fix: resolve the region audit like the pack

scripts/region_audit.py

@@ -19,7 +19,7 @@ from pathlib import Path
 sys.path.insert(0, str(Path(__file__).resolve().parent))
 
 import region
-from common import load_emulator_profiles, load_database, parse_md5_list
+from common import load_database, load_emulator_profiles, resolve_local_file
 
 # No-Intro filename tokens: full English territory names.
 NOINTRO = {
@@ -100,24 +100,21 @@ def catalog_regions(db: dict) -> dict[str, tuple[set[str], str]]:
     return out
 
 
-def resolve_sha1(file_entry: dict, db: dict) -> str | None:
-    """Resolve a profile file entry to a repo SHA1, or None when ambiguous."""
-    files = db["files"]
-    indexes = db["indexes"]
-    raw_sha1 = file_entry.get("sha1")
-    declared = raw_sha1 if isinstance(raw_sha1, list) else [raw_sha1]
-    sha1_hits = [str(value).lower() for value in declared if value]
-    sha1_hits = [value for value in sha1_hits if value in files]
-    if len(sha1_hits) == 1:
-        return sha1_hits[0]
-    for md5 in parse_md5_list(file_entry.get("md5")):
-        hit = indexes["by_md5"].get(md5)
-        if hit:
-            return hit
-    hits = indexes["by_name"].get(file_entry.get("name", ""), [])
-    if isinstance(hits, str):
-        hits = [hits]
-    return hits[0] if len(hits) == 1 else None
+def resolve_sha1(file_entry: dict, db: dict, by_path: dict | None = None) -> str | None:
+    """The repo SHA1 a profile entry resolves to, by the pack's own resolver.
+
+    A private resolver read sha1, then md5, then a lone name: entries that
+    identify their file by path (Dolphin's three IPL.bin), by crc32
+    (geargrafx) or by sha256 (mednafen) were never checked at all.
+    """
+    if by_path is None:
+        by_path = {entry.get("path", ""): sha1 for sha1, entry in db["files"].items()}
+    path, status = resolve_local_file(
+        file_entry, db, dest_hint=file_entry.get("path") or ""
+    )
+    if not path or status in ("not_found", "hash_mismatch"):
+        return None
+    return by_path.get(path)
 
 
 def build_report(profiles: dict, db: dict) -> dict:
@@ -130,6 +127,7 @@ def build_report(profiles: dict, db: dict) -> dict:
     for slugs, _name in catalog.values():
         vocabulary.update(slugs)
 
+    by_path = {entry.get("path", ""): sha1 for sha1, entry in db["files"].items()}
     for emu_name, profile in sorted(profiles.items()):
         if profile.get("type") in ("launcher", "alias"):
             continue
@@ -137,7 +135,7 @@ def build_report(profiles: dict, db: dict) -> dict:
             if not isinstance(f, dict) or not f.get("region"):
                 continue
             declared = region.normalize_declared(f["region"])
-            sha1 = resolve_sha1(f, db)
+            sha1 = resolve_sha1(dict(f, source_profile=emu_name), db, by_path)
             if not sha1 or sha1 not in catalog:
                 unchecked += 1
                 continue

tests/test_audit_regressions.py

@@ -939,17 +939,37 @@ class RegionRegressions(unittest.TestCase):
         )
         self.assertNotEqual(north_america, europe)
 
-    def test_region_audit_accepts_list_valued_md5(self):
-        sha1 = "a" * 40
-        md5 = "b" * 32
-        db = {
-            "files": {sha1: {}},
-            "indexes": {"by_md5": {md5: sha1}, "by_name": {}},
-        }
-        self.assertEqual(
-            region_audit.resolve_sha1({"name": "bios.bin", "md5": [md5]}, db),
-            sha1,
-        )
+    def test_region_audit_resolves_the_way_the_pack_does(self):
+        """A private resolver read sha1, md5 and a lone name: an entry
+        identified by its path (Dolphin's three IPL.bin) was never checked."""
+        import hashlib  # noqa: PLC0415
+
+        with tempfile.TemporaryDirectory(dir=TMP_ROOT) as directory:
+            previous = os.getcwd()
+            os.chdir(directory)
+            self.addCleanup(os.chdir, previous)
+            files, shas = {}, {}
+            for rel, payload in (("bios/GC/USA/IPL.bin", b"usa"), ("bios/GC/JAP/IPL.bin", b"jap")):
+                Path(rel).parent.mkdir(parents=True, exist_ok=True)
+                Path(rel).write_bytes(payload)
+                sha1 = hashlib.sha1(payload).hexdigest()
+                shas[rel] = sha1
+                files[sha1] = {"path": rel, "name": "IPL.bin", "size": len(payload), "sha1": sha1,
+                               "md5": hashlib.md5(payload).hexdigest(),
+                               "sha256": hashlib.sha256(payload).hexdigest(), "crc32": "00000001"}
+            from scripts import generate_db  # noqa: PLC0415
+
+            db = {"files": files, "indexes": generate_db.build_indexes(files, {})}
+            by_md5 = {e["md5"]: s for s, e in files.items()}
+            self.assertEqual(
+                region_audit.resolve_sha1({"name": "IPL.bin", "md5": [files[shas["bios/GC/USA/IPL.bin"]]["md5"]]}, db),
+                shas["bios/GC/USA/IPL.bin"],
+            )
+            self.assertEqual(
+                region_audit.resolve_sha1({"name": "IPL.bin", "path": "GC/JAP/IPL.bin"}, db),
+                shas["bios/GC/JAP/IPL.bin"],
+            )
+            self.assertTrue(by_md5)
 
 
 class ArchiveSecurityRegressions(unittest.TestCase):