Avoid overscheduling due to sibling manifest rule

bugbug/test_scheduling.py

@@ -914,6 +914,11 @@ def get_test_info(date: datetime) -> dict[str, Any]:
 
 manifest_by_path: dict[str, set[str]] | None = None
 
+# If a modified file is close to more manifests than this, it is too broad
+# (e.g. dom/moz.build) for the sibling heuristic to be informative, so we
+# don't schedule any of them and leave the decision to the model.
+MAX_SIBLING_MANIFESTS = 42
+
 
 def find_manifests_for_paths(repo_dir_str: str, paths: list[str]) -> set[str]:
     global manifest_by_path
@@ -1009,16 +1014,22 @@ def collect_support_files(value):
         # If a manifest, a test, or a support file is modified, run the manifest that includes it.
         if path in manifest_by_path:
             manifests.update(manifest_by_path[path])
-        else:
+        # Skip root-level files, otherwise we'd walk the whole repository and
+        # schedule every manifest.
+        elif (repo_dir / path).parent != repo_dir:
             # Find manifests that are in test subfolders close to a modified file (e.g. if dom/battery/BatteryManager.cpp is modified, we should run dom/battery/test/chrome.toml and dom/battery/test/mochitest.toml).
+            sibling_manifests: set[str] = set()
             for sibling in (repo_dir / path).parent.rglob("*"):
                 if sibling.is_dir() and repository.is_test(f"{str(sibling)}/"):
-                    manifests.update(
+                    sibling_manifests.update(
                         str(f.relative_to(repo_dir))
                         for f in sibling.rglob("*.toml")
                         if f.is_file()
                     )
 
+            if len(sibling_manifests) <= MAX_SIBLING_MANIFESTS:
+                manifests.update(sibling_manifests)
+
         # If a web-platform test or meta is modified, run the relevant web-platform folder.
         if not any(path.endswith(ignore) for ignore in ("/META.yml", "/README.md")):
             for base in ("testing/web-platform/mozilla", "testing/web-platform"):
@@ -1101,6 +1112,10 @@ def find_tasks_for_paths(
     # Any file in a folder close to a gtest folder is modified (e.g. dom/media/CubebUtils.cpp and we have dom/media/gtest/).
     if not select_gtest:
         for path in paths:
+            # Skip root-level files, otherwise we'd walk the whole repository.
+            if (repo_dir / path).parent == repo_dir:
+                continue
+
             for sibling in (repo_dir / path).parent.rglob("*"):
                 if sibling.is_dir() and any(
                     part in _GTEST_FOLDERS for part in sibling.parts

tests/test_test_scheduling.py

@@ -1078,6 +1078,41 @@ def test_find_manifests_for_paths(tmp_path) -> None:
         "test/chrome.toml"
     }
 
+    # A root-level file that is not referenced by any manifest must not
+    # schedule every manifest in the repository.
+    (tmp_path / "mach").touch()
+    assert test_scheduling.find_manifests_for_paths(str(tmp_path), ["mach"]) == set()
+
+    # A file close to too many manifests (e.g. dom/moz.build) must not
+    # schedule all of them.
+    (tmp_path / "hub" / "moz.build").parent.mkdir(parents=True)
+    (tmp_path / "hub" / "moz.build").touch()
+    for i in range(test_scheduling.MAX_SIBLING_MANIFESTS):
+        (tmp_path / "hub" / f"component{i}" / "test").mkdir(parents=True)
+        (tmp_path / "hub" / f"component{i}" / "test" / "mochitest.toml").touch()
+
+    assert (
+        len(test_scheduling.find_manifests_for_paths(str(tmp_path), ["hub/moz.build"]))
+        == test_scheduling.MAX_SIBLING_MANIFESTS
+    )
+
+    (tmp_path / "hub" / "one_more" / "test").mkdir(parents=True)
+    (tmp_path / "hub" / "one_more" / "test" / "mochitest.toml").touch()
+
+    assert (
+        test_scheduling.find_manifests_for_paths(str(tmp_path), ["hub/moz.build"])
+        == set()
+    )
+
+    # The cap applies per path, so a narrow file is still scheduled when
+    # modified together with a broad one.
+    assert test_scheduling.find_manifests_for_paths(
+        str(tmp_path), ["hub/moz.build", "dom/battery/BatteryManager.cpp"]
+    ) == {
+        "dom/battery/test/mochitest.toml",
+        "dom/battery/test/chrome.toml",
+    }
+
     assert test_scheduling.find_manifests_for_paths(
         str(tmp_path), ["test/test_resolve_uris_ipc.js"]
     ) == {