fix: resolve profile entries with their owner

scripts/cross_reference.py

@@ -242,10 +242,17 @@ class FileConsumers:
         self.platforms: dict[str, set[str]] = {}
         self.emulators: dict[str, set[str]] = {}
 
-    def sha1_of(self, entry: dict, dest: str = "") -> str | None:
+    def sha1_of(self, entry: dict, dest: str = "", owner: str = "") -> str | None:
-        """The collected file an entry resolves to, or None."""
+        """The collected file an entry resolves to, or None.
+
+        A profile entry names its owner, as the builder does: without it the
+        name step chose another tree's copy, and the page credited the file
+        the pack does not ship.
+        """
         if not entry.get("name"):
             return None
+        if owner:
+            entry = {**entry, "source_profile": owner}
         path, status = resolve_local_file(
             entry, self._db, self._zip_contents, dest_hint=dest
         )
@@ -266,7 +273,7 @@ class FileConsumers:
             if profile.get("type") == "alias":
                 continue
             for entry in profile.get("files", []):
-                sha1 = self.sha1_of(entry, entry.get("path") or "")
+                sha1 = self.sha1_of(entry, entry.get("path") or "", owner=name)
                 if sha1:
                     self.emulators.setdefault(sha1, set()).add(name)
 

scripts/packextras.py

@@ -213,7 +213,9 @@ def _agnostic_scan_extras(
             if not isinstance(candidate, dict):
                 continue
             local, status = resolve_local_file(
-                candidate, db, dest_hint=candidate.get("path", "")
+                {**candidate, "source_profile": emu_name},
+                db,
+                dest_hint=candidate.get("path", ""),
             )
             if not local or "/" not in local:
                 continue

scripts/slots.py

@@ -175,8 +175,9 @@ def profile_claims(
             if not dest:
                 continue
             full = f"{base_dest}/{dest}" if base_dest else dest
+            # The owner's own copy is the one the builder ships.
             local, status = resolve_local_file(
-                entry,
+                {**entry, "source_profile": emu_name},
                 db,
                 zip_contents,
                 dest_hint=dest,

tests/test_owner_resolution.py

@@ -0,0 +1,65 @@
+"""Every reader of a profile entry names its owner, as the builder does.
+
+The resolver prefers the owning profile's copy when a name has several. The
+builder passes `source_profile`; the site's consumer index, the slot claims
+and the agnostic scan did not, and resolved 95 entries to other files: the
+site credited aethersx2's game_controller_db.txt to armsx2 and left the copy
+armsx2's pack ships without a reader, and MAME's neogeo.zip without MAME.
+"""
+
+from __future__ import annotations
+
+import sys
+import unittest
+from pathlib import Path
+
+REPO_ROOT = Path(__file__).resolve().parent.parent
+sys.path.insert(0, str(REPO_ROOT / "scripts"))
+
+import cross_reference  # noqa: E402
+import slots  # noqa: E402
+from common import (  # noqa: E402
+    build_zip_contents_index,
+    load_database,
+    load_emulator_profiles,
+)
+
+
+class ReadersNameTheOwner(unittest.TestCase):
+    @classmethod
+    def setUpClass(cls):
+        db_path = REPO_ROOT / "database.json"
+        if not db_path.exists():
+            raise unittest.SkipTest("database.json is not built")
+        cls.db = load_database(str(db_path))
+        cls.profiles = load_emulator_profiles(str(REPO_ROOT / "emulators"))
+        cls.sha1_by_path = {r["path"]: s for s, r in cls.db["files"].items()}
+
+    def test_the_site_credits_the_copy_the_pack_ships(self):
+        consumers = cross_reference.FileConsumers(
+            self.db, build_zip_contents_index(self.db)
+        )
+        consumers.add_emulators(self.profiles)
+        for path, owner in (
+            ("bios/Other/armsx2/game_controller_db.txt", "armsx2"),
+            ("bios/Arcade/MAME/neogeo.zip", "mame"),
+        ):
+            with self.subTest(path=path):
+                sha1 = self.sha1_by_path.get(path)
+                if sha1 is None:
+                    self.skipTest(f"{path} is not collected")
+                self.assertIn(owner, consumers.emulators.get(sha1, set()))
+
+    def test_slot_claims_resolve_the_owners_tree(self):
+        claims = slots.profile_claims({"pcem": self.profiles["pcem"]}, self.db)
+        held = {
+            claim.destination.rsplit("/", 1)[-1]: claim.local_path for claim in claims
+        }
+        local = held.get("ide_xt.bin")
+        if local is None:
+            self.skipTest("ide_xt.bin is not claimed")
+        self.assertTrue(local.startswith("bios/Other/pcem/"), local)
+
+
+if __name__ == "__main__":
+    unittest.main()