feat: follow mame clones in the gap analysis

scripts/common.py

@@ -783,7 +783,7 @@ def resolve_local_file(
 
     # MAME clone fallback: if a file was deduped, resolve via canonical
     if _depth < 3 and not has_strong_hash:
-        clone_map = _get_mame_clone_map()
+        clone_map = get_mame_clone_map()
         canonical = clone_map.get(name)
         if canonical and canonical != name:
             canonical_entry = {"name": canonical}
@@ -835,7 +835,10 @@ def resolve_local_file(
         return "data_dir_hash_exact"
 
     data_dir_mismatch: str | None = None
-    if data_dir_registry:
+    # Without a hash the cache walk matches on filename alone, which is the
+    # step an unsourceable entry has to skip: hiscore.dat names one file per
+    # driver set, so FBNeo's copy would answer for MAME's.
+    if data_dir_registry and (has_strong_hash or not unsourceable):
         for _dd_key, dd_entry in data_dir_registry.items():
             cache_dir = dd_entry.get("local_cache", "")
             if not cache_dir or not os.path.isdir(cache_dir):
@@ -890,7 +893,7 @@ def resolve_local_file(
 _mame_clone_map_cache: dict[str, str] | None = None
 
 
-def _get_mame_clone_map() -> dict[str, str]:
+def get_mame_clone_map() -> dict[str, str]:
     """Load and cache the MAME clone map (clone_name -> canonical_name)."""
     global _mame_clone_map_cache
     if _mame_clone_map_cache is not None:

scripts/cross_reference.py

@@ -20,6 +20,7 @@ from pathlib import Path
 
 sys.path.insert(0, os.path.dirname(__file__))
 from common import (
+    get_mame_clone_map,
     list_registered_platforms,
     load_database,
     load_emulator_profiles,
@@ -157,6 +158,16 @@ def _resolve_source(
     # bios/ via by_path_suffix (regional variants)
     if by_path_suffix and fname in by_path_suffix:
         return "bios"
+    # bios/ under the canonical MAME set name, as resolve_local_file does:
+    # a renamed archive is held once, under the name the dedup kept.
+    canonical = get_mame_clone_map().get(fname)
+    if canonical and canonical != fname:
+        if canonical in by_name and _name_hit(canonical):
+            return "bios"
+        if canonical.lower() in by_name_lower and _name_hit(
+            by_name_lower[canonical.lower()]
+        ):
+            return "bios"
     # data/ supplemental index
     if data_names:
         if fname in data_names or key in data_names:

tests/test_e2e.py

@@ -3114,6 +3114,33 @@ class TestE2E(unittest.TestCase):
         self.assertEqual(os.path.basename(path), "data_only.bin")
         self.assertEqual(status, "data_dir")
 
+    def test_167b_unsourceable_skips_data_dir_name_match(self):
+        """The cache walk matches on filename, so unsourceable has to skip it."""
+        data_dir = os.path.join(self.root, "data", "other-core")
+        os.makedirs(data_dir, exist_ok=True)
+        with open(os.path.join(data_dir, "shared_name.dat"), "wb") as f:
+            f.write(b"ANOTHER CORE DEFINITIONS")
+
+        registry = {"other-core": {"local_cache": data_dir}}
+
+        fe = {"name": "shared_name.dat", "unsourceable": "ships inside the package"}
+        path, status = resolve_local_file(fe, self.db, data_dir_registry=registry)
+        self.assertIsNone(path)
+        self.assertEqual(status, "not_found")
+
+        # A declared hash is content proof, so the cache still answers for it.
+        import hashlib
+
+        digest = hashlib.sha1(b"ANOTHER CORE DEFINITIONS").hexdigest()
+        fe = {
+            "name": "shared_name.dat",
+            "sha1": digest,
+            "unsourceable": "ships inside the package",
+        }
+        path, status = resolve_local_file(fe, self.db, data_dir_registry=registry)
+        self.assertIsNotNone(path)
+        self.assertEqual(status, "data_dir_hash_exact")
+
     def test_168_generate_truth_basic(self):
         """generate_platform_truth resolves cores and builds system truth."""
         import yaml as _yaml
@@ -5991,6 +6018,19 @@ struct BurnDriver BurnDrvneogeo = {
         entry = {"name": "chip.rom", "aliases": ["OTHER.rom"], "crc32": "deadbeef"}
         self.assertEqual(self._gap_source(entry, db), "missing")
 
+    def test_gap_analysis_resolves_renamed_mame_set(self):
+        """A renamed set is held under the name dedup kept, as verify reads it."""
+        import cross_reference
+
+        db = self._gap_db("v4bios.zip", "a" * 40, "465c4e1c")
+        original = cross_reference.get_mame_clone_map
+        cross_reference.get_mame_clone_map = lambda: {"bctvidbs.zip": "v4bios.zip"}
+        try:
+            entry = {"name": "bctvidbs.zip", "required": True}
+            self.assertEqual(self._gap_source(entry, db), "bios")
+        finally:
+            cross_reference.get_mame_clone_map = original
+
 
 if __name__ == "__main__":
     unittest.main()