 'c|l|w|5,0'))
 		self.assertEqual(self.errors, [])
 		self.assertEqual(self.calls(), ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal',
 			'chat.update'])
@@ -1086,7 +1086,7 @@ def test_a_press_redraws_the_chart_in_place(self):
 	def test_a_chart_not_ready_to_show_is_shown_a_moment_later(self):
 		"""Slack refusing the blocks of a chart just uploaded (invalid_blocks) is asked again after a wait."""
 		self.refusals['chat.postMessage'] = ['invalid_blocks']
-		worker.process_request('slack', 'C1', '<@U1>', '.price', BOT, 'U')
+		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U'))
 		self.assertEqual((self.calls()[-2:], self.waits), (['chat.postMessage', 'chat.postMessage'], [1]))
 		self.assertEqual((self.sent, self.errors), ([], []))
 
@@ -1094,7 +1094,7 @@ def test_buttons_slack_will_not_show_leave_a_plain_chart(self):
 		"""When Slack will not show the chart with its buttons, the chart is uploaded into the channel without them."""
 		self.refusals['chat.postMessage'] = ['invalid_blocks'] * 4
 		with contextlib.redirect_stderr(io.StringIO()):
-			worker.process_request('slack', 'C1', '<@U1>', '.price', BOT, 'U')
+			worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U'))
 		self.assertEqual(self.calls()[-3:], ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal'])
 		self.assertEqual(self.posts[-1][1]['data']['channel_id'], 'C1')
 		self.assertEqual((self.waits, self.sent, self.errors), ([1, 2, 4], [], []))
@@ -1102,7 +1102,7 @@ def test_buttons_slack_will_not_show_leave_a_plain_chart(self):
 	def test_without_slackOutgoingWebhook_a_chart_has_no_buttons(self):
 		"""Without slackOutgoingWebhook, no press could reach finbot, so a chart goes without buttons."""
 		self.env(slackOutgoingWebhook='')
-		worker.process_request('slack', 'C1', '<@U1>', '.price', BOT, 'U')
+		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U'))
 		self.assertEqual(self.calls(), ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal'])
 		self.assertEqual(self.posts[-1][1]['data']['channel_id'], 'C1')
 
@@ -1125,7 +1125,7 @@ def post(url, **kwargs):
 			return record(url, **kwargs)
 		self.patch(transports.net, 'post', post)
 		with contextlib.redirect_stderr(io.StringIO()):
-			worker.process_request('slack', 'C1', '<@U1>', '.price', BOT, 'U')
+			worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U'))
 		self.assertEqual(self.calls(), ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal',
 			'chat.postMessage'])
 		self.assertEqual(self.errors, [])
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 690c9200..7e6fa191 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -508,7 +508,7 @@ def press(self, data):
 		self.patch(transports.TelegramBot, 'edit_chart', lambda dest, msg, img, caption, buttons=None:
 			edits.append((caption.split('\n')[0], transports.TelegramBot.markup(buttons))))
 		self.patch(telegram, 'answerCallbackQuery', lambda cid, text=None: toasts.append(text))
-		worker.process_callback('telegram', 'cb', '55', '9', data)
+		worker.process_callback(inbound.Press('telegram', 'cb', '55', '9', data))
 		return edits, toasts
 
 	def test_pressing_1d_rebuilds_todays_list_and_moves_the_marker(self):
diff --git a/util/record_outputs.py b/util/record_outputs.py
index e31626d0..2fc1a2a5 100755
--- a/util/record_outputs.py
+++ b/util/record_outputs.py
@@ -10,7 +10,7 @@
 
 --keep DIR keeps the two recordings; --full shows changed lines whole. The exit status is 0 when the recordings match,
 1 when they differ."""
-import argparse, difflib, hashlib, io, json, os, random, subprocess, sys, tempfile, unittest
+import argparse, difflib, hashlib, inspect, io, json, os, random, subprocess, sys, tempfile, unittest
 from pathlib import Path
 
 REPO = Path(__file__).resolve().parent.parent
@@ -55,6 +55,13 @@ def record(output):
 		real_sending = ((webhook, 'payload_wrapper', common.REAL_PAYLOAD_WRAPPER),
 			(webhook, 'sendPhoto', common.REAL_SEND_PHOTO), (webhook, 'report_error', common.REAL_REPORT_ERROR))
 
+	def handing(func, kind, *fields):
+		"""A step that hands func the Message or Press (kind, its name in lib.inbound) of these fields, or the fields
+		one by one to a commit from before it took the object."""
+		if len(inspect.signature(func).parameters) > 1:
+			return lambda: func(*fields)
+		return lambda: func(getattr(common.inbound, kind)(*fields))
+
 	class Record(common.FinbotCase):
 		"""The steps, as one test, so that FinbotCase's fixtures (and their cleanup) apply."""
 		def runTest(self):
@@ -95,12 +102,12 @@ def step(label, run):
 				steps.append({'step': label, 'posts': list(posts), 'raised': raised})
 			for service, chat, user in (('telegram', '55', '@u'), ('slack', 'C1', '<@U1>')):
 				for text in COMMANDS:
-					run = lambda: worker.process_request(service, chat, user, text, common.BOT, 'U')
-					step(f'{service} {text}', run)
+					step(f'{service} {text}', handing(worker.process_request, 'Message', service, chat, user, text,
+						common.BOT, 'U'))
 			for service, chat, message in (('telegram', '55', '9'), ('slack', 'C1', '1.5')):
 				for data in BUTTONS:
 					label = f'button {data}' if service == 'telegram' else f'{service} button {data}'
-					step(label, lambda: worker.process_callback(service, 'cb', chat, message, data))
+					step(label, handing(worker.process_callback, 'Press', service, 'cb', chat, message, data))
 			self.env(webhooks=WEBHOOKS, telegramChatID='1')
 			for where in ('', ' (slackChannelID)'): # Slack's incoming webhook, then the channel as the bot
 				if where:

From 54e9faaf40fc102b1bc444239dd7652a231666c9 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 10:24:27 +0000
Subject: [PATCH 67/81] inbound: reading an update is apart from how it arrives

A network's class read its updates and also held its webhook: the path, the secret header or signature, the form a
Slack press comes in. The readers (Telegram, Slack) now only read, and a webhook class per network (TelegramWebhook,
SlackWebhook) takes the request in and hands the update to its reader. bot.take(network, update) does what follows
for any way of arriving (read it, drop a repeat, hand it to the worker), so a pull mode (getUpdates, Socket Mode, a
gateway) only has to fetch updates and call it. A test hands updates over without a webhook.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md           |   7 +-
 bot.py              |  24 +++---
 lib/inbound.py      | 178 ++++++++++++++++++++++++--------------------
 tests/test_edges.py |  16 ++++
 4 files changed, 133 insertions(+), 92 deletions(-)

diff --git a/CLAUDE.md b/CLAUDE.md
index db2e495c..080d940f 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -34,9 +34,10 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 
 ## Code map (open the one file you need)
 - `bot.py`: WSGI entry (route, verify, parse, then a thread of `lib/worker.py` per message or button press); SIGHUP
-  reload and SIGUSR1 cache flush, each in a thread. `lib/inbound.py`: a class per network that posts to the bot
-  (`Telegram`, `Slack`): its path, how it proves itself, and its updates as `Message`, `Press` or a `Response` given
-  straight back.
+  reload and SIGUSR1 cache flush, each in a thread. `lib/inbound.py`: a class per network that reads its updates
+  (`Telegram`, `Slack`: a `Message`, a `Press`, or a `Response` given straight back), apart from a class per way they
+  arrive (`TelegramWebhook`, `SlackWebhook`: the path, how it proves itself); `bot.take(network, update)` takes an
+  update however it arrived.
 - `lib/worker.py`: `COMMANDS`, a table of `Command(names, handler, usage, section, buttons, percent, period,
   keywords)`, which also makes `.help` and the Telegram command keyboard, kept current in DMs. `read()` finds the
   command by its whole name (any other first word that may be a ticker is `.SYMBOL`) and `read_args()` its `Args`: a
diff --git a/bot.py b/bot.py
index 305ea05d..1b5ec23f 100755
--- a/bot.py
+++ b/bot.py
@@ -63,10 +63,9 @@ def main(environ, start_response):
 		return [b'']
 
 def handle(environ, start_response):
-	"""Route a request by its path to the network that posts there, turn it away unless it proves it comes from that
-	network (before its body is read, when the proof is a header), read the update (CONTENT_LENGTH bytes, up to
-	MAX_BODY), turn away repeats, and hand a message or a button press to lib/worker in a thread of its own: the network
-	gets its answer at once."""
+	"""Route a request by its path to the webhook of the network that posts there, turn it away unless it proves it
+	comes from that network (before its body is read, when the proof is a header), read the update (CONTENT_LENGTH
+	bytes, up to MAX_BODY) and take it (take): the network gets its answer at once."""
 	def respond(response):
 		"""Answer the request: a status line and a body."""
 		start_response(response.status, [('Content-type', 'application/json')])
@@ -103,14 +102,19 @@ def print_headers():
 		print(f"[{source.service}]: refused update", json.dumps(shown, indent=4, default=str), file=sys.stderr)
 		return respond(inbound.FORBIDDEN)
 	environ[AUTHENTICATED] = True
-	event = source.parse(update)
+	return respond(take(source.network, update))
+
+def take(network, update):
+	"""An update from a network (an inbound.Network reads it), however it arrived: what it says goes to lib/worker in a
+	thread of its own (dispatch), unless it is a repeat. Returns the inbound.Response for the network."""
+	event = network.parse(update)
 	if isinstance(event, inbound.Response):
-		return respond(event)
-	if seen_before(source.service, source.update_id(update)):
-		print(f"[{source.service}]: update {source.update_id(update)} seen already, not run again", file=sys.stderr)
-		return respond(inbound.OK)
+		return event
+	if seen_before(network.service, network.update_id(update)):
+		print(f"[{network.service}]: update {network.update_id(update)} seen already, not run again", file=sys.stderr)
+		return inbound.OK
 	dispatch(event)
-	return respond(inbound.OK)
+	return inbound.OK
 
 def dispatch(event):
 	"""Hand a message or a button press to lib/worker, in a thread of its own."""
diff --git a/lib/inbound.py b/lib/inbound.py
index 84376299..65f7adb5 100644
--- a/lib/inbound.py
+++ b/lib/inbound.py
@@ -1,6 +1,8 @@
-"""What the networks post to the bot: one class per network that sends updates, which knows the path it posts to, how
-it proves itself and how to read an update. An update becomes a Message to answer, a Press of a button, or a Response
-the network gets straight back (a challenge, or nothing to do)."""
+"""What the networks send the bot. Reading an update is apart from how it arrives: a class per network reads its updates
+(Telegram, Slack), and a class per way of arriving takes them in (a webhook: the path the network posts to, how it
+proves itself, the body), so that another way (a pull mode) can hand its updates to the same reader. An update becomes
+a Message to answer, a Press of a button, or a Response the network gets straight back (a challenge, or nothing to
+do)."""
 import dataclasses
 import hashlib
 import hmac
@@ -62,56 +64,26 @@ def same_secret(given, expected):
 	return bool(expected) and given is not None and hmac.compare_digest(given.encode(), expected.encode())
 
 
-class Webhook:
-	"""A network posting updates to the bot; subclasses give its path, how it proves itself and what an update says."""
+class Network:
+	"""How a network's updates read, however they arrive."""
 	service = ''
 
-	def path(self):
-		"""The URL path the network posts to, or '' when finbot.env sets none."""
-		return ''
-
-	def refuse_unread(self, environ):
-		"""Why to turn the request away before its body is read (a secret header missing or wrong), else None."""
-		return None
-
-	def decode(self, body):
-		"""The update in the request body, a dict. Raises ValueError when the body is not one."""
-		update = json.loads(body)
-		if not isinstance(update, dict):
-			raise ValueError('not a JSON object')
-		return update
-
-	def refuse(self, environ, body, update):
-		"""Why to refuse the request once read (the body's signature or a secret in it missing or wrong), else None."""
-		return None
-
 	def update_id(self, update):
 		"""The update's id, the same when the network resends it (unanswered in time), or None if it gives none."""
 		return None
 
 	def parse(self, update):
-		"""What the update says: a Message, a Press, or a Response to give straight back."""
+		"""What the update (a dict) says: a Message, a Press, or a Response to give straight back."""
 		raise NotImplementedError
 
 
-class Telegram(Webhook):
-	"""Telegram's updates, each sent with the secret header that setWebhook gave it."""
+class Telegram(Network):
+	"""Telegram's updates: messages and button presses."""
 	service = 'telegram'
 	bot_name = None # the bot's @username, from telegram.register() as bot.py starts
 	IGNORED = {'sticker', 'animation', 'voice', 'video', 'video_note', 'audio', 'document', 'contact', 'location',
 		'venue', 'poll', 'dice'} # nothing to answer, nothing worth logging
 
-	def path(self):
-		"""The path of telegramOutgoingWebhook."""
-		return urlparse(config.env.telegramOutgoingWebhook).path if config.env.telegramOutgoingWebhook else ''
-
-	def refuse_unread(self, environ):
-		"""The header must carry telegramOutgoingToken."""
-		secret = environ.get('HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN')
-		if not same_secret(secret, config.env.telegramOutgoingToken):
-			return f"Telegram authorisation header {'not present' if secret is None else 'is present but incorrect'}"
-		return None
-
 	def update_id(self, update):
 		"""Telegram's update_id."""
 		return update.get('update_id')
@@ -151,47 +123,11 @@ def parse(self, update):
 		return Message(self.service, chat_id, user, text, self.bot_name, user_name, str(message.get('message_id', '')))
 
 
-class Slack(Webhook):
-	"""Slack's Events API, and its interactivity (a button pressed), at the same URL: each request signed with the app's
-	signing secret, and carrying its verification token."""
+class Slack(Network):
+	"""Slack's Events API (messages) and its interactivity (a button pressed)."""
 	service = 'slack'
 	BY_PEOPLE = (None, 'thread_broadcast') # message subtypes a person writes; the rest are bot posts, edits, joins...
 
-	def path(self):
-		"""The path of slackOutgoingWebhook."""
-		return urlparse(config.env.slackOutgoingWebhook).path if config.env.slackOutgoingWebhook else ''
-
-	def refuse(self, environ, body, update):
-		"""With slackSigningSecret, the request must carry Slack's signature of its body, made in the last five minutes;
-		without it, the body must carry slackOutgoingToken (the older way, which Slack has deprecated)."""
-		if config.env.slackSigningSecret:
-			return self.signature_problem(environ, body, config.env.slackSigningSecret)
-		if 'token' not in update:
-			return "Slack authorisation field not present"
-		if not same_secret(str(update['token']), config.env.slackOutgoingToken):
-			return "Slack auth incorrect"
-		return None
-
-	@staticmethod
-	def signature_problem(environ, body, secret):
-		"""Why the request's signature fails, or None: X-Slack-Signature must be v0= and the HMAC-SHA256, under the
-		signing secret, of v0:<X-Slack-Request-Timestamp>:<the body>, and the timestamp within five minutes of now."""
-		timestamp = environ.get('HTTP_X_SLACK_REQUEST_TIMESTAMP', '')
-		signature = environ.get('HTTP_X_SLACK_SIGNATURE')
-		if signature is None or not (timestamp.isascii() and timestamp.isdigit()):
-			return "Slack signature not present"
-		if abs(time.time() - int(timestamp)) > SIGNATURE_AGE:
-			return "Slack signature too old (a replay, or a clock is wrong)"
-		signed = b'v0:' + timestamp.encode() + b':' + body
-		expected = 'v0=' + hmac.new(secret.encode(), signed, hashlib.sha256).hexdigest()
-		return None if same_secret(signature, expected) else "Slack signature incorrect"
-
-	def decode(self, body):
-		"""The update: an event comes as JSON, a button press as a form whose `payload` field holds the JSON."""
-		if body.startswith(b'payload='):
-			body = parse_qs(body.decode('utf-8', 'replace')).get('payload', [''])[0].encode()
-		return super().decode(body)
-
 	def update_id(self, update):
 		"""Slack's event_id (a button press has none)."""
 		return update.get('event_id')
@@ -237,9 +173,93 @@ def press(self, update):
 			container['message_ts'], value, who)
 
 
-TELEGRAM, SLACK = Telegram(), Slack()
+class Webhook:
+	"""Updates a network posts to the bot, for its reader (a Network); subclasses give the path it posts to and how it
+	proves itself."""
+	def __init__(self, network):
+		"""network: the reader of its updates."""
+		self.network = network
+		self.service = network.service
+
+	def path(self):
+		"""The URL path the network posts to, or '' when finbot.env sets none."""
+		return ''
+
+	def refuse_unread(self, environ):
+		"""Why to turn the request away before its body is read (a secret header missing or wrong), else None."""
+		return None
+
+	def decode(self, body):
+		"""The update in the request body, a dict. Raises ValueError when the body is not one."""
+		update = json.loads(body)
+		if not isinstance(update, dict):
+			raise ValueError('not a JSON object')
+		return update
+
+	def refuse(self, environ, body, update):
+		"""Why to refuse the request once read (the body's signature or a secret in it missing or wrong), else None."""
+		return None
+
+
+class TelegramWebhook(Webhook):
+	"""Telegram posting to telegramOutgoingWebhook, each update with the secret header that setWebhook gave it."""
+
+	def path(self):
+		"""The path of telegramOutgoingWebhook."""
+		return urlparse(config.env.telegramOutgoingWebhook).path if config.env.telegramOutgoingWebhook else ''
+
+	def refuse_unread(self, environ):
+		"""The header must carry telegramOutgoingToken."""
+		secret = environ.get('HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN')
+		if not same_secret(secret, config.env.telegramOutgoingToken):
+			return f"Telegram authorisation header {'not present' if secret is None else 'is present but incorrect'}"
+		return None
+
+
+class SlackWebhook(Webhook):
+	"""Slack posting events and button presses to slackOutgoingWebhook (the app's Events and Interactivity URL), each
+	request signed with the app's signing secret, and carrying its verification token."""
+
+	def path(self):
+		"""The path of slackOutgoingWebhook."""
+		return urlparse(config.env.slackOutgoingWebhook).path if config.env.slackOutgoingWebhook else ''
+
+	def refuse(self, environ, body, update):
+		"""With slackSigningSecret, the request must carry Slack's signature of its body, made in the last five minutes;
+		without it, the body must carry slackOutgoingToken (the older way, which Slack has deprecated)."""
+		if config.env.slackSigningSecret:
+			return self.signature_problem(environ, body, config.env.slackSigningSecret)
+		if 'token' not in update:
+			return "Slack authorisation field not present"
+		if not same_secret(str(update['token']), config.env.slackOutgoingToken):
+			return "Slack auth incorrect"
+		return None
+
+	@staticmethod
+	def signature_problem(environ, body, secret):
+		"""Why the request's signature fails, or None: X-Slack-Signature must be v0= and the HMAC-SHA256, under the
+		signing secret, of v0:<X-Slack-Request-Timestamp>:<the body>, and the timestamp within five minutes of now."""
+		timestamp = environ.get('HTTP_X_SLACK_REQUEST_TIMESTAMP', '')
+		signature = environ.get('HTTP_X_SLACK_SIGNATURE')
+		if signature is None or not (timestamp.isascii() and timestamp.isdigit()):
+			return "Slack signature not present"
+		if abs(time.time() - int(timestamp)) > SIGNATURE_AGE:
+			return "Slack signature too old (a replay, or a clock is wrong)"
+		signed = b'v0:' + timestamp.encode() + b':' + body
+		expected = 'v0=' + hmac.new(secret.encode(), signed, hashlib.sha256).hexdigest()
+		return None if same_secret(signature, expected) else "Slack signature incorrect"
+
+	def decode(self, body):
+		"""The update: an event comes as JSON, a button press as a form whose `payload` field holds the JSON."""
+		if body.startswith(b'payload='):
+			body = parse_qs(body.decode('utf-8', 'replace')).get('payload', [''])[0].encode()
+		return super().decode(body)
+
+
+TELEGRAM, SLACK = Telegram(), Slack() # the readers
+WEBHOOKS = (TelegramWebhook(TELEGRAM), SlackWebhook(SLACK))
 
 
 def route(path):
-	"""The network that posts to this path, or None."""
-	return next((source for source in (TELEGRAM, SLACK) if source.path() and path == source.path()), None)
+	"""The webhook whose path this is, or None."""
+	return next((hook for hook in WEBHOOKS if hook.path() and path == hook.path()), None)
diff --git a/tests/test_edges.py b/tests/test_edges.py
index efc100a0..5ac10f0d 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -394,6 +394,22 @@ def test_a_telegram_update_sent_again_runs_once(self):
 		gevent.sleep(0.05)
 		self.assertEqual(len(calls), 1)
 
+	def test_an_update_is_taken_however_it_arrived(self):
+		"""Reading an update needs no webhook: one handed over another way (as a pull mode would) reaches the worker as
+		from the webhook, Telegram's message and Slack's alike, and a repeat runs once."""
+		got = []
+		self.patch(worker, 'process_request', got.append)
+		message = {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A', 'username': 'a'}, 'text': '.price'}
+		event = {'type': 'event_callback', 'event_id': 'Ev9', 'authorizations': [{'user_id': 'UBOT'}],
+			'event': {'type': 'message', 'channel': 'C1', 'user': 'U1', 'text': '.trades', 'ts': '1.2'}}
+		with contextlib.redirect_stderr(io.StringIO()):
+			for _ in range(2):
+				self.assertEqual(bot.take(inbound.TELEGRAM, {'update_id': 70, 'message': message}), inbound.OK)
+				self.assertEqual(bot.take(inbound.SLACK, event), inbound.OK)
+		gevent.sleep(0.05) # the worker runs in a (gevent) thread
+		self.assertEqual([(m.service, m.chat_id, m.text) for m in got], [('telegram', '5', '.price'),
+			('slack', 'C1', '.trades')])
+
 	def test_button_presses_are_logged(self):
 		"""A button press is logged with who pressed it (handle, first name or unknown), the chat and the button."""
 		calls = []

From f3433bc1663a5c3e0ea2e802c83a86d8a071ee1f Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 10:26:51 +0000
Subject: [PATCH 68/81] delivery: deliver() takes chart specs only; the
 scheduled destinations in a plain list

- Every report gives its chart as a render.Chart, which deliver() draws where it can go; the path for a chart drawn
  already served only two tests, which give a spec now.
- default_chats() lists the destinations plainly, Telegram, then Slack (the channel as the bot, else the incoming
  webhook), then Discord, where it worked them out from the order of the webhooks plus a Slack afterthought.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/delivery.py         | 60 +++++++++++++++++------------------------
 tests/test_providers.py |  9 ++++---
 2 files changed, 30 insertions(+), 39 deletions(-)

diff --git a/lib/delivery.py b/lib/delivery.py
index 47aef285..ebdf48c4 100644
--- a/lib/delivery.py
+++ b/lib/delivery.py
@@ -13,9 +13,8 @@
 
 @dataclasses.dataclass
 class Report:
-	"""What a report says: its lines; perhaps a chart, a render.Chart that deliver() draws only for a destination that
-	takes charts (or one drawn already, a PNG file object); and buttons, rows of buttons.Button (or, on Telegram, a
-	keyboard markup of its own)."""
+	"""What a report says: its lines; perhaps a chart, a render.Chart spec that deliver() draws only for a destination
+	that takes charts; and buttons, rows of buttons.Button (or, on Telegram, a keyboard markup of its own)."""
 	lines: list
 	chart: object = None
 	buttons: list = None
@@ -31,28 +30,24 @@ def reply_to(service, chat, message=None):
 	return kind(chat, message) if kind else None
 
 
-def scheduled(service, url):
-	"""Where a scheduled report goes on a network: Telegram's telegramChatID; Slack's slackChannelID as the bot, with
-	slackBotToken (so that charts go too), else its incoming webhook; Discord's webhook."""
-	env = config.env
-	if service == 'telegram':
-		return transports.TelegramBot(env.telegramChatID)
-	if service == 'slack' and env.slackChannelID and env.slackBotToken:
-		return transports.SlackBot(env.slackChannelID)
-	return {'slack': transports.SlackWebhook, 'discord': transports.DiscordWebhook}[service](url)
-
-
 def default_chats():
-	"""Where a scheduled report goes: each network's destination (scheduled), Telegram's left out without a
-	telegramChatID. Raises RuntimeError when there is nowhere to post."""
+	"""Where a scheduled report goes, in this order: Telegram's telegramChatID; Slack's slackChannelID as the bot, with
+	slackBotToken (so that charts go too), else its incoming webhook; Discord's webhook. Raises RuntimeError when there
+	is nowhere to post."""
 	env = config.env
-	services = [service for service in env.webhooks if service != 'telegram' or env.telegramChatID]
-	if env.slackChannelID and env.slackBotToken and 'slack' not in services: # the bot needs no incoming webhook
-		services.append('slack')
-	if not services:
+	chats = []
+	if env.telegramChatID and 'telegram' in env.webhooks:
+		chats.append(transports.TelegramBot(env.telegramChatID))
+	if env.slackChannelID and env.slackBotToken:
+		chats.append(transports.SlackBot(env.slackChannelID))
+	elif 'slack' in env.webhooks:
+		chats.append(transports.SlackWebhook(env.webhooks['slack']))
+	if 'discord' in env.webhooks:
+		chats.append(transports.DiscordWebhook(env.webhooks['discord']))
+	if not chats:
 		raise RuntimeError('nowhere to post: finbot.env sets no telegramChatID, slackChannelID, slack_webhook or'
 			' discord_webhook')
-	return [scheduled(service, env.webhooks.get(service)) for service in services]
+	return chats
 
 
 def targets(dest):
@@ -61,20 +56,15 @@ def targets(dest):
 
 
 def deliver(dest, report):
-	"""Post a report: its chart, the lines as its caption, where the destination takes charts and the graph preference
-	is on; else the lines, buttons under the last message. A chart the network refuses goes as text without its buttons,
-	unless it may have arrived after all (a timeout). Returns the network's answers, one per message."""
-	if report.chart is None:
-		return dest.text(report.lines, report.buttons)
-	chart, caption = report.chart, '\n'.join(map(str, report.lines))
-	chartable = dest.takes_charts and config.prefs.graph
-	if isinstance(chart, render.Chart): # a spec: drawn only where it can go
-		if not (chartable and chart.rows):
-			return dest.text(report.lines)
-		chart, caption = render.draw(chart), caption if report.lines else chart.title
-	elif not chartable:
-		return dest.text(caption.split('\n'))
-	answer = dest.chart(chart, caption, report.buttons)
+	"""Post a report: its chart, drawn here only where it can go (a destination that takes charts, with the graph
+	preference on), with the lines as its caption; else the lines, and the buttons under the last message unless they
+	were the chart's. A chart the network refuses goes as text without its buttons, unless it may have arrived after all
+	(a timeout). Returns the network's answers, one per message."""
+	chart = report.chart
+	if chart is None or not (dest.takes_charts and config.prefs.graph and chart.rows):
+		return dest.text(report.lines, report.buttons if chart is None else None)
+	caption = '\n'.join(map(str, report.lines)) or chart.title
+	answer = dest.chart(render.draw(chart), caption, report.buttons)
 	if answer is transports.MAYBE_SENT:
 		return []
 	if answer:
diff --git a/tests/test_providers.py b/tests/test_providers.py
index bdb820c5..b5f7e74c 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -41,6 +41,7 @@ def get(url, **kwargs):
 
 
 REAL_CREDENTIALS = yahoo.credentials
+CHART = render.Chart('bars', 'Heading', rows=(('A', 1.0),)) # a chart for the sending tests, which draw it as b'png'
 REAL_TOKEN = sharesight._token
 
 
@@ -788,6 +789,7 @@ def test_slack_replies_upload_their_charts(self):
 	def test_a_slack_upload_that_fails_goes_out_as_text(self):
 		"""A Slack upload failing at any step goes as text, but not on a last-step timeout, when it may have arrived."""
 		self.env(slackBotToken='xoxb-test')
+		self.patch(render, 'draw', lambda chart: io.BytesIO(b'png'))
 
 		good = self.upload_steps()
 		slow = transports.requests.exceptions.ReadTimeout('slow')
@@ -801,8 +803,7 @@ def test_a_slack_upload_that_fails_goes_out_as_text(self):
 				self.slack_answers(*outcomes)
 				self.sent.clear()
 				with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
-					report = delivery.Report(['*Heading*', 'line 2'], io.BytesIO(b'png'))
-					delivery.deliver(transports.SlackBot('C1'), report)
+					delivery.deliver(transports.SlackBot('C1'), delivery.Report(['*Heading*', 'line 2'], CHART))
 				self.assertEqual(self.sent, expect)
 
 	def test_scheduled_reports_to_slack_carry_charts_only_as_the_bot(self):
@@ -863,6 +864,7 @@ def test_the_graph_preference_off_sends_no_chart_anywhere(self):
 
 	def test_a_chart_telegram_refuses_goes_out_as_text(self):
 		"""A chart Telegram refuses or that fails goes as text, but not after a timeout, when it may have arrived."""
+		self.patch(render, 'draw', lambda chart: io.BytesIO(b'png'))
 		def post(outcome):
 			"""A net.post that answers with outcome, or raises it."""
 			def fake(url, **kwargs):
@@ -880,8 +882,7 @@ def fake(url, **kwargs):
 				self.sent.clear()
 				self.patch(transports.net, 'post', post(outcome))
 				with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
-					report = delivery.Report(['Heading', 'line 2'], io.BytesIO(b'png'), [[]])
-					delivery.deliver(transports.TelegramBot('5'), report)
+					delivery.deliver(transports.TelegramBot('5'), delivery.Report(['Heading', 'line 2'], CHART, [[]]))
 				self.assertEqual(self.sent, expect)
 
 

From ac0550bca242c7ded6e7746f8b8a69e30e5cca8a Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 10:29:05 +0000
Subject: [PATCH 69/81] buttons: the data a button carries is made and read in
 one file

A period button's data was made in buttons.keyboard but taken apart in worker.process_callback, and a .jobs button's
was made in about.py and taken apart in worker again. buttons.parse reads both (a ChartPress or a JobPress, or None),
beside keyboard and a new job_button, so the format lives in one file; malformed .jobs data is still refused by the
schedule's check, as before.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/buttons.py       | 36 ++++++++++++++++++++++++++++++++++--
 lib/reports/about.py |  2 +-
 lib/worker.py        | 31 +++++++++++++++----------------
 tests/test_util.py   | 11 +++++++++++
 4 files changed, 61 insertions(+), 19 deletions(-)

diff --git a/lib/buttons.py b/lib/buttons.py
index 86b9dbc0..cf8b0982 100644
--- a/lib/buttons.py
+++ b/lib/buttons.py
@@ -1,5 +1,6 @@
-"""Buttons under a message: the period buttons (1D | 7D | 1M | 1Y | Max) of a chart, and the callback data that carries
-what its builder needs (charts.BUILDERS). Each transport draws Button rows its own way."""
+"""Buttons under a message, and what their data says when one is pressed (parse): a chart's period buttons (1D | 7D |
+1M | 1Y | Max), whose data carries what its builder needs (charts.BUILDERS), and a .jobs command's button. Each
+transport draws Button rows its own way."""
 import hashlib
 import typing
 
@@ -12,6 +13,20 @@ class Button(typing.NamedTuple):
 	active: bool = False
 
 
+class ChartPress(typing.NamedTuple):
+	"""A chart's period button pressed: the chart's kind, the period, and its builder's arguments (resolve them)."""
+	kind: str
+	period: str
+	ref: str
+
+
+class JobPress(typing.NamedTuple):
+	"""A .jobs command's button pressed: the command's number in the list and its checksum, both for
+	jobs.Scheduler.pick to check."""
+	number: str
+	checksum: str
+
+
 PERIODS = {'d': ('1D', 1), 'w': ('7D', 7), 'm': ('1M', 30), 'y': ('1Y', 365), 'x': ('Max', None)}
 BUTTONS = {'h': 'wmyx', 'p': 'wmyx', 'c': 'wmyx', 'f': 'wmyx', 'l': 'dwmyx'} # chart kind -> its buttons, five at most
 # (Discord's limit for a row): h history, p a stock's price chart, c compare, f performance, l the price list
@@ -55,3 +70,20 @@ def keyboard(kind, args, active=None):
 	(charts.BUILDERS): a ticker, tickers, a portfolio, or a price list's threshold and top, as strings."""
 	ref = _ref(args)
 	return [[Button(PERIODS[key][0], f"c|{kind}|{key}|{ref}", key == active) for key in BUTTONS[kind]]]
+
+def job_button(number, label, checksum):
+	"""A .jobs command's button: 'N. label', cut to 30 characters, its data the number and the command's checksum."""
+	return Button(f"{number}. {label}"[:30], f"j|{number}|{checksum}")
+
+def parse(data):
+	"""What a pressed button's data asks for: a JobPress for a .jobs button (whatever follows 'j|', which pick refuses
+	unless it is a command's number and checksum), a ChartPress for a period button of a known kind and period, or None
+	for data of neither (from an older finbot, or forged)."""
+	kind, _, rest = data.partition('|')
+	if kind == 'j':
+		number, _, checksum = rest.partition('|')
+		return JobPress(number, checksum)
+	parts = rest.split('|', 2)
+	if kind == 'c' and len(parts) == 3 and parts[0] in BUTTONS and parts[1] in BUTTONS[parts[0]]:
+		return ChartPress(*parts)
+	return None
diff --git a/lib/reports/about.py b/lib/reports/about.py
index 3f5ebb85..4c05e363 100644
--- a/lib/reports/about.py
+++ b/lib/reports/about.py
@@ -20,7 +20,7 @@ def prepare_jobs_payload(service, run=None):
 	payload.append("\nRun one with .jobs run NUMBER" + (' or tap it' if service == 'telegram' else ''))
 	markup = None
 	if service == 'telegram':
-		row = [buttons.Button(f"{number}. {jobs.display(label)}"[:30], f"j|{number}|{jobs.checksum(label)}")
+		row = [buttons.job_button(number, jobs.display(label), jobs.checksum(label))
 			for number, (label, _, _) in enumerate(commands, 1)]
 		markup = [row[i:i + 2] for i in range(0, len(row), 2)]
 	return payload, markup
diff --git a/lib/worker.py b/lib/worker.py
index d1c203d0..19a01802 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -41,16 +41,16 @@ def reply(self, lines, chart=None, buttons=None):
 		can go) and buttons."""
 		return delivery.deliver(self.dest, delivery.Report(lines, chart, buttons))
 
-def process_job_callback(press):
-	"""A command button of .jobs (an inbound.Press): run that command now and answer in the chat."""
+def process_job_callback(press, pressed):
+	"""A command button of .jobs (an inbound.Press; pressed, what its data says): run that command now and answer in
+	the chat."""
 	service = dialects.of(press.service)
 	chat = delivery.reply_to(service, press.chat_id)
 	try:
 		if jobs.current is None: # the scheduler is off now
 			raise jobs.ScheduleError('the scheduler is off')
-		_, number, crc = press.data.split('|') # only a command offered by .jobs can be picked; anything else is refused
-		command = jobs.current.pick(number, crc)
-	except (ValueError, jobs.ScheduleError):
+		command = jobs.current.pick(pressed.number, pressed.checksum) # only a command .jobs offers; anything else fails
+	except jobs.ScheduleError:
 		chat.acknowledge(press.press_id, 'Buttons updated, try again')
 		payload, markup = about.prepare_jobs_payload(service, '') # the list as it is now, in place of the stale one
 		chat.edit_text(press.message_id, '\n'.join(payload), markup or [])
@@ -65,23 +65,22 @@ def process_job_callback(press):
 def process_callback(press):
 	"""A button pressed (an inbound.Press): run a scheduled command from .jobs, or rebuild a chart message for the
 	chosen period and edit it in place."""
-	if press.data.startswith('j|'):
-		return process_job_callback(press)
+	pressed = buttons.parse(press.data)
+	if isinstance(pressed, buttons.JobPress):
+		return process_job_callback(press, pressed)
 	service = dialects.of(press.service)
-	parts = press.data.split('|', 3)
-	valid = len(parts) == 4 and parts[0] == 'c' and parts[1] in buttons.BUTTONS and parts[2] in buttons.BUTTONS[parts[1]]
 	chat = delivery.reply_to(service, press.chat_id)
 	# answering now shows Telegram's toast bubble ("Loading 1Y…") while the new chart is built
-	chat.acknowledge(press.press_id, f"Loading {buttons.PERIODS[parts[2]][0]}…" if valid else None)
-	if not valid:
+	chat.acknowledge(press.press_id, f"Loading {buttons.PERIODS[pressed.period][0]}…" if pressed else None)
+	if not pressed:
 		return
 	try:
-		_, kind, period, ref = parts
-		args = buttons.resolve(ref)
-		caption, chart = charts.build(kind, args, period, service)
+		args = buttons.resolve(pressed.ref)
+		caption, chart = charts.build(pressed.kind, args, pressed.period, service)
 		if chart is None: # too few prices in that period to draw a line
-			raise answer.Answer(f"Too few prices to draw {buttons.PERIODS[period][0]}")
-		chat.edit_chart(press.message_id, render.draw(chart), caption, buttons.keyboard(kind, args, period))
+			raise answer.Answer(f"Too few prices to draw {buttons.PERIODS[pressed.period][0]}")
+		markup = buttons.keyboard(pressed.kind, args, pressed.period)
+		chat.edit_chart(press.message_id, render.draw(chart), caption, markup)
 	except answer.Answer as e: # nothing to draw for that period: said in the chat, as a command would
 		delivery.deliver(chat, delivery.Report(e.lines))
 	except Exception as e:
diff --git a/tests/test_util.py b/tests/test_util.py
index f5232ed1..721750c1 100644
--- a/tests/test_util.py
+++ b/tests/test_util.py
@@ -104,6 +104,17 @@ def test_date_range_english(self):
 		with self.assertRaises(TypeError):
 			formats.date_range_english(None, end) # callers must pass a period
 
+	def test_button_data_reads_back(self):
+		"""A button's data says what was pressed: a period button of a known kind and period (as keyboard() makes
+		them), any .jobs button (pick checks it), and nothing for anything else."""
+		self.assertEqual([buttons.parse(button.data) for button in buttons.keyboard('l', ['5', '0'])[0]],
+			[buttons.ChartPress('l', key, '5,0') for key in 'dwmyx'])
+		self.assertEqual(buttons.parse(buttons.job_button(8, 'trades', 'abcd').data), buttons.JobPress('8', 'abcd'))
+		self.assertEqual(buttons.parse('j|8'), buttons.JobPress('8', ''))
+		for data in ('c|h|d|T1', 'c|z|w|T1', 'c|h|w', 'x|h|w|T1', '', 'c'):
+			with self.subTest(data=data):
+				self.assertIsNone(buttons.parse(data))
+
 	def test_strip_url(self):
 		"""A link's text is its address without the scheme or www."""
 		for url, text in (('https://www.apple.com', 'apple.com'), ('http://example.org/a?b', 'example.org/a?b'),

From 1726b4a21eb7c13277f4cd1613af50a55f6fa96f Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 10:33:07 +0000
Subject: [PATCH 70/81] Charts with period buttons: the first view is built as
 a button press builds it

.history's first view came from its button builder, but .compare's and .price SYMBOL PERIOD's were made by other
code, so the chart a command posted and the one its own button redraws could differ. All three go through the
builder now (worker.chart_reply), and the price report lost the code that drew one stock's line, which only that view
used. One difference: with performance_use_sharesight on, .price SYMBOL PERIOD used to give Sharesight's figure and no
chart; it gives Yahoo's figure and line now, as its buttons always did (the lists still take Sharesight's figures).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 lib/charts.py        | 14 ++++++++------
 lib/reports/price.py | 34 ++++++++++++++--------------------
 lib/worker.py        | 29 +++++++++++++++++------------
 3 files changed, 39 insertions(+), 38 deletions(-)

diff --git a/lib/charts.py b/lib/charts.py
index fd54443d..0dbf7781 100644
--- a/lib/charts.py
+++ b/lib/charts.py
@@ -1,4 +1,5 @@
-"""A period button pressed: the chart rebuilt for that period, by the builder of the chart's kind."""
+"""A chart with period buttons, built by its kind's builder: the command's first view, and again for the period of a
+button pressed, so that both look alike."""
 from lib import answer
 from lib import buttons
 from lib import formats
@@ -12,7 +13,8 @@ def history_chart(service, period, days, ticker):
 	return '\n'.join(lines), chart
 
 def price_chart(service, period, days, ticker):
-	""".price SYMBOL: one stock's change over the period, and its line."""
+	""".price SYMBOL [period]: one stock's change over the period (all its history for Max), from Yahoo, and its
+	line."""
 	market_data = yahoo.fetch([ticker])
 	if ticker not in market_data:
 		raise answer.Answer(f"no data found for {ticker}")
@@ -46,9 +48,9 @@ def price_list_chart(service, period, days, threshold, top):
 
 BUILDERS = {'h': history_chart, 'p': price_chart, 'c': compare_chart, 'f': performance_chart, 'l': price_list_chart}
 
-def build(kind, args, period, service):
-	"""(caption, chart spec or None) of a chart kind at a button period (None: the command's default view), from the
-	arguments its buttons carry (buttons.keyboard)."""
+def build(kind, args, service, period=None, days=None):
+	"""(caption, chart spec or None) of a chart kind from the arguments its buttons carry (buttons.keyboard): over a
+	button period, or over `days` (a command's own span: .compare 3m), or the command's default view."""
 	if kind not in BUILDERS:
 		raise ValueError(f"unknown chart kind {kind}")
-	return BUILDERS[kind](service, period, buttons.PERIODS[period][1] if period else None, *args)
+	return BUILDERS[kind](service, period, buttons.PERIODS[period][1] if period else days, *args)
diff --git a/lib/reports/price.py b/lib/reports/price.py
index c245cc3d..8cbbee8f 100644
--- a/lib/reports/price.py
+++ b/lib/reports/price.py
@@ -17,17 +17,16 @@
 from lib.reports import history
 
 def gather(specific_stock=None, days=None):
-	"""(tickers, market data, the line chart or None) for a price report on the stocks followed or one (Yahoo's
-	symbol): Yahoo's quotes, and over `days` each one's change as percent_change_period: a holding's from Sharesight,
-	the others' from Yahoo's history, fetched together up front, with the line chart of the one stock asked about
-	(performance_use_sharesight: Sharesight's figures only, and the one stock's even when the others take Yahoo's)."""
+	"""(tickers, market data) for a price report on the stocks followed or one (Yahoo's symbol): Yahoo's quotes, and
+	over `days` each one's change as percent_change_period: a holding's from Sharesight, the others' from Yahoo's
+	history, fetched together up front (performance_use_sharesight: Sharesight's figures only, and the one stock's
+	even when the others take Yahoo's)."""
 	tickers = watchlist.covered(specific_stock)
 	market_data = yahoo.fetch(tickers)
 	if config.prefs.demote_volatile: # trigger_bar compares market caps, in US dollars
 		yahoo.add_usd_market_caps(market_data)
-	chart = None
 	if not days:
-		return tickers, market_data, chart
+		return tickers, market_data
 	# Sharesight reports a holding's performance only for the time it was held: held for 1Y and asked for 5Y, it gives 1Y
 	if not specific_stock or config.prefs.performance_use_sharesight:
 		for portfolio in sharesight.portfolios():
@@ -37,22 +36,20 @@ def gather(specific_stock=None, days=None):
 				elif ticker in tickers:
 					print("Notice: price.py:", ticker, "has no data", file=sys.stderr)
 	if config.prefs.performance_use_sharesight:
-		return tickers, market_data, chart
+		return tickers, market_data
 	missing = [ticker for ticker in market_data if 'percent_change_period' not in market_data[ticker]]
 	if not specific_stock:
 		yahoo.prefetch_history(missing)
 	for ticker in missing:
 		try:
-			changes, chart = history.price_line(ticker, days, market_data[ticker])
+			changes, _ = history.price_line(ticker, days, market_data[ticker])
 		except RuntimeError as e: # no history: the one stock asked about says so, a list leaves it out
 			if specific_stock:
 				raise
 			print(ticker, e, file=sys.stderr)
 			continue
 		market_data[ticker]['percent_change_period'] = changes[days]
-	if not specific_stock:
-		chart = None # the line of one stock only
-	return tickers, market_data, chart
+	return tickers, market_data
 
 def needs_beta(quote):
 	"""Whether trigger_bar must know this ticker's beta: a small cap between US$150M and US$1B, which counts as volatile when its
@@ -91,12 +88,11 @@ def trigger_bar(ticker, quote, threshold, details):
 class Moves:
 	"""What a price report shows, the same for every chat (moves() works it out once): the moves that pass, (ticker,
 	percent, quote) best first, the countries of their exchanges, whether a market of the stocks is in session, the date
-	of the freshest price, and the one stock's line chart, if any."""
+	of the freshest price."""
 	rows: list
 	exchanges: set
 	in_session: bool
 	as_of: datetime.date
-	chart: object = None
 
 def change(mode, quote, now, interactive):
 	"""A stock's move in a mode of MODES, or None when it has none to report: not in session (midsession), quoted more
@@ -121,7 +117,7 @@ def moves(mode='interday', threshold=None, days=None, specific_stock=None, inter
 	when one passes its trigger_bar."""
 	if threshold is None:
 		threshold = config.prefs.price_percent
-	_, market_data, chart = gather(specific_stock, days)
+	_, market_data = gather(specific_stock, days)
 	now = datetime.datetime.now()
 	rows, exchanges = [], set()
 	for ticker, quote in market_data.items():
@@ -137,7 +133,7 @@ def moves(mode='interday', threshold=None, days=None, specific_stock=None, inter
 	stamps = [quote['regularMarketTime'] for quote in market_data.values()]
 	as_of = datetime.datetime.fromtimestamp(max(stamps)).date() if stamps else datetime.date.today()
 	in_session = any(quote['marketState'] == 'REGULAR' for quote in market_data.values())
-	return Moves(rows, exchanges, in_session, as_of, chart)
+	return Moves(rows, exchanges, in_session, as_of)
 
 def report(service, found, mode='interday', threshold=None, days=None, top=None, user='', specific_stock=None,
 		interactive=False):
@@ -151,7 +147,7 @@ def report(service, found, mode='interday', threshold=None, days=None, top=None,
 	if not lines:
 		return [nothing(mode, threshold, user, specific_stock, found.in_session, days)] if interactive else [], None
 	if specific_stock:
-		return lines, found.chart
+		return lines, None
 	label = lambda rows: [(formats.short_ticker(ticker), percent) for ticker, percent, _ in rows]
 	on = formats.date_short(found.as_of, True)
 	span = formats.date_range_english(days, found.as_of) if days else on
@@ -196,12 +192,10 @@ def run(dest=None, *, mode='interday', threshold=None, days=None, top=None, user
 		threshold = config.prefs.price_percent
 	found = moves(mode, threshold, days, specific_stock, interactive)
 	def markup(chat):
-		"""A chart's period buttons, where the chat takes buttons: one ticker's chart, and the daily and N-day lists,
-		but not the intraday ones."""
+		"""A chart's period buttons, where the chat takes buttons: the daily and N-day lists, but not the intraday
+		ones."""
 		if not chat.takes_buttons:
 			return None
-		if specific_stock:
-			return buttons.keyboard('p', [specific_stock], buttons.period_for_days(days))
 		if mode in ('interday', 'period'):
 			return buttons.keyboard('l', [f"{threshold:g}", str(top or 0)], buttons.period_for_days(days, 'd' if not days else None))
 		return None
diff --git a/lib/worker.py b/lib/worker.py
index 19a01802..b2507896 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -15,7 +15,7 @@
 from lib import render
 from lib import buttons
 from lib import charts
-from lib.reports import about, cal, history, holdings, performance, price, profile, rankings, shorts, trades
+from lib.reports import about, cal, holdings, performance, price, profile, rankings, shorts, trades
 from lib import sharesight
 from lib import transports
 from lib import jobs
@@ -76,7 +76,7 @@ def process_callback(press):
 		return
 	try:
 		args = buttons.resolve(pressed.ref)
-		caption, chart = charts.build(pressed.kind, args, pressed.period, service)
+		caption, chart = charts.build(pressed.kind, args, service, period=pressed.period)
 		if chart is None: # too few prices in that period to draw a line
 			raise answer.Answer(f"Too few prices to draw {buttons.PERIODS[pressed.period][0]}")
 		markup = buttons.keyboard(pressed.kind, args, pressed.period)
@@ -245,10 +245,13 @@ def on_session(request, args):
 
 def on_price(request, args):
 	""".price [percent|SYMBOL|top|bottom] [period]: a bare number is a percent, as on the other commands; a period has
-	its unit."""
+	its unit. One stock over a period is its chart, as its period buttons draw it."""
+	symbol = None if args.keyword else ticker(args)
+	if symbol and args.days:
+		return chart_reply(request, 'p', [symbol], args.days)
 	top = 10 if args.keyword else None # top or bottom: one chart shows both ends
 	price.run(request.dest, mode='period' if args.days else 'interday', threshold=0 if top else args.percent,
-		days=args.days, top=top, user=request.user, specific_stock=None if top else ticker(args))
+		days=args.days, top=top, user=request.user, specific_stock=symbol)
 
 def on_shorts(request, args):
 	""".shorts [percent|SYMBOL]: the most shorted stocks followed, or one stock's short interest."""
@@ -308,19 +311,21 @@ def on_compare(request, args):
 	tickers = list(dict.fromkeys(yahoo.symbols(args.words))) # each once, in the order asked
 	if len(tickers) < 2:
 		raise Usage
-	payload, chart = history.prepare_compare(request.service, tickers, args.days)
-	markup = None
-	if request.dest.takes_buttons:
-		markup = buttons.keyboard('c', tickers, buttons.period_for_days(args.days) if args.days else 'x')
-	request.reply(payload, chart, markup)
+	chart_reply(request, 'c', tickers, args.days)
 
 def on_history(request, args):
 	""".history SYMBOL: the stock's change over every period, with its price chart, and period buttons where they go."""
 	if len(args.words) != 1:
 		raise Usage
-	symbol = ticker(args)
-	caption, chart = charts.build('h', [symbol], None, request.service)
-	markup = buttons.keyboard('h', [symbol], 'x') if chart and request.dest.takes_buttons else None
+	chart_reply(request, 'h', [ticker(args)])
+
+def chart_reply(request, kind, args, days=None):
+	"""The first view of a chart with period buttons, built as a button press builds it (charts.build): its caption,
+	chart and buttons where they go, the period shown marked (Max when days is None)."""
+	caption, chart = charts.build(kind, args, request.service, days=days)
+	markup = None
+	if chart and request.dest.takes_buttons:
+		markup = buttons.keyboard(kind, args, buttons.period_for_days(days) if days else 'x')
 	request.reply(caption.split('\n'), chart, markup)
 
 def on_plan(request, args):

From a31190f973f499961bda3db9da1ef9a9d1773976 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 10:40:06 +0000
Subject: [PATCH 71/81] Slack: .jobs buttons in Block Kit; a pressed chart's
 earlier file is deleted

- .jobs on Slack lists the commands with a button each, two to a row of Block Kit, where presses reach finbot (the
  bot token and slackOutgoingWebhook), and says to tap one; a press of a stale button puts the current list in its
  place (SlackBot.edit_text, chat.update). Whether a chat gets the buttons is now its transport's say, not a test for
  Telegram.
- A period button pressed under a Slack chart uploads the new chart, shows it, and then deletes the file it replaced
  (D28): nothing reuses an uploaded chart, since a period shown again is uploaded again, so deleting costs nothing
  and the bot's files no longer pile up. Each chart message's file is noted in finbot_slack_charts.json (the newest
  500); a message from before this, or noted no longer, keeps its earlier file. files.delete needs the files:write
  scope the uploads already have.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 README.md               |  2 +-
 doc/slack.md            |  7 +++-
 lib/reports/about.py    |  9 +++--
 lib/transports.py       | 86 ++++++++++++++++++++++++++++++++---------
 lib/worker.py           |  4 +-
 tests/test_jobs.py      | 36 ++++++++++++++---
 tests/test_providers.py | 19 +++++++++
 7 files changed, 131 insertions(+), 32 deletions(-)

diff --git a/README.md b/README.md
index 1203e000..260668be 100644
--- a/README.md
+++ b/README.md
@@ -639,7 +639,7 @@ sudo systemctl reload finbot        # the bot re-reads the file and logs how man
 ```
 The bot logs the number of jobs when it starts or reloads, and each job as it runs (`journalctl -u finbot`); a line it cannot use is reported by name. If the edited file has no usable line the previous schedule keeps running.
 
-**To run a job from chat**, send `.jobs` (also `.cron`, `.crons` or `.schedule`): it lists the scheduled commands, each once and without times, comments or `.py` (`cleanup` and `podcasts` are scheduled but not listed), and on Telegram adds a button per command. `.jobs run 2` (or tapping the button) runs that command now, and `.jobs run price.py close` runs it by name; anything not in the schedule is refused, and so is a chain such as `a; b` (run them one at a time). A manual run is exactly a cron run: it reports to all configured chat networks, updates the same state (seen trades, ratings and so on), and does not change the schedule. Anyone who can message the bot can do this; a command that is already running is not started twice.
+**To run a job from chat**, send `.jobs` (also `.cron`, `.crons` or `.schedule`): it lists the scheduled commands, each once and without times, comments or `.py` (`cleanup` and `podcasts` are scheduled but not listed), and on Telegram and Slack (with its buttons set up) adds a button per command. `.jobs run 2` (or tapping the button) runs that command now, and `.jobs run price.py close` runs it by name; anything not in the schedule is refused, and so is a chain such as `a; b` (run them one at a time). A manual run is exactly a cron run: it reports to all configured chat networks, updates the same state (seen trades, ratings and so on), and does not change the schedule. Anyone who can message the bot can do this; a command that is already running is not started twice.
 
 Behaviour worth knowing:
 * A run that was due while the bot was down (a restart, a deploy) still fires once if it was due within the last `scheduler_catchup` minutes (default 5). A restart right after a run does not repeat it.
diff --git a/doc/slack.md b/doc/slack.md
index afee3300..bd314e5a 100644
--- a/doc/slack.md
+++ b/doc/slack.md
@@ -105,7 +105,10 @@ slackChannelID=C0123456789
 
 A chart is uploaded with Slack's file upload API (`files.getUploadURLExternal`, then `files.completeUploadExternal`).
 With its period buttons, it goes in a message of blocks: the report, the chart (uploaded into no channel, and shown by
-its id) and the buttons. A press uploads the chart for that period and shows it in the message in place of the last
-one (Slack cannot replace a file; the earlier one stays among the bot's files, but in no channel). If Slack will not
+its id) and the buttons. A press uploads the chart for that period, shows it in the message in place of the last one
+(Slack cannot replace a file) and deletes the file it replaced, so the bot's files do not pile up. If Slack will not
 show a chart with its buttons, the chart is uploaded into the channel without them; if an upload fails (a missing scope,
 say), the report goes out as text. The log says why.
+
+`.jobs` lists the scheduled commands with a button for each, as on Telegram; a button from a list that has since
+changed puts the current list in its place.
diff --git a/lib/reports/about.py b/lib/reports/about.py
index 4c05e363..4339191a 100644
--- a/lib/reports/about.py
+++ b/lib/reports/about.py
@@ -3,9 +3,10 @@
 from lib import formats
 from lib import jobs
 
-def prepare_jobs_payload(service, run=None):
+def prepare_jobs_payload(service, run=None, tappable=False):
 	"""`.jobs`: the scheduled commands, each once; with `run` (a number in that list, or a command's text) that command
-	run now, as cron would. Returns (payload, buttons): Telegram gets a button per command, two to a row."""
+	run now, as cron would. Returns (payload, buttons): with tappable (the chat takes buttons) a button per command, two
+	to a row."""
 	scheduler = jobs.current
 	if scheduler is None:
 		return ['Scheduled jobs are off (scheduler=False)'], None
@@ -17,9 +18,9 @@ def prepare_jobs_payload(service, run=None):
 	payload = [service.bold('Scheduled commands:')]
 	for number, (label, _, _) in enumerate(commands, 1):
 		payload.append(f"{number}. {jobs.display(label)}" + (' (running)' if label in scheduler.busy else ''))
-	payload.append("\nRun one with .jobs run NUMBER" + (' or tap it' if service == 'telegram' else ''))
+	payload.append("\nRun one with .jobs run NUMBER" + (' or tap it' if tappable else ''))
 	markup = None
-	if service == 'telegram':
+	if tappable:
 		row = [buttons.job_button(number, jobs.display(label), jobs.checksum(label))
 			for number, (label, _, _) in enumerate(commands, 1)]
 		markup = [row[i:i + 2] for i in range(0, len(row), 2)]
diff --git a/lib/transports.py b/lib/transports.py
index 02754730..5e4ed38c 100644
--- a/lib/transports.py
+++ b/lib/transports.py
@@ -12,10 +12,13 @@
 from lib import config
 from lib import dialects
 from lib import net
+from lib import state
 from lib import telegram
 
 MAYBE_SENT = object() # a read timeout on the request that posts a chart: it may have arrived, so no text copy follows
 SUPPRESS_EMBEDS = 4 # Discord message flag: no link previews
+SLACK_CHARTS = 'finbot_slack_charts.json' # 'channel ts' of a chart message -> the file it shows (deleted once replaced)
+SLACK_CHARTS_KEPT = 500 # chart messages noted, newest last: an older one's file stays when a press replaces it
 NO_MENTIONS = {'parse': []} # Discord: text that looks like @everyone or a mention pings nobody
 
 
@@ -280,15 +283,19 @@ def typing_done(self):
 			print("Slack reactions.remove failed:", net.redact(e), file=sys.stderr)
 
 	def post_text(self, text, buttons=None):
-		"""chat.postMessage, without link previews. Slack's answer, or None when it failed or refused (logged): Slack
-		refuses with HTTP 200 and ok false (not_in_channel, say), so a refusal is read from the answer."""
+		"""chat.postMessage, without link previews, and with buttons (a Block Kit message) where they work. Slack's
+		answer, or None when it failed or refused (logged): Slack refuses with HTTP 200 and ok false (not_in_channel,
+		say), so a refusal is read from the answer."""
 		if not config.env.slackBotToken:
 			print(f"ERROR slack: a reply to channel {self.where} needs slackBotToken in finbot.env; not sent",
 				file=sys.stderr)
 			return None
+		blocks = {}
+		if buttons and not isinstance(buttons, dict) and self.takes_buttons:
+			blocks['blocks'] = json.dumps(text_blocks(text, buttons))
 		try:
 			answer = slack_api('chat.postMessage', channel=self.where, text=text, unfurl_links='false',
-				unfurl_media='false')
+				unfurl_media='false', **blocks)
 		except Exception as e:
 			print("Slack chat.postMessage to", self.where, "failed:", net.redact(e), file=sys.stderr)
 			return None
@@ -339,23 +346,54 @@ def unshared(self, data):
 	def post_blocks(self, data, caption, buttons):
 		"""The chart in a Block Kit message, the report above it and the period buttons under it. Slack's answer, or
 		MAYBE_SENT when posting the message timed out (it may have arrived); raises RuntimeError when Slack refuses."""
-		blocks = chart_blocks(caption, buttons, self.unshared(data))
+		file_id = self.unshared(data)
+		blocks = chart_blocks(caption, buttons, file_id)
 		try:
-			return until_shown(lambda: slack_api('chat.postMessage', channel=self.where, text=caption[:SECTION_LIMIT],
-				blocks=json.dumps(blocks), unfurl_links='false', unfurl_media='false'))
+			answer = until_shown(lambda: slack_api('chat.postMessage', channel=self.where,
+				text=caption[:SECTION_LIMIT], blocks=json.dumps(blocks), unfurl_links='false', unfurl_media='false'))
 		except requests.exceptions.ReadTimeout as e:
 			print("Slack chat.postMessage timed out, the chart may still arrive:", net.redact(e), file=sys.stderr)
 			return MAYBE_SENT
+		if answer.get('ts'):
+			self.shows(answer['ts'], file_id)
+		return answer
 
 	def edit_chart(self, message, image, caption, buttons=None):
 		"""Show another period in a chart message: the new chart uploaded (a file cannot be replaced) and shown in its
-		image block, with its caption and buttons. Raises RuntimeError when Slack refuses."""
+		image block, with its caption and buttons, then the file it showed before deleted (nothing reuses an uploaded
+		chart: a period shown again is uploaded again). Raises RuntimeError when Slack refuses the update."""
 		if hasattr(image, 'seek'):
 			image.seek(0)
 		file_id = self.unshared(image.read() if hasattr(image, 'read') else bytes(image))
 		blocks = chart_blocks(caption, buttons or [], file_id)
 		until_shown(lambda: slack_api('chat.update', channel=self.where, ts=message, text=caption[:SECTION_LIMIT],
 			blocks=json.dumps(blocks)))
+		replaced = self.shows(message, file_id)
+		if replaced and replaced != file_id:
+			try:
+				slack_api('files.delete', file=replaced)
+			except Exception as e: # it stays among the bot's files, in no channel
+				print("Slack files.delete failed:", net.redact(e), file=sys.stderr)
+
+	def shows(self, message, file_id):
+		"""Note that a chart message shows this file now (SLACK_CHARTS): the file it showed before, if noted."""
+		key = f'{self.where} {message}'
+		with state.edit(SLACK_CHARTS, {}) as shown:
+			replaced = shown.pop(key, None)
+			shown[key] = file_id
+			for old in list(shown)[:-SLACK_CHARTS_KEPT]:
+				del shown[old]
+		return replaced
+
+	def edit_text(self, message, text, buttons=None):
+		"""Edit a message's text, and its buttons unless buttons is None ([] removes them); a failure is only logged."""
+		data = {'channel': self.where, 'ts': message, 'text': text}
+		if buttons is not None:
+			data['blocks'] = json.dumps(text_blocks(text, buttons))
+		try:
+			slack_api('chat.update', **data)
+		except Exception as e:
+			print("Slack chat.update failed:", net.redact(e), file=sys.stderr)
 
 
 class SlackWebhook(Transport):
@@ -423,18 +461,30 @@ def slack_api(method, **data):
 SECTION_LIMIT = 3000 # characters of text in a Block Kit section
 
 
+def section_blocks(text):
+	"""Block Kit sections of a message's text, as many as its length needs (none for no text)."""
+	return [{'type': 'section', 'text': {'type': 'mrkdwn', 'text': '\n'.join(part)}}
+		for part in chunked(text.split('\n'), SECTION_LIMIT, 1000) if '\n'.join(part).strip()]
+
+
+def actions_block(row):
+	"""A Block Kit row of buttons, the one shown now marked."""
+	return {'type': 'actions', 'elements': [{'type': 'button', 'text': {'type': 'plain_text', 'text': button.label},
+		'value': button.data, 'action_id': button.data, **({'style': 'primary'} if button.active else {})}
+		for button in row]}
+
+
 def chart_blocks(caption, buttons, file_id):
-	"""Block Kit for a chart message: the report (in as many sections as its length needs), the chart (an uploaded file,
-	by its id) and the buttons, the one shown now marked."""
-	blocks = [{'type': 'section', 'text': {'type': 'mrkdwn', 'text': '\n'.join(part)}}
-		for part in chunked(caption.split('\n'), SECTION_LIMIT, 1000) if '\n'.join(part).strip()]
-	blocks.append({'type': 'image', 'slack_file': {'id': file_id}, 'alt_text': 'chart'})
-	elements = [{'type': 'button', 'text': {'type': 'plain_text', 'text': button.label}, 'value': button.data,
-		'action_id': button.data, **({'style': 'primary'} if button.active else {})}
-		for row in buttons for button in row]
-	if elements:
-		blocks.append({'type': 'actions', 'elements': elements})
-	return blocks
+	"""Block Kit for a chart message: the report, the chart (an uploaded file, by its id) and the buttons in one
+	row."""
+	row = [button for line in buttons for button in line]
+	return section_blocks(caption) + [{'type': 'image', 'slack_file': {'id': file_id}, 'alt_text': 'chart'}] + \
+		([actions_block(row)] if row else [])
+
+
+def text_blocks(text, buttons):
+	"""Block Kit for a message with buttons: the text, then the buttons, a row of them to a block."""
+	return section_blocks(text) + [actions_block(row) for row in buttons if row]
 
 
 def until_shown(post, waits=(1, 2, 4)):
diff --git a/lib/worker.py b/lib/worker.py
index b2507896..11b22641 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -52,7 +52,7 @@ def process_job_callback(press, pressed):
 		command = jobs.current.pick(pressed.number, pressed.checksum) # only a command .jobs offers; anything else fails
 	except jobs.ScheduleError:
 		chat.acknowledge(press.press_id, 'Buttons updated, try again')
-		payload, markup = about.prepare_jobs_payload(service, '') # the list as it is now, in place of the stale one
+		payload, markup = about.prepare_jobs_payload(service, '', chat.takes_buttons) # the list now, for the stale one
 		chat.edit_text(press.message_id, '\n'.join(payload), markup or [])
 		return
 	chat.acknowledge(press.press_id, f'Running {jobs.display(command[0])}…'[:100])
@@ -354,7 +354,7 @@ def on_jobs(request, args):
 	""".jobs [run NUMBER|COMMAND]: the schedule, or one of its commands run now."""
 	if bool(args.keyword) != bool(args.words): # run without a command, or words without run
 		raise Usage
-	payload, markup = about.prepare_jobs_payload(request.service, args.rest or None)
+	payload, markup = about.prepare_jobs_payload(request.service, args.rest or None, request.dest.takes_buttons)
 	request.reply(payload, buttons=markup)
 
 def on_profile(request, args):
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
index 434fa081..6d0e01b4 100644
--- a/tests/test_jobs.py
+++ b/tests/test_jobs.py
@@ -577,16 +577,42 @@ def test_list_in_chat_is_just_the_commands(self):
 		self.s.busy.add('price.py 7')
 		self.assertIn('6. price 7 (running)', '\n'.join(map(str, self.list_in_chat())))
 
-	def test_telegram_gets_a_button_per_command_and_slack_does_not(self):
-		"""On Telegram each command gets a button whose data fits in 64 bytes; Slack gets none."""
-		payload, markup = about.prepare_jobs_payload(TELEGRAM, '')
+	def test_a_chat_that_takes_buttons_gets_one_per_command(self):
+		"""Where buttons work each command gets one, whose data fits in Telegram's 64 bytes, and the list says to tap
+		it; elsewhere there are none, and it does not."""
+		payload, markup = about.prepare_jobs_payload(TELEGRAM, '', tappable=True)
 		pressable = [b for row in markup for b in row]
 		self.assertEqual([b.label for b in pressable], ['1. performance 7', '2. performance 30', '3. performance 90',
 			'4. performance 180', '5. performance 365', '6. price 7', '7. price close', '8. trades'])
 		for b in pressable:
 			self.assertLessEqual(len(b.data.encode()), 64)
+		self.assertTrue(payload[-1].endswith('or tap it'))
 		payload, markup = about.prepare_jobs_payload(SLACK, '')
 		self.assertIsNone(markup)
+		self.assertTrue(payload[-1].endswith('.jobs run NUMBER'))
+
+	def test_slack_gets_the_buttons_in_block_kit_and_a_stale_list_is_replaced(self):
+		"""With the bot token and slackOutgoingWebhook, Slack's .jobs list carries a button per command, two to a row
+		of Block Kit, and a stale button's press puts the current list in its place."""
+		self.env(slackBotToken='xoxb-test', slackOutgoingWebhook='https://example.com/slack')
+		self.patch(transports.Transport, 'text', REAL_TEXT)
+		posts = []
+		self.patch(transports.net, 'post', lambda url, **kwargs: posts.append((url, kwargs)) or
+			mock.Mock(status_code=200, json=lambda: {'ok': True, 'ts': '1.5'}))
+		with contextlib.redirect_stderr(io.StringIO()):
+			worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.jobs', BOT, 'U'))
+		url, kwargs = posts[-1]
+		blocks = json.loads(kwargs['data']['blocks'])
+		self.assertTrue(url.endswith('/chat.postMessage'))
+		self.assertEqual([block['type'] for block in blocks], ['section'] + ['actions'] * 4)
+		self.assertEqual([button['text']['text'] for button in blocks[4]['elements']], ['7. price close', '8. trades'])
+		with contextlib.redirect_stderr(io.StringIO()):
+			worker.process_callback(inbound.Press('slack', 't1', 'C1', '1.5', 'j|7|0000'))
+		url, kwargs = posts[-1]
+		self.assertTrue(url.endswith('/chat.update'))
+		self.assertEqual((kwargs['data']['channel'], kwargs['data']['ts']), ('C1', '1.5'))
+		self.assertEqual(len(json.loads(kwargs['data']['blocks'])), 5)
+		self.assertEqual(self.ran, [])
 
 	def test_the_chat_command_passes_the_keyboard_on(self):
 		""".jobs sends its buttons with the list."""
@@ -599,7 +625,7 @@ def test_button_press_runs_that_command(self):
 		"""Pressing a command's button runs it, says so in a toast, and reports how it went."""
 		sent = []
 		self.patch(telegram, 'answerCallbackQuery', lambda cid, text=None: sent.append(text))
-		markup = about.prepare_jobs_payload(TELEGRAM, '')[1]
+		markup = about.prepare_jobs_payload(TELEGRAM, '', tappable=True)[1]
 		data = next(b.data for row in markup for b in row if b.label.startswith('7.'))
 		self.sent.clear()
 		worker.process_callback(inbound.Press('telegram', 'cb1', '55', '9', data))
@@ -647,7 +673,7 @@ def test_button_from_an_old_schedule_updates_the_message(self):
 		chat, mid, text, markup = self.edits[0]
 		self.assertEqual((chat, mid), ('55', '9')) # the message with the old buttons is edited in place
 		self.assertIn('7. price close', text)
-		self.assertEqual(markup, about.prepare_jobs_payload(TELEGRAM, '')[1]) # with the current buttons
+		self.assertEqual(markup, about.prepare_jobs_payload(TELEGRAM, '', tappable=True)[1]) # the current buttons
 		self.patch(jobs, 'current', None) # scheduler off: the list is replaced by the notice and the buttons removed
 		worker.process_callback(inbound.Press('telegram', 'cb1', '55', '9', 'j|7|0000'))
 		self.assertEqual(self.edits[-1][2:], ('Scheduled jobs are off (scheduler=False)', [])) # no buttons left
diff --git a/tests/test_providers.py b/tests/test_providers.py
index b5f7e74c..acfbf490 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -1084,6 +1084,25 @@ def test_a_press_redraws_the_chart_in_place(self):
 		self.assertEqual(blocks[1]['slack_file'], {'id': 'F1'})
 		self.assertEqual([button.get('style') for button in blocks[2]['elements']], [None, 'primary'] + [None] * 3)
 
+	def test_a_press_deletes_the_chart_it_replaced(self):
+		"""Each press uploads its period's chart and, once the message shows it, deletes the file shown before: no
+		chart is reused, so the bot's files do not pile up. The file a press shows is never deleted."""
+		uploads = iter(f'F{n}' for n in range(1, 10))
+		record = transports.net.post
+		def post(url, **kwargs):
+			"""Answer as Slack does, with a new file id for each upload."""
+			answer = record(url, **kwargs)
+			if url.endswith('/files.getUploadURLExternal'):
+				return Resp({'ok': True, 'upload_url': self.UPLOAD_URL, 'file_id': next(uploads)})
+			return answer
+		self.patch(transports.net, 'post', post)
+		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U')) # F1, in message 1.5
+		for _ in range(2):
+			worker.process_callback(inbound.Press('slack', 't1', 'C1', '1.5', 'c|l|w|5,0')) # F2, then F3
+		deleted = [kwargs['data']['file'] for url, kwargs in self.posts if url.endswith('/files.delete')]
+		self.assertEqual((deleted, self.errors), (['F1', 'F2'], []))
+		self.assertEqual(self.blocks('chat.update')[1]['slack_file'], {'id': 'F3'})
+
 	def test_a_chart_not_ready_to_show_is_shown_a_moment_later(self):
 		"""Slack refusing the blocks of a chart just uploaded (invalid_blocks) is asked again after a wait."""
 		self.refusals['chat.postMessage'] = ['invalid_blocks']

From 501412c06e96cfd9ce6c6c7d098b30cf27d1c168 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 10:42:04 +0000
Subject: [PATCH 72/81] Retire chunk_maxlines: a long report splits by size
 alone

chunk_maxlines capped the lines per message only of a report already too long for one message, which the README did
not say (it read as a split of any longer report). The size-aware split does the job, so the preference goes (D23): a
long report now fills each message up to the network's limit instead of 20 lines. preferences.conf files that still
set it load as before, and the log at start says to delete the line, as for alliterate.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 README.md                 |  3 +--
 defaults/preferences.conf |  1 -
 lib/config.py             |  4 ++--
 lib/transports.py         | 17 ++++++++---------
 tests/test_config.py      |  6 +++---
 tests/test_finbot.py      | 10 +++++-----
 6 files changed, 19 insertions(+), 22 deletions(-)

diff --git a/README.md b/README.md
index 260668be..82da6103 100644
--- a/README.md
+++ b/README.md
@@ -506,7 +506,6 @@ The preferences, read by the bot itself. A change takes effect on a reload. A tr
 | `hyperlink` | `True` | Hyperlink tickers in reports |
 | `hyperlinkFooter` | `True` | Add a website link footer to stock lookups |
 | `hyperlinkProvider` | `google` | Where ticker links go: `google` or `yahoo` |
-| `chunk_maxlines` | `20` | Maximum lines per message before a report is split |
 | `exchange_suffix` | `AX` | Regional: Yahoo's suffix for your exchange (`AX` for the ASX, `NZ`, `L`, `TO` and so on). A bare symbol Yahoo does not know is tried again with it, in every command that takes a symbol (`BHP` becomes `BHP.AX`); empty, a symbol is taken as written |
 | `country_code` | `AU` | Regional: your country, for the reminder report |
 | `timezone` | `Australia/Melbourne` | Regional: your time zone, for the reminder report |
@@ -517,7 +516,7 @@ The preferences, read by the bot itself. A change takes effect on a reload. A tr
 | `debug` | `False` | Extra diagnostics in the log |
 | `print_headers` | `False` | Log inbound request headers |
 
-A key in the wrong file, or one that is not listed, stops finbot starting; on a reload it is reported and nothing is applied. A key finbot no longer reads (`alliterate`) is ignored, and the log at start says to delete it.
+A key in the wrong file, or one that is not listed, stops finbot starting; on a reload it is reported and nothing is applied. A key finbot no longer reads (`alliterate`, `chunk_maxlines`) is ignored, and the log at start says to delete it.
 
 edit `preferences.conf`, then reload if running. The service log shows what changed (`config: future_days 30 -> 45`). If the file has a mistake, the reload reports it (`ERROR config: ...`), applies nothing and leaves finbot running on its previous settings. Check an edit before reloading with `util/check_config.py /opt/finbot.env`. Edit `finbot.env`, then restart: a running bot cannot see it change.
 
diff --git a/defaults/preferences.conf b/defaults/preferences.conf
index 5f15f0e6..fbef112a 100644
--- a/defaults/preferences.conf
+++ b/defaults/preferences.conf
@@ -31,7 +31,6 @@ graph=True
 hyperlink=True
 hyperlinkFooter=True
 hyperlinkProvider=google
-chunk_maxlines=20
 
 # Regional: where you are.
 # exchange_suffix: a ticker Yahoo does not know is tried again on your exchange, by Yahoo's suffix for it (AX: BHP is
diff --git a/lib/config.py b/lib/config.py
index f9ba4548..970afaf5 100644
--- a/lib/config.py
+++ b/lib/config.py
@@ -21,7 +21,8 @@
 	'slackOutgoingToken', 'slackOutgoingWebhook', 'slackSigningSecret', 'telegramBotToken', 'telegramOutgoingToken',
 	'telegramOutgoingWebhook') # never printed
 PREFERENCES_FILE = 'preferences.conf'
-RETIRED = ('alliterate',) # preferences.conf keys that do nothing any more: accepted, with a note at start to delete
+RETIRED = ('alliterate', 'chunk_maxlines') # preferences.conf keys that do nothing any more: accepted, with a note at
+# start to delete them
 PASSED = 'FINBOT_PREFERENCES_PASSED' # set by the bot for its job processes: the preferences arrive in the environment, do not read the file
 OLD_ENV = REPO_DIR / '.env' # the single config file of old versions, which python-dotenv loaded; nothing reads it now
 
@@ -246,7 +247,6 @@ class Prefs:
 	default or a module constant, so that a reload is seen."""
 	cache: bool = True
 	cache_seconds: int = 43200
-	chunk_maxlines: int = 20
 	country_code: str = 'AU'
 	debug: bool = False
 	demote_leveraged: bool = True
diff --git a/lib/transports.py b/lib/transports.py
index 5e4ed38c..d386ebdb 100644
--- a/lib/transports.py
+++ b/lib/transports.py
@@ -77,9 +77,8 @@ def is_typing(self):
 		return bool(self._notices and self._notices.is_alive())
 
 	def text(self, lines, buttons=None):
-		"""Post lines of text: one message, or several when they are too long for one (then at most chunk_maxlines lines
-		each), buttons under the last; the typing notice stops first. Returns the network's answer for each message
-		(None for one that failed)."""
+		"""Post lines of text: one message, or several when they are too long for one, buttons under the last; the
+		typing notice stops first. Returns the network's answer for each message (None for one that failed)."""
 		self.stop_typing()
 		lines = [str(line) for line in lines]
 		if not lines:
@@ -87,7 +86,7 @@ def text(self, lines, buttons=None):
 		limit = self.dialect.text_limit
 		if len('\n'.join(lines)) <= limit:
 			return [self.post_text('\n'.join(lines), buttons)]
-		chunks = chunked(lines, limit, config.prefs.chunk_maxlines)
+		chunks = chunked(lines, limit)
 		answers = []
 		for i, chunk in enumerate(chunks):
 			if i:
@@ -464,7 +463,7 @@ def slack_api(method, **data):
 def section_blocks(text):
 	"""Block Kit sections of a message's text, as many as its length needs (none for no text)."""
 	return [{'type': 'section', 'text': {'type': 'mrkdwn', 'text': '\n'.join(part)}}
-		for part in chunked(text.split('\n'), SECTION_LIMIT, 1000) if '\n'.join(part).strip()]
+		for part in chunked(text.split('\n'), SECTION_LIMIT) if '\n'.join(part).strip()]
 
 
 def actions_block(row):
@@ -501,14 +500,14 @@ def until_shown(post, waits=(1, 2, 4)):
 	return post()
 
 
-def chunked(lines, limit, max_lines):
-	"""The lines in messages of at most max_lines lines and limit characters each, whichever comes first; a line longer
-	than limit on its own is cut short."""
+def chunked(lines, limit):
+	"""The lines in messages of at most limit characters each, whole lines to a message; a line longer than limit on
+	its own is cut short."""
 	chunks, current, size = [], [], 0
 	for line in lines:
 		if len(line) > limit:
 			line = line[:limit - 1] + '…'
-		if current and (len(current) >= max_lines or size + 1 + len(line) > limit):
+		if current and size + 1 + len(line) > limit:
 			chunks.append(current)
 			current, size = [], 0
 		size += len(line) + (1 if current else 0)
diff --git a/tests/test_config.py b/tests/test_config.py
index 8cbbdee0..b527850c 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -132,11 +132,11 @@ def test_a_retired_key_is_accepted_and_said_to_do_nothing(self):
 		"""A key finbot no longer reads (alliterate) does not stop a start, as an unknown key would: it is ignored, and
 		the log at start says to delete it."""
 		path = Path(common.temp_dir(self), 'preferences.conf')
-		path.write_text('alliterate=True\nfuture_days=45\n')
+		path.write_text('alliterate=True\nchunk_maxlines=20\nfuture_days=45\n')
 		values = config.read_preferences(path)
 		self.assertEqual(config.preference_values(config._getenv_for(values)).future_days, 45)
-		self.assertEqual(config.retired_notes(values),
-			['alliterate in preferences.conf does nothing any more: delete the line'])
+		notes = [f'{key} in preferences.conf does nothing any more: delete the line' for key in config.RETIRED]
+		self.assertEqual(config.retired_notes(values), notes)
 		self.assertEqual(check_config.notes(path), config.retired_notes(values))
 
 	def test_unreadable_file_is_an_error_naming_it(self):
diff --git a/tests/test_finbot.py b/tests/test_finbot.py
index 73a5b55e..79cd70b3 100644
--- a/tests/test_finbot.py
+++ b/tests/test_finbot.py
@@ -28,13 +28,13 @@ def post(url, **kwargs):
 		self.assertIn('reply_markup', posted[-1])
 		self.assertTrue(all('reply_markup' not in p for p in posted[:-1]))
 
-	def test_chunks_respect_both_the_line_count_and_the_size(self):
-		"""A message ends at chunk_maxlines lines or the size limit, whichever first; an overlong line is cut short."""
-		chunks = transports.chunked(['a' * 30] * 10 + ['b' * 500], limit=100, max_lines=4)
-		self.assertEqual([len(chunk) for chunk in chunks], [3, 3, 3, 1, 1]) # 3 x 30 plus newlines is 92; a fourth would pass 100
+	def test_chunks_respect_the_size(self):
+		"""A message ends before the line that would take it past the size limit; an overlong line is cut short."""
+		chunks = transports.chunked(['a' * 30] * 10 + ['b' * 500], limit=100)
+		self.assertEqual([len(chunk) for chunk in chunks], [3, 3, 3, 1, 1]) # 3 x 30 and 2 newlines: 92; 4 pass 100
 		self.assertTrue(all(len('\n'.join(chunk)) <= 100 for chunk in chunks))
 		self.assertEqual(chunks[-1], ['b' * 99 + '…']) # a line over the limit on its own is cut short
-		self.assertEqual(transports.chunked(['x'] * 9, limit=4096, max_lines=4), [['x'] * 4, ['x'] * 4, ['x']])
+		self.assertEqual(transports.chunked(['x'] * 9, limit=4096), [['x'] * 9]) # however many lines fit
 
 	def test_zoneinfo_offset_is_not_lmt(self):
 		"""zoneinfo gives Melbourne its real offset, not the local mean time pytz once gave for old dates."""

From 84eae66792bba2838ab9b5d4c5136fec168aaa5e Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 11:06:28 +0000
Subject: [PATCH 73/81] Every line within 120 columns, and a test that keeps it
 so (D24)

418 lines of code, tests, shipped defaults, units and tools were over 120 columns (tabs counting 4), which every edit
quotes whole. They are rewrapped: comments and docstrings re-flowed (a one-line docstring that no longer fit was
reworded to fit, as the convention asks), code split after a comma inside its brackets or given a name for a part, the
odd long string joined from two. Nothing runs differently: the tests pass and the recorder finds every command,
button and scheduled report the same. A test in test_finbot holds every line to 120 columns from now on; README.md,
CLAUDE.md and doc/ are left out, since CI runs no tests for a change to those alone, and so are data files.

Along the way the tests' repeated "environment without finbot's settings" became common.bare_environment().

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 .claude/hooks/session-start.sh |   3 +-
 CLAUDE.md                      |   6 +-
 bot.py                         |  40 +++++-----
 defaults/finbot.env            |  15 ++--
 defaults/preferences.conf      |  13 ++--
 defaults/schedule.cron         |   5 +-
 finbot.service                 |   6 +-
 lib/cache.py                   |  18 ++---
 lib/config.py                  |  85 ++++++++++++---------
 lib/dialects.py                |  15 ++--
 lib/formats.py                 |  14 ++--
 lib/jobrun.py                  |  14 ++--
 lib/jobs.py                    |  73 ++++++++++--------
 lib/net.py                     |  16 ++--
 lib/render.py                  | 100 +++++++++++++++----------
 lib/reports/cal.py             |   5 +-
 lib/reports/cleanup.py         |   4 +-
 lib/reports/holdings.py        |   3 +-
 lib/reports/performance.py     |   6 +-
 lib/reports/podcasts.py        |  16 ++--
 lib/reports/price.py           |  16 ++--
 lib/reports/profile.py         |  10 ++-
 lib/reports/rankings.py        |   9 ++-
 lib/reports/rating.py          |   2 +-
 lib/sharesight.py              |   3 +-
 lib/simplywallst.py            |   6 +-
 lib/state.py                   |  13 ++--
 lib/telegram.py                |  17 +++--
 lib/watchlist.py               |   4 +-
 lib/worker.py                  |   2 +-
 lib/yahoo.py                   |  52 ++++++++-----
 tests/common.py                |  30 ++++++--
 tests/test_charts.py           |  20 +++--
 tests/test_config.py           |  83 +++++++++++++--------
 tests/test_datadir.py          |  24 ++++--
 tests/test_edges.py            |  79 ++++++++++++--------
 tests/test_finbot.py           |  20 ++++-
 tests/test_jobs.py             | 131 +++++++++++++++++++++------------
 tests/test_providers.py        | 103 +++++++++++++++++---------
 tests/test_reports.py          | 130 ++++++++++++++++++++------------
 tests/test_util.py             |  20 +++--
 util/check_config.py           |  16 ++--
 util/google_namecheap_cron     |   3 +-
 util/iptables_telegram_cron    |   3 +-
 util/split_env.py              |  26 ++++---
 util/sync_defaults.py          |   7 +-
 46 files changed, 805 insertions(+), 481 deletions(-)

diff --git a/.claude/hooks/session-start.sh b/.claude/hooks/session-start.sh
index e3468826..f442c27a 100755
--- a/.claude/hooks/session-start.sh
+++ b/.claude/hooks/session-start.sh
@@ -13,7 +13,8 @@ PYTHON_VERSION="$(tr -d '[:space:]' < .python-version)"
 uv python install "$PYTHON_VERSION"
 
 # (re)create .venv only when it is missing or on a different Python
-if [ ! -x .venv/bin/python ] || [ "$(.venv/bin/python -c 'import platform; print(platform.python_version())')" != "$PYTHON_VERSION" ]; then
+if [ ! -x .venv/bin/python ] ||
+	[ "$(.venv/bin/python -c 'import platform; print(platform.python_version())')" != "$PYTHON_VERSION" ]; then
 	rm -rf .venv
 	uv venv --python "$PYTHON_VERSION" .venv
 fi
diff --git a/CLAUDE.md b/CLAUDE.md
index 080d940f..14ba1158 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -75,7 +75,7 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - Which test file: commands and scheduled reports `test_reports`; failures, bad input, WSGI and Slack events `test_edges`;
   Yahoo/Sharesight parsing, sending, Telegram API `test_providers`; scheduler `test_jobs`; config `test_config`; charts
   `test_charts` (refresh goldens with `UPDATE_GOLDEN=1`); markets, formats, links, cache and state files `test_util`; data dir
-  `test_datadir`; chunking, git_version `test_finbot`.
+  `test_datadir`; chunking, git_version, and the docstring, stdout and line-width checks `test_finbot`.
 
 ## Conventions
 - Read settings as `config.env.<name>` / `config.prefs.<name>` where they are used, never into a default or module constant (a
@@ -86,8 +86,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - A command that cannot give what was asked (no such portfolio, no prices) raises `answer.Answer(*lines)`, which the
   chat gets as a plain reply; `yahoo.NoData` is one. Any other exception becomes a one-line ⚠️ error report.
 - Every function and method has a docstring, tests included (a test checks): one line where it fits, reworded rather
-  than spilling a few words onto a second. Tabs everywhere (`.editorconfig`). New and changed lines stay within 120
-  columns, tabs counting 4 (an edit quotes whole lines).
+  than spilling a few words onto a second. Tabs everywhere (`.editorconfig`). Every line stays within 120 columns,
+  tabs counting 4 (an edit quotes whole lines); a test checks all but README.md, CLAUDE.md and `doc/`.
 - Log with `print(..., file=sys.stderr)`. systemd discards stdout (`StandardOutput=null`), so the bot and its reports never
   print there (a test checks); only the util/ tools, run by hand, do.
 - A report gives its chart as a `render.Chart` spec, not a drawing: `deliver()` draws it only for a chat that takes
diff --git a/bot.py b/bot.py
index 1b5ec23f..c3367c94 100755
--- a/bot.py
+++ b/bot.py
@@ -2,7 +2,8 @@
 
 import signal
 if __name__ == '__main__':
-	signal.signal(signal.SIGHUP, signal.SIG_IGN) # ignored until startup is done (the reload handler is installed last): its default action kills the bot
+	# ignored until startup is done (the reload handler is installed last): its default action kills the bot
+	signal.signal(signal.SIGHUP, signal.SIG_IGN)
 	signal.signal(signal.SIGUSR1, signal.SIG_IGN) # likewise the cache flush
 import gevent.monkey
 gevent.monkey.patch_all()
@@ -28,8 +29,9 @@
 _updates_seen = collections.OrderedDict() # (service, update id) of the last updates taken, oldest first
 
 def seen_before(service, update_id, keep=1000):
-	"""Whether this update was taken already: Telegram's update_id, Slack's event_id. Slack resends an event it got no answer to
-	within 3 seconds and Telegram one that failed, and a command must not run twice. An update without an id is never a repeat."""
+	"""Whether this update was taken already: Telegram's update_id, Slack's event_id. Slack resends an event it got no
+	answer to within 3 seconds and Telegram one that failed, and a command must not run twice. An update without an id
+	is never a repeat."""
 	if update_id is None:
 		return False
 	key = (service, update_id)
@@ -41,10 +43,10 @@ def seen_before(service, update_id, keep=1000):
 	return False
 
 def main(environ, start_response):
-	"""WSGI entry point. Never lets an exception escape. A failure handling an authenticated request is reported as one line to the
-	default channel and answered 200, so the service does not retry it into the chat again. Before authentication a failure is only
-	logged: a stranger must not be able to post into the chats. Either names the network, not the path: a webhook's
-	path is a secret (telegramOutgoingWebhook, slackOutgoingWebhook)."""
+	"""WSGI entry point. Never lets an exception escape. A failure handling an authenticated request is reported as one
+	line to the default channel and answered 200, so the service does not retry it into the chat again. Before
+	authentication a failure is only logged: a stranger must not be able to post into the chats. Either names the
+	network, not the path: a webhook's path is a secret (telegramOutgoingWebhook, slackOutgoingWebhook)."""
 	try:
 		return handle(environ, start_response)
 	except Exception as e:
@@ -128,13 +130,14 @@ def dispatch(event):
 	threading.Thread(target=target, args=(event,)).start()
 
 def git_version(checkout=None):
-	"""Short hash of the checked-out commit, read straight from the .git directory: no git binary to run, and no "dubious
-	ownership" refusal when the service user does not own the checkout. 'unknown' when it cannot be read."""
+	"""Short hash of the checked-out commit, read straight from the .git directory: no git binary to run, and no
+	"dubious ownership" refusal when the service user does not own the checkout. 'unknown' when it cannot be read."""
 	try:
 		git = Path(checkout or config.REPO_DIR) / '.git'
 		if git.is_file(): # a linked worktree or submodule: .git is a file saying "gitdir: <path>"
 			git = (git.parent / git.read_text().split(':', 1)[1].strip()).resolve()
-		common = (git / (git / 'commondir').read_text().strip()).resolve() if (git / 'commondir').is_file() else git # refs of a worktree
+		# refs of a worktree
+		common = (git / (git / 'commondir').read_text().strip()).resolve() if (git / 'commondir').is_file() else git
 		head = (git / 'HEAD').read_text().strip()
 		if head.startswith('ref:'):
 			ref = head.split(':', 1)[1].strip()
@@ -160,25 +163,27 @@ def run():
 	return lambda *args: threading.Thread(target=run, daemon=True).start()
 
 def reload_all(*args):
-	"""SIGHUP: re-read preferences.conf and, if the scheduler is running, schedule.cron. Each reports its own errors; a bad
-	preferences.conf leaves the running config untouched."""
+	"""SIGHUP: re-read preferences.conf and, if the scheduler is running, schedule.cron. Each reports its own errors; a
+	bad preferences.conf leaves the running config untouched."""
 	try:
 		changes = config.reload()
 		for change in changes:
 			print('config:', change, file=sys.stderr)
-		print(f"config: reloaded {config.preferences_path}, {len(changes)} change{'' if len(changes) == 1 else 's'}", file=sys.stderr)
+		print(f"config: reloaded {config.preferences_path}, {len(changes)} change{'' if len(changes) == 1 else 's'}",
+			file=sys.stderr)
 	except config.ConfigError as e:
 		config.print_errors(e)
 	if scheduler:
 		scheduler.reload()
 
 def flush_cache(*args):
-	"""SIGUSR1 (`systemctl kill -s USR1 finbot`): delete the cache and keep the state (watchlist, seen trades, ratings, schedule).
-	The Sharesight caches refill at once."""
+	"""SIGUSR1 (`systemctl kill -s USR1 finbot`): delete the cache and keep the state (watchlist, seen trades, ratings,
+	schedule). The Sharesight caches refill at once."""
 	try:
 		removed = cleanup.run(days=0)
 		sharesight.wake()
-		print(f"cache: flushed {removed} file{'' if removed == 1 else 's'} from {config.env.cache_dir}", file=sys.stderr)
+		print(f"cache: flushed {removed} file{'' if removed == 1 else 's'} from {config.env.cache_dir}",
+			file=sys.stderr)
 	except Exception as e:
 		print('ERROR', delivery.error_line(e, 'cache flush'), file=sys.stderr)
 
@@ -220,7 +225,8 @@ def flush_cache(*args):
 	if config.env.telegramBotToken:
 		threading.Thread(target=worker.refresh_keyboards, daemon=True).start() # stale DM keyboards, in the background
 	if config.env.sharesight_auth.get('client_id'):
-		threading.Thread(target=sharesight.keep_warm, daemon=True).start() # keep the portfolio list and holdings report cached, while cache is on
+		# keep the portfolio list and holdings report cached, while cache is on
+		threading.Thread(target=sharesight.keep_warm, daemon=True).start()
 	if scheduler:
 		jobs.current = scheduler # for the .jobs chat command
 		threading.Thread(target=scheduler.run_forever, daemon=True).start() # the scheduled reports, formerly cron jobs
diff --git a/defaults/finbot.env b/defaults/finbot.env
index 282edafe..cdbb8783 100644
--- a/defaults/finbot.env
+++ b/defaults/finbot.env
@@ -4,18 +4,23 @@
 #   docker:   --env-file /opt/finbot.env
 #   init.d:   set -a; . /opt/finbot.env; set +a; ./bot.py
 # Keep it private: chmod 600, owned by root. It holds the credentials.
-# Flat KEY=value lines: no spaces around "=", no quotes (docker keeps quotes literally), no trailing comments; comments on their own line.
-# Everything else (thresholds, windows, display, caching, ...) is in preferences.conf, which the bot reloads without a restart.
-# A preference key in this file is an error: finbot refuses to start. Lint both files with: util/check_config.py /opt/finbot.env
+# Flat KEY=value lines: no spaces around "=", no quotes (docker keeps quotes literally), no trailing comments; comments
+# on their own line.
+# Everything else (thresholds, windows, display, caching, ...) is in preferences.conf, which the bot reloads without a
+# restart.
+# A preference key in this file is an error: finbot refuses to start. Lint both files with:
+#   util/check_config.py /opt/finbot.env
 
-# Where state and cache live. Leave unset under systemd with StateDirectory=finbot (it uses $STATE_DIRECTORY, /var/lib/finbot).
+# Where state and cache live. Leave unset under systemd with StateDirectory=finbot (it uses $STATE_DIRECTORY,
+# /var/lib/finbot).
 # DATA_DIR=/var/lib/finbot
 
 # The listening socket
 ip=127.0.0.1
 port=5000
 
-# Run the scheduled reports (schedule.cron) inside the bot, and how many minutes late a run missed while the bot was down may still fire
+# Run the scheduled reports (schedule.cron) inside the bot, and how many minutes late a run missed while the bot was
+# down may still fire
 scheduler=True
 scheduler_catchup=5
 
diff --git a/defaults/preferences.conf b/defaults/preferences.conf
index fbef112a..95f52c02 100644
--- a/defaults/preferences.conf
+++ b/defaults/preferences.conf
@@ -1,17 +1,20 @@
 # Preferences for finbot: how the bot behaves. The bot reads this file itself, at start and on every reload.
 #   Apply an edit without a restart:  sudo systemctl reload finbot     (docker: docker kill -s HUP <container>)
 #   Check an edit first:              util/check_config.py /path/to/finbot.env
-# Flat KEY=value lines, one per line. This is NOT an INI file: no [sections]. No spaces around "=", no quotes, no trailing comments;
-# put comments on their own line. A key that is not listed here is an error and finbot refuses to start.
-# Each value shown is the built-in default. The startup keys (credentials, ip, port, telegramChatID, scheduler, ...) are NOT set here:
-# they live in finbot.env, the startup file, which the launcher loads and the bot never reads. Changing them needs a restart.
+# Flat KEY=value lines, one per line. This is NOT an INI file: no [sections]. No spaces around "=", no quotes, no
+# trailing comments; put comments on their own line. A key that is not listed here is an error and finbot refuses to
+# start.
+# Each value shown is the built-in default. The startup keys (credentials, ip, port, telegramChatID, scheduler, ...)
+# are NOT set here: they live in finbot.env, the startup file, which the launcher loads and the bot never reads.
+# Changing them needs a restart.
 
 # Reports: how far back and forward to look, and when to alert
 past_days=28
 future_days=30
 price_percent=5
 shorts_percent=15
-# .buy, .sell and .forward pe list only stocks that at least this many analysts cover: with fewer, a consensus says little
+# .buy, .sell and .forward pe list only stocks that at least this many analysts cover: with fewer, a consensus says
+# little
 min_analysts=7
 # price alerts: volatile and leveraged stocks need a bigger move; the multiplier is how much bigger
 demote_volatile=True
diff --git a/defaults/schedule.cron b/defaults/schedule.cron
index d11f5fe4..bb52b055 100644
--- a/defaults/schedule.cron
+++ b/defaults/schedule.cron
@@ -1,5 +1,6 @@
-# The scheduled reports, run by the bot (bot.py). Same time syntax as crontab(5): minute hour day-of-month month day-of-week,
-# or an @macro such as @daily. The command is a finbot script and its arguments; separate several with ";" to run them in turn.
+# The scheduled reports, run by the bot (bot.py). Same time syntax as crontab(5): minute hour day-of-month month
+# day-of-week, or an @macro such as @daily. The command is a finbot script and its arguments; separate several with ";"
+# to run them in turn.
 # Times are in the server's own time zone, like cron; add a line "CRON_TZ=Australia/Melbourne" above a job to pin one.
 # To change the schedule, copy this file to the data directory (util/sync_defaults.py --apply), edit the copy, then:
 #   systemctl reload finbot
diff --git a/finbot.service b/finbot.service
index 2fff0c7a..fd74abf0 100644
--- a/finbot.service
+++ b/finbot.service
@@ -8,7 +8,8 @@ User=CHANGEME
 ExecStart=/home/CHANGEME/finbot/bot.py
 # systemctl reload finbot re-reads preferences.conf and schedule.cron without a restart
 ExecReload=/bin/kill -HUP $MAINPID
-# Refuse to start on a mistake in finbot.env or preferences.conf (the bot never reads finbot.env itself, so this is where its typos are caught). "+" runs it as root.
+# Refuse to start on a mistake in finbot.env or preferences.conf (the bot never reads finbot.env itself, so this is
+# where its typos are caught). "+" runs it as root.
 ExecStartPre=+/home/CHANGEME/finbot/util/check_config.py /opt/finbot.env
 Restart=on-failure
 RestartSec=30s
@@ -16,7 +17,8 @@ StandardOutput=null
 StandardError=journal
 SyslogIdentifier=finbot
 Environment=PYTHONUNBUFFERED=1
-# finbot.env is the startup file: systemd loads it (as root), the bot only inherits the values. chmod 600. A change needs a restart.
+# finbot.env is the startup file: systemd loads it (as root), the bot only inherits the values. chmod 600. A change
+# needs a restart.
 EnvironmentFile=/opt/finbot.env
 
 [Install]
diff --git a/lib/cache.py b/lib/cache.py
index a7a31c36..7c8bb7ac 100644
--- a/lib/cache.py
+++ b/lib/cache.py
@@ -1,5 +1,5 @@
-"""Cache files in DATA_DIR/cache: copies of Yahoo and Sharesight answers and of drawn charts, each good for so many seconds.
-Anything here may be deleted at any time (cleanup.py, a flush); what must last lives in lib/state.py."""
+"""Cache files in DATA_DIR/cache: copies of Yahoo and Sharesight answers and of drawn charts, each good for so many
+seconds. Anything here may be deleted at any time (cleanup.py, a flush); what must last lives in lib/state.py."""
 import io
 import json
 import os
@@ -25,7 +25,7 @@ def age(name):
 
 
 def _fresh(name, seconds):
-	"""The cache file's path if it was written less than `seconds` ago (the cache_seconds preference when None), else None."""
+	"""The cache file's path if written less than `seconds` ago (cache_seconds when None), else None."""
 	if seconds is None:
 		seconds = config.prefs.cache_seconds
 	where, written = path(name), age(name)
@@ -40,7 +40,7 @@ def _fresh(name, seconds):
 
 
 def _read_fresh(name, seconds, read):
-	"""read(path) of the cache file if it is fresh, else None; also None if the file goes (cleanup, a flush) after the check."""
+	"""read(path) of the cache file if fresh, else None; also None if it goes (cleanup, a flush) after the check."""
 	where = _fresh(name, seconds)
 	if not where:
 		return None
@@ -51,7 +51,7 @@ def _read_fresh(name, seconds, read):
 
 
 def read(name, seconds=None):
-	"""A cache file's JSON, or None when it is missing or older than `seconds` (the cache_seconds preference when None)."""
+	"""A cache file's JSON, or None when it is missing or older than `seconds` (cache_seconds when None)."""
 	return _read_fresh(name, seconds, lambda where: json.loads(where.read_text(encoding="utf-8")))
 
 
@@ -61,7 +61,7 @@ def read_bytes(name, seconds=None):
 
 
 def load(name):
-	"""A cache file's JSON whatever its age, or None: for an answer that carries its own expiry (the Sharesight token)."""
+	"""A cache file's JSON whatever its age, or None: for an answer with its own expiry (the Sharesight token)."""
 	where = Path(path(name))
 	try:
 		return json.loads(where.read_text(encoding="utf-8"))
@@ -96,9 +96,9 @@ def write_bytes(name, data):
 
 
 def write_file(where, data, sync=False):
-	"""Write `data` (bytes) to `where` whole or not at all, mode 0640: it goes to a temp file beside it, which then replaces the
-	file, so a failure part way (a full disk, a kill) leaves the old file as it was. sync=True also flushes it to disk, for state
-	that must survive a power cut. The process umask is left alone."""
+	"""Write `data` (bytes) to `where` whole or not at all, mode 0640: it goes to a temp file beside it, which then
+	replaces the file, so a failure part way (a full disk, a kill) leaves the old file as it was. sync=True also flushes
+	it to disk, for state that must survive a power cut. The process umask is left alone."""
 	where = os.path.realpath(where) # through a symlink, as an in-place write would
 	fd, tmp = tempfile.mkstemp(dir=os.path.dirname(where), prefix=os.path.basename(where) + '.', suffix='.tmp')
 	try:
diff --git a/lib/config.py b/lib/config.py
index 970afaf5..90569936 100644
--- a/lib/config.py
+++ b/lib/config.py
@@ -7,23 +7,26 @@
 from pathlib import Path
 
 # TO CONFIGURE, DO NOT EDIT THIS FILE. There are two config files:
-#   finbot.env       the startup file: DATA_DIR, ip, port, credentials, telegramChatID, scheduler. Never read by the bot: the launcher loads it
-#                    into the environment (systemd EnvironmentFile=, docker --env-file, init.d `set -a; . finbot.env`). A change needs a restart.
-#   preferences.conf how the bot behaves: <DATA_DIR>/preferences.conf, read by the bot itself at start and on every reload (SIGHUP).
+#   finbot.env       the startup file: DATA_DIR, ip, port, credentials, telegramChatID, scheduler. Never read by the
+#                    bot: the launcher loads it into the environment (systemd EnvironmentFile=, docker --env-file,
+#                    init.d `set -a; . finbot.env`). A change needs a restart.
+#   preferences.conf how the bot behaves: <DATA_DIR>/preferences.conf, read by the bot itself at start and on every
+#                    reload (SIGHUP).
 
 REPO_DIR = Path(__file__).resolve().parent.parent
 
-STARTUP_KEYS = ('DATA_DIR', 'ip', 'port', 'telegramChatID', 'scheduler', 'scheduler_catchup',
-	'sharesight_code', 'sharesight_client_id', 'sharesight_client_secret', 'slack_webhook', 'discord_webhook', 'slackBotToken',
+STARTUP_KEYS = ('DATA_DIR', 'ip', 'port', 'telegramChatID', 'scheduler', 'scheduler_catchup', 'sharesight_code',
+	'sharesight_client_id', 'sharesight_client_secret', 'slack_webhook', 'discord_webhook', 'slackBotToken',
 	'slackChannelID', 'slackOutgoingToken', 'slackOutgoingWebhook', 'slackSigningSecret', 'telegramBotToken',
 	'telegramOutgoingToken', 'telegramOutgoingWebhook')
-SECRET_KEYS = ('sharesight_code', 'sharesight_client_id', 'sharesight_client_secret', 'slack_webhook', 'discord_webhook', 'slackBotToken',
-	'slackOutgoingToken', 'slackOutgoingWebhook', 'slackSigningSecret', 'telegramBotToken', 'telegramOutgoingToken',
-	'telegramOutgoingWebhook') # never printed
+SECRET_KEYS = ('sharesight_code', 'sharesight_client_id', 'sharesight_client_secret', 'slack_webhook',
+	'discord_webhook', 'slackBotToken', 'slackOutgoingToken', 'slackOutgoingWebhook', 'slackSigningSecret',
+	'telegramBotToken', 'telegramOutgoingToken', 'telegramOutgoingWebhook') # never printed
 PREFERENCES_FILE = 'preferences.conf'
 RETIRED = ('alliterate', 'chunk_maxlines') # preferences.conf keys that do nothing any more: accepted, with a note at
 # start to delete them
-PASSED = 'FINBOT_PREFERENCES_PASSED' # set by the bot for its job processes: the preferences arrive in the environment, do not read the file
+# set by the bot for its job processes: the preferences arrive in the environment, do not read the file
+PASSED = 'FINBOT_PREFERENCES_PASSED'
 OLD_ENV = REPO_DIR / '.env' # the single config file of old versions, which python-dotenv loaded; nothing reads it now
 
 
@@ -40,7 +43,8 @@ def key_problem(key, this_keys, other_keys, other_file):
 	if key in this_keys:
 		return None
 	if key in other_keys:
-		return f"{key} belongs in {other_file}" + (' (a change needs a restart)' if other_file == 'finbot.env' else ' (it is reloadable)')
+		why = ' (a change needs a restart)' if other_file == 'finbot.env' else ' (it is reloadable)'
+		return f"{key} belongs in {other_file}" + why
 	if key == 'volatile_multiplier':
 		return 'volatile_multiplier was renamed to volatility_multiplier'
 	near = difflib.get_close_matches(key, list(this_keys) + list(other_keys), n=1)
@@ -48,8 +52,8 @@ def key_problem(key, this_keys, other_keys, other_file):
 
 
 def parse_assignments(text, path, classify):
-	"""KEY=value lines -> ({key: value}, [errors]). Blank and # lines are skipped, one pair of matching quotes is stripped, `classify(key)`
-	returns a reason to refuse the key. Every error names the file and line."""
+	"""KEY=value lines -> ({key: value}, [errors]). Blank and # lines are skipped, one pair of matching quotes is
+	stripped, `classify(key)` returns a reason to refuse the key. Every error names the file and line."""
 	values, errors = {}, []
 	for number, raw in enumerate(text.splitlines(), 1):
 		line = raw.strip()
@@ -88,7 +92,7 @@ def read_text(path):
 
 
 def read_preferences(path):
-	"""{key: value} of a preferences.conf, None if the file does not exist. Raises ConfigError for anything wrong with it."""
+	"""{key: value} of a preferences.conf, None if there is none. Raises ConfigError for anything wrong with it."""
 	text = read_text(path)
 	if text is None:
 		return None
@@ -104,7 +108,8 @@ def read_startup_file(path):
 	text = read_text(path)
 	if text is None:
 		raise ConfigError(f'{path} does not exist')
-	values, errors = parse_assignments(text, path, lambda key: key_problem(key, STARTUP_KEYS, PREFERENCE_KEYS, PREFERENCES_FILE))
+	values, errors = parse_assignments(text, path,
+		lambda key: key_problem(key, STARTUP_KEYS, PREFERENCE_KEYS, PREFERENCES_FILE))
 	if errors:
 		raise ConfigError(errors)
 	return values
@@ -115,8 +120,8 @@ def read_startup_file(path):
 
 
 class _Reader:
-	"""Typed reads from a getenv(name, default); a bad value is remembered, the default used, and all problems raised together.
-	A blank value means unset: the default."""
+	"""Typed reads from a getenv(name, default); a bad value is remembered, the default used, and all problems raised
+	together. A blank value means unset: the default."""
 	def __init__(self, getenv):
 		"""getenv(name, default) supplies the values; the problems found collect in `errors` until check()."""
 		self.getenv = getenv
@@ -175,7 +180,7 @@ def check(self):
 
 @dataclasses.dataclass(frozen=True)
 class Env:
-	"""The startup settings: finbot.env, which the launcher loads into the environment. Fixed for the life of the process."""
+	"""The startup settings: finbot.env, which the launcher loads into the environment; fixed for the process."""
 	data_dir: str # state: watchlist, seen trades, ratings, the schedule's record, preferences.conf
 	cache_dir: str # data_dir/cache
 	ip: str
@@ -209,7 +214,8 @@ def startup_values(getenv):
 	if getenv('slack_webhook'):
 		hooks['slack'] = getenv('slack_webhook').rstrip('/')
 	if getenv('discord_webhook'):
-		hooks['discord'] = getenv('discord_webhook').rstrip('/').removesuffix('/slack') # the native endpoint, not the Slack-compatible one
+		# the native endpoint, not the Slack-compatible one
+		hooks['discord'] = getenv('discord_webhook').rstrip('/').removesuffix('/slack')
 	if getenv('telegramBotToken'):
 		hooks['telegram'] = 'https://api.telegram.org/bot' + getenv('telegramBotToken').rstrip('/') + '/'
 	env = Env(
@@ -270,7 +276,8 @@ class Prefs:
 	timezone: str = 'Australia/Melbourne'
 	trades_use_yahoo: bool = True # set False if Yahoo breaks
 	volatility_multiplier: float = 2.0
-	source: dict = dataclasses.field(default_factory=dict, compare=False, repr=False) # the file's own lines, for job processes
+	# the file's own lines, for job processes
+	source: dict = dataclasses.field(default_factory=dict, compare=False, repr=False)
 
 
 def preference_values(getenv, source=None):
@@ -287,7 +294,8 @@ def preference_values(getenv, source=None):
 	return Prefs(**values, source=dict(source or {}))
 
 
-PREFERENCE_KEYS = tuple(field.name for field in dataclasses.fields(Prefs) if field.compare) # preferences.conf's keys: what a reload can change
+# preferences.conf's keys: what a reload can change
+PREFERENCE_KEYS = tuple(field.name for field in dataclasses.fields(Prefs) if field.compare)
 
 
 def _getenv_for(preferences):
@@ -320,7 +328,8 @@ def _fail(e):
 	_fail(e)
 
 for _dir in (env.data_dir, env.cache_dir):
-	if os.environ.get('FINBOT_CHECK_ONLY'): # util/check_config.py, possibly run as root: must not create directories the service user then cannot write
+	# util/check_config.py, possibly run as root: must not create directories the service user then cannot write
+	if os.environ.get('FINBOT_CHECK_ONLY'):
 		break
 	try:
 		os.makedirs(_dir, exist_ok=True)
@@ -338,8 +347,8 @@ def data_file(name):
 _data_files = {} # path -> (modification time, parsed JSON)
 
 def data_json(name, default=None):
-	"""A data file's JSON (the user's copy or the shipped default, as data_file has it), or `default` when there is neither. Parsed
-	again only when the file changes; callers must not modify what they get."""
+	"""A data file's JSON (the user's copy or the shipped default, as data_file has it), or `default` when there is
+	neither. Parsed again only when the file changes; callers must not modify what they get."""
 	path = data_file(name)
 	try:
 		mtime = os.path.getmtime(path)
@@ -354,8 +363,9 @@ def data_json(name, default=None):
 
 
 def reload():
-	"""Re-read preferences.conf and apply it: config.prefs is replaced in one assignment, so a request never sees half of a reload.
-	All or nothing. Returns one line per change; raises ConfigError (nothing applied) if the file is missing, unreadable or invalid."""
+	"""Re-read preferences.conf and apply it: config.prefs is replaced in one assignment, so a request never sees half
+	of a reload. All or nothing. Returns one line per change; raises ConfigError (nothing applied) if the file is
+	missing, unreadable or invalid."""
 	global prefs
 	new = read_preferences(preferences_path)
 	if new is None:
@@ -373,29 +383,34 @@ def child_environment():
 
 
 def check_startup_environment():
-	"""For bot.py at start. Preferences must come from preferences.conf, not from the environment that finbot.env fills, and a
-	leftover .env from before the split stops the start: its settings would otherwise be silently ignored."""
+	"""For bot.py at start. Preferences must come from preferences.conf, not from the environment that finbot.env fills,
+	and a leftover .env from before the split stops the start: its settings would otherwise be silently ignored."""
 	if os.environ.get(PASSED):
 		return
-	problems = [f'{key} is a preference; move it from finbot.env to {PREFERENCES_FILE}' for key in PREFERENCE_KEYS if key in os.environ]
+	problems = [f'{key} is a preference; move it from finbot.env to {PREFERENCES_FILE}'
+		for key in PREFERENCE_KEYS if key in os.environ]
 	if 'volatile_multiplier' in os.environ: # the old name, from before the split: it never worked as a preference
-		problems.append(f'volatile_multiplier was renamed to volatility_multiplier, a preference; move it from finbot.env to {PREFERENCES_FILE}')
+		problems.append('volatile_multiplier was renamed to volatility_multiplier, a preference; move it from'
+			f' finbot.env to {PREFERENCES_FILE}')
 	if OLD_ENV.exists():
-		problems.append(f'{OLD_ENV} is the old single config file, which finbot no longer reads; it will not start while the file is there')
-		problems.append(f'split it with {REPO_DIR}/util/split_env.py, which writes finbot.env.new and preferences.conf.new beside it; put finbot.env '
-			f'where your launcher loads it and {PREFERENCES_FILE} in {env.data_dir}, then delete {OLD_ENV}')
+		problems.append(f'{OLD_ENV} is the old single config file, which finbot no longer reads; it will not start'
+			' while the file is there')
+		problems.append(f'split it with {REPO_DIR}/util/split_env.py, which writes finbot.env.new and'
+			f' preferences.conf.new beside it; put finbot.env where your launcher loads it and {PREFERENCES_FILE} in'
+			f' {env.data_dir}, then delete {OLD_ENV}')
 	if problems:
 		raise ConfigError(problems)
 
 
 def startup_report():
-	"""Lines for the log at start: where the config came from (a count, never names or values; util/check_config.py names the problems)."""
+	"""Lines for the log at start: where the config came from (counts only; check_config names the problems)."""
 	provided = [key for key in STARTUP_KEYS if key in os.environ]
 	lines = [f'startup environment provides {len(provided)} of {len(STARTUP_KEYS)} startup keys (values hidden)']
 	if os.path.isfile(preferences_path):
 		lines.append(f'preferences from {preferences_path} ({len(prefs.source)} set; reload with SIGHUP)')
 	else:
-		lines.append(f'no {PREFERENCES_FILE} in {env.data_dir}, using the defaults; reload will have nothing to re-read')
+		lines.append(f'no {PREFERENCES_FILE} in {env.data_dir}, using the defaults; reload will have nothing to'
+			' re-read')
 	return lines + retired_notes(prefs.source)
 
 
@@ -406,7 +421,7 @@ def retired_notes(preferences):
 
 
 def startup_warnings():
-	"""Problems in finbot.env that do not stop the start, a line each: a setting that leaves part of the bot unable to work."""
+	"""Settings in finbot.env that do not stop the start but keep part of the bot from working, a line each."""
 	warnings = []
 	if env.slackOutgoingWebhook and not env.slackBotToken:
 		warnings.append('slackOutgoingWebhook is set but slackBotToken is not: Slack commands will get no reply')
diff --git a/lib/dialects.py b/lib/dialects.py
index bc71e89d..c6d2f131 100644
--- a/lib/dialects.py
+++ b/lib/dialects.py
@@ -1,6 +1,6 @@
-"""How each chat network marks text up: bold, italic, struck-through, links, user text made safe, and message sizes. A dialect is
-also the network's name (TELEGRAM == 'telegram'), so it can stand wherever the name did: reports take one as `service` and write
-`service.bold(text)`. dialects.of('slack') gives the one for a name."""
+"""How each chat network marks text up: bold, italic, struck-through, links, user text made safe, and message sizes. A
+dialect is also the network's name (TELEGRAM == 'telegram'), so it can stand wherever the name did: reports take one as
+`service` and write `service.bold(text)`. dialects.of('slack') gives the one for a name."""
 import html
 import re
 
@@ -36,9 +36,9 @@ def link(self, url, text):
 		return self._link(url, text) if config.prefs.hyperlink else text
 
 	def escape(self, text):
-		"""Text a user wrote (a .plan entry, a name) made safe to show: Telegram's HTML and Slack's markup both take &amp; &lt; &gt;.
-		It is unescaped first because it may arrive escaped already (inbound escapes Telegram names as they arrive):
-		escaping twice would show &amp;amp;."""
+		"""Text a user wrote (a .plan entry, a name) made safe to show: Telegram's HTML and Slack's markup both take
+		&amp; &lt; &gt;. It is unescaped first because it may arrive escaped already (inbound escapes Telegram names as
+		they arrive): escaping twice would show &amp;amp;."""
 		return html.escape(html.unescape(str(text)), quote=False)
 
 	def visible(self, text):
@@ -79,7 +79,8 @@ class Discord(Dialect):
 
 	def _link(self, url, text):
 		"""A masked link: <url> stops a preview; brackets in the text and parentheses in the url would end it early."""
-		return '[' + text.replace('[', '\\[').replace(']', '\\]') + '](<' + url.replace('(', '%28').replace(')', '%29') + '>)'
+		return '[' + text.replace('[', '\\[').replace(']', '\\]') + '](<' + url.replace('(', '%28').replace(')',
+			'%29') + '>)'
 
 	def escape(self, text):
 		"""The plain text: Discord's markdown shows entities as written."""
diff --git a/lib/formats.py b/lib/formats.py
index e7fa7864..bbb90c62 100644
--- a/lib/formats.py
+++ b/lib/formats.py
@@ -7,7 +7,8 @@
 	('BlackRock Institutional Trust Company N.A.', ''), ('Listed Funds Trust', ''), ('GraniteShares ETF Trust', ''),
 	('ProShares Trust', ''), ('VanEck ETF Trust', ''), ('Direxion Shares ETF Trust', ''), (' of California', ''),
 	('Magnificent', 'Mag'), ('First Trust NASDAQ Clean Edge Green Energy Index Fund', 'First Trust Clean Energy'),
-	('First Trust Exchange-Traded Fund III First Trust Nasdaq Clean Edge Clean Energy Index Fund', 'First Trust Green Energy'),
+	('First Trust Exchange-Traded Fund III First Trust Nasdaq Clean Edge Clean Energy Index Fund',
+		'First Trust Green Energy'),
 	('Mirae Asset Global Investments (Hong Kong) Limited', ''), ('Space Exploration Technologies', 'SpaceX'),
 	('Exchange Traded Funds Series OFC', ''), ('Global X Global X', 'Global X'), ('Tema ETF Trust', ''), (' AUD', ' '),
 	(' USD', ' '), ('New York Shares', ''), ('China Electric Vehicle and Battery ETF', 'China EV ETF'),
@@ -18,15 +19,18 @@
 ]
 
 TITLE_SHORTENINGS = [ # and these after the suffixes come off
-	('Microbalifesciences', 'Microba Life Sciences'), ('Walt Disney Co (The)', 'Disney'), ('Lisenergylimited', 'LI-S Energy'),
+	('Microbalifesciences', 'Microba Life Sciences'), ('Walt Disney Co (The)', 'Disney'),
+	('Lisenergylimited', 'LI-S Energy'),
 	('Invesco Capital Management LLC ', ''), ('QUALCOMM', 'Qualcomm'), ('Ordinary Shares Class A', ''),
 	('Ordinary Shares Class C', ''), ('Lbt Innovations', 'LBT Innovations'),
 	('Vanguard Information Technology', 'Vanguard Infotech'), ('The ', ''), (' Shares', ''), ('Rea ', 'REA '),
-	('Csl ', 'CSL '), ('Battery ', 'Batt '), ('.com', ''), ('Etf', 'ETF'), ('N.V.', ''), (' NV ', ' '), ('New York Re', ''),
+	('Csl ', 'CSL '), ('Battery ', 'Batt '), ('.com', ''), ('Etf', 'ETF'), ('N.V.', ''), (' NV ', ' '),
+	('New York Re', ''),
 	(' Australian', ' Aus'), (' Australia', ' Aus'), (' Infrastructure', ' Infra'), (' Manufacturing Company', ' '),
 	(' Limited', ' '), (' Ltd', ' '), (' Holdings', ' '), (' Holding', ' '), (' (Holdings)', ''), (' Corporation', ' '),
 	(' Incorporated', ' '), (' incorporated', ' '), (' Technologies', ' '), (' Technology', ' '), (' Enterprises', ' '),
-	(' Enterprise', ' '), (' Enterpr', ' '), (' Ventures', ' '), (' Resources', ' '), (' (Intrada', ''), (' Structured', ''),
+	(' Enterprise', ' '), (' Enterpr', ' '), (' Ventures', ' '), (' Resources', ' '), (' (Intrada', ''),
+	(' Structured', ''),
 	(' Co.', ' '), (' Corp.', ' '), (' Corp', ' '), (' Tech ', ' '), (' Company', ' '), (' Group', ' '), (', Inc', ' '),
 	(' Inc', ' '), (' Plc', ' '), (' plc', ' '), (' Index', ' '), (' ADR', ' '), (' Daily', ' '), (' Long', ' '),
 	(' Bull', ' '), (' .', ' '), (' ,', ' ')
@@ -92,7 +96,7 @@ def date_short(d, year=False):
 	return f"{d.day} {d:%b}" + (f" {d.year}" if year else "")
 
 def date_range_english(days, end):
-	"""'28 Sep - 5 Oct 2026' for the `days` ending on `end`; both ends carry their year when the range crosses a year boundary."""
+	"""'28 Sep - 5 Oct 2026' for the `days` ending on `end`; both ends carry their year when the range crosses one."""
 	start = end - datetime.timedelta(days=days)
 	return f"{date_short(start, start.year != end.year)} - {date_short(end, True)}"
 
diff --git a/lib/jobrun.py b/lib/jobrun.py
index fb3680f3..3478ee1f 100644
--- a/lib/jobrun.py
+++ b/lib/jobrun.py
@@ -1,16 +1,20 @@
 """Runs one scheduled command in its own process: `python -m lib.jobrun lib.reports.<name> '<json kwargs>'`.
 
-The bot starts this for every scheduled or manual run, so a job's CPU-heavy work (charts, pandas) happens off the bot's event
-loop and cannot hold up incoming webhooks. Failures are reported to the channel and give the exit status REPORTED."""
+The bot starts this for every scheduled or manual run, so a job's CPU-heavy work (charts, pandas) happens off the
+bot's event loop and cannot hold up incoming webhooks. Failures are reported to the channel and give the exit status
+REPORTED."""
 import signal
 if __name__ == '__main__':
-	signal.signal(signal.SIGHUP, signal.SIG_IGN) # a reload is for the bot, but `systemctl kill -s HUP` signals every process of the unit
-	signal.signal(signal.SIGUSR1, signal.SIG_IGN) # likewise a flush (systemctl kill -s USR1); its default action would kill the job
+	# a reload is for the bot, but `systemctl kill -s HUP` signals every process of the unit
+	signal.signal(signal.SIGHUP, signal.SIG_IGN)
+	# likewise a flush (systemctl kill -s USR1); its default action would kill the job
+	signal.signal(signal.SIGUSR1, signal.SIG_IGN)
 import importlib, json, re, sys
 
 from lib import delivery
 
-REPORTED = 3 # exit status of a run that failed and said so in the chat; the bot reports any other non-zero status itself
+# exit status of a run that failed and said so in the chat; the bot reports any other non-zero status itself
+REPORTED = 3
 
 
 def main(argv):
diff --git a/lib/jobs.py b/lib/jobs.py
index dfed7b97..9a11b88e 100644
--- a/lib/jobs.py
+++ b/lib/jobs.py
@@ -1,8 +1,9 @@
 """The scheduled reports (formerly cron jobs), run by the bot itself.
 
-The schedule is a file in crontab syntax: five time fields (or an @macro) and then a finbot script with its arguments, e.g.
-`*/20 * * * Mon-Fri trades.py` or `28 21 * * Sat cal.py earnings; price.py 7`. Pasted crontab lines (`./price.py > /dev/null`,
-`{ a; b ;}`) are accepted too. Running in the bot means the jobs share its user, sandbox, environment and cache."""
+The schedule is a file in crontab syntax: five time fields (or an @macro) and then a finbot script with its
+arguments, e.g. `*/20 * * * Mon-Fri trades.py` or `28 21 * * Sat cal.py earnings; price.py 7`. Pasted crontab lines
+(`./price.py > /dev/null`, `{ a; b ;}`) are accepted too. Running in the bot means the jobs share its user, sandbox,
+environment and cache."""
 import datetime
 import hmac
 import json
@@ -27,19 +28,20 @@
 from lib import delivery
 from lib.reports import cal, cleanup, milestone, performance, podcasts, price, rating, reminder, shorts, trades
 
-SCRIPTS = {'cal': cal, 'cleanup': cleanup, 'milestone': milestone, 'performance': performance, 'podcasts': podcasts, 'price': price,
-	'rating': rating, 'reminder': reminder, 'shorts': shorts, 'trades': trades}
+SCRIPTS = {'cal': cal, 'cleanup': cleanup, 'milestone': milestone, 'performance': performance, 'podcasts': podcasts,
+	'price': price, 'rating': rating, 'reminder': reminder, 'shorts': shorts, 'trades': trades}
 current = None # the running Scheduler, set by bot.py; the chat commands use it
 HIDDEN = ('cleanup', 'podcasts') # scheduled, but not offered in the chat list
 STATE_FILE = 'finbot_jobs.json' # last fire time of every job, so a restart neither repeats a run nor loses one
 JOB_TIMEOUT = 5 * 60 # seconds each command may run before it is stopped and reported
 IGNORED_VARIABLES = ('MAILTO', 'SHELL', 'PATH', 'HOME', 'LOGNAME')
-OPERATOR = re.compile(r'\d*>>?&?\d*|&>>?|;|&&|\|\|?|&|[<`$(){}]') # what a shell reads as more than a word, outside quotes
+# what a shell reads as more than a word, outside quotes
+OPERATOR = re.compile(r'\d*>>?&?\d*|&>>?|;|&&|\|\|?|&|[<`$(){}]')
 REDIRECT = re.compile(r'\d*>>?&?\d*|&>>?') # > file, >> file, 2>&1, &> file
 
 
 def checksum(label):
-	"""Short stand-in for a command in a Telegram button, whose data is limited to 64 bytes: detects a changed schedule."""
+	"""A command's short stand-in in a button (Telegram allows 64 bytes of data): it detects a changed schedule."""
 	return format(zlib.crc32(label.encode()) & 0xffff, '04x')
 
 
@@ -47,7 +49,8 @@ def run_child(label, module, kwargs):
 	"""Run a command in its own process (python -m lib.jobrun) and return its exit status. The child inherits the
 	environment, user and output of the bot, so it behaves like the old cron job. One still running after JOB_TIMEOUT
 	is stopped and TimeoutError raised; any other interruption stops it too."""
-	env = dict(os.environ, **config.child_environment()) # the preferences this process runs on, not whatever the file says now
+	# the preferences this process runs on, not whatever the file says now
+	env = dict(os.environ, **config.child_environment())
 	args = [sys.executable, '-m', 'lib.jobrun', module.__name__, json.dumps(kwargs)]
 	proc = subprocess.Popen(args, cwd=config.REPO_DIR, env=env)
 	try:
@@ -95,8 +98,9 @@ class ScheduleError(ValueError):
 
 
 class ImpossibleTime(ScheduleError):
-	"""A time that cannot be (minute 61) or never comes (30 February). Other bad lines are reported and skipped, but this refuses the
-	whole file: finbot does not start with it, and a reload keeps the schedule that is running. `lines` says where, one per line."""
+	"""A time that cannot be (minute 61) or never comes (30 February). Other bad lines are reported and skipped, but
+	this refuses the whole file: finbot does not start with it, and a reload keeps the schedule that is running. `lines`
+	says where, one per line."""
 	def __init__(self, lines):
 		"""lines: where, one message or a list of them."""
 		self.lines = [lines] if isinstance(lines, str) else list(lines)
@@ -161,7 +165,7 @@ def flush():
 
 
 def split_commands(tokens):
-	"""Word lists, one per command: split on unquoted ; and &&, redirects and braces dropped, any other shell syntax refused."""
+	"""Word lists, one per command: split on unquoted ; and &&, redirects and braces dropped, other syntax refused."""
 	commands, current, target = [], [], False
 	for kind, value in tokens:
 		if target: # the file of a redirect
@@ -247,9 +251,10 @@ def parse_schedule(text, default_zone=None):
 
 
 class Scheduler:
-	"""Fires the jobs of a schedule file. `path` is a callable returning the file (the data dir copy wins over the default, so a
-	new copy is found on reload). `clock`, `sleep` and `spawn` can be replaced, which the tests do. Its lock covers
-	what the loop, a reload and runs from chat share: the jobs, their bases, the record of runs and what is running."""
+	"""Fires the jobs of a schedule file. `path` is a callable returning the file (the data dir copy wins over the
+	default, so a new copy is found on reload). `clock`, `sleep` and `spawn` can be replaced, which the tests do. Its
+	lock covers what the loop, a reload and runs from chat share: the jobs, their bases, the record of runs and what is
+	running."""
 
 	def __init__(self, path, catchup=None, clock=None, sleep=None, spawn=None, zone=None, runner=None):
 		"""path() gives the schedule file's path. catchup is how late, in minutes, a missed run may still start (the
@@ -269,16 +274,18 @@ def __init__(self, path, catchup=None, clock=None, sleep=None, spawn=None, zone=
 		self.lock = threading.Lock()
 
 	def load(self):
-		"""Read the schedule file. A line with an impossible time refuses the whole file: ImpossibleTime is raised, naming every such
-		line, and nothing changes. Other bad lines are reported and skipped; a file with no usable line keeps the schedule that is
-		running."""
+		"""Read the schedule file. A line with an impossible time refuses the whole file: ImpossibleTime is raised,
+		naming every such line, and nothing changes. Other bad lines are reported and skipped; a file with no usable
+		line keeps the schedule that is running."""
 		path = self.path()
 		try:
 			jobs, errors = parse_schedule(Path(path).read_text(encoding='utf-8'), self.zone)
 		except OSError as e:
 			delivery.report_error(e, context='schedule')
 			return False
-		if impossible := [f'{path} line {number}: {error}: {line}' for number, line, error in errors if isinstance(error, ImpossibleTime)]:
+		impossible = [f'{path} line {number}: {error}: {line}' for number, line, error in errors
+			if isinstance(error, ImpossibleTime)]
+		if impossible:
 			raise ImpossibleTime(impossible)
 		for number, line, error in errors:
 			delivery.report_error(ScheduleError(f'{path} line {number}: {error}: {line}'), context='schedule')
@@ -301,7 +308,7 @@ def load(self):
 		return True
 
 	def reload(self, *args):
-		"""SIGHUP: read the edited file. One with an impossible time is refused and reported; the running schedule carries on."""
+		"""SIGHUP: read the edited file. One with an impossible time is refused and reported; the schedule stays."""
 		try:
 			self.load()
 		except ImpossibleTime as e:
@@ -309,7 +316,7 @@ def reload(self, *args):
 			delivery.report_error(ScheduleError(problem), context='schedule')
 
 	def start_base(self, job, now):
-		"""Where to count from at start: a run missed within the catch-up window fires once; with no record, nothing is missed."""
+		"""Where to count from at start: a run missed in the catch-up window fires once; with no record, none is."""
 		last = self.last.get(job.key)
 		if not last:
 			return now
@@ -319,13 +326,16 @@ def start_base(self, job, now):
 			return now
 
 	def next_fire(self, job, after):
-		"""The first fire time after `after`, in the job's zone. croniter is given the naive local time and the zone is put back after:
-		croniter up to 6.0 (the distro packages) mishandles zoneinfo across DST changes, firing daily jobs twice or an hour late. A
-		time the clocks repeat fires once, the first time; one they skip (02:30 as they go forward) fires an hour later by the clock."""
+		"""The first fire time after `after`, in the job's zone. croniter is given the naive local time and the zone is
+		put back after: croniter up to 6.0 (the distro packages) mishandles zoneinfo across DST changes, firing daily
+		jobs twice or an hour late. A time the clocks repeat fires once, the first time; one they skip (02:30 as they go
+		forward) fires an hour later by the clock."""
 		times = croniter(job.expr, after.astimezone(job.zone).replace(tzinfo=None))
 		while True:
-			t = times.get_next(datetime.datetime).replace(tzinfo=job.zone).timestamp() # fold 0: the first of a repeated time
-			if t > after.timestamp(): # by the instant: comparing two times of one zone ignores which of a repeated hour they are in
+			# fold 0: the first of a repeated time
+			t = times.get_next(datetime.datetime).replace(tzinfo=job.zone).timestamp()
+			# by the instant: comparing two times of one zone ignores which of a repeated hour they are in
+			if t > after.timestamp():
 				return datetime.datetime.fromtimestamp(t, job.zone)
 
 	def due(self, now):
@@ -384,8 +394,9 @@ def execute(self, commands):
 		return outcome
 
 	def commands(self):
-		"""Every offered scheduled command once, sorted by name with numbers in numeric order: [(label, module, kwargs)]. The time,
-		comments, redirects and the other commands of a line are not part of it; HIDDEN scripts are left out."""
+		"""Every offered scheduled command once, sorted by name with numbers in numeric order: [(label, module,
+		kwargs)]. The time, comments, redirects and the other commands of a line are not part of it; HIDDEN scripts are
+		left out."""
 		unique = {}
 		hidden = {SCRIPTS[name] for name in HIDDEN}
 		for job in self.jobs:
@@ -396,7 +407,7 @@ def commands(self):
 		return [unique[label] for label in sorted(unique, key=natural)]
 
 	def select(self, text):
-		"""The scheduled command for `.jobs run <text>`: a number from the list, or the command itself. Raises ScheduleError."""
+		"""The scheduled command for `.jobs run <text>`: a number in the list, or the command. Raises ScheduleError."""
 		text = text.strip()
 		commands = self.commands()
 		if text.isdigit():
@@ -415,8 +426,8 @@ def select(self, text):
 		raise ScheduleError(f"'{text}' is not in the schedule; send .jobs to list the commands")
 
 	def pick(self, number, crc):
-		"""The offered command a button stands for. Button data is untrusted input: the number must be plain digits within the
-		list and the checksum must match that very command, else ScheduleError."""
+		"""The offered command a button stands for. Button data is untrusted input: the number must be plain digits
+		within the list and the checksum must match that very command, else ScheduleError."""
 		commands = self.commands()
 		if not (number.isascii() and number.isdigit() and 1 <= int(number) <= len(commands)):
 			raise ScheduleError('no such command')
@@ -430,7 +441,7 @@ def run_command(self, command):
 		return command[0], self.execute([command])
 
 	def run_now(self, text):
-		"""`.jobs run`: one scheduled command, now, as cron would run it; the schedule is unchanged. Returns (label, outcome)."""
+		"""`.jobs run`: one scheduled command now, as cron runs it; the schedule stays. Returns (label, outcome)."""
 		return self.run_command(self.select(text))
 
 	def tick(self):
diff --git a/lib/net.py b/lib/net.py
index 1ef6bb02..c5e59fa7 100644
--- a/lib/net.py
+++ b/lib/net.py
@@ -1,8 +1,9 @@
-"""Outbound HTTP. Every request goes through one session, so repeated calls to a host reuse its connection instead of paying a
-new TCP and TLS handshake each time: each typing notice and reply, each Yahoo ticker, each Sharesight report.
+"""Outbound HTTP. Every request goes through one session, so repeated calls to a host reuse its connection instead of
+paying a new TCP and TLS handshake each time: each typing notice and reply, each Yahoo ticker, each Sharesight report.
 
-The session keeps no cookies, so requests stay as stateless as they were without it. Yahoo's cookie travels in an explicit Cookie
-header; and the request that asks fc.yahoo.com for a new cookie must not carry the old one, or Yahoo would not set a new one."""
+The session keeps no cookies, so requests stay as stateless as they were without it. Yahoo's cookie travels in an
+explicit Cookie header; and the request that asks fc.yahoo.com for a new cookie must not carry the old one, or Yahoo
+would not set a new one."""
 import http.cookiejar
 import re
 import requests
@@ -10,7 +11,8 @@
 from lib import config
 
 session = requests.Session()
-session.cookies.set_policy(http.cookiejar.DefaultCookiePolicy(allowed_domains=[])) # no domain may set or receive a cookie
+# no domain may set or receive a cookie
+session.cookies.set_policy(http.cookiejar.DefaultCookiePolicy(allowed_domains=[]))
 CONNECT_TIMEOUT = 5 # seconds to wait for a host to answer the connection: one that is down costs this, not http_timeout
 
 def get(url, **kwargs):
@@ -27,8 +29,8 @@ def timeouts(kwargs):
 	return {'timeout': (CONNECT_TIMEOUT, config.prefs.http_timeout), **kwargs}
 
 def redact(text):
-	"""text (or an exception) without the secrets that request URLs, headers and error messages carry: Telegram bot tokens,
-	Slack and Discord webhook paths, bearer tokens, and token, crumb, key, secret and auth parameters."""
+	"""text (or an exception) without the secrets that request URLs, headers and error messages carry: Telegram bot
+	tokens, Slack and Discord webhook paths, bearer tokens, and token, crumb, key, secret and auth parameters."""
 	text = re.sub(r'bot\d+:[\w-]+', 'bot<redacted>', str(text))
 	text = re.sub(r'(/services/|/api/webhooks/)[^\s\'"?&]+', r'\1<redacted>', text)
 	text = re.sub(r'(?i)(bearer\s+)[\w.~+/=-]+', r'\1<redacted>', text)
diff --git a/lib/render.py b/lib/render.py
index 3649e740..c74e44cc 100644
--- a/lib/render.py
+++ b/lib/render.py
@@ -60,20 +60,21 @@ def style_date_axis(ax, start, end, ink):
 _render_pool = None # set by render_off_hub(): the bot draws its charts in one native thread
 
 def render_off_hub():
-	"""For the bot: draw charts in one native thread from now on, so that the gevent hub keeps serving every other chat while one
-	is drawn (the shorts pictogram takes over a second; in the hub that froze everything). One thread, because several at once
-	deadlock matplotlib under gevent's patched locks. matplotlib is imported here, at start, instead of by the first chart. Job
-	processes and the tests draw in place."""
+	"""For the bot: draw charts in one native thread from now on, so that the gevent hub keeps serving every other chat
+	while one is drawn (the shorts pictogram takes over a second; in the hub that froze everything). One thread, because
+	several at once deadlock matplotlib under gevent's patched locks. matplotlib is imported here, at start, instead of
+	by the first chart. Job processes and the tests draw in place."""
 	global _render_pool
 	import importlib
 	from gevent.threadpool import ThreadPool
-	for module in ('matplotlib.backends.backend_agg', 'matplotlib.dates', 'matplotlib.figure', 'matplotlib.patches', 'matplotlib.path', 'matplotlib.ticker'):
+	for module in ('matplotlib.backends.backend_agg', 'matplotlib.dates', 'matplotlib.figure', 'matplotlib.patches',
+			'matplotlib.path', 'matplotlib.ticker'):
 		importlib.import_module(module)
 	_render_pool = ThreadPool(1)
 
 def _drawn(draw):
-	"""A chart function that draws in the render pool when there is one. Only for a function that calls none of the others: the
-	pool's one thread waiting on itself would never finish."""
+	"""A chart function that draws in the render pool when there is one. Only for a function that calls none of the
+	others: the pool's one thread waiting on itself would never finish."""
 	@functools.wraps(draw)
 	def in_pool(*args, **kwargs):
 		"""Draw in the render thread when the bot has one (render_off_hub), else here."""
@@ -134,7 +135,7 @@ def _series(rows):
 	return pd.Series([float(close) for _, close in rows], index=pd.to_datetime([day for day, _ in rows]))
 
 def _png(fig, facecolor):
-	"""The figure as PNG bytes, rewound, or an upload sends an empty file. No bbox_inches='tight': it would change the 1280x800 size."""
+	"""The figure as PNG bytes, rewound (or an upload sends nothing); not bbox_inches='tight': it alters 1280x800."""
 	buf = io.BytesIO()
 	fig.savefig(buf, format='png', facecolor=facecolor)
 	buf.seek(0)
@@ -226,9 +227,10 @@ def draw_compare(chart):
 		label, y = rebased[i]
 		color = colors[i]
 		ax.plot([1.015], [p], marker='s', markersize=4, color=color, transform=ax.transAxes, clip_on=False, zorder=4)
-		ax.annotate(f"{label}  {value:+,.1f}%", xy=(y.index[-1], value), xycoords='data', xytext=(1.03, p), textcoords='axes fraction',
-			ha='left', va='center', fontsize=7.5, color=ink, annotation_clip=False,
-			arrowprops=dict(arrowstyle='-', color=color, linewidth=0.7, alpha=0.7, shrinkA=0, shrinkB=2, relpos=(0, 0.5)))
+		ax.annotate(f"{label}  {value:+,.1f}%", xy=(y.index[-1], value), xycoords='data', xytext=(1.03, p),
+			textcoords='axes fraction', ha='left', va='center', fontsize=7.5, color=ink, annotation_clip=False,
+			arrowprops=dict(arrowstyle='-', color=color, linewidth=0.7, alpha=0.7, shrinkA=0, shrinkB=2,
+				relpos=(0, 0.5)))
 	_heading(fig, chart.title, chart.subtitle)
 	return _png(fig, bg)
 
@@ -277,22 +279,26 @@ def draw_bars(chart):
 	axes_px = (top_edge - 0.02) * fig.get_figheight() * dpi
 	bar_px = 0.72 * axes_px / n
 	longest = max([len(str(label)) for _, label, _ in real] + [3])
-	em_px = min(bar_px * 1.3, 0.28 * (0.96 - left) * fig.get_figwidth() * dpi / (0.62 * longest)) # text height follows the bar thickness
+	# text height follows the bar thickness
+	em_px = min(bar_px * 1.3, 0.28 * (0.96 - left) * fig.get_figwidth() * dpi / (0.62 * longest))
 	size = max(5, min(30, em_px * 72 / dpi))
 	# each row's label sits beside the bar's base (left of the zero line for rising bars, right of it for falling ones)
 	base_labels = []
 	value_labels = []
-	drop = -0.365 * size # baseline sits half a cap height below the bar's centre, so the capitals are centred on the bar
+	# baseline sits half a cap height below the bar's centre, so the capitals are centred on the bar
+	drop = -0.365 * size
 	for y, label, v in real:
-		base_labels.append(ax.annotate(label, xy=(base, y), xytext=(-4, drop), textcoords='offset points', ha='right', va='baseline',
-			fontsize=size, color=ink, annotation_clip=False, zorder=5))
-		value_labels.append(ax.annotate(value_fmt(v), xy=(v, y), xytext=(4 if v >= 0 else -4, drop), textcoords='offset points',
-			ha='left' if v >= 0 else 'right', va='baseline', fontsize=size, color=ink, annotation_clip=False))
+		base_labels.append(ax.annotate(label, xy=(base, y), xytext=(-4, drop), textcoords='offset points', ha='right',
+			va='baseline', fontsize=size, color=ink, annotation_clip=False, zorder=5))
+		value_labels.append(ax.annotate(value_fmt(v), xy=(v, y), xytext=(4 if v >= 0 else -4, drop),
+			textcoords='offset points', ha='left' if v >= 0 else 'right', va='baseline', fontsize=size, color=ink,
+			annotation_clip=False))
 	renderer = canvas.get_renderer()
 	canvas.draw()
 	label_width = [t.get_window_extent(renderer).width for t in base_labels]
 	value_width = [t.get_window_extent(renderer).width for t in value_labels]
-	if xlim: # zoomed charts put the labels in the left margin: shrink the text until the widest label fits inside the image
+	# zoomed charts put the labels in the left margin: shrink the text until the widest label fits inside the image
+	if xlim:
 		factor = _shrink(base_labels, left * fig.get_figwidth() * dpi - 30) # leaves the 4pt gap before the bar base too
 		if factor < 1:
 			for t in value_labels:
@@ -302,9 +308,11 @@ def draw_bars(chart):
 			value_width = [t.get_window_extent(renderer).width for t in value_labels]
 	width = ax.get_window_extent(renderer).width
 	lo, hi = lo0, hi0
-	for _ in range(12): # the pixels-per-unit scale depends on the limits, which depend on the text beside the bars: settle it iteratively
+	# the pixels-per-unit scale depends on the limits, which depend on the text beside the bars: settle it iteratively
+	for _ in range(12):
 		per_unit = width / (hi - lo)
-		# room for the text outside the bars: each tip's value, and each base's label (zoomed charts keep those in the left margin)
+		# room for the text outside the bars: each tip's value, and each base's label (zoomed charts keep those in the
+		# left margin)
 		new_lo, new_hi = lo0, hi0
 		for (_, _, v), w, vw in zip(real, label_width, value_width):
 			if v >= 0:
@@ -325,8 +333,10 @@ def draw_bars(chart):
 	_heading(fig, chart.title, subtitle)
 	return _png(fig, bg)
 
-SHORTS_VERDICTS = ((5, 'board shorts'), (10, 'cargo shorts'), (15, 'boxers'), (20, 'gym shorts'), # five points each at the bottom
-	(30, 'cutoffs'), (40, 'short shorts'), (50, 'daisy dukes'), (60, 'hot pants'), (70, 'speedos'), (80, 'budgie smugglers'), (90, 'just a belt'),
+# five points each at the bottom
+SHORTS_VERDICTS = ((5, 'board shorts'), (10, 'cargo shorts'), (15, 'boxers'), (20, 'gym shorts'),
+	(30, 'cutoffs'), (40, 'short shorts'), (50, 'daisy dukes'), (60, 'hot pants'), (70, 'speedos'),
+	(80, 'budgie smugglers'), (90, 'just a belt'),
 	(float('inf'), 'never nudes')) # the extreme ones step in every 10 points
 
 def shorts_verdict(percent):
@@ -359,13 +369,14 @@ def P(px, py):
 			"""A point of the pair's box, from fractions of its width and height."""
 			return (x + px * w, y + py * h)
 		belt = '#' + ''.join(f"{int(int(color[i:i + 2], 16) * 0.7):02x}" for i in (1, 3, 5))
-		legs = Polygon([P(0.04, 0.2), P(0.96, 0.2), P(1.0, 0.97), P(0.56, 0.97), P(0.5, 0.58), P(0.44, 0.97), P(0.0, 0.97)],
-			closed=True, facecolor=color, edgecolor=ink, linewidth=lw, joinstyle='round')
+		legs = Polygon([P(0.04, 0.2), P(0.96, 0.2), P(1.0, 0.97), P(0.56, 0.97), P(0.5, 0.58), P(0.44, 0.97),
+			P(0.0, 0.97)], closed=True, facecolor=color, edgecolor=ink, linewidth=lw, joinstyle='round')
 		waist = Rectangle(P(0.04, 0.0), 0.92 * w, 0.22 * h, facecolor=belt, edgecolor=ink, linewidth=lw)
 		parts = [legs, waist]
 		for dx, dy in ((0.2, 0.5), (0.3, 0.78), (0.72, 0.5), (0.8, 0.78), (0.14, 0.8)): # polka dots
 			parts.append(Circle(P(dx, dy), 0.045 * w, facecolor='white', edgecolor='none'))
-		parts.append(Rectangle(P(0.47, 0.0), 0.06 * w, 0.22 * h, facecolor='#ffd23f', edgecolor=ink, linewidth=lw * 0.6)) # belt buckle
+		parts.append(Rectangle(P(0.47, 0.0), 0.06 * w, 0.22 * h, facecolor='#ffd23f', edgecolor=ink,
+			linewidth=lw * 0.6)) # belt buckle
 		clip = Rectangle((x - 2, y - 2), w * frac + 2, h + 4, transform=ax.transData) if frac < 1 else None
 		for part in parts:
 			ax.add_patch(part)
@@ -379,13 +390,15 @@ def P(px, py):
 	tilt = math.radians(-1) # beams turned 1 degree counter-clockwise (angles run clockwise here, as y points down)
 	for k in range(0, rays, 2): # every other wedge, so the rays alternate with the sky
 		a0, a1 = k * 2 * math.pi / rays + tilt, (k + 1) * 2 * math.pi / rays + tilt
-		ax.add_patch(Polygon([(sx, sy), (sx + 2000 * math.cos(a0), sy + 2000 * math.sin(a0)), (sx + 2000 * math.cos(a1), sy + 2000 * math.sin(a1))],
+		ax.add_patch(Polygon([(sx, sy), (sx + 2000 * math.cos(a0), sy + 2000 * math.sin(a0)),
+			(sx + 2000 * math.cos(a1), sy + 2000 * math.sin(a1))],
 			closed=True, facecolor=ray, edgecolor='none', zorder=-1))
 	def at(u, v):
 		"""A point on the sun's face, in sun radii from its centre with y down, in the axes' pixels."""
 		return (sx + u * R, sy + v * R)
 	ax.add_patch(Circle((sx, sy), R, facecolor='#ffb000', edgecolor='none'))
-	for dx, dy, r, c in ((-0.05, -0.06, 0.93, '#ffb90a'), (-0.1, -0.12, 0.78, '#ffc21f'), (-0.16, -0.2, 0.55, '#ffcb35')): # fake gradient
+	for dx, dy, r, c in ((-0.05, -0.06, 0.93, '#ffb90a'), (-0.1, -0.12, 0.78, '#ffc21f'),
+			(-0.16, -0.2, 0.55, '#ffcb35')): # fake gradient
 		ax.add_patch(Circle(at(dx, dy), r * R, facecolor=c, edgecolor='none'))
 	brown = '#5a3d37'
 	for (u, v), w, h in (((-0.46, -0.07), 0.21, 0.25), ((0.31, -0.37), 0.20, 0.24)): # lopsided eyes
@@ -393,20 +406,25 @@ def at(u, v):
 	nose = np.array([(-0.07, -0.35), (0.05, -0.26), (0.15, -0.13), (0.18, 0.0), (0.10, 0.10), (-0.03, 0.16)])
 	nose = nose.mean(axis=0) + 0.62 * (nose - nose.mean(axis=0)) # dainty: a smaller squiggle
 	pts = np.vstack([nose[0], nose, nose[-1]]) # Catmull-Rom through the points, so the squiggle is smooth
-	curve = [0.5 * ((2 * p1) + (-p0 + p2) * t + (2 * p0 - 5 * p1 + 4 * p2 - p3) * t * t + (-p0 + 3 * p1 - 3 * p2 + p3) * t ** 3)
+	curve = [0.5 * ((2 * p1) + (-p0 + p2) * t + (2 * p0 - 5 * p1 + 4 * p2 - p3) * t * t
+		+ (-p0 + 3 * p1 - 3 * p2 + p3) * t ** 3)
 		for p0, p1, p2, p3 in zip(pts, pts[1:], pts[2:], pts[3:]) for t in np.linspace(0, 1, 12)]
-	ax.plot([sx + u * R for u, _ in curve], [sy + v * R for _, v in curve], color='#e07b1a', linewidth=2.6, solid_capstyle='round')
-	mouth_pts = [(-0.47, 0.47), (0.73, 0.0), (0.75, 0.35), (0.45, 0.74), (0.1, 0.74), (-0.15, 0.74), (-0.42, 0.62), (-0.47, 0.47), (-0.47, 0.47)]
+	ax.plot([sx + u * R for u, _ in curve], [sy + v * R for _, v in curve], color='#e07b1a', linewidth=2.6,
+		solid_capstyle='round')
+	mouth_pts = [(-0.47, 0.47), (0.73, 0.0), (0.75, 0.35), (0.45, 0.74), (0.1, 0.74), (-0.15, 0.74), (-0.42, 0.62),
+		(-0.47, 0.47), (-0.47, 0.47)]
 	mc = (0.13, 0.37) # the mouth shrinks towards its middle
 	mouth = Path([at(mc[0] + 0.65 * (u - mc[0]), mc[1] + 0.65 * (v - mc[1])) for u, v in mouth_pts],
-		[Path.MOVETO, Path.LINETO, Path.CURVE4, Path.CURVE4, Path.CURVE4, Path.CURVE4, Path.CURVE4, Path.CURVE4, Path.CLOSEPOLY])
+		[Path.MOVETO, Path.LINETO, Path.CURVE4, Path.CURVE4, Path.CURVE4, Path.CURVE4, Path.CURVE4, Path.CURVE4,
+			Path.CLOSEPOLY])
 	ax.add_patch(PathPatch(mouth, facecolor='#2b2118', edgecolor='none'))
 	ax.add_patch(Rectangle((0, 748), W, H - 748, facecolor=sea, edgecolor='none'))
 	for cx in range(0, W + 36, 36):
 		ax.add_patch(Circle((cx, 748), 19, facecolor=foam, edgecolor='none'))
 
 	ax.text(40, 52, 'Shorts', color=ink, fontsize=21, fontweight='bold', ha='left', va='center')
-	ax.text(40, 104, f"each pair of shorts = {step}% of shares shorted", color=ink, fontsize=9.5, ha='left', va='center')
+	ax.text(40, 104, f"each pair of shorts = {step}% of shares shorted", color=ink, fontsize=9.5, ha='left',
+		va='center')
 
 	top, bottom = 205, 726 # below the sun
 	n = len(rows)
@@ -430,19 +448,23 @@ def at(u, v):
 		whole, frac = divmod(value / step, 1)
 		color = colors[i % len(colors)]
 		for k in range(int(whole) + (1 if frac > 0.02 else 0)):
-			shorts(x0 + k * icon_w, cy - icon_h / 2, icon_w * 0.92, icon_h, color, frac if k == int(whole) and frac > 0.02 else 1.0, lw)
+			shorts(x0 + k * icon_w, cy - icon_h / 2, icon_w * 0.92, icon_h, color,
+				frac if k == int(whole) and frac > 0.02 else 1.0, lw)
 		end = x0 + (value / step) * icon_w
-		pct_texts.append((ax.text(end + 14, cy, f"{round(value)}%", color=ink, fontsize=size, fontweight='bold', ha='left', va='center'), cy, value))
+		pct_texts.append((ax.text(end + 14, cy, f"{round(value)}%", color=ink, fontsize=size, fontweight='bold',
+			ha='left', va='center'), cy, value))
 	canvas.draw() # measure the percentages, then put each verdict right after its own
 	for text, cy, value in pct_texts:
 		x = text.get_window_extent(canvas.get_renderer()).x1 + 10
 		verdict = ax.text(x, cy, shorts_verdict(round(value)), color=ink, fontsize=size * 0.62, fontstyle='italic',
 			ha='left', va='center')
 		_shrink([verdict], W - 16 - x) # never run off the image
-	ax.text(W / 2, 784, 'past fashion not indicative of future fashion', color='white', fontsize=8, fontstyle='italic', ha='center', va='center')
+	ax.text(W / 2, 784, 'past fashion not indicative of future fashion', color='white', fontsize=8, fontstyle='italic',
+		ha='center', va='center')
 	line_x = x0 + threshold / step * icon_w
 	ax.plot([line_x, line_x], [top - 12, bottom], color=ink, linewidth=1.8, linestyle=(0, (5, 4)), zorder=0)
-	ax.text(line_x, top - 18, f"{threshold:g}% outrageous line", color=ink, fontsize=7.5, fontweight='bold', ha='center', va='bottom')
+	ax.text(line_x, top - 18, f"{threshold:g}% outrageous line", color=ink, fontsize=7.5, fontweight='bold',
+		ha='center', va='bottom')
 
 	return _png(fig, sky)
 
@@ -473,8 +495,10 @@ def draw_pie(chart):
 	step = min(0.075, 0.74 / max(len(rows), 1))
 	y = 0.5 + step * (len(rows) - 1) / 2 - 0.03 # legend block centred on the pie
 	for (label, value), color in zip(rows, colors):
-		fig.add_artist(Rectangle((0.57, y - 0.014), 0.018, 0.028, transform=fig.transFigure, facecolor=color, edgecolor='none'))
-		fig.text(0.60, y, label if len(label) <= 28 else label[:27] + '…', color=ink, fontsize=8.5, ha='left', va='center')
+		fig.add_artist(Rectangle((0.57, y - 0.014), 0.018, 0.028, transform=fig.transFigure, facecolor=color,
+			edgecolor='none'))
+		fig.text(0.60, y, label if len(label) <= 28 else label[:27] + '…', color=ink, fontsize=8.5, ha='left',
+			va='center')
 		fig.text(0.97, y, f"{value / total * 100:.1f}%", color=ink2, fontsize=8.5, ha='right', va='center')
 		y -= step
 	_heading(fig, chart.title, chart.subtitle)
diff --git a/lib/reports/cal.py b/lib/reports/cal.py
index c6ee23c3..35140379 100644
--- a/lib/reports/cal.py
+++ b/lib/reports/cal.py
@@ -10,10 +10,11 @@
 from lib import yahoo
 
 def primary_listings():
-	"""{ADR or second listing: its primary listing}, from finbot_adr.json: read when used, so an edit is seen without a restart."""
+	"""{ADR or second listing: its primary listing}, from finbot_adr.json, read when used: an edit needs no restart."""
 	return config.data_json('finbot_adr.json', {})
 
-EVENTS = {'earnings': ('earnings_date', 'Upcoming earnings'), 'ex-dividend': ('ex_dividend_date', 'Upcoming ex-dividends')} # -> (field, heading)
+EVENTS = {'earnings': ('earnings_date', 'Upcoming earnings'),
+	'ex-dividend': ('ex_dividend_date', 'Upcoming ex-dividends')} # -> (field, heading)
 
 def run(dest=None, *, days=None, specific_stock=None, event='earnings'):
 	"""Post the earnings or ex-dividend dates ahead (`event`) for the tracked stocks or one (Yahoo's symbol): to the
diff --git a/lib/reports/cleanup.py b/lib/reports/cleanup.py
index f38bc7f9..3ffb74cb 100644
--- a/lib/reports/cleanup.py
+++ b/lib/reports/cleanup.py
@@ -3,8 +3,8 @@
 from lib import config
 
 def run(*, days=30):
-	"""Delete the cache files (finbot_*) that were last written more than `days` days ago (days=0: all of them, a flush), but not
-	a temp file written in the last minute: that is an atomic write in flight (cache.write_file)."""
+	"""Delete the cache files (finbot_*) that were last written more than `days` days ago (days=0: all of them, a
+	flush), but not a temp file written in the last minute: that is an atomic write in flight (cache.write_file)."""
 	now = time.time()
 	cutoff = now - days * 86400
 	removed = 0
diff --git a/lib/reports/holdings.py b/lib/reports/holdings.py
index dc62f13b..d89c6e6a 100644
--- a/lib/reports/holdings.py
+++ b/lib/reports/holdings.py
@@ -9,7 +9,8 @@
 
 def holding_line(holding, service, text=None):
 	"""'🇦🇺 Title (CODE)' for a Sharesight holding, CODE (or `text`) linking to the holding on Sharesight."""
-	link = service.link(f"https://portfolio.sharesight.com/holdings/{holding['holding_id']}/dashboard", holding['code'] if text is None else text)
+	link = service.link(f"https://portfolio.sharesight.com/holdings/{holding['holding_id']}/dashboard",
+		holding['code'] if text is None else text)
 	return f"{markets.flag_from_market(holding['market_code'])} {formats.transform_title(holding['name'])} ({link})"
 
 def prepare_holdings_payload(portfolio_name, service):
diff --git a/lib/reports/performance.py b/lib/reports/performance.py
index 99316878..f45f1c2b 100644
--- a/lib/reports/performance.py
+++ b/lib/reports/performance.py
@@ -18,7 +18,7 @@ def gather(past_days, portfolio_select=None):
 	return [(portfolio, portfolio.gain(past_days)) for portfolio in sharesight.portfolios(portfolio_select)]
 
 def benchmark(ticker, market, text, past_days, service):
-	"""(report line, percent) of a benchmark over past_days; percent is None, and the line says so, when it has no history."""
+	"""(report line, percent) of a benchmark over past_days; percent None, and the line saying so, without history."""
 	try:
 		percent = yahoo.price_history(ticker, days=past_days)[0][past_days]
 	except Exception as e: # the rest of the report still goes out
@@ -28,8 +28,8 @@ def benchmark(ticker, market, text, past_days, service):
 	return f"{formats.get_emoji(percent)} {link} {formats.signed_percent(percent, 2)}", percent
 
 def report(service, past_days, portfolio_select=None, performance=None):
-	"""(payload, chart) of the performance report for one service: each portfolio and the benchmarks, best first. performance:
-	gather()'s, fetched when not given."""
+	"""(payload, chart) of the performance report for one service: each portfolio and the benchmarks, best first.
+	performance: gather()'s, fetched when not given."""
 	if performance is None:
 		performance = gather(past_days, portfolio_select)
 	rows = [] # (percent or None, line, chart label), one list for the caption and the chart, best first
diff --git a/lib/reports/podcasts.py b/lib/reports/podcasts.py
index 3be7da95..db9b401f 100644
--- a/lib/reports/podcasts.py
+++ b/lib/reports/podcasts.py
@@ -36,7 +36,7 @@ def format_duration(duration_str):
 	return f"{total_seconds // 3600:02d}:{total_seconds % 3600 // 60:02d}"
 
 def published(pub_raw):
-	"""An episode's publication time (in UTC where the feed gives no zone), or None where it gives none or garbles it."""
+	"""An episode's publication time (UTC where the feed gives no zone), or None where it gives none or garbles it."""
 	try:
 		when = parsedate_to_datetime(pub_raw)
 	except (TypeError, ValueError):
@@ -57,8 +57,9 @@ def how_long_ago(when):
 	return f"{delta.seconds}s ago"
 
 def media_link(item):
-	"""An episode's media: <media:content><media:player url>, else <media:content url>, else <enclosure url>; None without any."""
-	for element in (item.find(".//media:content/media:player", NAMESPACES), item.find("media:content", NAMESPACES), item.find("enclosure")):
+	"""An episode's media: <media:content><media:player url>, else <media:content url>, <enclosure url>, or None."""
+	for element in (item.find(".//media:content/media:player", NAMESPACES), item.find("media:content", NAMESPACES),
+			item.find("enclosure")):
 		if element is not None and element.attrib.get("url"):
 			return element.attrib["url"]
 	return None
@@ -76,8 +77,10 @@ def fetch_new_episodes(feed_url, seen):
 		if not guid or guid in seen or not media_url:
 			continue
 		when = published(item.findtext("pubDate"))
-		episodes.append({"podcast": podcast_name, "title": (item.findtext("title") or "Untitled").strip(), "url": media_url,
-			"ago": how_long_ago(when), "duration": format_duration(item.findtext("itunes:duration", namespaces=NAMESPACES)),
+		episodes.append({"podcast": podcast_name, "title": (item.findtext("title") or "Untitled").strip(),
+			"url": media_url,
+			"ago": how_long_ago(when),
+			"duration": format_duration(item.findtext("itunes:duration", namespaces=NAMESPACES)),
 			"published": when or datetime.min.replace(tzinfo=timezone.utc)})
 		guids.add(guid)
 	episodes.sort(key=lambda episode: episode["published"], reverse=True)
@@ -113,6 +116,7 @@ def job(argv):
 	"""A scheduled run's arguments: (label, run's keyword arguments); a bad one raises ValueError with the usage."""
 	match argv:
 		case [feed_url]:
-			return 'podcasts.py ' + hashlib.md5(feed_url.encode()).hexdigest()[:8], {'feed_url': feed_url} # the label matches the seen file's name
+			# the label matches the seen file's name
+			return 'podcasts.py ' + hashlib.md5(feed_url.encode()).hexdigest()[:8], {'feed_url': feed_url}
 		case _:
 			raise ValueError('"<podcast_feed_url>"')
diff --git a/lib/reports/price.py b/lib/reports/price.py
index 8cbbee8f..70f03824 100644
--- a/lib/reports/price.py
+++ b/lib/reports/price.py
@@ -27,7 +27,8 @@ def gather(specific_stock=None, days=None):
 		yahoo.add_usd_market_caps(market_data)
 	if not days:
 		return tickers, market_data
-	# Sharesight reports a holding's performance only for the time it was held: held for 1Y and asked for 5Y, it gives 1Y
+	# Sharesight reports a holding's performance only for the time it was held: held for 1Y and asked for 5Y, it gives
+	# 1Y
 	if not specific_stock or config.prefs.performance_use_sharesight:
 		for portfolio in sharesight.portfolios():
 			for ticker, gain in portfolio.gains(days).items():
@@ -52,14 +53,14 @@ def gather(specific_stock=None, days=None):
 	return tickers, market_data
 
 def needs_beta(quote):
-	"""Whether trigger_bar must know this ticker's beta: a small cap between US$150M and US$1B, which counts as volatile when its
-	beta is over 1.5. The beta is only in Yahoo's details."""
+	"""Whether trigger_bar must know this ticker's beta: a small cap between US$150M and US$1B, which counts as volatile
+	when its beta is over 1.5. The beta is only in Yahoo's details."""
 	return config.prefs.demote_volatile and 1.5e8 <= quote.get('market_cap_usd', float('inf')) < 1e9
 
 def trigger_bar(ticker, quote, threshold, details):
-	"""The move this ticker needs before it alone sends a scheduled report: more for leveraged ETFs (demote_leveraged) and volatile
-	small caps (demote_volatile; their market caps in US dollars), else the threshold itself. details: Yahoo's details of the
-	tickers that need_beta, fetched together."""
+	"""The move this ticker needs before it alone sends a scheduled report: more for leveraged ETFs (demote_leveraged)
+	and volatile small caps (demote_volatile; their market caps in US dollars), else the threshold itself. details:
+	Yahoo's details of the tickers that need_beta, fetched together."""
 	title = quote['profile_title'].lower()
 	if config.prefs.demote_leveraged and ('2x' in title or '3x' in title):
 		return threshold * 1.3
@@ -197,7 +198,8 @@ def markup(chat):
 		if not chat.takes_buttons:
 			return None
 		if mode in ('interday', 'period'):
-			return buttons.keyboard('l', [f"{threshold:g}", str(top or 0)], buttons.period_for_days(days, 'd' if not days else None))
+			return buttons.keyboard('l', [f"{threshold:g}", str(top or 0)],
+				buttons.period_for_days(days, 'd' if not days else None))
 		return None
 	for chat in delivery.targets(dest):
 		payload, graph = report(chat.dialect, found, mode, threshold, days, top, user, specific_stock, interactive)
diff --git a/lib/reports/profile.py b/lib/reports/profile.py
index a7cda1a5..e79c2f15 100644
--- a/lib/reports/profile.py
+++ b/lib/reports/profile.py
@@ -134,7 +134,8 @@ def trends(data):
 def outlook(service, data):
 	"""The analysts' forecasts and score, insiders' buying or selling, and the short interest."""
 	lines = []
-	for key, label in (('revenueEstimateY', 'Revenue forecast (1Y):'), ('earningsEstimateY', 'Earnings forecast (1Y):')):
+	for key, label in (('revenueEstimateY', 'Revenue forecast (1Y):'),
+			('earningsEstimateY', 'Earnings forecast (1Y):')):
 		if key in data:
 			growth = int(round(data[key]))
 			lines.append(service.bold(label) + f" {formats.signed_percent(growth)} " + ('' if growth > 0 else '⚠️ '))
@@ -164,7 +165,9 @@ def valuation(service, data, state):
 		if state != 'REGULAR' and 'prePostMarketPrice' in data:
 			price_line += f" ({data['prePostMarketPrice']:,.2f} after hrs)"
 		lines.append(price_line)
-	trailing_pe = str(int(round(data['price_to_earnings_trailing']))) if 'price_to_earnings_trailing' in data else 'N/A ⚠️ '
+	trailing_pe = 'N/A ⚠️ '
+	if 'price_to_earnings_trailing' in data:
+		trailing_pe = str(int(round(data['price_to_earnings_trailing'])))
 	if data.get('net_income', 0) > 0:
 		lines.append(service.bold("Trailing P/E:") + f" {trailing_pe}")
 	if data.get('netIncomeToCommon', 0) > 0 and data.get('sharesOutstanding') and 'regularMarketPrice' in data:
@@ -173,7 +176,8 @@ def valuation(service, data, state):
 	if 'price_to_earnings_forward' in data:
 		forward_pe = int(round(data['price_to_earnings_forward']))
 		warn = forward_pe < 0
-		if 'profile_industry' in data: # software is dear: it warns above 100, the rest above 30 or above their trailing P/E
+		# software is dear: it warns above 100, the rest above 30 or above their trailing P/E
+		if 'profile_industry' in data:
 			warn = warn or forward_pe > (100 if 'Software' in data['profile_industry'] else 30)
 			warn = warn or ('price_to_earnings_trailing' in data and forward_pe > int(trailing_pe))
 		lines.append(service.bold("Forward P/E:") + f" {forward_pe} " + ('⚠️ ' if warn else ''))
diff --git a/lib/reports/rankings.py b/lib/reports/rankings.py
index a37a6e42..908ca8c9 100644
--- a/lib/reports/rankings.py
+++ b/lib/reports/rankings.py
@@ -39,9 +39,10 @@ def marketcap_list(service, bottom=False, length=15):
 	return [service.bold(heading)] + [line for _, line in caps[:length]]
 
 def prepare_rating_payload(service, action, length=15):
-	"""(payload, chart) of the `length` tracked stocks rated `action` ('buy' or 'sell') by min_analysts or more analysts (fewer,
-	and the consensus says little: ASX small caps covered by two to six brokers are nearly all a perfect 1.0 buy), the strongest
-	rating first (1 is a strong buy, 5 a strong sell); on equal ratings, the one more analysts give first."""
+	"""(payload, chart) of the `length` tracked stocks rated `action` ('buy' or 'sell') by min_analysts or more analysts
+	(fewer, and the consensus says little: ASX small caps covered by two to six brokers are nearly all a perfect 1.0
+	buy), the strongest rating first (1 is a strong buy, 5 a strong sell); on equal ratings, the one more analysts give
+	first."""
 	found = [] # (rating, analysts, line, ticker)
 	for ticker, data in yahoo.fetch_details(watchlist.with_holdings()).items():
 		if 'recommend' not in data or data['recommend_analysts'] < config.prefs.min_analysts:
@@ -118,7 +119,7 @@ def ratio_list(service, kind, mode='top', length=15):
 	return [service.bold(heading)] + [line for _, _, line in rows], chart
 
 def prepare_beta_payload(service):
-	"""(payload, chart) of the tracked stocks with a beta over 1.5 and a market cap under US$2B, the highest beta first."""
+	"""(payload, chart) of the stocks followed with a beta over 1.5 and a market cap under US$2B, highest beta first."""
 	found = [] # (beta, line, chart row)
 	for ticker, data in yahoo.add_usd_market_caps(yahoo.fetch_details(watchlist.with_holdings())).items():
 		if 'beta' not in data:
diff --git a/lib/reports/rating.py b/lib/reports/rating.py
index 53514467..2670d64e 100644
--- a/lib/reports/rating.py
+++ b/lib/reports/rating.py
@@ -1,4 +1,4 @@
-"""Analysts' consensus rating changes since the last scheduled run (scheduled: rating.py). .buy and .sell are in rankings."""
+"""Analysts' consensus rating changes since the last scheduled run (rating.py); .buy and .sell are in rankings."""
 from lib import delivery
 from lib import state
 from lib import links
diff --git a/lib/sharesight.py b/lib/sharesight.py
index 13dc261a..9f220767 100644
--- a/lib/sharesight.py
+++ b/lib/sharesight.py
@@ -138,7 +138,8 @@ def holdings(self):
 
 	def tickers(self):
 		"""What it holds, as Yahoo tickers, sorted."""
-		return sorted({markets.transform_to_yahoo(each['code'], each['market_code']) for each in self.holdings().values()})
+		return sorted({markets.transform_to_yahoo(each['code'],
+			each['market_code']) for each in self.holdings().values()})
 
 	def trades(self, days=None):
 		"""Its trades of the past `days` (the past_days preference when None), each with the portfolio's name added."""
diff --git a/lib/simplywallst.py b/lib/simplywallst.py
index 206ea1ed..c2dc7d83 100644
--- a/lib/simplywallst.py
+++ b/lib/simplywallst.py
@@ -1,6 +1,7 @@
 from lib import config
 
-FILES = ['finbot_sws_custom.json', 'finbot_sws_asx200.json', 'finbot_sws_sp500.json', 'finbot_sws_nasdaq100.json'] # the first to list a ticker wins
+# the first to list a ticker wins
+FILES = ['finbot_sws_custom.json', 'finbot_sws_asx200.json', 'finbot_sws_sp500.json', 'finbot_sws_nasdaq100.json']
 
 def get_url(symbol, name, exchange):
 	"""The ticker's page on simplywall.st, from the lists in FILES; without one, a search that lands on it."""
@@ -8,5 +9,6 @@ def get_url(symbol, name, exchange):
 		url = config.data_json(file, {}).get(symbol)
 		if url:
 			return url
-	return 'https://www.google.com/search?q=site:simplywall.st+(' + name + '+' + exchange + ':' + symbol.split('.')[0] + ')+Stock&btnI'
+	query = f"site:simplywall.st+({name}+{exchange}:{symbol.split('.')[0]})+Stock"
+	return 'https://www.google.com/search?q=' + query + '&btnI'
 
diff --git a/lib/state.py b/lib/state.py
index f299349e..d10f32b9 100644
--- a/lib/state.py
+++ b/lib/state.py
@@ -1,5 +1,6 @@
-"""State files in DATA_DIR: what must last from one run to the next (the watchlist, plans, seen trades, ratings, record highs,
-the schedule's last runs, chart button refs, keyboard versions, seen podcast episodes). Written whole or not at all, and flushed."""
+"""State files in DATA_DIR: what must last from one run to the next (the watchlist, plans, seen trades, ratings, record
+highs, the schedule's last runs, chart button refs, keyboard versions, seen podcast episodes). Written whole or not at
+all, and flushed."""
 import contextlib
 import copy
 import json
@@ -37,10 +38,10 @@ def _lock(name):
 
 @contextlib.contextmanager
 def edit(name, default):
-	"""Change a state file in place: `with state.edit(name, {}) as data: data[key] = value`. The file is read as the block starts
-	(a copy of `default` when there is none yet) and written as the block ends without an error, if the block changed it.
-	Meanwhile no other edit of the same file runs in this process, so two chats changing it at once both count. Keep slow work,
-	a network call say, outside the block."""
+	"""Change a state file in place: `with state.edit(name, {}) as data: data[key] = value`. The file is read as the
+	block starts (a copy of `default` when there is none yet) and written as the block ends without an error, if the
+	block changed it. Meanwhile no other edit of the same file runs in this process, so two chats changing it at once
+	both count. Keep slow work, a network call say, outside the block."""
 	with _lock(name):
 		data = load(name, copy.deepcopy(default))
 		before = copy.deepcopy(data)
diff --git a/lib/telegram.py b/lib/telegram.py
index ecc16e1d..6920f751 100644
--- a/lib/telegram.py
+++ b/lib/telegram.py
@@ -4,7 +4,7 @@
 from lib import net
 
 def call(method, **kwargs):
-	"""POST a Bot API method and return its result. Raises RuntimeError with Telegram's description when it answers not ok."""
+	"""POST a Bot API method and return its result. Raises RuntimeError with Telegram's reason when it is not ok."""
 	response = net.post(config.env.webhooks['telegram'] + method, **kwargs)
 	try:
 		output = response.json()
@@ -15,7 +15,7 @@ def call(method, **kwargs):
 	return output['result']
 
 def try_call(method, **kwargs):
-	"""call(), for a call whose failure must not stop what the bot is doing: a failure is only logged (and None returned)."""
+	"""call(), where a failure must not stop what the bot is doing: it is only logged, and None returned."""
 	try:
 		return call(method, **kwargs)
 	except Exception as e:
@@ -24,7 +24,8 @@ def try_call(method, **kwargs):
 def setWebhook():
 	"""Have Telegram post messages and button presses to telegramOutgoingWebhook, signed with telegramOutgoingToken."""
 	print("registering", config.env.telegramOutgoingWebhook, file=sys.stderr)
-	params = {'url': config.env.telegramOutgoingWebhook, "allowed_updates": json.dumps(['message', 'callback_query']), 'secret_token': config.env.telegramOutgoingToken}
+	params = {'url': config.env.telegramOutgoingWebhook, "allowed_updates": json.dumps(['message', 'callback_query']),
+		'secret_token': config.env.telegramOutgoingToken}
 	return call('setWebhook', params=params)
 
 def delWebhook():
@@ -37,18 +38,18 @@ def getMe():
 	return call('getMe')
 
 def register():
-	"""Called once by bot.py as it starts, never on import (the job processes and the tests import this module too): unregister and
-	register the webhook again, which wakes up a stale connection on Telegram's side, then return the bot's @username. Raises
-	RuntimeError, or a requests error, if any of the three calls fails."""
+	"""Called once by bot.py as it starts, never on import (the job processes and the tests import this module too):
+	unregister and register the webhook again, which wakes up a stale connection on Telegram's side, then return the
+	bot's @username. Raises RuntimeError, or a requests error, if any of the three calls fails."""
 	delWebhook()
 	setWebhook()
 	return '@' + getMe()['username']
 
 def pinChatMessage(chat_id, message_id):
-	"""Pin a message, without the "pinned a message" notification. A failure is only logged: the message is there either way."""
+	"""Pin a message, without the "pinned a message" notice. A failure is only logged: the message is there anyway."""
 	try_call('pinChatMessage', json={'chat_id': chat_id, 'message_id': message_id, 'disable_notification': True})
 
 def answerCallbackQuery(callback_query_id, text=None):
-	"""Acknowledge a button press (stops the client's loading spinner), with `text` as a toast. A failure is only logged."""
+	"""Acknowledge a button press (the client's spinner stops), `text` as a toast. A failure is only logged."""
 	try_call('answerCallbackQuery', json={'callback_query_id': callback_query_id} | ({'text': text} if text else {}))
 
diff --git a/lib/watchlist.py b/lib/watchlist.py
index c693cf18..27a47af0 100644
--- a/lib/watchlist.py
+++ b/lib/watchlist.py
@@ -1,5 +1,5 @@
-"""The watchlist: tickers followed beside the Sharesight holdings, in DATA_DIR/finbot_watchlist.json. Every call reads the file,
-so the bot and its scheduled reports (other processes) always see the same list."""
+"""The watchlist: tickers followed beside the Sharesight holdings, in DATA_DIR/finbot_watchlist.json. Every call reads
+the file, so the bot and its scheduled reports (other processes) always see the same list."""
 from lib import config
 from lib import markets
 from lib import sharesight
diff --git a/lib/worker.py b/lib/worker.py
index 11b22641..4ec49f25 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -192,7 +192,7 @@ def taken(word):
 	return args
 
 def parse_percent(arg):
-	"""'5', '5%' or '2.5' -> the number. Raises ValueError for anything else, nan and inf included: those are tickers."""
+	"""'5', '5%' or '2.5' -> the number. Raises ValueError for anything else, nan and inf too: those are tickers."""
 	value = float(arg.removesuffix('%'))
 	if not math.isfinite(value):
 		raise ValueError(f"{arg} is not a percentage")
diff --git a/lib/yahoo.py b/lib/yahoo.py
index f1efdd25..54f213ae 100644
--- a/lib/yahoo.py
+++ b/lib/yahoo.py
@@ -117,7 +117,9 @@ def _convert(convert, value):
 	except (TypeError, ValueError, OverflowError):
 		return None
 
-QUOTE_FIELDS = ( # (key, Yahoo's name, conversion) for the fields of a v7 quote; one Yahoo leaves out, sends as null or garbles is left out
+# (key, Yahoo's name, conversion) for the fields of a v7 quote; one Yahoo leaves out, sends as null or garbles is left
+# out
+QUOTE_FIELDS = (
 	('percent_change', 'regularMarketChangePercent', _round2),
 	('dividend', 'trailingAnnualDividendRate', lambda v: round(float(v), 1)),
 	('percent_change_premarket', 'preMarketChangePercent', _round2),
@@ -135,8 +137,8 @@ def _convert(convert, value):
 )
 
 def fetch(tickers):
-	"""v7 quotes for many tickers in one request: {ticker: {...}}. Cached for 5 minutes (DO NOT CACHE LONGER). A ticker Yahoo gives
-	no name or price for is left out. Raises RuntimeError when Yahoo cannot be reached."""
+	"""v7 quotes for many tickers in one request: {ticker: {...}}. Cached for 5 minutes (DO NOT CACHE LONGER). A ticker
+	Yahoo gives no name or price for is left out. Raises RuntimeError when Yahoo cannot be reached."""
 	if not tickers:
 		print("no tickers provided to yahoo.fetch()", file=sys.stderr)
 		return {}
@@ -149,15 +151,18 @@ def fetch(tickers):
 	now = datetime.datetime.now().timestamp()
 	yahoo_output = {}
 	for item in data['quoteResponse']['result']:
-		title = next((name for name in (item.get('longName'), item.get('shortName')) if isinstance(name, str) and name), None)
+		title = next((name for name in (item.get('longName'), item.get('shortName')) if isinstance(name, str) and name),
+			None)
 		if not title or item.get('regularMarketPrice') is None:
 			continue
-		quote = {'profile_title': formats.transform_title(title), 'ticker': item['symbol'], 'regularMarketPrice': item['regularMarketPrice'],
+		quote = {'profile_title': formats.transform_title(title), 'ticker': item['symbol'],
+			'regularMarketPrice': item['regularMarketPrice'],
 			'currency': item.get('currency'), 'percent_change': 0.0, 'dividend': 0.0}
 		for key, field, convert in QUOTE_FIELDS:
 			if (value := _convert(convert, item.get(field))) is not None:
 				quote[key] = value
-		stamps = [item[name] for name in ('earningsTimestamp', 'earningsTimestampStart', 'earningsTimestampEnd') if isinstance(item.get(name), (int, float))]
+		stamps = [item[name] for name in ('earningsTimestamp', 'earningsTimestampStart',
+			'earningsTimestampEnd') if isinstance(item.get(name), (int, float))]
 		if stamps: # the first still to come, else the announced date
 			quote['earnings_date'] = next((stamp for stamp in stamps if stamp > now), stamps[0])
 		yahoo_output[item['symbol']] = quote
@@ -170,7 +175,8 @@ def _quote_answer(tickers):
 	quotes = dig(data, 'quoteResponse') or {}
 	return data if quotes.get('result') is not None and quotes.get('error') is None else None
 
-DETAIL_FIELDS = ( # (key, path in the quoteSummary result[, conversion]); a field Yahoo leaves out, sends as null or garbles is left out
+# (key, path in the quoteSummary result[, conversion]); a field Yahoo leaves out, sends as null or garbles is left out
+DETAIL_FIELDS = (
 	('profile_bio', ('summaryProfile', 'longBusinessSummary')),
 	('profile_city', ('summaryProfile', 'city')),
 	('profile_country', ('summaryProfile', 'country')),
@@ -219,7 +225,8 @@ def fetch_detail(ticker, seconds=None):
 	if result is None: # Yahoo answers 404 Not Found, or result=null, for a symbol it does not know
 		print(ticker, "is unknown to Yahoo", file=sys.stderr)
 		return {}
-	title = next((name for name in (dig(result, 'price', 'longName'), dig(result, 'price', 'shortName')) if isinstance(name, str) and name), None)
+	title = next((name for name in (dig(result, 'price', 'longName'),
+		dig(result, 'price', 'shortName')) if isinstance(name, str) and name), None)
 	if not title: # catches some delisted stocks like "DUB"
 		print(ticker, "has no name at Yahoo", file=sys.stderr)
 		return {}
@@ -245,8 +252,10 @@ def fetch_detail(ticker, seconds=None):
 			detail['revenueEstimateY'], detail['earningsEstimateY'] = _percent(estimates[0]), _percent(estimates[1])
 	# a quarter without an earnings figure is bad data: None for both, which trend_marks marks
 	if dig(quarterly, 0, 'revenue', 'raw') is not None:
-		detail['earningsQ'] = [None if dig(item, 'earnings', 'fmt') is None else dig(item, 'earnings', 'raw') for item in quarterly]
-		detail['revenueQ'] = [None if dig(item, 'earnings', 'fmt') is None else dig(item, 'revenue', 'raw') for item in quarterly]
+		detail['earningsQ'] = [None if dig(item, 'earnings', 'fmt') is None else dig(item, 'earnings',
+			'raw') for item in quarterly]
+		detail['revenueQ'] = [None if dig(item, 'earnings', 'fmt') is None else dig(item, 'revenue',
+			'raw') for item in quarterly]
 		yearly = _list(dig(result, 'earnings', 'financialsChart', 'yearly'))
 		detail['earningsY'] = [dig(item, 'earnings', 'raw') for item in yearly]
 		detail['revenueY'] = [dig(item, 'revenue', 'raw') for item in yearly]
@@ -369,7 +378,7 @@ def price_series(ticker):
 	return name, meta.get('fullExchangeName') or meta.get('exchangeName') or '', series
 
 def fetch_each(get, tickers, workers=8):
-	"""get(ticker) for every ticker, `workers` at a time so their waits overlap: [its result, or the exception it raised], in order."""
+	"""get(ticker) for each ticker, `workers` at a time so the waits overlap: [its result or the exception raised]."""
 	def one(ticker):
 		"""get(ticker), or the exception it raised."""
 		try:
@@ -379,19 +388,21 @@ def one(ticker):
 	with concurrent.futures.ThreadPoolExecutor(max_workers=workers) as executor:
 		return list(executor.map(one, tickers))
 
-SUBUNITS = {'GBp': 'GBP', 'GBX': 'GBP', 'ILA': 'ILS', 'ZAc': 'ZAR'} # Yahoo prices these markets in a subunit but gives market caps in the unit
+# Yahoo prices these markets in a subunit but gives market caps in the unit
+SUBUNITS = {'GBp': 'GBP', 'GBX': 'GBP', 'ILA': 'ILS', 'ZAc': 'ZAR'}
 
 def market_cap_currency(data):
-	"""The currency of a ticker's market cap: its quote currency, but the unit where Yahoo quotes in pence or the like."""
+	"""The currency of a ticker's market cap: its quote currency, or the unit where Yahoo quotes in pence and so on."""
 	return SUBUNITS.get(data.get('currency'), data.get('currency'))
 
 def add_usd_market_caps(market_data):
-	"""Put 'market_cap_usd' beside each 'market_cap', which Yahoo gives in the ticker's own currency, converted at Yahoo's
-	exchange rates, so that market caps in won, yen and dollars compare. A ticker whose currency Yahoo has no rate for gets
-	none. Returns market_data."""
-	wanted = {market_cap_currency(data) for data in market_data.values() if 'market_cap' in data and market_cap_currency(data)} - {'USD'}
+	"""Put 'market_cap_usd' beside each 'market_cap', which Yahoo gives in the ticker's own currency, converted at
+	Yahoo's exchange rates, so that market caps in won, yen and dollars compare. A ticker whose currency Yahoo has no
+	rate for gets none. Returns market_data."""
+	wanted = {market_cap_currency(data) for data in market_data.values() if 'market_cap' in data} - {None, '', 'USD'}
 	rates = {'USD': 1.0}
-	for symbol, quote in (fetch([f'{currency}=X' for currency in wanted]) if wanted else {}).items(): # KRW=X: won per US dollar
+	# KRW=X: won per US dollar
+	for symbol, quote in (fetch([f'{currency}=X' for currency in wanted]) if wanted else {}).items():
 		if quote.get('regularMarketPrice'):
 			rates[symbol.removesuffix('=X')] = 1 / quote['regularMarketPrice']
 	for data in market_data.values():
@@ -400,8 +411,9 @@ def add_usd_market_caps(market_data):
 	return market_data
 
 def fetch_details(tickers, strict=False):
-	"""fetch_detail of every ticker, fetched at the same time, in one {ticker: detail}. A ticker whose fetch failed is left out
-	(and logged), so that a list goes out without it; strict=True raises its error instead, for a report on the ticker asked for."""
+	"""fetch_detail of every ticker, fetched at the same time, in one {ticker: detail}. A ticker whose fetch failed is
+	left out (and logged), so that a list goes out without it; strict=True raises its error instead, for a report on the
+	ticker asked for."""
 	tickers = list(tickers)
 	details = {}
 	for ticker, detail in zip(tickers, fetch_each(fetch_detail, tickers)):
diff --git a/tests/common.py b/tests/common.py
index b0a0914b..bed5beaf 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -7,12 +7,15 @@
 ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
 sys.path.insert(0, ROOT)
 os.chdir(ROOT)
-os.environ.update(telegramOutgoingWebhook='https://example.com/telegram', telegramOutgoingToken='tok', telegramChatID='1')
+os.environ.update(telegramOutgoingWebhook='https://example.com/telegram', telegramOutgoingToken='tok',
+	telegramChatID='1')
 
-import bot # first, and once: it monkey-patches with gevent, which must happen before requests imports ssl and never mid-test
+# first, and once: it monkey-patches with gevent, which must happen before requests imports ssl and never mid-test
+import bot
 from lib import answer, buttons, cache, charts, config, delivery, dialects, formats, inbound, links, markets, render
 from lib import sharesight, shortman, state, telegram, transports, watchlist, worker, yahoo
-from lib.reports import about, cal, history, holdings, milestone, performance, price, profile, rankings, rating, reminder, shorts, trades
+from lib.reports import about, cal, history, holdings, milestone, performance, price, profile, rankings, rating
+from lib.reports import reminder, shorts, trades
 
 REAL = {(m.__name__, n): getattr(m, n) for m, n in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'),
 	(yahoo, 'price_history'), (yahoo, 'historic_high'),
@@ -20,6 +23,13 @@
 REAL_GET_HOLDINGS = watchlist.with_holdings
 
 
+def bare_environment():
+	"""os.environ without finbot's settings (the startup keys, the preferences, the flag that hands a job its
+	preferences): for a process that must start from the defaults."""
+	settings = {*config.STARTUP_KEYS, *config.PREFERENCE_KEYS, config.PASSED}
+	return {k: v for k, v in os.environ.items() if k not in settings}
+
+
 def temp_dir(case):
 	"""A new temporary directory, removed when the test `case` ends."""
 	path = tempfile.mkdtemp()
@@ -57,12 +67,14 @@ def market_entry(ticker):
 	threshold, sort and branch has rows."""
 	i = int(ticker[1:]) if ticker[1:].isdigit() else 0 # varied data so threshold/sort/sell branches all have rows
 	return {'marketState': STATES[i % 4], 'regularMarketTime': time.time(), 'quoteType': 'EQUITY',
-			'profile_title': 'Co ' + ticker, 'profile_exchange': 'NasdaqGS', 'percent_change': (i - 12) * 1.5, 'percent_change_premarket': (i - 12) * 1.5, 'percent_change_postmarket': (i - 12) * 1.5,
+			'profile_title': 'Co ' + ticker, 'profile_exchange': 'NasdaqGS', 'percent_change': (i - 12) * 1.5,
+			'percent_change_premarket': (i - 12) * 1.5, 'percent_change_postmarket': (i - 12) * 1.5,
 			'market_cap': 1e9, 'price_to_earnings_trailing': 9, 'price_to_earnings_forward': 9,
 			'beta': 2.0, 'short_percent': 20, 'fiftyTwoWeekHigh': 120.0, 'fiftyTwoWeekLow': 80.0,
 			'regularMarketPrice': 100.0, 'earnings_date': time.time() + 86400 * 3,
-			'ex_dividend_date': time.time() + 86400 * 3, 'recommend': 'buy' if i % 2 else 'sell', 'recommend_index': 1.2 if i % 2 else 4.2,
-			'recommend_analysts': 8, 'price_to_earnings_peg': 1.2, 'currency': 'USD'}
+			'ex_dividend_date': time.time() + 86400 * 3, 'recommend': 'buy' if i % 2 else 'sell',
+			'recommend_index': 1.2 if i % 2 else 4.2, 'recommend_analysts': 8, 'price_to_earnings_peg': 1.2,
+			'currency': 'USD'}
 
 
 class FinbotCase(unittest.TestCase):
@@ -108,7 +120,8 @@ def setUp(self):
 		self.patch(yahoo, 'prefetch_history', lambda ts: None)
 		self.patch(yahoo, 'historic_high', lambda t: (100.0, 50.0))
 		dates = pd.date_range(end=pd.Timestamp.today().normalize(), periods=400)
-		self.patch(yahoo, 'price_series', lambda t: ('Co ' + t, 'NASDAQ', pd.Series(range(100, 500), index=dates, name=t, dtype=float)))
+		self.patch(yahoo, 'price_series',
+			lambda t: ('Co ' + t, 'NASDAQ', pd.Series(range(100, 500), index=dates, name=t, dtype=float)))
 		self.patch(shortman, 'fetch', lambda md: md)
 		held = [{'id': i, 'instrument': {'code': t, 'market_code': 'NASDAQ'}, 'portfolio': {'name': 'Rob'},
 					 'capital_gain_percent': str(i * 4 - 40), 'value': '1000'} for i, t in enumerate(TICKERS)]
@@ -117,7 +130,8 @@ def setUp(self):
 			'holdings': held, 'currency_gain_percent': '0', 'capital_gain_percent': '2.5', 'total_gain_percent': '0'})
 		self.patch(sharesight, '_trades', lambda pid, days=None: [{
 			'id': 1, 'transaction_type': 'BUY', 'symbol': 'T1', 'market': 'NASDAQ',
-			'transaction_date': time.strftime('%Y-%m-%d'), 'quantity': '10', 'price': '5.0', 'value': '50', 'brokerage_currency_code': 'USD', 'holding_id': 7}])
+			'transaction_date': time.strftime('%Y-%m-%d'), 'quantity': '10', 'price': '5.0', 'value': '50',
+			'brokerage_currency_code': 'USD', 'holding_id': 7}])
 		self.typing = [] # every chat a test starts a typing notice in; none may still be sending it when the test ends
 		noted = lambda chat: self.typing.append(chat) or REAL_START_TYPING(chat)
 		self.patch(transports.Transport, 'start_typing', noted)
diff --git a/tests/test_charts.py b/tests/test_charts.py
index 94fae03f..9407a57d 100644
--- a/tests/test_charts.py
+++ b/tests/test_charts.py
@@ -99,7 +99,8 @@ def check_image(self, buf, name, bg=BG, edge_clean=True):
 		self.assertIn(tuple(int(c) for c in pixels[0, 0]), allowed, name)
 		if not edge_clean: # the shorts beach draws its sea right to the edge
 			return
-		edge = np.concatenate([pixels[:3].reshape(-1, 3), pixels[-3:].reshape(-1, 3), pixels[:, :3].reshape(-1, 3), pixels[:, -3:].reshape(-1, 3)])
+		edge = np.concatenate([pixels[:3].reshape(-1, 3), pixels[-3:].reshape(-1, 3), pixels[:, :3].reshape(-1, 3),
+			pixels[:, -3:].reshape(-1, 3)])
 		self.assertTrue((edge == bg).all(), f'{name}: something is drawn on the image edge (clipped text or bars?)')
 
 	def test_each_chart_type(self):
@@ -132,8 +133,10 @@ def test_shorts_chart_sizes_and_verdicts(self):
 				bg=SHORTS_BG, edge_clean=False)
 		self.check_image(shorts([('A VERY LONG LABEL', 15.0), ('B', 0.4)]), 'shorts long label and tiny value',
 			bg=SHORTS_BG, edge_clean=False)
-		tiers = [(0, 'board shorts'), (4.9, 'board shorts'), (5, 'cargo shorts'), (10, 'boxers'), (15, 'gym shorts'), (20, 'cutoffs'),
-				 (30, 'short shorts'), (40, 'daisy dukes'), (50, 'hot pants'), (60, 'speedos'), (70, 'budgie smugglers'),
+		tiers = [(0, 'board shorts'), (4.9, 'board shorts'), (5, 'cargo shorts'), (10, 'boxers'), (15, 'gym shorts'),
+			(20, 'cutoffs'),
+				 (30, 'short shorts'), (40, 'daisy dukes'), (50, 'hot pants'), (60, 'speedos'),
+				 (70, 'budgie smugglers'),
 				 (80, 'just a belt'), (90, 'never nudes'), (200, 'never nudes')]
 		self.assertEqual([render.shorts_verdict(p) for p, _ in tiers], [name for _, name in tiers])
 
@@ -156,7 +159,7 @@ def test_a_long_title_is_shrunk_to_fit(self):
 
 
 class RenderPool(unittest.TestCase):
-	"""The bot draws charts in one native thread (render.render_off_hub), so the gevent hub keeps serving while a chart is drawn."""
+	"""The bot draws charts in one native thread (render.render_off_hub): the hub keeps serving while one is drawn."""
 
 	def setUp(self):
 		"""Draw in the render thread, as the bot does, for this test only."""
@@ -170,7 +173,8 @@ def test_charts_asked_for_at_once_all_finish(self):
 		jobs += [gevent.spawn(bars, MIXED)]
 		jobs += [gevent.spawn(uncached, render.Chart('bars', 'T', rows=tuple(MIXED)))]
 		gevent.joinall(jobs, timeout=120)
-		self.assertTrue(all(job.successful() for job in jobs), 'a chart did not finish (several native threads deadlock matplotlib)')
+		self.assertTrue(all(job.successful() for job in jobs),
+			'a chart did not finish (several native threads deadlock matplotlib)')
 		self.assertTrue(all(decode(job.value).size == (1280, 800) for job in jobs))
 
 	def test_the_hub_keeps_serving_while_a_chart_is_drawn(self):
@@ -230,7 +234,8 @@ def test_first_row_is_on_top_and_spacers_are_skipped(self):
 		real = Axes.set_ylim
 		with mock.patch.object(Axes, 'set_ylim', lambda ax, *a, **k: (calls.append(a), real(ax, *a, **k))[1]):
 			bars(MIXED)
-		self.assertTrue(any(len(a) == 2 and a[0] > a[1] for a in calls), 'y axis must be inverted so row 0 is drawn first (top)')
+		self.assertTrue(any(len(a) == 2 and a[0] > a[1] for a in calls),
+			'y axis must be inverted so row 0 is drawn first (top)')
 
 	def test_pixels_follow_the_sign_of_the_data(self):
 		"""The drawn image is blue for gains and red for losses, in proportion to the rows."""
@@ -252,7 +257,8 @@ def test_session_tints_the_background(self):
 		self.assertEqual(plain, (0x14, 0x18, 0x1f))
 		self.assertEqual(close, plain) # at close stays untinted
 		self.assertTrue(mid[1] > mid[0] and mid[1] > plain[1]) # green
-		self.assertTrue(pre[0] > plain[0] and pre[1] > plain[1] and pre[0] > pre[2] and pre[1] > pre[2]) # yellow: red and green channels both lead blue
+		# yellow: red and green channels both lead blue
+		self.assertTrue(pre[0] > plain[0] and pre[1] > plain[1] and pre[0] > pre[2] and pre[1] > pre[2])
 		self.assertLess(max(abs(int(a) - int(b)) for a, b in zip(pre, plain)), 20) # subtle
 
 	def test_longer_bar_is_drawn_longer(self):
diff --git a/tests/test_config.py b/tests/test_config.py
index b527850c..5fa82144 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -1,4 +1,4 @@
-"""The two config files: finbot.env (the startup file, loaded by the launcher) and preferences.conf (read by the bot, reloadable)."""
+"""The two config files: finbot.env (the startup file, which the launcher loads) and preferences.conf (reloadable)."""
 import ast, contextlib, dataclasses, io, os, re, subprocess, sys, unittest
 from pathlib import Path
 from unittest import mock
@@ -67,8 +67,10 @@ def errors(self, text):
 
 	def test_comments_blanks_quotes_and_equals_in_values(self):
 		"""Comments and blank lines are skipped, quotes come off, and spaces and blanks in values are kept."""
-		values, _ = self.parse('# c\n\nfuture_days = 30\nhyperlinkProvider="google"\nexclude_portfolios=1 2\ncountry_code=\n')
-		self.assertEqual(values, {'future_days': '30', 'hyperlinkProvider': 'google', 'exclude_portfolios': '1 2', 'country_code': ''})
+		values, _ = self.parse('# c\n\nfuture_days = 30\nhyperlinkProvider="google"\nexclude_portfolios=1 2\n'
+			'country_code=\n')
+		self.assertEqual(values,
+			{'future_days': '30', 'hyperlinkProvider': 'google', 'exclude_portfolios': '1 2', 'country_code': ''})
 
 	def test_missing_file_is_none(self):
 		"""No preferences.conf is not an error: None."""
@@ -111,7 +113,8 @@ def test_invalid_values_are_named(self):
 
 	def test_booleans_take_the_usual_words_and_refuse_a_typo(self):
 		"""Booleans take true/false, yes/no, on/off and 1/0 in any case; a typo is an error (it used to mean False)."""
-		for word, value in (('True', True), ('yes', True), ('ON', True), ('1', True), ('false', False), ('No', False), ('off', False), ('0', False)):
+		for word, value in (('True', True), ('yes', True), ('ON', True), ('1', True), ('false', False), ('No', False),
+				('off', False), ('0', False)):
 			with self.subTest(word=word):
 				self.assertIs(config.preference_values(config._getenv_for({'graph': word})).graph, value)
 		with self.assertRaises(config.ConfigError) as caught: # it used to mean False without a word
@@ -156,11 +159,13 @@ def test_precedence_file_then_environment_then_default(self):
 	def test_the_old_multiplier_name_is_gone_everywhere(self):
 		"""volatile_multiplier is no longer read, and in the environment it stops the start with the new name."""
 		with mock.patch.dict(os.environ, {'volatile_multiplier': '3'}):
-			self.assertEqual(config.preference_values(config._getenv_for({})).volatility_multiplier, 2.0) # not read any more
+			# not read any more
+			self.assertEqual(config.preference_values(config._getenv_for({})).volatility_multiplier, 2.0)
 			with self.assertRaises(config.ConfigError) as caught:
 				config.check_startup_environment()
 		self.assertIn('volatile_multiplier was renamed to volatility_multiplier', caught.exception.errors[0])
-		self.assertEqual(config.preference_values(config._getenv_for({'volatility_multiplier': '4'})).volatility_multiplier, 4.0)
+		values = config.preference_values(config._getenv_for({'volatility_multiplier': '4'}))
+		self.assertEqual(values.volatility_multiplier, 4.0)
 
 	def test_startup_file_linting(self):
 		"""util/check_config.py reports bad lines in finbot.env first, then bad values, and nothing for a good file."""
@@ -200,17 +205,20 @@ def test_a_change_is_seen_at_once(self):
 		"""A reload applies every change, says which, and swaps in a new object, so no request sees half a reload."""
 		self.write('future_days=45\nprice_percent=7\nshorts_percent=12\nexclude_portfolios=1156263\n'
 			'exchange_suffix=nz\nhyperlinkProvider=yahoo\n'
-			'hyperlinkFooter=false\ndebug=1\ngraph=0\ndemote_volatile=0\nvolatility_multiplier=3\npast_days=14\nmin_analysts=3\n')
+			'hyperlinkFooter=false\ndebug=1\ngraph=0\ndemote_volatile=0\nvolatility_multiplier=3\npast_days=14\n'
+			'min_analysts=3\n')
 		before = config.prefs
 		changes = config.reload()
 		self.assertEqual(len(changes), 13)
 		self.assertIn('future_days 30 -> 45', changes)
 		self.assertIsNot(config.prefs, before) # one new object: no request sees half a reload
-		self.assertEqual((config.prefs.future_days, config.prefs.past_days, config.prefs.price_percent, config.prefs.min_analysts), (45, 14, 7.0, 3))
+		self.assertEqual((config.prefs.future_days, config.prefs.past_days, config.prefs.price_percent,
+			config.prefs.min_analysts), (45, 14, 7.0, 3))
 		prefs = config.prefs
 		self.assertEqual((prefs.exclude_portfolios, prefs.exchange_suffix, prefs.hyperlinkProvider,
 			prefs.hyperlinkFooter), (('1156263',), 'NZ', 'yahoo', False))
-		self.assertEqual((config.prefs.debug, config.prefs.graph, config.prefs.demote_volatile, config.prefs.volatility_multiplier), (True, False, False, 3.0))
+		self.assertEqual((config.prefs.debug, config.prefs.graph, config.prefs.demote_volatile,
+			config.prefs.volatility_multiplier), (True, False, False, 3.0))
 		self.assertEqual(config.reload(), []) # nothing more to change
 
 	def test_a_value_removed_from_the_file_goes_back_to_its_default(self):
@@ -240,7 +248,8 @@ def test_an_error_applies_nothing(self):
 		"""A reload that finds any error applies nothing."""
 		self.write('future_days=45\n')
 		config.reload()
-		for text in ('future_days=3\nprice_percent=abc\n', 'future_days=3\nfutur_days=4\n', 'future_days=3\nslack_webhook=x\n'):
+		for text in ('future_days=3\nprice_percent=abc\n', 'future_days=3\nfutur_days=4\n',
+				'future_days=3\nslack_webhook=x\n'):
 			self.write(text)
 			with self.assertRaises(config.ConfigError):
 				config.reload()
@@ -258,17 +267,19 @@ def test_a_missing_file_is_an_error_and_keeps_the_running_values(self):
 		self.file.write_text('')
 
 	def test_nothing_copies_the_config_or_freezes_a_preference(self):
-		"""Modules read config.prefs.<name> when they run, so that a reload is seen: no star import (which copied every name into
-		each module), and no function default or module constant computed from a preference."""
+		"""Modules read config.prefs.<name> when they run, so that a reload is seen: no star import (which copied every
+		name into each module), and no function default or module constant computed from a preference."""
 		def reads_prefs(node):
 			"""Whether an expression reads config.prefs."""
-			return any(isinstance(n, ast.Attribute) and n.attr == 'prefs' and isinstance(n.value, ast.Name) and n.value.id == 'config' for n in ast.walk(node))
+			return any(isinstance(n, ast.Attribute) and n.attr == 'prefs' and isinstance(n.value,
+				ast.Name) and n.value.id == 'config' for n in ast.walk(node))
 		offenders = []
 		for path in sorted([*(REPO / 'lib').rglob('*.py'), *REPO.glob('*.py'), *(REPO / 'util').glob('*.py')]):
 			tree = ast.parse(path.read_text())
 			for node in ast.walk(tree):
 				if isinstance(node, ast.ImportFrom) and node.module == 'lib.config':
-					offenders.append(f'{path.name}:{node.lineno} from lib.config import {", ".join(a.name for a in node.names)}')
+					names = ', '.join(a.name for a in node.names)
+					offenders.append(f'{path.name}:{node.lineno} from lib.config import {names}')
 				if isinstance(node, (ast.FunctionDef, ast.AsyncFunctionDef)):
 					for default in [*node.args.defaults, *[d for d in node.args.kw_defaults if d]]:
 						if reads_prefs(default):
@@ -289,8 +300,9 @@ def test_a_job_process_runs_on_the_snapshot_not_the_edited_file(self):
 		data = common.temp_dir(self)
 		Path(data, 'preferences.conf').write_text('future_days=99\n')
 		show = "from lib import config as c; print(c.prefs.future_days)"
-		run = lambda **extra: subprocess.run([sys.executable, '-c', show], cwd=REPO, capture_output=True, text=True, check=True,
-			env={**{k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k != config.PASSED}, 'DATA_DIR': data, **extra}).stdout.split()[-1]
+		parent = {k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k != config.PASSED}
+		run = lambda **extra: subprocess.run([sys.executable, '-c', show], cwd=REPO, capture_output=True, text=True,
+			check=True, env={**parent, 'DATA_DIR': data, **extra}).stdout.split()[-1]
 		self.assertEqual(run(**env), '45') # the parent's values, and the file is not read
 		self.assertEqual(run(), '99') # a script run by hand reads the file itself
 
@@ -317,8 +329,9 @@ def run_config(self, code='from lib import config as c; print(c.prefs.future_day
 		data = common.temp_dir(self)
 		if prefs is not None:
 			Path(data, 'preferences.conf').write_text(prefs)
-		e = {k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k not in config.STARTUP_KEYS and k != config.PASSED}
-		return subprocess.run([sys.executable, '-c', code], cwd=REPO, env={**e, 'DATA_DIR': data, **env}, capture_output=True, text=True)
+		env = {**common.bare_environment(), 'DATA_DIR': data, **env}
+		return subprocess.run([sys.executable, '-c', code], cwd=REPO, env=env,
+			capture_output=True, text=True)
 
 	def test_no_preferences_file_is_fine(self):
 		"""finbot starts without a preferences.conf, on the defaults."""
@@ -331,7 +344,8 @@ def test_a_preferences_file_is_read_at_start(self):
 
 	def test_a_broken_preferences_file_refuses_to_start(self):
 		"""A bad value, unknown key, startup key or malformed line in preferences.conf stops the start, saying why."""
-		for prefs, expect in (('future_days=abc\n', "future_days='abc' is not a whole number"), ('futur_days=1\n', 'unknown key futur_days'),
+		for prefs, expect in (('future_days=abc\n', "future_days='abc' is not a whole number"),
+			('futur_days=1\n', 'unknown key futur_days'),
 				('slack_webhook=x\n', 'belongs in finbot.env'), ('what\n', 'expected KEY=value')):
 			out = self.run_config(prefs=prefs)
 			self.assertEqual(out.returncode, 1, prefs)
@@ -342,7 +356,8 @@ def test_an_unreadable_preferences_file_refuses_to_start(self):
 		"""A preferences.conf that cannot be read stops the start."""
 		data = common.temp_dir(self)
 		Path(data, 'preferences.conf').mkdir()
-		out = subprocess.run([sys.executable, '-c', 'from lib import config'], cwd=REPO, env={**os.environ, 'DATA_DIR': data}, capture_output=True, text=True)
+		out = subprocess.run([sys.executable, '-c', 'from lib import config'], cwd=REPO,
+			env={**os.environ, 'DATA_DIR': data}, capture_output=True, text=True)
 		self.assertEqual(out.returncode, 1)
 		self.assertIn('cannot read', out.stderr)
 
@@ -359,7 +374,8 @@ def test_a_preference_in_the_environment_is_refused_by_the_bot(self):
 			with self.assertRaises(config.ConfigError) as caught:
 				config.check_startup_environment()
 		self.assertEqual(len(caught.exception.errors), 2)
-		self.assertIn('future_days is a preference; move it from finbot.env to preferences.conf', '\n'.join(caught.exception.errors))
+		self.assertIn('future_days is a preference; move it from finbot.env to preferences.conf',
+			'\n'.join(caught.exception.errors))
 		with mock.patch.dict(os.environ, {'future_days': '30', config.PASSED: '1'}):
 			config.check_startup_environment() # a job process gets its preferences this way
 		with mock.patch.dict(os.environ, {'ip': '1.2.3.4', 'telegramChatID': '5'}):
@@ -382,11 +398,13 @@ def test_a_leftover_env_in_the_checkout_refuses_to_start(self):
 				config.check_startup_environment()
 		self.assertEqual(len(caught.exception.errors), 2)
 		self.assertIn(f'{old} is the old single config file, which finbot no longer reads', caught.exception.errors[0])
-		self.assertIn('/util/split_env.py, which writes finbot.env.new and preferences.conf.new', caught.exception.errors[1])
+		self.assertIn('/util/split_env.py, which writes finbot.env.new and preferences.conf.new',
+			caught.exception.errors[1])
 
 	def test_a_discord_webhook_uses_the_native_endpoint(self):
 		"""A Discord webhook URL is used as is, its old /slack suffix removed."""
-		for url in ('https://discord.com/api/webhooks/1/abc', 'https://discord.com/api/webhooks/1/abc/slack/'): # the old setup's suffix
+		# the old setup's suffix
+		for url in ('https://discord.com/api/webhooks/1/abc', 'https://discord.com/api/webhooks/1/abc/slack/'):
 			values = config.startup_values(lambda name, default=None: url if name == 'discord_webhook' else default)
 			self.assertEqual(values.webhooks['discord'], 'https://discord.com/api/webhooks/1/abc')
 
@@ -440,7 +458,8 @@ def test_reload_all_reloads_preferences_and_the_schedule_independently(self):
 				bot.reload_all()
 			self.assertIn('config: future_days 30 -> 45', err.getvalue())
 			scheduler.reload.assert_called_once()
-			file.write_text('futur_days=1\n') # a preferences error does not stop the schedule reload, and the bot survives it
+			# a preferences error does not stop the schedule reload, and the bot survives it
+			file.write_text('futur_days=1\n')
 			err = io.StringIO()
 			with contextlib.redirect_stderr(err):
 				bot.reload_all()
@@ -462,7 +481,8 @@ def test_reload_with_the_scheduler_off_still_reloads_preferences(self):
 class Tools(unittest.TestCase):
 	def test_split_env(self):
 		"""split_env sorts an old .env into startup keys and prefs, renames volatile_multiplier, lists what it drops."""
-		startup, prefs, notes = split_env.split('slack_webhook=x\nfuture_days=3\nvolatile_multiplier=2\nslackToken=y\nSTATE_DIRECTORY=/a\nbogus=1\nnot a setting\n# c\nport=80\n')
+		startup, prefs, notes = split_env.split('slack_webhook=x\nfuture_days=3\nvolatile_multiplier=2\nslackToken=y\n'
+			'STATE_DIRECTORY=/a\nbogus=1\nnot a setting\n# c\nport=80\n')
 		self.assertEqual(startup, ['slack_webhook=x', 'port=80'])
 		self.assertEqual(prefs, ['future_days=3', 'volatility_multiplier=2'])
 		self.assertEqual(len(notes), 4)
@@ -470,7 +490,8 @@ def test_split_env(self):
 
 	def test_split_env_reads_the_old_dotenv_format_too(self):
 		"""split_env reads the old dotenv format: quotes, export, comments after values and # inside URLs."""
-		old = "# c\nfuture_days = '30' # note\nexport slack_webhook = \"https://x/#frag\"\nport = 80 # comment\nbare\ndiscord_webhook=http://y/#z\n\n"
+		old = ("# c\nfuture_days = '30' # note\nexport slack_webhook = \"https://x/#frag\"\nport = 80 # comment\nbare\n"
+			"discord_webhook=http://y/#z\n\n")
 		startup, prefs, notes = split_env.split(old)
 		self.assertEqual(startup, ['slack_webhook=https://x/#frag', 'port=80', 'discord_webhook=http://y/#z'])
 		self.assertEqual(prefs, ['future_days=30'])
@@ -481,7 +502,8 @@ def test_split_env_writes_next_to_the_file_and_never_overwrites(self):
 		them, and says so for a missing file."""
 		d = Path(common.temp_dir(self))
 		(d / 'old.env').write_text('port=80\nfuture_days=3\n')
-		run = lambda: subprocess.run([sys.executable, str(REPO / 'util' / 'split_env.py'), str(d / 'old.env')], capture_output=True, text=True)
+		run = lambda: subprocess.run([sys.executable, str(REPO / 'util' / 'split_env.py'), str(d / 'old.env')],
+			capture_output=True, text=True)
 		self.assertEqual(run().returncode, 0)
 		self.assertIn('port=80', (d / 'finbot.env.new').read_text())
 		self.assertIn('future_days=3', (d / 'preferences.conf.new').read_text())
@@ -489,7 +511,8 @@ def test_split_env_writes_next_to_the_file_and_never_overwrites(self):
 		out = run()
 		self.assertNotEqual(out.returncode, 0)
 		self.assertIn('refusing to overwrite', out.stderr)
-		out = subprocess.run([sys.executable, str(REPO / 'util' / 'split_env.py'), str(d / 'missing.env')], capture_output=True, text=True)
+		out = subprocess.run([sys.executable, str(REPO / 'util' / 'split_env.py'), str(d / 'missing.env')],
+			capture_output=True, text=True)
 		self.assertEqual((out.returncode, out.stderr.strip()), (1, f'{d}/missing.env does not exist: nothing to split'))
 
 	def test_check_config_script_exit_status_and_it_does_not_create_directories(self):
@@ -497,8 +520,8 @@ def test_check_config_script_exit_status_and_it_does_not_create_directories(self
 		d = Path(common.temp_dir(self))
 		(d / 'finbot.env').write_text('port=80\n')
 		data = d / 'data'
-		run = lambda: subprocess.run([sys.executable, str(REPO / 'util' / 'check_config.py'), str(d / 'finbot.env')], capture_output=True, text=True,
-			env={**os.environ, 'DATA_DIR': str(data)})
+		run = lambda: subprocess.run([sys.executable, str(REPO / 'util' / 'check_config.py'), str(d / 'finbot.env')],
+			capture_output=True, text=True, env={**os.environ, 'DATA_DIR': str(data)})
 		out = run()
 		self.assertEqual(out.returncode, 0, out.stderr)
 		self.assertFalse(data.exists()) # may run as root from systemd: must not create the data directory
diff --git a/tests/test_datadir.py b/tests/test_datadir.py
index 33e81464..8f3d3e6d 100644
--- a/tests/test_datadir.py
+++ b/tests/test_datadir.py
@@ -12,7 +12,8 @@ def run_config(code, **env):
 	"""What `code` prints when run in a fresh interpreter in the checkout, with DATA_DIR cleared and `env` added."""
 	e = {k: v for k, v in os.environ.items() if k != 'DATA_DIR'}
 	e.update(env)
-	return subprocess.run([sys.executable, '-c', code], cwd=REPO, env=e, capture_output=True, text=True, check=True).stdout.strip()
+	return subprocess.run([sys.executable, '-c', code], cwd=REPO, env=e, capture_output=True, text=True,
+		check=True).stdout.strip()
 
 
 class DataDirTests(FinbotCase):
@@ -37,7 +38,8 @@ def test_data_dir_moves_var_and_cache(self):
 
 	def test_config_reads_environment_only(self):
 		"""Settings come from the environment; python-dotenv is not loaded."""
-		out = run_config("import sys; from lib import config as c; print(c.prefs.price_percent, 'dotenv' in sys.modules)", price_percent='3.3')
+		code = "import sys; from lib import config as c; print(c.prefs.price_percent, 'dotenv' in sys.modules)"
+		out = run_config(code, price_percent='3.3')
 		self.assertEqual(out.split()[-2:], ['3.3', 'False'])
 
 	def test_state_directory_fallback(self):
@@ -47,17 +49,21 @@ def test_state_directory_fallback(self):
 			e = {k: v for k, v in os.environ.items() if k not in ('DATA_DIR', 'MPLCONFIGDIR')}
 			def run(**env):
 				"""The data dir, cache dir and MPLCONFIGDIR a fresh interpreter ends up with, given `env`."""
-				return subprocess.run([sys.executable, '-c', show], cwd=REPO, env=dict(e, **env), capture_output=True, text=True, check=True).stdout.split()[-3:]
+				return subprocess.run([sys.executable, '-c', show], cwd=REPO, env=dict(e, **env), capture_output=True,
+					text=True, check=True).stdout.split()[-3:]
 			self.assertEqual(run(STATE_DIRECTORY=d + ':/x/y'), [d, d + '/cache', d + '/cache/matplotlib'])
 			self.assertEqual(run(STATE_DIRECTORY='/x/y', DATA_DIR=d)[0], d) # DATA_DIR wins
 			self.assertEqual(run(DATA_DIR=d, MPLCONFIGDIR='/custom')[2], '/custom') # explicit value kept
 
 	def test_render_import_sets_mplconfigdir_without_home(self):
 		"""Importing render puts MPLCONFIGDIR in the cache before matplotlib loads: charts work without a home dir."""
-		e = {k: v for k, v in os.environ.items() if k not in ('MPLCONFIGDIR', 'XDG_CONFIG_HOME', 'XDG_CACHE_HOME', 'HOME')}
-		code = "import os; from lib import render; import matplotlib.dates; print(os.environ['MPLCONFIGDIR'])" # render loads matplotlib lazily
+		e = {k: v for k, v in os.environ.items() if k not in ('MPLCONFIGDIR', 'XDG_CONFIG_HOME', 'XDG_CACHE_HOME',
+			'HOME')}
+		# render loads matplotlib lazily
+		code = "import os; from lib import render; import matplotlib.dates; print(os.environ['MPLCONFIGDIR'])"
 		with tempfile.TemporaryDirectory() as d:
-			out = subprocess.run([sys.executable, '-c', code], cwd=REPO, env=dict(e, DATA_DIR=d), capture_output=True, text=True, check=True).stdout.strip()
+			out = subprocess.run([sys.executable, '-c', code], cwd=REPO, env=dict(e, DATA_DIR=d), capture_output=True,
+				text=True, check=True).stdout.strip()
 			self.assertEqual(out.split()[-1], d + '/cache/matplotlib')
 
 	def test_data_file_override_and_fallback(self):
@@ -86,12 +92,14 @@ def test_sync_defaults(self):
 		sync_defaults.sync(apply=True)
 		self.assertTrue(Path(config.env.data_dir, 'finbot_adr.json').exists())
 		self.assertTrue(Path(config.env.data_dir, 'preferences.conf').exists())
-		self.assertFalse(Path(config.env.data_dir, 'finbot.env').exists()) # the startup file is never copied into the data dir
+		# the startup file is never copied into the data dir
+		self.assertFalse(Path(config.env.data_dir, 'finbot.env').exists())
 		mine = Path(config.env.data_dir, 'finbot_adr.json')
 		mine.write_text('{"mine": 1}')
 		sync_defaults.sync(apply=True)
 		self.assertEqual(mine.read_text(), '{"mine": 1}') # never overwritten without --force
-		self.assertIn(('finbot_adr.json', 'skipped - destination differs'), sync_defaults.sync()) # existing files are skipped, and said so
+		# existing files are skipped, and said so
+		self.assertIn(('finbot_adr.json', 'skipped - destination differs'), sync_defaults.sync())
 		sync_defaults.sync(force=True)
 		self.assertNotEqual(mine.read_text(), '{"mine": 1}')
 		self.assertEqual(Path(str(mine) + '.bak').read_text(), '{"mine": 1}')
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 5ac10f0d..19a346b5 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -101,7 +101,8 @@ def test_a_bare_symbol_yahoo_does_not_know_is_tried_on_the_asx(self):
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts if t.endswith('.AX')})
 		self.assertIn('I added', str(self.command('.watchlist add BHP')))
 		self.assertIn('already tracking', str(self.command('.watchlist add BHP')))
-		self.assertIn('could not find', str(self.command('.watchlist add BHP.L'))) # a ticker with a market is not retried
+		# a ticker with a market is not retried
+		self.assertIn('could not find', str(self.command('.watchlist add BHP.L')))
 		self.assertEqual(watchlist.tickers(), ['BHP.AX'])
 		self.assertIn('I deleted', str(self.command('.watchlist del bhp')))
 		self.assertEqual(watchlist.tickers(), [])
@@ -114,7 +115,8 @@ def slow_fetch(tickers):
 			time.sleep(0.01)
 			return {t: market_entry(t) for t in tickers}
 		self.patch(yahoo, 'fetch', slow_fetch)
-		threads = [threading.Thread(target=holdings.prepare_watchlist, args=(TELEGRAM, '@u', 'add', t)) for t in ('T1', 'T2')]
+		threads = [threading.Thread(target=holdings.prepare_watchlist, args=(TELEGRAM, '@u', 'add', t)) for t in ('T1',
+			'T2')]
 		for thread in threads:
 			thread.start()
 		for thread in threads:
@@ -154,7 +156,8 @@ def no_history(t, *a, **k):
 
 	def test_history_without_a_ticker_or_with_extra_argument_gives_usage(self):
 		""".history with no ticker or an extra argument gets its usage, not the profile of a ticker called HISTORY."""
-		self.assertEqual(self.command('.history'), [('text', ['Usage: .history SYMBOL'])]) # not the profile of a ticker called HISTORY
+		# not the profile of a ticker called HISTORY
+		self.assertEqual(self.command('.history'), [('text', ['Usage: .history SYMBOL'])])
 		self.assertEqual(self.command('.history T1 1y'), [('text', ['Usage: .history SYMBOL'])])
 
 
@@ -162,7 +165,8 @@ class PortfolioNames(FinbotCase):
 	def test_an_unknown_portfolio_gets_the_same_answer_everywhere(self):
 		"""An unknown portfolio gets one answer from .trades, .performance, .holdings and .pie: the ones there are."""
 		for cmd in ('.trades 7d Nope', '.performance Nope', '.holdings Nope', '.pie Nope'):
-			with self.subTest(cmd=cmd): # it used to be a ValueError, "no Sharesight data found", or two wordings of this
+			# it used to be a ValueError, "no Sharesight data found", or two wordings of this
+			with self.subTest(cmd=cmd):
 				self.assertEqual(self.command(cmd), [('text', ['Nope portfolio not found. I only know about:', 'Rob'])])
 
 	def test_a_portfolio_name_in_any_case(self):
@@ -244,7 +248,8 @@ def test_a_bare_number_on_price_is_a_percent(self):
 
 	def test_top_is_gone(self):
 		""".top is a ticker now, like any word: the command is .price top."""
-		self.assertIn('Co TOP', str(self.command('.top 7d'))) # .price top 7d is the command; .top is a ticker now, like any word
+		# .price top 7d is the command; .top is a ticker now, like any word
+		self.assertIn('Co TOP', str(self.command('.top 7d')))
 
 	def test_a_ratio_command_reads_its_ticker_as_the_other_commands_do(self):
 		""".pe t1 reads its ticker in any case, and a trailing space after top is fine."""
@@ -292,7 +297,8 @@ def test_telegram_requires_the_secret_header(self):
 		"""A Telegram update without the secret header, or with a wrong one, is refused and not processed."""
 		message = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.help'}}
 		self.assertIn(b'Unauthorized', self.request(body=message)[0])
-		self.assertIn(b'Unauthorized', self.request(body=message, headers={'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'wrong'})[0])
+		self.assertIn(b'Unauthorized',
+			self.request(body=message, headers={'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'wrong'})[0])
 		self.assertEqual(self.sent, []) # nothing was processed
 
 	def test_an_empty_secret_proves_nothing(self):
@@ -311,7 +317,8 @@ def test_unknown_path_is_404(self):
 
 	def test_malformed_body_is_refused_not_a_crash(self):
 		"""A body that is not a JSON object is 400 once authenticated, and turned away unread before that."""
-		self.assertEqual(self.request(body=b'not json')[1], '403 Forbidden') # no secret: turned away before the body is read
+		# no secret: turned away before the body is read
+		self.assertEqual(self.request(body=b'not json')[1], '403 Forbidden')
 		for body in (b'not json', b'', b'[1, 2]', b'7'):
 			self.assertEqual(self.request(body=body, headers=self.SECRET)[1], '400 Bad Request', body)
 		self.assertEqual(self.errors, [])
@@ -340,10 +347,14 @@ def test_strangers_cannot_post_into_the_chats(self):
 
 		self.env(slackOutgoingToken='stok')
 
-		for path, body, headers, status in (('/', b'', {}, '404 Not Found'), ('/wp-login.php', b'log=x&pwd=y', {}, '404 Not Found'),
-				('/telegram', b'log=x', {}, '403 Forbidden'), ('/telegram', {'message': {}}, {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'wrong'}, '403 Forbidden'),
-				('/slack', b'', {}, '400 Bad Request'), ('/slack', b'token=stok', {}, '400 Bad Request'), ('/slack', {'type': 'event_callback'}, {}, '403 Forbidden'),
-				('/slack', {'token': 'wrong', 'type': 'event_callback'}, {}, '403 Forbidden'), ('/slack', {'token': 'stok'}, {}, '200 OK')):
+		for path, body, headers, status in (('/', b'', {}, '404 Not Found'),
+			('/wp-login.php', b'log=x&pwd=y', {}, '404 Not Found'),
+				('/telegram', b'log=x', {}, '403 Forbidden'),
+				('/telegram', {'message': {}}, {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'wrong'}, '403 Forbidden'),
+				('/slack', b'', {}, '400 Bad Request'), ('/slack', b'token=stok', {}, '400 Bad Request'),
+				('/slack', {'type': 'event_callback'}, {}, '403 Forbidden'),
+				('/slack', {'token': 'wrong', 'type': 'event_callback'}, {}, '403 Forbidden'),
+				('/slack', {'token': 'stok'}, {}, '200 OK')):
 			with self.subTest(path=path, body=body):
 				self.assertEqual(self.request(path, body, headers)[1], status)
 		self.assertEqual(self.errors, []) # nothing reached the chats
@@ -372,7 +383,8 @@ def test_a_reported_failure_names_the_network_not_its_secret_path(self):
 	def test_a_failure_after_authentication_is_reported_once_and_not_retried(self):
 		"""A failure on an authenticated update is reported once and answered 200, so Telegram doesn't send it again."""
 		broken = {'message': {'message_id': 1, 'from': {'first_name': 'A'}, 'text': '.help'}} # no chat
-		self.assertEqual(self.request(body=broken, headers=self.SECRET)[1], '200 OK') # a retry would only fail and report again
+		# a retry would only fail and report again
+		self.assertEqual(self.request(body=broken, headers=self.SECRET)[1], '200 OK')
 		self.assertEqual(len(self.errors), 1)
 		self.assertIn("KeyError: 'chat'", self.errors[0])
 
@@ -388,7 +400,8 @@ def test_a_telegram_update_sent_again_runs_once(self):
 		"""An update Telegram sends again runs once."""
 		calls = []
 		self.patch(worker, 'process_request', lambda *a: calls.append(a))
-		body = {'update_id': 7, 'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.price'}}
+		body = {'update_id': 7,
+			'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.price'}}
 		for _ in range(2): # Telegram sends an update again when it got no answer
 			self.assertEqual(self.request(body=body, headers=self.SECRET)[1], '200 OK')
 		gevent.sleep(0.05)
@@ -415,8 +428,10 @@ def test_button_presses_are_logged(self):
 		calls = []
 		self.patch(worker, 'process_callback', lambda *a: calls.append(a))
 		secret = {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'tok'}
-		for sender, who in (({'username': 'rob', 'first_name': 'Rob'}, '@rob'), ({'first_name': 'Ann'}, 'Ann'), ({}, 'unknown')):
-			body = json.dumps({'callback_query': {'id': '1', 'from': sender, 'data': 'j|7|abcd', 'message': {'message_id': 9, 'chat': {'id': 55}}}}).encode()
+		for sender, who in (({'username': 'rob', 'first_name': 'Rob'}, '@rob'), ({'first_name': 'Ann'}, 'Ann'),
+				({}, 'unknown')):
+			body = json.dumps({'callback_query': {'id': '1', 'from': sender, 'data': 'j|7|abcd',
+				'message': {'message_id': 9, 'chat': {'id': 55}}}}).encode()
 			err = io.StringIO()
 			with contextlib.redirect_stderr(err):
 				bot.main(wsgi_environ('/telegram', body, **secret), lambda *a, **k: None)
@@ -438,8 +453,8 @@ def setUp(self):
 
 	def event(self, event_id=None, **event):
 		"""Post a Slack message event to the WSGI handler and return the status."""
-		body = json.dumps({'token': 'stok', 'type': 'event_callback', 'authorizations': [{'user_id': 'UBOT'}], 'event_id': event_id,
-			'event': dict({'type': 'message', 'channel': 'C1', 'ts': '1.2'}, **event)}).encode()
+		body = json.dumps({'token': 'stok', 'type': 'event_callback', 'authorizations': [{'user_id': 'UBOT'}],
+			'event_id': event_id, 'event': dict({'type': 'message', 'channel': 'C1', 'ts': '1.2'}, **event)}).encode()
 		status = []
 		with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
 			bot.main(wsgi_environ('/slack', body), lambda s, *a, **k: status.append(s))
@@ -475,7 +490,8 @@ def test_the_message_goes_to_the_worker_with_its_ts(self):
 	def test_only_people_are_answered(self):
 		"""Messages from bots, this bot included, joins, edits and events without a user are not answered; people's
 		messages are, with mentions turned into @names."""
-		for event in ({'subtype': 'bot_message', 'bot_id': 'B1', 'text': "⚠️ /slack: KeyError: 'user'"}, # a webhook report
+		# a webhook report
+		for event in ({'subtype': 'bot_message', 'bot_id': 'B1', 'text': "⚠️ /slack: KeyError: 'user'"},
 					  {'user': 'UBOT', 'bot_id': 'B2', 'text': '.holdings: Please try again'}, # this bot's own reply
 					  {'user': 'UBOT', 'text': '.holdings'},
 					  {'subtype': 'channel_join', 'user': 'U1', 'text': '<@U1> has joined the channel'},
@@ -487,7 +503,8 @@ def test_only_people_are_answered(self):
 		self.assertEqual(self.event(user='U1', text='.price <@UBOT>'), '200 OK')
 		self.assertEqual(self.event(user='U2', subtype='thread_broadcast', text='.trades'), '200 OK')
 		self.assertEqual([(r.service, r.chat_id, r.user, r.text, r.bot_name) for r in self.requests],
-						 [('slack', 'C1', '<@U1>', '.price @UBOT', '@UBOT'), ('slack', 'C1', '<@U2>', '.trades', '@UBOT')])
+						 [('slack', 'C1', '<@U1>', '.price @UBOT', '@UBOT'),
+						 	('slack', 'C1', '<@U2>', '.trades', '@UBOT')])
 
 	def test_links_and_escapes_lose_their_markup(self):
 		"""Each link loses Slack's markup, however many there are (Slack links .AX tickers as domains), and Slack's
@@ -572,8 +589,8 @@ def test_unsigned_wrongly_signed_old_or_altered_requests_are_refused(self):
 
 
 class UserTextIsEscaped(FinbotCase):
-	"""Replies go out as Telegram HTML: text a user wrote is shown, never read as markup. One stray < in a .plan entry made Telegram
-	refuse that reply and every .plan reply after it."""
+	"""Replies go out as Telegram HTML: text a user wrote is shown, never read as markup. One stray < in a .plan entry
+	made Telegram refuse that reply and every .plan reply after it."""
 
 	def test_a_plan_is_shown_as_text(self):
 		"""A .plan entry is shown as text on Telegram and on Slack, escaped once, and a Slack mention stays markup."""
@@ -601,7 +618,7 @@ def test_a_telegram_name_is_escaped_where_it_arrives(self):
 
 
 class SecretsStayOutOfLogs(FinbotCase):
-	"""A failed request's error message carries its URL: the Telegram token, a Slack or Discord webhook path, a Yahoo crumb."""
+	"""A failed request's error carries its URL: the Telegram token, a Slack or Discord webhook path, a Yahoo crumb."""
 	SECRET_URL = 'https://api.telegram.org/bot123456:SECRET-tok_en/sendMessage?chat_id=1'
 
 	def stderr_of(self, call):
@@ -641,7 +658,7 @@ def test_the_traceback_of_a_reported_error_is_redacted(self):
 
 class SharedSession(unittest.TestCase):
 	def test_the_session_keeps_no_cookies(self):
-		"""Yahoo's cookie goes in an explicit header: a stored one would be sent to fc.yahoo.com, which then sets no new one."""
+		"""Yahoo's cookie goes in a header: a stored one would go to fc.yahoo.com, which then sets no new one."""
 		import email.message, urllib.request
 		from lib import net
 		headers = email.message.Message()
@@ -678,13 +695,15 @@ def test_sighup_during_startup_is_ignored_and_works_afterwards(self):
 		"""A SIGHUP while bot.py starts is ignored, where its default action would kill it; once serving, it reloads."""
 		skip_if_old_env(self)
 		with tempfile.TemporaryDirectory() as data:
-			env = {k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k not in config.STARTUP_KEYS}
+			env = bare_environment()
 			env.update(DATA_DIR=data, port='0', scheduler='False')
-			proc = subprocess.Popen([sys.executable, '-u', 'bot.py'], cwd=ROOT, env=env, stderr=subprocess.PIPE, stdout=subprocess.DEVNULL, text=True)
+			proc = subprocess.Popen([sys.executable, '-u', 'bot.py'], cwd=ROOT, env=env, stderr=subprocess.PIPE,
+				stdout=subprocess.DEVNULL, text=True)
 			self.addCleanup(proc.wait)
 			self.addCleanup(proc.kill)
 			deadline = time.monotonic() + 30
-			while not int(Path(f'/proc/{proc.pid}/status').read_text().split('SigIgn:')[1].split()[0], 16) & 1: # until SIGHUP is ignored
+			# until SIGHUP is ignored
+			while not int(Path(f'/proc/{proc.pid}/status').read_text().split('SigIgn:')[1].split()[0], 16) & 1:
 				self.assertLess(time.monotonic(), deadline)
 				time.sleep(0.001)
 			proc.send_signal(signal.SIGHUP) # still importing: the default action would kill it
@@ -713,13 +732,15 @@ def test_usr1_flushes_the_cache_and_keeps_the_state(self):
 			for name in cached: # the last: a write in flight
 				(cache / name).write_text('{}')
 			Path(data, 'finbot_watchlist.json').write_text('["T1"]') # state, not cache
-			env = {k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k not in config.STARTUP_KEYS}
+			env = bare_environment()
 			env.update(DATA_DIR=data, port='0', scheduler='False')
-			proc = subprocess.Popen([sys.executable, '-u', 'bot.py'], cwd=ROOT, env=env, stderr=subprocess.PIPE, stdout=subprocess.DEVNULL, text=True)
+			proc = subprocess.Popen([sys.executable, '-u', 'bot.py'], cwd=ROOT, env=env, stderr=subprocess.PIPE,
+				stdout=subprocess.DEVNULL, text=True)
 			self.addCleanup(proc.wait)
 			self.addCleanup(proc.kill)
 			deadline = time.monotonic() + 30
-			while not int(Path(f'/proc/{proc.pid}/status').read_text().split('SigIgn:')[1].split()[0], 16) & (1 << (signal.SIGUSR1 - 1)):
+			while not int(Path(f'/proc/{proc.pid}/status').read_text().split('SigIgn:')[1].split()[0],
+					16) & (1 << (signal.SIGUSR1 - 1)):
 				self.assertLess(time.monotonic(), deadline)
 				time.sleep(0.001)
 			proc.send_signal(signal.SIGUSR1) # still starting: ignored, where its default action would kill the bot
diff --git a/tests/test_finbot.py b/tests/test_finbot.py
index 79cd70b3..ff4a9e5f 100644
--- a/tests/test_finbot.py
+++ b/tests/test_finbot.py
@@ -59,7 +59,8 @@ def post(message):
 			return out, err.getvalue()
 		self.assertEqual(post({'sticker': {'file_id': 'x'}}), ([b''], '')) # stickers: silent, not logged
 		self.assertGreaterEqual(len(bot.git_version()), 7)
-		head = subprocess.run(['git', 'rev-parse', 'HEAD'], cwd=bot.__file__.rsplit('/', 1)[0], capture_output=True, text=True).stdout.strip()
+		head = subprocess.run(['git', 'rev-parse', 'HEAD'], cwd=bot.__file__.rsplit('/', 1)[0], capture_output=True,
+			text=True).stdout.strip()
 		self.assertEqual(bot.git_version(), head[:7]) # matches what git says
 
 
@@ -75,6 +76,20 @@ def test_every_function_has_a_docstring(self):
 					missing.append(f'{path.relative_to(ROOT)}:{node.lineno} {node.name}')
 		self.assertEqual(missing, [])
 
+	def test_every_line_fits_in_120_columns(self):
+		"""Every line of the code, the tests, the shipped defaults, the units, the tools and CI fits in 120
+		columns, tabs counting 4: an edit quotes whole lines. README.md, CLAUDE.md and doc/ are left out (CI runs no
+		tests for a change to those alone, so no test may read them), and so are data files."""
+		patterns = ('*.py', '*.service', 'lib/**/*.py', 'tests/*.py', 'util/*', 'defaults/*', '.github/workflows/*.yml',
+			'.claude/hooks/*')
+		files = sorted({path for pattern in patterns for path in Path(ROOT).glob(pattern)
+			if path.is_file() and path.suffix != '.json'})
+		long = [f'{path.relative_to(ROOT)}:{number}' for path in files
+			for number, line in enumerate(path.read_text(encoding='utf-8').splitlines(), 1)
+			if len(line.expandtabs(4)) > 120]
+		self.assertGreater(len(files), 60)
+		self.assertEqual(long, [])
+
 	def test_the_bot_and_its_reports_print_only_to_stderr(self):
 		"""systemd discards stdout (StandardOutput=null), so only util/ tools, run by hand, may print there."""
 		offenders = []
@@ -146,7 +161,8 @@ def test_unreadable_is_unknown(self):
 		"""No .git, a dangling ref, or a garbled HEAD all give 'unknown'."""
 		with tempfile.TemporaryDirectory() as empty, tempfile.TemporaryDirectory() as odd:
 			self.repo(odd, 'ref: refs/heads/gone')
-			self.assertEqual([bot.git_version(empty), bot.git_version(odd), bot.git_version('/nonexistent')], ['unknown'] * 3)
+			self.assertEqual([bot.git_version(empty), bot.git_version(odd), bot.git_version('/nonexistent')],
+				['unknown'] * 3)
 			self.repo(empty, 'garbage')
 			self.assertEqual(bot.git_version(empty), 'unknown')
 
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
index 6d0e01b4..359504c8 100644
--- a/tests/test_jobs.py
+++ b/tests/test_jobs.py
@@ -1,4 +1,4 @@
-"""The in-process scheduler (lib/jobs.py): schedule file parsing, each script's job(argv), and firing with a fake clock."""
+"""The in-process scheduler (lib/jobs.py): the schedule file, each script's job(argv), firing on a fake clock."""
 import contextlib, datetime, hashlib, io, json, os, signal, subprocess, sys, tempfile, threading, time, unittest
 import zoneinfo
 from pathlib import Path
@@ -57,22 +57,26 @@ def test_pasted_crontab_lines(self):
 	def test_quoted_arguments_may_hold_anything(self):
 		"""A quoted argument may hold shell characters; commands after a ; still split."""
 		url = 'https://example.com/feed?a=1&b=2;c>d $x |y'
-		parsed, errors = jobs.parse_schedule(f'01 0-10,23 * * Mon-Fri podcasts.py "{url}" > ~/out 2>&1 # hourly (UTC)\n', UTC)
+		line = f'01 0-10,23 * * Mon-Fri podcasts.py "{url}" > ~/out 2>&1 # hourly (UTC)\n'
+		parsed, errors = jobs.parse_schedule(line, UTC)
 		self.assertEqual(errors, [])
 		self.assertEqual(parsed[0].commands[0][2], {'feed_url': url})
 		parsed, errors = jobs.parse_schedule(f"5 4 * * * podcasts.py '{url}'; shorts.py\n", UTC)
-		self.assertEqual([(label[:9], kw) for label, _, kw in parsed[0].commands], [('podcasts.', {'feed_url': url}), ('shorts.py', {})])
+		self.assertEqual([(label[:9], kw) for label, _, kw in parsed[0].commands],
+			[('podcasts.', {'feed_url': url}), ('shorts.py', {})])
 
 	def test_unquoted_shell_syntax_is_still_refused(self):
 		"""Pipes, &&, a trailing &, $(), backticks, input redirects, subshells and an unterminated quote are refused."""
-		for line in ('0 1 * * * trades.py | tee x', '0 1 * * * trades.py && shorts.py & ', '0 1 * * * trades.py $(date)', '0 1 * * * trades.py `x`',
+		for line in ('0 1 * * * trades.py | tee x', '0 1 * * * trades.py && shorts.py & ',
+			'0 1 * * * trades.py $(date)', '0 1 * * * trades.py `x`',
 					 '0 1 * * * trades.py < in', '0 1 * * * (trades.py)', '0 1 * * * trades.py "unterminated'):
 			parsed, errors = jobs.parse_schedule(line + '\n', UTC)
 			self.assertEqual((parsed, len(errors)), ([], 1), line)
 
 	def test_redirects_and_braces_are_dropped(self):
 		"""Output redirects, braces, paths and trailing comments are dropped from a command."""
-		for command in ('trades.py > /dev/null', 'trades.py >/dev/null 2>&1', 'trades.py &> /dev/null', 'trades.py >> ~/x 2>&1',
+		for command in ('trades.py > /dev/null', 'trades.py >/dev/null 2>&1', 'trades.py &> /dev/null',
+			'trades.py >> ~/x 2>&1',
 						'{ trades.py ;} > ~/x', '/opt/finbot/trades.py # note (x)'):
 			parsed, errors = jobs.parse_schedule(f'0 1 * * * {command}\n', UTC)
 			self.assertEqual(errors, [], command)
@@ -86,7 +90,8 @@ def test_bin_run_wrapper_is_no_longer_accepted(self):
 
 	def test_bad_lines_are_skipped_and_named(self):
 		"""Bad lines are skipped and reported by line number; the good ones load."""
-		text = '0 1 * * * ls | wc -l\n0 1 * * * price.py bogus\n0 1 * * * nothere.py\n61 1 * * * trades.py\n0 1 * *\nFOO=bar\n5 4 * * * shorts.py\n'
+		text = ('0 1 * * * ls | wc -l\n0 1 * * * price.py bogus\n0 1 * * * nothere.py\n61 1 * * * trades.py\n0 1 * *\n'
+			'FOO=bar\n5 4 * * * shorts.py\n')
 		parsed, errors = jobs.parse_schedule(text, UTC)
 		self.assertEqual([j.expr for j in parsed], ['5 4 * * *'])
 		self.assertEqual([number for number, _, _ in errors], [1, 2, 3, 4, 5, 6])
@@ -94,7 +99,8 @@ def test_bad_lines_are_skipped_and_named(self):
 	def test_impossible_times_are_marked_and_reboot_is_just_unsupported(self):
 		"""Times that never come (30 February, minute 61, month 13, @bogus) are ImpossibleTime; @reboot is only
 		unsupported, and 29 February does come."""
-		for line in ('0 0 30 2 * trades.py', '0 9 31 4,6,9,11 * trades.py', '61 1 * * * trades.py', '0 0 * 13 * trades.py', '@bogus trades.py'):
+		for line in ('0 0 30 2 * trades.py', '0 9 31 4,6,9,11 * trades.py', '61 1 * * * trades.py',
+				'0 0 * 13 * trades.py', '@bogus trades.py'):
 			parsed, errors = jobs.parse_schedule(line + '\n', UTC)
 			self.assertEqual(parsed, [], line)
 			self.assertIsInstance(errors[0][2], jobs.ImpossibleTime, line)
@@ -104,7 +110,8 @@ def test_impossible_times_are_marked_and_reboot_is_just_unsupported(self):
 
 	def test_cron_tz_applies_to_the_lines_below(self):
 		"""CRON_TZ sets the zone of the lines below it; an unknown zone is skipped."""
-		parsed, _ = jobs.parse_schedule('1 1 * * * trades.py\nCRON_TZ=Australia/Melbourne\n2 2 * * * trades.py\nCRON_TZ=Nowhere/Land\n', UTC)
+		text = '1 1 * * * trades.py\nCRON_TZ=Australia/Melbourne\n2 2 * * * trades.py\nCRON_TZ=Nowhere/Land\n'
+		parsed, _ = jobs.parse_schedule(text, UTC)
 		self.assertEqual([j.zone.key for j in parsed], ['UTC', 'Australia/Melbourne'])
 
 	def test_local_zone_follows_tz(self):
@@ -140,7 +147,8 @@ def test_podcasts(self):
 		url = 'https://example.com/podcast.rss'
 		label, kwargs = podcasts.job([url])
 		self.assertEqual(kwargs, {'feed_url': url})
-		self.assertIn('finbot_podcast_seen_' + label.split()[1], podcasts.get_seen_filename(url)) # the label is the start of the state file's hash
+		# the label is the start of the state file's hash
+		self.assertIn('finbot_podcast_seen_' + label.split()[1], podcasts.get_seen_filename(url))
 		self.assertEqual(label, 'podcasts.py ' + hashlib.md5(url.encode()).hexdigest()[:8])
 		for bad in ([], [url, url]):
 			with self.assertRaises(ValueError):
@@ -175,11 +183,12 @@ def test_cli_runs_the_handler(self):
 
 	def test_each_scheduled_report_runs_from_its_top_level_script(self):
 		"""./price.py and the rest run lib/reports/<name>.py, so cron lines and runs by hand keep working."""
-		self.assertEqual(sorted(path.stem for path in Path(ROOT).glob('*.py') if path.stem not in ('bot', 'bot_send_stdin')),
+		self.assertEqual(sorted(path.stem for path in Path(ROOT).glob('*.py') if path.stem not in ('bot',
+			'bot_send_stdin')),
 						 sorted(jobs.SCRIPTS))
 		with tempfile.TemporaryDirectory() as data:
-			run = lambda *args: subprocess.run([sys.executable, *args], cwd=ROOT, capture_output=True, text=True, timeout=60,
-				env={**os.environ, 'DATA_DIR': data})
+			run = lambda *args: subprocess.run([sys.executable, *args], cwd=ROOT, capture_output=True, text=True,
+				timeout=60, env={**os.environ, 'DATA_DIR': data})
 			self.assertEqual(run('cleanup.py', '0').returncode, 0)
 			bad = run('price.py', 'bogus')
 			self.assertEqual(bad.returncode, 1)
@@ -193,7 +202,8 @@ def test_removes_only_old_finbot_cache_files(self):
 		self.env(cache_dir=folder)
 
 		old, new = time.time() - 40 * 86400, time.time() - 5 * 86400
-		for name, stamp in (('finbot_old.json', old), ('finbot_new.json', new), ('other_old.json', old), ('finbot_old.png', old)):
+		for name, stamp in (('finbot_old.json', old), ('finbot_new.json', new), ('other_old.json', old),
+				('finbot_old.png', old)):
 			path = os.path.join(folder, name)
 			Path(path).write_text('x')
 			os.utime(path, (stamp, stamp))
@@ -207,7 +217,8 @@ def test_a_temp_file_being_written_is_left_alone(self):
 		folder = temp_dir(self)
 		self.env(cache_dir=folder)
 
-		for name, stamp in (('finbot_a.json.x1.tmp', time.time() - 5), ('finbot_b.json.x2.tmp', time.time() - 40 * 86400)):
+		for name, stamp in (('finbot_a.json.x1.tmp', time.time() - 5),
+				('finbot_b.json.x2.tmp', time.time() - 40 * 86400)):
 			Path(folder, name).write_text('x')
 			os.utime(Path(folder, name), (stamp, stamp))
 		self.assertEqual(cleanup.run(days=0), 1) # a flush: the stale one goes, the write in flight stays
@@ -227,7 +238,8 @@ def flag(self, **env):
 		e = {k: v for k, v in os.environ.items() if k != 'scheduler'}
 		e.update(env)
 		repo = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
-		return subprocess.run([sys.executable, '-c', code], cwd=repo, env=e, capture_output=True, text=True, check=True).stdout.split()[-1]
+		return subprocess.run([sys.executable, '-c', code], cwd=repo, env=e, capture_output=True, text=True,
+			check=True).stdout.split()[-1]
 
 	def test_on_by_default_and_off_for_false(self):
 		"""The scheduler is on by default and with any true word, off with any false one."""
@@ -292,11 +304,13 @@ def test_a_run_missed_within_the_catch_up_window_fires_once(self):
 		state.save(jobs.STATE_FILE, first.last)
 		self.clock.now = at(13, 21, 43)
 		s = self.scheduler()
-		self.assertEqual([t for _, t in s.due(self.clock.now)], [at(13, 21, 40)]) # 21:20 is older than 5 minutes: skipped
+		# 21:20 is older than 5 minutes: skipped
+		self.assertEqual([t for _, t in s.due(self.clock.now)], [at(13, 21, 40)])
 
 	def test_a_corrupt_state_file_is_reported_and_the_schedule_starts_afresh(self):
 		"""A state file cut off mid-write is reported once, and the schedule runs as usual and writes it afresh."""
-		Path(config.env.data_dir, jobs.STATE_FILE).write_text('{"*/20 * * * Mon-Fri|UTC|trades.py": "2026-10-13T21:') # cut off mid-write
+		# cut off mid-write
+		Path(config.env.data_dir, jobs.STATE_FILE).write_text('{"*/20 * * * Mon-Fri|UTC|trades.py": "2026-10-13T21:')
 		s = self.scheduler()
 		self.assertEqual((s.last, len(self.errors)), ({}, 1))
 		self.assertIn('cannot read finbot_jobs.json', self.errors[0])
@@ -412,8 +426,8 @@ def sleep(seconds):
 
 
 class DaylightSaving(unittest.TestCase):
-	"""Melbourne's clocks go back an hour at 03:00 on 5 April 2026 and forward at 02:00 on 4 October 2026. The fire times must be
-	right on any croniter: the distro packages (up to 6.0) got zoneinfo's changes wrong."""
+	"""Melbourne's clocks go back an hour at 03:00 on 5 April 2026 and forward at 02:00 on 4 October 2026. The fire
+	times must be right on any croniter: the distro packages (up to 6.0) got zoneinfo's changes wrong."""
 	MEL = zoneinfo.ZoneInfo('Australia/Melbourne')
 
 	def fires(self, expr, start, end):
@@ -422,7 +436,8 @@ def fires(self, expr, start, end):
 		s = jobs.Scheduler(lambda: None, zone=self.MEL)
 		times, t = [], start
 		while (t := s.next_fire(job, t)) <= end:
-			times.append(t.astimezone(UTC).astimezone(self.MEL).strftime('%d %H:%M %Z')) # by way of UTC: the real local time
+			# by way of UTC: the real local time
+			times.append(t.astimezone(UTC).astimezone(self.MEL).strftime('%d %H:%M %Z'))
 		return times
 
 	def mel(self, *args, **kw):
@@ -431,21 +446,28 @@ def mel(self, *args, **kw):
 
 	def test_a_daily_job_fires_once_at_its_time(self):
 		"""A daily job fires at its local time each day across both changes."""
-		self.assertEqual(self.fires('0 9 * * *', self.mel(2026, 10, 2), self.mel(2026, 10, 6)), ['02 09:00 AEST', '03 09:00 AEST', '04 09:00 AEDT', '05 09:00 AEDT'])
-		self.assertEqual(self.fires('0 9 * * *', self.mel(2026, 4, 3), self.mel(2026, 4, 7)), ['03 09:00 AEDT', '04 09:00 AEDT', '05 09:00 AEST', '06 09:00 AEST'])
+		self.assertEqual(self.fires('0 9 * * *', self.mel(2026, 10, 2), self.mel(2026, 10, 6)),
+			['02 09:00 AEST', '03 09:00 AEST', '04 09:00 AEDT', '05 09:00 AEDT'])
+		self.assertEqual(self.fires('0 9 * * *', self.mel(2026, 4, 3), self.mel(2026, 4, 7)),
+			['03 09:00 AEDT', '04 09:00 AEDT', '05 09:00 AEST', '06 09:00 AEST'])
 
 	def test_a_skipped_or_repeated_time_fires_once(self):
 		"""A time the clocks skip fires an hour later, and a repeated one fires once."""
-		self.assertEqual(self.fires('30 2 * * *', self.mel(2026, 10, 3), self.mel(2026, 10, 6)), ['03 02:30 AEST', '04 03:30 AEDT', '05 02:30 AEDT'])
-		self.assertEqual(self.fires('30 2 * * *', self.mel(2026, 4, 4), self.mel(2026, 4, 7)), ['04 02:30 AEDT', '05 02:30 AEDT', '06 02:30 AEST'])
-		self.assertEqual(self.fires('*/20 * * * *', self.mel(2026, 10, 4, 1, 30), self.mel(2026, 10, 4, 3, 50)), ['04 01:40 AEST', '04 03:00 AEDT', '04 03:20 AEDT', '04 03:40 AEDT'])
+		self.assertEqual(self.fires('30 2 * * *', self.mel(2026, 10, 3), self.mel(2026, 10, 6)),
+			['03 02:30 AEST', '04 03:30 AEDT', '05 02:30 AEDT'])
+		self.assertEqual(self.fires('30 2 * * *', self.mel(2026, 4, 4), self.mel(2026, 4, 7)),
+			['04 02:30 AEDT', '05 02:30 AEDT', '06 02:30 AEST'])
+		self.assertEqual(self.fires('*/20 * * * *', self.mel(2026, 10, 4, 1, 30), self.mel(2026, 10, 4, 3, 50)),
+			['04 01:40 AEST', '04 03:00 AEDT', '04 03:20 AEDT', '04 03:40 AEDT'])
 		self.assertEqual(self.fires('*/20 * * * *', self.mel(2026, 4, 5, 1, 30), self.mel(2026, 4, 5, 3, 30)),
-						 ['05 01:40 AEDT', '05 02:00 AEDT', '05 02:20 AEDT', '05 02:40 AEDT', '05 03:00 AEST', '05 03:20 AEST'])
+						 ['05 01:40 AEDT', '05 02:00 AEDT', '05 02:20 AEDT', '05 02:40 AEDT', '05 03:00 AEST',
+						 	'05 03:20 AEST'])
 
 	def test_starting_in_the_repeated_hour_runs_nothing_twice(self):
 		"""Starting during the repeated hour runs nothing twice."""
 		second_time_round = self.mel(2026, 4, 5, 2, 10, fold=1).astimezone(UTC)
-		self.assertEqual(self.fires('*/20 * * * *', second_time_round, self.mel(2026, 4, 5, 3, 0, 30)), ['05 03:00 AEST'])
+		self.assertEqual(self.fires('*/20 * * * *', second_time_round, self.mel(2026, 4, 5, 3, 0, 30)),
+			['05 03:00 AEST'])
 		self.assertEqual(self.fires('30 2 * * *', second_time_round, self.mel(2026, 4, 6, 12)), ['06 02:30 AEST'])
 
 	def test_the_scheduler_runs_a_daily_job_once_across_the_change(self):
@@ -470,9 +492,10 @@ def test_the_bot_will_not_start_with_an_impossible_time(self):
 		skip_if_old_env(self)
 		with tempfile.TemporaryDirectory() as data:
 			Path(data, 'schedule.cron').write_text('*/20 * * * Mon-Fri trades.py\n0 9 31 6 * price.py close\n')
-			env = {k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k not in config.STARTUP_KEYS}
+			env = bare_environment()
 			env.update(DATA_DIR=data, port='0')
-			out = subprocess.run([sys.executable, 'bot.py'], cwd=ROOT, env=env, capture_output=True, text=True, timeout=120)
+			out = subprocess.run([sys.executable, 'bot.py'], cwd=ROOT, env=env, capture_output=True, text=True,
+				timeout=120)
 		self.assertEqual(out.returncode, 1, out.stderr)
 		self.assertIn(f"ERROR schedule: {data}/schedule.cron line 2: '0 9 31 6 *' never comes", out.stderr)
 
@@ -481,18 +504,20 @@ class ManualRuns(SchedulerBase):
 	def setUp(self):
 		"""A schedule with repeats, comments, braces and the commands that cannot be picked (cleanup, podcasts)."""
 		super().setUp()
-		self.file.write_text('# a comment\n*/20 * * * Mon-Fri trades.py > /dev/null\n28 21 * * Sat price.py close; trades.py # weekly\n'
-							 '5 4 * * * { price.py close ;}\n6 4 * * * price.py 7\n'
-							 '0 3 * * * performance.py 365; performance.py 7; performance.py 90; performance.py 30; performance.py 180\n'
-							 '0 2 * * * cleanup.py\n1 * * * * podcasts.py "https://example.com/feed"\n')
+		self.file.write_text('# a comment\n*/20 * * * Mon-Fri trades.py > /dev/null\n'
+			'28 21 * * Sat price.py close; trades.py # weekly\n5 4 * * * { price.py close ;}\n6 4 * * * price.py 7\n'
+			'0 3 * * * performance.py 365; performance.py 7; performance.py 90; performance.py 30; performance.py 180\n'
+			'0 2 * * * cleanup.py\n1 * * * * podcasts.py "https://example.com/feed"\n')
 		self.s = self.scheduler()
 		self.patch(jobs, 'current', self.s)
 		self.patch(jobs.SCRIPTS['price'], 'run', lambda **k: self.ran.append(('price', k)))
 
 	def test_commands_are_unique_clean_and_in_order(self):
 		"""The commands offered are each once, without redirects, sorted by name with numbers in numeric order."""
-		self.assertEqual([label for label, _, _ in self.s.commands()], ['performance.py 7', 'performance.py 30', 'performance.py 90', 'performance.py 180',
-																	   'performance.py 365', 'price.py 7', 'price.py close', 'trades.py'])
+		self.assertEqual([label for label, _, _ in self.s.commands()],
+			['performance.py 7', 'performance.py 30', 'performance.py 90', 'performance.py 180',
+																	   'performance.py 365', 'price.py 7',
+																	   'price.py close', 'trades.py'])
 
 	def test_select_by_number_or_command_text(self):
 		"""A command is picked by its number or its text, quotes, paths and redirects aside."""
@@ -503,7 +528,8 @@ def test_select_by_number_or_command_text(self):
 
 	def test_select_refuses_what_is_not_offered(self):
 		"""Numbers out of range, cleanup, podcasts, unscheduled commands, shell syntax and two commands are refused."""
-		for text in ('0', '9', 'cleanup.py', 'podcasts.py "https://example.com/feed"', 'price.py 99', 'cal.py earnings', 'nonsense', 'ls | wc', '', 'price.py close; trades.py'):
+		for text in ('0', '9', 'cleanup.py', 'podcasts.py "https://example.com/feed"', 'price.py 99', 'cal.py earnings',
+				'nonsense', 'ls | wc', '', 'price.py close; trades.py'):
 			with self.assertRaises(jobs.ScheduleError, msg=text):
 				self.s.select(text)
 
@@ -570,7 +596,8 @@ def test_list_in_chat_is_just_the_commands(self):
 		""".jobs and aliases number the commands, without times, comments or redirects, and mark one that is running."""
 		for alias in ('.jobs', '.cron', '.crons', '.schedule', '@' + BOT.lstrip('@') + ' jobs'):
 			text = '\n'.join(map(str, self.list_in_chat(alias)))
-			for line in ('1. performance 7', '2. performance 30', '5. performance 365', '6. price 7', '7. price close', '8. trades'):
+			for line in ('1. performance 7', '2. performance 30', '5. performance 365', '6. price 7', '7. price close',
+					'8. trades'):
 				self.assertIn(line, text, alias)
 			for noise in ('*/20', '#', 'Mon-Fri', 'next', 'last', '/dev/null', 'weekly', '.py', 'cleanup', 'podcasts'):
 				self.assertNotIn(noise, text, alias)
@@ -644,8 +671,10 @@ def test_forged_button_data_never_runs_anything_else(self):
 		"""Bad or forged button data runs nothing, refreshes buttons in place; no number picks cleanup or podcasts."""
 		self.stale_buttons()
 		good = jobs.checksum('trades.py')
-		for data in ('j|-1|' + good, 'j|0|' + good, 'j|9|' + good, 'j|8|0000', 'j|8|', 'j|8', 'j|8|' + good + '|x', 'j||' + good,
-					 'j|٨|' + good, 'j| 8|' + good, 'j|8.0|' + good, 'j|trades.py|' + good, 'j|1|' + good, 'j|8|price.py close',
+		for data in ('j|-1|' + good, 'j|0|' + good, 'j|9|' + good, 'j|8|0000', 'j|8|', 'j|8', 'j|8|' + good + '|x',
+			'j||' + good,
+					 'j|٨|' + good, 'j| 8|' + good, 'j|8.0|' + good, 'j|trades.py|' + good, 'j|1|' + good,
+					 'j|8|price.py close',
 					 'j|8|' + good.upper() + 'x', 'j|99999999999999999999|' + good):
 			self.sent.clear()
 			self.toasts.clear(); self.edits.clear()
@@ -728,20 +757,24 @@ def test_every_command_can_be_handed_to_a_child(self):
 	def test_a_real_child_runs_the_handler_and_reports_its_exit_status(self):
 		"""A real child runs the report and exits 0, or REPORTED when it failed and said so."""
 		real = jobs.subprocess.Popen
-		self.patch(jobs.subprocess, 'Popen', lambda args, **kw: real(args, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, **kw))
+		self.patch(jobs.subprocess, 'Popen',
+			lambda args, **kw: real(args, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, **kw))
 		with tempfile.TemporaryDirectory() as data:
 			env = {'DATA_DIR': data, 'telegramBotToken': '', 'slack_webhook': '', 'discord_webhook': ''}
 			with mock.patch.dict(os.environ, env):
 				self.assertEqual(jobs.run_child('cleanup.py', cleanup, {}), 0)
-				self.assertEqual(jobs.run_child('cleanup.py 0', cleanup, {'days': 'not a number'}), jobs.jobrun.REPORTED)
+				self.assertEqual(jobs.run_child('cleanup.py 0', cleanup, {'days': 'not a number'}),
+					jobs.jobrun.REPORTED)
 
 	def test_a_job_process_survives_a_reload_signal(self):
 		"""A job process ignores SIGHUP, which systemctl kill -s HUP sends to every process of the unit."""
 		with tempfile.TemporaryDirectory() as scripts, tempfile.TemporaryDirectory() as data:
 			Path(scripts, 'sleeper.py').write_text(
 				'import time\ndef run():\n    print("started", flush=True)\n    time.sleep(1)\n')
-			env = {**os.environ, 'PYTHONPATH': scripts, 'DATA_DIR': data, 'telegramBotToken': '', 'slack_webhook': '', 'discord_webhook': ''}
-			proc = subprocess.Popen([sys.executable, '-m', 'lib.jobrun', 'sleeper', '{}'], cwd=ROOT, env=env, stdout=subprocess.PIPE, text=True)
+			env = {**os.environ, 'PYTHONPATH': scripts, 'DATA_DIR': data, 'telegramBotToken': '', 'slack_webhook': '',
+				'discord_webhook': ''}
+			proc = subprocess.Popen([sys.executable, '-m', 'lib.jobrun', 'sleeper', '{}'], cwd=ROOT, env=env,
+				stdout=subprocess.PIPE, text=True)
 			self.addCleanup(proc.kill)
 			for line in proc.stdout:
 				if line.strip() == 'started':
@@ -753,7 +786,8 @@ def test_a_script_that_cannot_be_imported_is_reported(self):
 		"""A report that cannot be imported is reported like any other failure."""
 		with tempfile.TemporaryDirectory() as data:
 			env = {**os.environ, 'DATA_DIR': data, 'telegramBotToken': '', 'slack_webhook': '', 'discord_webhook': ''}
-			out = subprocess.run([sys.executable, '-m', 'lib.jobrun', 'nosuchscript', '{}'], cwd=ROOT, env=env, capture_output=True, text=True)
+			out = subprocess.run([sys.executable, '-m', 'lib.jobrun', 'nosuchscript', '{}'], cwd=ROOT, env=env,
+				capture_output=True, text=True)
 		self.assertEqual(out.returncode, jobs.jobrun.REPORTED, out.stderr)
 		self.assertIn('⚠️ nosuchscript.py: ModuleNotFoundError', out.stderr)
 
@@ -762,7 +796,8 @@ def test_scheduler_reports_a_child_that_ended_without_saying_why(self):
 		reported itself is not reported twice."""
 		file = Path(temp_dir(self), 'schedule.cron')
 		file.write_text('* * * * * trades.py\n')
-		for status, reason in ((-signal.SIGHUP, 'killed by SIGHUP'), (-signal.SIGKILL, 'killed by SIGKILL'), (1, 'exited with status 1 without saying why'),
+		for status, reason in ((-signal.SIGHUP, 'killed by SIGHUP'), (-signal.SIGKILL, 'killed by SIGKILL'),
+			(1, 'exited with status 1 without saying why'),
 							   (jobs.jobrun.REPORTED, None)): # the child reported this one itself
 			with self.subTest(status=status):
 				self.errors.clear()
@@ -774,9 +809,11 @@ def test_scheduler_reports_a_child_that_ended_without_saying_why(self):
 
 	def test_a_report_script_loads_neither_pandas_nor_the_scheduler(self):
 		"""A job imports only its script; price once pulled in pandas and every script (via charts, reports, jobs)."""
-		code = 'import sys; from lib.reports import price, performance, trades, shorts, cal; print(sorted(m for m in ("pandas", "lib.jobs", "matplotlib") if m in sys.modules))'
+		code = ('import sys; from lib.reports import price, performance, trades, shorts, cal; '
+			'print(sorted(m for m in ("pandas", "lib.jobs", "matplotlib") if m in sys.modules))')
 		with tempfile.TemporaryDirectory() as data:
-			out = subprocess.run([sys.executable, '-c', code], cwd=ROOT, capture_output=True, text=True, env={**os.environ, 'DATA_DIR': data})
+			out = subprocess.run([sys.executable, '-c', code], cwd=ROOT, capture_output=True, text=True,
+				env={**os.environ, 'DATA_DIR': data})
 		self.assertEqual(out.stdout.strip().splitlines()[-1:], ['[]'], out.stderr)
 
 	def test_the_event_loop_keeps_running_while_a_child_works(self):
diff --git a/tests/test_providers.py b/tests/test_providers.py
index acfbf490..aa492a59 100644
--- a/tests/test_providers.py
+++ b/tests/test_providers.py
@@ -148,7 +148,8 @@ class YahooDetail(ProviderCase):
 	def combined():
 		"""quoteSummary as Yahoo returns it for several modules at once, assembled from the per-module samples."""
 		merged = {}
-		for module in ('price', 'summaryProfile', 'summaryDetail', 'defaultKeyStatistics', 'financialData', 'calendarEvents', 'earnings'):
+		for module in ('price', 'summaryProfile', 'summaryDetail', 'defaultKeyStatistics', 'financialData',
+				'calendarEvents', 'earnings'):
 			merged.update(sample('yahoo', f'sample_yahoo_v10_{module}.json')['quoteSummary']['result'][0])
 		return {'quoteSummary': {'result': [merged], 'error': None}}
 
@@ -256,16 +257,19 @@ def chart(closes, failing=()):
 		"""A v8 chart response with one daily bar per close, the last one today (14:30 UTC, New York's open)."""
 		today = datetime.datetime.now(datetime.timezone.utc).replace(hour=14, minute=30, second=0, microsecond=0)
 		stamps = [int((today - datetime.timedelta(days=len(closes) - 1 - i)).timestamp()) for i in range(len(closes))]
-		meta = {'currency': 'USD', 'symbol': 'T1', 'exchangeTimezoneName': 'America/New_York', 'regularMarketTime': stamps[-1],
+		meta = {'currency': 'USD', 'symbol': 'T1', 'exchangeTimezoneName': 'America/New_York',
+			'regularMarketTime': stamps[-1],
 				'regularMarketPrice': closes[-1], 'longName': 'Co T1'}
 		quote = {'open': closes, 'high': closes, 'low': closes, 'close': closes, 'volume': [1] * len(closes)}
-		return {'chart': {'result': [{'meta': meta, 'timestamp': stamps, 'indicators': {'quote': [quote]}}], 'error': None}}
+		return {'chart': {'result': [{'meta': meta, 'timestamp': stamps, 'indicators': {'quote': [quote]}}],
+			'error': None}}
 
 	def test_1d_is_the_change_since_the_previous_close(self):
 		"""1D is the change since the previous close, and 5D the week's; with a quote, its price and today's change
 		count, the chart's being up to cache_seconds old."""
 		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
-		self.http(yahoo, Resp(self.chart([90] * 30 + [100, 105, 110, 115, 120, 116.4]))) # up 20% in a week, down 3% today
+		# up 20% in a week, down 3% today
+		self.http(yahoo, Resp(self.chart([90] * 30 + [100, 105, 110, 115, 120, 116.4])))
 		table, line = yahoo.price_history('T1')
 		self.assertEqual(table['1D'], -3)
 		self.assertGreater(table['5D'], 10) # the week, not the day
@@ -340,7 +344,8 @@ def slack(text):
 	def test_the_chart_request_carries_the_cookie_of_its_crumb(self):
 		"""A chart request sends the cookie its crumb belongs to."""
 		headers = []
-		self.patch(yahoo.net, 'get', lambda url, **kwargs: headers.append(kwargs.get('headers', {})) or Resp(self.chart([100.0] * 5)))
+		self.patch(yahoo.net, 'get',
+			lambda url, **kwargs: headers.append(kwargs.get('headers', {})) or Resp(self.chart([100.0] * 5)))
 		yahoo.fetch_chart_json('T1', full=True)
 		self.assertEqual(headers[0].get('Cookie'), 'A=B') # the crumb used to go without its cookie
 
@@ -387,7 +392,8 @@ def test_portfolio_exclusion_and_inclusion(self):
 
 	def test_portfolio_failures_raise(self):
 		"""An error answer, Sharesight unreachable or no portfolios is a RuntimeError."""
-		for failure in (Resp({'error': 'nope', 'error_code': 'x'}, 401), ConnectionError('down'), Resp({'portfolios': []})):
+		for failure in (Resp({'error': 'nope', 'error_code': 'x'}, 401), ConnectionError('down'),
+				Resp({'portfolios': []})):
 			with self.subTest(failure=failure):
 				self.http(sharesight, failure)
 				with self.assertRaises(RuntimeError):
@@ -456,7 +462,8 @@ def test_performance_and_holdings_from_real_response(self):
 	def test_performance_retries_when_sharesight_is_busy(self):
 		"""While Sharesight says too many requests are in flight, the request is tried again."""
 		busy = Resp({'error': 'Too many parallel requests', 'error_code': 'x'}, 429)
-		getter = self.http(sharesight, busy, busy, Resp(sample('sharesight', 'finbot_sharesight_performance_684141_0.json')))
+		getter = self.http(sharesight, busy, busy,
+			Resp(sample('sharesight', 'finbot_sharesight_performance_684141_0.json')))
 		self.assertIn('holdings', sharesight._performance(684141, 0))
 		self.assertEqual(len(getter.calls), 3)
 
@@ -504,7 +511,8 @@ def test_cold_cache_is_filled_then_left_alone(self):
 		sharesight.warm()
 		self.assertEqual(len(calls), 3) # portfolio list + one holdings report per portfolio (2)
 		self.assertEqual(sum('/performance' in c for c in calls), 2)
-		self.assertTrue(all('start_date=' + datetime.date.today().isoformat() in c for c in calls if '/performance' in c)) # days=0
+		today = 'start_date=' + datetime.date.today().isoformat()
+		self.assertTrue(all(today in c for c in calls if '/performance' in c)) # days=0
 		sharesight.warm()
 		self.assertEqual(len(calls), 3) # nothing refetched while fresh
 
@@ -513,7 +521,8 @@ def test_refreshes_before_the_ttl_runs_out(self):
 		calls = self.serve_sharesight()
 		sharesight.warm()
 		calls.clear()
-		self.age(sharesight.performance_cache_file(684141, 0), 0.95 * config.prefs.performance_cache_seconds) # 95% of its lifetime
+		# 95% of its lifetime
+		self.age(sharesight.performance_cache_file(684141, 0), 0.95 * config.prefs.performance_cache_seconds)
 		sharesight.warm()
 		self.assertEqual(len(calls), 1)
 		self.assertIn('684141/performance', calls[0])
@@ -683,7 +692,8 @@ def test_error_reports_are_escaped_for_telegram_and_slack(self):
 		self.patch(transports.Transport, 'text', REAL_TEXT)
 		for kind in (transports.TelegramBot, transports.SlackWebhook, transports.DiscordWebhook):
 			self.patch(kind, 'post_text', lambda dest, text, buttons=None: sent.append((dest.name, text)))
-		self.env(webhooks={'telegram': 'https://api.telegram.org/botX/', 'slack': 'https://hooks.slack.com/x', 'discord': 'https://discord.com/x/slack'})
+		self.env(webhooks={'telegram': 'https://api.telegram.org/botX/', 'slack': 'https://hooks.slack.com/x',
+			'discord': 'https://discord.com/x/slack'})
 		self.env(telegramChatID='5')
 
 		error = RuntimeError("404 for ZZ at https://x/?a=1&crumb=abc; '<' not supported between <urllib3.connection>")
@@ -739,7 +749,8 @@ def post(url, **kwargs):
 
 	def upload_steps(self):
 		"""Slack's answers to the three upload steps, each a success."""
-		return [Resp({'ok': True, 'upload_url': self.UPLOAD_URL, 'file_id': 'F1'}), Resp(None, text='OK - 3'), Resp({'ok': True, 'files': [{'id': 'F1'}]})]
+		return [Resp({'ok': True, 'upload_url': self.UPLOAD_URL, 'file_id': 'F1'}), Resp(None, text='OK - 3'),
+			Resp({'ok': True, 'files': [{'id': 'F1'}]})]
 
 	def test_where_charts_can_go(self):
 		"""Charts always go to Telegram and Discord; Slack gets them as the bot, with the bot token, but not through its
@@ -776,15 +787,20 @@ def test_slack_replies_upload_their_charts(self):
 				self.sent.clear(); self.errors.clear()
 				worker.process_request(inbound.Message('slack', 'C1', '<@U1>', cmd, BOT, 'U'))
 				self.assertEqual(self.errors, [])
-				self.assertEqual([url for url, _ in calls], ['https://slack.com/api/files.getUploadURLExternal', self.UPLOAD_URL,
+				self.assertEqual([url for url, _ in calls],
+					['https://slack.com/api/files.getUploadURLExternal', self.UPLOAD_URL,
 															 'https://slack.com/api/files.completeUploadExternal'])
 				(_, ticket), (_, upload), (_, share) = calls
 				self.assertEqual(ticket['headers'], {'Authorization': 'Bearer xoxb-test'})
 				self.assertTrue(upload['data'].startswith(b'\x89PNG')) # the chart's bytes, as the body
-				self.assertEqual((ticket['data']['filename'], ticket['data']['length']), ('chart.png', len(upload['data'])))
-				self.assertEqual((share['data']['channel_id'], json.loads(share['data']['files'])), ('C1', [{'id': 'F1'}]))
-				self.assertTrue(share['data']['initial_comment'].startswith(heading), share['data']['initial_comment'][:80]) # the report
-				self.assertNotIn(share['data']['initial_comment'].split('\n'), [payload for _, payload in self.sent]) # not sent again as text
+				self.assertEqual((ticket['data']['filename'], ticket['data']['length']),
+					('chart.png', len(upload['data'])))
+				self.assertEqual((share['data']['channel_id'], json.loads(share['data']['files'])),
+					('C1', [{'id': 'F1'}]))
+				self.assertTrue(share['data']['initial_comment'].startswith(heading),
+					share['data']['initial_comment'][:80]) # the report
+				# not sent again as text
+				self.assertNotIn(share['data']['initial_comment'].split('\n'), [payload for _, payload in self.sent])
 
 	def test_a_slack_upload_that_fails_goes_out_as_text(self):
 		"""A Slack upload failing at any step goes as text, but not on a last-step timeout, when it may have arrived."""
@@ -794,7 +810,8 @@ def test_a_slack_upload_that_fails_goes_out_as_text(self):
 		good = self.upload_steps()
 		slow = transports.requests.exceptions.ReadTimeout('slow')
 		text = [('text', ['*Heading*', 'line 2'])]
-		for outcomes, expect in (([Resp({'ok': False, 'error': 'missing_scope'})], text), ([ConnectionError('down')], text),
+		for outcomes, expect in (([Resp({'ok': False, 'error': 'missing_scope'})], text),
+			([ConnectionError('down')], text),
 								 ([good[0], Resp(None, status=500, text='nope')], text),
 								 ([good[0], good[1], Resp({'ok': False, 'error': 'not_in_channel'})], text),
 								 ([good[0], good[1], slow], []), # it may have been shared after all
@@ -811,7 +828,8 @@ def test_scheduled_reports_to_slack_carry_charts_only_as_the_bot(self):
 		their charts."""
 		self.env(slackBotToken='xoxb-test')
 
-		self.env(webhooks={**config.env.webhooks, 'slack': 'https://hooks.slack.com/services/x'}) # an incoming webhook cannot upload
+		# an incoming webhook cannot upload
+		self.env(webhooks={**config.env.webhooks, 'slack': 'https://hooks.slack.com/services/x'})
 		sent = []
 		self.patch(transports.Transport, 'chart', lambda dest, image, caption, buttons=None:
 			sent.append((dest.name, 'chart')) or {'ok': True})
@@ -837,7 +855,8 @@ def refuse(*a, **k):
 			raise AssertionError('a chart was drawn for Slack')
 		self.patch(render, 'draw', refuse) # every chart is drawn there
 		self.env(webhooks={**config.env.webhooks, 'slack': 'https://hooks.slack.com/services/x'})
-		for cmd in ('.beta', '.pie Rob', '.buy', '.sell', '.peg', '.price', '.price top', '.price T1 7d', '.performance 7d', '.shorts', '.history T1', '.compare T1 T2'):
+		for cmd in ('.beta', '.pie Rob', '.buy', '.sell', '.peg', '.price', '.price top', '.price T1 7d',
+				'.performance 7d', '.shorts', '.history T1', '.compare T1 T2'):
 			with self.subTest(cmd=cmd):
 				self.sent.clear(); self.errors.clear()
 				worker.process_request(inbound.Message('slack', 'C1', '<@U1>', cmd, BOT, 'U'))
@@ -854,7 +873,8 @@ def test_the_graph_preference_off_sends_no_chart_anywhere(self):
 		posts = []
 		self.patch(transports.net, 'post', lambda url, **k: posts.append(url) or Resp({'ok': True, 'result': True}))
 		for service, chat in (('telegram', '55'), ('slack', 'C1')):
-			for cmd in ('.price', '.price top', '.performance 7d', '.shorts', '.buy', '.peg', '.beta', '.compare T1 T2', '.pie Rob'):
+			for cmd in ('.price', '.price top', '.performance 7d', '.shorts', '.buy', '.peg', '.beta', '.compare T1 T2',
+					'.pie Rob'):
 				with self.subTest(service=service, cmd=cmd):
 					self.sent.clear(); self.errors.clear()
 					worker.process_request(inbound.Message(service, chat, '@u', cmd, BOT, 'U'))
@@ -874,7 +894,8 @@ def fake(url, **kwargs):
 				return outcome
 			return fake
 		text = [('text', ['Heading', 'line 2'])]
-		for outcome, expect in ((Resp({'ok': False, 'error_code': 400, 'description': "Bad Request: can't parse entities"}, status=400), text),
+		for outcome, expect in ((Resp({'ok': False, 'error_code': 400,
+			'description': "Bad Request: can't parse entities"}, status=400), text),
 								(Resp(None, status=502), text), (ConnectionError('down'), text),
 								(transports.requests.exceptions.ReadTimeout('slow'), []), # it may be there
 								(Resp({'ok': True, 'result': {'message_id': 1}}), [])):
@@ -887,7 +908,7 @@ def fake(url, **kwargs):
 
 
 class DiscordWebhook(FinbotCase):
-	"""Discord's own webhook endpoint (the Slack-compatible one takes no flags): no link previews, no pings, charts attached."""
+	"""Discord's own webhook endpoint (the Slack-like one takes no flags): no previews, no pings, charts attached."""
 	URL = 'https://discord.com/api/webhooks/1/abc'
 
 	def setUp(self):
@@ -912,7 +933,8 @@ def test_a_scheduled_report_carries_its_chart(self):
 		(url, kwargs), = self.posts
 		self.assertEqual(url, self.URL + '?wait=true')
 		payload = json.loads(kwargs['data']['payload_json'])
-		self.assertEqual((payload['flags'], payload['allowed_mentions'], payload['attachments'][0]['filename']), (4, {'parse': []}, 'chart.png'))
+		self.assertEqual((payload['flags'], payload['allowed_mentions'], payload['attachments'][0]['filename']),
+			(4, {'parse': []}, 'chart.png'))
 		self.assertTrue(payload['content'].startswith('**≥ 5% at close'), payload['content'][:60])
 		self.assertLessEqual(len(payload['content']), 2000)
 		name, image, kind = kwargs['files']['files[0]']
@@ -973,7 +995,8 @@ def test_a_reply_without_a_bot_token_is_logged_not_sent(self):
 			for cmd in ('.help', '.price', '.compare T1 T2'):
 				worker.process_request(inbound.Message('slack', 'C1', '<@U1>', cmd, BOT, 'U'))
 		self.assertEqual((self.posts, self.errors), ([], []))
-		self.assertEqual(err.getvalue().count('ERROR slack: a reply to channel C1 needs slackBotToken in finbot.env; not sent'), 3)
+		refusal = 'ERROR slack: a reply to channel C1 needs slackBotToken in finbot.env; not sent'
+		self.assertEqual(err.getvalue().count(refusal), 3)
 
 	def test_a_command_at_work_shows_an_hourglass_on_its_message(self):
 		"""While a command works, ⏳ is on the message it answers (Slack has no typing notice for bots); it comes off
@@ -1016,7 +1039,8 @@ def test_the_slack_api_says_what_is_missing(self):
 		"""slack_api without the bot token says so (it used to be a TypeError)."""
 		self.env(slackBotToken=False)
 
-		with self.assertRaisesRegex(RuntimeError, 'Slack files.getUploadURLExternal: needs slackBotToken in finbot.env'):
+		with self.assertRaisesRegex(RuntimeError,
+				'Slack files.getUploadURLExternal: needs slackBotToken in finbot.env'):
 			transports.slack_api('files.getUploadURLExternal', filename='chart.png', length=1) # it was a TypeError
 		self.assertEqual(self.posts, [])
 
@@ -1152,7 +1176,7 @@ def post(url, **kwargs):
 
 
 class TelegramWebhook(FinbotCase):
-	"""bot.py registers the webhook once as it starts. Importing lib.telegram, as every job process does, touches nothing."""
+	"""bot.py registers the webhook once as it starts; importing lib.telegram, as each job does, does nothing."""
 
 	def setUp(self):
 		"""A Telegram bot with an outgoing webhook configured."""
@@ -1175,14 +1199,15 @@ def post(url, **kwargs):
 
 	def test_a_refused_call_raises(self):
 		"""A call Telegram refuses raises with its description."""
-		self.patch(telegram.net, 'post', lambda url, **k: Resp({'ok': False, 'description': 'Bad Request: bad webhook: HTTPS url must be provided'}))
+		self.patch(telegram.net, 'post',
+			lambda url, **k: Resp({'ok': False, 'description': 'Bad Request: bad webhook: HTTPS url must be provided'}))
 		with contextlib.redirect_stderr(io.StringIO()), self.assertRaises(RuntimeError) as caught:
 			telegram.register()
 		self.assertIn('HTTPS url must be provided', str(caught.exception))
 
 	def test_edits_toasts_and_typing_go_through_call(self):
-		"""A chart edit that Telegram refuses is an error (unless nothing changed); a toast, a text edit or a typing notice that
-		fails, or an edit lost on the way, is only logged."""
+		"""A chart edit that Telegram refuses is an error (unless nothing changed); a toast, a text edit or a typing
+		notice that fails, or an edit lost on the way, is only logged."""
 		def answer(outcome):
 			"""A net.post that answers with outcome, or raises it."""
 			def post(url, **kwargs):
@@ -1192,9 +1217,11 @@ def post(url, **kwargs):
 				return outcome
 			return post
 		refused = Resp({'ok': False, 'description': 'Bad Request: message to edit not found'})
-		unchanged = Resp({'ok': False, 'description': 'Bad Request: message is not modified: specified new message content is the same'})
+		unchanged = Resp({'ok': False,
+			'description': 'Bad Request: message is not modified: specified new message content is the same'})
 		chat = transports.TelegramBot('5')
-		for outcome, raises in ((refused, True), (unchanged, False), (ConnectionError('down'), False), (Resp({'ok': True, 'result': True}), False)):
+		for outcome, raises in ((refused, True), (unchanged, False), (ConnectionError('down'), False),
+				(Resp({'ok': True, 'result': True}), False)):
 			with self.subTest(outcome=outcome):
 				self.patch(telegram.net, 'post', answer(outcome))
 				with contextlib.redirect_stderr(io.StringIO()):
@@ -1216,18 +1243,22 @@ def test_importing_never_calls_telegram(self):
 				'print("quiet")')
 		with tempfile.TemporaryDirectory() as data:
 			out = subprocess.run([sys.executable, '-c', code], cwd=ROOT, capture_output=True, text=True,
-				env={**os.environ, 'DATA_DIR': data, 'telegramBotToken': '123:abc', 'telegramOutgoingWebhook': 'https://example.com/telegram'})
+				env={**os.environ, 'DATA_DIR': data, 'telegramBotToken': '123:abc',
+					'telegramOutgoingWebhook': 'https://example.com/telegram'})
 		self.assertEqual(out.stdout.split()[-1:], ['quiet'], out.stderr)
 
 	def test_the_bot_will_not_start_unregistered_and_never_logs_the_token(self):
 		"""bot.py exits when the webhook cannot be registered, and never logs the token."""
 		skip_if_old_env(self)
 		with tempfile.TemporaryDirectory() as data:
-			env = {k: v for k, v in os.environ.items() if k not in config.PREFERENCE_KEYS and k not in config.STARTUP_KEYS and 'proxy' not in k.lower()}
-			env.update(DATA_DIR=data, port='0', scheduler='False', telegramBotToken='123456:SECRET-token', telegramOutgoingToken='tok',
+			env = {k: v for k, v in bare_environment().items() if 'proxy' not in k.lower()}
+			env.update(DATA_DIR=data, port='0', scheduler='False', telegramBotToken='123456:SECRET-token',
+				telegramOutgoingToken='tok',
 					   telegramOutgoingWebhook='https://example.com/telegram',
-					   HTTPS_PROXY='http://127.0.0.1:9', https_proxy='http://127.0.0.1:9') # unreachable, so the registration fails at once and offline
-			out = subprocess.run([sys.executable, 'bot.py'], cwd=ROOT, env=env, capture_output=True, text=True, timeout=120)
+					   # unreachable, so the registration fails at once and offline
+					   HTTPS_PROXY='http://127.0.0.1:9', https_proxy='http://127.0.0.1:9')
+			out = subprocess.run([sys.executable, 'bot.py'], cwd=ROOT, env=env, capture_output=True, text=True,
+				timeout=120)
 		self.assertEqual(out.returncode, 1, out.stderr)
 		self.assertIn('ERROR ⚠️ telegram webhook:', out.stderr)
 		self.assertNotIn('SECRET-token', out.stderr + out.stdout)
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 7e6fa191..1ad02325 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -17,7 +17,8 @@ def test_text_commands(self):
 	def test_chart_commands(self):
 		"""Every chart command answers with a chart."""
 		for cmd in ('.price', '.price 7d', '.price top', '.price bottom', '.session', '.premarket', '.postmarket',
-					'.performance', '.performance 7d', '.performance 1y', '.shorts', '.peg', '.buy', '.sell', '.beta', '.price top 7d',
+					'.performance', '.performance 7d', '.performance 1y', '.shorts', '.peg', '.buy', '.sell', '.beta',
+					'.price top 7d',
 					'.forward pe', '.fpe bottom'):
 			with self.subTest(cmd=cmd):
 				out = self.command(cmd)
@@ -107,12 +108,17 @@ def test_prepare_rating(self):
 			self.assertTrue(rankings.prepare_rating_payload(TELEGRAM, action, length=5)[0])
 
 	def test_ratings_list_the_strongest_first(self):
-		"""1 is a strong buy and 5 a strong sell: .sell used to list the mildest sells first. On equal ratings, more analysts first;
-		fewer than 7 analysts and the stock is left out (T7, T8: the ASX small caps that are always a perfect 1.0)."""
-		ratings = {'T1': ('buy', 1.5, 7), 'T2': ('strong_buy', 1.2, 12), 'T3': ('buy', 1.5, 9), 'T7': ('strong_buy', 1.0, 6),
-				   'T4': ('sell', 4.2, 7), 'T5': ('underperform', 4.6, 10), 'T6': ('sell', 4.2, 8), 'T8': ('strong_buy', 1.0, 2)}
+		"""1 is a strong buy and 5 a strong sell: .sell used to list the mildest sells first. On equal ratings, more
+		analysts first; fewer than 7 analysts and the stock is left out (T7, T8: the ASX small caps that are always a
+		perfect 1.0)."""
+		ratings = {'T1': ('buy', 1.5, 7), 'T2': ('strong_buy', 1.2, 12), 'T3': ('buy', 1.5, 9),
+			'T7': ('strong_buy', 1.0, 6),
+				   'T4': ('sell', 4.2, 7), 'T5': ('underperform', 4.6, 10), 'T6': ('sell', 4.2, 8),
+				   'T8': ('strong_buy', 1.0, 2)}
 		self.patch(watchlist, 'with_holdings', lambda: list(ratings))
-		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | dict(zip(('recommend', 'recommend_index', 'recommend_analysts'), ratings[t]))})
+		self.patch(yahoo, 'fetch_detail',
+			lambda t, *a, **k: {t: market_entry(t) | dict(zip(('recommend', 'recommend_index', 'recommend_analysts'),
+			ratings[t]))})
 		def order(action):
 			"""(the tickers listed, the tickers charted) of a ratings report, in order."""
 			payload, chart = rankings.prepare_rating_payload(TELEGRAM, action)
@@ -121,11 +127,14 @@ def order(action):
 		self.assertEqual(order('sell'), (['T5', 'T6', 'T4'], ['T5', 'T6', 'T4']))
 		self.assertIn('(7+ analysts)', rankings.prepare_rating_payload(TELEGRAM, 'buy')[0][0])
 		self.patch(watchlist, 'with_holdings', lambda: ['T7', 'T8'])
-		self.assertEqual(rankings.prepare_rating_payload(TELEGRAM, 'buy'), (['No tracked stocks rated buy by 7+ analysts'], None))
+		self.assertEqual(rankings.prepare_rating_payload(TELEGRAM, 'buy'),
+			(['No tracked stocks rated buy by 7+ analysts'], None))
 
 	def test_a_small_cap_quoted_in_won_needs_the_bigger_move_too(self):
-		"""demote_volatile compares market caps in US dollars: unconverted, a small cap quoted in won was never small."""
-		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | ({'regularMarketPrice': 1380.0} if t == 'KRW=X' else {'currency': 'KRW', 'market_cap': 1e11}) for t in ts})
+		"""demote_volatile compares market caps in US dollars: unconverted, a small cap in won was never small."""
+		self.patch(yahoo, 'fetch',
+			lambda ts: {t: market_entry(t) | ({'regularMarketPrice': 1380.0} if t == 'KRW=X' else {'currency': 'KRW',
+			'market_cap': 1e11}) for t in ts})
 		self.prefs(demote_volatile=True)
 
 		quote = price.gather()[1]['T1']
@@ -146,9 +155,10 @@ def test_a_short_top_list_repeats_none_of_its_best_as_the_worst(self):
 				self.assertEqual(chart.subtitle, '% change over the past 3 months') # it said "in 3 months"
 
 	def test_the_small_caps_betas_are_fetched_together(self):
-		"""A scheduled price report asks for the beta of each small cap that moved (US$150M to US$1B): together, not one by one."""
+		"""A scheduled price report asks for the betas of the small caps that moved (US$150M to US$1B) together."""
 		calls = []
-		self.patch(yahoo, 'fetch_details', lambda tickers, strict=False: calls.append(sorted(tickers)) or {t: market_entry(t) for t in tickers})
+		self.patch(yahoo, 'fetch_details',
+			lambda tickers, strict=False: calls.append(sorted(tickers)) or {t: market_entry(t) for t in tickers})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: self.fail('a beta fetched on its own'))
 		self.prefs(demote_volatile=True)
 
@@ -158,10 +168,11 @@ def test_the_small_caps_betas_are_fetched_together(self):
 		self.assertTrue(calls[0])
 
 	def test_forward_pe_lists_need_as_many_analysts_as_buy(self):
-		"""A forward P/E is analysts' estimate: with few of them it says little, as a perfect 1.0 buy from two brokers does."""
+		"""A forward P/E is analysts' estimate: from few, it says little, as a perfect 1.0 buy from two brokers does."""
 		analysts = {'T1': 7, 'T2': 6, 'T3': 12, 'T4': 2}
 		self.patch(watchlist, 'with_holdings', lambda: list(analysts))
-		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) | {'price_to_earnings_forward': -5 if t == 'T3' else 20} for t in ts})
+		self.patch(yahoo, 'fetch',
+			lambda ts: {t: market_entry(t) | {'price_to_earnings_forward': -5 if t == 'T3' else 20} for t in ts})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'recommend_analysts': analysts[t]}})
 		def listed(mode='top'):
 			"""(heading, the tickers listed, the chart) of a forward P/E list."""
@@ -175,22 +186,26 @@ def listed(mode='top'):
 		self.assertIn('Co T2', rankings.ratio_line(TELEGRAM, 'T2', 'forward pe')) # one ticker asked for: shown
 		self.prefs(min_analysts=13) # the one setting moves .buy and .forward pe together
 		self.assertEqual(listed()[0], 'No tracked stocks with 13+ analysts have a forward pe ratio')
-		self.assertEqual(rankings.prepare_rating_payload(TELEGRAM, 'buy')[0], ['No tracked stocks rated buy by 13+ analysts'])
+		self.assertEqual(rankings.prepare_rating_payload(TELEGRAM, 'buy')[0],
+			['No tracked stocks rated buy by 13+ analysts'])
 
 	def test_market_caps_compare_in_us_dollars(self):
 		"""Yahoo gives a market cap in the stock's own currency: unconverted, a won market cap topped every list."""
-		stocks = {'K1.KS': ('KRW', 1.01e15), 'K2.KS': ('KRW', 5e11), 'U1': ('USD', 5.39e12), 'L1.L': ('GBp', 6.25e10), 'Z1': ('XYZ', 1e12)}
+		stocks = {'K1.KS': ('KRW', 1.01e15), 'K2.KS': ('KRW', 5e11), 'U1': ('USD', 5.39e12), 'L1.L': ('GBp', 6.25e10),
+			'Z1': ('XYZ', 1e12)}
 		fx = {'KRW=X': 1380.0, 'GBP=X': 0.79} # units per US dollar; nothing for XYZ
 		def fetch(tickers):
 			"""Quotes in each stock's own currency, and the exchange rates."""
-			return {t: market_entry(t) | {'regularMarketPrice': fx[t]} if t in fx else market_entry(t) | {'currency': stocks[t][0], 'market_cap': stocks[t][1], 'beta': 2.0}
-					for t in tickers if t in fx or t in stocks}
+			own = lambda t: {'currency': stocks[t][0], 'market_cap': stocks[t][1], 'beta': 2.0}
+			return {t: market_entry(t) | ({'regularMarketPrice': fx[t]} if t in fx else own(t))
+				for t in tickers if t in fx or t in stocks}
 		self.patch(yahoo, 'fetch', fetch)
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: fetch([t]))
 		self.patch(watchlist, 'with_holdings', lambda: list(stocks))
 		payload = rankings.marketcap_list(TELEGRAM)
 		self.assertIn('by market cap (USD)', payload[0])
-		self.assertEqual([line.split('Co ')[1].split(' ')[0] for line in payload[1:]], ['U1', 'K1.KS', 'L1.L', 'K2.KS']) # Z1: no rate, left out
+		# Z1: no rate, left out
+		self.assertEqual([line.split('Co ')[1].split(' ')[0] for line in payload[1:]], ['U1', 'K1.KS', 'L1.L', 'K2.KS'])
 		self.assertIn('mkt cap: 731.88 B', payload[2]) # 1.01 Q won
 		self.assertIn('mkt cap: KRW 1.01 Q (USD 731.88 B)', rankings.marketcap_line(TELEGRAM, 'K1.KS'))
 		self.assertIn('mkt cap: GBP 62.50 B (USD 79.11 B)', rankings.marketcap_line(TELEGRAM, 'L1.L')) # pence: pounds
@@ -229,7 +244,8 @@ def test_prepare_beta(self):
 		self.assertIn('Beta over 1.5', payload[0])
 		self.assertEqual(len(chart.rows), len(payload) - 1) # a bar per stock listed
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'beta': 1.0}})
-		self.assertEqual(rankings.prepare_beta_payload(TELEGRAM), (['No tracked stocks have a beta over 1.5 and mkt cap under US$2B'], None))
+		self.assertEqual(rankings.prepare_beta_payload(TELEGRAM),
+			(['No tracked stocks have a beta over 1.5 and mkt cap under US$2B'], None))
 
 	def test_prepare_profile(self):
 		"""A company profile is built."""
@@ -237,11 +253,15 @@ def test_prepare_profile(self):
 
 	def test_a_profile_leaves_out_what_yahoo_did_not_send(self):
 		"""Each of these fields missing (or a zero equity) used to crash the whole profile."""
-		full = market_entry('T1') | {'insiderBuy': 1, 'insiderSell': 0, 'insiderBuyValue': 5e6, 'insiderSellValue': 0, 'netIncomeToCommon': 1e6,
-									 'sharesOutstanding': 1e6, 'shareholder_equity': 1e9, 'total_debt': 9e8, 'total_cash': 1e8, 'profile_industry': 'Tools'}
-		for missing in ('marketState', 'profile_exchange', 'market_cap', 'currency', 'regularMarketPrice', 'sharesOutstanding', 'percent_change'):
+		full = market_entry('T1') | {'insiderBuy': 1, 'insiderSell': 0, 'insiderBuyValue': 5e6, 'insiderSellValue': 0,
+			'netIncomeToCommon': 1e6,
+									 'sharesOutstanding': 1e6, 'shareholder_equity': 1e9, 'total_debt': 9e8,
+									 'total_cash': 1e8, 'profile_industry': 'Tools'}
+		for missing in ('marketState', 'profile_exchange', 'market_cap', 'currency', 'regularMarketPrice',
+				'sharesOutstanding', 'percent_change'):
 			with self.subTest(missing=missing):
-				self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: {k: v for k, v in full.items() if k != missing}})
+				self.patch(yahoo, 'fetch_detail',
+					lambda t, *a, **k: {t: {k: v for k, v in full.items() if k != missing}})
 				self.assertIn('Co T1', profile.prepare_profile_payload(TELEGRAM, '@u', 'T1')[0])
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: full | {'shareholder_equity': 0}})
 		self.assertNotIn('Net debt/equity', str(profile.prepare_profile_payload(TELEGRAM, '@u', 'T1')))
@@ -254,10 +274,13 @@ def test_a_profile_signs_its_changes(self):
 
 	def test_a_profile_warns_of_a_ps_ratio_only_above_15(self):
 		"""The PS ratio line warns only above 15; the forward P/E's warning used to show on it too."""
-		for ps, warned in ((12.3, False), (15.1, True)): # forward P/E 99 warns (above 30): its warning used to show on the PS line too
+		# forward P/E 99 warns (above 30): its warning used to show on the PS line too
+		for ps, warned in ((12.3, False), (15.1, True)):
 			with self.subTest(ps=ps):
-				self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'price_to_sales': ps, 'price_to_earnings_forward': 99, 'profile_industry': 'Tools'}})
-				line = next(line for line in profile.prepare_profile_payload(TELEGRAM, '@u', 'T1') if 'PS ratio' in line)
+				self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | {'price_to_sales': ps,
+					'price_to_earnings_forward': 99, 'profile_industry': 'Tools'}})
+				line = next(line for line in profile.prepare_profile_payload(TELEGRAM, '@u',
+					'T1') if 'PS ratio' in line)
 				self.assertEqual('⚠️' in line, warned, line)
 
 	def test_prepare_watchlist(self):
@@ -309,9 +332,10 @@ def test_only_the_scheduled_earnings_report_is_pinned(self):
 		self.assertEqual(pins, ['1'])
 
 	def test_earnings_skip_an_adr_whose_primary_listing_yahoo_does_not_know(self):
-		"""T1 has no earnings date, so its primary listing is looked up; Yahoo knowing nothing of it used to crash the report."""
+		"""T1 has no earnings date, so its primary listing is looked up: Yahoo not knowing it crashed the report."""
 		self.patch(cal, 'primary_listings', lambda: {'T1': 'T1.XX'})
-		self.patch(yahoo, 'fetch', lambda ts: {t: {k: v for k, v in market_entry(t).items() if not (t == 'T1' and k == 'earnings_date')} for t in ts})
+		undated = lambda t: {k: v for k, v in market_entry(t).items() if not (t == 'T1' and k == 'earnings_date')}
+		self.patch(yahoo, 'fetch', lambda ts: {t: undated(t) for t in ts})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {} if t == 'T1.XX' else {t: market_entry(t)})
 		cal.run(event='earnings')
 		self.assertIn('Co T2 ', str(self.sent))
@@ -333,7 +357,8 @@ def test_a_trade_is_in_its_market_currency_not_the_brokerage_one(self):
 	def test_trades_asks_yahoo_only_for_new_trades(self):
 		"""A scheduled trades run asks Yahoo only about trades it has not reported, and nothing once all are known."""
 		fetched = []
-		self.patch(yahoo, 'fetch', lambda tickers: fetched.append(set(tickers)) or {t: market_entry(t) for t in tickers})
+		self.patch(yahoo, 'fetch',
+			lambda tickers: fetched.append(set(tickers)) or {t: market_entry(t) for t in tickers})
 		trades.run()
 		self.assertEqual(fetched, [{'T1'}])
 		fetched.clear()
@@ -342,7 +367,8 @@ def test_trades_asks_yahoo_only_for_new_trades(self):
 
 	def test_a_scheduled_run_keeps_the_trades_another_run_recorded_meanwhile(self):
 		"""The seen trades are read again as the run ends, so a trade another run recorded meanwhile is kept."""
-		meanwhile = lambda tickers: state.save('finbot_sharesight_trades.json', [999]) or {t: market_entry(t) for t in tickers}
+		meanwhile = lambda tickers: state.save('finbot_sharesight_trades.json',
+			[999]) or {t: market_entry(t) for t in tickers}
 		self.patch(yahoo, 'fetch', meanwhile) # called after the seen trades were read, before the new ones are recorded
 		trades.run()
 		known = state.load('finbot_sharesight_trades.json')
@@ -354,9 +380,10 @@ def test_milestone(self):
 		self.assertTrue(self.sent)
 
 	def test_milestone_records_a_ticker_only_with_its_new_values(self):
-		"""A first sighting without 52-week values used to be recorded as 0/0, so the next run reported a record high for it."""
+		"""A first sighting without 52-week values was recorded as 0/0: the next run reported a record high for it."""
 		self.patch(yahoo, 'historic_high', lambda t: (150.0, 50.0)) # beyond the 52-week range (120/80): nothing new
-		self.patch(yahoo, 'fetch', lambda ts: {t: {k: v for k, v in market_entry(t).items() if not k.startswith('fiftyTwoWeek')} for t in ts})
+		self.patch(yahoo, 'fetch',
+			lambda ts: {t: {k: v for k, v in market_entry(t).items() if not k.startswith('fiftyTwoWeek')} for t in ts})
 		milestone.run()
 		self.assertFalse(state.load('finbot_milestone.json'))
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts})
@@ -375,8 +402,8 @@ def test_a_record_kept_from_ten_years_is_seeded_again_from_the_whole_history(sel
 		self.assertEqual(state.load('finbot_milestone.json')['T1'], {'all_time_high': 150.0, 'all_time_low': 50.0})
 
 	def test_milestone_records_are_all_time(self):
-		"""A new 52-week high below the all-time high is no record. The record used to become the 52-week high after the first
-		run, so a stock far below its peak posted a record high each time it set a new 52-week one."""
+		"""A new 52-week high below the all-time high is no record. The record used to become the 52-week high after the
+		first run, so a stock far below its peak posted a record high each time it set a new 52-week one."""
 		self.patch(yahoo, 'historic_high', lambda t: (150.0, 50.0))
 		milestone.run() # first sighting: the 52-week range (120/80) is inside the history (150/50)
 		for high, expected in ((130.0, False), (160.0, True), (155.0, False)):
@@ -404,7 +431,8 @@ def test_rating_keeps_the_last_rating_of_a_ticker_missing_from_a_run(self):
 		rating.run() # records every ticker's rating
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {} if t == 'T1' else {t: market_entry(t)})
 		rating.run() # no T1 this time: its rating used to be dropped from the record
-		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | ({'recommend': 'sell', 'recommend_index': 4.5} if t == 'T1' else {})})
+		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t) | ({'recommend': 'sell',
+			'recommend_index': 4.5} if t == 'T1' else {})})
 		self.sent.clear()
 		rating.run()
 		self.assertIn('Co T1 ', str(self.sent)) # so its change from buy to sell went unreported
@@ -415,8 +443,9 @@ def test_an_interactive_rating_records_nothing(self):
 		self.assertIsNone(state.load('finbot_rating.json'))
 
 	def test_slack_webhook_without_a_bot_token_gets_scheduled_reports(self):
-		"""Scheduled reports reach Slack through its incoming webhook, which takes no channel and no token. They used to send
-		Telegram's chat id along, which made write() need slackBotToken and crash without one, before Telegram got its copy."""
+		"""Scheduled reports reach Slack through its incoming webhook, which takes no channel and no token. They used to
+		send Telegram's chat id along, which made write() need slackBotToken and crash without one, before Telegram got
+		its copy."""
 		self.env(slackBotToken=False)
 
 		self.env(webhooks={**config.env.webhooks, 'slack': 'https://hooks.slack.com/services/T/B/x'})
@@ -425,7 +454,8 @@ def test_slack_webhook_without_a_bot_token_gets_scheduled_reports(self):
 		posts = []
 		def post(url, **kwargs):
 			"""Record (host, whether a bot token was sent, the channel) and answer ok."""
-			posts.append((url.split('/')[2], 'Authorization' in kwargs.get('headers', {}), (kwargs.get('json') or {}).get('channel')))
+			posts.append((url.split('/')[2], 'Authorization' in kwargs.get('headers', {}),
+				(kwargs.get('json') or {}).get('channel')))
 			return mock.Mock(status_code=200, json=lambda: {'ok': True, 'result': {'message_id': 1, 'chat': {'id': 1}}})
 		self.patch(transports.net, 'post', post)
 		for report in (price.run, shorts.run, milestone.run):
@@ -575,7 +605,8 @@ def test_format_helpers(self):
 		self.assertEqual(podcasts.format_duration('62:05'), '01:02')
 		self.assertEqual(podcasts.format_duration('nonsense'), '')
 		self.assertTrue(podcasts.how_long_ago(podcasts.published('Mon, 01 Jan 2024 00:00:00 +0000')).endswith('d ago'))
-		self.assertEqual(podcasts.published('Mon, 01 Jan 2024 00:00:00 -0000').tzinfo, datetime.timezone.utc) # no zone given: UTC
+		# no zone given: UTC
+		self.assertEqual(podcasts.published('Mon, 01 Jan 2024 00:00:00 -0000').tzinfo, datetime.timezone.utc)
 		self.assertIsNone(podcasts.published('yesterday'))
 
 	def test_new_episodes_are_posted_once_newest_first(self):
@@ -583,14 +614,18 @@ def test_new_episodes_are_posted_once_newest_first(self):
 		from lib.reports import podcasts
 		def item(n, date):
 			"""A feed item, its date optional."""
-			return f'<item><guid>g{n}</guid><title>Show: ep {n}</title><enclosure url="https://x/{n}.mp3"/>{date}<itunes:duration>600</itunes:duration></item>'
+			return (f'<item><guid>g{n}</guid><title>Show: ep {n}</title><enclosure url="https://x/{n}.mp3"/>{date}'
+				'<itunes:duration>600</itunes:duration></item>')
 		feed = ('<rss xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd"><channel><title>Show</title>'
-				+ item(1, '<pubDate>Mon, 01 Jan 2024 00:00:00 +0000</pubDate>') + item(2, '') + item(3, '<pubDate>Tue, 02 Jan 2024 00:00:00 GMT</pubDate>')
+				+ item(1, '<pubDate>Mon, 01 Jan 2024 00:00:00 +0000</pubDate>') + item(2, '') + item(3,
+					'<pubDate>Tue, 02 Jan 2024 00:00:00 GMT</pubDate>')
 				+ '</channel></rss>').encode()
 		self.patch(podcasts.net, 'get', lambda url, **k: mock.Mock(content=feed, raise_for_status=lambda: None))
 		podcasts.run(feed_url='https://example.com/feed')
-		self.assertEqual([line.split(' (')[0] for line in self.sent[0][1]], ['📣 Show: <a href="https://x/3.mp3">ep 3</a> 00:10',
-						 '📣 Show: <a href="https://x/1.mp3">ep 1</a> 00:10', '📣 Show: <a href="https://x/2.mp3">ep 2</a> 00:10'])
+		self.assertEqual([line.split(' (')[0] for line in self.sent[0][1]],
+			['📣 Show: <a href="https://x/3.mp3">ep 3</a> 00:10',
+						 '📣 Show: <a href="https://x/1.mp3">ep 1</a> 00:10',
+						 '📣 Show: <a href="https://x/2.mp3">ep 2</a> 00:10'])
 		self.sent.clear()
 		podcasts.run(feed_url='https://example.com/feed')
 		self.assertEqual(self.sent, []) # all seen now
@@ -602,8 +637,10 @@ def setUp(self):
 		super().setUp()
 		def holdings(portfolio):
 			"""The five holdings, whatever the portfolio."""
-			data = [('A', 'Semiconductors', 'Electronic Technology', 500), ('B', 'Semiconductors', 'Electronic Technology', 300),
-					('C', 'Biotechnology', 'Health Technology', 200), ('D', None, None, 100), ('E', 'Gold', 'Minerals', 0)]
+			data = [('A', 'Semiconductors', 'Electronic Technology', 500),
+				('B', 'Semiconductors', 'Electronic Technology', 300),
+					('C', 'Biotechnology', 'Health Technology', 200), ('D', None, None, 100),
+					('E', 'Gold', 'Minerals', 0)]
 			return {c + ':NASDAQ': {'code': c, 'market_code': 'NASDAQ', 'industry_classification_name': i,
 									'sector_classification_name': s, 'value': v} for c, i, s, v in data}
 		self.patch(sharesight.Portfolio, 'holdings', holdings)
@@ -623,7 +660,8 @@ def test_by_sector_and_payload(self):
 
 	def test_folds_into_other_and_renders(self):
 		"""Over eight slices, the smallest fold into Other, and the chart draws."""
-		held = {f'T{i}:X': {'code': f'T{i}', 'industry_classification_name': f'Ind {i}', 'value': 100 - i} for i in range(12)}
+		held = {f'T{i}:X': {'code': f'T{i}', 'industry_classification_name': f'Ind {i}',
+			'value': 100 - i} for i in range(12)}
 		self.patch(sharesight.Portfolio, 'holdings', lambda portfolio: held)
 		payload, chart = holdings.prepare_pie('Rob', 'industry', TELEGRAM)
 		self.assertEqual(len(chart.rows), 8)
diff --git a/tests/test_util.py b/tests/test_util.py
index 721750c1..5965c47f 100644
--- a/tests/test_util.py
+++ b/tests/test_util.py
@@ -1,4 +1,4 @@
-"""Pure helpers (lib/markets.py, formats.py, links.py) and the cache and state files: table-driven pass and fail cases."""
+"""Pure helpers (lib/markets.py, formats.py, links.py) and the cache and state files: table-driven cases."""
 import datetime, json, threading, unittest
 from pathlib import Path
 from tests.common import *
@@ -39,7 +39,8 @@ def test_each_market_has_its_own_flag_and_currency(self):
 				('CPSE', 'NOVO-B.CO', '🇩🇰', 'DKK'), ('XETR', 'SAP.DE', '🇩🇪', 'EUR'), ('AMS', 'ASML.AS', '🇳🇱', 'EUR'),
 				('EURONEXT', None, '🇪🇺', 'EUR'), ('HKG', '0700.HK', '🇭🇰', None), ('NASDAQ', 'AAPL', '🇺🇸', 'USD')):
 			with self.subTest(market=market):
-				self.assertEqual((markets.flag_from_market(market), markets.currency_from_market(market)), (flag, currency))
+				self.assertEqual((markets.flag_from_market(market), markets.currency_from_market(market)),
+					(flag, currency))
 				if ticker:
 					self.assertEqual(markets.flag_from_ticker(ticker), flag)
 
@@ -66,7 +67,8 @@ def test_no_yahoo_suffix_is_also_a_market_code(self):
 
 	def test_make_paragraphs_keeps_every_word(self):
 		"""make_paragraphs breaks a long description into paragraphs at sentence ends without losing a word."""
-		self.assertEqual(formats.make_paragraphs('Acme makes widgets. It sells them.'), 'Acme makes widgets. It sells them.')
+		self.assertEqual(formats.make_paragraphs('Acme makes widgets. It sells them.'),
+			'Acme makes widgets. It sells them.')
 		long = ' '.join(['word'] * 22) + ' end. Tail sentence here.'
 		self.assertEqual(formats.make_paragraphs(long), ' '.join(['word'] * 22) + ' end.\n\nTail sentence here.')
 
@@ -75,7 +77,8 @@ def test_transform_title(self):
 		self.assertEqual(formats.transform_title('APPLE INC'), 'Apple')
 		self.assertEqual(formats.transform_title('Acme FPO'), 'Acme')
 		self.assertEqual(formats.transform_title('Mixed Case Name'), 'Mixed Case Name')
-		for title, expected in (('Allianz Se', 'Allianz'), ('Foo SE AG', 'Foo'), ('Johnson & Johnson &', 'Johnson & Johnson'),
+		for title, expected in (('Allianz Se', 'Allianz'), ('Foo SE AG', 'Foo'),
+			('Johnson & Johnson &', 'Johnson & Johnson'),
 								('Black and White and', 'Black and White'), ('Big Holdings Things Ltd', 'Big Things')):
 			with self.subTest(title=title): # a suffix comes off the end only; no run of spaces is left behind
 				self.assertEqual(formats.transform_title(title), expected)
@@ -84,7 +87,8 @@ def test_transform_title(self):
 class Formatting(unittest.TestCase):
 	def test_signed_percent(self):
 		"""A signed percent: a plus for gains, thousands separators, and the decimals asked for."""
-		for value, decimals, expected in ((12, None, '+12%'), (-3.5, None, '-3.5%'), (0, None, '0%'), (1234, None, '+1,234%'),
+		for value, decimals, expected in ((12, None, '+12%'), (-3.5, None, '-3.5%'), (0, None, '0%'),
+			(1234, None, '+1,234%'),
 										  (2.345, 1, '+2.3%'), (-0.04, 1, '-0.0%')):
 			with self.subTest(value=value):
 				self.assertEqual(formats.signed_percent(value, decimals), expected)
@@ -100,7 +104,8 @@ def test_date_range_english(self):
 		end = datetime.date(2026, 10, 5)
 		self.assertEqual(formats.date_range_english(7, end), '28 Sep - 5 Oct 2026')
 		self.assertEqual(formats.date_range_english(30, end), '5 Sep - 5 Oct 2026')
-		self.assertEqual(formats.date_range_english(10, datetime.date(2026, 1, 5)), '26 Dec 2025 - 5 Jan 2026') # crosses a year
+		# crosses a year
+		self.assertEqual(formats.date_range_english(10, datetime.date(2026, 1, 5)), '26 Dec 2025 - 5 Jan 2026')
 		with self.assertRaises(TypeError):
 			formats.date_range_english(None, end) # callers must pass a period
 
@@ -124,7 +129,8 @@ def test_strip_url(self):
 
 	def test_ordinal(self):
 		"""Ordinal suffixes, the teens included."""
-		for number, suffix in ((1, 'st'), (2, 'nd'), (3, 'rd'), (4, 'th'), (11, 'th'), (12, 'th'), (13, 'th'), (21, 'st'), (112, 'th')):
+		for number, suffix in ((1, 'st'), (2, 'nd'), (3, 'rd'), (4, 'th'), (11, 'th'), (12, 'th'), (13, 'th'),
+				(21, 'st'), (112, 'th')):
 			with self.subTest(number=number):
 				self.assertEqual(formats.ordinal(number), suffix)
 
diff --git a/util/check_config.py b/util/check_config.py
index ef0615b7..75da666d 100755
--- a/util/check_config.py
+++ b/util/check_config.py
@@ -3,9 +3,10 @@
 
   util/check_config.py /opt/finbot.env [/var/lib/finbot/preferences.conf]
 
-finbot.env (the startup file) is never read by the bot, so this is the only place its typos are caught: an unknown key, a preference key in it,
-a malformed line or a bad value. preferences.conf defaults to <DATA_DIR>/preferences.conf and is checked the way the bot checks it at start
-and on reload. Exits 1 and prints every problem if either file is wrong, so a launcher can refuse to start (see the README)."""
+finbot.env (the startup file) is never read by the bot, so this is the only place its typos are caught: an unknown key,
+a preference key in it, a malformed line or a bad value. preferences.conf defaults to <DATA_DIR>/preferences.conf and is
+checked the way the bot checks it at start and on reload. Exits 1 and prints every problem if either file is wrong, so a
+launcher can refuse to start (see the README)."""
 
 import argparse
 import os
@@ -13,7 +14,8 @@
 from pathlib import Path
 
 if __name__ == '__main__':
-	os.environ['FINBOT_CHECK_ONLY'] = '1' # importing lib.config must not create the data directories (this may run as root, from systemd)
+	# importing lib.config must not create the data directories (this may run as root, from systemd)
+	os.environ['FINBOT_CHECK_ONLY'] = '1'
 sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
 from lib import config
 
@@ -48,13 +50,15 @@ def notes(preferences_path):
 if __name__ == '__main__':
 	ap = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter)
 	ap.add_argument('startup', help='the finbot.env to check (the startup file)')
-	ap.add_argument('preferences', nargs='?', default=config.preferences_path, help='default: <DATA_DIR>/preferences.conf')
+	ap.add_argument('preferences', nargs='?', default=config.preferences_path,
+		help='default: <DATA_DIR>/preferences.conf')
 	args = ap.parse_args()
 	problems = check(args.startup, args.preferences)
 	for line in problems:
 		print('ERROR config:', line, file=sys.stderr)
 	if not problems:
-		print('config: ok', args.startup, args.preferences if Path(args.preferences).is_file() else '(no preferences.conf, defaults)')
+		print('config: ok', args.startup,
+			args.preferences if Path(args.preferences).is_file() else '(no preferences.conf, defaults)')
 		for line in notes(args.preferences):
 			print('config:', line)
 	sys.exit(1 if problems else 0)
diff --git a/util/google_namecheap_cron b/util/google_namecheap_cron
index f0db1902..39cc97f9 100644
--- a/util/google_namecheap_cron
+++ b/util/google_namecheap_cron
@@ -29,7 +29,8 @@ get_new_ip(){
 
 update_provider(){
 	if [[ $PROVIDER == namecheap ]]; then
-		curl -s "https://dynamicdns.park-your-domain.com/update?host=${RECORD}&domain=${BASEDOMAIN}&password=${PASSWORD}&ip=${newip}" > /dev/null
+		url="https://dynamicdns.park-your-domain.com/update?host=${RECORD}&domain=${BASEDOMAIN}"
+		curl -s "${url}&password=${PASSWORD}&ip=${newip}" > /dev/null
 	elif [[ $(is_ipv4 $PROVIDER) ]]; then # RFC 2136
 		sudo nsupdate -4 -l -k $KEYFILE <<EOF
 server $PROVIDER
diff --git a/util/iptables_telegram_cron b/util/iptables_telegram_cron
index 83ed51e3..44a9ab9b 100644
--- a/util/iptables_telegram_cron
+++ b/util/iptables_telegram_cron
@@ -34,7 +34,8 @@ flush() {
 	iptables -N HTTP &>/dev/null || true
 	iptables -F HTTP
 	# note: blocking port 80 would break certbot http-01
-	iptables -C INPUT -p tcp -m multiport --dport 8443,443 -j HTTP &>/dev/null || iptables -I INPUT -p tcp -m multiport --dport 8443,443 -j HTTP
+	iptables -C INPUT -p tcp -m multiport --dport 8443,443 -j HTTP &>/dev/null ||
+		iptables -I INPUT -p tcp -m multiport --dport 8443,443 -j HTTP
 }
 
 stage() {
diff --git a/util/split_env.py b/util/split_env.py
index e9b81653..9033ca96 100755
--- a/util/split_env.py
+++ b/util/split_env.py
@@ -3,10 +3,10 @@
 
   util/split_env.py [old file]      (default: the .env in the checkout, which finbot will not start with)
 
-Reads the old file in either format: dotenv (key = 'value' # comment, export key=value) or plain KEY=value. Writes finbot.env.new
-and preferences.conf.new next to it; it never overwrites anything. Review them, then move them into place: finbot.env wherever your
-launcher loads it from, preferences.conf into DATA_DIR. Then delete the old file. Keys it cannot place (removed or unknown) are listed
-and left out."""
+Reads the old file in either format: dotenv (key = 'value' # comment, export key=value) or plain KEY=value. Writes
+finbot.env.new and preferences.conf.new next to it; it never overwrites anything. Review them, then move them into
+place: finbot.env wherever your launcher loads it from, preferences.conf into DATA_DIR. Then delete the old file. Keys
+it cannot place (removed or unknown) are listed and left out."""
 
 import argparse
 import os
@@ -24,8 +24,9 @@
 
 
 def parse_line(line):
-	"""(key, value) of a line of an old env file, None for a blank or comment line, (line, None) for one that is not KEY=value.
-	Takes the dotenv format too: an `export ` prefix, spaces around =, quotes, and a trailing # comment are dropped."""
+	"""(key, value) of a line of an old env file, None for a blank or comment line, (line, None) for one that is not
+	KEY=value. Takes the dotenv format too: an `export ` prefix, spaces around =, quotes, and a trailing # comment are
+	dropped."""
 	line = line.strip()
 	if not line or line.startswith('#'):
 		return None
@@ -68,14 +69,18 @@ def split(text, path='.env'):
 
 if __name__ == '__main__':
 	ap = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter)
-	ap.add_argument('file', nargs='?', default=str(config.OLD_ENV), help=f'the old env file (default: {config.OLD_ENV})')
+	ap.add_argument('file', nargs='?', default=str(config.OLD_ENV),
+		help=f'the old env file (default: {config.OLD_ENV})')
 	args = ap.parse_args()
 	src = Path(args.file)
 	if not src.is_file():
 		sys.exit(f'{src} does not exist: nothing to split')
 	startup, preferences, notes = split(src.read_text(), src)
-	outputs = {src.with_name('finbot.env.new'): ('# finbot.env: the startup file, loaded by your launcher; a change needs a restart.', startup, 0o600),
-		src.with_name('preferences.conf.new'): ('# preferences.conf: read by the bot; reload with systemctl reload finbot.', preferences, 0o644)}
+	outputs = {
+		src.with_name('finbot.env.new'):
+			('# finbot.env: the startup file, loaded by your launcher; a change needs a restart.', startup, 0o600),
+		src.with_name('preferences.conf.new'):
+			('# preferences.conf: read by the bot; reload with systemctl reload finbot.', preferences, 0o644)}
 	if any(path.exists() for path in outputs):
 		sys.exit('refusing to overwrite: ' + ', '.join(str(p) for p in outputs if p.exists()))
 	for path, (header, lines, mode) in outputs.items():
@@ -84,4 +89,5 @@ def split(text, path='.env'):
 		print(f'wrote {path} ({len(lines)} keys)', file=sys.stderr)
 	for note in notes:
 		print(note, file=sys.stderr)
-	print(f'next: put finbot.env where your launcher loads it and preferences.conf in the data directory (DATA_DIR), then delete {src}', file=sys.stderr)
+	print('next: put finbot.env where your launcher loads it and preferences.conf in the data directory (DATA_DIR),'
+		f' then delete {src}', file=sys.stderr)
diff --git a/util/sync_defaults.py b/util/sync_defaults.py
index 3716e6cb..081018cb 100755
--- a/util/sync_defaults.py
+++ b/util/sync_defaults.py
@@ -22,7 +22,9 @@ def sync(apply=False, force=False, dest=None):
 		if not src.is_file():
 			continue
 		if src.name == 'finbot.env':
-			continue # the startup file is not data: it holds credentials, lives wherever the launcher loads it from, and the bot never reads it
+			# the startup file is not data: it holds credentials, lives wherever the launcher loads it from, and the bot
+			# never reads it
+			continue
 		name = src.name
 		target = dest_dir / name
 		if not target.exists():
@@ -51,5 +53,6 @@ def sync(apply=False, force=False, dest=None):
 	args = ap.parse_args()
 	print(f"{config.REPO_DIR / 'defaults'} -> {Path(args.dest or config.env.data_dir).resolve()}")
 	for name, outcome in sync(args.apply or args.force, args.force, args.dest):
-		label = {'copied': '', 'skipped - destination identical': '', 'skipped - destination differs': '[differs]'}.get(outcome, f'[{outcome}]')
+		label = {'copied': '', 'skipped - destination identical': '',
+			'skipped - destination differs': '[differs]'}.get(outcome, f'[{outcome}]')
 		print(f"{name} {label}".rstrip())

From cb0f4127483018f01089c29e6b02489f087228ce Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 11:15:54 +0000
Subject: [PATCH 74/81] tests: the three largest files split by subject;
 CLAUDE.md's maps brought up to date (D25, N7)

test_providers.py (65 KB), test_jobs.py (44 KB) and test_edges.py (42 KB) were the biggest files a session opened. They
are split by subject, as moves only (the same 400 tests pass):
- test_yahoo, test_sharesight, test_sending (Telegram, Slack and Discord posting) and test_telegram_api, from
  test_providers; the stand-ins they share (sample, Resp, serve, ProviderCase) moved into tests/common.py;
- test_schedule (the schedule file, job arguments) and test_scheduler (firing, daylight saving, runs from chat, child
  processes), from test_jobs;
- test_inbound (webhook requests, Slack events and signatures, text escaped on arrival), from test_edges, which keeps
  the failures, bad arguments, secrets in logs and signals.

CLAUDE.md: the new test map, the code map for buttons, charts, markets and the test helpers, the worker's entry points,
the README's size (36 KB), and its own lines wrapped at 120 from the Facts down.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md                                 |   71 +-
 tests/common.py                           |   75 +-
 tests/test_edges.py                       |  346 +-----
 tests/test_inbound.py                     |  351 ++++++
 tests/test_providers.py                   | 1268 ---------------------
 tests/test_schedule.py                    |  187 +++
 tests/{test_jobs.py => test_scheduler.py} |  188 +--
 tests/test_sending.py                     |  549 +++++++++
 tests/test_sharesight.py                  |  274 +++++
 tests/test_telegram_api.py                |   96 ++
 tests/test_yahoo.py                       |  314 +++++
 11 files changed, 1893 insertions(+), 1826 deletions(-)
 create mode 100644 tests/test_inbound.py
 delete mode 100644 tests/test_providers.py
 create mode 100644 tests/test_schedule.py
 rename tests/{test_jobs.py => test_scheduler.py} (77%)
 create mode 100644 tests/test_sending.py
 create mode 100644 tests/test_sharesight.py
 create mode 100644 tests/test_telegram_api.py
 create mode 100644 tests/test_yahoo.py

diff --git a/CLAUDE.md b/CLAUDE.md
index 14ba1158..e3cf60c8 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -21,10 +21,12 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
 - Python 3.13.5; tests are offline and need no network (`tests/`). CI runs them on the latest 3.x and 3.13.5, except for
   pull requests that change only README.md, CLAUDE.md, `doc/*.md` or `img/*.png` (install, compile and tests are
   skipped; the allowlist in `.github/workflows/ci.yml` must never include a path a test reads).
-- Production runs the Debian system packages (including `python3-croniter`, for the scheduler in `lib/jobs.py`); `requirements.txt` is for pip/uv users and CI.
-- A new preference goes in three places: a `Prefs` field with its default (`lib/config.py`; its type says how it is read),
-  `defaults/preferences.conf`, and the README's preferences table. The tests catch a key missing from the template or a
-  default that differs, but not the README (no test may read README.md: CI skips the tests for a README-only change).
+- Production runs the Debian system packages (including `python3-croniter`, for the scheduler in `lib/jobs.py`);
+  `requirements.txt` is for pip/uv users and CI.
+- A new preference goes in three places: a `Prefs` field with its default (`lib/config.py`; its type says how it is
+  read), `defaults/preferences.conf`, and the README's preferences table. The tests catch a key missing from the
+  template or a default that differs, but not the README (no test may read README.md: CI skips the tests for a
+  README-only change).
 - If Yahoo and Sharesight are unreachable (the cloud environment's network policy can block them), check behaviour
   against the offline fixtures in `tests/common.py`.
 - For a refactor, `util/record_outputs.py` (or `--against COMMIT`) records what every command, button and scheduled
@@ -43,7 +45,7 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   command by its whole name (any other first word that may be a ticker is `.SYMBOL`) and `read_args()` its `Args`: a
   percent, a period and a keyword from either end, the words between; a handler raises `Usage` for arguments it cannot
   take, and gets a ticker with `ticker(args)`. `Request` (`request.reply(lines, chart, buttons)`); each command gets
-  the typing notice while it works.
+  the typing notice while it works. `process_request(message)` and `process_callback(press)` take inbound's objects.
 - `lib/reports/`: what finbot says. Scheduled (each has `run(dest=None, **kwargs)`, which posts to the chat that asked
   or else the default chats, and `job(argv)`, its schedule-line parser): `price performance trades shorts cal rating
   milestone reminder podcasts cleanup`; the top-level `price.py` and the rest are shims that run them. Chat-only
@@ -53,12 +55,14 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   `deliver(dest, report)`, the one-line error report. `lib/transports.py`: a class per way of posting (`TelegramBot`,
   `SlackBot`, `SlackWebhook`, `DiscordWebhook`): splitting, captions, charts, buttons, edits, and the typing notice
   (`start_typing`: after half a second; Slack's ⏳ on the message), which `text()` and `chart()` stop before posting.
-  `lib/telegram.py`: Bot API calls.
-- `lib/buttons.py`: `Button` and a chart's period buttons (each transport draws them). `lib/charts.py`: a period
-  button press rebuilds its chart.
-- `lib/dialects.py`: each network's markup and message sizes. `lib/markets.py`: market tables, flags, ticker translation.
-  `lib/formats.py`: names, units, dates, percents in words. `lib/links.py`: stock links. `lib/render.py`: every chart,
-  `Chart` specs and `draw()`.
+  `lib/telegram.py`: Bot API calls (`call`, `try_call`).
+- `lib/buttons.py`: `Button`, a chart's period buttons (`keyboard`), a `.jobs` command's (`job_button`), and what a
+  pressed button's data asks for (`parse`); each transport draws them. `lib/charts.py`: a chart with period buttons,
+  built by its kind's builder (`build`) for the command's first view (`worker.chart_reply`) and for each press.
+- `lib/dialects.py`: each network's markup and message sizes. `lib/markets.py`: one table by market code (flag,
+  currency, Yahoo suffix, Google code; Sharesight's codes, so TSE is Toronto), and `code(name)` for any name an exchange
+  goes by. `lib/formats.py`: names, units, dates, percents in words. `lib/links.py`: stock links. `lib/render.py`:
+  every chart, `Chart` specs and `draw()`.
 - `lib/yahoo.py lib/sharesight.py lib/shortman.py lib/simplywallst.py`: data sources. Yahoo's and Sharesight's answers
   are cached in `DATA_DIR/cache` as sent, not parsed, so a parser fix needs no flush (`cache.cached`).
   `yahoo.symbol(typed)` is Yahoo's symbol for a ticker as written, tried on the local exchange (`exchange_suffix`)
@@ -67,29 +71,36 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   `trades(days)`.
 - `lib/cache.py`: cache files (any may vanish at any time). `lib/state.py`: state files in `DATA_DIR`, which must last;
   change one with `with state.edit(name, default) as data:`, never load-then-save. `lib/watchlist.py`: the watchlist.
-- `lib/config.py`: `config.env` (finbot.env, fixed at start) and `config.prefs` (preferences.conf, a new object on each reload).
-  `lib/jobs.py` + `lib/jobrun.py`: the scheduler, its shared state under one lock; each command runs in a child
-  process, stopped after `JOB_TIMEOUT`.
-- `tests/common.py`: `FinbotCase` (offline fixtures). `self.command('.price')` returns what was sent: `('text', lines)` or
-  `('photo', size, caption)`. `wsgi_environ(path, body, **headers)` is a request for `bot.main`.
-- Which test file: commands and scheduled reports `test_reports`; failures, bad input, WSGI and Slack events `test_edges`;
-  Yahoo/Sharesight parsing, sending, Telegram API `test_providers`; scheduler `test_jobs`; config `test_config`; charts
-  `test_charts` (refresh goldens with `UPDATE_GOLDEN=1`); markets, formats, links, cache and state files `test_util`; data dir
-  `test_datadir`; chunking, git_version, and the docstring, stdout and line-width checks `test_finbot`.
+- `lib/config.py`: `config.env` (finbot.env, fixed at start) and `config.prefs` (preferences.conf, a new object on
+  each reload). `lib/jobs.py` + `lib/jobrun.py`: the scheduler, its shared state under one lock; each command runs in
+  a child process, stopped after `JOB_TIMEOUT`.
+- `tests/common.py`: `FinbotCase` (offline fixtures). `self.command('.price')` returns what was sent:
+  `('text', lines)` or `('photo', size, caption)`. `wsgi_environ(path, body, **headers)` is a request for `bot.main`.
+  `ProviderCase` runs the real parsers on the answers saved under `doc/` (`sample`), as `Resp` and `serve` return
+  them; `temp_dir(self)` is a directory removed after the test, `bare_environment()` the environment without finbot's
+  settings.
+- Which test file: commands and scheduled reports `test_reports`; failures, bad input, secrets in logs, signals
+  `test_edges`; webhook requests, Slack events and signatures `test_inbound`; Yahoo `test_yahoo`; Sharesight
+  `test_sharesight`; posting to each network `test_sending`; registering Telegram's webhook `test_telegram_api`; the
+  schedule file and job arguments `test_schedule`; the scheduler and manual runs `test_scheduler`; config
+  `test_config`; charts `test_charts` (refresh goldens with `UPDATE_GOLDEN=1`); markets, formats, links, buttons,
+  cache and state files `test_util`; data dir `test_datadir`; chunking, git_version, and the docstring, stdout and
+  line-width checks `test_finbot`.
 
 ## Conventions
-- Read settings as `config.env.<name>` / `config.prefs.<name>` where they are used, never into a default or module constant (a
-  reload must be seen); in tests, `self.env(...)` / `self.prefs(...)` change them.
-- A report is a list of lines. `service` is a dialect (`lib/dialects.py`), which equals the network's name: markup with
-  `service.bold(text)`, `.italic`, `.strike`, `.link(url, text)`, user-written text through `service.escape(text)`, stock links
-  through `links.finance_link(symbol, exchange, service)`. Tests pass `TELEGRAM`, `SLACK`, `DISCORD`.
+- Read settings as `config.env.<name>` / `config.prefs.<name>` where they are used, never into a default or module
+  constant (a reload must be seen); in tests, `self.env(...)` / `self.prefs(...)` change them.
+- A report is a list of lines. `service` is a dialect (`lib/dialects.py`), which equals the network's name: markup
+  with `service.bold(text)`, `.italic`, `.strike`, `.link(url, text)`, user-written text through
+  `service.escape(text)`, stock links through `links.finance_link(symbol, exchange, service)`. Tests pass `TELEGRAM`,
+  `SLACK`, `DISCORD`.
 - A command that cannot give what was asked (no such portfolio, no prices) raises `answer.Answer(*lines)`, which the
   chat gets as a plain reply; `yahoo.NoData` is one. Any other exception becomes a one-line ⚠️ error report.
 - Every function and method has a docstring, tests included (a test checks): one line where it fits, reworded rather
   than spilling a few words onto a second. Tabs everywhere (`.editorconfig`). Every line stays within 120 columns,
   tabs counting 4 (an edit quotes whole lines); a test checks all but README.md, CLAUDE.md and `doc/`.
-- Log with `print(..., file=sys.stderr)`. systemd discards stdout (`StandardOutput=null`), so the bot and its reports never
-  print there (a test checks); only the util/ tools, run by hand, do.
+- Log with `print(..., file=sys.stderr)`. systemd discards stdout (`StandardOutput=null`), so the bot and its reports
+  never print there (a test checks); only the util/ tools, run by hand, do.
 - A report gives its chart as a `render.Chart` spec, not a drawing: `deliver()` draws it only for a chat that takes
   charts, through `render.draw()`, which caches the image by the spec's hash (and the drawing code's).
 - A chart renderer in `lib/render.py` takes its `Chart` and is listed in `RENDERERS` by kind; it gets `@_drawn` (the
@@ -98,8 +109,8 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   buttons make its `.help` lines and keyboard buttons, and the README command list.
 
 ## Saving tokens
-- Run the tests quietly: `.venv/bin/python -m unittest discover -s tests -b` prints only failures (about 0.4 KB when green). One
-  module: `.venv/bin/python -m unittest tests.test_reports -b`.
-- README.md is 38 KB: grep for the section instead of reading it whole. Each network's setup is in its own guide,
+- Run the tests quietly: `.venv/bin/python -m unittest discover -s tests -b` prints only failures (about 0.4 KB when
+  green). One module: `.venv/bin/python -m unittest tests.test_reports -b`.
+- README.md is 36 KB: grep for the section instead of reading it whole. Each network's setup is in its own guide,
   `doc/telegram.md`, `doc/slack.md` and `doc/discord.md`: a change to a network's setup or keys goes there. `doc/yahoo/`
   and `doc/sharesight/` hold large API samples that the tests replay: inspect them with python, don't read them.
diff --git a/tests/common.py b/tests/common.py
index bed5beaf..ff2c2bdd 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -1,6 +1,8 @@
 """Shared fixture: points finbot at a temp cache dir and replaces every external service
-(Telegram, Yahoo, Sharesight, Short Man) with canned data. Nothing here touches the network."""
-import collections, dataclasses, io, os, shutil, sys, tempfile, time, unittest
+(Telegram, Yahoo, Sharesight, Short Man) with canned data. Nothing here touches the network. ProviderCase runs the real
+parsers instead, on the API answers saved under doc/ (sample) as a stand-in network returns them (Resp, serve)."""
+import collections, dataclasses, io, json, os, shutil, sys, tempfile, time, unittest, zoneinfo
+from pathlib import Path
 import pandas as pd
 from unittest import mock
 
@@ -170,3 +172,72 @@ def command(self, text, chat='55'):
 		worker.process_request(inbound.Message('telegram', chat, '@u', text, BOT, 'U'))
 		self.assertEqual(self.errors, [], f'{text} crashed') # the worker reports crashes instead of raising
 		return list(self.sent)
+
+
+DOC = os.path.join(ROOT, 'doc')
+
+
+def sample(*path):
+	"""A saved API response from doc/."""
+	with open(os.path.join(DOC, *path), encoding='utf-8') as f:
+		return json.load(f)
+
+
+class Resp:
+	def __init__(self, payload=None, status=200, text=''):
+		"""A response with this JSON payload (None: not JSON), status and text."""
+		self.payload, self.status_code, self.text = payload, status, text or json.dumps(payload)
+	def json(self):
+		"""The payload, or ValueError when it is not JSON."""
+		if self.payload is None:
+			raise ValueError('no json')
+		return self.payload
+
+
+def serve(*responses):
+	"""net.get replacement: replays the responses in order (exceptions are raised); the last one repeats."""
+	queue = list(responses)
+	calls = []
+	def get(url, **kwargs):
+		"""The next response, or raise it when it is an exception."""
+		calls.append(url)
+		item = queue.pop(0) if len(queue) > 1 else queue[0]
+		if isinstance(item, Exception):
+			raise item
+		return item
+	get.calls = calls
+	return get
+
+
+REAL_CREDENTIALS = yahoo.credentials
+
+
+CHART = render.Chart('bars', 'Heading', rows=(('A', 1.0),)) # a chart for the sending tests, which draw it as b'png'
+
+
+REAL_TOKEN = sharesight._token
+
+
+class ProviderCase(FinbotCase):
+	def setUp(self):
+		"""Real parsers instead of the fixture's canned data, with a fixed cookie, crumb and token, and no waiting."""
+		super().setUp()
+		# these tests exercise the real parsers, so undo the canned data from the base fixture
+		for module, name in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'),
+				(sharesight, '_portfolios'), (sharesight, '_trades'), (sharesight, '_performance')):
+			self.patch(module, name, REAL[(module.__name__, name)])
+		self.patch(yahoo, 'credentials', lambda renew=False: ('A=B', 'crumb'))
+		self.patch(sharesight, '_token', lambda: 'token')
+		self.patch(sharesight.time, 'sleep', lambda s: None)
+
+	def http(self, module, *responses):
+		"""Have module's net.get replay these responses; returns the stand-in, whose calls lists the URLs asked for."""
+		getter = serve(*responses)
+		self.patch(module.net, 'get', getter)
+		return getter
+
+
+UTC = zoneinfo.ZoneInfo('UTC')
+
+
+DEFAULT_SCHEDULE = Path(ROOT, 'defaults', 'schedule.cron')
diff --git a/tests/test_edges.py b/tests/test_edges.py
index 19a346b5..f68de779 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -1,13 +1,9 @@
-"""Failure and no-data scenarios: empty or broken upstream data, bad requests, rejected callers."""
-import contextlib, hashlib, hmac, io, json, os, signal, subprocess, sys, tempfile, threading, time, unittest
+"""Failure and no-data scenarios: empty or broken upstream data, bad arguments, unknown portfolios, secrets
+kept out of logs, and the bot's signals."""
+import contextlib, io, os, signal, subprocess, sys, tempfile, threading, time, unittest
 from pathlib import Path
-from urllib.parse import urlencode
-import gevent
 from tests.common import *
 
-PRESS = {'type': 'block_actions', 'token': 'stok', 'trigger_id': 't1', 'user': {'id': 'U1', 'username': 'rob'},
-	'container': {'type': 'message', 'message_ts': '1.5', 'channel_id': 'C1'},
-	'actions': [{'action_id': 'c|h|w|T1', 'value': 'c|h|w|T1', 'type': 'button'}]} # a Slack period button pressed
 
 COMMANDS = ('.price', '.price 7d', '.price top', '.session', '.premarket', '.performance 7d', '.shorts', '.peg', '.buy',
 			'.sell', '.beta', '.pe', '.marketcap', '.earnings', '.dividend', '.trades', '.holdings', '.watchlist',
@@ -281,342 +277,6 @@ def test_a_ticker_yahoo_does_not_know_is_tried_on_the_local_exchange(self):
 				self.assertIn('Co BHP.AX', str(self.command(message)))
 
 
-class WsgiRejections(FinbotCase):
-	SECRET = {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'tok'}
-
-	def request(self, path='/telegram', body=None, headers=None):
-		"""(the body, the status) the WSGI handler answers a request with."""
-		body = body if isinstance(body, bytes) else json.dumps(body if body is not None else {}).encode()
-		env = wsgi_environ(path, body, **(headers or {}))
-		status = []
-		with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
-			out = bot.main(env, lambda s, *a, **k: status.append(s))
-		return b''.join(out), status[-1]
-
-	def test_telegram_requires_the_secret_header(self):
-		"""A Telegram update without the secret header, or with a wrong one, is refused and not processed."""
-		message = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.help'}}
-		self.assertIn(b'Unauthorized', self.request(body=message)[0])
-		self.assertIn(b'Unauthorized',
-			self.request(body=message, headers={'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'wrong'})[0])
-		self.assertEqual(self.sent, []) # nothing was processed
-
-	def test_an_empty_secret_proves_nothing(self):
-		"""With no secret configured, a request carrying an empty one is refused, on Telegram and on Slack."""
-		self.env(telegramOutgoingToken='', slackOutgoingWebhook='https://example.com/slack', slackOutgoingToken='')
-		message = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.help'}}
-		empty = {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': ''}
-		self.assertEqual(self.request(body=message, headers=empty)[1], '403 Forbidden')
-		self.assertEqual(self.request('/slack', {'token': '', 'type': 'url_verification', 'challenge': 'c'})[1],
-			'403 Forbidden')
-		self.assertEqual((self.sent, self.errors), ([], []))
-
-	def test_unknown_path_is_404(self):
-		"""An unknown path is 404."""
-		self.assertEqual(self.request(path='/nope')[1], '404 Not Found')
-
-	def test_malformed_body_is_refused_not_a_crash(self):
-		"""A body that is not a JSON object is 400 once authenticated, and turned away unread before that."""
-		# no secret: turned away before the body is read
-		self.assertEqual(self.request(body=b'not json')[1], '403 Forbidden')
-		for body in (b'not json', b'', b'[1, 2]', b'7'):
-			self.assertEqual(self.request(body=body, headers=self.SECRET)[1], '400 Bad Request', body)
-		self.assertEqual(self.errors, [])
-
-	def test_only_the_body_is_read_and_a_huge_one_not_at_all(self):
-		"""Only CONTENT_LENGTH bytes are read, since a server need not end the stream there; a body over MAX_BODY is
-		refused unread, and a length that is not a number is a bad request."""
-		self.patch(worker, 'process_request', lambda *a: None)
-		def post(body, length):
-			"""(the status, the bytes read) for this body and CONTENT_LENGTH, with more on the stream after it."""
-			env = {**wsgi_environ('/telegram', body, **self.SECRET), 'CONTENT_LENGTH': length}
-			env['wsgi.input'] = stream = io.BytesIO(body + b'{"the next request": 1}')
-			status = []
-			with contextlib.redirect_stderr(io.StringIO()):
-				bot.main(env, lambda s, *a, **k: status.append(s))
-			return status[-1], stream.tell()
-		message = {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.help'}
-		update = json.dumps({'message': message}).encode()
-		self.assertEqual(post(update, str(len(update))), ('200 OK', len(update)))
-		self.assertEqual(post(update, str(bot.MAX_BODY + 1)), ('413 Payload Too Large', 0))
-		self.assertEqual(post(update, 'lots'), ('400 Bad Request', 0))
-
-	def test_strangers_cannot_post_into_the_chats(self):
-		"""Strangers (wrong paths, no or wrong secrets, malformed bodies) are refused, and nothing reaches the chats."""
-		self.env(slackOutgoingWebhook='https://example.com/slack')
-
-		self.env(slackOutgoingToken='stok')
-
-		for path, body, headers, status in (('/', b'', {}, '404 Not Found'),
-			('/wp-login.php', b'log=x&pwd=y', {}, '404 Not Found'),
-				('/telegram', b'log=x', {}, '403 Forbidden'),
-				('/telegram', {'message': {}}, {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'wrong'}, '403 Forbidden'),
-				('/slack', b'', {}, '400 Bad Request'), ('/slack', b'token=stok', {}, '400 Bad Request'),
-				('/slack', {'type': 'event_callback'}, {}, '403 Forbidden'),
-				('/slack', {'token': 'wrong', 'type': 'event_callback'}, {}, '403 Forbidden'),
-				('/slack', {'token': 'stok'}, {}, '200 OK')):
-			with self.subTest(path=path, body=body):
-				self.assertEqual(self.request(path, body, headers)[1], status)
-		self.assertEqual(self.errors, []) # nothing reached the chats
-		self.assertEqual(self.sent, [])
-
-	def test_a_failure_before_authentication_is_only_logged(self):
-		"""A failure while routing, before authentication, is logged and answered 500, never posted to the chats."""
-		def broken(*a, **k):
-			"""Fail while routing."""
-			raise RuntimeError('boom')
-		self.patch(inbound, 'urlparse', broken) # fails while routing, before the secret is even looked at
-		env = wsgi_environ('/telegram', b'{}', **self.SECRET)
-		status = []
-		with contextlib.redirect_stderr(io.StringIO()) as err:
-			bot.main(env, lambda s, *a, **k: status.append(s))
-		self.assertEqual((status[-1][:3], self.errors), ('500', []))
-		self.assertIn('boom', err.getvalue())
-
-	def test_a_reported_failure_names_the_network_not_its_secret_path(self):
-		"""The error line of a failed update names the network, not the webhook's path, which is a secret."""
-		self.env(telegramOutgoingWebhook='https://example.com/hook-s3cret')
-		broken = {'message': {'message_id': 1, 'from': {'first_name': 'A'}, 'text': '.help'}} # no chat
-		self.assertEqual(self.request('/hook-s3cret', broken, self.SECRET)[1], '200 OK')
-		self.assertEqual(self.errors, ["⚠️ telegram: KeyError: 'chat'"])
-
-	def test_a_failure_after_authentication_is_reported_once_and_not_retried(self):
-		"""A failure on an authenticated update is reported once and answered 200, so Telegram doesn't send it again."""
-		broken = {'message': {'message_id': 1, 'from': {'first_name': 'A'}, 'text': '.help'}} # no chat
-		# a retry would only fail and report again
-		self.assertEqual(self.request(body=broken, headers=self.SECRET)[1], '200 OK')
-		self.assertEqual(len(self.errors), 1)
-		self.assertIn("KeyError: 'chat'", self.errors[0])
-
-	def test_telegram_edge_updates(self):
-		"""Edits, button presses without a message, and messages without text are answered and left alone."""
-		secret = self.SECRET
-		self.assertEqual(self.request(body={'edited_message': {}}, headers=secret)[0], b'Unsupported')
-		self.assertEqual(self.request(body={'callback_query': {'id': '1'}}, headers=secret)[0], b'Unsupported')
-		odd = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'new_chat_members': []}}
-		self.assertIn(b'Unhandled', self.request(body=odd, headers=secret)[0])
-
-	def test_a_telegram_update_sent_again_runs_once(self):
-		"""An update Telegram sends again runs once."""
-		calls = []
-		self.patch(worker, 'process_request', lambda *a: calls.append(a))
-		body = {'update_id': 7,
-			'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.price'}}
-		for _ in range(2): # Telegram sends an update again when it got no answer
-			self.assertEqual(self.request(body=body, headers=self.SECRET)[1], '200 OK')
-		gevent.sleep(0.05)
-		self.assertEqual(len(calls), 1)
-
-	def test_an_update_is_taken_however_it_arrived(self):
-		"""Reading an update needs no webhook: one handed over another way (as a pull mode would) reaches the worker as
-		from the webhook, Telegram's message and Slack's alike, and a repeat runs once."""
-		got = []
-		self.patch(worker, 'process_request', got.append)
-		message = {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A', 'username': 'a'}, 'text': '.price'}
-		event = {'type': 'event_callback', 'event_id': 'Ev9', 'authorizations': [{'user_id': 'UBOT'}],
-			'event': {'type': 'message', 'channel': 'C1', 'user': 'U1', 'text': '.trades', 'ts': '1.2'}}
-		with contextlib.redirect_stderr(io.StringIO()):
-			for _ in range(2):
-				self.assertEqual(bot.take(inbound.TELEGRAM, {'update_id': 70, 'message': message}), inbound.OK)
-				self.assertEqual(bot.take(inbound.SLACK, event), inbound.OK)
-		gevent.sleep(0.05) # the worker runs in a (gevent) thread
-		self.assertEqual([(m.service, m.chat_id, m.text) for m in got], [('telegram', '5', '.price'),
-			('slack', 'C1', '.trades')])
-
-	def test_button_presses_are_logged(self):
-		"""A button press is logged with who pressed it (handle, first name or unknown), the chat and the button."""
-		calls = []
-		self.patch(worker, 'process_callback', lambda *a: calls.append(a))
-		secret = {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'tok'}
-		for sender, who in (({'username': 'rob', 'first_name': 'Rob'}, '@rob'), ({'first_name': 'Ann'}, 'Ann'),
-				({}, 'unknown')):
-			body = json.dumps({'callback_query': {'id': '1', 'from': sender, 'data': 'j|7|abcd',
-				'message': {'message_id': 9, 'chat': {'id': 55}}}}).encode()
-			err = io.StringIO()
-			with contextlib.redirect_stderr(err):
-				bot.main(wsgi_environ('/telegram', body, **secret), lambda *a, **k: None)
-			self.assertIn(f'[Telegram button]: {who} chat 55 j|7|abcd', err.getvalue())
-
-
-class SlackEvents(FinbotCase):
-	"""Slack sends every message in a subscribed channel, the bot's own posts included."""
-
-	def setUp(self):
-		"""Slack commands configured, and the worker replaced by a list of what it would have run."""
-		super().setUp()
-		self.env(slackOutgoingWebhook='https://example.com/slack')
-
-		self.env(slackOutgoingToken='stok')
-
-		self.requests = []
-		self.patch(worker, 'process_request', self.requests.append)
-
-	def event(self, event_id=None, **event):
-		"""Post a Slack message event to the WSGI handler and return the status."""
-		body = json.dumps({'token': 'stok', 'type': 'event_callback', 'authorizations': [{'user_id': 'UBOT'}],
-			'event_id': event_id, 'event': dict({'type': 'message', 'channel': 'C1', 'ts': '1.2'}, **event)}).encode()
-		status = []
-		with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
-			bot.main(wsgi_environ('/slack', body), lambda s, *a, **k: status.append(s))
-		gevent.sleep(0.05) # the worker runs in a (gevent) thread
-		return status[-1]
-
-	def press(self, payload):
-		"""Post a button press as Slack's interactivity does (a form, the JSON in its payload field): the status."""
-		status = []
-		with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
-			body = urlencode({'payload': json.dumps(payload)}).encode()
-			bot.main(wsgi_environ('/slack', body), lambda s, *a, **k: status.append(s))
-		gevent.sleep(0.05) # the worker runs in a (gevent) thread
-		return status[-1]
-
-	def test_a_button_press_reaches_the_worker(self):
-		"""A period button pressed under a chart reaches the worker, posted to the events URL: the channel, the message
-		and the button's value. A press with a wrong token is refused, and one not under a message goes nowhere."""
-		presses = []
-		self.patch(worker, 'process_callback', presses.append)
-		self.assertEqual(self.press(PRESS), '200 OK')
-		self.assertEqual(presses, [inbound.Press('slack', 't1', 'C1', '1.5', 'c|h|w|T1', '@rob')])
-		self.assertEqual(self.press({**PRESS, 'token': 'wrong'}), '403 Forbidden')
-		self.assertEqual(self.press({**PRESS, 'container': {'type': 'view', 'view_id': 'V1'}}), '200 OK')
-		self.assertEqual(self.press({**PRESS, 'actions': []}), '200 OK')
-		self.assertEqual(len(presses), 1)
-
-	def test_the_message_goes_to_the_worker_with_its_ts(self):
-		"""A message reaches the worker with its ts, the message's id: finbot puts ⏳ on it while it works."""
-		self.assertEqual(self.event(user='U1', text='.price'), '200 OK')
-		self.assertEqual(self.requests[0].message_id, '1.2')
-
-	def test_only_people_are_answered(self):
-		"""Messages from bots, this bot included, joins, edits and events without a user are not answered; people's
-		messages are, with mentions turned into @names."""
-		# a webhook report
-		for event in ({'subtype': 'bot_message', 'bot_id': 'B1', 'text': "⚠️ /slack: KeyError: 'user'"},
-					  {'user': 'UBOT', 'bot_id': 'B2', 'text': '.holdings: Please try again'}, # this bot's own reply
-					  {'user': 'UBOT', 'text': '.holdings'},
-					  {'subtype': 'channel_join', 'user': 'U1', 'text': '<@U1> has joined the channel'},
-					  {'subtype': 'message_changed', 'message': {'text': '.price'}},
-					  {'text': '.price'}): # no user at all
-			with self.subTest(event=event):
-				self.assertEqual(self.event(**event), '200 OK')
-		self.assertEqual((self.requests, self.errors), ([], []))
-		self.assertEqual(self.event(user='U1', text='.price <@UBOT>'), '200 OK')
-		self.assertEqual(self.event(user='U2', subtype='thread_broadcast', text='.trades'), '200 OK')
-		self.assertEqual([(r.service, r.chat_id, r.user, r.text, r.bot_name) for r in self.requests],
-						 [('slack', 'C1', '<@U1>', '.price @UBOT', '@UBOT'),
-						 	('slack', 'C1', '<@U2>', '.trades', '@UBOT')])
-
-	def test_links_and_escapes_lose_their_markup(self):
-		"""Each link loses Slack's markup, however many there are (Slack links .AX tickers as domains), and Slack's
-		escapes become characters again: .buy &amp; sell runs both reports."""
-		for text, expected in (('.compare <http://BHP.AX|BHP.AX> <http://CBA.AX|CBA.AX>', '.compare BHP.AX CBA.AX'),
-				('.price <https://example.com>', '.price https://example.com'), ('.buy &amp; sell', '.buy & sell'),
-				('.plan buy &lt;BHP&gt;', '.plan buy <BHP>')):
-			with self.subTest(text=text):
-				self.assertEqual(self.event(user='U1', text=text), '200 OK')
-				self.assertEqual(self.requests[-1].text, expected)
-
-	def test_a_retried_event_runs_once(self):
-		"""An event Slack sends again (same event_id) runs once."""
-		for _ in range(2): # Slack resends an event it got no answer to within 3 seconds, with the same event_id
-			self.assertEqual(self.event('Ev1', user='U1', text='.price'), '200 OK')
-		self.assertEqual(self.event('Ev2', user='U1', text='.price'), '200 OK')
-		self.assertEqual(len(self.requests), 2)
-
-	def test_slack_does_not_change_the_telegram_bot_name(self):
-		"""A Slack event leaves the Telegram bot's name alone."""
-		self.patch(inbound.TELEGRAM, 'bot_name', '@finbot_tg')
-		self.event(user='U1', text='.price')
-		self.assertEqual(inbound.TELEGRAM.bot_name, '@finbot_tg')
-
-
-class SlackSigning(FinbotCase):
-	"""With slackSigningSecret, a Slack request must carry Slack's signature of its body, made in the last five
-	minutes; the verification token is then neither needed nor enough."""
-
-	def setUp(self):
-		"""Slack commands with a signing secret, and the worker replaced by a list of what it would have run."""
-		super().setUp()
-		self.env(slackOutgoingWebhook='https://example.com/slack', slackOutgoingToken='stok', slackSigningSecret='sss')
-		self.requests = []
-		self.patch(worker, 'process_request', self.requests.append)
-
-	def post(self, secret='sss', age=0, token='stok', headers=None, tamper=False):
-		"""Post a Slack message event signed with `secret` `age` seconds ago (or with these headers instead) and
-		return the status; tamper changes the body after signing."""
-		body = json.dumps({'token': token, 'type': 'event_callback', 'authorizations': [{'user_id': 'UBOT'}],
-			'event': {'type': 'message', 'channel': 'C1', 'user': 'U1', 'text': '.price'}}).encode()
-		timestamp = str(int(time.time()) - age)
-		mac = hmac.new(secret.encode(), b'v0:' + timestamp.encode() + b':' + body, hashlib.sha256).hexdigest()
-		if headers is None:
-			headers = {'HTTP_X_SLACK_REQUEST_TIMESTAMP': timestamp, 'HTTP_X_SLACK_SIGNATURE': 'v0=' + mac}
-		env = wsgi_environ('/slack', body.replace(b'.price', b'.trades') if tamper else body, **headers)
-		status = []
-		with contextlib.redirect_stderr(io.StringIO()):
-			bot.main(env, lambda s, *a, **k: status.append(s))
-		gevent.sleep(0.05) # the worker runs in a (gevent) thread
-		return status[-1]
-
-	def test_a_button_press_is_signed_over_its_form(self):
-		"""A button press is signed like an event, over the form it comes as; altered, it is refused."""
-		presses = []
-		self.patch(worker, 'process_callback', presses.append)
-		body = urlencode({'payload': json.dumps({**PRESS, 'token': 'not needed'})}).encode()
-		timestamp = str(int(time.time()))
-		mac = hmac.new(b'sss', b'v0:' + timestamp.encode() + b':' + body, hashlib.sha256).hexdigest()
-		headers = {'HTTP_X_SLACK_REQUEST_TIMESTAMP': timestamp, 'HTTP_X_SLACK_SIGNATURE': 'v0=' + mac}
-		for sent, status in ((body, '200 OK'), (body.replace(b'T1', b'T2'), '403 Forbidden')):
-			answers = []
-			with contextlib.redirect_stderr(io.StringIO()):
-				bot.main(wsgi_environ('/slack', sent, **headers), lambda s, *a, **k: answers.append(s))
-			gevent.sleep(0.05)
-			self.assertEqual(answers, [status])
-		self.assertEqual(presses, [inbound.Press('slack', 't1', 'C1', '1.5', 'c|h|w|T1', '@rob')])
-
-	def test_a_signed_request_is_answered_and_the_token_is_not_needed(self):
-		"""A request signed with the secret is answered, whatever token it carries."""
-		self.assertEqual([self.post(), self.post(token='wrong')], ['200 OK'] * 2)
-		self.assertEqual(len(self.requests), 2)
-
-	def test_unsigned_wrongly_signed_old_or_altered_requests_are_refused(self):
-		"""No signature (the right token is not enough), a wrong secret, a timestamp more than five minutes away or
-		not a number, and a body changed after signing are all refused."""
-		for case in (dict(headers={}), dict(secret='wrong'), dict(age=400), dict(age=-400), dict(tamper=True),
-				dict(headers={'HTTP_X_SLACK_REQUEST_TIMESTAMP': 'soon', 'HTTP_X_SLACK_SIGNATURE': 'v0=00'})):
-			with self.subTest(**case):
-				self.assertEqual(self.post(**case), '403 Forbidden')
-		self.assertEqual((self.requests, self.errors), ([], []))
-
-
-class UserTextIsEscaped(FinbotCase):
-	"""Replies go out as Telegram HTML: text a user wrote is shown, never read as markup. One stray < in a .plan entry
-	made Telegram refuse that reply and every .plan reply after it."""
-
-	def test_a_plan_is_shown_as_text(self):
-		"""A .plan entry is shown as text on Telegram and on Slack, escaped once, and a Slack mention stays markup."""
-		reply = self.command('.plan buy <cheap> & hold')[0][1]
-		self.assertIn('<i>buy &lt;cheap&gt; &amp; hold</i>', reply[0])
-		self.assertEqual(self.command('.plan')[0][1], reply) # and the same way again
-		self.sent.clear()
-		# inbound has unescaped Slack's text
-		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.plan sell <dear>', BOT, 'U'))
-		text = '\n'.join(self.sent[0][1])
-		self.assertIn('*<@U1>*', text) # a Slack mention stays markup
-		self.assertIn('_sell &lt;dear&gt;_', text) # escaped once, not twice
-		self.assertIn('_buy &lt;cheap&gt; &amp; hold_', text)
-
-	def test_a_telegram_name_is_escaped_where_it_arrives(self):
-		"""A Telegram first name is escaped as the update arrives, in the handle and the name."""
-		requests = []
-		self.patch(worker, 'process_request', requests.append)
-		body = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A<b>&'}, 'text': '.hello'}}
-		env = wsgi_environ('/telegram', json.dumps(body).encode(), HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN='tok')
-		with contextlib.redirect_stderr(io.StringIO()):
-			bot.main(env, lambda *a, **k: None)
-		gevent.sleep(0.05) # the worker runs in a (gevent) thread
-		self.assertEqual((requests[0].user, requests[0].user_name), ('@A&lt;b&gt;&amp;',) * 2) # the handle and the name
-
-
 class SecretsStayOutOfLogs(FinbotCase):
 	"""A failed request's error carries its URL: the Telegram token, a Slack or Discord webhook path, a Yahoo crumb."""
 	SECRET_URL = 'https://api.telegram.org/bot123456:SECRET-tok_en/sendMessage?chat_id=1'
diff --git a/tests/test_inbound.py b/tests/test_inbound.py
new file mode 100644
index 00000000..71641921
--- /dev/null
+++ b/tests/test_inbound.py
@@ -0,0 +1,351 @@
+"""What reaches the bot: webhook requests turned away or taken, Slack's events, presses and signatures,
+and user text escaped where it arrives."""
+import contextlib, hashlib, hmac, io, json, time, unittest
+from urllib.parse import urlencode
+import gevent
+from tests.common import *
+
+
+PRESS = {'type': 'block_actions', 'token': 'stok', 'trigger_id': 't1', 'user': {'id': 'U1', 'username': 'rob'},
+	'container': {'type': 'message', 'message_ts': '1.5', 'channel_id': 'C1'},
+	'actions': [{'action_id': 'c|h|w|T1', 'value': 'c|h|w|T1', 'type': 'button'}]} # a Slack period button pressed
+
+
+class WsgiRejections(FinbotCase):
+	SECRET = {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'tok'}
+
+	def request(self, path='/telegram', body=None, headers=None):
+		"""(the body, the status) the WSGI handler answers a request with."""
+		body = body if isinstance(body, bytes) else json.dumps(body if body is not None else {}).encode()
+		env = wsgi_environ(path, body, **(headers or {}))
+		status = []
+		with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
+			out = bot.main(env, lambda s, *a, **k: status.append(s))
+		return b''.join(out), status[-1]
+
+	def test_telegram_requires_the_secret_header(self):
+		"""A Telegram update without the secret header, or with a wrong one, is refused and not processed."""
+		message = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.help'}}
+		self.assertIn(b'Unauthorized', self.request(body=message)[0])
+		self.assertIn(b'Unauthorized',
+			self.request(body=message, headers={'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'wrong'})[0])
+		self.assertEqual(self.sent, []) # nothing was processed
+
+	def test_an_empty_secret_proves_nothing(self):
+		"""With no secret configured, a request carrying an empty one is refused, on Telegram and on Slack."""
+		self.env(telegramOutgoingToken='', slackOutgoingWebhook='https://example.com/slack', slackOutgoingToken='')
+		message = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.help'}}
+		empty = {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': ''}
+		self.assertEqual(self.request(body=message, headers=empty)[1], '403 Forbidden')
+		self.assertEqual(self.request('/slack', {'token': '', 'type': 'url_verification', 'challenge': 'c'})[1],
+			'403 Forbidden')
+		self.assertEqual((self.sent, self.errors), ([], []))
+
+	def test_unknown_path_is_404(self):
+		"""An unknown path is 404."""
+		self.assertEqual(self.request(path='/nope')[1], '404 Not Found')
+
+	def test_malformed_body_is_refused_not_a_crash(self):
+		"""A body that is not a JSON object is 400 once authenticated, and turned away unread before that."""
+		# no secret: turned away before the body is read
+		self.assertEqual(self.request(body=b'not json')[1], '403 Forbidden')
+		for body in (b'not json', b'', b'[1, 2]', b'7'):
+			self.assertEqual(self.request(body=body, headers=self.SECRET)[1], '400 Bad Request', body)
+		self.assertEqual(self.errors, [])
+
+	def test_only_the_body_is_read_and_a_huge_one_not_at_all(self):
+		"""Only CONTENT_LENGTH bytes are read, since a server need not end the stream there; a body over MAX_BODY is
+		refused unread, and a length that is not a number is a bad request."""
+		self.patch(worker, 'process_request', lambda *a: None)
+		def post(body, length):
+			"""(the status, the bytes read) for this body and CONTENT_LENGTH, with more on the stream after it."""
+			env = {**wsgi_environ('/telegram', body, **self.SECRET), 'CONTENT_LENGTH': length}
+			env['wsgi.input'] = stream = io.BytesIO(body + b'{"the next request": 1}')
+			status = []
+			with contextlib.redirect_stderr(io.StringIO()):
+				bot.main(env, lambda s, *a, **k: status.append(s))
+			return status[-1], stream.tell()
+		message = {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.help'}
+		update = json.dumps({'message': message}).encode()
+		self.assertEqual(post(update, str(len(update))), ('200 OK', len(update)))
+		self.assertEqual(post(update, str(bot.MAX_BODY + 1)), ('413 Payload Too Large', 0))
+		self.assertEqual(post(update, 'lots'), ('400 Bad Request', 0))
+
+	def test_strangers_cannot_post_into_the_chats(self):
+		"""Strangers (wrong paths, no or wrong secrets, malformed bodies) are refused, and nothing reaches the chats."""
+		self.env(slackOutgoingWebhook='https://example.com/slack')
+
+		self.env(slackOutgoingToken='stok')
+
+		for path, body, headers, status in (('/', b'', {}, '404 Not Found'),
+			('/wp-login.php', b'log=x&pwd=y', {}, '404 Not Found'),
+				('/telegram', b'log=x', {}, '403 Forbidden'),
+				('/telegram', {'message': {}}, {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'wrong'}, '403 Forbidden'),
+				('/slack', b'', {}, '400 Bad Request'), ('/slack', b'token=stok', {}, '400 Bad Request'),
+				('/slack', {'type': 'event_callback'}, {}, '403 Forbidden'),
+				('/slack', {'token': 'wrong', 'type': 'event_callback'}, {}, '403 Forbidden'),
+				('/slack', {'token': 'stok'}, {}, '200 OK')):
+			with self.subTest(path=path, body=body):
+				self.assertEqual(self.request(path, body, headers)[1], status)
+		self.assertEqual(self.errors, []) # nothing reached the chats
+		self.assertEqual(self.sent, [])
+
+	def test_a_failure_before_authentication_is_only_logged(self):
+		"""A failure while routing, before authentication, is logged and answered 500, never posted to the chats."""
+		def broken(*a, **k):
+			"""Fail while routing."""
+			raise RuntimeError('boom')
+		self.patch(inbound, 'urlparse', broken) # fails while routing, before the secret is even looked at
+		env = wsgi_environ('/telegram', b'{}', **self.SECRET)
+		status = []
+		with contextlib.redirect_stderr(io.StringIO()) as err:
+			bot.main(env, lambda s, *a, **k: status.append(s))
+		self.assertEqual((status[-1][:3], self.errors), ('500', []))
+		self.assertIn('boom', err.getvalue())
+
+	def test_a_reported_failure_names_the_network_not_its_secret_path(self):
+		"""The error line of a failed update names the network, not the webhook's path, which is a secret."""
+		self.env(telegramOutgoingWebhook='https://example.com/hook-s3cret')
+		broken = {'message': {'message_id': 1, 'from': {'first_name': 'A'}, 'text': '.help'}} # no chat
+		self.assertEqual(self.request('/hook-s3cret', broken, self.SECRET)[1], '200 OK')
+		self.assertEqual(self.errors, ["⚠️ telegram: KeyError: 'chat'"])
+
+	def test_a_failure_after_authentication_is_reported_once_and_not_retried(self):
+		"""A failure on an authenticated update is reported once and answered 200, so Telegram doesn't send it again."""
+		broken = {'message': {'message_id': 1, 'from': {'first_name': 'A'}, 'text': '.help'}} # no chat
+		# a retry would only fail and report again
+		self.assertEqual(self.request(body=broken, headers=self.SECRET)[1], '200 OK')
+		self.assertEqual(len(self.errors), 1)
+		self.assertIn("KeyError: 'chat'", self.errors[0])
+
+	def test_telegram_edge_updates(self):
+		"""Edits, button presses without a message, and messages without text are answered and left alone."""
+		secret = self.SECRET
+		self.assertEqual(self.request(body={'edited_message': {}}, headers=secret)[0], b'Unsupported')
+		self.assertEqual(self.request(body={'callback_query': {'id': '1'}}, headers=secret)[0], b'Unsupported')
+		odd = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'new_chat_members': []}}
+		self.assertIn(b'Unhandled', self.request(body=odd, headers=secret)[0])
+
+	def test_a_telegram_update_sent_again_runs_once(self):
+		"""An update Telegram sends again runs once."""
+		calls = []
+		self.patch(worker, 'process_request', lambda *a: calls.append(a))
+		body = {'update_id': 7,
+			'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A'}, 'text': '.price'}}
+		for _ in range(2): # Telegram sends an update again when it got no answer
+			self.assertEqual(self.request(body=body, headers=self.SECRET)[1], '200 OK')
+		gevent.sleep(0.05)
+		self.assertEqual(len(calls), 1)
+
+	def test_an_update_is_taken_however_it_arrived(self):
+		"""Reading an update needs no webhook: one handed over another way (as a pull mode would) reaches the worker as
+		from the webhook, Telegram's message and Slack's alike, and a repeat runs once."""
+		got = []
+		self.patch(worker, 'process_request', got.append)
+		message = {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A', 'username': 'a'}, 'text': '.price'}
+		event = {'type': 'event_callback', 'event_id': 'Ev9', 'authorizations': [{'user_id': 'UBOT'}],
+			'event': {'type': 'message', 'channel': 'C1', 'user': 'U1', 'text': '.trades', 'ts': '1.2'}}
+		with contextlib.redirect_stderr(io.StringIO()):
+			for _ in range(2):
+				self.assertEqual(bot.take(inbound.TELEGRAM, {'update_id': 70, 'message': message}), inbound.OK)
+				self.assertEqual(bot.take(inbound.SLACK, event), inbound.OK)
+		gevent.sleep(0.05) # the worker runs in a (gevent) thread
+		self.assertEqual([(m.service, m.chat_id, m.text) for m in got], [('telegram', '5', '.price'),
+			('slack', 'C1', '.trades')])
+
+	def test_button_presses_are_logged(self):
+		"""A button press is logged with who pressed it (handle, first name or unknown), the chat and the button."""
+		calls = []
+		self.patch(worker, 'process_callback', lambda *a: calls.append(a))
+		secret = {'HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN': 'tok'}
+		for sender, who in (({'username': 'rob', 'first_name': 'Rob'}, '@rob'), ({'first_name': 'Ann'}, 'Ann'),
+				({}, 'unknown')):
+			body = json.dumps({'callback_query': {'id': '1', 'from': sender, 'data': 'j|7|abcd',
+				'message': {'message_id': 9, 'chat': {'id': 55}}}}).encode()
+			err = io.StringIO()
+			with contextlib.redirect_stderr(err):
+				bot.main(wsgi_environ('/telegram', body, **secret), lambda *a, **k: None)
+			self.assertIn(f'[Telegram button]: {who} chat 55 j|7|abcd', err.getvalue())
+
+
+class SlackEvents(FinbotCase):
+	"""Slack sends every message in a subscribed channel, the bot's own posts included."""
+
+	def setUp(self):
+		"""Slack commands configured, and the worker replaced by a list of what it would have run."""
+		super().setUp()
+		self.env(slackOutgoingWebhook='https://example.com/slack')
+
+		self.env(slackOutgoingToken='stok')
+
+		self.requests = []
+		self.patch(worker, 'process_request', self.requests.append)
+
+	def event(self, event_id=None, **event):
+		"""Post a Slack message event to the WSGI handler and return the status."""
+		body = json.dumps({'token': 'stok', 'type': 'event_callback', 'authorizations': [{'user_id': 'UBOT'}],
+			'event_id': event_id, 'event': dict({'type': 'message', 'channel': 'C1', 'ts': '1.2'}, **event)}).encode()
+		status = []
+		with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
+			bot.main(wsgi_environ('/slack', body), lambda s, *a, **k: status.append(s))
+		gevent.sleep(0.05) # the worker runs in a (gevent) thread
+		return status[-1]
+
+	def press(self, payload):
+		"""Post a button press as Slack's interactivity does (a form, the JSON in its payload field): the status."""
+		status = []
+		with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
+			body = urlencode({'payload': json.dumps(payload)}).encode()
+			bot.main(wsgi_environ('/slack', body), lambda s, *a, **k: status.append(s))
+		gevent.sleep(0.05) # the worker runs in a (gevent) thread
+		return status[-1]
+
+	def test_a_button_press_reaches_the_worker(self):
+		"""A period button pressed under a chart reaches the worker, posted to the events URL: the channel, the message
+		and the button's value. A press with a wrong token is refused, and one not under a message goes nowhere."""
+		presses = []
+		self.patch(worker, 'process_callback', presses.append)
+		self.assertEqual(self.press(PRESS), '200 OK')
+		self.assertEqual(presses, [inbound.Press('slack', 't1', 'C1', '1.5', 'c|h|w|T1', '@rob')])
+		self.assertEqual(self.press({**PRESS, 'token': 'wrong'}), '403 Forbidden')
+		self.assertEqual(self.press({**PRESS, 'container': {'type': 'view', 'view_id': 'V1'}}), '200 OK')
+		self.assertEqual(self.press({**PRESS, 'actions': []}), '200 OK')
+		self.assertEqual(len(presses), 1)
+
+	def test_the_message_goes_to_the_worker_with_its_ts(self):
+		"""A message reaches the worker with its ts, the message's id: finbot puts ⏳ on it while it works."""
+		self.assertEqual(self.event(user='U1', text='.price'), '200 OK')
+		self.assertEqual(self.requests[0].message_id, '1.2')
+
+	def test_only_people_are_answered(self):
+		"""Messages from bots, this bot included, joins, edits and events without a user are not answered; people's
+		messages are, with mentions turned into @names."""
+		# a webhook report
+		for event in ({'subtype': 'bot_message', 'bot_id': 'B1', 'text': "⚠️ /slack: KeyError: 'user'"},
+					  {'user': 'UBOT', 'bot_id': 'B2', 'text': '.holdings: Please try again'}, # this bot's own reply
+					  {'user': 'UBOT', 'text': '.holdings'},
+					  {'subtype': 'channel_join', 'user': 'U1', 'text': '<@U1> has joined the channel'},
+					  {'subtype': 'message_changed', 'message': {'text': '.price'}},
+					  {'text': '.price'}): # no user at all
+			with self.subTest(event=event):
+				self.assertEqual(self.event(**event), '200 OK')
+		self.assertEqual((self.requests, self.errors), ([], []))
+		self.assertEqual(self.event(user='U1', text='.price <@UBOT>'), '200 OK')
+		self.assertEqual(self.event(user='U2', subtype='thread_broadcast', text='.trades'), '200 OK')
+		self.assertEqual([(r.service, r.chat_id, r.user, r.text, r.bot_name) for r in self.requests],
+						 [('slack', 'C1', '<@U1>', '.price @UBOT', '@UBOT'),
+						 	('slack', 'C1', '<@U2>', '.trades', '@UBOT')])
+
+	def test_links_and_escapes_lose_their_markup(self):
+		"""Each link loses Slack's markup, however many there are (Slack links .AX tickers as domains), and Slack's
+		escapes become characters again: .buy &amp; sell runs both reports."""
+		for text, expected in (('.compare <http://BHP.AX|BHP.AX> <http://CBA.AX|CBA.AX>', '.compare BHP.AX CBA.AX'),
+				('.price <https://example.com>', '.price https://example.com'), ('.buy &amp; sell', '.buy & sell'),
+				('.plan buy &lt;BHP&gt;', '.plan buy <BHP>')):
+			with self.subTest(text=text):
+				self.assertEqual(self.event(user='U1', text=text), '200 OK')
+				self.assertEqual(self.requests[-1].text, expected)
+
+	def test_a_retried_event_runs_once(self):
+		"""An event Slack sends again (same event_id) runs once."""
+		for _ in range(2): # Slack resends an event it got no answer to within 3 seconds, with the same event_id
+			self.assertEqual(self.event('Ev1', user='U1', text='.price'), '200 OK')
+		self.assertEqual(self.event('Ev2', user='U1', text='.price'), '200 OK')
+		self.assertEqual(len(self.requests), 2)
+
+	def test_slack_does_not_change_the_telegram_bot_name(self):
+		"""A Slack event leaves the Telegram bot's name alone."""
+		self.patch(inbound.TELEGRAM, 'bot_name', '@finbot_tg')
+		self.event(user='U1', text='.price')
+		self.assertEqual(inbound.TELEGRAM.bot_name, '@finbot_tg')
+
+
+class SlackSigning(FinbotCase):
+	"""With slackSigningSecret, a Slack request must carry Slack's signature of its body, made in the last five
+	minutes; the verification token is then neither needed nor enough."""
+
+	def setUp(self):
+		"""Slack commands with a signing secret, and the worker replaced by a list of what it would have run."""
+		super().setUp()
+		self.env(slackOutgoingWebhook='https://example.com/slack', slackOutgoingToken='stok', slackSigningSecret='sss')
+		self.requests = []
+		self.patch(worker, 'process_request', self.requests.append)
+
+	def post(self, secret='sss', age=0, token='stok', headers=None, tamper=False):
+		"""Post a Slack message event signed with `secret` `age` seconds ago (or with these headers instead) and
+		return the status; tamper changes the body after signing."""
+		body = json.dumps({'token': token, 'type': 'event_callback', 'authorizations': [{'user_id': 'UBOT'}],
+			'event': {'type': 'message', 'channel': 'C1', 'user': 'U1', 'text': '.price'}}).encode()
+		timestamp = str(int(time.time()) - age)
+		mac = hmac.new(secret.encode(), b'v0:' + timestamp.encode() + b':' + body, hashlib.sha256).hexdigest()
+		if headers is None:
+			headers = {'HTTP_X_SLACK_REQUEST_TIMESTAMP': timestamp, 'HTTP_X_SLACK_SIGNATURE': 'v0=' + mac}
+		env = wsgi_environ('/slack', body.replace(b'.price', b'.trades') if tamper else body, **headers)
+		status = []
+		with contextlib.redirect_stderr(io.StringIO()):
+			bot.main(env, lambda s, *a, **k: status.append(s))
+		gevent.sleep(0.05) # the worker runs in a (gevent) thread
+		return status[-1]
+
+	def test_a_button_press_is_signed_over_its_form(self):
+		"""A button press is signed like an event, over the form it comes as; altered, it is refused."""
+		presses = []
+		self.patch(worker, 'process_callback', presses.append)
+		body = urlencode({'payload': json.dumps({**PRESS, 'token': 'not needed'})}).encode()
+		timestamp = str(int(time.time()))
+		mac = hmac.new(b'sss', b'v0:' + timestamp.encode() + b':' + body, hashlib.sha256).hexdigest()
+		headers = {'HTTP_X_SLACK_REQUEST_TIMESTAMP': timestamp, 'HTTP_X_SLACK_SIGNATURE': 'v0=' + mac}
+		for sent, status in ((body, '200 OK'), (body.replace(b'T1', b'T2'), '403 Forbidden')):
+			answers = []
+			with contextlib.redirect_stderr(io.StringIO()):
+				bot.main(wsgi_environ('/slack', sent, **headers), lambda s, *a, **k: answers.append(s))
+			gevent.sleep(0.05)
+			self.assertEqual(answers, [status])
+		self.assertEqual(presses, [inbound.Press('slack', 't1', 'C1', '1.5', 'c|h|w|T1', '@rob')])
+
+	def test_a_signed_request_is_answered_and_the_token_is_not_needed(self):
+		"""A request signed with the secret is answered, whatever token it carries."""
+		self.assertEqual([self.post(), self.post(token='wrong')], ['200 OK'] * 2)
+		self.assertEqual(len(self.requests), 2)
+
+	def test_unsigned_wrongly_signed_old_or_altered_requests_are_refused(self):
+		"""No signature (the right token is not enough), a wrong secret, a timestamp more than five minutes away or
+		not a number, and a body changed after signing are all refused."""
+		for case in (dict(headers={}), dict(secret='wrong'), dict(age=400), dict(age=-400), dict(tamper=True),
+				dict(headers={'HTTP_X_SLACK_REQUEST_TIMESTAMP': 'soon', 'HTTP_X_SLACK_SIGNATURE': 'v0=00'})):
+			with self.subTest(**case):
+				self.assertEqual(self.post(**case), '403 Forbidden')
+		self.assertEqual((self.requests, self.errors), ([], []))
+
+
+class UserTextIsEscaped(FinbotCase):
+	"""Replies go out as Telegram HTML: text a user wrote is shown, never read as markup. One stray < in a .plan entry
+	made Telegram refuse that reply and every .plan reply after it."""
+
+	def test_a_plan_is_shown_as_text(self):
+		"""A .plan entry is shown as text on Telegram and on Slack, escaped once, and a Slack mention stays markup."""
+		reply = self.command('.plan buy <cheap> & hold')[0][1]
+		self.assertIn('<i>buy &lt;cheap&gt; &amp; hold</i>', reply[0])
+		self.assertEqual(self.command('.plan')[0][1], reply) # and the same way again
+		self.sent.clear()
+		# inbound has unescaped Slack's text
+		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.plan sell <dear>', BOT, 'U'))
+		text = '\n'.join(self.sent[0][1])
+		self.assertIn('*<@U1>*', text) # a Slack mention stays markup
+		self.assertIn('_sell &lt;dear&gt;_', text) # escaped once, not twice
+		self.assertIn('_buy &lt;cheap&gt; &amp; hold_', text)
+
+	def test_a_telegram_name_is_escaped_where_it_arrives(self):
+		"""A Telegram first name is escaped as the update arrives, in the handle and the name."""
+		requests = []
+		self.patch(worker, 'process_request', requests.append)
+		body = {'message': {'message_id': 1, 'chat': {'id': 5}, 'from': {'first_name': 'A<b>&'}, 'text': '.hello'}}
+		env = wsgi_environ('/telegram', json.dumps(body).encode(), HTTP_X_TELEGRAM_BOT_API_SECRET_TOKEN='tok')
+		with contextlib.redirect_stderr(io.StringIO()):
+			bot.main(env, lambda *a, **k: None)
+		gevent.sleep(0.05) # the worker runs in a (gevent) thread
+		self.assertEqual((requests[0].user, requests[0].user_name), ('@A&lt;b&gt;&amp;',) * 2) # the handle and the name
+
+
+if __name__ == '__main__':
+	unittest.main()
diff --git a/tests/test_providers.py b/tests/test_providers.py
deleted file mode 100644
index aa492a59..00000000
--- a/tests/test_providers.py
+++ /dev/null
@@ -1,1268 +0,0 @@
-"""Yahoo, Sharesight and Telegram parsing and failure handling, driven by the real API responses saved under doc/.
-HTTP is mocked at lib.net, so the production parsing code runs unchanged."""
-import contextlib, datetime, io, json, os, subprocess, threading, unittest
-import gevent
-from unittest import mock
-from tests.common import *
-
-DOC = os.path.join(ROOT, 'doc')
-
-
-def sample(*path):
-	"""A saved API response from doc/."""
-	with open(os.path.join(DOC, *path), encoding='utf-8') as f:
-		return json.load(f)
-
-
-class Resp:
-	def __init__(self, payload=None, status=200, text=''):
-		"""A response with this JSON payload (None: not JSON), status and text."""
-		self.payload, self.status_code, self.text = payload, status, text or json.dumps(payload)
-	def json(self):
-		"""The payload, or ValueError when it is not JSON."""
-		if self.payload is None:
-			raise ValueError('no json')
-		return self.payload
-
-
-def serve(*responses):
-	"""net.get replacement: replays the responses in order (exceptions are raised); the last one repeats."""
-	queue = list(responses)
-	calls = []
-	def get(url, **kwargs):
-		"""The next response, or raise it when it is an exception."""
-		calls.append(url)
-		item = queue.pop(0) if len(queue) > 1 else queue[0]
-		if isinstance(item, Exception):
-			raise item
-		return item
-	get.calls = calls
-	return get
-
-
-REAL_CREDENTIALS = yahoo.credentials
-CHART = render.Chart('bars', 'Heading', rows=(('A', 1.0),)) # a chart for the sending tests, which draw it as b'png'
-REAL_TOKEN = sharesight._token
-
-
-class ProviderCase(FinbotCase):
-	def setUp(self):
-		"""Real parsers instead of the fixture's canned data, with a fixed cookie, crumb and token, and no waiting."""
-		super().setUp()
-		# these tests exercise the real parsers, so undo the canned data from the base fixture
-		for module, name in ((yahoo, 'fetch'), (yahoo, 'fetch_detail'), (yahoo, 'price_series'),
-				(sharesight, '_portfolios'), (sharesight, '_trades'), (sharesight, '_performance')):
-			self.patch(module, name, REAL[(module.__name__, name)])
-		self.patch(yahoo, 'credentials', lambda renew=False: ('A=B', 'crumb'))
-		self.patch(sharesight, '_token', lambda: 'token')
-		self.patch(sharesight.time, 'sleep', lambda s: None)
-
-	def http(self, module, *responses):
-		"""Have module's net.get replay these responses; returns the stand-in, whose calls lists the URLs asked for."""
-		getter = serve(*responses)
-		self.patch(module.net, 'get', getter)
-		return getter
-
-
-class YahooQuotes(ProviderCase):
-	def test_parses_real_v7_quote(self):
-		"""A real v7 quote parses into finbot's fields."""
-		self.http(yahoo, Resp(sample('yahoo', 'sample_yahoo_v7.json')))
-		data = yahoo.fetch(['AAPL'])['AAPL']
-		self.assertEqual(data['profile_title'], 'Apple')
-		self.assertEqual(data['percent_change'], 0.09)
-		self.assertEqual(data['percent_change_postmarket'], -0.13)
-		self.assertEqual(data['quoteType'], 'EQUITY')
-		self.assertEqual(data['marketState'], 'CLOSED')
-		self.assertEqual(data['profile_exchange'], 'NasdaqGS')
-		self.assertGreater(data['market_cap'], 1e12)
-
-	def test_second_call_is_served_from_cache(self):
-		"""A second fetch of the same ticker comes from the cache."""
-		getter = self.http(yahoo, Resp(sample('yahoo', 'sample_yahoo_v7.json')))
-		yahoo.fetch(['AAPL'])
-		yahoo.fetch(['AAPL'])
-		self.assertEqual(len(getter.calls), 1)
-
-	def test_falls_back_to_second_host(self):
-		"""When the first Yahoo host fails, the second is asked."""
-		getter = self.http(yahoo, Resp(status=500), Resp(sample('yahoo', 'sample_yahoo_v7.json')))
-		self.assertIn('AAPL', yahoo.fetch(['AAPL']))
-		self.assertIn('query1', getter.calls[1])
-
-	def test_all_hosts_failing_raises(self):
-		"""Every host failing, or Yahoo unreachable, is a RuntimeError."""
-		for failure in (Resp(status=500), ConnectionError('down')):
-			with self.subTest(failure=failure):
-				self.http(yahoo, failure)
-				with self.assertRaises(RuntimeError):
-					yahoo.fetch(['AAPL'])
-
-	def test_empty_result_is_none(self):
-		"""An empty result is no quotes."""
-		self.http(yahoo, Resp({'quoteResponse': {'result': None, 'error': {'code': 'x'}}}))
-		self.assertEqual(yahoo.fetch(['NOPE']), {})
-
-	def test_an_error_answer_is_not_cached(self):
-		"""Yahoo's error answer is no quotes, and is not cached: the next fetch asks again."""
-		error = Resp({'quoteResponse': {'result': None, 'error': {'code': 'x'}}})
-		self.http(yahoo, error, Resp(sample('yahoo', 'sample_yahoo_v7.json')))
-		self.assertEqual(yahoo.fetch(['AAPL']), {})
-		self.assertIn('AAPL', yahoo.fetch(['AAPL']))
-
-	def test_the_cache_holds_yahoos_answer_so_a_parser_fix_needs_no_flush(self):
-		"""Quotes are cached as Yahoo sent them, not parsed: a change to the parsing applies to a cached answer."""
-		body = sample('yahoo', 'sample_yahoo_v7.json')
-		self.http(yahoo, Resp(body))
-		yahoo.fetch(['AAPL'])
-		self.http(yahoo, ConnectionError('down')) # from here on, only the cache answers
-		self.patch(yahoo, 'QUOTE_FIELDS', yahoo.QUOTE_FIELDS + (('short_name', 'shortName', None),))
-		self.assertEqual(yahoo.fetch(['AAPL'])['AAPL']['short_name'], 'Apple Inc.')
-
-	def test_with_the_cache_off_every_fetch_asks_yahoo(self):
-		"""With the cache preference off, every fetch asks Yahoo, and nothing is written to the cache."""
-		self.prefs(cache=False)
-		getter = self.http(yahoo, Resp(sample('yahoo', 'sample_yahoo_v7.json')))
-		yahoo.fetch(['AAPL'])
-		yahoo.fetch(['AAPL'])
-		self.assertEqual(len(getter.calls), 2)
-		self.assertEqual([name for name in os.listdir(config.env.cache_dir) if name.startswith('finbot_yahoo_')], [])
-
-	def test_no_tickers_is_none(self):
-		"""No tickers asked for: no quotes, and no request."""
-		self.assertEqual(yahoo.fetch([]), {})
-
-	def test_incomplete_quotes_are_skipped(self):
-		"""A quote without a price or a name is left out."""
-		body = sample('yahoo', 'sample_yahoo_v7.json')
-		good = body['quoteResponse']['result'][0]
-		no_price = {k: v for k, v in good.items() if k != 'regularMarketPrice'} | {'symbol': 'NOPRICE'}
-		no_name = {k: v for k, v in good.items() if k not in ('longName', 'shortName')} | {'symbol': 'NONAME'}
-		body['quoteResponse']['result'] = [good, no_price, no_name]
-		self.http(yahoo, Resp(body))
-		self.assertEqual(list(yahoo.fetch(['AAPL', 'NOPRICE', 'NONAME'])), ['AAPL'])
-
-
-class YahooDetail(ProviderCase):
-	@staticmethod
-	def combined():
-		"""quoteSummary as Yahoo returns it for several modules at once, assembled from the per-module samples."""
-		merged = {}
-		for module in ('price', 'summaryProfile', 'summaryDetail', 'defaultKeyStatistics', 'financialData',
-				'calendarEvents', 'earnings'):
-			merged.update(sample('yahoo', f'sample_yahoo_v10_{module}.json')['quoteSummary']['result'][0])
-		return {'quoteSummary': {'result': [merged], 'error': None}}
-
-	def test_parses_real_quote_summary(self):
-		"""A real quoteSummary parses, the company's description included."""
-		self.http(yahoo, Resp(self.combined()))
-		data = yahoo.fetch_detail('NVDA')['NVDA']
-		self.assertEqual(data['profile_title'], 'NVIDIA')
-		self.assertTrue(data['profile_bio'])
-
-	def test_a_cached_detail_needs_no_cookie(self):
-		"""A detail served from the cache takes neither the cookie nor the crumb, nor their lock."""
-		self.http(yahoo, Resp(self.combined()))
-		yahoo.fetch_detail('NVDA')
-		def refuse():
-			"""Fail: the credentials must not be needed."""
-			raise AssertionError('a cached detail took the Yahoo auth lock')
-		self.patch(yahoo, 'credentials', refuse)
-		self.assertIn('NVDA', yahoo.fetch_detail('NVDA'))
-
-	def test_the_cache_holds_yahoos_answer_so_a_parser_fix_needs_no_flush(self):
-		"""A detail is cached as Yahoo sent it, not parsed: a change to the parsing applies to a cached answer."""
-		self.http(yahoo, Resp(self.combined()))
-		yahoo.fetch_detail('NVDA')
-		self.http(yahoo, ConnectionError('down')) # from here on, only the cache answers
-		self.patch(yahoo, 'DETAIL_FIELDS', yahoo.DETAIL_FIELDS + (('short_name', ('price', 'shortName')),))
-		self.assertEqual(yahoo.fetch_detail('NVDA')['NVDA']['short_name'], 'NVIDIA Corporation')
-
-	def test_a_symbol_yahoo_does_not_know_is_empty_and_an_outage_an_error(self):
-		"""A detail of a symbol Yahoo does not know (404 Not Found) is empty, an answer; Yahoo unreachable or refusing
-		is an error, where it used to read as an unknown symbol ("I could not find TSLA")."""
-		unknown = {'quoteSummary': {'result': None, 'error': {'code': 'Not Found', 'description': 'Quote not found'}}}
-		self.http(yahoo, Resp(unknown, status=404))
-		self.assertEqual(yahoo.fetch_detail('DRNA'), {})
-		for failure in (Resp(status=404), Resp(status=500), ConnectionError('down')):
-			with self.subTest(failure=failure):
-				self.http(yahoo, failure)
-				with self.assertRaises(RuntimeError):
-					yahoo.fetch_detail('TSLA')
-
-	def test_missing_names_give_empty(self):
-		"""A detail without a name, or without a result, is empty."""
-		body = self.combined()
-		body['quoteSummary']['result'][0]['price'].update(longName=None, shortName=None)
-		self.http(yahoo, Resp(body))
-		self.assertEqual(yahoo.fetch_detail('DUB'), {})
-		self.http(yahoo, Resp({'quoteSummary': {'result': None, 'error': {}}}))
-		self.assertEqual(yahoo.fetch_detail('GONE'), {})
-
-
-class YahooChart(ProviderCase):
-	def test_real_chart_to_bars(self):
-		"""A real v8 chart becomes its daily bars: the day, the open and the close."""
-		rows = yahoo.bars(sample('yahoo', 'sample_yahoo_v8_chart.json'))
-		self.assertEqual(len(rows), 2)
-		self.assertTrue(all(isinstance(day, datetime.date) and close for day, _, close in rows))
-
-	def test_a_bar_is_dated_in_its_exchanges_time_zone(self):
-		"""An ASX bar, stamped at Sydney's open (23:00 UTC the day before), is dated the day it traded: it used to be
-		dated the day before, so N-day changes and chart dates slipped a session."""
-		open_sydney = datetime.datetime(2026, 1, 5, 23, 0, tzinfo=datetime.timezone.utc) # 10:00 on 6 Jan in Sydney
-		chart = {'chart': {'result': [{'meta': {'exchangeTimezoneName': 'Australia/Sydney'},
-			'timestamp': [int(open_sydney.timestamp())], 'indicators': {'quote': [{'open': [1.0], 'close': [2.0]}]}}]}}
-		self.assertEqual(yahoo.bars(chart), [(datetime.date(2026, 1, 6), 1.0, 2.0)])
-
-	def test_a_refused_crumb_is_renewed_once(self):
-		"""Yahoo refusing the crumb (401) gets a new cookie and crumb, and the request is asked again, once."""
-		renewed = []
-		fresh = lambda renew=False: renewed.append(renew) or ('A=B', 'new' if renew else 'old')
-		self.patch(yahoo, 'credentials', fresh)
-		getter = self.http(yahoo, Resp(status=401), Resp(status=401), Resp(sample('yahoo', 'sample_yahoo_v7.json')))
-		self.assertIn('AAPL', yahoo.fetch(['AAPL']))
-		crumbs = [url.rsplit('crumb=', 1)[1] for url in getter.calls]
-		self.assertEqual((renewed, crumbs), ([False, True], ['old', 'old', 'new']))
-		self.http(yahoo, Resp(status=401))
-		with self.assertRaises(RuntimeError): # a new crumb refused too: an error, not a loop
-			yahoo.fetch(['MSFT'])
-
-	def test_the_credentials_are_one_file_renewed_together(self):
-		"""The cookie and the crumb Yahoo issued for it are cached together and renewed together; no cookie, or no
-		crumb, is an error, where a stale fallback pair turned it into a refusal later."""
-		self.patch(yahoo, 'credentials', REAL_CREDENTIALS)
-		cookie = Resp(None, status=404)
-		cookie.headers = {'Set-Cookie': 'A3=d=AQAB; Max-Age=31557600; Domain=.yahoo.com; Path=/; Secure'}
-		getter = self.http(yahoo, cookie, Resp(None, text='crumb1'))
-		self.assertEqual(yahoo.credentials(), ('A3=d=AQAB', 'crumb1'))
-		self.assertEqual(yahoo.credentials(), ('A3=d=AQAB', 'crumb1')) # from the cache now
-		self.assertEqual(len(getter.calls), 2)
-		nothing = Resp(None, status=404)
-		nothing.headers = {}
-		self.http(yahoo, nothing)
-		with self.assertRaisesRegex(RuntimeError, 'no cookie'):
-			yahoo.credentials(renew=True)
-
-	def test_fetch_chart_json_failures_raise(self):
-		"""A chart Yahoo will not give is a RuntimeError."""
-		for failure in (Resp(status=404), ConnectionError('down')):
-			with self.subTest(failure=failure):
-				self.http(yahoo, failure)
-				with self.assertRaises(RuntimeError):
-					yahoo.fetch_chart_json('AAPL', full=True)
-
-	@staticmethod
-	def chart(closes, failing=()):
-		"""A v8 chart response with one daily bar per close, the last one today (14:30 UTC, New York's open)."""
-		today = datetime.datetime.now(datetime.timezone.utc).replace(hour=14, minute=30, second=0, microsecond=0)
-		stamps = [int((today - datetime.timedelta(days=len(closes) - 1 - i)).timestamp()) for i in range(len(closes))]
-		meta = {'currency': 'USD', 'symbol': 'T1', 'exchangeTimezoneName': 'America/New_York',
-			'regularMarketTime': stamps[-1],
-				'regularMarketPrice': closes[-1], 'longName': 'Co T1'}
-		quote = {'open': closes, 'high': closes, 'low': closes, 'close': closes, 'volume': [1] * len(closes)}
-		return {'chart': {'result': [{'meta': meta, 'timestamp': stamps, 'indicators': {'quote': [quote]}}],
-			'error': None}}
-
-	def test_1d_is_the_change_since_the_previous_close(self):
-		"""1D is the change since the previous close, and 5D the week's; with a quote, its price and today's change
-		count, the chart's being up to cache_seconds old."""
-		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
-		# up 20% in a week, down 3% today
-		self.http(yahoo, Resp(self.chart([90] * 30 + [100, 105, 110, 115, 120, 116.4])))
-		table, line = yahoo.price_history('T1')
-		self.assertEqual(table['1D'], -3)
-		self.assertGreater(table['5D'], 10) # the week, not the day
-		table, line = yahoo.price_history('T1', quote={'regularMarketPrice': 125.0, 'percent_change': 7.4})
-		self.assertEqual((table['1D'], line.price), (7, 125.0))
-
-	def test_a_record_is_from_the_whole_history(self):
-		"""historic_high reads the whole history (the chart from period1=0), its first day aside: records seeded
-		from 10 years were no all-time records."""
-		urls = []
-		chart = self.chart([5.0, 50.0, 10.0, 20.0])
-		self.patch(yahoo, 'historic_high', REAL[('lib.yahoo', 'historic_high')])
-		self.patch(yahoo.net, 'get', lambda url, **kwargs: urls.append(url) or Resp(chart))
-		self.assertEqual(yahoo.historic_high('T1'), (50.0, 10.0)) # 5.0, the first day, aside
-		self.assertIn('period1=0&', urls[0])
-
-	def test_a_symbol_yahoo_does_not_know_is_no_data(self):
-		"""Yahoo's 404 for a symbol it does not know is NoData, an answer; any other 404 is an error, not an answer."""
-		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
-		unknown = {'chart': {'result': None, 'error': {'code': 'Not Found', 'description': 'No data found'}}}
-		self.http(yahoo, Resp(unknown, status=404))
-		with self.assertRaises(yahoo.NoData) as caught:
-			yahoo.price_history('XYZ', 7)
-		self.assertEqual(caught.exception.lines, ['no price history for XYZ'])
-		self.http(yahoo, Resp(status=404)) # an endpoint that moved, say
-		with self.assertRaises(RuntimeError) as caught:
-			yahoo.price_history('XYZ', 7)
-		self.assertNotIsInstance(caught.exception, answer.Answer)
-
-	def test_a_failed_chart_fetch_raises(self):
-		"""A chart refused (429) or unreachable is a RuntimeError, which callers turn into leaving the ticker out."""
-		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
-		for failure in (Resp(status=429), ConnectionError('down')):
-			with self.subTest(failure=failure):
-				self.http(yahoo, failure)
-				with self.assertRaises(RuntimeError):
-					yahoo.price_history('AAPL', 7)
-
-	def test_one_failed_ticker_does_not_take_the_report_down(self):
-		"""One ticker without a chart is left out of a price list, a benchmark without one is said to have none, and
-		only a report on that one ticker fails, in one line."""
-		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
-		quotes = lambda ts: {t: market_entry(t) | {'regularMarketPrice': 110.0} for t in ts} # charts over HTTP
-		self.patch(yahoo, 'fetch', quotes)
-		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t)})
-		self.patch(sharesight, '_portfolios', lambda seconds=None: [{'name': 'Rob', 'id': 1}])
-		self.patch(sharesight, '_performance', lambda pid, days=None, seconds=None: # every ticker needs its history
-			{'holdings': [], 'capital_gain_percent': '2.5'})
-		good = Resp(self.chart([100.0] * 20 + [110.0]))
-		def get(url, **kwargs):
-			"""A chart for every ticker but T3 and SPY, which get 404."""
-			return Resp(status=404) if '/chart/T3?' in url or '/chart/SPY?' in url else good
-		self.patch(yahoo.net, 'get', get)
-		def slack(text):
-			"""Run a command on Slack, where reports come as text, and return the text."""
-			self.sent.clear(); self.errors.clear()
-			worker.process_request(inbound.Message('slack', 'C1', '<@U1>', text, BOT, 'U'))
-			self.assertEqual(self.errors, [], text)
-			return '\n'.join(map(str, self.sent[-1][1]))
-		price_list = slack('.price 7d') # T3 is left out, the rest are listed
-		self.assertNotIn('Co T3 (', price_list)
-		self.assertIn('Co T4 (', price_list)
-		report = slack('.performance 7d') # the S&P 500 benchmark has no history
-		self.assertIn('⚠️ S&P 500: no price history', report)
-		self.assertIn('NASDAQ 100', report)
-		self.command('.T3') # the profile goes out without its history lines
-		self.sent.clear(); self.errors.clear()
-		worker.process_request(inbound.Message('telegram', '55', '@u', '.price T3 7d', BOT, 'U'))
-		self.assertEqual(len(self.errors), 1) # the one ticker asked for: one line saying why
-		self.assertTrue(self.errors[0].startswith('⚠️ .price: RuntimeError: 404 error for T3'))
-
-	def test_the_chart_request_carries_the_cookie_of_its_crumb(self):
-		"""A chart request sends the cookie its crumb belongs to."""
-		headers = []
-		self.patch(yahoo.net, 'get',
-			lambda url, **kwargs: headers.append(kwargs.get('headers', {})) or Resp(self.chart([100.0] * 5)))
-		yahoo.fetch_chart_json('T1', full=True)
-		self.assertEqual(headers[0].get('Cookie'), 'A=B') # the crumb used to go without its cookie
-
-	def test_with_the_cache_off_no_chart_is_written(self):
-		"""With the cache preference off, a chart is drawn but not written to the cache."""
-		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
-		self.prefs(cache=False)
-		self.http(yahoo, Resp(self.chart([100.0] * 30 + [110.0])))
-		changes, line = yahoo.price_history('T1', 7)
-		self.assertTrue(render.draw(history.line_chart('T1', changes, line, 7)).read().startswith(b'\x89PNG'))
-		self.assertEqual([name for name in os.listdir(config.env.cache_dir) if name.startswith('finbot_chart_')], [])
-
-	def test_price_series_raises_on_failure_and_on_empty_history(self):
-		"""price_series raises when Yahoo fails and when the history has no closes."""
-		self.http(yahoo, Resp(status=500))
-		with self.assertRaises(RuntimeError):
-			yahoo.price_series('AAPL')
-		empty = sample('yahoo', 'sample_yahoo_v8_chart.json')
-		empty['chart']['result'][0]['indicators']['quote'][0]['close'] = [None, None]
-		self.http(yahoo, Resp(empty))
-		with self.assertRaises(RuntimeError):
-			yahoo.price_series('EMPTY')
-
-
-def portfolios():
-	"""{name: id} of the portfolios to report on."""
-	return {portfolio.name: portfolio.id for portfolio in sharesight.portfolios()}
-
-
-class SharesightApi(ProviderCase):
-	def test_portfolios_from_real_response(self):
-		"""The portfolio list parses from a real answer."""
-		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_portfolios.json')))
-		self.assertEqual(portfolios(), {'Rob': 445825, 'RobSMSF': 684141})
-
-	def test_portfolio_exclusion_and_inclusion(self):
-		"""exclude_portfolios and include_portfolios pick the portfolios reported on."""
-		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_portfolios.json')))
-		self.prefs(exclude_portfolios=['445825'])
-
-		self.assertEqual(portfolios(), {'RobSMSF': 684141})
-		self.prefs(exclude_portfolios=[], include_portfolios=['445825'])
-		self.assertEqual(portfolios(), {'Rob': 445825})
-
-	def test_portfolio_failures_raise(self):
-		"""An error answer, Sharesight unreachable or no portfolios is a RuntimeError."""
-		for failure in (Resp({'error': 'nope', 'error_code': 'x'}, 401), ConnectionError('down'),
-				Resp({'portfolios': []})):
-			with self.subTest(failure=failure):
-				self.http(sharesight, failure)
-				with self.assertRaises(RuntimeError):
-					sharesight.portfolios()
-
-	def test_an_answer_that_is_not_json_says_so(self):
-		"""An HTML error page from Sharesight is a RuntimeError naming the request and the status, not a JSON parse
-		error."""
-		self.http(sharesight, Resp(None, 502, '<html>Bad gateway</html>'))
-		with self.assertRaisesRegex(RuntimeError, 'Sharesight trades: a 502 answer that is not JSON'):
-			sharesight.Portfolio('RobSMSF', 684141).trades(1)
-
-	def test_trades_from_real_response(self):
-		"""Trades parse from a real answer, tagged with their portfolio, which the cache leaves out: it holds the answer
-		as sent."""
-		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_trades_684141_1.json')))
-		trades = sharesight.Portfolio('RobSMSF', 684141).trades(1)
-		self.assertEqual(len(trades), 2)
-		self.assertEqual(trades[0]['portfolio'], 'RobSMSF')
-		self.assertEqual(trades[0]['symbol'], 'FANG')
-		self.assertEqual(cache.read('finbot_sharesight_trades_684141_1.json'), sample('sharesight',
-			'finbot_sharesight_trades_684141_1.json'))
-
-	def test_trade_failures_raise(self):
-		"""Sharesight unreachable, or an error answer, is a RuntimeError."""
-		self.http(sharesight, ConnectionError('down'))
-		with self.assertRaises(RuntimeError):
-			sharesight.Portfolio('RobSMSF', 684141).trades(1)
-		self.http(sharesight, Resp({'trades': [], 'error': 'bad', 'error_code': 'x'}, 400))
-		with self.assertRaisesRegex(RuntimeError, 'Sharesight trades: bad'):
-			sharesight.Portfolio('RobSMSF', 684141).trades(2)
-
-	def test_an_oauth_error_is_reported_with_its_description(self):
-		"""Sharesight's refusal of the credentials says why, in OAuth's words (it used to be a KeyError on its missing
-		error_code)."""
-		self.patch(sharesight, '_token', REAL_TOKEN)
-		refusal = {'error': 'invalid_client', 'error_description': 'Client authentication failed'}
-		self.patch(sharesight.net, 'post', lambda url, **kwargs: Resp(refusal, 401))
-		with self.assertRaisesRegex(RuntimeError, 'Sharesight auth: invalid_client: Client authentication failed'):
-			sharesight.portfolios()
-
-	def test_an_answer_without_what_was_asked_for_is_an_error(self):
-		"""A refusal without an error, or a 200 without the list asked for, is an error saying so, not a KeyError
-		later."""
-		for answer in (Resp({'message': 'nope'}, 403), Resp({'links': {}})):
-			with self.subTest(status=answer.status_code):
-				self.http(sharesight, answer)
-				why = f'Sharesight portfolios: a {answer.status_code} answer without portfolios'
-				with self.assertRaisesRegex(RuntimeError, why):
-					sharesight.portfolios()
-
-	def test_performance_and_holdings_from_real_response(self):
-		"""A real holdings report parses: its tickers come out sorted, US ones without a suffix, and the gains are the
-		portfolio's and each holding's."""
-		sent = sample('sharesight', 'finbot_sharesight_performance_684141_0.json')
-		self.http(sharesight, Resp(sent))
-		self.assertEqual(len(sharesight._performance(684141, 0)['holdings']), 18)
-		portfolio = sharesight.Portfolio('RobSMSF', 684141)
-		tickers = portfolio.tickers()
-		self.assertIn('ARM', tickers) # US venues carry no Yahoo suffix
-		self.assertEqual(tickers, sorted(tickers))
-		self.assertEqual(portfolio.gain(0), float(sent['report']['capital_gain_percent']))
-		self.assertEqual(sorted(portfolio.gains(0)), tickers)
-		self.assertTrue(all(isinstance(gain, float) for gain in portfolio.gains(0).values()))
-
-	def test_performance_retries_when_sharesight_is_busy(self):
-		"""While Sharesight says too many requests are in flight, the request is tried again."""
-		busy = Resp({'error': 'Too many parallel requests', 'error_code': 'x'}, 429)
-		getter = self.http(sharesight, busy, busy,
-			Resp(sample('sharesight', 'finbot_sharesight_performance_684141_0.json')))
-		self.assertIn('holdings', sharesight._performance(684141, 0))
-		self.assertEqual(len(getter.calls), 3)
-
-	def test_performance_gives_up_after_repeated_busy_or_hard_errors(self):
-		"""Busy every time, a hard error, or unreachable: a RuntimeError in the end."""
-		for failure in (Resp({'error': 'Too many parallel requests', 'error_code': 'x'}, 429),
-						Resp({'error': 'forbidden', 'error_code': 'x'}, 403), ConnectionError('down')):
-			with self.subTest(failure=failure):
-				self.http(sharesight, failure)
-				with self.assertRaises(RuntimeError):
-					sharesight._performance(684141, 7)
-
-	def test_an_answer_without_a_report_says_so(self):
-		"""A performance answer without a report is an error naming the portfolio, and is not cached."""
-		self.http(sharesight, Resp({}))
-		with self.assertRaisesRegex(RuntimeError, 'performance of portfolio 684141: a 200 answer without report'):
-			sharesight._performance(684141, 7)
-		self.assertIsNone(cache.read(sharesight.performance_cache_file(684141, 7))) # and is not cached
-
-
-class SharesightKeepWarm(ProviderCase):
-	"""The background refresher keeps the portfolio list and the holdings report (days=0) cached ahead of their TTLs."""
-
-	def serve_sharesight(self):
-		"""Answer with the real portfolio list and holdings report; returns the URLs asked for."""
-		portfolios = sample('sharesight', 'finbot_sharesight_portfolios.json')
-		performance = sample('sharesight', 'finbot_sharesight_performance_684141_0.json')
-		calls = []
-		def get(url, **kwargs):
-			"""The holdings report or the portfolio list, as asked."""
-			calls.append(url)
-			return Resp(performance if '/performance' in url else portfolios)
-		self.patch(sharesight.net, 'get', get)
-		return calls
-
-	def age(self, cache_file, seconds):
-		"""Make a cache file `seconds` older."""
-		path = os.path.join(config.env.cache_dir, cache_file)
-		old = os.path.getmtime(path) - seconds
-		os.utime(path, (old, old))
-
-	def test_cold_cache_is_filled_then_left_alone(self):
-		"""A cold cache is filled (the list and each portfolio's holdings, as of today), then left alone while fresh."""
-		calls = self.serve_sharesight()
-		sharesight.warm()
-		self.assertEqual(len(calls), 3) # portfolio list + one holdings report per portfolio (2)
-		self.assertEqual(sum('/performance' in c for c in calls), 2)
-		today = 'start_date=' + datetime.date.today().isoformat()
-		self.assertTrue(all(today in c for c in calls if '/performance' in c)) # days=0
-		sharesight.warm()
-		self.assertEqual(len(calls), 3) # nothing refetched while fresh
-
-	def test_refreshes_before_the_ttl_runs_out(self):
-		"""A cache file at 95% of its lifetime is fetched again, and only that one."""
-		calls = self.serve_sharesight()
-		sharesight.warm()
-		calls.clear()
-		# 95% of its lifetime
-		self.age(sharesight.performance_cache_file(684141, 0), 0.95 * config.prefs.performance_cache_seconds)
-		sharesight.warm()
-		self.assertEqual(len(calls), 1)
-		self.assertIn('684141/performance', calls[0])
-		calls.clear()
-		self.age(sharesight.PORTFOLIOS_CACHE, 0.95 * config.prefs.cache_seconds)
-		sharesight.warm()
-		self.assertEqual(len(calls), 1)
-		self.assertTrue(calls[0].endswith('/portfolios'))
-
-	def test_a_request_never_waits_on_a_cache_the_job_has_warmed(self):
-		"""Once warmed, the list and the holdings report need no request."""
-		calls = self.serve_sharesight()
-		sharesight.warm()
-		calls.clear()
-		self.assertEqual(portfolios(), {'Rob': 445825, 'RobSMSF': 684141})
-		self.assertEqual(len(sharesight._performance(684141, 0)['holdings']), 18)
-		self.assertEqual(calls, [])
-
-	def test_every_holding_is_served_from_the_warmed_cache(self):
-		"""Every portfolio's tickers come from the warmed cache, with no request."""
-		calls = self.serve_sharesight()
-		sharesight.warm()
-		calls.clear()
-		tickers = sharesight.tickers() # the portfolio list and each portfolio's holdings report, no cache of its own
-		self.assertIn('ARM', tickers)
-		self.assertEqual(calls, [])
-
-	def test_failed_refresh_keeps_the_old_cache(self):
-		"""A refresh that fails keeps the cache it would have replaced."""
-		self.serve_sharesight()
-		sharesight.warm()
-		self.age(sharesight.PORTFOLIOS_CACHE, 0.95 * config.prefs.cache_seconds)
-		self.http(sharesight, ConnectionError('down'))
-		with self.assertRaises(RuntimeError):
-			sharesight.warm()
-		self.assertEqual(portfolios(), {'Rob': 445825, 'RobSMSF': 684141}) # still served from the cache
-
-	def test_a_pass_says_when_the_next_is_due(self):
-		"""A pass reads each cache file once and gives the seconds until the next refresh: 90% of the shortest cache
-		lifetime once everything is fresh."""
-		self.serve_sharesight()
-		due = sharesight.warm() # a cold cache, filled
-		shortest = min(config.prefs.cache_seconds, config.prefs.performance_cache_seconds)
-		self.assertAlmostEqual(due, 0.9 * shortest, delta=5)
-		reads = []
-		real = cache.read
-		self.patch(cache, 'read', lambda name, *a, **k: (reads.append(name), real(name, *a, **k))[1])
-		self.age(sharesight.performance_cache_file(684141, 0), 600)
-		due = sharesight.warm()
-		self.assertAlmostEqual(due, 0.9 * config.prefs.performance_cache_seconds - 600, delta=5)
-		files = [sharesight.PORTFOLIOS_CACHE] + [sharesight.performance_cache_file(id, 0) for id in (445825, 684141)]
-		self.assertEqual(sorted(reads), sorted(files))
-
-	class Stopped(Exception):
-		"""Raised by the stand-in for keep_warm's wake event, to end the loop."""
-
-	def waits(self, count, on_wait=lambda n: None, **kwargs):
-		"""Run keep_warm until it has waited `count` times, noting each wait instead of sleeping (on_wait(n) at the
-		nth); returns the waits."""
-		waits, stopped = [], self.Stopped
-		class Wake:
-			"""keep_warm's wake event, which nobody sets."""
-			def wait(self, seconds):
-				"""Note the wait instead of sleeping, and end the loop after the last."""
-				waits.append(seconds)
-				on_wait(len(waits))
-				if len(waits) == count:
-					raise stopped
-			def clear(self):
-				"""Nothing to clear."""
-		self.patch(sharesight, '_wake', Wake())
-		with self.assertRaises(self.Stopped):
-			sharesight.keep_warm(**kwargs)
-		return waits
-
-	def test_loop_survives_failures_and_backs_off(self):
-		"""keep_warm survives a failure, retries slowly, then sleeps until the next cache is due."""
-		outcomes = [RuntimeError('boom'), 3000]
-		def warm():
-			"""Fail once, then succeed."""
-			outcome = outcomes.pop(0)
-			if isinstance(outcome, Exception):
-				raise outcome
-			return outcome
-		self.patch(sharesight, 'warm', warm)
-		self.assertEqual(self.waits(2, retry=300), [300, 3000]) # a slow retry after the failure, then until one is due
-
-	def test_with_cache_off_nothing_is_fetched_until_a_reload_turns_it_on(self):
-		"""With cache off, keep_warm idles without asking Sharesight, and starts when a reload turns it on."""
-		passes = []
-		self.patch(sharesight, 'warm', lambda: passes.append(1) or 3000)
-		self.prefs(cache=False)
-		def reload(n):
-			"""At the second wait, switch the cache on, as a reload would."""
-			if n == 2:
-				config.prefs = dataclasses.replace(config.prefs, cache=True)
-		self.assertEqual(self.waits(4, reload, idle=60), [60, 60, 3000, 3000]) # no request every 5 seconds while off
-		self.assertEqual(len(passes), 2)
-
-	def test_a_flush_wakes_the_loop_to_refill_at_once(self):
-		"""A flush wakes keep_warm, which refills the caches at once instead of when they are next due."""
-		passes = []
-		self.patch(sharesight, 'warm', lambda: passes.append(1) or 3000)
-		loop = gevent.spawn(sharesight.keep_warm)
-		self.addCleanup(loop.kill)
-		gevent.sleep(0.05)
-		self.assertEqual(len(passes), 1) # then it sleeps until the next cache is due, 50 minutes away
-		sharesight.wake()
-		gevent.sleep(0.05)
-		self.assertEqual(len(passes), 2)
-
-
-class TelegramSending(FinbotCase):
-	"""A message posted to Telegram against failing endpoints: the real send path, not the base fixture's recorder."""
-
-	def post(self, outcome):
-		"""Post 'hi' to a Telegram chat, the request answering with outcome (or raising it)."""
-		def fake(url, **kwargs):
-			"""The outcome, or raise it."""
-			if isinstance(outcome, Exception):
-				raise outcome
-			return outcome
-		self.patch(transports.net, 'post', fake)
-		return transports.TelegramBot('5').post_text('hi')
-
-	def test_success_returns_the_message(self):
-		"""A sent message returns the message, as Telegram gives it."""
-		self.assertEqual(self.post(Resp({'ok': True, 'result': {'message_id': 7}})), {'message_id': 7})
-
-	def test_a_refusal_or_failure_returns_none_and_logs_why(self):
-		"""A refused or failed send returns None, and the log has Telegram's reason; a 200 that is not JSON is no
-		message either."""
-		refusal = Resp({'ok': False, 'error_code': 400, 'description': "Bad Request: can't parse entities"}, 400)
-		for outcome in (refusal, Resp(status=429), ConnectionError('down'), Resp(None)):
-			with self.subTest(outcome=outcome), contextlib.redirect_stderr(io.StringIO()):
-				self.assertIsNone(self.post(outcome))
-		self.assertIn("can't parse entities", self.logged(refusal))
-
-	def logged(self, outcome):
-		"""What posting with this outcome logs."""
-		with contextlib.redirect_stderr(io.StringIO()) as log:
-			self.post(outcome)
-		return log.getvalue()
-
-	def test_a_chart_goes_without_a_parameter_sendphoto_does_not_take(self):
-		"""sendPhoto is sent the chat, caption and markup, and no link-preview flag: a photo has no preview."""
-		sent = []
-		self.patch(transports.net, 'post', lambda url, **kwargs: sent.append((url, kwargs)) or
-			Resp({'ok': True, 'result': {'message_id': 8}}))
-		self.assertEqual(transports.TelegramBot('5').post_chart(io.BytesIO(b'png'), 'caption'), {'message_id': 8})
-		url, kwargs = sent[0]
-		self.assertTrue(url.endswith('/sendPhoto'))
-		self.assertEqual(kwargs['data'], {'chat_id': '5', 'caption': 'caption', 'parse_mode': 'HTML',
-			'disable_notification': True})
-
-	def test_error_line_redacts_secrets_and_truncates(self):
-		"""An error line has no token or crumb in it, and is cut to about 200 characters."""
-		line = delivery.error_line(RuntimeError('GET /bot123456:ABC-def_1/send?crumb=SECRET&x=1 failed'), 'ctx')
-		self.assertNotIn('ABC-def_1', line)
-		self.assertNotIn('SECRET', line)
-		self.assertTrue(line.startswith('⚠️ ctx: '))
-		self.assertLessEqual(len(delivery.error_line('x' * 1000)), 205)
-
-	def test_error_reports_are_escaped_for_telegram_and_slack(self):
-		"""An error line is escaped for Telegram's HTML and Slack (the same), and left as it is for Discord."""
-		sent = []
-		self.patch(transports.Transport, 'text', REAL_TEXT)
-		for kind in (transports.TelegramBot, transports.SlackWebhook, transports.DiscordWebhook):
-			self.patch(kind, 'post_text', lambda dest, text, buttons=None: sent.append((dest.name, text)))
-		self.env(webhooks={'telegram': 'https://api.telegram.org/botX/', 'slack': 'https://hooks.slack.com/x',
-			'discord': 'https://discord.com/x/slack'})
-		self.env(telegramChatID='5')
-
-		error = RuntimeError("404 for ZZ at https://x/?a=1&crumb=abc; '<' not supported between <urllib3.connection>")
-		with contextlib.redirect_stderr(io.StringIO()):
-			REAL_REPORT_ERROR(error, context='.compare <b>')
-			REAL_REPORT_ERROR(error, transports.TelegramBot('55'), context='.compare')
-		texts = dict(sent[:3])
-		self.assertIn('crumb=&lt;redacted&gt;', texts['telegram']) # Telegram's HTML parser rejects a bare <redacted>
-		self.assertNotIn('<', texts['telegram'] + sent[3][1])
-		self.assertIn('.compare &lt;b&gt;: RuntimeError', texts['telegram'])
-		self.assertEqual(texts['slack'], texts['telegram'])
-		self.assertIn('crumb=<redacted>', texts['discord'])
-
-	def test_report_error_never_raises(self):
-		"""report_error survives a failed send."""
-		self.patch(transports.Transport, 'text', mock.Mock(side_effect=RuntimeError('send failed')))
-		with contextlib.redirect_stderr(io.StringIO()):
-			REAL_REPORT_ERROR(ValueError('boom'), transports.TelegramBot('5'))
-			REAL_REPORT_ERROR(ValueError('boom'))
-
-	def test_guarded_reports_and_exits_nonzero(self):
-		"""guarded reports a failure and exits 1; a success returns its value."""
-		reported = []
-		self.patch(delivery, 'report_error', lambda e, **k: reported.append(e))
-		with self.assertRaises(SystemExit) as raised:
-			delivery.guarded('job', lambda: 1 / 0)
-		self.assertEqual(raised.exception.code, 1)
-		self.assertEqual(len(reported), 1)
-		self.assertEqual(delivery.guarded('job', lambda: 'fine'), 'fine')
-
-
-class ChartDelivery(FinbotCase):
-	"""Charts reach Telegram and Slack replies (uploaded); a report whose chart fails still goes, as text."""
-	UPLOAD_URL = 'https://files.slack.com/upload/v1/abc'
-
-	def setUp(self):
-		"""The real Transport.chart, in place of the fixture's recorder."""
-		super().setUp()
-		self.patch(transports.Transport, 'chart', REAL_CHART)
-
-	def slack_answers(self, *outcomes):
-		"""net.post for the upload steps: the outcomes in turn (an exception is raised). Returns the calls made."""
-		calls = []
-		def post(url, **kwargs):
-			"""Record the call and answer with the next outcome, or raise it."""
-			calls.append((url, kwargs))
-			outcome = outcomes[len(calls) - 1]
-			if isinstance(outcome, Exception):
-				raise outcome
-			return outcome
-		self.patch(transports.net, 'post', post)
-		return calls
-
-	def upload_steps(self):
-		"""Slack's answers to the three upload steps, each a success."""
-		return [Resp({'ok': True, 'upload_url': self.UPLOAD_URL, 'file_id': 'F1'}), Resp(None, text='OK - 3'),
-			Resp({'ok': True, 'files': [{'id': 'F1'}]})]
-
-	def test_where_charts_can_go(self):
-		"""Charts always go to Telegram and Discord; Slack gets them as the bot, with the bot token, but not through its
-		incoming webhook."""
-		self.env(slackBotToken='xoxb-test')
-
-		kinds = (transports.TelegramBot, transports.SlackBot, transports.SlackWebhook, transports.DiscordWebhook)
-		self.assertEqual([kind(None).takes_charts for kind in kinds], [True, True, False, True])
-		self.env(slackBotToken=False)
-		self.assertFalse(transports.SlackBot(None).takes_charts)
-
-	def test_scheduled_reports_go_to_slack_as_the_bot_with_slackChannelID(self):
-		"""With slackChannelID and the bot token, scheduled reports reach Slack as the bot in place of the incoming
-		webhook, which it does not need; without the token, the webhook carries them."""
-		hook = 'https://hooks.slack.com/services/x'
-		telegram = config.env.webhooks['telegram']
-		self.env(webhooks={'telegram': telegram, 'slack': hook}, telegramChatID='1', slackChannelID='C9',
-			slackBotToken='xoxb-test')
-		self.assertEqual(list(map(repr, delivery.default_chats())), ['<TelegramBot 1>', '<SlackBot C9>'])
-		self.env(slackBotToken='')
-		self.assertEqual(list(map(repr, delivery.default_chats())), ['<TelegramBot 1>', f'<SlackWebhook {hook}>'])
-		self.env(webhooks={'telegram': telegram}, slackBotToken='xoxb-test')
-		self.assertEqual(list(map(repr, delivery.default_chats())), ['<TelegramBot 1>', '<SlackBot C9>'])
-		self.env(webhooks={}, telegramChatID='') # the bot alone
-		self.assertEqual(list(map(repr, delivery.default_chats())), ['<SlackBot C9>'])
-
-	def test_slack_replies_upload_their_charts(self):
-		"""A Slack reply uploads its chart in three steps (bot token), the report as its message, with no text copy."""
-		self.env(slackBotToken='xoxb-test')
-
-		for cmd, heading in (('.beta', '*Beta over 1.5'), ('.price', '*≥ 5% at close'), ('.compare T1 T2', '*')):
-			with self.subTest(cmd=cmd):
-				calls = self.slack_answers(*self.upload_steps())
-				self.sent.clear(); self.errors.clear()
-				worker.process_request(inbound.Message('slack', 'C1', '<@U1>', cmd, BOT, 'U'))
-				self.assertEqual(self.errors, [])
-				self.assertEqual([url for url, _ in calls],
-					['https://slack.com/api/files.getUploadURLExternal', self.UPLOAD_URL,
-															 'https://slack.com/api/files.completeUploadExternal'])
-				(_, ticket), (_, upload), (_, share) = calls
-				self.assertEqual(ticket['headers'], {'Authorization': 'Bearer xoxb-test'})
-				self.assertTrue(upload['data'].startswith(b'\x89PNG')) # the chart's bytes, as the body
-				self.assertEqual((ticket['data']['filename'], ticket['data']['length']),
-					('chart.png', len(upload['data'])))
-				self.assertEqual((share['data']['channel_id'], json.loads(share['data']['files'])),
-					('C1', [{'id': 'F1'}]))
-				self.assertTrue(share['data']['initial_comment'].startswith(heading),
-					share['data']['initial_comment'][:80]) # the report
-				# not sent again as text
-				self.assertNotIn(share['data']['initial_comment'].split('\n'), [payload for _, payload in self.sent])
-
-	def test_a_slack_upload_that_fails_goes_out_as_text(self):
-		"""A Slack upload failing at any step goes as text, but not on a last-step timeout, when it may have arrived."""
-		self.env(slackBotToken='xoxb-test')
-		self.patch(render, 'draw', lambda chart: io.BytesIO(b'png'))
-
-		good = self.upload_steps()
-		slow = transports.requests.exceptions.ReadTimeout('slow')
-		text = [('text', ['*Heading*', 'line 2'])]
-		for outcomes, expect in (([Resp({'ok': False, 'error': 'missing_scope'})], text),
-			([ConnectionError('down')], text),
-								 ([good[0], Resp(None, status=500, text='nope')], text),
-								 ([good[0], good[1], Resp({'ok': False, 'error': 'not_in_channel'})], text),
-								 ([good[0], good[1], slow], []), # it may have been shared after all
-								 (good, [])):
-			with self.subTest(outcomes=outcomes):
-				self.slack_answers(*outcomes)
-				self.sent.clear()
-				with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
-					delivery.deliver(transports.SlackBot('C1'), delivery.Report(['*Heading*', 'line 2'], CHART))
-				self.assertEqual(self.sent, expect)
-
-	def test_scheduled_reports_to_slack_carry_charts_only_as_the_bot(self):
-		"""Scheduled reports reach Slack's incoming webhook as text; the Slack bot (slackChannelID) and Telegram get
-		their charts."""
-		self.env(slackBotToken='xoxb-test')
-
-		# an incoming webhook cannot upload
-		self.env(webhooks={**config.env.webhooks, 'slack': 'https://hooks.slack.com/services/x'})
-		sent = []
-		self.patch(transports.Transport, 'chart', lambda dest, image, caption, buttons=None:
-			sent.append((dest.name, 'chart')) or {'ok': True})
-		self.patch(transports.Transport, 'text', lambda dest, lines, buttons=None:
-			sent.append((dest.name, 'text')) or [])
-		for report in (price.run, lambda: performance.run(past_days=7), shorts.run):
-			with self.subTest(report=report):
-				sent.clear()
-				report()
-				self.assertEqual(sorted(set(sent)), [('slack', 'text'), ('telegram', 'chart')])
-				self.env(slackChannelID='C9') # as the bot, the report's chart goes too
-				sent.clear()
-				report()
-				self.assertEqual(sorted(set(sent)), [('slack', 'chart'), ('telegram', 'chart')])
-				self.env(slackChannelID='')
-
-	def test_slack_without_a_bot_token_gets_text_and_no_chart_is_drawn(self):
-		"""Without the bot token, Slack replies are text and no chart is drawn for them."""
-		self.env(slackBotToken=False)
-
-		def refuse(*a, **k):
-			"""Fail: no chart should be drawn."""
-			raise AssertionError('a chart was drawn for Slack')
-		self.patch(render, 'draw', refuse) # every chart is drawn there
-		self.env(webhooks={**config.env.webhooks, 'slack': 'https://hooks.slack.com/services/x'})
-		for cmd in ('.beta', '.pie Rob', '.buy', '.sell', '.peg', '.price', '.price top', '.price T1 7d',
-				'.performance 7d', '.shorts', '.history T1', '.compare T1 T2'):
-			with self.subTest(cmd=cmd):
-				self.sent.clear(); self.errors.clear()
-				worker.process_request(inbound.Message('slack', 'C1', '<@U1>', cmd, BOT, 'U'))
-				self.assertEqual(self.errors, [])
-				self.assertTrue(self.sent)
-				self.assertEqual({kind for kind, *_ in self.sent}, {'text'})
-
-	def test_the_graph_preference_off_sends_no_chart_anywhere(self):
-		"""graph=False means no charts, as the README says; it used to hold back only the price-line charts."""
-		self.prefs(graph=False)
-
-		self.env(slackBotToken='xoxb-test')
-
-		posts = []
-		self.patch(transports.net, 'post', lambda url, **k: posts.append(url) or Resp({'ok': True, 'result': True}))
-		for service, chat in (('telegram', '55'), ('slack', 'C1')):
-			for cmd in ('.price', '.price top', '.performance 7d', '.shorts', '.buy', '.peg', '.beta', '.compare T1 T2',
-					'.pie Rob'):
-				with self.subTest(service=service, cmd=cmd):
-					self.sent.clear(); self.errors.clear()
-					worker.process_request(inbound.Message(service, chat, '@u', cmd, BOT, 'U'))
-					self.assertEqual(self.errors, [])
-					self.assertEqual({kind for kind, *_ in self.sent}, {'text'})
-		self.assertEqual([url for url in posts if 'sendPhoto' in url or 'files.' in url], [])
-
-	def test_a_chart_telegram_refuses_goes_out_as_text(self):
-		"""A chart Telegram refuses or that fails goes as text, but not after a timeout, when it may have arrived."""
-		self.patch(render, 'draw', lambda chart: io.BytesIO(b'png'))
-		def post(outcome):
-			"""A net.post that answers with outcome, or raises it."""
-			def fake(url, **kwargs):
-				"""The outcome, or raise it."""
-				if isinstance(outcome, Exception):
-					raise outcome
-				return outcome
-			return fake
-		text = [('text', ['Heading', 'line 2'])]
-		for outcome, expect in ((Resp({'ok': False, 'error_code': 400,
-			'description': "Bad Request: can't parse entities"}, status=400), text),
-								(Resp(None, status=502), text), (ConnectionError('down'), text),
-								(transports.requests.exceptions.ReadTimeout('slow'), []), # it may be there
-								(Resp({'ok': True, 'result': {'message_id': 1}}), [])):
-			with self.subTest(outcome=outcome):
-				self.sent.clear()
-				self.patch(transports.net, 'post', post(outcome))
-				with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
-					delivery.deliver(transports.TelegramBot('5'), delivery.Report(['Heading', 'line 2'], CHART, [[]]))
-				self.assertEqual(self.sent, expect)
-
-
-class DiscordWebhook(FinbotCase):
-	"""Discord's own webhook endpoint (the Slack-like one takes no flags): no previews, no pings, charts attached."""
-	URL = 'https://discord.com/api/webhooks/1/abc'
-
-	def setUp(self):
-		"""Discord as the only destination, sent through the real path, its posts recorded."""
-		super().setUp()
-		self.env(webhooks={'discord': self.URL})
-		self.patch(transports.Transport, 'text', REAL_TEXT)
-		self.patch(transports.Transport, 'chart', REAL_CHART)
-		self.posts = []
-		self.patch(transports.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or Resp({'id': '1'}))
-
-	def test_text_goes_to_the_native_endpoint_without_previews_or_pings(self):
-		"""Text goes to the native endpoint with ?wait=true, without previews or mentions."""
-		transports.DiscordWebhook(self.URL).post_text('hello @everyone')
-		url, kwargs = self.posts[0]
-		self.assertEqual(url, self.URL + '?wait=true') # without wait Discord answers 204, which counted as a failure
-		self.assertEqual(kwargs['json'], {'content': 'hello @everyone', 'flags': 4, 'allowed_mentions': {'parse': []}})
-
-	def test_a_scheduled_report_carries_its_chart(self):
-		"""A scheduled Discord report has its chart attached and the report as its message, within 2,000 characters."""
-		price.run()
-		(url, kwargs), = self.posts
-		self.assertEqual(url, self.URL + '?wait=true')
-		payload = json.loads(kwargs['data']['payload_json'])
-		self.assertEqual((payload['flags'], payload['allowed_mentions'], payload['attachments'][0]['filename']),
-			(4, {'parse': []}, 'chart.png'))
-		self.assertTrue(payload['content'].startswith('**≥ 5% at close'), payload['content'][:60])
-		self.assertLessEqual(len(payload['content']), 2000)
-		name, image, kind = kwargs['files']['files[0]']
-		self.assertEqual((name, kind), ('chart.png', 'image/png'))
-		self.assertTrue(image.read().startswith(b'\x89PNG'))
-
-	def test_a_long_report_is_split_by_size(self):
-		"""A long report is split into messages of at most 2,000 characters."""
-		line = 'x' * 150
-		transports.DiscordWebhook(self.URL).text([line] * 30)
-		sizes = [len(kwargs['json']['content']) for _, kwargs in self.posts]
-		self.assertGreater(len(sizes), 2)
-		self.assertTrue(all(size <= 2000 for size in sizes), sizes) # 20 lines a chunk used to make 3,019 characters
-
-	def test_links_are_masked_links(self):
-		"""Links are masked links, brackets and parentheses escaped."""
-		self.assertEqual(DISCORD.link('https://x.example/a(b)', 'T[1]'), '[T\\[1\\]](<https://x.example/a%28b%29>)')
-
-
-class SlackSending(FinbotCase):
-	"""Slack replies go through chat.postMessage with the bot token; scheduled reports through the incoming webhook."""
-
-	def setUp(self):
-		"""The real send path, its posts recorded."""
-		super().setUp()
-		self.patch(transports.Transport, 'text', REAL_TEXT)
-		self.patch(transports.Transport, 'chart', REAL_CHART)
-		self.posts = []
-		self.patch(transports.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or Resp({'ok': True}))
-
-	def test_link_previews_are_turned_off_in_the_body(self):
-		"""Link previews are turned off in the body, where Slack reads them, for replies and for the webhook."""
-		self.env(slackBotToken='xoxb-test')
-
-		transports.SlackBot('C1').post_text('hi')
-		transports.SlackWebhook('https://hooks.slack.com/services/T/B/x').post_text('hi')
-		(_, reply), (_, hook) = self.posts
-		self.assertEqual((reply['data']['unfurl_links'], reply['data']['unfurl_media']), ('false', 'false'))
-		self.assertEqual((hook['json']['unfurl_links'], hook['json']['unfurl_media']), (False, False))
-		for kwargs in (reply, hook): # they used to go in the headers, where Slack does not look
-			self.assertNotIn('unfurl_links', kwargs['headers'])
-
-	def test_a_reply_slack_refuses_is_logged_as_failed(self):
-		"""Slack refuses with HTTP 200 and ok false (the bot is not in the channel, say): a failure, logged as one."""
-		self.env(slackBotToken='xoxb-test')
-		self.patch(transports.net, 'post', lambda url, **kwargs: Resp({'ok': False, 'error': 'not_in_channel'}))
-		with contextlib.redirect_stderr(io.StringIO()) as err:
-			self.assertIsNone(transports.SlackBot('C1').post_text('hi'))
-		self.assertIn('Slack chat.postMessage to C1 failed: Slack chat.postMessage: not_in_channel', err.getvalue())
-		self.assertNotIn('OK', err.getvalue())
-
-	def test_a_reply_without_a_bot_token_is_logged_not_sent(self):
-		"""A Slack reply without the bot token is logged, once each, and not sent."""
-		self.env(slackBotToken=False)
-
-		err = io.StringIO()
-		with contextlib.redirect_stderr(err), contextlib.redirect_stdout(io.StringIO()):
-			for cmd in ('.help', '.price', '.compare T1 T2'):
-				worker.process_request(inbound.Message('slack', 'C1', '<@U1>', cmd, BOT, 'U'))
-		self.assertEqual((self.posts, self.errors), ([], []))
-		refusal = 'ERROR slack: a reply to channel C1 needs slackBotToken in finbot.env; not sent'
-		self.assertEqual(err.getvalue().count(refusal), 3)
-
-	def test_a_command_at_work_shows_an_hourglass_on_its_message(self):
-		"""While a command works, ⏳ is on the message it answers (Slack has no typing notice for bots); it comes off
-		once the answer is out, so the answer does not wait for it. Only with that message and the bot token."""
-		self.env(slackBotToken='xoxb-test')
-		self.patch(transports.Transport, 'TYPING', (0, 7, 10))
-		added = threading.Event()
-		def post(url, **kwargs):
-			"""Record the post; note when ⏳ is on."""
-			self.posts.append((url, kwargs))
-			if url.endswith('/reactions.add'):
-				added.set()
-			return Resp({'ok': True})
-		self.patch(transports.net, 'post', post)
-		help_text = about.prepare_help
-		self.patch(about, 'prepare_help', lambda *a: (added.wait(5), help_text(*a))[1]) # a command slow enough for ⏳
-		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.help', BOT, 'U', '1.2'))
-		self.assertEqual([url.rsplit('/', 1)[-1] for url, _ in self.posts],
-			['reactions.add', 'chat.postMessage', 'reactions.remove'])
-		reaction = {'channel': 'C1', 'timestamp': '1.2', 'name': 'hourglass_flowing_sand'}
-		self.assertEqual((self.posts[0][1]['data'], self.posts[2][1]['data']), (reaction, reaction))
-		self.assertFalse(transports.SlackBot('C1').shows_typing) # no message to put it on
-		self.env(slackBotToken='')
-		self.assertFalse(transports.SlackBot('C1', '1.2').shows_typing)
-
-	def test_an_hourglass_slack_refuses_is_tried_once(self):
-		"""⏳ that Slack refuses (a missing scope) is logged and not tried again, and nothing is taken off."""
-		self.env(slackBotToken='xoxb-test')
-		refuse = Resp({'ok': False, 'error': 'missing_scope'})
-		self.patch(transports.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or refuse)
-		chat = transports.SlackBot('C1', '1.2')
-		with contextlib.redirect_stderr(io.StringIO()) as err:
-			chat.typing()
-			chat.typing()
-			chat.typing_done()
-		self.assertEqual([url for url, _ in self.posts], ['https://slack.com/api/reactions.add'])
-		self.assertIn('Slack reactions.add failed: Slack reactions.add: missing_scope', err.getvalue())
-
-	def test_the_slack_api_says_what_is_missing(self):
-		"""slack_api without the bot token says so (it used to be a TypeError)."""
-		self.env(slackBotToken=False)
-
-		with self.assertRaisesRegex(RuntimeError,
-				'Slack files.getUploadURLExternal: needs slackBotToken in finbot.env'):
-			transports.slack_api('files.getUploadURLExternal', filename='chart.png', length=1) # it was a TypeError
-		self.assertEqual(self.posts, [])
-
-
-class SlackButtons(FinbotCase):
-	"""A chart's period buttons on Slack: a Block Kit message (the report, the chart uploaded into no channel, the
-	buttons) that a press redraws in place, where the presses reach finbot (the bot token and slackOutgoingWebhook)."""
-	UPLOAD_URL = 'https://files.slack.com/upload/v1/abc'
-
-	def setUp(self):
-		"""The real send path, its posts recorded and answered as Slack would; no waiting."""
-		super().setUp()
-		self.patch(transports.Transport, 'text', REAL_TEXT)
-		self.patch(transports.Transport, 'chart', REAL_CHART)
-		self.env(slackBotToken='xoxb-test', slackOutgoingWebhook='https://example.com/slack')
-		self.posts, self.refusals, self.waits = [], {}, []
-		self.patch(transports.time, 'sleep', self.waits.append)
-		def post(url, **kwargs):
-			"""Record the post and answer it: a refusal set for the method, while there is one left, else success."""
-			self.posts.append((url, kwargs))
-			method = url.rsplit('/', 1)[-1]
-			if self.refusals.get(method):
-				return Resp({'ok': False, 'error': self.refusals[method].pop(0)})
-			if url == self.UPLOAD_URL:
-				return Resp(None, text='OK - 3')
-			return Resp({'ok': True, 'upload_url': self.UPLOAD_URL, 'file_id': 'F1', 'files': [{'id': 'F1'}],
-				'ts': '1.5'})
-		self.patch(transports.net, 'post', post)
-
-	def calls(self):
-		"""The methods posted, in order (the upload as 'upload')."""
-		return ['upload' if url == self.UPLOAD_URL else url.rsplit('/', 1)[-1] for url, _ in self.posts]
-
-	def blocks(self, method):
-		"""The blocks of the last call of this method."""
-		return json.loads([kwargs for url, kwargs in self.posts if url.endswith('/' + method)][-1]['data']['blocks'])
-
-	def test_a_chart_reply_carries_its_period_buttons(self):
-		"""A chart reply is a Block Kit message: the report, the chart (uploaded into no channel, shown by its id) and
-		the period buttons, the one shown marked; nothing goes as text."""
-		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U'))
-		self.assertEqual(self.errors, [])
-		self.assertEqual(self.calls(), ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal',
-			'chat.postMessage'])
-		self.assertNotIn('channel_id', self.posts[2][1]['data']) # in no channel: the message shows it
-		blocks = self.blocks('chat.postMessage')
-		self.assertEqual([block['type'] for block in blocks], ['section', 'image', 'actions'])
-		self.assertTrue(blocks[0]['text']['text'].startswith('*≥ 5% at close'))
-		self.assertEqual(blocks[1]['slack_file'], {'id': 'F1'})
-		buttons_shown = blocks[2]['elements']
-		self.assertEqual([button['text']['text'] for button in buttons_shown], ['1D', '7D', '1M', '1Y', 'Max'])
-		self.assertEqual([button.get('style') for button in buttons_shown], ['primary'] + [None] * 4)
-		self.assertEqual(buttons_shown[1]['value'], 'c|l|w|5,0')
-
-	def test_a_press_redraws_the_chart_in_place(self):
-		"""A press uploads the chart for that period (a file cannot be replaced) and puts it in the message with
-		chat.update, the period pressed now marked; Telegram is not asked to answer the press."""
-		worker.process_callback(inbound.Press('slack', 't1', 'C1', '1.5', 'c|l|w|5,0'))
-		self.assertEqual(self.errors, [])
-		self.assertEqual(self.calls(), ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal',
-			'chat.update'])
-		update = self.posts[-1][1]['data']
-		self.assertEqual((update['channel'], update['ts']), ('C1', '1.5'))
-		blocks = self.blocks('chat.update')
-		self.assertEqual(blocks[1]['slack_file'], {'id': 'F1'})
-		self.assertEqual([button.get('style') for button in blocks[2]['elements']], [None, 'primary'] + [None] * 3)
-
-	def test_a_press_deletes_the_chart_it_replaced(self):
-		"""Each press uploads its period's chart and, once the message shows it, deletes the file shown before: no
-		chart is reused, so the bot's files do not pile up. The file a press shows is never deleted."""
-		uploads = iter(f'F{n}' for n in range(1, 10))
-		record = transports.net.post
-		def post(url, **kwargs):
-			"""Answer as Slack does, with a new file id for each upload."""
-			answer = record(url, **kwargs)
-			if url.endswith('/files.getUploadURLExternal'):
-				return Resp({'ok': True, 'upload_url': self.UPLOAD_URL, 'file_id': next(uploads)})
-			return answer
-		self.patch(transports.net, 'post', post)
-		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U')) # F1, in message 1.5
-		for _ in range(2):
-			worker.process_callback(inbound.Press('slack', 't1', 'C1', '1.5', 'c|l|w|5,0')) # F2, then F3
-		deleted = [kwargs['data']['file'] for url, kwargs in self.posts if url.endswith('/files.delete')]
-		self.assertEqual((deleted, self.errors), (['F1', 'F2'], []))
-		self.assertEqual(self.blocks('chat.update')[1]['slack_file'], {'id': 'F3'})
-
-	def test_a_chart_not_ready_to_show_is_shown_a_moment_later(self):
-		"""Slack refusing the blocks of a chart just uploaded (invalid_blocks) is asked again after a wait."""
-		self.refusals['chat.postMessage'] = ['invalid_blocks']
-		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U'))
-		self.assertEqual((self.calls()[-2:], self.waits), (['chat.postMessage', 'chat.postMessage'], [1]))
-		self.assertEqual((self.sent, self.errors), ([], []))
-
-	def test_buttons_slack_will_not_show_leave_a_plain_chart(self):
-		"""When Slack will not show the chart with its buttons, the chart is uploaded into the channel without them."""
-		self.refusals['chat.postMessage'] = ['invalid_blocks'] * 4
-		with contextlib.redirect_stderr(io.StringIO()):
-			worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U'))
-		self.assertEqual(self.calls()[-3:], ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal'])
-		self.assertEqual(self.posts[-1][1]['data']['channel_id'], 'C1')
-		self.assertEqual((self.waits, self.sent, self.errors), ([1, 2, 4], [], []))
-
-	def test_without_slackOutgoingWebhook_a_chart_has_no_buttons(self):
-		"""Without slackOutgoingWebhook, no press could reach finbot, so a chart goes without buttons."""
-		self.env(slackOutgoingWebhook='')
-		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U'))
-		self.assertEqual(self.calls(), ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal'])
-		self.assertEqual(self.posts[-1][1]['data']['channel_id'], 'C1')
-
-	def test_a_long_report_takes_several_sections(self):
-		"""A report longer than a section takes holds is split into sections, and an empty one takes none."""
-		caption = '\n'.join('x' * 99 for _ in range(70)) # 6,999 characters
-		blocks = transports.chart_blocks(caption, [], 'F1')
-		self.assertEqual([block['type'] for block in blocks], ['section'] * 3 + ['image'])
-		self.assertTrue(all(len(block['text']['text']) <= 3000 for block in blocks[:3]))
-		self.assertEqual([block['type'] for block in transports.chart_blocks('', [], 'F1')], ['image'])
-
-	def test_a_chart_message_that_times_out_is_not_posted_again(self):
-		"""A chart with its buttons whose chat.postMessage times out may have arrived: no plain copy follows."""
-		record = transports.net.post
-		def post(url, **kwargs):
-			"""Time out on chat.postMessage, as a slow Slack does."""
-			if url.endswith('/chat.postMessage'):
-				self.posts.append((url, kwargs))
-				raise transports.requests.exceptions.ReadTimeout('slow')
-			return record(url, **kwargs)
-		self.patch(transports.net, 'post', post)
-		with contextlib.redirect_stderr(io.StringIO()):
-			worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U'))
-		self.assertEqual(self.calls(), ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal',
-			'chat.postMessage'])
-		self.assertEqual(self.errors, [])
-
-
-class TelegramWebhook(FinbotCase):
-	"""bot.py registers the webhook once as it starts; importing lib.telegram, as each job does, does nothing."""
-
-	def setUp(self):
-		"""A Telegram bot with an outgoing webhook configured."""
-		super().setUp()
-		self.env(webhooks={'telegram': 'https://api.telegram.org/botX/'})
-		self.env(telegramOutgoingWebhook='https://example.com/telegram')
-
-
-	def test_register_unregisters_registers_and_names_the_bot(self):
-		"""register() removes the webhook, sets it again and returns the bot's @name."""
-		calls = []
-		def post(url, **kwargs):
-			"""Record the method and URL, and answer like Telegram."""
-			calls.append((url.rsplit('/', 1)[1], kwargs.get('params', {}).get('url')))
-			return Resp({'ok': True, 'result': {'username': 'finbot'} if url.endswith('getMe') else True})
-		self.patch(telegram.net, 'post', post)
-		with contextlib.redirect_stderr(io.StringIO()):
-			self.assertEqual(telegram.register(), '@finbot')
-		self.assertEqual(calls, [('setWebhook', ''), ('setWebhook', 'https://example.com/telegram'), ('getMe', None)])
-
-	def test_a_refused_call_raises(self):
-		"""A call Telegram refuses raises with its description."""
-		self.patch(telegram.net, 'post',
-			lambda url, **k: Resp({'ok': False, 'description': 'Bad Request: bad webhook: HTTPS url must be provided'}))
-		with contextlib.redirect_stderr(io.StringIO()), self.assertRaises(RuntimeError) as caught:
-			telegram.register()
-		self.assertIn('HTTPS url must be provided', str(caught.exception))
-
-	def test_edits_toasts_and_typing_go_through_call(self):
-		"""A chart edit that Telegram refuses is an error (unless nothing changed); a toast, a text edit or a typing
-		notice that fails, or an edit lost on the way, is only logged."""
-		def answer(outcome):
-			"""A net.post that answers with outcome, or raises it."""
-			def post(url, **kwargs):
-				"""The outcome, or raise it."""
-				if isinstance(outcome, Exception):
-					raise outcome
-				return outcome
-			return post
-		refused = Resp({'ok': False, 'description': 'Bad Request: message to edit not found'})
-		unchanged = Resp({'ok': False,
-			'description': 'Bad Request: message is not modified: specified new message content is the same'})
-		chat = transports.TelegramBot('5')
-		for outcome, raises in ((refused, True), (unchanged, False), (ConnectionError('down'), False),
-				(Resp({'ok': True, 'result': True}), False)):
-			with self.subTest(outcome=outcome):
-				self.patch(telegram.net, 'post', answer(outcome))
-				with contextlib.redirect_stderr(io.StringIO()):
-					if raises:
-						with self.assertRaisesRegex(RuntimeError, 'not found'):
-							chat.edit_chart('9', io.BytesIO(b'png'), 'caption')
-					else:
-						chat.edit_chart('9', io.BytesIO(b'png'), 'caption')
-					telegram.answerCallbackQuery('1', 'Loading…')
-					chat.edit_text('9', 'text')
-					chat.typing()
-
-	def test_importing_never_calls_telegram(self):
-		"""Importing the modules a job process uses makes no request."""
-		code = ('import requests\n'
-				'def refuse(*a, **k): raise SystemExit("network call on import")\n'
-				'requests.Session.request = refuse\n'
-				'from lib import telegram, yahoo, worker, jobrun\n'
-				'print("quiet")')
-		with tempfile.TemporaryDirectory() as data:
-			out = subprocess.run([sys.executable, '-c', code], cwd=ROOT, capture_output=True, text=True,
-				env={**os.environ, 'DATA_DIR': data, 'telegramBotToken': '123:abc',
-					'telegramOutgoingWebhook': 'https://example.com/telegram'})
-		self.assertEqual(out.stdout.split()[-1:], ['quiet'], out.stderr)
-
-	def test_the_bot_will_not_start_unregistered_and_never_logs_the_token(self):
-		"""bot.py exits when the webhook cannot be registered, and never logs the token."""
-		skip_if_old_env(self)
-		with tempfile.TemporaryDirectory() as data:
-			env = {k: v for k, v in bare_environment().items() if 'proxy' not in k.lower()}
-			env.update(DATA_DIR=data, port='0', scheduler='False', telegramBotToken='123456:SECRET-token',
-				telegramOutgoingToken='tok',
-					   telegramOutgoingWebhook='https://example.com/telegram',
-					   # unreachable, so the registration fails at once and offline
-					   HTTPS_PROXY='http://127.0.0.1:9', https_proxy='http://127.0.0.1:9')
-			out = subprocess.run([sys.executable, 'bot.py'], cwd=ROOT, env=env, capture_output=True, text=True,
-				timeout=120)
-		self.assertEqual(out.returncode, 1, out.stderr)
-		self.assertIn('ERROR ⚠️ telegram webhook:', out.stderr)
-		self.assertNotIn('SECRET-token', out.stderr + out.stdout)
-
-
-if __name__ == '__main__':
-	unittest.main()
diff --git a/tests/test_schedule.py b/tests/test_schedule.py
new file mode 100644
index 00000000..59e0bc93
--- /dev/null
+++ b/tests/test_schedule.py
@@ -0,0 +1,187 @@
+"""The schedule file (lib/jobs.py): its lines parsed, each script's job(argv), and a schedule the bot refuses."""
+import hashlib, io, os, subprocess, sys, tempfile, unittest
+from pathlib import Path
+from unittest import mock
+from tests.common import *
+from lib import jobs
+from lib.reports import cal, cleanup, performance, podcasts, price
+
+
+class ParseTests(unittest.TestCase):
+	def test_shipped_schedule_has_no_errors(self):
+		"""defaults/schedule.cron parses without errors, its Saturday line into its five commands."""
+		parsed, errors = jobs.parse_schedule(DEFAULT_SCHEDULE.read_text(), UTC)
+		self.assertEqual(errors, [])
+		self.assertEqual(len(parsed), 9)
+		weekly = next(j for j in parsed if j.expr == '28 21 * * Sat')
+		self.assertEqual([label for label, _, _ in weekly.commands],
+						 ['cal.py earnings', 'cal.py ex-dividend', 'price.py 7', 'performance.py 7', 'rating.py'])
+
+	def test_pasted_crontab_lines(self):
+		"""Lines pasted from a crontab work: variables, paths, redirects, braces, comments and @daily."""
+		text = ('MAILTO=me\n'
+				'*/20 * * * Mon-Fri ~/finbot/trades.py > ~/finbot.trades 2>&1\n'
+				'28  21 * * Sat { ./cal.py earnings; ./price.py 7 ;} &> /dev/null # weekly (UTC)\n'
+				'@daily /opt/finbot/reminder.py\n')
+		parsed, errors = jobs.parse_schedule(text, UTC)
+		self.assertEqual(errors, [])
+		self.assertEqual([[label for label, _, _ in j.commands] for j in parsed],
+						 [['trades.py'], ['cal.py earnings', 'price.py 7'], ['reminder.py']])
+
+	def test_quoted_arguments_may_hold_anything(self):
+		"""A quoted argument may hold shell characters; commands after a ; still split."""
+		url = 'https://example.com/feed?a=1&b=2;c>d $x |y'
+		line = f'01 0-10,23 * * Mon-Fri podcasts.py "{url}" > ~/out 2>&1 # hourly (UTC)\n'
+		parsed, errors = jobs.parse_schedule(line, UTC)
+		self.assertEqual(errors, [])
+		self.assertEqual(parsed[0].commands[0][2], {'feed_url': url})
+		parsed, errors = jobs.parse_schedule(f"5 4 * * * podcasts.py '{url}'; shorts.py\n", UTC)
+		self.assertEqual([(label[:9], kw) for label, _, kw in parsed[0].commands],
+			[('podcasts.', {'feed_url': url}), ('shorts.py', {})])
+
+	def test_unquoted_shell_syntax_is_still_refused(self):
+		"""Pipes, &&, a trailing &, $(), backticks, input redirects, subshells and an unterminated quote are refused."""
+		for line in ('0 1 * * * trades.py | tee x', '0 1 * * * trades.py && shorts.py & ',
+			'0 1 * * * trades.py $(date)', '0 1 * * * trades.py `x`',
+					 '0 1 * * * trades.py < in', '0 1 * * * (trades.py)', '0 1 * * * trades.py "unterminated'):
+			parsed, errors = jobs.parse_schedule(line + '\n', UTC)
+			self.assertEqual((parsed, len(errors)), ([], 1), line)
+
+	def test_redirects_and_braces_are_dropped(self):
+		"""Output redirects, braces, paths and trailing comments are dropped from a command."""
+		for command in ('trades.py > /dev/null', 'trades.py >/dev/null 2>&1', 'trades.py &> /dev/null',
+			'trades.py >> ~/x 2>&1',
+						'{ trades.py ;} > ~/x', '/opt/finbot/trades.py # note (x)'):
+			parsed, errors = jobs.parse_schedule(f'0 1 * * * {command}\n', UTC)
+			self.assertEqual(errors, [], command)
+			self.assertEqual([label for label, _, _ in parsed[0].commands], ['trades.py'], command)
+
+	def test_bin_run_wrapper_is_no_longer_accepted(self):
+		"""The old bin/run wrapper and a cd before the command are refused."""
+		for command in ('bin/run trades.py', 'cd /opt/finbot; trades.py'):
+			parsed, errors = jobs.parse_schedule(f'0 1 * * * {command}\n', UTC)
+			self.assertEqual((parsed, len(errors)), ([], 1), command)
+
+	def test_bad_lines_are_skipped_and_named(self):
+		"""Bad lines are skipped and reported by line number; the good ones load."""
+		text = ('0 1 * * * ls | wc -l\n0 1 * * * price.py bogus\n0 1 * * * nothere.py\n61 1 * * * trades.py\n0 1 * *\n'
+			'FOO=bar\n5 4 * * * shorts.py\n')
+		parsed, errors = jobs.parse_schedule(text, UTC)
+		self.assertEqual([j.expr for j in parsed], ['5 4 * * *'])
+		self.assertEqual([number for number, _, _ in errors], [1, 2, 3, 4, 5, 6])
+
+	def test_impossible_times_are_marked_and_reboot_is_just_unsupported(self):
+		"""Times that never come (30 February, minute 61, month 13, @bogus) are ImpossibleTime; @reboot is only
+		unsupported, and 29 February does come."""
+		for line in ('0 0 30 2 * trades.py', '0 9 31 4,6,9,11 * trades.py', '61 1 * * * trades.py',
+				'0 0 * 13 * trades.py', '@bogus trades.py'):
+			parsed, errors = jobs.parse_schedule(line + '\n', UTC)
+			self.assertEqual(parsed, [], line)
+			self.assertIsInstance(errors[0][2], jobs.ImpossibleTime, line)
+		parsed, errors = jobs.parse_schedule('@reboot trades.py\n0 0 29 2 * trades.py\n', UTC)
+		self.assertEqual([j.expr for j in parsed], ['0 0 29 2 *']) # a leap day does come
+		self.assertNotIsInstance(errors[0][2], jobs.ImpossibleTime)
+
+	def test_cron_tz_applies_to_the_lines_below(self):
+		"""CRON_TZ sets the zone of the lines below it; an unknown zone is skipped."""
+		text = '1 1 * * * trades.py\nCRON_TZ=Australia/Melbourne\n2 2 * * * trades.py\nCRON_TZ=Nowhere/Land\n'
+		parsed, _ = jobs.parse_schedule(text, UTC)
+		self.assertEqual([j.zone.key for j in parsed], ['UTC', 'Australia/Melbourne'])
+
+	def test_local_zone_follows_tz(self):
+		"""The local zone follows TZ."""
+		with mock.patch.dict(os.environ, {'TZ': 'Australia/Melbourne'}):
+			self.assertEqual(jobs.local_zone().key, 'Australia/Melbourne')
+
+
+class JobArguments(unittest.TestCase):
+	def test_price(self):
+		"""price.py takes one mode or a number of days (the top and bottom ten over it); anything else is its usage."""
+		self.assertEqual(price.job(['midsession']), ('price.py midsession', {'mode': 'midsession'}))
+		self.assertEqual(price.job(['premarket'])[1], {'mode': 'premarket'})
+		self.assertEqual(price.job(['interday'])[1], {'mode': 'interday'})
+		self.assertEqual(price.job(['close'])[1], {'mode': 'close'})
+		self.assertEqual(price.job(['28']), ('price.py 28', {'mode': 'period', 'threshold': 0, 'days': 28, 'top': 10}))
+		for bad in ([], ['soon'], ['7', '8']):
+			with self.assertRaises(ValueError):
+				price.job(bad)
+
+	def test_cal_and_performance(self):
+		"""cal.py takes earnings or ex-dividend; performance.py an optional number of days."""
+		self.assertEqual(cal.job(['earnings']), ('cal.py earnings', {'event': 'earnings'}))
+		self.assertEqual(cal.job(['ex-dividend']), ('cal.py ex-dividend', {'event': 'ex-dividend'}))
+		self.assertEqual(performance.job([]), ('performance.py', {}))
+		self.assertEqual(performance.job(['7']), ('performance.py 7', {'past_days': 7}))
+		for bad, module in ((['x'], cal), ([], cal), (['x'], performance)):
+			with self.assertRaises(ValueError):
+				module.job(bad)
+
+	def test_podcasts(self):
+		"""podcasts.py takes one feed URL; its label carries the start of the hash that names its state file."""
+		url = 'https://example.com/podcast.rss'
+		label, kwargs = podcasts.job([url])
+		self.assertEqual(kwargs, {'feed_url': url})
+		# the label is the start of the state file's hash
+		self.assertIn('finbot_podcast_seen_' + label.split()[1], podcasts.get_seen_filename(url))
+		self.assertEqual(label, 'podcasts.py ' + hashlib.md5(url.encode()).hexdigest()[:8])
+		for bad in ([], [url, url]):
+			with self.assertRaises(ValueError):
+				podcasts.job(bad)
+
+	def test_cleanup_arguments(self):
+		"""cleanup.py takes an optional number of days."""
+		self.assertEqual(cleanup.job([]), ('cleanup.py', {}))
+		self.assertEqual(cleanup.job(['7']), ('cleanup.py 7', {'days': 7}))
+		for bad in (['x'], ['1', '2']):
+			with self.assertRaises(ValueError):
+				cleanup.job(bad)
+
+	def test_plain_scripts(self):
+		"""The other reports take no arguments."""
+		for name, module in jobs.SCRIPTS.items():
+			if name not in ('cal', 'cleanup', 'performance', 'podcasts', 'price'):
+				self.assertEqual(module.job([]), (name + '.py', {}))
+
+	def test_cli_prints_usage_and_exits(self):
+		"""A bad argument on the command line prints the usage and exits 1."""
+		with mock.patch('sys.stderr', new_callable=io.StringIO) as err, self.assertRaises(SystemExit) as exit_:
+			delivery.cli('price.py', price.job, lambda **k: None, ['bogus'])
+		self.assertEqual(exit_.exception.code, 1)
+		self.assertIn('Usage: price.py [midsession', err.getvalue())
+
+	def test_cli_runs_the_handler(self):
+		"""Good arguments run the report with what job() made of them."""
+		calls = []
+		delivery.cli('price.py', price.job, lambda **k: calls.append(k), ['midsession'])
+		self.assertEqual(calls, [{'mode': 'midsession'}])
+
+	def test_each_scheduled_report_runs_from_its_top_level_script(self):
+		"""./price.py and the rest run lib/reports/<name>.py, so cron lines and runs by hand keep working."""
+		self.assertEqual(sorted(path.stem for path in Path(ROOT).glob('*.py') if path.stem not in ('bot',
+			'bot_send_stdin')),
+						 sorted(jobs.SCRIPTS))
+		with tempfile.TemporaryDirectory() as data:
+			run = lambda *args: subprocess.run([sys.executable, *args], cwd=ROOT, capture_output=True, text=True,
+				timeout=60, env={**os.environ, 'DATA_DIR': data})
+			self.assertEqual(run('cleanup.py', '0').returncode, 0)
+			bad = run('price.py', 'bogus')
+			self.assertEqual(bad.returncode, 1)
+			self.assertIn('Usage: price.py [midsession', bad.stderr)
+
+
+class StartupRefusal(unittest.TestCase):
+	def test_the_bot_will_not_start_with_an_impossible_time(self):
+		"""bot.py will not start while the schedule holds a time that never comes, and says which."""
+		skip_if_old_env(self)
+		with tempfile.TemporaryDirectory() as data:
+			Path(data, 'schedule.cron').write_text('*/20 * * * Mon-Fri trades.py\n0 9 31 6 * price.py close\n')
+			env = bare_environment()
+			env.update(DATA_DIR=data, port='0')
+			out = subprocess.run([sys.executable, 'bot.py'], cwd=ROOT, env=env, capture_output=True, text=True,
+				timeout=120)
+		self.assertEqual(out.returncode, 1, out.stderr)
+		self.assertIn(f"ERROR schedule: {data}/schedule.cron line 2: '0 9 31 6 *' never comes", out.stderr)
+
+
+if __name__ == '__main__':
+	unittest.main()
diff --git a/tests/test_jobs.py b/tests/test_scheduler.py
similarity index 77%
rename from tests/test_jobs.py
rename to tests/test_scheduler.py
index 359504c8..729dd60d 100644
--- a/tests/test_jobs.py
+++ b/tests/test_scheduler.py
@@ -1,15 +1,13 @@
-"""The in-process scheduler (lib/jobs.py): the schedule file, each script's job(argv), firing on a fake clock."""
-import contextlib, datetime, hashlib, io, json, os, signal, subprocess, sys, tempfile, threading, time, unittest
+"""The in-process scheduler (lib/jobs.py): firing on a fake clock, across the clocks' changes, runs from chat,
+and each command in a child process."""
+import contextlib, datetime, io, json, os, signal, subprocess, sys, tempfile, threading, time, unittest
 import zoneinfo
 from pathlib import Path
 from unittest import mock
 import gevent
 from tests.common import *
 from lib import jobs
-from lib.reports import cal, cleanup, performance, podcasts, price
-
-UTC = zoneinfo.ZoneInfo('UTC')
-DEFAULT_SCHEDULE = Path(os.path.dirname(os.path.dirname(os.path.abspath(__file__)))) / 'defaults' / 'schedule.cron'
+from lib.reports import cleanup, podcasts, price
 
 
 def run_in_process(label, module, kwargs):
@@ -33,168 +31,6 @@ def __call__(self):
 		return self.now
 
 
-class ParseTests(unittest.TestCase):
-	def test_shipped_schedule_has_no_errors(self):
-		"""defaults/schedule.cron parses without errors, its Saturday line into its five commands."""
-		parsed, errors = jobs.parse_schedule(DEFAULT_SCHEDULE.read_text(), UTC)
-		self.assertEqual(errors, [])
-		self.assertEqual(len(parsed), 9)
-		weekly = next(j for j in parsed if j.expr == '28 21 * * Sat')
-		self.assertEqual([label for label, _, _ in weekly.commands],
-						 ['cal.py earnings', 'cal.py ex-dividend', 'price.py 7', 'performance.py 7', 'rating.py'])
-
-	def test_pasted_crontab_lines(self):
-		"""Lines pasted from a crontab work: variables, paths, redirects, braces, comments and @daily."""
-		text = ('MAILTO=me\n'
-				'*/20 * * * Mon-Fri ~/finbot/trades.py > ~/finbot.trades 2>&1\n'
-				'28  21 * * Sat { ./cal.py earnings; ./price.py 7 ;} &> /dev/null # weekly (UTC)\n'
-				'@daily /opt/finbot/reminder.py\n')
-		parsed, errors = jobs.parse_schedule(text, UTC)
-		self.assertEqual(errors, [])
-		self.assertEqual([[label for label, _, _ in j.commands] for j in parsed],
-						 [['trades.py'], ['cal.py earnings', 'price.py 7'], ['reminder.py']])
-
-	def test_quoted_arguments_may_hold_anything(self):
-		"""A quoted argument may hold shell characters; commands after a ; still split."""
-		url = 'https://example.com/feed?a=1&b=2;c>d $x |y'
-		line = f'01 0-10,23 * * Mon-Fri podcasts.py "{url}" > ~/out 2>&1 # hourly (UTC)\n'
-		parsed, errors = jobs.parse_schedule(line, UTC)
-		self.assertEqual(errors, [])
-		self.assertEqual(parsed[0].commands[0][2], {'feed_url': url})
-		parsed, errors = jobs.parse_schedule(f"5 4 * * * podcasts.py '{url}'; shorts.py\n", UTC)
-		self.assertEqual([(label[:9], kw) for label, _, kw in parsed[0].commands],
-			[('podcasts.', {'feed_url': url}), ('shorts.py', {})])
-
-	def test_unquoted_shell_syntax_is_still_refused(self):
-		"""Pipes, &&, a trailing &, $(), backticks, input redirects, subshells and an unterminated quote are refused."""
-		for line in ('0 1 * * * trades.py | tee x', '0 1 * * * trades.py && shorts.py & ',
-			'0 1 * * * trades.py $(date)', '0 1 * * * trades.py `x`',
-					 '0 1 * * * trades.py < in', '0 1 * * * (trades.py)', '0 1 * * * trades.py "unterminated'):
-			parsed, errors = jobs.parse_schedule(line + '\n', UTC)
-			self.assertEqual((parsed, len(errors)), ([], 1), line)
-
-	def test_redirects_and_braces_are_dropped(self):
-		"""Output redirects, braces, paths and trailing comments are dropped from a command."""
-		for command in ('trades.py > /dev/null', 'trades.py >/dev/null 2>&1', 'trades.py &> /dev/null',
-			'trades.py >> ~/x 2>&1',
-						'{ trades.py ;} > ~/x', '/opt/finbot/trades.py # note (x)'):
-			parsed, errors = jobs.parse_schedule(f'0 1 * * * {command}\n', UTC)
-			self.assertEqual(errors, [], command)
-			self.assertEqual([label for label, _, _ in parsed[0].commands], ['trades.py'], command)
-
-	def test_bin_run_wrapper_is_no_longer_accepted(self):
-		"""The old bin/run wrapper and a cd before the command are refused."""
-		for command in ('bin/run trades.py', 'cd /opt/finbot; trades.py'):
-			parsed, errors = jobs.parse_schedule(f'0 1 * * * {command}\n', UTC)
-			self.assertEqual((parsed, len(errors)), ([], 1), command)
-
-	def test_bad_lines_are_skipped_and_named(self):
-		"""Bad lines are skipped and reported by line number; the good ones load."""
-		text = ('0 1 * * * ls | wc -l\n0 1 * * * price.py bogus\n0 1 * * * nothere.py\n61 1 * * * trades.py\n0 1 * *\n'
-			'FOO=bar\n5 4 * * * shorts.py\n')
-		parsed, errors = jobs.parse_schedule(text, UTC)
-		self.assertEqual([j.expr for j in parsed], ['5 4 * * *'])
-		self.assertEqual([number for number, _, _ in errors], [1, 2, 3, 4, 5, 6])
-
-	def test_impossible_times_are_marked_and_reboot_is_just_unsupported(self):
-		"""Times that never come (30 February, minute 61, month 13, @bogus) are ImpossibleTime; @reboot is only
-		unsupported, and 29 February does come."""
-		for line in ('0 0 30 2 * trades.py', '0 9 31 4,6,9,11 * trades.py', '61 1 * * * trades.py',
-				'0 0 * 13 * trades.py', '@bogus trades.py'):
-			parsed, errors = jobs.parse_schedule(line + '\n', UTC)
-			self.assertEqual(parsed, [], line)
-			self.assertIsInstance(errors[0][2], jobs.ImpossibleTime, line)
-		parsed, errors = jobs.parse_schedule('@reboot trades.py\n0 0 29 2 * trades.py\n', UTC)
-		self.assertEqual([j.expr for j in parsed], ['0 0 29 2 *']) # a leap day does come
-		self.assertNotIsInstance(errors[0][2], jobs.ImpossibleTime)
-
-	def test_cron_tz_applies_to_the_lines_below(self):
-		"""CRON_TZ sets the zone of the lines below it; an unknown zone is skipped."""
-		text = '1 1 * * * trades.py\nCRON_TZ=Australia/Melbourne\n2 2 * * * trades.py\nCRON_TZ=Nowhere/Land\n'
-		parsed, _ = jobs.parse_schedule(text, UTC)
-		self.assertEqual([j.zone.key for j in parsed], ['UTC', 'Australia/Melbourne'])
-
-	def test_local_zone_follows_tz(self):
-		"""The local zone follows TZ."""
-		with mock.patch.dict(os.environ, {'TZ': 'Australia/Melbourne'}):
-			self.assertEqual(jobs.local_zone().key, 'Australia/Melbourne')
-
-
-class JobArguments(unittest.TestCase):
-	def test_price(self):
-		"""price.py takes one mode or a number of days (the top and bottom ten over it); anything else is its usage."""
-		self.assertEqual(price.job(['midsession']), ('price.py midsession', {'mode': 'midsession'}))
-		self.assertEqual(price.job(['premarket'])[1], {'mode': 'premarket'})
-		self.assertEqual(price.job(['interday'])[1], {'mode': 'interday'})
-		self.assertEqual(price.job(['close'])[1], {'mode': 'close'})
-		self.assertEqual(price.job(['28']), ('price.py 28', {'mode': 'period', 'threshold': 0, 'days': 28, 'top': 10}))
-		for bad in ([], ['soon'], ['7', '8']):
-			with self.assertRaises(ValueError):
-				price.job(bad)
-
-	def test_cal_and_performance(self):
-		"""cal.py takes earnings or ex-dividend; performance.py an optional number of days."""
-		self.assertEqual(cal.job(['earnings']), ('cal.py earnings', {'event': 'earnings'}))
-		self.assertEqual(cal.job(['ex-dividend']), ('cal.py ex-dividend', {'event': 'ex-dividend'}))
-		self.assertEqual(performance.job([]), ('performance.py', {}))
-		self.assertEqual(performance.job(['7']), ('performance.py 7', {'past_days': 7}))
-		for bad, module in ((['x'], cal), ([], cal), (['x'], performance)):
-			with self.assertRaises(ValueError):
-				module.job(bad)
-
-	def test_podcasts(self):
-		"""podcasts.py takes one feed URL; its label carries the start of the hash that names its state file."""
-		url = 'https://example.com/podcast.rss'
-		label, kwargs = podcasts.job([url])
-		self.assertEqual(kwargs, {'feed_url': url})
-		# the label is the start of the state file's hash
-		self.assertIn('finbot_podcast_seen_' + label.split()[1], podcasts.get_seen_filename(url))
-		self.assertEqual(label, 'podcasts.py ' + hashlib.md5(url.encode()).hexdigest()[:8])
-		for bad in ([], [url, url]):
-			with self.assertRaises(ValueError):
-				podcasts.job(bad)
-
-	def test_cleanup_arguments(self):
-		"""cleanup.py takes an optional number of days."""
-		self.assertEqual(cleanup.job([]), ('cleanup.py', {}))
-		self.assertEqual(cleanup.job(['7']), ('cleanup.py 7', {'days': 7}))
-		for bad in (['x'], ['1', '2']):
-			with self.assertRaises(ValueError):
-				cleanup.job(bad)
-
-	def test_plain_scripts(self):
-		"""The other reports take no arguments."""
-		for name, module in jobs.SCRIPTS.items():
-			if name not in ('cal', 'cleanup', 'performance', 'podcasts', 'price'):
-				self.assertEqual(module.job([]), (name + '.py', {}))
-
-	def test_cli_prints_usage_and_exits(self):
-		"""A bad argument on the command line prints the usage and exits 1."""
-		with mock.patch('sys.stderr', new_callable=io.StringIO) as err, self.assertRaises(SystemExit) as exit_:
-			delivery.cli('price.py', price.job, lambda **k: None, ['bogus'])
-		self.assertEqual(exit_.exception.code, 1)
-		self.assertIn('Usage: price.py [midsession', err.getvalue())
-
-	def test_cli_runs_the_handler(self):
-		"""Good arguments run the report with what job() made of them."""
-		calls = []
-		delivery.cli('price.py', price.job, lambda **k: calls.append(k), ['midsession'])
-		self.assertEqual(calls, [{'mode': 'midsession'}])
-
-	def test_each_scheduled_report_runs_from_its_top_level_script(self):
-		"""./price.py and the rest run lib/reports/<name>.py, so cron lines and runs by hand keep working."""
-		self.assertEqual(sorted(path.stem for path in Path(ROOT).glob('*.py') if path.stem not in ('bot',
-			'bot_send_stdin')),
-						 sorted(jobs.SCRIPTS))
-		with tempfile.TemporaryDirectory() as data:
-			run = lambda *args: subprocess.run([sys.executable, *args], cwd=ROOT, capture_output=True, text=True,
-				timeout=60, env={**os.environ, 'DATA_DIR': data})
-			self.assertEqual(run('cleanup.py', '0').returncode, 0)
-			bad = run('price.py', 'bogus')
-			self.assertEqual(bad.returncode, 1)
-			self.assertIn('Usage: price.py [midsession', bad.stderr)
-
-
 class CleanupJob(FinbotCase):
 	def test_removes_only_old_finbot_cache_files(self):
 		"""cleanup deletes finbot's cache files older than the days given, not newer ones, other files or dirs."""
@@ -486,20 +322,6 @@ def test_the_scheduler_runs_a_daily_job_once_across_the_change(self):
 		self.assertEqual(fired, ['04 09:00', '05 09:00'])
 
 
-class StartupRefusal(unittest.TestCase):
-	def test_the_bot_will_not_start_with_an_impossible_time(self):
-		"""bot.py will not start while the schedule holds a time that never comes, and says which."""
-		skip_if_old_env(self)
-		with tempfile.TemporaryDirectory() as data:
-			Path(data, 'schedule.cron').write_text('*/20 * * * Mon-Fri trades.py\n0 9 31 6 * price.py close\n')
-			env = bare_environment()
-			env.update(DATA_DIR=data, port='0')
-			out = subprocess.run([sys.executable, 'bot.py'], cwd=ROOT, env=env, capture_output=True, text=True,
-				timeout=120)
-		self.assertEqual(out.returncode, 1, out.stderr)
-		self.assertIn(f"ERROR schedule: {data}/schedule.cron line 2: '0 9 31 6 *' never comes", out.stderr)
-
-
 class ManualRuns(SchedulerBase):
 	def setUp(self):
 		"""A schedule with repeats, comments, braces and the commands that cannot be picked (cleanup, podcasts)."""
@@ -734,6 +556,7 @@ def test_job_is_still_a_ticker(self):
 		""".job is a ticker, not .jobs."""
 		self.assertNotIn('Scheduled', str(self.command('.job')))
 
+
 class ChildProcess(FinbotCase):
 	"""Jobs run in their own process, so heavy work cannot hold up the bot's web server."""
 
@@ -865,6 +688,5 @@ def test_each_command_of_a_line_has_a_deadline_of_its_own(self):
 		self.assertEqual(self.errors, [])
 
 
-
 if __name__ == '__main__':
 	unittest.main()
diff --git a/tests/test_sending.py b/tests/test_sending.py
new file mode 100644
index 00000000..27f1e9eb
--- /dev/null
+++ b/tests/test_sending.py
@@ -0,0 +1,549 @@
+"""Posting to the networks against their answers and failures: Telegram, Slack (text, charts, Block Kit
+buttons) and Discord."""
+import contextlib, io, json, threading, unittest
+from unittest import mock
+from tests.common import *
+
+
+class TelegramSending(FinbotCase):
+	"""A message posted to Telegram against failing endpoints: the real send path, not the base fixture's recorder."""
+
+	def post(self, outcome):
+		"""Post 'hi' to a Telegram chat, the request answering with outcome (or raising it)."""
+		def fake(url, **kwargs):
+			"""The outcome, or raise it."""
+			if isinstance(outcome, Exception):
+				raise outcome
+			return outcome
+		self.patch(transports.net, 'post', fake)
+		return transports.TelegramBot('5').post_text('hi')
+
+	def test_success_returns_the_message(self):
+		"""A sent message returns the message, as Telegram gives it."""
+		self.assertEqual(self.post(Resp({'ok': True, 'result': {'message_id': 7}})), {'message_id': 7})
+
+	def test_a_refusal_or_failure_returns_none_and_logs_why(self):
+		"""A refused or failed send returns None, and the log has Telegram's reason; a 200 that is not JSON is no
+		message either."""
+		refusal = Resp({'ok': False, 'error_code': 400, 'description': "Bad Request: can't parse entities"}, 400)
+		for outcome in (refusal, Resp(status=429), ConnectionError('down'), Resp(None)):
+			with self.subTest(outcome=outcome), contextlib.redirect_stderr(io.StringIO()):
+				self.assertIsNone(self.post(outcome))
+		self.assertIn("can't parse entities", self.logged(refusal))
+
+	def logged(self, outcome):
+		"""What posting with this outcome logs."""
+		with contextlib.redirect_stderr(io.StringIO()) as log:
+			self.post(outcome)
+		return log.getvalue()
+
+	def test_a_chart_goes_without_a_parameter_sendphoto_does_not_take(self):
+		"""sendPhoto is sent the chat, caption and markup, and no link-preview flag: a photo has no preview."""
+		sent = []
+		self.patch(transports.net, 'post', lambda url, **kwargs: sent.append((url, kwargs)) or
+			Resp({'ok': True, 'result': {'message_id': 8}}))
+		self.assertEqual(transports.TelegramBot('5').post_chart(io.BytesIO(b'png'), 'caption'), {'message_id': 8})
+		url, kwargs = sent[0]
+		self.assertTrue(url.endswith('/sendPhoto'))
+		self.assertEqual(kwargs['data'], {'chat_id': '5', 'caption': 'caption', 'parse_mode': 'HTML',
+			'disable_notification': True})
+
+	def test_error_line_redacts_secrets_and_truncates(self):
+		"""An error line has no token or crumb in it, and is cut to about 200 characters."""
+		line = delivery.error_line(RuntimeError('GET /bot123456:ABC-def_1/send?crumb=SECRET&x=1 failed'), 'ctx')
+		self.assertNotIn('ABC-def_1', line)
+		self.assertNotIn('SECRET', line)
+		self.assertTrue(line.startswith('⚠️ ctx: '))
+		self.assertLessEqual(len(delivery.error_line('x' * 1000)), 205)
+
+	def test_error_reports_are_escaped_for_telegram_and_slack(self):
+		"""An error line is escaped for Telegram's HTML and Slack (the same), and left as it is for Discord."""
+		sent = []
+		self.patch(transports.Transport, 'text', REAL_TEXT)
+		for kind in (transports.TelegramBot, transports.SlackWebhook, transports.DiscordWebhook):
+			self.patch(kind, 'post_text', lambda dest, text, buttons=None: sent.append((dest.name, text)))
+		self.env(webhooks={'telegram': 'https://api.telegram.org/botX/', 'slack': 'https://hooks.slack.com/x',
+			'discord': 'https://discord.com/x/slack'})
+		self.env(telegramChatID='5')
+
+		error = RuntimeError("404 for ZZ at https://x/?a=1&crumb=abc; '<' not supported between <urllib3.connection>")
+		with contextlib.redirect_stderr(io.StringIO()):
+			REAL_REPORT_ERROR(error, context='.compare <b>')
+			REAL_REPORT_ERROR(error, transports.TelegramBot('55'), context='.compare')
+		texts = dict(sent[:3])
+		self.assertIn('crumb=&lt;redacted&gt;', texts['telegram']) # Telegram's HTML parser rejects a bare <redacted>
+		self.assertNotIn('<', texts['telegram'] + sent[3][1])
+		self.assertIn('.compare &lt;b&gt;: RuntimeError', texts['telegram'])
+		self.assertEqual(texts['slack'], texts['telegram'])
+		self.assertIn('crumb=<redacted>', texts['discord'])
+
+	def test_report_error_never_raises(self):
+		"""report_error survives a failed send."""
+		self.patch(transports.Transport, 'text', mock.Mock(side_effect=RuntimeError('send failed')))
+		with contextlib.redirect_stderr(io.StringIO()):
+			REAL_REPORT_ERROR(ValueError('boom'), transports.TelegramBot('5'))
+			REAL_REPORT_ERROR(ValueError('boom'))
+
+	def test_guarded_reports_and_exits_nonzero(self):
+		"""guarded reports a failure and exits 1; a success returns its value."""
+		reported = []
+		self.patch(delivery, 'report_error', lambda e, **k: reported.append(e))
+		with self.assertRaises(SystemExit) as raised:
+			delivery.guarded('job', lambda: 1 / 0)
+		self.assertEqual(raised.exception.code, 1)
+		self.assertEqual(len(reported), 1)
+		self.assertEqual(delivery.guarded('job', lambda: 'fine'), 'fine')
+
+
+class ChartDelivery(FinbotCase):
+	"""Charts reach Telegram and Slack replies (uploaded); a report whose chart fails still goes, as text."""
+	UPLOAD_URL = 'https://files.slack.com/upload/v1/abc'
+
+	def setUp(self):
+		"""The real Transport.chart, in place of the fixture's recorder."""
+		super().setUp()
+		self.patch(transports.Transport, 'chart', REAL_CHART)
+
+	def slack_answers(self, *outcomes):
+		"""net.post for the upload steps: the outcomes in turn (an exception is raised). Returns the calls made."""
+		calls = []
+		def post(url, **kwargs):
+			"""Record the call and answer with the next outcome, or raise it."""
+			calls.append((url, kwargs))
+			outcome = outcomes[len(calls) - 1]
+			if isinstance(outcome, Exception):
+				raise outcome
+			return outcome
+		self.patch(transports.net, 'post', post)
+		return calls
+
+	def upload_steps(self):
+		"""Slack's answers to the three upload steps, each a success."""
+		return [Resp({'ok': True, 'upload_url': self.UPLOAD_URL, 'file_id': 'F1'}), Resp(None, text='OK - 3'),
+			Resp({'ok': True, 'files': [{'id': 'F1'}]})]
+
+	def test_where_charts_can_go(self):
+		"""Charts always go to Telegram and Discord; Slack gets them as the bot, with the bot token, but not through its
+		incoming webhook."""
+		self.env(slackBotToken='xoxb-test')
+
+		kinds = (transports.TelegramBot, transports.SlackBot, transports.SlackWebhook, transports.DiscordWebhook)
+		self.assertEqual([kind(None).takes_charts for kind in kinds], [True, True, False, True])
+		self.env(slackBotToken=False)
+		self.assertFalse(transports.SlackBot(None).takes_charts)
+
+	def test_scheduled_reports_go_to_slack_as_the_bot_with_slackChannelID(self):
+		"""With slackChannelID and the bot token, scheduled reports reach Slack as the bot in place of the incoming
+		webhook, which it does not need; without the token, the webhook carries them."""
+		hook = 'https://hooks.slack.com/services/x'
+		telegram = config.env.webhooks['telegram']
+		self.env(webhooks={'telegram': telegram, 'slack': hook}, telegramChatID='1', slackChannelID='C9',
+			slackBotToken='xoxb-test')
+		self.assertEqual(list(map(repr, delivery.default_chats())), ['<TelegramBot 1>', '<SlackBot C9>'])
+		self.env(slackBotToken='')
+		self.assertEqual(list(map(repr, delivery.default_chats())), ['<TelegramBot 1>', f'<SlackWebhook {hook}>'])
+		self.env(webhooks={'telegram': telegram}, slackBotToken='xoxb-test')
+		self.assertEqual(list(map(repr, delivery.default_chats())), ['<TelegramBot 1>', '<SlackBot C9>'])
+		self.env(webhooks={}, telegramChatID='') # the bot alone
+		self.assertEqual(list(map(repr, delivery.default_chats())), ['<SlackBot C9>'])
+
+	def test_slack_replies_upload_their_charts(self):
+		"""A Slack reply uploads its chart in three steps (bot token), the report as its message, with no text copy."""
+		self.env(slackBotToken='xoxb-test')
+
+		for cmd, heading in (('.beta', '*Beta over 1.5'), ('.price', '*≥ 5% at close'), ('.compare T1 T2', '*')):
+			with self.subTest(cmd=cmd):
+				calls = self.slack_answers(*self.upload_steps())
+				self.sent.clear(); self.errors.clear()
+				worker.process_request(inbound.Message('slack', 'C1', '<@U1>', cmd, BOT, 'U'))
+				self.assertEqual(self.errors, [])
+				self.assertEqual([url for url, _ in calls],
+					['https://slack.com/api/files.getUploadURLExternal', self.UPLOAD_URL,
+															 'https://slack.com/api/files.completeUploadExternal'])
+				(_, ticket), (_, upload), (_, share) = calls
+				self.assertEqual(ticket['headers'], {'Authorization': 'Bearer xoxb-test'})
+				self.assertTrue(upload['data'].startswith(b'\x89PNG')) # the chart's bytes, as the body
+				self.assertEqual((ticket['data']['filename'], ticket['data']['length']),
+					('chart.png', len(upload['data'])))
+				self.assertEqual((share['data']['channel_id'], json.loads(share['data']['files'])),
+					('C1', [{'id': 'F1'}]))
+				self.assertTrue(share['data']['initial_comment'].startswith(heading),
+					share['data']['initial_comment'][:80]) # the report
+				# not sent again as text
+				self.assertNotIn(share['data']['initial_comment'].split('\n'), [payload for _, payload in self.sent])
+
+	def test_a_slack_upload_that_fails_goes_out_as_text(self):
+		"""A Slack upload failing at any step goes as text, but not on a last-step timeout, when it may have arrived."""
+		self.env(slackBotToken='xoxb-test')
+		self.patch(render, 'draw', lambda chart: io.BytesIO(b'png'))
+
+		good = self.upload_steps()
+		slow = transports.requests.exceptions.ReadTimeout('slow')
+		text = [('text', ['*Heading*', 'line 2'])]
+		for outcomes, expect in (([Resp({'ok': False, 'error': 'missing_scope'})], text),
+			([ConnectionError('down')], text),
+								 ([good[0], Resp(None, status=500, text='nope')], text),
+								 ([good[0], good[1], Resp({'ok': False, 'error': 'not_in_channel'})], text),
+								 ([good[0], good[1], slow], []), # it may have been shared after all
+								 (good, [])):
+			with self.subTest(outcomes=outcomes):
+				self.slack_answers(*outcomes)
+				self.sent.clear()
+				with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
+					delivery.deliver(transports.SlackBot('C1'), delivery.Report(['*Heading*', 'line 2'], CHART))
+				self.assertEqual(self.sent, expect)
+
+	def test_scheduled_reports_to_slack_carry_charts_only_as_the_bot(self):
+		"""Scheduled reports reach Slack's incoming webhook as text; the Slack bot (slackChannelID) and Telegram get
+		their charts."""
+		self.env(slackBotToken='xoxb-test')
+
+		# an incoming webhook cannot upload
+		self.env(webhooks={**config.env.webhooks, 'slack': 'https://hooks.slack.com/services/x'})
+		sent = []
+		self.patch(transports.Transport, 'chart', lambda dest, image, caption, buttons=None:
+			sent.append((dest.name, 'chart')) or {'ok': True})
+		self.patch(transports.Transport, 'text', lambda dest, lines, buttons=None:
+			sent.append((dest.name, 'text')) or [])
+		for report in (price.run, lambda: performance.run(past_days=7), shorts.run):
+			with self.subTest(report=report):
+				sent.clear()
+				report()
+				self.assertEqual(sorted(set(sent)), [('slack', 'text'), ('telegram', 'chart')])
+				self.env(slackChannelID='C9') # as the bot, the report's chart goes too
+				sent.clear()
+				report()
+				self.assertEqual(sorted(set(sent)), [('slack', 'chart'), ('telegram', 'chart')])
+				self.env(slackChannelID='')
+
+	def test_slack_without_a_bot_token_gets_text_and_no_chart_is_drawn(self):
+		"""Without the bot token, Slack replies are text and no chart is drawn for them."""
+		self.env(slackBotToken=False)
+
+		def refuse(*a, **k):
+			"""Fail: no chart should be drawn."""
+			raise AssertionError('a chart was drawn for Slack')
+		self.patch(render, 'draw', refuse) # every chart is drawn there
+		self.env(webhooks={**config.env.webhooks, 'slack': 'https://hooks.slack.com/services/x'})
+		for cmd in ('.beta', '.pie Rob', '.buy', '.sell', '.peg', '.price', '.price top', '.price T1 7d',
+				'.performance 7d', '.shorts', '.history T1', '.compare T1 T2'):
+			with self.subTest(cmd=cmd):
+				self.sent.clear(); self.errors.clear()
+				worker.process_request(inbound.Message('slack', 'C1', '<@U1>', cmd, BOT, 'U'))
+				self.assertEqual(self.errors, [])
+				self.assertTrue(self.sent)
+				self.assertEqual({kind for kind, *_ in self.sent}, {'text'})
+
+	def test_the_graph_preference_off_sends_no_chart_anywhere(self):
+		"""graph=False means no charts, as the README says; it used to hold back only the price-line charts."""
+		self.prefs(graph=False)
+
+		self.env(slackBotToken='xoxb-test')
+
+		posts = []
+		self.patch(transports.net, 'post', lambda url, **k: posts.append(url) or Resp({'ok': True, 'result': True}))
+		for service, chat in (('telegram', '55'), ('slack', 'C1')):
+			for cmd in ('.price', '.price top', '.performance 7d', '.shorts', '.buy', '.peg', '.beta', '.compare T1 T2',
+					'.pie Rob'):
+				with self.subTest(service=service, cmd=cmd):
+					self.sent.clear(); self.errors.clear()
+					worker.process_request(inbound.Message(service, chat, '@u', cmd, BOT, 'U'))
+					self.assertEqual(self.errors, [])
+					self.assertEqual({kind for kind, *_ in self.sent}, {'text'})
+		self.assertEqual([url for url in posts if 'sendPhoto' in url or 'files.' in url], [])
+
+	def test_a_chart_telegram_refuses_goes_out_as_text(self):
+		"""A chart Telegram refuses or that fails goes as text, but not after a timeout, when it may have arrived."""
+		self.patch(render, 'draw', lambda chart: io.BytesIO(b'png'))
+		def post(outcome):
+			"""A net.post that answers with outcome, or raises it."""
+			def fake(url, **kwargs):
+				"""The outcome, or raise it."""
+				if isinstance(outcome, Exception):
+					raise outcome
+				return outcome
+			return fake
+		text = [('text', ['Heading', 'line 2'])]
+		for outcome, expect in ((Resp({'ok': False, 'error_code': 400,
+			'description': "Bad Request: can't parse entities"}, status=400), text),
+								(Resp(None, status=502), text), (ConnectionError('down'), text),
+								(transports.requests.exceptions.ReadTimeout('slow'), []), # it may be there
+								(Resp({'ok': True, 'result': {'message_id': 1}}), [])):
+			with self.subTest(outcome=outcome):
+				self.sent.clear()
+				self.patch(transports.net, 'post', post(outcome))
+				with contextlib.redirect_stderr(io.StringIO()), contextlib.redirect_stdout(io.StringIO()):
+					delivery.deliver(transports.TelegramBot('5'), delivery.Report(['Heading', 'line 2'], CHART, [[]]))
+				self.assertEqual(self.sent, expect)
+
+
+class DiscordWebhook(FinbotCase):
+	"""Discord's own webhook endpoint (the Slack-like one takes no flags): no previews, no pings, charts attached."""
+	URL = 'https://discord.com/api/webhooks/1/abc'
+
+	def setUp(self):
+		"""Discord as the only destination, sent through the real path, its posts recorded."""
+		super().setUp()
+		self.env(webhooks={'discord': self.URL})
+		self.patch(transports.Transport, 'text', REAL_TEXT)
+		self.patch(transports.Transport, 'chart', REAL_CHART)
+		self.posts = []
+		self.patch(transports.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or Resp({'id': '1'}))
+
+	def test_text_goes_to_the_native_endpoint_without_previews_or_pings(self):
+		"""Text goes to the native endpoint with ?wait=true, without previews or mentions."""
+		transports.DiscordWebhook(self.URL).post_text('hello @everyone')
+		url, kwargs = self.posts[0]
+		self.assertEqual(url, self.URL + '?wait=true') # without wait Discord answers 204, which counted as a failure
+		self.assertEqual(kwargs['json'], {'content': 'hello @everyone', 'flags': 4, 'allowed_mentions': {'parse': []}})
+
+	def test_a_scheduled_report_carries_its_chart(self):
+		"""A scheduled Discord report has its chart attached and the report as its message, within 2,000 characters."""
+		price.run()
+		(url, kwargs), = self.posts
+		self.assertEqual(url, self.URL + '?wait=true')
+		payload = json.loads(kwargs['data']['payload_json'])
+		self.assertEqual((payload['flags'], payload['allowed_mentions'], payload['attachments'][0]['filename']),
+			(4, {'parse': []}, 'chart.png'))
+		self.assertTrue(payload['content'].startswith('**≥ 5% at close'), payload['content'][:60])
+		self.assertLessEqual(len(payload['content']), 2000)
+		name, image, kind = kwargs['files']['files[0]']
+		self.assertEqual((name, kind), ('chart.png', 'image/png'))
+		self.assertTrue(image.read().startswith(b'\x89PNG'))
+
+	def test_a_long_report_is_split_by_size(self):
+		"""A long report is split into messages of at most 2,000 characters."""
+		line = 'x' * 150
+		transports.DiscordWebhook(self.URL).text([line] * 30)
+		sizes = [len(kwargs['json']['content']) for _, kwargs in self.posts]
+		self.assertGreater(len(sizes), 2)
+		self.assertTrue(all(size <= 2000 for size in sizes), sizes) # 20 lines a chunk used to make 3,019 characters
+
+	def test_links_are_masked_links(self):
+		"""Links are masked links, brackets and parentheses escaped."""
+		self.assertEqual(DISCORD.link('https://x.example/a(b)', 'T[1]'), '[T\\[1\\]](<https://x.example/a%28b%29>)')
+
+
+class SlackSending(FinbotCase):
+	"""Slack replies go through chat.postMessage with the bot token; scheduled reports through the incoming webhook."""
+
+	def setUp(self):
+		"""The real send path, its posts recorded."""
+		super().setUp()
+		self.patch(transports.Transport, 'text', REAL_TEXT)
+		self.patch(transports.Transport, 'chart', REAL_CHART)
+		self.posts = []
+		self.patch(transports.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or Resp({'ok': True}))
+
+	def test_link_previews_are_turned_off_in_the_body(self):
+		"""Link previews are turned off in the body, where Slack reads them, for replies and for the webhook."""
+		self.env(slackBotToken='xoxb-test')
+
+		transports.SlackBot('C1').post_text('hi')
+		transports.SlackWebhook('https://hooks.slack.com/services/T/B/x').post_text('hi')
+		(_, reply), (_, hook) = self.posts
+		self.assertEqual((reply['data']['unfurl_links'], reply['data']['unfurl_media']), ('false', 'false'))
+		self.assertEqual((hook['json']['unfurl_links'], hook['json']['unfurl_media']), (False, False))
+		for kwargs in (reply, hook): # they used to go in the headers, where Slack does not look
+			self.assertNotIn('unfurl_links', kwargs['headers'])
+
+	def test_a_reply_slack_refuses_is_logged_as_failed(self):
+		"""Slack refuses with HTTP 200 and ok false (the bot is not in the channel, say): a failure, logged as one."""
+		self.env(slackBotToken='xoxb-test')
+		self.patch(transports.net, 'post', lambda url, **kwargs: Resp({'ok': False, 'error': 'not_in_channel'}))
+		with contextlib.redirect_stderr(io.StringIO()) as err:
+			self.assertIsNone(transports.SlackBot('C1').post_text('hi'))
+		self.assertIn('Slack chat.postMessage to C1 failed: Slack chat.postMessage: not_in_channel', err.getvalue())
+		self.assertNotIn('OK', err.getvalue())
+
+	def test_a_reply_without_a_bot_token_is_logged_not_sent(self):
+		"""A Slack reply without the bot token is logged, once each, and not sent."""
+		self.env(slackBotToken=False)
+
+		err = io.StringIO()
+		with contextlib.redirect_stderr(err), contextlib.redirect_stdout(io.StringIO()):
+			for cmd in ('.help', '.price', '.compare T1 T2'):
+				worker.process_request(inbound.Message('slack', 'C1', '<@U1>', cmd, BOT, 'U'))
+		self.assertEqual((self.posts, self.errors), ([], []))
+		refusal = 'ERROR slack: a reply to channel C1 needs slackBotToken in finbot.env; not sent'
+		self.assertEqual(err.getvalue().count(refusal), 3)
+
+	def test_a_command_at_work_shows_an_hourglass_on_its_message(self):
+		"""While a command works, ⏳ is on the message it answers (Slack has no typing notice for bots); it comes off
+		once the answer is out, so the answer does not wait for it. Only with that message and the bot token."""
+		self.env(slackBotToken='xoxb-test')
+		self.patch(transports.Transport, 'TYPING', (0, 7, 10))
+		added = threading.Event()
+		def post(url, **kwargs):
+			"""Record the post; note when ⏳ is on."""
+			self.posts.append((url, kwargs))
+			if url.endswith('/reactions.add'):
+				added.set()
+			return Resp({'ok': True})
+		self.patch(transports.net, 'post', post)
+		help_text = about.prepare_help
+		self.patch(about, 'prepare_help', lambda *a: (added.wait(5), help_text(*a))[1]) # a command slow enough for ⏳
+		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.help', BOT, 'U', '1.2'))
+		self.assertEqual([url.rsplit('/', 1)[-1] for url, _ in self.posts],
+			['reactions.add', 'chat.postMessage', 'reactions.remove'])
+		reaction = {'channel': 'C1', 'timestamp': '1.2', 'name': 'hourglass_flowing_sand'}
+		self.assertEqual((self.posts[0][1]['data'], self.posts[2][1]['data']), (reaction, reaction))
+		self.assertFalse(transports.SlackBot('C1').shows_typing) # no message to put it on
+		self.env(slackBotToken='')
+		self.assertFalse(transports.SlackBot('C1', '1.2').shows_typing)
+
+	def test_an_hourglass_slack_refuses_is_tried_once(self):
+		"""⏳ that Slack refuses (a missing scope) is logged and not tried again, and nothing is taken off."""
+		self.env(slackBotToken='xoxb-test')
+		refuse = Resp({'ok': False, 'error': 'missing_scope'})
+		self.patch(transports.net, 'post', lambda url, **kwargs: self.posts.append((url, kwargs)) or refuse)
+		chat = transports.SlackBot('C1', '1.2')
+		with contextlib.redirect_stderr(io.StringIO()) as err:
+			chat.typing()
+			chat.typing()
+			chat.typing_done()
+		self.assertEqual([url for url, _ in self.posts], ['https://slack.com/api/reactions.add'])
+		self.assertIn('Slack reactions.add failed: Slack reactions.add: missing_scope', err.getvalue())
+
+	def test_the_slack_api_says_what_is_missing(self):
+		"""slack_api without the bot token says so (it used to be a TypeError)."""
+		self.env(slackBotToken=False)
+
+		with self.assertRaisesRegex(RuntimeError,
+				'Slack files.getUploadURLExternal: needs slackBotToken in finbot.env'):
+			transports.slack_api('files.getUploadURLExternal', filename='chart.png', length=1) # it was a TypeError
+		self.assertEqual(self.posts, [])
+
+
+class SlackButtons(FinbotCase):
+	"""A chart's period buttons on Slack: a Block Kit message (the report, the chart uploaded into no channel, the
+	buttons) that a press redraws in place, where the presses reach finbot (the bot token and slackOutgoingWebhook)."""
+	UPLOAD_URL = 'https://files.slack.com/upload/v1/abc'
+
+	def setUp(self):
+		"""The real send path, its posts recorded and answered as Slack would; no waiting."""
+		super().setUp()
+		self.patch(transports.Transport, 'text', REAL_TEXT)
+		self.patch(transports.Transport, 'chart', REAL_CHART)
+		self.env(slackBotToken='xoxb-test', slackOutgoingWebhook='https://example.com/slack')
+		self.posts, self.refusals, self.waits = [], {}, []
+		self.patch(transports.time, 'sleep', self.waits.append)
+		def post(url, **kwargs):
+			"""Record the post and answer it: a refusal set for the method, while there is one left, else success."""
+			self.posts.append((url, kwargs))
+			method = url.rsplit('/', 1)[-1]
+			if self.refusals.get(method):
+				return Resp({'ok': False, 'error': self.refusals[method].pop(0)})
+			if url == self.UPLOAD_URL:
+				return Resp(None, text='OK - 3')
+			return Resp({'ok': True, 'upload_url': self.UPLOAD_URL, 'file_id': 'F1', 'files': [{'id': 'F1'}],
+				'ts': '1.5'})
+		self.patch(transports.net, 'post', post)
+
+	def calls(self):
+		"""The methods posted, in order (the upload as 'upload')."""
+		return ['upload' if url == self.UPLOAD_URL else url.rsplit('/', 1)[-1] for url, _ in self.posts]
+
+	def blocks(self, method):
+		"""The blocks of the last call of this method."""
+		return json.loads([kwargs for url, kwargs in self.posts if url.endswith('/' + method)][-1]['data']['blocks'])
+
+	def test_a_chart_reply_carries_its_period_buttons(self):
+		"""A chart reply is a Block Kit message: the report, the chart (uploaded into no channel, shown by its id) and
+		the period buttons, the one shown marked; nothing goes as text."""
+		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U'))
+		self.assertEqual(self.errors, [])
+		self.assertEqual(self.calls(), ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal',
+			'chat.postMessage'])
+		self.assertNotIn('channel_id', self.posts[2][1]['data']) # in no channel: the message shows it
+		blocks = self.blocks('chat.postMessage')
+		self.assertEqual([block['type'] for block in blocks], ['section', 'image', 'actions'])
+		self.assertTrue(blocks[0]['text']['text'].startswith('*≥ 5% at close'))
+		self.assertEqual(blocks[1]['slack_file'], {'id': 'F1'})
+		buttons_shown = blocks[2]['elements']
+		self.assertEqual([button['text']['text'] for button in buttons_shown], ['1D', '7D', '1M', '1Y', 'Max'])
+		self.assertEqual([button.get('style') for button in buttons_shown], ['primary'] + [None] * 4)
+		self.assertEqual(buttons_shown[1]['value'], 'c|l|w|5,0')
+
+	def test_a_press_redraws_the_chart_in_place(self):
+		"""A press uploads the chart for that period (a file cannot be replaced) and puts it in the message with
+		chat.update, the period pressed now marked; Telegram is not asked to answer the press."""
+		worker.process_callback(inbound.Press('slack', 't1', 'C1', '1.5', 'c|l|w|5,0'))
+		self.assertEqual(self.errors, [])
+		self.assertEqual(self.calls(), ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal',
+			'chat.update'])
+		update = self.posts[-1][1]['data']
+		self.assertEqual((update['channel'], update['ts']), ('C1', '1.5'))
+		blocks = self.blocks('chat.update')
+		self.assertEqual(blocks[1]['slack_file'], {'id': 'F1'})
+		self.assertEqual([button.get('style') for button in blocks[2]['elements']], [None, 'primary'] + [None] * 3)
+
+	def test_a_press_deletes_the_chart_it_replaced(self):
+		"""Each press uploads its period's chart and, once the message shows it, deletes the file shown before: no
+		chart is reused, so the bot's files do not pile up. The file a press shows is never deleted."""
+		uploads = iter(f'F{n}' for n in range(1, 10))
+		record = transports.net.post
+		def post(url, **kwargs):
+			"""Answer as Slack does, with a new file id for each upload."""
+			answer = record(url, **kwargs)
+			if url.endswith('/files.getUploadURLExternal'):
+				return Resp({'ok': True, 'upload_url': self.UPLOAD_URL, 'file_id': next(uploads)})
+			return answer
+		self.patch(transports.net, 'post', post)
+		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U')) # F1, in message 1.5
+		for _ in range(2):
+			worker.process_callback(inbound.Press('slack', 't1', 'C1', '1.5', 'c|l|w|5,0')) # F2, then F3
+		deleted = [kwargs['data']['file'] for url, kwargs in self.posts if url.endswith('/files.delete')]
+		self.assertEqual((deleted, self.errors), (['F1', 'F2'], []))
+		self.assertEqual(self.blocks('chat.update')[1]['slack_file'], {'id': 'F3'})
+
+	def test_a_chart_not_ready_to_show_is_shown_a_moment_later(self):
+		"""Slack refusing the blocks of a chart just uploaded (invalid_blocks) is asked again after a wait."""
+		self.refusals['chat.postMessage'] = ['invalid_blocks']
+		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U'))
+		self.assertEqual((self.calls()[-2:], self.waits), (['chat.postMessage', 'chat.postMessage'], [1]))
+		self.assertEqual((self.sent, self.errors), ([], []))
+
+	def test_buttons_slack_will_not_show_leave_a_plain_chart(self):
+		"""When Slack will not show the chart with its buttons, the chart is uploaded into the channel without them."""
+		self.refusals['chat.postMessage'] = ['invalid_blocks'] * 4
+		with contextlib.redirect_stderr(io.StringIO()):
+			worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U'))
+		self.assertEqual(self.calls()[-3:], ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal'])
+		self.assertEqual(self.posts[-1][1]['data']['channel_id'], 'C1')
+		self.assertEqual((self.waits, self.sent, self.errors), ([1, 2, 4], [], []))
+
+	def test_without_slackOutgoingWebhook_a_chart_has_no_buttons(self):
+		"""Without slackOutgoingWebhook, no press could reach finbot, so a chart goes without buttons."""
+		self.env(slackOutgoingWebhook='')
+		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U'))
+		self.assertEqual(self.calls(), ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal'])
+		self.assertEqual(self.posts[-1][1]['data']['channel_id'], 'C1')
+
+	def test_a_long_report_takes_several_sections(self):
+		"""A report longer than a section takes holds is split into sections, and an empty one takes none."""
+		caption = '\n'.join('x' * 99 for _ in range(70)) # 6,999 characters
+		blocks = transports.chart_blocks(caption, [], 'F1')
+		self.assertEqual([block['type'] for block in blocks], ['section'] * 3 + ['image'])
+		self.assertTrue(all(len(block['text']['text']) <= 3000 for block in blocks[:3]))
+		self.assertEqual([block['type'] for block in transports.chart_blocks('', [], 'F1')], ['image'])
+
+	def test_a_chart_message_that_times_out_is_not_posted_again(self):
+		"""A chart with its buttons whose chat.postMessage times out may have arrived: no plain copy follows."""
+		record = transports.net.post
+		def post(url, **kwargs):
+			"""Time out on chat.postMessage, as a slow Slack does."""
+			if url.endswith('/chat.postMessage'):
+				self.posts.append((url, kwargs))
+				raise transports.requests.exceptions.ReadTimeout('slow')
+			return record(url, **kwargs)
+		self.patch(transports.net, 'post', post)
+		with contextlib.redirect_stderr(io.StringIO()):
+			worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U'))
+		self.assertEqual(self.calls(), ['files.getUploadURLExternal', 'upload', 'files.completeUploadExternal',
+			'chat.postMessage'])
+		self.assertEqual(self.errors, [])
+
+
+if __name__ == '__main__':
+	unittest.main()
diff --git a/tests/test_sharesight.py b/tests/test_sharesight.py
new file mode 100644
index 00000000..966834e5
--- /dev/null
+++ b/tests/test_sharesight.py
@@ -0,0 +1,274 @@
+"""Sharesight's portfolios, holdings, gains and trades on the answers saved under doc/, its failures,
+and the cache warmer."""
+import datetime, os, unittest
+import gevent
+from tests.common import *
+
+
+def portfolios():
+	"""{name: id} of the portfolios to report on."""
+	return {portfolio.name: portfolio.id for portfolio in sharesight.portfolios()}
+
+
+class SharesightApi(ProviderCase):
+	def test_portfolios_from_real_response(self):
+		"""The portfolio list parses from a real answer."""
+		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_portfolios.json')))
+		self.assertEqual(portfolios(), {'Rob': 445825, 'RobSMSF': 684141})
+
+	def test_portfolio_exclusion_and_inclusion(self):
+		"""exclude_portfolios and include_portfolios pick the portfolios reported on."""
+		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_portfolios.json')))
+		self.prefs(exclude_portfolios=['445825'])
+
+		self.assertEqual(portfolios(), {'RobSMSF': 684141})
+		self.prefs(exclude_portfolios=[], include_portfolios=['445825'])
+		self.assertEqual(portfolios(), {'Rob': 445825})
+
+	def test_portfolio_failures_raise(self):
+		"""An error answer, Sharesight unreachable or no portfolios is a RuntimeError."""
+		for failure in (Resp({'error': 'nope', 'error_code': 'x'}, 401), ConnectionError('down'),
+				Resp({'portfolios': []})):
+			with self.subTest(failure=failure):
+				self.http(sharesight, failure)
+				with self.assertRaises(RuntimeError):
+					sharesight.portfolios()
+
+	def test_an_answer_that_is_not_json_says_so(self):
+		"""An HTML error page from Sharesight is a RuntimeError naming the request and the status, not a JSON parse
+		error."""
+		self.http(sharesight, Resp(None, 502, '<html>Bad gateway</html>'))
+		with self.assertRaisesRegex(RuntimeError, 'Sharesight trades: a 502 answer that is not JSON'):
+			sharesight.Portfolio('RobSMSF', 684141).trades(1)
+
+	def test_trades_from_real_response(self):
+		"""Trades parse from a real answer, tagged with their portfolio, which the cache leaves out: it holds the answer
+		as sent."""
+		self.http(sharesight, Resp(sample('sharesight', 'finbot_sharesight_trades_684141_1.json')))
+		trades = sharesight.Portfolio('RobSMSF', 684141).trades(1)
+		self.assertEqual(len(trades), 2)
+		self.assertEqual(trades[0]['portfolio'], 'RobSMSF')
+		self.assertEqual(trades[0]['symbol'], 'FANG')
+		self.assertEqual(cache.read('finbot_sharesight_trades_684141_1.json'), sample('sharesight',
+			'finbot_sharesight_trades_684141_1.json'))
+
+	def test_trade_failures_raise(self):
+		"""Sharesight unreachable, or an error answer, is a RuntimeError."""
+		self.http(sharesight, ConnectionError('down'))
+		with self.assertRaises(RuntimeError):
+			sharesight.Portfolio('RobSMSF', 684141).trades(1)
+		self.http(sharesight, Resp({'trades': [], 'error': 'bad', 'error_code': 'x'}, 400))
+		with self.assertRaisesRegex(RuntimeError, 'Sharesight trades: bad'):
+			sharesight.Portfolio('RobSMSF', 684141).trades(2)
+
+	def test_an_oauth_error_is_reported_with_its_description(self):
+		"""Sharesight's refusal of the credentials says why, in OAuth's words (it used to be a KeyError on its missing
+		error_code)."""
+		self.patch(sharesight, '_token', REAL_TOKEN)
+		refusal = {'error': 'invalid_client', 'error_description': 'Client authentication failed'}
+		self.patch(sharesight.net, 'post', lambda url, **kwargs: Resp(refusal, 401))
+		with self.assertRaisesRegex(RuntimeError, 'Sharesight auth: invalid_client: Client authentication failed'):
+			sharesight.portfolios()
+
+	def test_an_answer_without_what_was_asked_for_is_an_error(self):
+		"""A refusal without an error, or a 200 without the list asked for, is an error saying so, not a KeyError
+		later."""
+		for answer in (Resp({'message': 'nope'}, 403), Resp({'links': {}})):
+			with self.subTest(status=answer.status_code):
+				self.http(sharesight, answer)
+				why = f'Sharesight portfolios: a {answer.status_code} answer without portfolios'
+				with self.assertRaisesRegex(RuntimeError, why):
+					sharesight.portfolios()
+
+	def test_performance_and_holdings_from_real_response(self):
+		"""A real holdings report parses: its tickers come out sorted, US ones without a suffix, and the gains are the
+		portfolio's and each holding's."""
+		sent = sample('sharesight', 'finbot_sharesight_performance_684141_0.json')
+		self.http(sharesight, Resp(sent))
+		self.assertEqual(len(sharesight._performance(684141, 0)['holdings']), 18)
+		portfolio = sharesight.Portfolio('RobSMSF', 684141)
+		tickers = portfolio.tickers()
+		self.assertIn('ARM', tickers) # US venues carry no Yahoo suffix
+		self.assertEqual(tickers, sorted(tickers))
+		self.assertEqual(portfolio.gain(0), float(sent['report']['capital_gain_percent']))
+		self.assertEqual(sorted(portfolio.gains(0)), tickers)
+		self.assertTrue(all(isinstance(gain, float) for gain in portfolio.gains(0).values()))
+
+	def test_performance_retries_when_sharesight_is_busy(self):
+		"""While Sharesight says too many requests are in flight, the request is tried again."""
+		busy = Resp({'error': 'Too many parallel requests', 'error_code': 'x'}, 429)
+		getter = self.http(sharesight, busy, busy,
+			Resp(sample('sharesight', 'finbot_sharesight_performance_684141_0.json')))
+		self.assertIn('holdings', sharesight._performance(684141, 0))
+		self.assertEqual(len(getter.calls), 3)
+
+	def test_performance_gives_up_after_repeated_busy_or_hard_errors(self):
+		"""Busy every time, a hard error, or unreachable: a RuntimeError in the end."""
+		for failure in (Resp({'error': 'Too many parallel requests', 'error_code': 'x'}, 429),
+						Resp({'error': 'forbidden', 'error_code': 'x'}, 403), ConnectionError('down')):
+			with self.subTest(failure=failure):
+				self.http(sharesight, failure)
+				with self.assertRaises(RuntimeError):
+					sharesight._performance(684141, 7)
+
+	def test_an_answer_without_a_report_says_so(self):
+		"""A performance answer without a report is an error naming the portfolio, and is not cached."""
+		self.http(sharesight, Resp({}))
+		with self.assertRaisesRegex(RuntimeError, 'performance of portfolio 684141: a 200 answer without report'):
+			sharesight._performance(684141, 7)
+		self.assertIsNone(cache.read(sharesight.performance_cache_file(684141, 7))) # and is not cached
+
+
+class SharesightKeepWarm(ProviderCase):
+	"""The background refresher keeps the portfolio list and the holdings report (days=0) cached ahead of their TTLs."""
+
+	def serve_sharesight(self):
+		"""Answer with the real portfolio list and holdings report; returns the URLs asked for."""
+		portfolios = sample('sharesight', 'finbot_sharesight_portfolios.json')
+		performance = sample('sharesight', 'finbot_sharesight_performance_684141_0.json')
+		calls = []
+		def get(url, **kwargs):
+			"""The holdings report or the portfolio list, as asked."""
+			calls.append(url)
+			return Resp(performance if '/performance' in url else portfolios)
+		self.patch(sharesight.net, 'get', get)
+		return calls
+
+	def age(self, cache_file, seconds):
+		"""Make a cache file `seconds` older."""
+		path = os.path.join(config.env.cache_dir, cache_file)
+		old = os.path.getmtime(path) - seconds
+		os.utime(path, (old, old))
+
+	def test_cold_cache_is_filled_then_left_alone(self):
+		"""A cold cache is filled (the list and each portfolio's holdings, as of today), then left alone while fresh."""
+		calls = self.serve_sharesight()
+		sharesight.warm()
+		self.assertEqual(len(calls), 3) # portfolio list + one holdings report per portfolio (2)
+		self.assertEqual(sum('/performance' in c for c in calls), 2)
+		today = 'start_date=' + datetime.date.today().isoformat()
+		self.assertTrue(all(today in c for c in calls if '/performance' in c)) # days=0
+		sharesight.warm()
+		self.assertEqual(len(calls), 3) # nothing refetched while fresh
+
+	def test_refreshes_before_the_ttl_runs_out(self):
+		"""A cache file at 95% of its lifetime is fetched again, and only that one."""
+		calls = self.serve_sharesight()
+		sharesight.warm()
+		calls.clear()
+		# 95% of its lifetime
+		self.age(sharesight.performance_cache_file(684141, 0), 0.95 * config.prefs.performance_cache_seconds)
+		sharesight.warm()
+		self.assertEqual(len(calls), 1)
+		self.assertIn('684141/performance', calls[0])
+		calls.clear()
+		self.age(sharesight.PORTFOLIOS_CACHE, 0.95 * config.prefs.cache_seconds)
+		sharesight.warm()
+		self.assertEqual(len(calls), 1)
+		self.assertTrue(calls[0].endswith('/portfolios'))
+
+	def test_a_request_never_waits_on_a_cache_the_job_has_warmed(self):
+		"""Once warmed, the list and the holdings report need no request."""
+		calls = self.serve_sharesight()
+		sharesight.warm()
+		calls.clear()
+		self.assertEqual(portfolios(), {'Rob': 445825, 'RobSMSF': 684141})
+		self.assertEqual(len(sharesight._performance(684141, 0)['holdings']), 18)
+		self.assertEqual(calls, [])
+
+	def test_every_holding_is_served_from_the_warmed_cache(self):
+		"""Every portfolio's tickers come from the warmed cache, with no request."""
+		calls = self.serve_sharesight()
+		sharesight.warm()
+		calls.clear()
+		tickers = sharesight.tickers() # the portfolio list and each portfolio's holdings report, no cache of its own
+		self.assertIn('ARM', tickers)
+		self.assertEqual(calls, [])
+
+	def test_failed_refresh_keeps_the_old_cache(self):
+		"""A refresh that fails keeps the cache it would have replaced."""
+		self.serve_sharesight()
+		sharesight.warm()
+		self.age(sharesight.PORTFOLIOS_CACHE, 0.95 * config.prefs.cache_seconds)
+		self.http(sharesight, ConnectionError('down'))
+		with self.assertRaises(RuntimeError):
+			sharesight.warm()
+		self.assertEqual(portfolios(), {'Rob': 445825, 'RobSMSF': 684141}) # still served from the cache
+
+	def test_a_pass_says_when_the_next_is_due(self):
+		"""A pass reads each cache file once and gives the seconds until the next refresh: 90% of the shortest cache
+		lifetime once everything is fresh."""
+		self.serve_sharesight()
+		due = sharesight.warm() # a cold cache, filled
+		shortest = min(config.prefs.cache_seconds, config.prefs.performance_cache_seconds)
+		self.assertAlmostEqual(due, 0.9 * shortest, delta=5)
+		reads = []
+		real = cache.read
+		self.patch(cache, 'read', lambda name, *a, **k: (reads.append(name), real(name, *a, **k))[1])
+		self.age(sharesight.performance_cache_file(684141, 0), 600)
+		due = sharesight.warm()
+		self.assertAlmostEqual(due, 0.9 * config.prefs.performance_cache_seconds - 600, delta=5)
+		files = [sharesight.PORTFOLIOS_CACHE] + [sharesight.performance_cache_file(id, 0) for id in (445825, 684141)]
+		self.assertEqual(sorted(reads), sorted(files))
+
+	class Stopped(Exception):
+		"""Raised by the stand-in for keep_warm's wake event, to end the loop."""
+
+	def waits(self, count, on_wait=lambda n: None, **kwargs):
+		"""Run keep_warm until it has waited `count` times, noting each wait instead of sleeping (on_wait(n) at the
+		nth); returns the waits."""
+		waits, stopped = [], self.Stopped
+		class Wake:
+			"""keep_warm's wake event, which nobody sets."""
+			def wait(self, seconds):
+				"""Note the wait instead of sleeping, and end the loop after the last."""
+				waits.append(seconds)
+				on_wait(len(waits))
+				if len(waits) == count:
+					raise stopped
+			def clear(self):
+				"""Nothing to clear."""
+		self.patch(sharesight, '_wake', Wake())
+		with self.assertRaises(self.Stopped):
+			sharesight.keep_warm(**kwargs)
+		return waits
+
+	def test_loop_survives_failures_and_backs_off(self):
+		"""keep_warm survives a failure, retries slowly, then sleeps until the next cache is due."""
+		outcomes = [RuntimeError('boom'), 3000]
+		def warm():
+			"""Fail once, then succeed."""
+			outcome = outcomes.pop(0)
+			if isinstance(outcome, Exception):
+				raise outcome
+			return outcome
+		self.patch(sharesight, 'warm', warm)
+		self.assertEqual(self.waits(2, retry=300), [300, 3000]) # a slow retry after the failure, then until one is due
+
+	def test_with_cache_off_nothing_is_fetched_until_a_reload_turns_it_on(self):
+		"""With cache off, keep_warm idles without asking Sharesight, and starts when a reload turns it on."""
+		passes = []
+		self.patch(sharesight, 'warm', lambda: passes.append(1) or 3000)
+		self.prefs(cache=False)
+		def reload(n):
+			"""At the second wait, switch the cache on, as a reload would."""
+			if n == 2:
+				config.prefs = dataclasses.replace(config.prefs, cache=True)
+		self.assertEqual(self.waits(4, reload, idle=60), [60, 60, 3000, 3000]) # no request every 5 seconds while off
+		self.assertEqual(len(passes), 2)
+
+	def test_a_flush_wakes_the_loop_to_refill_at_once(self):
+		"""A flush wakes keep_warm, which refills the caches at once instead of when they are next due."""
+		passes = []
+		self.patch(sharesight, 'warm', lambda: passes.append(1) or 3000)
+		loop = gevent.spawn(sharesight.keep_warm)
+		self.addCleanup(loop.kill)
+		gevent.sleep(0.05)
+		self.assertEqual(len(passes), 1) # then it sleeps until the next cache is due, 50 minutes away
+		sharesight.wake()
+		gevent.sleep(0.05)
+		self.assertEqual(len(passes), 2)
+
+
+if __name__ == '__main__':
+	unittest.main()
diff --git a/tests/test_telegram_api.py b/tests/test_telegram_api.py
new file mode 100644
index 00000000..4c2ef64d
--- /dev/null
+++ b/tests/test_telegram_api.py
@@ -0,0 +1,96 @@
+"""Telegram's Bot API: the webhook registered once, as bot.py starts."""
+import contextlib, io, os, subprocess, unittest
+from tests.common import *
+
+
+class TelegramWebhook(FinbotCase):
+	"""bot.py registers the webhook once as it starts; importing lib.telegram, as each job does, does nothing."""
+
+	def setUp(self):
+		"""A Telegram bot with an outgoing webhook configured."""
+		super().setUp()
+		self.env(webhooks={'telegram': 'https://api.telegram.org/botX/'})
+		self.env(telegramOutgoingWebhook='https://example.com/telegram')
+
+
+	def test_register_unregisters_registers_and_names_the_bot(self):
+		"""register() removes the webhook, sets it again and returns the bot's @name."""
+		calls = []
+		def post(url, **kwargs):
+			"""Record the method and URL, and answer like Telegram."""
+			calls.append((url.rsplit('/', 1)[1], kwargs.get('params', {}).get('url')))
+			return Resp({'ok': True, 'result': {'username': 'finbot'} if url.endswith('getMe') else True})
+		self.patch(telegram.net, 'post', post)
+		with contextlib.redirect_stderr(io.StringIO()):
+			self.assertEqual(telegram.register(), '@finbot')
+		self.assertEqual(calls, [('setWebhook', ''), ('setWebhook', 'https://example.com/telegram'), ('getMe', None)])
+
+	def test_a_refused_call_raises(self):
+		"""A call Telegram refuses raises with its description."""
+		self.patch(telegram.net, 'post',
+			lambda url, **k: Resp({'ok': False, 'description': 'Bad Request: bad webhook: HTTPS url must be provided'}))
+		with contextlib.redirect_stderr(io.StringIO()), self.assertRaises(RuntimeError) as caught:
+			telegram.register()
+		self.assertIn('HTTPS url must be provided', str(caught.exception))
+
+	def test_edits_toasts_and_typing_go_through_call(self):
+		"""A chart edit that Telegram refuses is an error (unless nothing changed); a toast, a text edit or a typing
+		notice that fails, or an edit lost on the way, is only logged."""
+		def answer(outcome):
+			"""A net.post that answers with outcome, or raises it."""
+			def post(url, **kwargs):
+				"""The outcome, or raise it."""
+				if isinstance(outcome, Exception):
+					raise outcome
+				return outcome
+			return post
+		refused = Resp({'ok': False, 'description': 'Bad Request: message to edit not found'})
+		unchanged = Resp({'ok': False,
+			'description': 'Bad Request: message is not modified: specified new message content is the same'})
+		chat = transports.TelegramBot('5')
+		for outcome, raises in ((refused, True), (unchanged, False), (ConnectionError('down'), False),
+				(Resp({'ok': True, 'result': True}), False)):
+			with self.subTest(outcome=outcome):
+				self.patch(telegram.net, 'post', answer(outcome))
+				with contextlib.redirect_stderr(io.StringIO()):
+					if raises:
+						with self.assertRaisesRegex(RuntimeError, 'not found'):
+							chat.edit_chart('9', io.BytesIO(b'png'), 'caption')
+					else:
+						chat.edit_chart('9', io.BytesIO(b'png'), 'caption')
+					telegram.answerCallbackQuery('1', 'Loading…')
+					chat.edit_text('9', 'text')
+					chat.typing()
+
+	def test_importing_never_calls_telegram(self):
+		"""Importing the modules a job process uses makes no request."""
+		code = ('import requests\n'
+				'def refuse(*a, **k): raise SystemExit("network call on import")\n'
+				'requests.Session.request = refuse\n'
+				'from lib import telegram, yahoo, worker, jobrun\n'
+				'print("quiet")')
+		with tempfile.TemporaryDirectory() as data:
+			out = subprocess.run([sys.executable, '-c', code], cwd=ROOT, capture_output=True, text=True,
+				env={**os.environ, 'DATA_DIR': data, 'telegramBotToken': '123:abc',
+					'telegramOutgoingWebhook': 'https://example.com/telegram'})
+		self.assertEqual(out.stdout.split()[-1:], ['quiet'], out.stderr)
+
+	def test_the_bot_will_not_start_unregistered_and_never_logs_the_token(self):
+		"""bot.py exits when the webhook cannot be registered, and never logs the token."""
+		skip_if_old_env(self)
+		with tempfile.TemporaryDirectory() as data:
+			env = {k: v for k, v in bare_environment().items() if 'proxy' not in k.lower()}
+			env.update(DATA_DIR=data, port='0', scheduler='False', telegramBotToken='123456:SECRET-token',
+				telegramOutgoingToken='tok',
+					   telegramOutgoingWebhook='https://example.com/telegram',
+					   # unreachable, so the registration fails at once and offline
+					   HTTPS_PROXY='http://127.0.0.1:9', https_proxy='http://127.0.0.1:9')
+			out = subprocess.run([sys.executable, 'bot.py'], cwd=ROOT, env=env, capture_output=True, text=True,
+				timeout=120)
+		self.assertEqual(out.returncode, 1, out.stderr)
+		self.assertIn('ERROR ⚠️ telegram webhook:', out.stderr)
+		self.assertNotIn('SECRET-token', out.stderr + out.stdout)
+
+
+if __name__ == '__main__':
+	unittest.main()
diff --git a/tests/test_yahoo.py b/tests/test_yahoo.py
new file mode 100644
index 00000000..1511a4aa
--- /dev/null
+++ b/tests/test_yahoo.py
@@ -0,0 +1,314 @@
+"""Yahoo's quotes, details and charts: the real parsers on the answers saved under doc/, and failures.
+HTTP is mocked at lib.net, so the production parsing code runs unchanged."""
+import datetime, os, unittest
+from tests.common import *
+
+
+class YahooQuotes(ProviderCase):
+	def test_parses_real_v7_quote(self):
+		"""A real v7 quote parses into finbot's fields."""
+		self.http(yahoo, Resp(sample('yahoo', 'sample_yahoo_v7.json')))
+		data = yahoo.fetch(['AAPL'])['AAPL']
+		self.assertEqual(data['profile_title'], 'Apple')
+		self.assertEqual(data['percent_change'], 0.09)
+		self.assertEqual(data['percent_change_postmarket'], -0.13)
+		self.assertEqual(data['quoteType'], 'EQUITY')
+		self.assertEqual(data['marketState'], 'CLOSED')
+		self.assertEqual(data['profile_exchange'], 'NasdaqGS')
+		self.assertGreater(data['market_cap'], 1e12)
+
+	def test_second_call_is_served_from_cache(self):
+		"""A second fetch of the same ticker comes from the cache."""
+		getter = self.http(yahoo, Resp(sample('yahoo', 'sample_yahoo_v7.json')))
+		yahoo.fetch(['AAPL'])
+		yahoo.fetch(['AAPL'])
+		self.assertEqual(len(getter.calls), 1)
+
+	def test_falls_back_to_second_host(self):
+		"""When the first Yahoo host fails, the second is asked."""
+		getter = self.http(yahoo, Resp(status=500), Resp(sample('yahoo', 'sample_yahoo_v7.json')))
+		self.assertIn('AAPL', yahoo.fetch(['AAPL']))
+		self.assertIn('query1', getter.calls[1])
+
+	def test_all_hosts_failing_raises(self):
+		"""Every host failing, or Yahoo unreachable, is a RuntimeError."""
+		for failure in (Resp(status=500), ConnectionError('down')):
+			with self.subTest(failure=failure):
+				self.http(yahoo, failure)
+				with self.assertRaises(RuntimeError):
+					yahoo.fetch(['AAPL'])
+
+	def test_empty_result_is_none(self):
+		"""An empty result is no quotes."""
+		self.http(yahoo, Resp({'quoteResponse': {'result': None, 'error': {'code': 'x'}}}))
+		self.assertEqual(yahoo.fetch(['NOPE']), {})
+
+	def test_an_error_answer_is_not_cached(self):
+		"""Yahoo's error answer is no quotes, and is not cached: the next fetch asks again."""
+		error = Resp({'quoteResponse': {'result': None, 'error': {'code': 'x'}}})
+		self.http(yahoo, error, Resp(sample('yahoo', 'sample_yahoo_v7.json')))
+		self.assertEqual(yahoo.fetch(['AAPL']), {})
+		self.assertIn('AAPL', yahoo.fetch(['AAPL']))
+
+	def test_the_cache_holds_yahoos_answer_so_a_parser_fix_needs_no_flush(self):
+		"""Quotes are cached as Yahoo sent them, not parsed: a change to the parsing applies to a cached answer."""
+		body = sample('yahoo', 'sample_yahoo_v7.json')
+		self.http(yahoo, Resp(body))
+		yahoo.fetch(['AAPL'])
+		self.http(yahoo, ConnectionError('down')) # from here on, only the cache answers
+		self.patch(yahoo, 'QUOTE_FIELDS', yahoo.QUOTE_FIELDS + (('short_name', 'shortName', None),))
+		self.assertEqual(yahoo.fetch(['AAPL'])['AAPL']['short_name'], 'Apple Inc.')
+
+	def test_with_the_cache_off_every_fetch_asks_yahoo(self):
+		"""With the cache preference off, every fetch asks Yahoo, and nothing is written to the cache."""
+		self.prefs(cache=False)
+		getter = self.http(yahoo, Resp(sample('yahoo', 'sample_yahoo_v7.json')))
+		yahoo.fetch(['AAPL'])
+		yahoo.fetch(['AAPL'])
+		self.assertEqual(len(getter.calls), 2)
+		self.assertEqual([name for name in os.listdir(config.env.cache_dir) if name.startswith('finbot_yahoo_')], [])
+
+	def test_no_tickers_is_none(self):
+		"""No tickers asked for: no quotes, and no request."""
+		self.assertEqual(yahoo.fetch([]), {})
+
+	def test_incomplete_quotes_are_skipped(self):
+		"""A quote without a price or a name is left out."""
+		body = sample('yahoo', 'sample_yahoo_v7.json')
+		good = body['quoteResponse']['result'][0]
+		no_price = {k: v for k, v in good.items() if k != 'regularMarketPrice'} | {'symbol': 'NOPRICE'}
+		no_name = {k: v for k, v in good.items() if k not in ('longName', 'shortName')} | {'symbol': 'NONAME'}
+		body['quoteResponse']['result'] = [good, no_price, no_name]
+		self.http(yahoo, Resp(body))
+		self.assertEqual(list(yahoo.fetch(['AAPL', 'NOPRICE', 'NONAME'])), ['AAPL'])
+
+
+class YahooDetail(ProviderCase):
+	@staticmethod
+	def combined():
+		"""quoteSummary as Yahoo returns it for several modules at once, assembled from the per-module samples."""
+		merged = {}
+		for module in ('price', 'summaryProfile', 'summaryDetail', 'defaultKeyStatistics', 'financialData',
+				'calendarEvents', 'earnings'):
+			merged.update(sample('yahoo', f'sample_yahoo_v10_{module}.json')['quoteSummary']['result'][0])
+		return {'quoteSummary': {'result': [merged], 'error': None}}
+
+	def test_parses_real_quote_summary(self):
+		"""A real quoteSummary parses, the company's description included."""
+		self.http(yahoo, Resp(self.combined()))
+		data = yahoo.fetch_detail('NVDA')['NVDA']
+		self.assertEqual(data['profile_title'], 'NVIDIA')
+		self.assertTrue(data['profile_bio'])
+
+	def test_a_cached_detail_needs_no_cookie(self):
+		"""A detail served from the cache takes neither the cookie nor the crumb, nor their lock."""
+		self.http(yahoo, Resp(self.combined()))
+		yahoo.fetch_detail('NVDA')
+		def refuse():
+			"""Fail: the credentials must not be needed."""
+			raise AssertionError('a cached detail took the Yahoo auth lock')
+		self.patch(yahoo, 'credentials', refuse)
+		self.assertIn('NVDA', yahoo.fetch_detail('NVDA'))
+
+	def test_the_cache_holds_yahoos_answer_so_a_parser_fix_needs_no_flush(self):
+		"""A detail is cached as Yahoo sent it, not parsed: a change to the parsing applies to a cached answer."""
+		self.http(yahoo, Resp(self.combined()))
+		yahoo.fetch_detail('NVDA')
+		self.http(yahoo, ConnectionError('down')) # from here on, only the cache answers
+		self.patch(yahoo, 'DETAIL_FIELDS', yahoo.DETAIL_FIELDS + (('short_name', ('price', 'shortName')),))
+		self.assertEqual(yahoo.fetch_detail('NVDA')['NVDA']['short_name'], 'NVIDIA Corporation')
+
+	def test_a_symbol_yahoo_does_not_know_is_empty_and_an_outage_an_error(self):
+		"""A detail of a symbol Yahoo does not know (404 Not Found) is empty, an answer; Yahoo unreachable or refusing
+		is an error, where it used to read as an unknown symbol ("I could not find TSLA")."""
+		unknown = {'quoteSummary': {'result': None, 'error': {'code': 'Not Found', 'description': 'Quote not found'}}}
+		self.http(yahoo, Resp(unknown, status=404))
+		self.assertEqual(yahoo.fetch_detail('DRNA'), {})
+		for failure in (Resp(status=404), Resp(status=500), ConnectionError('down')):
+			with self.subTest(failure=failure):
+				self.http(yahoo, failure)
+				with self.assertRaises(RuntimeError):
+					yahoo.fetch_detail('TSLA')
+
+	def test_missing_names_give_empty(self):
+		"""A detail without a name, or without a result, is empty."""
+		body = self.combined()
+		body['quoteSummary']['result'][0]['price'].update(longName=None, shortName=None)
+		self.http(yahoo, Resp(body))
+		self.assertEqual(yahoo.fetch_detail('DUB'), {})
+		self.http(yahoo, Resp({'quoteSummary': {'result': None, 'error': {}}}))
+		self.assertEqual(yahoo.fetch_detail('GONE'), {})
+
+
+class YahooChart(ProviderCase):
+	def test_real_chart_to_bars(self):
+		"""A real v8 chart becomes its daily bars: the day, the open and the close."""
+		rows = yahoo.bars(sample('yahoo', 'sample_yahoo_v8_chart.json'))
+		self.assertEqual(len(rows), 2)
+		self.assertTrue(all(isinstance(day, datetime.date) and close for day, _, close in rows))
+
+	def test_a_bar_is_dated_in_its_exchanges_time_zone(self):
+		"""An ASX bar, stamped at Sydney's open (23:00 UTC the day before), is dated the day it traded: it used to be
+		dated the day before, so N-day changes and chart dates slipped a session."""
+		open_sydney = datetime.datetime(2026, 1, 5, 23, 0, tzinfo=datetime.timezone.utc) # 10:00 on 6 Jan in Sydney
+		chart = {'chart': {'result': [{'meta': {'exchangeTimezoneName': 'Australia/Sydney'},
+			'timestamp': [int(open_sydney.timestamp())], 'indicators': {'quote': [{'open': [1.0], 'close': [2.0]}]}}]}}
+		self.assertEqual(yahoo.bars(chart), [(datetime.date(2026, 1, 6), 1.0, 2.0)])
+
+	def test_a_refused_crumb_is_renewed_once(self):
+		"""Yahoo refusing the crumb (401) gets a new cookie and crumb, and the request is asked again, once."""
+		renewed = []
+		fresh = lambda renew=False: renewed.append(renew) or ('A=B', 'new' if renew else 'old')
+		self.patch(yahoo, 'credentials', fresh)
+		getter = self.http(yahoo, Resp(status=401), Resp(status=401), Resp(sample('yahoo', 'sample_yahoo_v7.json')))
+		self.assertIn('AAPL', yahoo.fetch(['AAPL']))
+		crumbs = [url.rsplit('crumb=', 1)[1] for url in getter.calls]
+		self.assertEqual((renewed, crumbs), ([False, True], ['old', 'old', 'new']))
+		self.http(yahoo, Resp(status=401))
+		with self.assertRaises(RuntimeError): # a new crumb refused too: an error, not a loop
+			yahoo.fetch(['MSFT'])
+
+	def test_the_credentials_are_one_file_renewed_together(self):
+		"""The cookie and the crumb Yahoo issued for it are cached together and renewed together; no cookie, or no
+		crumb, is an error, where a stale fallback pair turned it into a refusal later."""
+		self.patch(yahoo, 'credentials', REAL_CREDENTIALS)
+		cookie = Resp(None, status=404)
+		cookie.headers = {'Set-Cookie': 'A3=d=AQAB; Max-Age=31557600; Domain=.yahoo.com; Path=/; Secure'}
+		getter = self.http(yahoo, cookie, Resp(None, text='crumb1'))
+		self.assertEqual(yahoo.credentials(), ('A3=d=AQAB', 'crumb1'))
+		self.assertEqual(yahoo.credentials(), ('A3=d=AQAB', 'crumb1')) # from the cache now
+		self.assertEqual(len(getter.calls), 2)
+		nothing = Resp(None, status=404)
+		nothing.headers = {}
+		self.http(yahoo, nothing)
+		with self.assertRaisesRegex(RuntimeError, 'no cookie'):
+			yahoo.credentials(renew=True)
+
+	def test_fetch_chart_json_failures_raise(self):
+		"""A chart Yahoo will not give is a RuntimeError."""
+		for failure in (Resp(status=404), ConnectionError('down')):
+			with self.subTest(failure=failure):
+				self.http(yahoo, failure)
+				with self.assertRaises(RuntimeError):
+					yahoo.fetch_chart_json('AAPL', full=True)
+
+	@staticmethod
+	def chart(closes, failing=()):
+		"""A v8 chart response with one daily bar per close, the last one today (14:30 UTC, New York's open)."""
+		today = datetime.datetime.now(datetime.timezone.utc).replace(hour=14, minute=30, second=0, microsecond=0)
+		stamps = [int((today - datetime.timedelta(days=len(closes) - 1 - i)).timestamp()) for i in range(len(closes))]
+		meta = {'currency': 'USD', 'symbol': 'T1', 'exchangeTimezoneName': 'America/New_York',
+			'regularMarketTime': stamps[-1],
+				'regularMarketPrice': closes[-1], 'longName': 'Co T1'}
+		quote = {'open': closes, 'high': closes, 'low': closes, 'close': closes, 'volume': [1] * len(closes)}
+		return {'chart': {'result': [{'meta': meta, 'timestamp': stamps, 'indicators': {'quote': [quote]}}],
+			'error': None}}
+
+	def test_1d_is_the_change_since_the_previous_close(self):
+		"""1D is the change since the previous close, and 5D the week's; with a quote, its price and today's change
+		count, the chart's being up to cache_seconds old."""
+		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
+		# up 20% in a week, down 3% today
+		self.http(yahoo, Resp(self.chart([90] * 30 + [100, 105, 110, 115, 120, 116.4])))
+		table, line = yahoo.price_history('T1')
+		self.assertEqual(table['1D'], -3)
+		self.assertGreater(table['5D'], 10) # the week, not the day
+		table, line = yahoo.price_history('T1', quote={'regularMarketPrice': 125.0, 'percent_change': 7.4})
+		self.assertEqual((table['1D'], line.price), (7, 125.0))
+
+	def test_a_record_is_from_the_whole_history(self):
+		"""historic_high reads the whole history (the chart from period1=0), its first day aside: records seeded
+		from 10 years were no all-time records."""
+		urls = []
+		chart = self.chart([5.0, 50.0, 10.0, 20.0])
+		self.patch(yahoo, 'historic_high', REAL[('lib.yahoo', 'historic_high')])
+		self.patch(yahoo.net, 'get', lambda url, **kwargs: urls.append(url) or Resp(chart))
+		self.assertEqual(yahoo.historic_high('T1'), (50.0, 10.0)) # 5.0, the first day, aside
+		self.assertIn('period1=0&', urls[0])
+
+	def test_a_symbol_yahoo_does_not_know_is_no_data(self):
+		"""Yahoo's 404 for a symbol it does not know is NoData, an answer; any other 404 is an error, not an answer."""
+		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
+		unknown = {'chart': {'result': None, 'error': {'code': 'Not Found', 'description': 'No data found'}}}
+		self.http(yahoo, Resp(unknown, status=404))
+		with self.assertRaises(yahoo.NoData) as caught:
+			yahoo.price_history('XYZ', 7)
+		self.assertEqual(caught.exception.lines, ['no price history for XYZ'])
+		self.http(yahoo, Resp(status=404)) # an endpoint that moved, say
+		with self.assertRaises(RuntimeError) as caught:
+			yahoo.price_history('XYZ', 7)
+		self.assertNotIsInstance(caught.exception, answer.Answer)
+
+	def test_a_failed_chart_fetch_raises(self):
+		"""A chart refused (429) or unreachable is a RuntimeError, which callers turn into leaving the ticker out."""
+		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
+		for failure in (Resp(status=429), ConnectionError('down')):
+			with self.subTest(failure=failure):
+				self.http(yahoo, failure)
+				with self.assertRaises(RuntimeError):
+					yahoo.price_history('AAPL', 7)
+
+	def test_one_failed_ticker_does_not_take_the_report_down(self):
+		"""One ticker without a chart is left out of a price list, a benchmark without one is said to have none, and
+		only a report on that one ticker fails, in one line."""
+		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
+		quotes = lambda ts: {t: market_entry(t) | {'regularMarketPrice': 110.0} for t in ts} # charts over HTTP
+		self.patch(yahoo, 'fetch', quotes)
+		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t)})
+		self.patch(sharesight, '_portfolios', lambda seconds=None: [{'name': 'Rob', 'id': 1}])
+		self.patch(sharesight, '_performance', lambda pid, days=None, seconds=None: # every ticker needs its history
+			{'holdings': [], 'capital_gain_percent': '2.5'})
+		good = Resp(self.chart([100.0] * 20 + [110.0]))
+		def get(url, **kwargs):
+			"""A chart for every ticker but T3 and SPY, which get 404."""
+			return Resp(status=404) if '/chart/T3?' in url or '/chart/SPY?' in url else good
+		self.patch(yahoo.net, 'get', get)
+		def slack(text):
+			"""Run a command on Slack, where reports come as text, and return the text."""
+			self.sent.clear(); self.errors.clear()
+			worker.process_request(inbound.Message('slack', 'C1', '<@U1>', text, BOT, 'U'))
+			self.assertEqual(self.errors, [], text)
+			return '\n'.join(map(str, self.sent[-1][1]))
+		price_list = slack('.price 7d') # T3 is left out, the rest are listed
+		self.assertNotIn('Co T3 (', price_list)
+		self.assertIn('Co T4 (', price_list)
+		report = slack('.performance 7d') # the S&P 500 benchmark has no history
+		self.assertIn('⚠️ S&P 500: no price history', report)
+		self.assertIn('NASDAQ 100', report)
+		self.command('.T3') # the profile goes out without its history lines
+		self.sent.clear(); self.errors.clear()
+		worker.process_request(inbound.Message('telegram', '55', '@u', '.price T3 7d', BOT, 'U'))
+		self.assertEqual(len(self.errors), 1) # the one ticker asked for: one line saying why
+		self.assertTrue(self.errors[0].startswith('⚠️ .price: RuntimeError: 404 error for T3'))
+
+	def test_the_chart_request_carries_the_cookie_of_its_crumb(self):
+		"""A chart request sends the cookie its crumb belongs to."""
+		headers = []
+		self.patch(yahoo.net, 'get',
+			lambda url, **kwargs: headers.append(kwargs.get('headers', {})) or Resp(self.chart([100.0] * 5)))
+		yahoo.fetch_chart_json('T1', full=True)
+		self.assertEqual(headers[0].get('Cookie'), 'A=B') # the crumb used to go without its cookie
+
+	def test_with_the_cache_off_no_chart_is_written(self):
+		"""With the cache preference off, a chart is drawn but not written to the cache."""
+		self.patch(yahoo, 'price_history', REAL[('lib.yahoo', 'price_history')])
+		self.prefs(cache=False)
+		self.http(yahoo, Resp(self.chart([100.0] * 30 + [110.0])))
+		changes, line = yahoo.price_history('T1', 7)
+		self.assertTrue(render.draw(history.line_chart('T1', changes, line, 7)).read().startswith(b'\x89PNG'))
+		self.assertEqual([name for name in os.listdir(config.env.cache_dir) if name.startswith('finbot_chart_')], [])
+
+	def test_price_series_raises_on_failure_and_on_empty_history(self):
+		"""price_series raises when Yahoo fails and when the history has no closes."""
+		self.http(yahoo, Resp(status=500))
+		with self.assertRaises(RuntimeError):
+			yahoo.price_series('AAPL')
+		empty = sample('yahoo', 'sample_yahoo_v8_chart.json')
+		empty['chart']['result'][0]['indicators']['quote'][0]['close'] = [None, None]
+		self.http(yahoo, Resp(empty))
+		with self.assertRaises(RuntimeError):
+			yahoo.price_series('EMPTY')
+
+
+if __name__ == '__main__':
+	unittest.main()

From e5dcdbbc85878dfb47c99b9bf8dca8330034795a Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 11:44:41 +0000
Subject: [PATCH 75/81] Price moves over a period: the list and one stock's
 view take the same figures

A list over a period (.price 7d, price.py 7) took each holding's figure from
Sharesight even with performance_use_sharesight off, while one stock's view
(.price BHP 7d) drew Yahoo's history and the figure that goes with it. Now
both take the same source:
- off (the default): Yahoo's price history for every stock, which is the
  change its chart draws, and covers the watchlist too;
- on: each holding's capital gain from Sharesight, which counts only the time
  it was held, for the list and for one stock, whose view then comes without
  a chart: Sharesight has no price history to draw.

The README's row for the preference says so.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 README.md             |  2 +-
 lib/reports/price.py  | 17 +++++++----------
 lib/worker.py         |  6 ++++--
 tests/common.py       |  4 +++-
 tests/test_reports.py | 18 ++++++++++++++++++
 5 files changed, 33 insertions(+), 14 deletions(-)

diff --git a/README.md b/README.md
index 82da6103..2d5d1adc 100644
--- a/README.md
+++ b/README.md
@@ -501,7 +501,7 @@ The preferences, read by the bot itself. A change takes effect on a reload. A tr
 | `exclude_portfolios` | | Sharesight portfolio IDs to leave out, space separated |
 | `include_portfolios` | | Sharesight portfolio IDs to report on, space separated (empty: all that are not excluded) |
 | `trades_use_yahoo` | `True` | Use Yahoo for trades; set `False` if Yahoo breaks |
-| `performance_use_sharesight` | `False` | Use Sharesight rather than Yahoo for performance over a period |
+| `performance_use_sharesight` | `False` | Price moves over a period (`.price 7d`, `price.py 7`) from Sharesight rather than Yahoo: each holding's capital gain, which counts only the time it was held; watchlist stocks are left out, and one stock's moves (`.price BHP 7d`) come without a chart. Off, every figure comes from Yahoo's price history, the same as the charts |
 | `graph` | `True` | Send charts with reports |
 | `hyperlink` | `True` | Hyperlink tickers in reports |
 | `hyperlinkFooter` | `True` | Add a website link footer to stock lookups |
diff --git a/lib/reports/price.py b/lib/reports/price.py
index 70f03824..44b3f97c 100644
--- a/lib/reports/price.py
+++ b/lib/reports/price.py
@@ -18,30 +18,27 @@
 
 def gather(specific_stock=None, days=None):
 	"""(tickers, market data) for a price report on the stocks followed or one (Yahoo's symbol): Yahoo's quotes, and
-	over `days` each one's change as percent_change_period: a holding's from Sharesight, the others' from Yahoo's
-	history, fetched together up front (performance_use_sharesight: Sharesight's figures only, and the one stock's
-	even when the others take Yahoo's)."""
+	over `days` each one's change as percent_change_period, from its Yahoo history (the change its chart shows), the
+	histories fetched together up front. With performance_use_sharesight, a holding's capital gain from Sharesight
+	instead, which counts only the time it was held (held for 1Y and asked for 5Y, it gives 1Y), and none for the
+	others."""
 	tickers = watchlist.covered(specific_stock)
 	market_data = yahoo.fetch(tickers)
 	if config.prefs.demote_volatile: # trigger_bar compares market caps, in US dollars
 		yahoo.add_usd_market_caps(market_data)
 	if not days:
 		return tickers, market_data
-	# Sharesight reports a holding's performance only for the time it was held: held for 1Y and asked for 5Y, it gives
-	# 1Y
-	if not specific_stock or config.prefs.performance_use_sharesight:
+	if config.prefs.performance_use_sharesight:
 		for portfolio in sharesight.portfolios():
 			for ticker, gain in portfolio.gains(days).items():
 				if ticker in market_data:
 					market_data[ticker]['percent_change_period'] = gain
 				elif ticker in tickers:
 					print("Notice: price.py:", ticker, "has no data", file=sys.stderr)
-	if config.prefs.performance_use_sharesight:
 		return tickers, market_data
-	missing = [ticker for ticker in market_data if 'percent_change_period' not in market_data[ticker]]
 	if not specific_stock:
-		yahoo.prefetch_history(missing)
-	for ticker in missing:
+		yahoo.prefetch_history(list(market_data))
+	for ticker in list(market_data):
 		try:
 			changes, _ = history.price_line(ticker, days, market_data[ticker])
 		except RuntimeError as e: # no history: the one stock asked about says so, a list leaves it out
diff --git a/lib/worker.py b/lib/worker.py
index 4ec49f25..e3102f3e 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -8,6 +8,7 @@
 import sys
 import typing
 from lib import answer
+from lib import config
 from lib import dialects
 from lib import state
 from lib import formats
@@ -245,9 +246,10 @@ def on_session(request, args):
 
 def on_price(request, args):
 	""".price [percent|SYMBOL|top|bottom] [period]: a bare number is a percent, as on the other commands; a period has
-	its unit. One stock over a period is its chart, as its period buttons draw it."""
+	its unit. One stock over a period is its chart, as its period buttons draw it, from the history that the lists
+	take their figures from too (not with performance_use_sharesight: Sharesight has no history to draw)."""
 	symbol = None if args.keyword else ticker(args)
-	if symbol and args.days:
+	if symbol and args.days and not config.prefs.performance_use_sharesight:
 		return chart_reply(request, 'p', [symbol], args.days)
 	top = 10 if args.keyword else None # top or bottom: one chart shows both ends
 	price.run(request.dest, mode='period' if args.days else 'interday', threshold=0 if top else args.percent,
diff --git a/tests/common.py b/tests/common.py
index ff2c2bdd..e9ef0c0e 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -117,8 +117,10 @@ def setUp(self):
 		self.patch(watchlist, 'with_holdings', lambda: TICKERS)
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t)})
+		moved = {t: float(i * 4 - 40) for i, t in enumerate(TICKERS)} # over any period: the holdings' gains below
 		self.patch(yahoo, 'price_history', lambda t, days=None, quote=None: # a history with no line to draw
-			({days: 1.5} if days else {'Max': 5, '1Y': 3}, yahoo.Line('Co ' + t, 'NasdaqGS', 'USD', 100.0, ())))
+			({days: moved.get(t, 1.5)} if days else {'Max': 5, '1Y': 3}, yahoo.Line('Co ' + t, 'NasdaqGS', 'USD', 100.0,
+			())))
 		self.patch(yahoo, 'prefetch_history', lambda ts: None)
 		self.patch(yahoo, 'historic_high', lambda t: (100.0, 50.0))
 		dates = pd.date_range(end=pd.Timestamp.today().normalize(), periods=400)
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 1ad02325..78fa14d1 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -1,6 +1,7 @@
 """Every chat command and scheduled report produces output on mocked data."""
 import unittest
 import datetime
+import re
 from tests.common import *
 
 
@@ -14,6 +15,23 @@ def test_text_commands(self):
 			with self.subTest(cmd=cmd):
 				self.assertTrue(self.command(cmd), f'{cmd} sent nothing')
 
+	def test_a_period_list_and_one_stocks_view_give_the_same_figure(self):
+		"""Over a period, the list and one stock's view take a holding's figure from the same place: Yahoo's history,
+		which its chart draws (-36% for T1), or with performance_use_sharesight Sharesight's gain over the time it was
+		held (+33%), and then no chart, which Sharesight cannot draw. The list used to take Sharesight's either way."""
+		holding = {'id': 1, 'instrument': {'code': 'T1', 'market_code': 'NASDAQ'}, 'capital_gain_percent': '33'}
+		self.patch(sharesight, '_performance', lambda pid, days=None, seconds=None:
+			{'capital_gain_percent': '0', 'holdings': [holding]})
+		def t1(text):
+			"""T1's figure in what the command sends as text."""
+			return re.search(r'Co T1 \(.*?\) ([+-]\d+%)', str(self.command(text))).group(1)
+		self.prefs(graph=False) # the list as text: a chart's caption is recorded by its heading
+		self.assertEqual((t1('.price 7d'), t1('.price T1 7d')), ('-36%', '-36%'))
+		self.prefs(performance_use_sharesight=True)
+		self.assertEqual((t1('.price 7d'), t1('.price T1 7d')), ('+33%', '+33%'))
+		self.prefs(graph=True)
+		self.assertEqual(self.command('.price T1 7d')[0][0], 'text') # no chart: Sharesight has no history
+
 	def test_chart_commands(self):
 		"""Every chart command answers with a chart."""
 		for cmd in ('.price', '.price 7d', '.price top', '.price bottom', '.session', '.premarket', '.postmarket',

From c1eeee3d7bbc7a65b2d00a3801dc99a7bc1b9444 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 11:53:30 +0000
Subject: [PATCH 76/81] Slack: a period pressed within 15 minutes of its upload
 shows that upload again

A press on a chart's period button uploaded the chart afresh every time and
deleted the file it replaced, so switching back to a period meant building,
drawing and uploading it again, with Slack's wait for a new file to become
showable. Now the upload of each period is noted with the report that came
with it, and a press for that period within 15 minutes, on this message or
another, shows that upload again in one chat.update: nothing is built, drawn
or uploaded. If Slack will not show it (a file deleted outside finbot, say),
the press builds and uploads as before.

An upload is deleted once no chart message shows it and it is a minute past
those 15 minutes (the minute lets a press that is reusing it finish showing
it); a file a message shows is never deleted. The chart messages noted the
old way are forgotten, their files left alone.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 doc/slack.md          |  9 +++--
 lib/transports.py     | 90 ++++++++++++++++++++++++++++++++++---------
 lib/worker.py         |  4 +-
 tests/test_sending.py | 52 +++++++++++++++++++++----
 4 files changed, 123 insertions(+), 32 deletions(-)

diff --git a/doc/slack.md b/doc/slack.md
index bd314e5a..038d6150 100644
--- a/doc/slack.md
+++ b/doc/slack.md
@@ -105,10 +105,11 @@ slackChannelID=C0123456789
 
 A chart is uploaded with Slack's file upload API (`files.getUploadURLExternal`, then `files.completeUploadExternal`).
 With its period buttons, it goes in a message of blocks: the report, the chart (uploaded into no channel, and shown by
-its id) and the buttons. A press uploads the chart for that period, shows it in the message in place of the last one
-(Slack cannot replace a file) and deletes the file it replaced, so the bot's files do not pile up. If Slack will not
-show a chart with its buttons, the chart is uploaded into the channel without them; if an upload fails (a missing scope,
-say), the report goes out as text. The log says why.
+its id) and the buttons. A press shows the chart for that period in the message in place of the last one: the upload of
+that period from the last 15 minutes, for this message or another, with the report that came with it, or else the chart
+drawn and uploaded anew (Slack cannot replace a file). An upload that no message shows is deleted once those 15 minutes
+are past, so the bot's files do not pile up. If Slack will not show a chart with its buttons, the chart is uploaded into
+the channel without them; if an upload fails (a missing scope, say), the report goes out as text. The log says why.
 
 `.jobs` lists the scheduled commands with a button for each, as on Telegram; a button from a list that has since
 changed puts the current list in its place.
diff --git a/lib/transports.py b/lib/transports.py
index d386ebdb..a4e05cde 100644
--- a/lib/transports.py
+++ b/lib/transports.py
@@ -17,8 +17,9 @@
 
 MAYBE_SENT = object() # a read timeout on the request that posts a chart: it may have arrived, so no text copy follows
 SUPPRESS_EMBEDS = 4 # Discord message flag: no link previews
-SLACK_CHARTS = 'finbot_slack_charts.json' # 'channel ts' of a chart message -> the file it shows (deleted once replaced)
-SLACK_CHARTS_KEPT = 500 # chart messages noted, newest last: an older one's file stays when a press replaces it
+SLACK_CHARTS = 'finbot_slack_charts.json' # Slack's chart messages: the file each shows, and the charts uploaded
+SLACK_CHARTS_KEPT = 500 # chart messages noted, newest last: an older one shows its file for good
+SLACK_REUSE_SECONDS = 15 * 60 # a period pressed within this long of its chart's upload shows that upload again
 NO_MENTIONS = {'parse': []} # Discord: text that looks like @everyone or a mention pings nobody
 
 
@@ -119,6 +120,11 @@ def typing_done(self):
 	def acknowledge(self, press, text=None):
 		"""Answer a button press, where the network waits for that (Telegram, with `text` as a toast)."""
 
+	def reshow(self, message, buttons):
+		"""Show a chart message's period (the button marked) again as it was posted lately, where the network keeps
+		that: True when it did. Here never: each period is drawn and sent anew."""
+		return False
+
 	def _post_json(self, url, payload, headers=None):
 		"""POST a JSON message: the answer ({} when it is not JSON), or None when the network refused it or failed."""
 		try:
@@ -354,36 +360,77 @@ def post_blocks(self, data, caption, buttons):
 			print("Slack chat.postMessage timed out, the chart may still arrive:", net.redact(e), file=sys.stderr)
 			return MAYBE_SENT
 		if answer.get('ts'):
-			self.shows(answer['ts'], file_id)
+			self.delete(self.shows(answer['ts'], file_id, caption, buttons))
 		return answer
 
+	def reshow(self, message, buttons):
+		"""Show a chart message's period again from its upload of under SLACK_REUSE_SECONDS ago (for this message or
+		another), with the report it came with: nothing is drawn or uploaded. True when it did; False when there is no
+		such upload, or Slack refuses it (a file deleted outside finbot, say)."""
+		view, now = marked(buttons), time.time()
+		uploads = state.load(SLACK_CHARTS, {}).get('uploads', {})
+		recent = [(at, file_id, caption) for file_id, (at, shows, caption) in uploads.items()
+			if view and shows == view and now - at < SLACK_REUSE_SECONDS]
+		if not recent:
+			return False
+		_, file_id, caption = max(recent)
+		try:
+			slack_api('chat.update', channel=self.where, ts=message, text=caption[:SECTION_LIMIT],
+				blocks=json.dumps(chart_blocks(caption, buttons, file_id)))
+		except Exception as e:
+			print("Slack chat.update with an earlier upload failed, uploading anew:", net.redact(e), file=sys.stderr)
+			return False
+		self.delete(self.shows(message, file_id))
+		return True
+
 	def edit_chart(self, message, image, caption, buttons=None):
 		"""Show another period in a chart message: the new chart uploaded (a file cannot be replaced) and shown in its
-		image block, with its caption and buttons, then the file it showed before deleted (nothing reuses an uploaded
-		chart: a period shown again is uploaded again). Raises RuntimeError when Slack refuses the update."""
+		image block, with its caption and buttons. A press for that period within SLACK_REUSE_SECONDS shows this upload
+		again (reshow). Raises RuntimeError when Slack refuses the update."""
 		if hasattr(image, 'seek'):
 			image.seek(0)
 		file_id = self.unshared(image.read() if hasattr(image, 'read') else bytes(image))
 		blocks = chart_blocks(caption, buttons or [], file_id)
 		until_shown(lambda: slack_api('chat.update', channel=self.where, ts=message, text=caption[:SECTION_LIMIT],
 			blocks=json.dumps(blocks)))
-		replaced = self.shows(message, file_id)
-		if replaced and replaced != file_id:
+		self.delete(self.shows(message, file_id, caption, buttons))
+
+	def shows(self, message, file_id, caption=None, buttons=None):
+		"""Note that a chart message shows this file now; given its report, that the file was uploaded now for the
+		period its buttons mark (see reshow). The uploads to delete: those no message shows, a minute past their reuse
+		(a press reusing one has that minute to show it)."""
+		now = time.time()
+		with state.edit(SLACK_CHARTS, {}) as noted:
+			if set(noted) != {'shown', 'uploads'}: # new, or the {'channel ts': file} of before, whose files stay
+				noted.clear()
+				noted.update(shown={}, uploads={})
+			shown, uploads = noted['shown'], noted['uploads'] # 'channel ts': file; file: [time, period, report]
+			if caption is not None:
+				uploads[file_id] = [now, marked(buttons), caption]
+			key = f'{self.where} {message}'
+			shown.pop(key, None)
+			shown[key] = file_id
+			for old in list(shown)[:-SLACK_CHARTS_KEPT]: # its message may show the file for good
+				uploads.pop(shown.pop(old), None)
+			showing = set(shown.values())
+			gone = [old for old, (at, _, _) in uploads.items()
+				if old not in showing and now - at > SLACK_REUSE_SECONDS + 60]
+			for old in gone:
+				del uploads[old]
+			for upload in uploads.values(): # past reuse, a file shown is kept only to be deleted later
+				if now - upload[0] >= SLACK_REUSE_SECONDS:
+					upload[1:] = [None, None]
+		return gone
+
+	def delete(self, files):
+		"""Delete uploaded charts that no message shows any more; a failure is only logged (the file stays among the
+		bot's files, in no channel)."""
+		for file_id in files:
 			try:
-				slack_api('files.delete', file=replaced)
-			except Exception as e: # it stays among the bot's files, in no channel
+				slack_api('files.delete', file=file_id)
+			except Exception as e:
 				print("Slack files.delete failed:", net.redact(e), file=sys.stderr)
 
-	def shows(self, message, file_id):
-		"""Note that a chart message shows this file now (SLACK_CHARTS): the file it showed before, if noted."""
-		key = f'{self.where} {message}'
-		with state.edit(SLACK_CHARTS, {}) as shown:
-			replaced = shown.pop(key, None)
-			shown[key] = file_id
-			for old in list(shown)[:-SLACK_CHARTS_KEPT]:
-				del shown[old]
-		return replaced
-
 	def edit_text(self, message, text, buttons=None):
 		"""Edit a message's text, and its buttons unless buttons is None ([] removes them); a failure is only logged."""
 		data = {'channel': self.where, 'ts': message, 'text': text}
@@ -481,6 +528,11 @@ def chart_blocks(caption, buttons, file_id):
 		([actions_block(row)] if row else [])
 
 
+def marked(buttons):
+	"""The data of the button marked as shown now, which names a chart message's period (None when none is)."""
+	return next((button.data for row in buttons or [] for button in row if button.active), None)
+
+
 def text_blocks(text, buttons):
 	"""Block Kit for a message with buttons: the text, then the buttons, a row of them to a block."""
 	return section_blocks(text) + [actions_block(row) for row in buttons if row]
diff --git a/lib/worker.py b/lib/worker.py
index e3102f3e..4eb0d736 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -77,10 +77,12 @@ def process_callback(press):
 		return
 	try:
 		args = buttons.resolve(pressed.ref)
+		markup = buttons.keyboard(pressed.kind, args, pressed.period)
+		if chat.reshow(press.message_id, markup): # that period as posted lately, where the network keeps it (Slack)
+			return
 		caption, chart = charts.build(pressed.kind, args, service, period=pressed.period)
 		if chart is None: # too few prices in that period to draw a line
 			raise answer.Answer(f"Too few prices to draw {buttons.PERIODS[pressed.period][0]}")
-		markup = buttons.keyboard(pressed.kind, args, pressed.period)
 		chat.edit_chart(press.message_id, render.draw(chart), caption, markup)
 	except answer.Answer as e: # nothing to draw for that period: said in the chat, as a command would
 		delivery.deliver(chat, delivery.Report(e.lines))
diff --git a/tests/test_sending.py b/tests/test_sending.py
index 27f1e9eb..f2a61d9b 100644
--- a/tests/test_sending.py
+++ b/tests/test_sending.py
@@ -478,9 +478,8 @@ def test_a_press_redraws_the_chart_in_place(self):
 		self.assertEqual(blocks[1]['slack_file'], {'id': 'F1'})
 		self.assertEqual([button.get('style') for button in blocks[2]['elements']], [None, 'primary'] + [None] * 3)
 
-	def test_a_press_deletes_the_chart_it_replaced(self):
-		"""Each press uploads its period's chart and, once the message shows it, deletes the file shown before: no
-		chart is reused, so the bot's files do not pile up. The file a press shows is never deleted."""
+	def uploading(self):
+		"""Answer each upload with a new file id, F1 first, as Slack does."""
 		uploads = iter(f'F{n}' for n in range(1, 10))
 		record = transports.net.post
 		def post(url, **kwargs):
@@ -490,12 +489,49 @@ def post(url, **kwargs):
 				return Resp({'ok': True, 'upload_url': self.UPLOAD_URL, 'file_id': next(uploads)})
 			return answer
 		self.patch(transports.net, 'post', post)
-		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U')) # F1, in message 1.5
-		for _ in range(2):
-			worker.process_callback(inbound.Press('slack', 't1', 'C1', '1.5', 'c|l|w|5,0')) # F2, then F3
-		deleted = [kwargs['data']['file'] for url, kwargs in self.posts if url.endswith('/files.delete')]
-		self.assertEqual((deleted, self.errors), (['F1', 'F2'], []))
+
+	def test_a_period_shown_again_within_15_minutes_shows_its_upload_again(self):
+		"""A press for a period whose chart was uploaded in the last 15 minutes, for this message or another, shows
+		that upload again with the report it came with: nothing is built or uploaded. Past that, the period is built and
+		uploaded anew, and the uploads no message shows are deleted, never one a message shows."""
+		self.uploading()
+		built = []
+		build = charts.build
+		self.patch(charts, 'build', lambda *args, **kwargs: built.append(args) or build(*args, **kwargs))
+		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U')) # 1D: F1, in message 1.5
+		worker.process_callback(inbound.Press('slack', 't1', 'C1', '1.5', 'c|l|w|5,0')) # 7D: F2
+		del self.posts[:], built[:]
+		worker.process_callback(inbound.Press('slack', 't2', 'C1', '2.5', 'c|l|d|5,0')) # 1D in another message: F1
+		self.assertEqual((self.calls(), built, self.errors), (['chat.update'], [], []))
+		blocks = self.blocks('chat.update')
+		self.assertEqual(blocks[1]['slack_file'], {'id': 'F1'})
+		self.assertTrue(blocks[0]['text']['text'].startswith('*≥ 5% at close'))
+		with state.edit(transports.SLACK_CHARTS, {}) as noted: # 16 minutes on
+			for upload in noted['uploads'].values():
+				upload[0] -= 16 * 60 + 1
+		worker.process_callback(inbound.Press('slack', 't1', 'C1', '1.5', 'c|l|d|5,0')) # 1D again: F3
 		self.assertEqual(self.blocks('chat.update')[1]['slack_file'], {'id': 'F3'})
+		deleted = [kwargs['data']['file'] for url, kwargs in self.posts if url.endswith('/files.delete')]
+		self.assertEqual((deleted, len(built), self.errors), (['F2'], 1, [])) # message 2.5 shows F1
+
+	def test_an_upload_slack_will_not_show_again_is_uploaded_anew(self):
+		"""When Slack refuses to show an earlier upload again (a file deleted outside finbot, say), the period is
+		built and uploaded anew."""
+		self.uploading()
+		worker.process_request(inbound.Message('slack', 'C1', '<@U1>', '.price', BOT, 'U')) # 1D: F1
+		self.refusals['chat.update'] = ['invalid_blocks']
+		with contextlib.redirect_stderr(io.StringIO()):
+			worker.process_callback(inbound.Press('slack', 't1', 'C1', '1.5', 'c|l|d|5,0'))
+		self.assertEqual(self.calls()[-5:], ['chat.update', 'files.getUploadURLExternal', 'upload',
+			'files.completeUploadExternal', 'chat.update'])
+		self.assertEqual((self.blocks('chat.update')[1]['slack_file'], self.errors), ({'id': 'F2'}, []))
+
+	def test_the_charts_noted_before_uploads_were_reused_are_left_alone(self):
+		"""The chart messages noted as before ({'channel ts': file}) are forgotten, their files never deleted."""
+		state.save(transports.SLACK_CHARTS, {'C1 1.5': 'F0'})
+		worker.process_callback(inbound.Press('slack', 't1', 'C1', '1.5', 'c|l|w|5,0'))
+		self.assertNotIn('files.delete', self.calls())
+		self.assertEqual(state.load(transports.SLACK_CHARTS)['shown'], {'C1 1.5': 'F1'})
 
 	def test_a_chart_not_ready_to_show_is_shown_a_moment_later(self):
 		"""Slack refusing the blocks of a chart just uploaded (invalid_blocks) is asked again after a wait."""

From 875a4bd5379a2c0516db457cd2aa1f883e023d82 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 12:10:50 +0000
Subject: [PATCH 77/81] yahoo_warm.py: the price histories of every stock
 followed, fetched afresh every 8 hours

A period list or a stock's chart reads each ticker's 10-year daily history
from the cache, where it is kept for cache_seconds (12 hours by default);
the first request after it runs out waited on Yahoo, one request per ticker.
A new scheduled job, yahoo_warm.py, fetches every followed stock's history
afresh (the watchlist and the holdings; the watchlist alone while Sharesight
cannot be asked), and the default schedule runs it every 8 hours, so none
runs out. With the cache preference off it fetches nothing, since nothing
would be kept. It is housekeeping, so .jobs does not offer it.

Sharesight's warmer keeps its own timing in the bot (it refreshes each cache
at 90% of its lifetime, and at once after a flush); its thread is now named
sharesight_warm, the name the README gives it beside yahoo_warm.py.

- yahoo.prefetch_history takes the cache age to fetch past (0: all afresh)
  and returns the tickers that failed.
- lib/jobrun.py takes a module name with an underscore.
- A data directory with its own schedule.cron needs the line:
  07 */8 * * * yahoo_warm.py

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md                 |  6 +++---
 README.md                 |  4 ++--
 bot.py                    |  2 +-
 defaults/schedule.cron    |  4 +++-
 lib/jobrun.py             |  2 +-
 lib/jobs.py               |  6 ++++--
 lib/reports/__init__.py   |  9 ++++----
 lib/reports/yahoo_warm.py | 23 ++++++++++++++++++++
 lib/yahoo.py              | 11 +++++++---
 tests/test_schedule.py    |  6 ++++--
 tests/test_scheduler.py   | 44 ++++++++++++++++++++++++++++++++++++++-
 yahoo_warm.py             |  5 +++++
 12 files changed, 102 insertions(+), 20 deletions(-)
 create mode 100644 lib/reports/yahoo_warm.py
 create mode 100755 yahoo_warm.py

diff --git a/CLAUDE.md b/CLAUDE.md
index e3cf60c8..ca10b6f6 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -48,9 +48,9 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   the typing notice while it works. `process_request(message)` and `process_callback(press)` take inbound's objects.
 - `lib/reports/`: what finbot says. Scheduled (each has `run(dest=None, **kwargs)`, which posts to the chat that asked
   or else the default chats, and `job(argv)`, its schedule-line parser): `price performance trades shorts cal rating
-  milestone reminder podcasts cleanup`; the top-level `price.py` and the rest are shims that run them. Chat-only
-  (`prepare_*`): `about` (.help .jobs), `history` (.history .compare), `holdings` (.holdings .who .pie .watchlist),
-  `profile` (.SYMBOL), `rankings` (.marketcap .pe .fpe .peg .beta .buy .sell).
+  milestone reminder podcasts cleanup yahoo_warm`; the top-level `price.py` and the rest are shims that run them.
+  Chat-only (`prepare_*`): `about` (.help .jobs), `history` (.history .compare), `holdings` (.holdings .who .pie
+  .watchlist), `profile` (.SYMBOL), `rankings` (.marketcap .pe .fpe .peg .beta .buy .sell).
 - `lib/delivery.py`: `Report(lines, chart, buttons)`, the destinations (the chat that asked, or the default chats),
   `deliver(dest, report)`, the one-line error report. `lib/transports.py`: a class per way of posting (`TelegramBot`,
   `SlackBot`, `SlackWebhook`, `DiscordWebhook`): splitting, captions, charts, buttons, edits, and the typing notice
diff --git a/README.md b/README.md
index 2d5d1adc..4a962e13 100644
--- a/README.md
+++ b/README.md
@@ -588,7 +588,7 @@ kill -USR1 <pid of bot.py>                  # anything else
 The log says `cache: flushed N files`. To install the `finbot-flush` shortcut: `sudo cp -v finbot-flush.service /etc/systemd/system/ &&
 sudo systemctl daemon-reload`. `finbot.service` itself needs no change. With the bot stopped, `cleanup.py 0` deletes the same files.
 
-While `bot.py` is running with Sharesight configured and `cache=True` (a reload that switches `cache` pauses or resumes it), a background job keeps the Sharesight portfolio list (`cache_seconds`) and each portfolio's holdings report (1 day, `performance_cache_seconds` cache) refreshed at 90% of their cache lifetime, so chat requests do not wait on Sharesight. If a refresh fails, the previous cache keeps being served and the job retries after 5 minutes.
+Two warmers keep the caches full, so that chat requests do not wait on Yahoo or Sharesight. The scheduled `yahoo_warm.py` fetches the price history of every stock followed afresh every 8 hours (see [Scheduling](#scheduling)); a data directory with its own `schedule.cron` needs its line, `07 */8 * * * yahoo_warm.py`. And while `bot.py` is running with Sharesight configured and `cache=True` (a reload that switches `cache` pauses or resumes it), its `sharesight_warm` thread keeps the Sharesight portfolio list (`cache_seconds`) and each portfolio's holdings report (1 day, `performance_cache_seconds` cache) refreshed at 90% of their cache lifetime, so chat requests do not wait on Sharesight. If a refresh fails, the previous cache keeps being served and the job retries after 5 minutes.
 
 ### Hyperlinks
 These `preferences.conf` options relate to hyperlinks:
@@ -615,7 +615,7 @@ hyperlinkProvider=yahoo
 ## Scheduling
 The scheduled reports are started by `bot.py`, each run in its own child process (`python -m lib.jobrun`), so they use the same user, sandbox, environment and data directory as the bot, while heavy work (charts, large reports) cannot hold up incoming chat webhooks. A run that exceeds 5 minutes is stopped. There is no cron to set up, which matters under systemd `DynamicUser=` where only the bot's own user can read the state directory. **This is on by default.** Remove the equivalent entries from any old crontab before the first start, otherwise every job runs twice; or set `scheduler=False` in the environment file to keep using cron.
 
-The schedulable scripts are `trades.py`, `price.py`, `reminder.py`, `cal.py`, `shorts.py`, `performance.py`, `rating.py`, `milestone.py`, `podcasts.py "<feed url>"` and `cleanup.py [days]` (deletes cache files older than `days`, default 30). Other jobs, such as a DNS updater or backups, stay in cron or systemd timers.
+The schedulable scripts are `trades.py`, `price.py`, `reminder.py`, `cal.py`, `shorts.py`, `performance.py`, `rating.py`, `milestone.py`, `podcasts.py "<feed url>"`, `cleanup.py [days]` (deletes cache files older than `days`, default 30) and `yahoo_warm.py` (fetches the price histories of every stock followed afresh, so that period lists and charts never wait on Yahoo; the default schedule runs it every 8 hours, within the 12 hours of `cache_seconds`). Other jobs, such as a DNS updater or backups, stay in cron or systemd timers.
 
 The schedule is `defaults/schedule.cron`, written in **cron-compatible syntax**: five time fields (minute, hour, day of month, month, day of week; `*/20`, `Mon-Fri`, ranges and lists work) or an `@macro` such as `@daily`, then a finbot script and its arguments. Quote an argument that has spaces or special characters (a feed URL with `&`, say). Several commands separated by `;` run one after another. `MAILTO=`, `SHELL=`, `PATH=` and `HOME=` lines are ignored. Anything else a crontab allows (a user column, pipes, other programs) is not supported: such a line is reported at startup and skipped. A time that cannot be or never comes (minute 61, 30 February) is different: finbot refuses to start with it, and a reload with it is refused, leaving the running schedule as it was. Existing crontab lines can be pasted as they are: `> /dev/null` redirects, `{ a; b ;}` groups and a path before the script name are understood.
 ```
diff --git a/bot.py b/bot.py
index c3367c94..7ebe9283 100755
--- a/bot.py
+++ b/bot.py
@@ -226,7 +226,7 @@ def flush_cache(*args):
 		threading.Thread(target=worker.refresh_keyboards, daemon=True).start() # stale DM keyboards, in the background
 	if config.env.sharesight_auth.get('client_id'):
 		# keep the portfolio list and holdings report cached, while cache is on
-		threading.Thread(target=sharesight.keep_warm, daemon=True).start()
+		threading.Thread(target=sharesight.keep_warm, daemon=True, name='sharesight_warm').start()
 	if scheduler:
 		jobs.current = scheduler # for the .jobs chat command
 		threading.Thread(target=scheduler.run_forever, daemon=True).start() # the scheduled reports, formerly cron jobs
diff --git a/defaults/schedule.cron b/defaults/schedule.cron
index bb52b055..36fcfd97 100644
--- a/defaults/schedule.cron
+++ b/defaults/schedule.cron
@@ -22,7 +22,9 @@
 # Weekly
 28  21 * * Sat cal.py earnings; cal.py ex-dividend; price.py 7; performance.py 7; rating.py
 
-# Housekeeping: delete cache files older than 30 days
+# Housekeeping: fetch the price histories of the stocks followed afresh every 8 hours (they are cached 12 hours, so
+# period lists and charts never wait on Yahoo); delete cache files older than 30 days
+07 */8 * * * yahoo_warm.py
 51  18 * * Sat cleanup.py 30
 
 # Monthly
diff --git a/lib/jobrun.py b/lib/jobrun.py
index 3478ee1f..adcea41f 100644
--- a/lib/jobrun.py
+++ b/lib/jobrun.py
@@ -20,7 +20,7 @@
 def main(argv):
 	"""Run one scheduled command: argv is the report's module and its JSON keyword arguments. Returns the exit status: 0
 	when it ran, REPORTED when it failed and said so in the chat, 2 for bad arguments."""
-	if len(argv) != 2 or not re.fullmatch(r'[a-z]+(\.[a-z]+)*', argv[0]):
+	if len(argv) != 2 or not re.fullmatch(r'[a-z][a-z_]*(\.[a-z][a-z_]*)*', argv[0]):
 		print('usage: python -m lib.jobrun <module> <json kwargs>', file=sys.stderr)
 		return 2
 	module = argv[0]
diff --git a/lib/jobs.py b/lib/jobs.py
index 9a11b88e..535ae37d 100644
--- a/lib/jobs.py
+++ b/lib/jobs.py
@@ -27,11 +27,13 @@
 from lib import jobrun
 from lib import delivery
 from lib.reports import cal, cleanup, milestone, performance, podcasts, price, rating, reminder, shorts, trades
+from lib.reports import yahoo_warm
 
 SCRIPTS = {'cal': cal, 'cleanup': cleanup, 'milestone': milestone, 'performance': performance, 'podcasts': podcasts,
-	'price': price, 'rating': rating, 'reminder': reminder, 'shorts': shorts, 'trades': trades}
+	'price': price, 'rating': rating, 'reminder': reminder, 'shorts': shorts, 'trades': trades,
+	'yahoo_warm': yahoo_warm}
 current = None # the running Scheduler, set by bot.py; the chat commands use it
-HIDDEN = ('cleanup', 'podcasts') # scheduled, but not offered in the chat list
+HIDDEN = ('cleanup', 'podcasts', 'yahoo_warm') # scheduled, but not offered in the chat list
 STATE_FILE = 'finbot_jobs.json' # last fire time of every job, so a restart neither repeats a run nor loses one
 JOB_TIMEOUT = 5 * 60 # seconds each command may run before it is stopped and reported
 IGNORED_VARIABLES = ('MAILTO', 'SHELL', 'PATH', 'HOME', 'LOGNAME')
diff --git a/lib/reports/__init__.py b/lib/reports/__init__.py
index ac36214b..5a93b6f8 100644
--- a/lib/reports/__init__.py
+++ b/lib/reports/__init__.py
@@ -1,8 +1,9 @@
 """What finbot says. A module per scheduled report: cal, cleanup (of the cache), milestone, performance, podcasts,
-price, rating, reminder, shorts and trades, each with job(argv), which parses its schedule line, and run(dest=None,
-**kwargs), which runs it, to the chat that asked (dest) or else the default chats; the top-level script of the same
-name runs it from the command line. And the chat-only reports: about (.help, .jobs), history (.history, .compare),
-holdings (.holdings, .who, .pie, .watchlist), profile (.SYMBOL) and rankings (.marketcap, .pe and the rest)."""
+price, rating, reminder, shorts, trades and yahoo_warm (which fills the cache), each with job(argv), which parses its
+schedule line, and run(dest=None, **kwargs), which runs it, to the chat that asked (dest) or else the default chats;
+the top-level script of the same name runs it from the command line. And the chat-only reports: about (.help, .jobs),
+history (.history, .compare), holdings (.holdings, .who, .pie, .watchlist), profile (.SYMBOL) and rankings (.marketcap,
+.pe and the rest)."""
 import os
 import sys
 
diff --git a/lib/reports/yahoo_warm.py b/lib/reports/yahoo_warm.py
new file mode 100644
index 00000000..f8f3e259
--- /dev/null
+++ b/lib/reports/yahoo_warm.py
@@ -0,0 +1,23 @@
+"""Keeps the price history of every stock followed fresh in the cache (scheduled: yahoo_warm.py), so that a period list
+or a chart never waits on Yahoo. Sharesight's caches have their own warmer, the bot's sharesight_warm thread."""
+import sys
+from lib import config, watchlist, yahoo
+
+def run():
+	"""Fetch afresh the price history of every stock followed (the watchlist and the holdings), so each is cached for
+	cache_seconds (12 hours by default) from now: the default schedule runs this every 8 hours, before any runs out.
+	With the cache preference off nothing would be kept, so nothing is fetched. Returns how many it fetched."""
+	if not config.prefs.cache:
+		return 0
+	try:
+		tickers = watchlist.with_holdings()
+	except Exception as e: # Sharesight unreachable, say: the watchlist's at least
+		print("yahoo_warm: no holdings from Sharesight:", e, file=sys.stderr)
+		tickers = watchlist.tickers()
+	failed = yahoo.prefetch_history(tickers, seconds=0)
+	print(f"yahoo_warm: {len(tickers) - len(failed)} of {len(tickers)} price histories fetched", file=sys.stderr)
+	return len(tickers) - len(failed)
+
+def job(argv):
+	"""Arguments of a scheduled run: yahoo_warm.py takes none."""
+	return 'yahoo_warm.py', {}
diff --git a/lib/yahoo.py b/lib/yahoo.py
index 54f213ae..c9a0e9c5 100644
--- a/lib/yahoo.py
+++ b/lib/yahoo.py
@@ -444,12 +444,17 @@ def symbol(typed):
 	"""Yahoo's symbol for one ticker as written in a chat (see symbols): BHP may be BHP.AX."""
 	return symbols([typed])[0]
 
-def prefetch_history(tickers, workers=8):
-	"""Warm the price-history cache for many tickers at once, so the per-ticker lookups that follow are cache hits."""
+def prefetch_history(tickers, workers=8, seconds=None):
+	"""Warm the price-history cache for many tickers at once, so the per-ticker lookups that follow are cache hits: a
+	history is fetched unless its cache file is under `seconds` old (cache_seconds when None; 0 fetches each afresh).
+	Returns the tickers whose history could not be fetched, each logged."""
 	tickers = list(tickers)
-	for ticker, result in zip(tickers, fetch_each(fetch_chart_json, tickers, workers)):
+	failed = []
+	for ticker, result in zip(tickers, fetch_each(lambda t: fetch_chart_json(t, seconds=seconds), tickers, workers)):
 		if isinstance(result, Exception):
 			print("prefetch failed for", ticker, result, file=sys.stderr)
+			failed.append(ticker)
+	return failed
 
 def fetch_chart_json(ticker, days=3665, seconds=None, full=False):
 	"""Yahoo's daily chart for the last `days`, or with full=True its entire history (cached separately). Raises NoData
diff --git a/tests/test_schedule.py b/tests/test_schedule.py
index 59e0bc93..34a9585f 100644
--- a/tests/test_schedule.py
+++ b/tests/test_schedule.py
@@ -9,10 +9,12 @@
 
 class ParseTests(unittest.TestCase):
 	def test_shipped_schedule_has_no_errors(self):
-		"""defaults/schedule.cron parses without errors, its Saturday line into its five commands."""
+		"""defaults/schedule.cron parses without errors, its Saturday line into its five commands; the price histories
+		are fetched afresh every 8 hours, within the 12 hours they are cached."""
 		parsed, errors = jobs.parse_schedule(DEFAULT_SCHEDULE.read_text(), UTC)
 		self.assertEqual(errors, [])
-		self.assertEqual(len(parsed), 9)
+		self.assertEqual(len(parsed), 10)
+		self.assertIn(('07 */8 * * *', ['yahoo_warm.py']), [(j.expr, [c[0] for c in j.commands]) for j in parsed])
 		weekly = next(j for j in parsed if j.expr == '28 21 * * Sat')
 		self.assertEqual([label for label, _, _ in weekly.commands],
 						 ['cal.py earnings', 'cal.py ex-dividend', 'price.py 7', 'performance.py 7', 'rating.py'])
diff --git a/tests/test_scheduler.py b/tests/test_scheduler.py
index 729dd60d..d371a248 100644
--- a/tests/test_scheduler.py
+++ b/tests/test_scheduler.py
@@ -7,7 +7,9 @@
 import gevent
 from tests.common import *
 from lib import jobs
-from lib.reports import cleanup, podcasts, price
+from lib.reports import cleanup, podcasts, price, yahoo_warm
+
+PREFETCH_HISTORY = yahoo.prefetch_history # FinbotCase replaces it; the warm job's test puts it back
 
 
 def run_in_process(label, module, kwargs):
@@ -67,6 +69,46 @@ def test_missing_cache_dir_is_fine(self):
 		self.assertEqual(cleanup.run(days=30), 0)
 
 
+class YahooWarmJob(FinbotCase):
+	def setUp(self):
+		"""The real prefetch, Yahoo answering each history with the ticker's name, except T2's, which fails."""
+		super().setUp()
+		self.patch(yahoo, 'prefetch_history', PREFETCH_HISTORY)
+		def answer(ticker, days, full):
+			"""A history that names its ticker; none for T2."""
+			if ticker == 'T2':
+				raise RuntimeError('Yahoo unreachable')
+			return {'history of': ticker}
+		self.patch(yahoo, '_chart_answer', answer)
+
+	def test_every_followed_stocks_history_is_fetched_afresh(self):
+		"""yahoo_warm fetches the price history of every stock followed even while its cache is fresh, so that none
+		runs out between runs: it counts the ones fetched, and a failure is only logged."""
+		cache.write('finbot_yahoo_history_T1.json', {'history of': 'an hour ago'})
+		with contextlib.redirect_stderr(io.StringIO()) as log:
+			self.assertEqual(yahoo_warm.run(), len(TICKERS) - 1)
+		self.assertEqual(cache.read('finbot_yahoo_history_T1.json'), {'history of': 'T1'})
+		self.assertIn(f'yahoo_warm: {len(TICKERS) - 1} of {len(TICKERS)} price histories fetched', log.getvalue())
+
+	def test_without_sharesight_the_watchlist_is_fetched(self):
+		"""While Sharesight cannot be asked for the holdings, the watchlist's histories are fetched."""
+		def unreachable():
+			"""Sharesight failing, as watchlist.with_holdings meets it."""
+			raise RuntimeError('Sharesight auth: down')
+		self.patch(watchlist, 'with_holdings', unreachable)
+		self.patch(watchlist, 'tickers', lambda: ['W1'])
+		with contextlib.redirect_stderr(io.StringIO()) as log:
+			self.assertEqual(yahoo_warm.run(), 1)
+		self.assertEqual(cache.read('finbot_yahoo_history_W1.json'), {'history of': 'W1'})
+		self.assertIn('no holdings from Sharesight', log.getvalue())
+
+	def test_with_the_cache_off_nothing_is_fetched(self):
+		"""With the cache preference off nothing would be kept, so nothing is fetched."""
+		self.prefs(cache=False)
+		self.patch(yahoo, '_chart_answer', None)
+		self.assertEqual(yahoo_warm.run(), 0)
+
+
 class SchedulerFlag(unittest.TestCase):
 	def flag(self, **env):
 		"""config.env.scheduler in a fresh interpreter, given `env`."""
diff --git a/yahoo_warm.py b/yahoo_warm.py
new file mode 100755
index 00000000..ed868699
--- /dev/null
+++ b/yahoo_warm.py
@@ -0,0 +1,5 @@
+#!/usr/bin/env python3
+"""./yahoo_warm.py [arguments]: runs lib/reports/yahoo_warm.py once, as its schedule line would."""
+if __name__ == '__main__':
+	from lib import reports
+	reports.cli(__file__)

From 5d4f9645d850a30fa1d7e5130ba72f6b68a9fd32 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 12:13:32 +0000
Subject: [PATCH 78/81] .jobs: the system jobs listed under System, without a
 number or a button

.jobs left the scheduled commands that are not run from chat (cleanup,
podcasts, and now yahoo_warm) out of its list entirely. It now lists them
after the numbered commands under a System heading, unnumbered and without
buttons, marked when running like the rest: they run on their schedule only.
`.jobs run cleanup.py` says it is a system job instead of that it is not in
the schedule, and a schedule of system jobs alone lists them rather than
saying there are no jobs.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 README.md               |  2 +-
 lib/jobs.py             | 16 +++++++++-------
 lib/reports/about.py    | 26 ++++++++++++++++----------
 tests/test_scheduler.py | 16 +++++++++++++---
 4 files changed, 39 insertions(+), 21 deletions(-)

diff --git a/README.md b/README.md
index 4a962e13..7c51abcd 100644
--- a/README.md
+++ b/README.md
@@ -638,7 +638,7 @@ sudo systemctl reload finbot        # the bot re-reads the file and logs how man
 ```
 The bot logs the number of jobs when it starts or reloads, and each job as it runs (`journalctl -u finbot`); a line it cannot use is reported by name. If the edited file has no usable line the previous schedule keeps running.
 
-**To run a job from chat**, send `.jobs` (also `.cron`, `.crons` or `.schedule`): it lists the scheduled commands, each once and without times, comments or `.py` (`cleanup` and `podcasts` are scheduled but not listed), and on Telegram and Slack (with its buttons set up) adds a button per command. `.jobs run 2` (or tapping the button) runs that command now, and `.jobs run price.py close` runs it by name; anything not in the schedule is refused, and so is a chain such as `a; b` (run them one at a time). A manual run is exactly a cron run: it reports to all configured chat networks, updates the same state (seen trades, ratings and so on), and does not change the schedule. Anyone who can message the bot can do this; a command that is already running is not started twice.
+**To run a job from chat**, send `.jobs` (also `.cron`, `.crons` or `.schedule`): it lists the scheduled commands, each once and without times, comments or `.py`, and on Telegram and Slack (with its buttons set up) adds a button per command. The system jobs (`cleanup`, `podcasts` and `yahoo_warm`) follow under _System_, without a number or a button: they run on their schedule only. `.jobs run 2` (or tapping the button) runs that command now, and `.jobs run price.py close` runs it by name; anything not in the schedule is refused, and so is a chain such as `a; b` (run them one at a time). A manual run is exactly a cron run: it reports to all configured chat networks, updates the same state (seen trades, ratings and so on), and does not change the schedule. Anyone who can message the bot can do this; a command that is already running is not started twice.
 
 Behaviour worth knowing:
 * A run that was due while the bot was down (a restart, a deploy) still fires once if it was due within the last `scheduler_catchup` minutes (default 5). A restart right after a run does not repeat it.
diff --git a/lib/jobs.py b/lib/jobs.py
index 535ae37d..604a608f 100644
--- a/lib/jobs.py
+++ b/lib/jobs.py
@@ -33,7 +33,7 @@
 	'price': price, 'rating': rating, 'reminder': reminder, 'shorts': shorts, 'trades': trades,
 	'yahoo_warm': yahoo_warm}
 current = None # the running Scheduler, set by bot.py; the chat commands use it
-HIDDEN = ('cleanup', 'podcasts', 'yahoo_warm') # scheduled, but not offered in the chat list
+HIDDEN = ('cleanup', 'podcasts', 'yahoo_warm') # scheduled, but not run from chat: .jobs lists them under System
 STATE_FILE = 'finbot_jobs.json' # last fire time of every job, so a restart neither repeats a run nor loses one
 JOB_TIMEOUT = 5 * 60 # seconds each command may run before it is stopped and reported
 IGNORED_VARIABLES = ('MAILTO', 'SHELL', 'PATH', 'HOME', 'LOGNAME')
@@ -395,15 +395,15 @@ def execute(self, commands):
 					self.busy.discard(label)
 		return outcome
 
-	def commands(self):
-		"""Every offered scheduled command once, sorted by name with numbers in numeric order: [(label, module,
-		kwargs)]. The time, comments, redirects and the other commands of a line are not part of it; HIDDEN scripts are
-		left out."""
+	def commands(self, hidden=False):
+		"""Every scheduled command offered in chat once, or with hidden=True each of the HIDDEN scripts' (System),
+		sorted by name with numbers in numeric order: [(label, module, kwargs)]. The time, comments, redirects and the
+		other commands of a line are not part of it."""
 		unique = {}
-		hidden = {SCRIPTS[name] for name in HIDDEN}
+		system = {SCRIPTS[name] for name in HIDDEN}
 		for job in self.jobs:
 			for command in job.commands:
-				if command[1] not in hidden:
+				if (command[1] in system) == hidden:
 					unique.setdefault(command[0], command)
 		natural = lambda label: [int(part) if part.isdigit() else part for part in re.split(r'(\d+)', label.lower())]
 		return [unique[label] for label in sorted(unique, key=natural)]
@@ -425,6 +425,8 @@ def select(self, text):
 		for command in commands:
 			if command[0] == wanted[0][0]:
 				return command
+		if any(command[0] == wanted[0][0] for command in self.commands(hidden=True)):
+			raise ScheduleError(f'{display(wanted[0][0])} is a system job: it runs on its schedule only')
 		raise ScheduleError(f"'{text}' is not in the schedule; send .jobs to list the commands")
 
 	def pick(self, number, crc):
diff --git a/lib/reports/about.py b/lib/reports/about.py
index 4339191a..b794220f 100644
--- a/lib/reports/about.py
+++ b/lib/reports/about.py
@@ -4,23 +4,29 @@
 from lib import jobs
 
 def prepare_jobs_payload(service, run=None, tappable=False):
-	"""`.jobs`: the scheduled commands, each once; with `run` (a number in that list, or a command's text) that command
-	run now, as cron would. Returns (payload, buttons): with tappable (the chat takes buttons) a button per command, two
-	to a row."""
+	"""`.jobs`: the scheduled commands, each once and numbered, then the system jobs (HIDDEN) under their own heading,
+	unnumbered: they run on their schedule only. With `run` (a number in that list, or a command's text) that command
+	run now, as cron would. Returns (payload, buttons): with tappable (the chat takes buttons) a button per numbered
+	command, two to a row."""
 	scheduler = jobs.current
 	if scheduler is None:
 		return ['Scheduled jobs are off (scheduler=False)'], None
 	if run:
 		return [run_job_line(scheduler, run)], None
-	commands = scheduler.commands()
-	if not commands:
+	commands, system = scheduler.commands(), scheduler.commands(hidden=True)
+	if not commands and not system:
 		return ['No scheduled jobs'], None
-	payload = [service.bold('Scheduled commands:')]
-	for number, (label, _, _) in enumerate(commands, 1):
-		payload.append(f"{number}. {jobs.display(label)}" + (' (running)' if label in scheduler.busy else ''))
-	payload.append("\nRun one with .jobs run NUMBER" + (' or tap it' if tappable else ''))
+	shown = lambda label: jobs.display(label) + (' (running)' if label in scheduler.busy else '')
+	payload = []
+	if commands:
+		payload = [service.bold('Scheduled commands:')] + [f"{number}. {shown(label)}"
+			for number, (label, _, _) in enumerate(commands, 1)]
+	if system:
+		payload += ([''] if payload else []) + [service.bold('System:')] + [shown(label) for label, _, _ in system]
+	if commands:
+		payload.append("\nRun one with .jobs run NUMBER" + (' or tap it' if tappable else ''))
 	markup = None
-	if tappable:
+	if tappable and commands:
 		row = [buttons.job_button(number, jobs.display(label), jobs.checksum(label))
 			for number, (label, _, _) in enumerate(commands, 1)]
 		markup = [row[i:i + 2] for i in range(0, len(row), 2)]
diff --git a/tests/test_scheduler.py b/tests/test_scheduler.py
index d371a248..45e54090 100644
--- a/tests/test_scheduler.py
+++ b/tests/test_scheduler.py
@@ -396,6 +396,8 @@ def test_select_refuses_what_is_not_offered(self):
 				'nonsense', 'ls | wc', '', 'price.py close; trades.py'):
 			with self.assertRaises(jobs.ScheduleError, msg=text):
 				self.s.select(text)
+		with self.assertRaisesRegex(jobs.ScheduleError, '^cleanup is a system job: it runs on its schedule only$'):
+			self.s.select('cleanup.py')
 
 	def test_run_now_runs_one_command_and_leaves_the_schedule_alone(self):
 		"""run_now runs one command and leaves the schedule's timing alone."""
@@ -457,17 +459,25 @@ def list_in_chat(self, text='.jobs'):
 		return self.command(text)[0][1]
 
 	def test_list_in_chat_is_just_the_commands(self):
-		""".jobs and aliases number the commands, without times, comments or redirects, and mark one that is running."""
+		""".jobs and aliases number the commands, without times, comments or redirects, and mark one that is running;
+		the system jobs follow under their own heading, without a number."""
+		feed = jobs.display(podcasts.job(['https://example.com/feed'])[0]) # podcasts and the feed's hash
 		for alias in ('.jobs', '.cron', '.crons', '.schedule', '@' + BOT.lstrip('@') + ' jobs'):
 			text = '\n'.join(map(str, self.list_in_chat(alias)))
 			for line in ('1. performance 7', '2. performance 30', '5. performance 365', '6. price 7', '7. price close',
-					'8. trades'):
+					'8. trades\n\n<b>System:</b>\ncleanup\n' + feed + '\n'):
 				self.assertIn(line, text, alias)
-			for noise in ('*/20', '#', 'Mon-Fri', 'next', 'last', '/dev/null', 'weekly', '.py', 'cleanup', 'podcasts'):
+			for noise in ('*/20', '#', 'Mon-Fri', 'next', 'last', '/dev/null', 'weekly', '.py'):
 				self.assertNotIn(noise, text, alias)
 		self.s.busy.add('price.py 7')
 		self.assertIn('6. price 7 (running)', '\n'.join(map(str, self.list_in_chat())))
 
+	def test_a_schedule_of_system_jobs_alone_lists_them(self):
+		"""A schedule of system jobs alone lists them, with nothing to run or tap."""
+		self.file.write_text('0 2 * * * cleanup.py\n')
+		self.patch(jobs, 'current', self.scheduler())
+		self.assertEqual(about.prepare_jobs_payload(TELEGRAM, '', tappable=True), (['<b>System:</b>', 'cleanup'], None))
+
 	def test_a_chat_that_takes_buttons_gets_one_per_command(self):
 		"""Where buttons work each command gets one, whose data fits in Telegram's 64 bytes, and the list says to tap
 		it; elsewhere there are none, and it does not."""

From c27ed724da29b7275be34c1ee634c552c56ac416 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 12:17:16 +0000
Subject: [PATCH 79/81] Price moves over a period always from Yahoo:
 performance_use_sharesight retired

With Yahoo's price histories kept warm (yahoo_warm.py), Yahoo answers a
period list faster than Sharesight, whose report is per portfolio and per
period, kept 30 minutes and fetched one at a time; and its figure is the
better one: the stock's move over the whole period (Sharesight's capital
gain counts only the time a holding was held), for the watchlist too, and
the change its chart draws. So the Sharesight path goes:
- .price 7d, price.py 7 and .price BHP 7d take every figure from Yahoo's
  history, and one stock over a period is always its chart;
- performance_use_sharesight is retired: a preferences.conf that still sets
  it starts as before, with a note to delete the line;
- Portfolio.gains, which only that path read, goes.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md                 |  3 +--
 README.md                 |  1 -
 defaults/preferences.conf |  1 -
 lib/config.py             |  5 ++---
 lib/reports/price.py      | 13 +------------
 lib/sharesight.py         |  7 -------
 lib/worker.py             |  5 ++---
 tests/common.py           |  2 +-
 tests/test_config.py      |  6 +++---
 tests/test_reports.py     |  9 ++-------
 tests/test_sharesight.py  |  6 ++----
 11 files changed, 14 insertions(+), 44 deletions(-)

diff --git a/CLAUDE.md b/CLAUDE.md
index ca10b6f6..b608385e 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -67,8 +67,7 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   are cached in `DATA_DIR/cache` as sent, not parsed, so a parser fix needs no flush (`cache.cached`).
   `yahoo.symbol(typed)` is Yahoo's symbol for a ticker as written, tried on the local exchange (`exchange_suffix`)
   when Yahoo does not know it bare: every ticker a chat gives goes through it. `sharesight.portfolios(name=None)` gives
-  `Portfolio` objects: `holdings()`, `tickers()`, `gain(days)` (the portfolio's), `gains(days)` (each holding's),
-  `trades(days)`.
+  `Portfolio` objects: `holdings()`, `tickers()`, `gain(days)` (the portfolio's), `trades(days)`.
 - `lib/cache.py`: cache files (any may vanish at any time). `lib/state.py`: state files in `DATA_DIR`, which must last;
   change one with `with state.edit(name, default) as data:`, never load-then-save. `lib/watchlist.py`: the watchlist.
 - `lib/config.py`: `config.env` (finbot.env, fixed at start) and `config.prefs` (preferences.conf, a new object on
diff --git a/README.md b/README.md
index 7c51abcd..a1834a8c 100644
--- a/README.md
+++ b/README.md
@@ -501,7 +501,6 @@ The preferences, read by the bot itself. A change takes effect on a reload. A tr
 | `exclude_portfolios` | | Sharesight portfolio IDs to leave out, space separated |
 | `include_portfolios` | | Sharesight portfolio IDs to report on, space separated (empty: all that are not excluded) |
 | `trades_use_yahoo` | `True` | Use Yahoo for trades; set `False` if Yahoo breaks |
-| `performance_use_sharesight` | `False` | Price moves over a period (`.price 7d`, `price.py 7`) from Sharesight rather than Yahoo: each holding's capital gain, which counts only the time it was held; watchlist stocks are left out, and one stock's moves (`.price BHP 7d`) come without a chart. Off, every figure comes from Yahoo's price history, the same as the charts |
 | `graph` | `True` | Send charts with reports |
 | `hyperlink` | `True` | Hyperlink tickers in reports |
 | `hyperlinkFooter` | `True` | Add a website link footer to stock lookups |
diff --git a/defaults/preferences.conf b/defaults/preferences.conf
index 95f52c02..1f29649d 100644
--- a/defaults/preferences.conf
+++ b/defaults/preferences.conf
@@ -27,7 +27,6 @@ include_portfolios=
 
 # Data sources: set trades_use_yahoo to False if Yahoo breaks
 trades_use_yahoo=True
-performance_use_sharesight=False
 
 # Display
 graph=True
diff --git a/lib/config.py b/lib/config.py
index 90569936..ada38520 100644
--- a/lib/config.py
+++ b/lib/config.py
@@ -23,8 +23,8 @@
 	'discord_webhook', 'slackBotToken', 'slackOutgoingToken', 'slackOutgoingWebhook', 'slackSigningSecret',
 	'telegramBotToken', 'telegramOutgoingToken', 'telegramOutgoingWebhook') # never printed
 PREFERENCES_FILE = 'preferences.conf'
-RETIRED = ('alliterate', 'chunk_maxlines') # preferences.conf keys that do nothing any more: accepted, with a note at
-# start to delete them
+RETIRED = ('alliterate', 'chunk_maxlines', 'performance_use_sharesight') # preferences.conf keys that do nothing any
+# more: accepted, with a note at start to delete them
 # set by the bot for its job processes: the preferences arrive in the environment, do not read the file
 PASSED = 'FINBOT_PREFERENCES_PASSED'
 OLD_ENV = REPO_DIR / '.env' # the single config file of old versions, which python-dotenv loaded; nothing reads it now
@@ -269,7 +269,6 @@ class Prefs:
 	min_analysts: int = 7 # .buy, .sell and .forward pe list only stocks this many analysts cover
 	past_days: int = 28
 	performance_cache_seconds: int = 1800 # Sharesight performance reports (30 minutes)
-	performance_use_sharesight: bool = False
 	price_percent: float = 5.0
 	print_headers: bool = False
 	shorts_percent: int = 15
diff --git a/lib/reports/price.py b/lib/reports/price.py
index 44b3f97c..3a904e2d 100644
--- a/lib/reports/price.py
+++ b/lib/reports/price.py
@@ -5,7 +5,6 @@
 import sys
 
 from lib import config
-from lib import sharesight
 from lib import delivery
 from lib import formats
 from lib import links
@@ -19,23 +18,13 @@
 def gather(specific_stock=None, days=None):
 	"""(tickers, market data) for a price report on the stocks followed or one (Yahoo's symbol): Yahoo's quotes, and
 	over `days` each one's change as percent_change_period, from its Yahoo history (the change its chart shows), the
-	histories fetched together up front. With performance_use_sharesight, a holding's capital gain from Sharesight
-	instead, which counts only the time it was held (held for 1Y and asked for 5Y, it gives 1Y), and none for the
-	others."""
+	histories fetched together up front."""
 	tickers = watchlist.covered(specific_stock)
 	market_data = yahoo.fetch(tickers)
 	if config.prefs.demote_volatile: # trigger_bar compares market caps, in US dollars
 		yahoo.add_usd_market_caps(market_data)
 	if not days:
 		return tickers, market_data
-	if config.prefs.performance_use_sharesight:
-		for portfolio in sharesight.portfolios():
-			for ticker, gain in portfolio.gains(days).items():
-				if ticker in market_data:
-					market_data[ticker]['percent_change_period'] = gain
-				elif ticker in tickers:
-					print("Notice: price.py:", ticker, "has no data", file=sys.stderr)
-		return tickers, market_data
 	if not specific_stock:
 		yahoo.prefetch_history(list(market_data))
 	for ticker in list(market_data):
diff --git a/lib/sharesight.py b/lib/sharesight.py
index 9f220767..53b909de 100644
--- a/lib/sharesight.py
+++ b/lib/sharesight.py
@@ -123,13 +123,6 @@ def gain(self, days=None):
 		"""Its capital gain over the past `days` (the past_days preference when None), in percent."""
 		return float(_performance(self.id, days)['capital_gain_percent'])
 
-	def gains(self, days=None):
-		"""{Yahoo ticker: capital gain in percent} of what it holds, over the past `days` (the past_days preference when
-		None), or over the time it was held where that is shorter."""
-		report = _performance(self.id, days)
-		return {markets.transform_to_yahoo(each['instrument']['code'], each['instrument']['market_code']):
-			float(each['capital_gain_percent']) for each in report['holdings']}
-
 	def holdings(self):
 		"""{'CODE:MARKET': Sharesight's instrument, with the holding's id and value added} of what it holds."""
 		return {each['instrument']['code'] + ':' + each['instrument']['market_code']:
diff --git a/lib/worker.py b/lib/worker.py
index 4eb0d736..db14f579 100755
--- a/lib/worker.py
+++ b/lib/worker.py
@@ -8,7 +8,6 @@
 import sys
 import typing
 from lib import answer
-from lib import config
 from lib import dialects
 from lib import state
 from lib import formats
@@ -249,9 +248,9 @@ def on_session(request, args):
 def on_price(request, args):
 	""".price [percent|SYMBOL|top|bottom] [period]: a bare number is a percent, as on the other commands; a period has
 	its unit. One stock over a period is its chart, as its period buttons draw it, from the history that the lists
-	take their figures from too (not with performance_use_sharesight: Sharesight has no history to draw)."""
+	take their figures from too."""
 	symbol = None if args.keyword else ticker(args)
-	if symbol and args.days and not config.prefs.performance_use_sharesight:
+	if symbol and args.days:
 		return chart_reply(request, 'p', [symbol], args.days)
 	top = 10 if args.keyword else None # top or bottom: one chart shows both ends
 	price.run(request.dest, mode='period' if args.days else 'interday', threshold=0 if top else args.percent,
diff --git a/tests/common.py b/tests/common.py
index e9ef0c0e..d11f9e61 100644
--- a/tests/common.py
+++ b/tests/common.py
@@ -117,7 +117,7 @@ def setUp(self):
 		self.patch(watchlist, 'with_holdings', lambda: TICKERS)
 		self.patch(yahoo, 'fetch', lambda ts: {t: market_entry(t) for t in ts})
 		self.patch(yahoo, 'fetch_detail', lambda t, *a, **k: {t: market_entry(t)})
-		moved = {t: float(i * 4 - 40) for i, t in enumerate(TICKERS)} # over any period: the holdings' gains below
+		moved = {t: float(i * 4 - 40) for i, t in enumerate(TICKERS)} # each stock's move over any period
 		self.patch(yahoo, 'price_history', lambda t, days=None, quote=None: # a history with no line to draw
 			({days: moved.get(t, 1.5)} if days else {'Max': 5, '1Y': 3}, yahoo.Line('Co ' + t, 'NasdaqGS', 'USD', 100.0,
 			())))
diff --git a/tests/test_config.py b/tests/test_config.py
index 5fa82144..1dd7cfaa 100644
--- a/tests/test_config.py
+++ b/tests/test_config.py
@@ -132,10 +132,10 @@ def test_a_blank_value_means_the_default(self):
 		self.assertEqual(config.preference_values(config._getenv_for({})).exchange_suffix, 'AX') # unset: the default
 
 	def test_a_retired_key_is_accepted_and_said_to_do_nothing(self):
-		"""A key finbot no longer reads (alliterate) does not stop a start, as an unknown key would: it is ignored, and
-		the log at start says to delete it."""
+		"""A key finbot no longer reads (alliterate, say) does not stop a start, as an unknown key would: it is ignored,
+		and the log at start says to delete it."""
 		path = Path(common.temp_dir(self), 'preferences.conf')
-		path.write_text('alliterate=True\nchunk_maxlines=20\nfuture_days=45\n')
+		path.write_text(''.join(f'{key}=True\n' for key in config.RETIRED) + 'future_days=45\n')
 		values = config.read_preferences(path)
 		self.assertEqual(config.preference_values(config._getenv_for(values)).future_days, 45)
 		notes = [f'{key} in preferences.conf does nothing any more: delete the line' for key in config.RETIRED]
diff --git a/tests/test_reports.py b/tests/test_reports.py
index 78fa14d1..d6ca0cd8 100644
--- a/tests/test_reports.py
+++ b/tests/test_reports.py
@@ -16,9 +16,8 @@ def test_text_commands(self):
 				self.assertTrue(self.command(cmd), f'{cmd} sent nothing')
 
 	def test_a_period_list_and_one_stocks_view_give_the_same_figure(self):
-		"""Over a period, the list and one stock's view take a holding's figure from the same place: Yahoo's history,
-		which its chart draws (-36% for T1), or with performance_use_sharesight Sharesight's gain over the time it was
-		held (+33%), and then no chart, which Sharesight cannot draw. The list used to take Sharesight's either way."""
+		"""Over a period, the list and one stock's view take a holding's figure from its Yahoo history, which its chart
+		draws, not Sharesight's gain over the time it was held (+33% here), which the list used to take."""
 		holding = {'id': 1, 'instrument': {'code': 'T1', 'market_code': 'NASDAQ'}, 'capital_gain_percent': '33'}
 		self.patch(sharesight, '_performance', lambda pid, days=None, seconds=None:
 			{'capital_gain_percent': '0', 'holdings': [holding]})
@@ -27,10 +26,6 @@ def t1(text):
 			return re.search(r'Co T1 \(.*?\) ([+-]\d+%)', str(self.command(text))).group(1)
 		self.prefs(graph=False) # the list as text: a chart's caption is recorded by its heading
 		self.assertEqual((t1('.price 7d'), t1('.price T1 7d')), ('-36%', '-36%'))
-		self.prefs(performance_use_sharesight=True)
-		self.assertEqual((t1('.price 7d'), t1('.price T1 7d')), ('+33%', '+33%'))
-		self.prefs(graph=True)
-		self.assertEqual(self.command('.price T1 7d')[0][0], 'text') # no chart: Sharesight has no history
 
 	def test_chart_commands(self):
 		"""Every chart command answers with a chart."""
diff --git a/tests/test_sharesight.py b/tests/test_sharesight.py
index 966834e5..241d9c78 100644
--- a/tests/test_sharesight.py
+++ b/tests/test_sharesight.py
@@ -81,8 +81,8 @@ def test_an_answer_without_what_was_asked_for_is_an_error(self):
 					sharesight.portfolios()
 
 	def test_performance_and_holdings_from_real_response(self):
-		"""A real holdings report parses: its tickers come out sorted, US ones without a suffix, and the gains are the
-		portfolio's and each holding's."""
+		"""A real holdings report parses: its tickers come out sorted, US ones without a suffix, and its gain is the
+		portfolio's."""
 		sent = sample('sharesight', 'finbot_sharesight_performance_684141_0.json')
 		self.http(sharesight, Resp(sent))
 		self.assertEqual(len(sharesight._performance(684141, 0)['holdings']), 18)
@@ -91,8 +91,6 @@ def test_performance_and_holdings_from_real_response(self):
 		self.assertIn('ARM', tickers) # US venues carry no Yahoo suffix
 		self.assertEqual(tickers, sorted(tickers))
 		self.assertEqual(portfolio.gain(0), float(sent['report']['capital_gain_percent']))
-		self.assertEqual(sorted(portfolio.gains(0)), tickers)
-		self.assertTrue(all(isinstance(gain, float) for gain in portfolio.gains(0).values()))
 
 	def test_performance_retries_when_sharesight_is_busy(self):
 		"""While Sharesight says too many requests are in flight, the request is tried again."""

From 8bc4fe42e4d5f028713a6c87ba6d65a674c41a3f Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 12:40:42 +0000
Subject: [PATCH 80/81] sharesight_warm.py: Sharesight's caches refreshed on a
 fixed schedule, like Yahoo's

The bot kept Sharesight's portfolio list and holdings reports warm from a
thread of its own, which refreshed each cache at 90% of its lifetime and
again at once after a flush. It is now a scheduled job like yahoo_warm.py,
on a fixed interval inside a cache lifetime made a little longer:
- sharesight_warm.py fetches the portfolio list and each portfolio's
  holdings report afresh; the default schedule runs it every 30 minutes
  (at :14 and :44, clear of trades.py);
- performance_cache_seconds defaults to 40 minutes (was 30), so the holdings
  never run out between runs;
- without Sharesight configured, or with the cache off, it fetches nothing;
  a failure is logged, not reported in the chat: the caches it would have
  replaced are served while fresh, and the next run tries again;
- .jobs lists it under System.

The thread goes, with sharesight.keep_warm and wake: after a flush, the
warm jobs fill the cache again on their next runs (a request meanwhile
fetches what it needs). sharesight.warm() now simply fetches both afresh.

A data directory with its own schedule.cron needs the line
`14,44 * * * * sharesight_warm.py`; a preferences.conf that sets
performance_cache_seconds=1800 should raise it, or drop the line.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 CLAUDE.md                      |   6 +-
 README.md                      |  16 +++--
 bot.py                         |   7 +-
 defaults/preferences.conf      |   2 +-
 defaults/schedule.cron         |   6 +-
 lib/config.py                  |   2 +-
 lib/jobs.py                    |   7 +-
 lib/reports/__init__.py        |  10 +--
 lib/reports/sharesight_warm.py |  22 +++++++
 lib/reports/yahoo_warm.py      |   2 +-
 lib/sharesight.py              |  39 +++--------
 sharesight_warm.py             |   5 ++
 tests/test_schedule.py         |  10 +--
 tests/test_scheduler.py        |  28 +++++++-
 tests/test_sharesight.py       | 115 +++------------------------------
 15 files changed, 107 insertions(+), 170 deletions(-)
 create mode 100644 lib/reports/sharesight_warm.py
 create mode 100755 sharesight_warm.py

diff --git a/CLAUDE.md b/CLAUDE.md
index b608385e..597836fa 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -48,9 +48,9 @@ An open PR takes in every commit pushed to its branch, so a PR opened for one sm
   the typing notice while it works. `process_request(message)` and `process_callback(press)` take inbound's objects.
 - `lib/reports/`: what finbot says. Scheduled (each has `run(dest=None, **kwargs)`, which posts to the chat that asked
   or else the default chats, and `job(argv)`, its schedule-line parser): `price performance trades shorts cal rating
-  milestone reminder podcasts cleanup yahoo_warm`; the top-level `price.py` and the rest are shims that run them.
-  Chat-only (`prepare_*`): `about` (.help .jobs), `history` (.history .compare), `holdings` (.holdings .who .pie
-  .watchlist), `profile` (.SYMBOL), `rankings` (.marketcap .pe .fpe .peg .beta .buy .sell).
+  milestone reminder podcasts cleanup sharesight_warm yahoo_warm`; the top-level `price.py` and the rest are shims that
+  run them. Chat-only (`prepare_*`): `about` (.help .jobs), `history` (.history .compare), `holdings` (.holdings .who
+  .pie .watchlist), `profile` (.SYMBOL), `rankings` (.marketcap .pe .fpe .peg .beta .buy .sell).
 - `lib/delivery.py`: `Report(lines, chart, buttons)`, the destinations (the chat that asked, or the default chats),
   `deliver(dest, report)`, the one-line error report. `lib/transports.py`: a class per way of posting (`TelegramBot`,
   `SlackBot`, `SlackWebhook`, `DiscordWebhook`): splitting, captions, charts, buttons, edits, and the typing notice
diff --git a/README.md b/README.md
index a1834a8c..bbbe9f79 100644
--- a/README.md
+++ b/README.md
@@ -510,7 +510,7 @@ The preferences, read by the bot itself. A change takes effect on a reload. A tr
 | `timezone` | `Australia/Melbourne` | Regional: your time zone, for the reminder report |
 | `cache` | `True` | Cache Yahoo and Sharesight responses |
 | `cache_seconds` | `43200` | Cache lifetime in seconds (12 hours) |
-| `performance_cache_seconds` | `1800` | Cache lifetime of Sharesight performance reports in seconds |
+| `performance_cache_seconds` | `2400` | Cache lifetime of Sharesight performance reports in seconds: longer than the 30 minutes between `sharesight_warm.py` runs, so the holdings never run out |
 | `http_timeout` | `18` | Seconds to wait on outbound HTTP requests |
 | `debug` | `False` | Extra diagnostics in the log |
 | `print_headers` | `False` | Log inbound request headers |
@@ -568,10 +568,10 @@ Many objects are cached for 12 hours by default (`cache_seconds`). Cache is cont
 ```
 cache=True
 cache_seconds=43200
-performance_cache_seconds=1800
+performance_cache_seconds=2400
 ```
 
-`performance_cache_seconds` (default 30 minutes) is the cache lifetime of Sharesight performance reports.
+`performance_cache_seconds` (default 40 minutes) is the cache lifetime of Sharesight performance reports.
 
 Charts are cached by what they show (with `cache=True`): the same chart, for another chat, a later run or a period button, is drawn once, and a chart of new figures is a new chart, so a cached one is never stale. The scheduled `cleanup.py` removes cache files after 30 days.
 
@@ -587,7 +587,11 @@ kill -USR1 <pid of bot.py>                  # anything else
 The log says `cache: flushed N files`. To install the `finbot-flush` shortcut: `sudo cp -v finbot-flush.service /etc/systemd/system/ &&
 sudo systemctl daemon-reload`. `finbot.service` itself needs no change. With the bot stopped, `cleanup.py 0` deletes the same files.
 
-Two warmers keep the caches full, so that chat requests do not wait on Yahoo or Sharesight. The scheduled `yahoo_warm.py` fetches the price history of every stock followed afresh every 8 hours (see [Scheduling](#scheduling)); a data directory with its own `schedule.cron` needs its line, `07 */8 * * * yahoo_warm.py`. And while `bot.py` is running with Sharesight configured and `cache=True` (a reload that switches `cache` pauses or resumes it), its `sharesight_warm` thread keeps the Sharesight portfolio list (`cache_seconds`) and each portfolio's holdings report (1 day, `performance_cache_seconds` cache) refreshed at 90% of their cache lifetime, so chat requests do not wait on Sharesight. If a refresh fails, the previous cache keeps being served and the job retries after 5 minutes.
+Two scheduled jobs keep the caches full, each on a fixed interval shorter than the cache lifetime, so that chat requests do not wait on Sharesight or Yahoo (see [Scheduling](#scheduling)): `sharesight_warm.py` fetches the Sharesight portfolio list and each portfolio's holdings report afresh every 30 minutes (`performance_cache_seconds`: 40 minutes), and `yahoo_warm.py` the price history of every stock followed every 8 hours (`cache_seconds`: 12 hours). Neither fetches anything with `cache=False`, and `sharesight_warm.py` nothing without Sharesight configured. A refresh that fails is logged, the previous cache keeps being served while fresh, and the next run tries again. After a flush, the next runs fill the cache again. A data directory with its own `schedule.cron` needs their lines:
+```
+14,44 * * * * sharesight_warm.py
+07 */8 * * * yahoo_warm.py
+```
 
 ### Hyperlinks
 These `preferences.conf` options relate to hyperlinks:
@@ -614,7 +618,7 @@ hyperlinkProvider=yahoo
 ## Scheduling
 The scheduled reports are started by `bot.py`, each run in its own child process (`python -m lib.jobrun`), so they use the same user, sandbox, environment and data directory as the bot, while heavy work (charts, large reports) cannot hold up incoming chat webhooks. A run that exceeds 5 minutes is stopped. There is no cron to set up, which matters under systemd `DynamicUser=` where only the bot's own user can read the state directory. **This is on by default.** Remove the equivalent entries from any old crontab before the first start, otherwise every job runs twice; or set `scheduler=False` in the environment file to keep using cron.
 
-The schedulable scripts are `trades.py`, `price.py`, `reminder.py`, `cal.py`, `shorts.py`, `performance.py`, `rating.py`, `milestone.py`, `podcasts.py "<feed url>"`, `cleanup.py [days]` (deletes cache files older than `days`, default 30) and `yahoo_warm.py` (fetches the price histories of every stock followed afresh, so that period lists and charts never wait on Yahoo; the default schedule runs it every 8 hours, within the 12 hours of `cache_seconds`). Other jobs, such as a DNS updater or backups, stay in cron or systemd timers.
+The schedulable scripts are `trades.py`, `price.py`, `reminder.py`, `cal.py`, `shorts.py`, `performance.py`, `rating.py`, `milestone.py`, `podcasts.py "<feed url>"`, `cleanup.py [days]` (deletes cache files older than `days`, default 30), `sharesight_warm.py` and `yahoo_warm.py` (they fill the cache: see [Caching](#caching)). Other jobs, such as a DNS updater or backups, stay in cron or systemd timers.
 
 The schedule is `defaults/schedule.cron`, written in **cron-compatible syntax**: five time fields (minute, hour, day of month, month, day of week; `*/20`, `Mon-Fri`, ranges and lists work) or an `@macro` such as `@daily`, then a finbot script and its arguments. Quote an argument that has spaces or special characters (a feed URL with `&`, say). Several commands separated by `;` run one after another. `MAILTO=`, `SHELL=`, `PATH=` and `HOME=` lines are ignored. Anything else a crontab allows (a user column, pipes, other programs) is not supported: such a line is reported at startup and skipped. A time that cannot be or never comes (minute 61, 30 February) is different: finbot refuses to start with it, and a reload with it is refused, leaving the running schedule as it was. Existing crontab lines can be pasted as they are: `> /dev/null` redirects, `{ a; b ;}` groups and a path before the script name are understood.
 ```
@@ -637,7 +641,7 @@ sudo systemctl reload finbot        # the bot re-reads the file and logs how man
 ```
 The bot logs the number of jobs when it starts or reloads, and each job as it runs (`journalctl -u finbot`); a line it cannot use is reported by name. If the edited file has no usable line the previous schedule keeps running.
 
-**To run a job from chat**, send `.jobs` (also `.cron`, `.crons` or `.schedule`): it lists the scheduled commands, each once and without times, comments or `.py`, and on Telegram and Slack (with its buttons set up) adds a button per command. The system jobs (`cleanup`, `podcasts` and `yahoo_warm`) follow under _System_, without a number or a button: they run on their schedule only. `.jobs run 2` (or tapping the button) runs that command now, and `.jobs run price.py close` runs it by name; anything not in the schedule is refused, and so is a chain such as `a; b` (run them one at a time). A manual run is exactly a cron run: it reports to all configured chat networks, updates the same state (seen trades, ratings and so on), and does not change the schedule. Anyone who can message the bot can do this; a command that is already running is not started twice.
+**To run a job from chat**, send `.jobs` (also `.cron`, `.crons` or `.schedule`): it lists the scheduled commands, each once and without times, comments or `.py`, and on Telegram and Slack (with its buttons set up) adds a button per command. The system jobs (`cleanup`, `podcasts`, `sharesight_warm` and `yahoo_warm`) follow under _System_, without a number or a button: they run on their schedule only. `.jobs run 2` (or tapping the button) runs that command now, and `.jobs run price.py close` runs it by name; anything not in the schedule is refused, and so is a chain such as `a; b` (run them one at a time). A manual run is exactly a cron run: it reports to all configured chat networks, updates the same state (seen trades, ratings and so on), and does not change the schedule. Anyone who can message the bot can do this; a command that is already running is not started twice.
 
 Behaviour worth knowing:
 * A run that was due while the bot was down (a restart, a deploy) still fires once if it was due within the last `scheduler_catchup` minutes (default 5). A restart right after a run does not repeat it.
diff --git a/bot.py b/bot.py
index 7ebe9283..9241e98c 100755
--- a/bot.py
+++ b/bot.py
@@ -16,7 +16,6 @@
 from lib import inbound
 from lib import worker
 from lib import jobs
-from lib import sharesight
 from lib import net
 from lib import telegram
 from lib import delivery
@@ -178,10 +177,9 @@ def reload_all(*args):
 
 def flush_cache(*args):
 	"""SIGUSR1 (`systemctl kill -s USR1 finbot`): delete the cache and keep the state (watchlist, seen trades, ratings,
-	schedule). The Sharesight caches refill at once."""
+	schedule). The warm jobs fill it again on their next runs, and any request meanwhile fetches what it needs."""
 	try:
 		removed = cleanup.run(days=0)
-		sharesight.wake()
 		print(f"cache: flushed {removed} file{'' if removed == 1 else 's'} from {config.env.cache_dir}",
 			file=sys.stderr)
 	except Exception as e:
@@ -224,9 +222,6 @@ def flush_cache(*args):
 			sys.exit(1)
 	if config.env.telegramBotToken:
 		threading.Thread(target=worker.refresh_keyboards, daemon=True).start() # stale DM keyboards, in the background
-	if config.env.sharesight_auth.get('client_id'):
-		# keep the portfolio list and holdings report cached, while cache is on
-		threading.Thread(target=sharesight.keep_warm, daemon=True, name='sharesight_warm').start()
 	if scheduler:
 		jobs.current = scheduler # for the .jobs chat command
 		threading.Thread(target=scheduler.run_forever, daemon=True).start() # the scheduled reports, formerly cron jobs
diff --git a/defaults/preferences.conf b/defaults/preferences.conf
index 1f29649d..b8b69bed 100644
--- a/defaults/preferences.conf
+++ b/defaults/preferences.conf
@@ -45,7 +45,7 @@ timezone=Australia/Melbourne
 # Caching and network, in seconds
 cache=True
 cache_seconds=43200
-performance_cache_seconds=1800
+performance_cache_seconds=2400
 http_timeout=18
 
 # Diagnostics
diff --git a/defaults/schedule.cron b/defaults/schedule.cron
index 36fcfd97..5020d035 100644
--- a/defaults/schedule.cron
+++ b/defaults/schedule.cron
@@ -22,8 +22,10 @@
 # Weekly
 28  21 * * Sat cal.py earnings; cal.py ex-dividend; price.py 7; performance.py 7; rating.py
 
-# Housekeeping: fetch the price histories of the stocks followed afresh every 8 hours (they are cached 12 hours, so
-# period lists and charts never wait on Yahoo); delete cache files older than 30 days
+# Housekeeping: fetch afresh the Sharesight portfolio list and holdings every 30 minutes (cached 40 minutes) and the
+# price histories of the stocks followed every 8 hours (cached 12 hours), so requests never wait on Sharesight or
+# Yahoo; delete cache files older than 30 days
+14,44 * * * * sharesight_warm.py
 07 */8 * * * yahoo_warm.py
 51  18 * * Sat cleanup.py 30
 
diff --git a/lib/config.py b/lib/config.py
index ada38520..7a22a430 100644
--- a/lib/config.py
+++ b/lib/config.py
@@ -268,7 +268,7 @@ class Prefs:
 	include_portfolios: tuple = () # portfolio ids
 	min_analysts: int = 7 # .buy, .sell and .forward pe list only stocks this many analysts cover
 	past_days: int = 28
-	performance_cache_seconds: int = 1800 # Sharesight performance reports (30 minutes)
+	performance_cache_seconds: int = 2400 # Sharesight performance reports (40 minutes: sharesight_warm runs every 30)
 	price_percent: float = 5.0
 	print_headers: bool = False
 	shorts_percent: int = 15
diff --git a/lib/jobs.py b/lib/jobs.py
index 604a608f..ed3fa0d5 100644
--- a/lib/jobs.py
+++ b/lib/jobs.py
@@ -27,13 +27,14 @@
 from lib import jobrun
 from lib import delivery
 from lib.reports import cal, cleanup, milestone, performance, podcasts, price, rating, reminder, shorts, trades
-from lib.reports import yahoo_warm
+from lib.reports import sharesight_warm, yahoo_warm
 
 SCRIPTS = {'cal': cal, 'cleanup': cleanup, 'milestone': milestone, 'performance': performance, 'podcasts': podcasts,
 	'price': price, 'rating': rating, 'reminder': reminder, 'shorts': shorts, 'trades': trades,
-	'yahoo_warm': yahoo_warm}
+	'sharesight_warm': sharesight_warm, 'yahoo_warm': yahoo_warm}
 current = None # the running Scheduler, set by bot.py; the chat commands use it
-HIDDEN = ('cleanup', 'podcasts', 'yahoo_warm') # scheduled, but not run from chat: .jobs lists them under System
+# scheduled, but not run from chat: .jobs lists them under System
+HIDDEN = ('cleanup', 'podcasts', 'sharesight_warm', 'yahoo_warm')
 STATE_FILE = 'finbot_jobs.json' # last fire time of every job, so a restart neither repeats a run nor loses one
 JOB_TIMEOUT = 5 * 60 # seconds each command may run before it is stopped and reported
 IGNORED_VARIABLES = ('MAILTO', 'SHELL', 'PATH', 'HOME', 'LOGNAME')
diff --git a/lib/reports/__init__.py b/lib/reports/__init__.py
index 5a93b6f8..c62a590f 100644
--- a/lib/reports/__init__.py
+++ b/lib/reports/__init__.py
@@ -1,9 +1,9 @@
 """What finbot says. A module per scheduled report: cal, cleanup (of the cache), milestone, performance, podcasts,
-price, rating, reminder, shorts, trades and yahoo_warm (which fills the cache), each with job(argv), which parses its
-schedule line, and run(dest=None, **kwargs), which runs it, to the chat that asked (dest) or else the default chats;
-the top-level script of the same name runs it from the command line. And the chat-only reports: about (.help, .jobs),
-history (.history, .compare), holdings (.holdings, .who, .pie, .watchlist), profile (.SYMBOL) and rankings (.marketcap,
-.pe and the rest)."""
+price, rating, reminder, shorts, trades, and sharesight_warm and yahoo_warm (which fill the cache), each with
+job(argv), which parses its schedule line, and run(dest=None, **kwargs), which runs it, to the chat that asked (dest)
+or else the default chats; the top-level script of the same name runs it from the command line. And the chat-only
+reports: about (.help, .jobs), history (.history, .compare), holdings (.holdings, .who, .pie, .watchlist), profile
+(.SYMBOL) and rankings (.marketcap, .pe and the rest)."""
 import os
 import sys
 
diff --git a/lib/reports/sharesight_warm.py b/lib/reports/sharesight_warm.py
new file mode 100644
index 00000000..666dae07
--- /dev/null
+++ b/lib/reports/sharesight_warm.py
@@ -0,0 +1,22 @@
+"""Keeps Sharesight's portfolio list and holdings reports fresh in the cache (scheduled: sharesight_warm.py), so that a
+request never waits on Sharesight. The price histories have their own, yahoo_warm.py."""
+import sys
+from lib import config, sharesight
+
+def run():
+	"""Fetch afresh the portfolio list and each portfolio's holdings report, which every report starts from: the
+	holdings report is cached for performance_cache_seconds (40 minutes by default), and the default schedule runs
+	this every 30 minutes. Nothing without Sharesight configured, or with the cache preference off (nothing would be
+	kept). A failure is only logged: the caches it would have replaced are served while fresh, and the next run tries
+	again. Returns how many portfolios it refreshed."""
+	if not (config.prefs.cache and config.env.sharesight_auth.get('client_id')):
+		return 0
+	try:
+		return len(sharesight.warm())
+	except Exception as e:
+		print("sharesight_warm: refresh failed:", e, file=sys.stderr)
+		return 0
+
+def job(argv):
+	"""Arguments of a scheduled run: sharesight_warm.py takes none."""
+	return 'sharesight_warm.py', {}
diff --git a/lib/reports/yahoo_warm.py b/lib/reports/yahoo_warm.py
index f8f3e259..9a75d7d6 100644
--- a/lib/reports/yahoo_warm.py
+++ b/lib/reports/yahoo_warm.py
@@ -1,5 +1,5 @@
 """Keeps the price history of every stock followed fresh in the cache (scheduled: yahoo_warm.py), so that a period list
-or a chart never waits on Yahoo. Sharesight's caches have their own warmer, the bot's sharesight_warm thread."""
+or a chart never waits on Yahoo. Sharesight's caches have their own, sharesight_warm.py."""
 import sys
 from lib import config, watchlist, yahoo
 
diff --git a/lib/sharesight.py b/lib/sharesight.py
index 53b909de..79960a6f 100644
--- a/lib/sharesight.py
+++ b/lib/sharesight.py
@@ -161,35 +161,12 @@ def tickers():
 	return sorted({ticker for portfolio in portfolios() for ticker in portfolio.tickers()})
 
 
-def warm(margin=0.9):
-	"""Refresh the portfolio list and each portfolio's holdings report (days=0) once they reach `margin` of their cache
-	lifetime, so that a request never waits on Sharesight. Returns the seconds until the next one is due."""
-	lifetimes = {PORTFOLIOS_CACHE: config.prefs.cache_seconds}
-	for portfolio in portfolios(seconds=margin * config.prefs.cache_seconds):
-		_performance(portfolio.id, 0, margin * config.prefs.performance_cache_seconds)
-		lifetimes[performance_cache_file(portfolio.id, 0)] = config.prefs.performance_cache_seconds
-	return max(0, min(margin * lifetime - (cache.age(name) or 0) for name, lifetime in lifetimes.items()))
+def warm():
+	"""Fetch afresh the portfolio list and each portfolio's holdings report (days=0), so that a request never waits on
+	Sharesight (scheduled: sharesight_warm.py). Returns the portfolios. Raises as the requests do, and then the caches
+	it would have replaced stay."""
+	found = portfolios(seconds=0)
+	for portfolio in found:
+		_performance(portfolio.id, 0, seconds=0)
+	return found
 
-
-_wake = threading.Event() # set by wake(): the caches were flushed, refill them now
-
-
-def wake():
-	"""Have keep_warm refill the caches now instead of when they would next be due."""
-	_wake.set()
-
-
-def keep_warm(min_wait=5, retry=300, idle=60):
-	"""Background loop: keep the portfolio list and the holdings reports cached, sleeping until the next one is due
-	(warm). While the cache preference is off (a reload can switch it either way) it fetches nothing, since nothing
-	would be written, and looks again every `idle` seconds. Never raises; after a failure it waits `retry` seconds."""
-	while True:
-		wait = idle
-		if config.prefs.cache:
-			try:
-				wait = max(min_wait, warm())
-			except Exception as e:
-				print("Sharesight cache refresh failed:", e, file=sys.stderr)
-				wait = retry
-		_wake.wait(wait)
-		_wake.clear()
diff --git a/sharesight_warm.py b/sharesight_warm.py
new file mode 100755
index 00000000..b04e2850
--- /dev/null
+++ b/sharesight_warm.py
@@ -0,0 +1,5 @@
+#!/usr/bin/env python3
+"""./sharesight_warm.py [arguments]: runs lib/reports/sharesight_warm.py once, as its schedule line would."""
+if __name__ == '__main__':
+	from lib import reports
+	reports.cli(__file__)
diff --git a/tests/test_schedule.py b/tests/test_schedule.py
index 34a9585f..884bdc33 100644
--- a/tests/test_schedule.py
+++ b/tests/test_schedule.py
@@ -9,12 +9,14 @@
 
 class ParseTests(unittest.TestCase):
 	def test_shipped_schedule_has_no_errors(self):
-		"""defaults/schedule.cron parses without errors, its Saturday line into its five commands; the price histories
-		are fetched afresh every 8 hours, within the 12 hours they are cached."""
+		"""defaults/schedule.cron parses without errors, its Saturday line into its five commands; the warm jobs run
+		within their caches' lifetimes: Sharesight's every 30 minutes, Yahoo's every 8 hours."""
 		parsed, errors = jobs.parse_schedule(DEFAULT_SCHEDULE.read_text(), UTC)
 		self.assertEqual(errors, [])
-		self.assertEqual(len(parsed), 10)
-		self.assertIn(('07 */8 * * *', ['yahoo_warm.py']), [(j.expr, [c[0] for c in j.commands]) for j in parsed])
+		self.assertEqual(len(parsed), 11)
+		lines = [(j.expr, [c[0] for c in j.commands]) for j in parsed]
+		self.assertIn(('14,44 * * * *', ['sharesight_warm.py']), lines)
+		self.assertIn(('07 */8 * * *', ['yahoo_warm.py']), lines)
 		weekly = next(j for j in parsed if j.expr == '28 21 * * Sat')
 		self.assertEqual([label for label, _, _ in weekly.commands],
 						 ['cal.py earnings', 'cal.py ex-dividend', 'price.py 7', 'performance.py 7', 'rating.py'])
diff --git a/tests/test_scheduler.py b/tests/test_scheduler.py
index 45e54090..ddbaab88 100644
--- a/tests/test_scheduler.py
+++ b/tests/test_scheduler.py
@@ -7,7 +7,7 @@
 import gevent
 from tests.common import *
 from lib import jobs
-from lib.reports import cleanup, podcasts, price, yahoo_warm
+from lib.reports import cleanup, podcasts, price, sharesight_warm, yahoo_warm
 
 PREFETCH_HISTORY = yahoo.prefetch_history # FinbotCase replaces it; the warm job's test puts it back
 
@@ -109,6 +109,32 @@ def test_with_the_cache_off_nothing_is_fetched(self):
 		self.assertEqual(yahoo_warm.run(), 0)
 
 
+class SharesightWarmJob(FinbotCase):
+	def test_the_portfolios_are_refreshed_where_sharesight_is_configured(self):
+		"""sharesight_warm refreshes the portfolio list and the holdings (sharesight.warm) and counts the portfolios;
+		without Sharesight configured, or with the cache preference off (nothing would be kept), it fetches nothing."""
+		self.patch(sharesight, 'warm', lambda: ['Rob', 'RobSMSF'])
+		self.env(sharesight_auth={'client_id': 'id'})
+		self.assertEqual(sharesight_warm.run(), 2)
+		self.prefs(cache=False)
+		self.assertEqual(sharesight_warm.run(), 0)
+		self.prefs(cache=True)
+		self.env(sharesight_auth={'client_id': None})
+		self.assertEqual(sharesight_warm.run(), 0)
+
+	def test_a_failure_is_logged_not_reported(self):
+		"""A refresh that fails is only logged, not reported in the chat: the next run tries again."""
+		def down():
+			"""Sharesight unreachable, as warm meets it."""
+			raise RuntimeError('Sharesight portfolios: unreachable')
+		self.patch(sharesight, 'warm', down)
+		self.env(sharesight_auth={'client_id': 'id'})
+		with contextlib.redirect_stderr(io.StringIO()) as log:
+			self.assertEqual(sharesight_warm.run(), 0)
+		self.assertIn('sharesight_warm: refresh failed: Sharesight portfolios: unreachable', log.getvalue())
+		self.assertEqual(self.errors, [])
+
+
 class SchedulerFlag(unittest.TestCase):
 	def flag(self, **env):
 		"""config.env.scheduler in a fresh interpreter, given `env`."""
diff --git a/tests/test_sharesight.py b/tests/test_sharesight.py
index 241d9c78..1d094a1a 100644
--- a/tests/test_sharesight.py
+++ b/tests/test_sharesight.py
@@ -1,7 +1,6 @@
 """Sharesight's portfolios, holdings, gains and trades on the answers saved under doc/, its failures,
 and the cache warmer."""
-import datetime, os, unittest
-import gevent
+import datetime, unittest
 from tests.common import *
 
 
@@ -117,8 +116,9 @@ def test_an_answer_without_a_report_says_so(self):
 		self.assertIsNone(cache.read(sharesight.performance_cache_file(684141, 7))) # and is not cached
 
 
-class SharesightKeepWarm(ProviderCase):
-	"""The background refresher keeps the portfolio list and the holdings report (days=0) cached ahead of their TTLs."""
+class SharesightWarm(ProviderCase):
+	"""warm() fetches the portfolio list and the holdings reports (days=0) afresh, for sharesight_warm.py to run on its
+	schedule."""
 
 	def serve_sharesight(self):
 		"""Answer with the real portfolio list and holdings report; returns the URLs asked for."""
@@ -132,38 +132,16 @@ def get(url, **kwargs):
 		self.patch(sharesight.net, 'get', get)
 		return calls
 
-	def age(self, cache_file, seconds):
-		"""Make a cache file `seconds` older."""
-		path = os.path.join(config.env.cache_dir, cache_file)
-		old = os.path.getmtime(path) - seconds
-		os.utime(path, (old, old))
-
-	def test_cold_cache_is_filled_then_left_alone(self):
-		"""A cold cache is filled (the list and each portfolio's holdings, as of today), then left alone while fresh."""
+	def test_the_list_and_each_holdings_report_are_fetched_afresh(self):
+		"""warm fetches the list and each portfolio's holdings as of today, afresh even while their caches are fresh."""
 		calls = self.serve_sharesight()
-		sharesight.warm()
+		self.assertEqual([portfolio.name for portfolio in sharesight.warm()], ['Rob', 'RobSMSF'])
 		self.assertEqual(len(calls), 3) # portfolio list + one holdings report per portfolio (2)
 		self.assertEqual(sum('/performance' in c for c in calls), 2)
 		today = 'start_date=' + datetime.date.today().isoformat()
 		self.assertTrue(all(today in c for c in calls if '/performance' in c)) # days=0
 		sharesight.warm()
-		self.assertEqual(len(calls), 3) # nothing refetched while fresh
-
-	def test_refreshes_before_the_ttl_runs_out(self):
-		"""A cache file at 95% of its lifetime is fetched again, and only that one."""
-		calls = self.serve_sharesight()
-		sharesight.warm()
-		calls.clear()
-		# 95% of its lifetime
-		self.age(sharesight.performance_cache_file(684141, 0), 0.95 * config.prefs.performance_cache_seconds)
-		sharesight.warm()
-		self.assertEqual(len(calls), 1)
-		self.assertIn('684141/performance', calls[0])
-		calls.clear()
-		self.age(sharesight.PORTFOLIOS_CACHE, 0.95 * config.prefs.cache_seconds)
-		sharesight.warm()
-		self.assertEqual(len(calls), 1)
-		self.assertTrue(calls[0].endswith('/portfolios'))
+		self.assertEqual(len(calls), 6)
 
 	def test_a_request_never_waits_on_a_cache_the_job_has_warmed(self):
 		"""Once warmed, the list and the holdings report need no request."""
@@ -184,89 +162,14 @@ def test_every_holding_is_served_from_the_warmed_cache(self):
 		self.assertEqual(calls, [])
 
 	def test_failed_refresh_keeps_the_old_cache(self):
-		"""A refresh that fails keeps the cache it would have replaced."""
+		"""A refresh that fails raises, and keeps the cache it would have replaced."""
 		self.serve_sharesight()
 		sharesight.warm()
-		self.age(sharesight.PORTFOLIOS_CACHE, 0.95 * config.prefs.cache_seconds)
 		self.http(sharesight, ConnectionError('down'))
 		with self.assertRaises(RuntimeError):
 			sharesight.warm()
 		self.assertEqual(portfolios(), {'Rob': 445825, 'RobSMSF': 684141}) # still served from the cache
 
-	def test_a_pass_says_when_the_next_is_due(self):
-		"""A pass reads each cache file once and gives the seconds until the next refresh: 90% of the shortest cache
-		lifetime once everything is fresh."""
-		self.serve_sharesight()
-		due = sharesight.warm() # a cold cache, filled
-		shortest = min(config.prefs.cache_seconds, config.prefs.performance_cache_seconds)
-		self.assertAlmostEqual(due, 0.9 * shortest, delta=5)
-		reads = []
-		real = cache.read
-		self.patch(cache, 'read', lambda name, *a, **k: (reads.append(name), real(name, *a, **k))[1])
-		self.age(sharesight.performance_cache_file(684141, 0), 600)
-		due = sharesight.warm()
-		self.assertAlmostEqual(due, 0.9 * config.prefs.performance_cache_seconds - 600, delta=5)
-		files = [sharesight.PORTFOLIOS_CACHE] + [sharesight.performance_cache_file(id, 0) for id in (445825, 684141)]
-		self.assertEqual(sorted(reads), sorted(files))
-
-	class Stopped(Exception):
-		"""Raised by the stand-in for keep_warm's wake event, to end the loop."""
-
-	def waits(self, count, on_wait=lambda n: None, **kwargs):
-		"""Run keep_warm until it has waited `count` times, noting each wait instead of sleeping (on_wait(n) at the
-		nth); returns the waits."""
-		waits, stopped = [], self.Stopped
-		class Wake:
-			"""keep_warm's wake event, which nobody sets."""
-			def wait(self, seconds):
-				"""Note the wait instead of sleeping, and end the loop after the last."""
-				waits.append(seconds)
-				on_wait(len(waits))
-				if len(waits) == count:
-					raise stopped
-			def clear(self):
-				"""Nothing to clear."""
-		self.patch(sharesight, '_wake', Wake())
-		with self.assertRaises(self.Stopped):
-			sharesight.keep_warm(**kwargs)
-		return waits
-
-	def test_loop_survives_failures_and_backs_off(self):
-		"""keep_warm survives a failure, retries slowly, then sleeps until the next cache is due."""
-		outcomes = [RuntimeError('boom'), 3000]
-		def warm():
-			"""Fail once, then succeed."""
-			outcome = outcomes.pop(0)
-			if isinstance(outcome, Exception):
-				raise outcome
-			return outcome
-		self.patch(sharesight, 'warm', warm)
-		self.assertEqual(self.waits(2, retry=300), [300, 3000]) # a slow retry after the failure, then until one is due
-
-	def test_with_cache_off_nothing_is_fetched_until_a_reload_turns_it_on(self):
-		"""With cache off, keep_warm idles without asking Sharesight, and starts when a reload turns it on."""
-		passes = []
-		self.patch(sharesight, 'warm', lambda: passes.append(1) or 3000)
-		self.prefs(cache=False)
-		def reload(n):
-			"""At the second wait, switch the cache on, as a reload would."""
-			if n == 2:
-				config.prefs = dataclasses.replace(config.prefs, cache=True)
-		self.assertEqual(self.waits(4, reload, idle=60), [60, 60, 3000, 3000]) # no request every 5 seconds while off
-		self.assertEqual(len(passes), 2)
-
-	def test_a_flush_wakes_the_loop_to_refill_at_once(self):
-		"""A flush wakes keep_warm, which refills the caches at once instead of when they are next due."""
-		passes = []
-		self.patch(sharesight, 'warm', lambda: passes.append(1) or 3000)
-		loop = gevent.spawn(sharesight.keep_warm)
-		self.addCleanup(loop.kill)
-		gevent.sleep(0.05)
-		self.assertEqual(len(passes), 1) # then it sleeps until the next cache is due, 50 minutes away
-		sharesight.wake()
-		gevent.sleep(0.05)
-		self.assertEqual(len(passes), 2)
-
 
 if __name__ == '__main__':
 	unittest.main()

From d4fbd25a66cc870baa49a7bca250365d51751f21 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Thu, 15 Oct 2026 12:53:08 +0000
Subject: [PATCH 81/81] A flush runs both warm jobs straight away

After a flush, the cache stayed empty until each warm job's next run (up to
30 minutes for Sharesight, 8 hours for Yahoo), so requests meanwhile waited
on the sources. Now the flush, once the files are deleted, runs
sharesight_warm.py and yahoo_warm.py at once, one after the other in a
thread of their own (a reload does not wait for them): through the
scheduler, which does not start one already running on its schedule, or
with the scheduler off each in a child process. The log says when they
start and when they are done.

yahoo_warm.py asks Sharesight for the holdings only where Sharesight is
configured, as sharesight_warm.py does: without it, the watchlist's
histories are fetched and no request goes to Sharesight.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JVEkZofdPjQ4jtXFKipeUX
---
 README.md                      |  8 +++----
 bot.py                         |  4 +++-
 lib/jobs.py                    | 18 +++++++++++++++
 lib/reports/sharesight_warm.py |  8 +++----
 lib/reports/yahoo_warm.py      | 18 ++++++++-------
 tests/test_edges.py            | 16 ++++++++++++-
 tests/test_scheduler.py        | 41 ++++++++++++++++++++++++++++++++--
 7 files changed, 93 insertions(+), 20 deletions(-)

diff --git a/README.md b/README.md
index bbbe9f79..1a33be11 100644
--- a/README.md
+++ b/README.md
@@ -576,18 +576,18 @@ performance_cache_seconds=2400
 Charts are cached by what they show (with `cache=True`): the same chart, for another chat, a later run or a period button, is drawn once, and a chart of new figures is a new chart, so a cached one is never stale. The scheduled `cleanup.py` removes cache files after 30 days.
 
 To flush the cache, so that everything is fetched afresh, signal the running bot: it deletes its cache files (Yahoo and Sharesight
-answers, charts, the Yahoo cookie) and keeps its state (watchlist, seen trades, ratings, schedule), then refills the Sharesight caches
-at once. There is no `systemctl flush`, since systemctl takes no custom verbs; any of these does it:
+answers, charts, the Yahoo cookie) and keeps its state (watchlist, seen trades, ratings, schedule), then runs both warm jobs
+(`sharesight_warm.py`, `yahoo_warm.py`) straight away to fill the cache again. There is no `systemctl flush`, since systemctl takes no custom verbs; any of these does it:
 ```
 sudo systemctl kill -s USR1 finbot          # systemd
 sudo systemctl start finbot-flush           # the same, once finbot-flush.service is installed (see below)
 docker kill -s USR1 <container>             # Docker
 kill -USR1 <pid of bot.py>                  # anything else
 ```
-The log says `cache: flushed N files`. To install the `finbot-flush` shortcut: `sudo cp -v finbot-flush.service /etc/systemd/system/ &&
+The log says `cache: flushed N files`, then `cache: warm jobs done` once the cache is full again. To install the `finbot-flush` shortcut: `sudo cp -v finbot-flush.service /etc/systemd/system/ &&
 sudo systemctl daemon-reload`. `finbot.service` itself needs no change. With the bot stopped, `cleanup.py 0` deletes the same files.
 
-Two scheduled jobs keep the caches full, each on a fixed interval shorter than the cache lifetime, so that chat requests do not wait on Sharesight or Yahoo (see [Scheduling](#scheduling)): `sharesight_warm.py` fetches the Sharesight portfolio list and each portfolio's holdings report afresh every 30 minutes (`performance_cache_seconds`: 40 minutes), and `yahoo_warm.py` the price history of every stock followed every 8 hours (`cache_seconds`: 12 hours). Neither fetches anything with `cache=False`, and `sharesight_warm.py` nothing without Sharesight configured. A refresh that fails is logged, the previous cache keeps being served while fresh, and the next run tries again. After a flush, the next runs fill the cache again. A data directory with its own `schedule.cron` needs their lines:
+Two scheduled jobs keep the caches full, each on a fixed interval shorter than the cache lifetime, so that chat requests do not wait on Sharesight or Yahoo (see [Scheduling](#scheduling)): `sharesight_warm.py` fetches the Sharesight portfolio list and each portfolio's holdings report afresh every 30 minutes (`performance_cache_seconds`: 40 minutes), and `yahoo_warm.py` the price history of every stock followed every 8 hours (`cache_seconds`: 12 hours). Neither fetches anything with `cache=False`, and `sharesight_warm.py` nothing without Sharesight configured. A refresh that fails is logged, the previous cache keeps being served while fresh, and the next run tries again. A flush runs both at once (see above). A data directory with its own `schedule.cron` needs their lines:
 ```
 14,44 * * * * sharesight_warm.py
 07 */8 * * * yahoo_warm.py
diff --git a/bot.py b/bot.py
index 9241e98c..e7f4ab2e 100755
--- a/bot.py
+++ b/bot.py
@@ -177,13 +177,15 @@ def reload_all(*args):
 
 def flush_cache(*args):
 	"""SIGUSR1 (`systemctl kill -s USR1 finbot`): delete the cache and keep the state (watchlist, seen trades, ratings,
-	schedule). The warm jobs fill it again on their next runs, and any request meanwhile fetches what it needs."""
+	schedule), then fill it again straight away with the warm jobs, in a thread of their own (a reload need not wait
+	for them)."""
 	try:
 		removed = cleanup.run(days=0)
 		print(f"cache: flushed {removed} file{'' if removed == 1 else 's'} from {config.env.cache_dir}",
 			file=sys.stderr)
 	except Exception as e:
 		print('ERROR', delivery.error_line(e, 'cache flush'), file=sys.stderr)
+	jobs.start_thread(jobs.warm_now, scheduler)
 
 if __name__ == '__main__':
 	print(f'Starting finbot {git_version()}', file=sys.stderr)
diff --git a/lib/jobs.py b/lib/jobs.py
index ed3fa0d5..e6db402d 100644
--- a/lib/jobs.py
+++ b/lib/jobs.py
@@ -35,6 +35,7 @@
 current = None # the running Scheduler, set by bot.py; the chat commands use it
 # scheduled, but not run from chat: .jobs lists them under System
 HIDDEN = ('cleanup', 'podcasts', 'sharesight_warm', 'yahoo_warm')
+WARM = 'sharesight_warm.py; yahoo_warm.py' # the jobs that fill the cache, run at once after a flush too
 STATE_FILE = 'finbot_jobs.json' # last fire time of every job, so a restart neither repeats a run nor loses one
 JOB_TIMEOUT = 5 * 60 # seconds each command may run before it is stopped and reported
 IGNORED_VARIABLES = ('MAILTO', 'SHELL', 'PATH', 'HOME', 'LOGNAME')
@@ -81,6 +82,23 @@ def start_thread(target, *args):
 	threading.Thread(target=target, args=args, daemon=True).start()
 
 
+def warm_now(scheduler=None):
+	"""Run the warm jobs (WARM) now, one after the other, each in its own child process: through the scheduler when it
+	runs, so that one already running on its schedule is not started twice; else directly, a failure reported in the
+	chat as the scheduler reports one."""
+	commands = parse_command(WARM)
+	print('cache: filling it again:', ', '.join(label for label, _, _ in commands), file=sys.stderr)
+	if scheduler:
+		scheduler.execute(commands)
+	else:
+		for label, module, kwargs in commands:
+			try:
+				run_child(label, module, kwargs)
+			except Exception as e:
+				delivery.report_error(e, context=label)
+	print('cache: warm jobs done', file=sys.stderr)
+
+
 def exit_reason(status):
 	"""Why a job process ended, when it did not report it itself: killed by a signal, or failed before it could say."""
 	if status < 0:
diff --git a/lib/reports/sharesight_warm.py b/lib/reports/sharesight_warm.py
index 666dae07..85606a98 100644
--- a/lib/reports/sharesight_warm.py
+++ b/lib/reports/sharesight_warm.py
@@ -5,10 +5,10 @@
 
 def run():
 	"""Fetch afresh the portfolio list and each portfolio's holdings report, which every report starts from: the
-	holdings report is cached for performance_cache_seconds (40 minutes by default), and the default schedule runs
-	this every 30 minutes. Nothing without Sharesight configured, or with the cache preference off (nothing would be
-	kept). A failure is only logged: the caches it would have replaced are served while fresh, and the next run tries
-	again. Returns how many portfolios it refreshed."""
+	holdings report is cached for performance_cache_seconds (40 minutes by default); the default schedule runs this
+	every 30 minutes, and a flush at once. Nothing without Sharesight configured, or with the cache preference off
+	(nothing would be kept). A failure is only logged: the caches it would have replaced are served while fresh, and
+	the next run tries again. Returns how many portfolios it refreshed."""
 	if not (config.prefs.cache and config.env.sharesight_auth.get('client_id')):
 		return 0
 	try:
diff --git a/lib/reports/yahoo_warm.py b/lib/reports/yahoo_warm.py
index 9a75d7d6..2a53f912 100644
--- a/lib/reports/yahoo_warm.py
+++ b/lib/reports/yahoo_warm.py
@@ -4,16 +4,18 @@
 from lib import config, watchlist, yahoo
 
 def run():
-	"""Fetch afresh the price history of every stock followed (the watchlist and the holdings), so each is cached for
-	cache_seconds (12 hours by default) from now: the default schedule runs this every 8 hours, before any runs out.
-	With the cache preference off nothing would be kept, so nothing is fetched. Returns how many it fetched."""
+	"""Fetch afresh the price history of every stock followed (the watchlist, and the holdings where Sharesight is
+	configured), so each is cached for cache_seconds (12 hours by default) from now: the default schedule runs this
+	every 8 hours, before any runs out, and a flush at once. With the cache preference off nothing would be kept, so
+	nothing is fetched. Returns how many it fetched."""
 	if not config.prefs.cache:
 		return 0
-	try:
-		tickers = watchlist.with_holdings()
-	except Exception as e: # Sharesight unreachable, say: the watchlist's at least
-		print("yahoo_warm: no holdings from Sharesight:", e, file=sys.stderr)
-		tickers = watchlist.tickers()
+	tickers = watchlist.tickers()
+	if config.env.sharesight_auth.get('client_id'):
+		try:
+			tickers = watchlist.with_holdings()
+		except Exception as e: # Sharesight unreachable, say: the watchlist's at least
+			print("yahoo_warm: no holdings from Sharesight:", e, file=sys.stderr)
 	failed = yahoo.prefetch_history(tickers, seconds=0)
 	print(f"yahoo_warm: {len(tickers) - len(failed)} of {len(tickers)} price histories fetched", file=sys.stderr)
 	return len(tickers) - len(failed)
diff --git a/tests/test_edges.py b/tests/test_edges.py
index f68de779..b25ec1e1 100644
--- a/tests/test_edges.py
+++ b/tests/test_edges.py
@@ -348,6 +348,18 @@ def work():
 		self.assertEqual(events, ['start', 'end', 'start', 'end']) # never two at once
 
 
+class CacheFlush(FinbotCase):
+	def test_a_flush_fills_the_cache_again_at_once(self):
+		"""Once the cache is deleted, a flush starts the warm jobs, through the scheduler, in a thread of their own."""
+		started = []
+		self.patch(bot.jobs, 'start_thread', lambda target, *args: started.append((target, args)))
+		self.patch(bot, 'scheduler', 'the scheduler')
+		with contextlib.redirect_stderr(io.StringIO()) as log:
+			bot.flush_cache()
+		self.assertIn('cache: flushed', log.getvalue())
+		self.assertEqual(started, [(bot.jobs.warm_now, ('the scheduler',))])
+
+
 class StartupSignals(unittest.TestCase):
 	"""A reload (SIGHUP) that arrives while the bot is still starting must not kill it."""
 
@@ -392,6 +404,7 @@ def test_usr1_flushes_the_cache_and_keeps_the_state(self):
 			for name in cached: # the last: a write in flight
 				(cache / name).write_text('{}')
 			Path(data, 'finbot_watchlist.json').write_text('["T1"]') # state, not cache
+			Path(data, 'preferences.conf').write_text('cache=False\n') # the warm jobs a flush starts fetch nothing
 			env = bare_environment()
 			env.update(DATA_DIR=data, port='0', scheduler='False')
 			proc = subprocess.Popen([sys.executable, '-u', 'bot.py'], cwd=ROOT, env=env, stderr=subprocess.PIPE,
@@ -415,9 +428,10 @@ def test_usr1_flushes_the_cache_and_keeps_the_state(self):
 			proc.send_signal(signal.SIGUSR1)
 			for line in proc.stderr:
 				lines.append(line)
-				if 'cache: flushed' in line:
+				if 'cache: warm jobs done' in line: # each in a child process, as the scheduler is off
 					break
 			self.assertIn('cache: flushed 2 files', ''.join(lines))
+			self.assertIn('cache: filling it again: sharesight_warm.py, yahoo_warm.py', ''.join(lines))
 			self.assertEqual(sorted(os.listdir(cache)), ['finbot_x.json.abc.tmp', 'matplotlib'])
 			self.assertTrue(Path(data, 'finbot_watchlist.json').exists())
 
diff --git a/tests/test_scheduler.py b/tests/test_scheduler.py
index ddbaab88..9668777b 100644
--- a/tests/test_scheduler.py
+++ b/tests/test_scheduler.py
@@ -71,9 +71,11 @@ def test_missing_cache_dir_is_fine(self):
 
 class YahooWarmJob(FinbotCase):
 	def setUp(self):
-		"""The real prefetch, Yahoo answering each history with the ticker's name, except T2's, which fails."""
+		"""The real prefetch, Yahoo answering each history with the ticker's name, except T2's, which fails; Sharesight
+		configured."""
 		super().setUp()
 		self.patch(yahoo, 'prefetch_history', PREFETCH_HISTORY)
+		self.env(sharesight_auth={'client_id': 'id'})
 		def answer(ticker, days, full):
 			"""A history that names its ticker; none for T2."""
 			if ticker == 'T2':
@@ -91,9 +93,12 @@ def test_every_followed_stocks_history_is_fetched_afresh(self):
 		self.assertIn(f'yahoo_warm: {len(TICKERS) - 1} of {len(TICKERS)} price histories fetched', log.getvalue())
 
 	def test_without_sharesight_the_watchlist_is_fetched(self):
-		"""While Sharesight cannot be asked for the holdings, the watchlist's histories are fetched."""
+		"""While Sharesight cannot be asked for the holdings, the watchlist's histories are fetched; without Sharesight
+		configured, it is not asked."""
+		asked = []
 		def unreachable():
 			"""Sharesight failing, as watchlist.with_holdings meets it."""
+			asked.append(1)
 			raise RuntimeError('Sharesight auth: down')
 		self.patch(watchlist, 'with_holdings', unreachable)
 		self.patch(watchlist, 'tickers', lambda: ['W1'])
@@ -101,6 +106,8 @@ def unreachable():
 			self.assertEqual(yahoo_warm.run(), 1)
 		self.assertEqual(cache.read('finbot_yahoo_history_W1.json'), {'history of': 'W1'})
 		self.assertIn('no holdings from Sharesight', log.getvalue())
+		self.env(sharesight_auth={'client_id': None})
+		self.assertEqual((yahoo_warm.run(), len(asked)), (1, 1))
 
 	def test_with_the_cache_off_nothing_is_fetched(self):
 		"""With the cache preference off nothing would be kept, so nothing is fetched."""
@@ -175,6 +182,36 @@ def scheduler(self, catchup=5):
 		return s
 
 
+class WarmNow(SchedulerBase):
+	def test_the_warm_jobs_run_through_the_scheduler(self):
+		"""warm_now runs both warm jobs through the scheduler, which does not start one that is running already."""
+		warmed = []
+		self.patch(jobs.SCRIPTS['sharesight_warm'], 'run', lambda: warmed.append('sharesight'))
+		self.patch(jobs.SCRIPTS['yahoo_warm'], 'run', lambda: warmed.append('yahoo'))
+		s = self.scheduler()
+		with contextlib.redirect_stderr(io.StringIO()) as log:
+			jobs.warm_now(s)
+			s.busy.add('yahoo_warm.py') # on its schedule
+			jobs.warm_now(s)
+		self.assertEqual(warmed, ['sharesight', 'yahoo', 'sharesight'])
+		self.assertIn('cache: filling it again: sharesight_warm.py, yahoo_warm.py', log.getvalue())
+
+	def test_without_a_scheduler_each_runs_in_a_child_process(self):
+		"""With the scheduler off, each warm job runs in a child process of its own, a failure reported."""
+		started = []
+		def child(label, module, kwargs):
+			"""Note the child process started; the second one runs out of time."""
+			started.append(label)
+			if label == 'yahoo_warm.py':
+				raise TimeoutError('still running after 5 minutes')
+			return 0
+		self.patch(jobs, 'run_child', child)
+		with contextlib.redirect_stderr(io.StringIO()):
+			jobs.warm_now()
+		self.assertEqual(started, ['sharesight_warm.py', 'yahoo_warm.py'])
+		self.assertTrue(any('still running after 5 minutes' in e for e in self.errors))
+
+
 class SchedulerTests(SchedulerBase):
 	def test_fires_on_the_minute_on_weekdays_only(self):
 		"""A job fires on its minute, once, on weekdays only; runs missed over days coalesce into one."""