refactor: index companion files by directory

scripts/packextras.py

@@ -446,12 +446,28 @@ def _companions(
     wanted |= {f"{stem}.{ext}" for ext in extensions}
     return sorted(
         (sha1, entry)
-        for sha1, entry in files_db.items()
-        if entry.get("path", "").rsplit("/", 1)[0] + "/" == directory
-        and str(entry.get("name", "")).lower() in wanted
+        for sha1, entry in _files_by_directory(files_db).get(directory, ())
+        if str(entry.get("name", "")).lower() in wanted
     )
 
 
+# The database a directory index was built from, with the index: one walk
+# per database instead of one per scanned image (1 023 walks of 19 041
+# entries in verify --all).
+_DIRECTORY_INDEX: tuple[dict, dict[str, list[tuple[str, dict]]]] | None = None
+
+
+def _files_by_directory(files_db: dict) -> dict[str, list[tuple[str, dict]]]:
+    global _DIRECTORY_INDEX
+    if _DIRECTORY_INDEX is None or _DIRECTORY_INDEX[0] is not files_db:
+        index: dict[str, list[tuple[str, dict]]] = {}
+        for sha1, entry in files_db.items():
+            directory = entry.get("path", "").rsplit("/", 1)[0] + "/"
+            index.setdefault(directory, []).append((sha1, entry))
+        _DIRECTORY_INDEX = (files_db, index)
+    return _DIRECTORY_INDEX[1]
+
+
 def _archive_prefix_extras(
     profiles: dict,
     relevant: set,

tests/test_audit_regressions.py

@@ -3698,5 +3698,24 @@ class AZipMemberIsHashedOnce(unittest.TestCase):
         self.assertEqual(len(opened), 1)
 
 
+
+class CompanionsComeFromADirectoryIndex(unittest.TestCase):
+    """_companions walked the whole database per scanned image."""
+
+    def test_same_answer_one_walk(self):
+        from scripts import packextras
+
+        files_db = {
+            "a": {"path": "bios/Sony/PS2/ps2.bin", "name": "ps2.bin"},
+            "b": {"path": "bios/Sony/PS2/ps2.bin.rom1", "name": "ps2.bin.rom1"},
+            "c": {"path": "bios/Sony/PS2/ps2.nvm", "name": "ps2.nvm"},
+            "d": {"path": "bios/Other/ps2.nvm", "name": "ps2.nvm"},
+        }
+        found = packextras._companions("bios/Sony/PS2/ps2.bin", "ps2.bin", {"rom1", "nvm"}, files_db)
+        self.assertEqual([sha1 for sha1, _ in found], ["b", "c"])
+        index = packextras._files_by_directory(files_db)
+        self.assertIs(packextras._files_by_directory(files_db), index)
+
+
 if __name__ == "__main__":
     unittest.main()