Feature set 175.

src/MoogleAPI.Web/Infrastructure/Battle/BattlePool.cs

@@ -20,8 +20,19 @@ public class BattlePool(AppDbContext db, HybridCache cache)
     /// only playable when both sides come from the same one.
     /// </summary>
     /// <remarks>
-    /// Absent for want of stats: VIII, whose enemy articles give HP as level-scaling
-    /// coefficients rather than a number, and XI, XIV and XVI, which publish almost none.
+    /// Absent for want of stats: XI, XIV and XVI. The first two publish none at all — not one
+    /// of their 419 articles carries an enemy stats infobox, because most of those rows are
+    /// races and family pages rather than single enemies, and the specialist wikis leave the
+    /// same fields blank.
+    /// <para>
+    /// VIII is a different case and no longer a data one. Its articles give stats as
+    /// coefficients of a level curve rather than numbers, so nothing could be read off the
+    /// wikitext; 85 of its 118 enemies now hold the values the wiki's own table renders at
+    /// level 50, and 68% of those have an elemental weakness — between VI and V, so the bar
+    /// II fails is comfortably cleared. What it does not have is moves: 14 of the 85 list an
+    /// ability, against a pool where most do. Adding it is a gameplay decision rather than a
+    /// blocked one.
+    /// </para>
     /// <para>
     /// II is absent for a different reason. It has HP and art for 166 monsters, so it looks
     /// playable — but exactly one of them has a listed elemental weakness, against 31–91% in

src/MoogleAPI.Web/Infrastructure/Wiki/WikiClient.cs

@@ -529,10 +529,21 @@ private static string[] MagicDefenseFields(string? release) =>
     internal static string? ParseFieldList(string wikitext, params string[] fieldNames) =>
         ParseFieldList(wikitext, StatPrefix, fieldNames);
 
-    // Character infoboxes name an ability field per release — "ffviir abilities",
-    // "ffviir2 abilities" — so the closed platform list that guards the enemy stat fields is
-    // too strict here. Character infoboxes have no equivalent of "bribe gil" to guard against.
-    private const string ReleasePrefix = @"(?:[a-z0-9]+\s+){0,2}";
+    /// <summary>
+    /// A tag naming a numbered entry, or a remake of one: <c>ffvii</c>, <c>ffviir</c>,
+    /// <c>ffviir2</c>. Written longest-first so that <c>ffxiii</c> cannot match as <c>ffxi</c>.
+    /// </summary>
+    private const string MainlineTag =
+        @"ff(?:xvi|xv|xiv|xiii|xii|xi|x|ix|viii|vii|vi|v|iv|iii|ii|i)r?2?";
+
+    // Character infoboxes name a field per release — "ffviir abilities", "ffviir2 abilities" —
+    // so the closed platform list that guards the enemy stat fields is too strict here. It
+    // cannot be an open wildcard either: one article covers a character's whole career, and a
+    // spin-off's fields sit right beside the numbered entry's. Reading any tag is what gave
+    // Ashe and Penelo their Revenant Wings jobs ("xiirw job"), Auron the ability list from
+    // X-2's Creature Creator ("x2 cc abilities"), and Zack a weapon out of Crisis Core
+    // ("viicc weapon") for a game he is not playable in.
+    private const string ReleasePrefix = $@"(?:{MainlineTag}\s+)?";
 
     internal static string? ParseCharacterFieldList(string wikitext, params string[] fieldNames) =>
         ParseFieldList(wikitext, ReleasePrefix, fieldNames);
@@ -549,21 +560,22 @@ private static string[] MagicDefenseFields(string? release) =>
     /// The prefix cannot be allowed to swallow a qualifier, though — every one of these
     /// articles also carries <c>|ultimate weapon=</c>, naming a specific late-game item rather
     /// than the class of arms the character uses. Matching it would call Cloud's weapon "Ultima
-    /// Weapon" instead of "Broadswords".
+    /// Weapon" instead of "Broadswords". Admitting only <see cref="MainlineTag"/> rejects it
+    /// along with every other title's fields.
     /// </para>
     /// </remarks>
     internal static string? ParseCharacterField(string wikitext, params string[] fieldNames)
     {
         foreach (var field in fieldNames)
         {
             foreach (Match match in Regex.Matches(wikitext,
-                         $@"^\|\s*((?:[a-z0-9]+\s+){{0,2}}){Regex.Escape(field)}\s*=\s*(.+)$",
+                         $@"^\|\s*({ReleasePrefix}){Regex.Escape(field)}\s*=\s*(.+)$",
                          RegexOptions.IgnoreCase | RegexOptions.Multiline))
             {
                 if (QualifiedField.IsMatch(match.Groups[1].Value)) continue;
 
                 var cleaned = CleanFieldValue(match.Groups[2].Value);
-                if (cleaned is not null) return cleaned;
+                if (cleaned is not null && !IsPlaceholder(cleaned)) return cleaned;
             }
         }
 
@@ -959,6 +971,12 @@ private static bool ShouldRetry(HttpRequestException ex) =>
         // Strip refs and HTML tags
         value = Regex.Replace(value, @"<ref\b[^>]*/?>.*?</ref>", "", RegexOptions.Singleline);
         value = Regex.Replace(value, @"<ref\b[^>]*/?>", "");
+        // Editor notes. The unterminated form is not a typo to tolerate but the normal case
+        // here: the comment opens on the field's line and closes several lines later, and this
+        // value was cut at the line end long before it got here. Final Fantasy XV's Y'jhimei
+        // had an occupation of "<!--Physical desc." for want of this.
+        value = Regex.Replace(value, @"<!--.*?-->", "", RegexOptions.Singleline);
+        value = Regex.Replace(value, @"<!--.*", "");
         // <br> separates multiple values in an infobox field. Dropping it like any other tag
         // fuses them together ("First Shield of RosariaMarquess of Rosaria"), so it has to
         // become a real separator before the generic tag strip runs.

tests/MoogleAPI.Tests/WikiParsingTests.cs

@@ -183,6 +183,82 @@ public void ParsesCharacterAbilityLists(string line, string expected)
             WikiClient.ParseCharacterFieldList("{{infobox character\n" + line + "\n}}", "abilities", "ability"));
     }
 
+    /// <summary>
+    /// One article covers a character's whole career, so a spin-off's fields sit right beside
+    /// the numbered entry's. Only a mainline tag counts.
+    /// </summary>
+    [Fact]
+    public void IgnoresFieldsBelongingToADifferentTitle()
+    {
+        const string vaan = """
+            {{infobox character
+            |name=Vaan
+            |xiirw job=Sky Pirate
+            |ffta2 job=Sky Pirate
+            }}
+            """;
+
+        Assert.Null(WikiClient.ParseCharacterField(vaan, "job", "class"));
+
+        const string auron = """
+            {{infobox character
+            |name=Auron
+            |ffx limit break=[[Bushido (Final Fantasy X)|Bushido]]
+            |x2 cc abilities=[[Full Break]], [[Dragon Fang (Bushido)|Dragon Fang]]
+            }}
+            """;
+
+        Assert.Equal("Bushido",
+            WikiClient.ParseCharacterFieldList(auron, "abilities", "ability", "limit break"));
+    }
+
+    /// <summary>
+    /// A remake of the numbered entry is still that entry — Cloud's arms are stated only under
+    /// the per-release tags.
+    /// </summary>
+    [Theory]
+    [InlineData("ffvii")]
+    [InlineData("ffviir")]
+    [InlineData("ffviir2")]
+    [InlineData("")]
+    public void ReadsAMainlineOrRemakeTag(string tag)
+    {
+        var line = $"|{tag}{(tag.Length > 0 ? " " : "")}weapon=Broadswords";
+
+        Assert.Equal("Broadswords",
+            WikiClient.ParseCharacterField("{{infobox character\n" + line + "\n}}", "weapon"));
+    }
+
+    [Theory]
+    [InlineData("|ultimate weapon=Ultima Weapon")]
+    [InlineData("|ffvii ultimate weapon=Ultima Weapon")]
+    [InlineData("|viicc weapon=SOLDIER Sword")]
+    [InlineData("|ivtay weapon=Spears")]
+    public void RejectsAQualifiedOrSpinOffWeapon(string line)
+    {
+        Assert.Null(WikiClient.ParseCharacterField("{{infobox character\n" + line + "\n}}", "weapon"));
+    }
+
+    /// <summary>
+    /// An editor's comment usually opens on the field's line and closes further down the
+    /// article, so the value arrives here with the comment unterminated.
+    /// </summary>
+    [Theory]
+    [InlineData("|occupation=<!--Physical desc.", null)]
+    [InlineData("|occupation=<!-- fill this in --> Florist", "Florist")]
+    [InlineData("|occupation=Florist<!--per the ultimania-->", "Florist")]
+    public void StripsEditorComments(string line, string? expected)
+    {
+        Assert.Equal(expected,
+            WikiClient.ParseInfoboxField("{{infobox character\n" + line + "\n}}", "occupation"));
+    }
+
+    [Fact]
+    public void TreatsAStatedNoneAsNoValue()
+    {
+        Assert.Null(WikiClient.ParseCharacterField("{{infobox character\n|job=None\n}}", "job"));
+    }
+
     [Fact]
     public void UnwrapsWikilinksToTheirDisplayText()
     {