From dce9c6e8f11acbaa2248b98753cc921e7dad269a Mon Sep 17 00:00:00 2001
From: Abdessamad Derraz <3028866+Abdess@users.noreply.github.com>
Date: Sat, 17 Oct 2026 18:21:27 +0200
Subject: [PATCH] fix: hold the MAME clone step to name rules

---
 scripts/common.py               | 16 ++++++++++----
 tests/test_audit_regressions.py | 39 +++++++++++++++++++++++++++++++++
 2 files changed, 51 insertions(+), 4 deletions(-)

diff --git a/scripts/common.py b/scripts/common.py
index 3d84adb3..2520359e 100644
--- a/scripts/common.py
+++ b/scripts/common.py
@@ -1335,12 +1335,20 @@ def resolve_local_file(
                     if os.path.exists(path):
                         return path, "zip_exact"
 
-    # MAME clone fallback: if a file was deduped, resolve via canonical
-    if _depth < 3 and not has_strong_hash:
+    # MAME clone fallback: if a file was deduped, resolve via canonical. The
+    # clone name is a name step like the others: an unsourceable entry skips
+    # it, and the size the code checks travels with it (dedup only merges
+    # identical bytes, so the canonical copy has the clone's size).
+    if _depth < 3 and not has_strong_hash and not unsourceable:
         clone_map = get_mame_clone_map(db.get("bios_dir"))
         canonical = clone_map.get(name)
         if canonical and canonical != name:
-            canonical_entry = {"name": canonical}
+            canonical_entry = {
+                field: file_entry[field]
+                for field in ("size", "min_size", "max_size", "validation")
+                if field in file_entry
+            }
+            canonical_entry["name"] = canonical
             result = resolve_local_file(
                 canonical_entry,
                 db,
@@ -1349,7 +1357,7 @@ def resolve_local_file(
                 _depth=_depth + 1,
                 data_dir_registry=data_dir_registry,
             )
-            if result[0]:
+            if result[0] and result[1] != "hash_mismatch":
                 return result[0], "mame_clone"
 
     # Data directory fallback: scan data/ caches for matching filename
diff --git a/tests/test_audit_regressions.py b/tests/test_audit_regressions.py
index fcf0af22..3eba6746 100644
--- a/tests/test_audit_regressions.py
+++ b/tests/test_audit_regressions.py
@@ -2936,3 +2936,42 @@ class ADataDirectoryCopyPassesTheNameChecks(unittest.TestCase):
     def test_another_adler32_is_a_contradiction(self):
         status = self.resolve(validation=["adler32"], known_hash_adler32="0xdeadbeef")[1]
         self.assertEqual(status, "hash_mismatch")
+
+
+class AMameCloneNameIsANameStep(unittest.TestCase):
+    """The clone fallback relaunched resolution on the bare canonical name:
+    an unsourceable entry and a size the code checks were both served the
+    parent set."""
+
+    def setUp(self):
+        self.tmp = Path(tempfile.mkdtemp(dir=TMP_ROOT))
+        self.addCleanup(lambda: __import__("shutil").rmtree(self.tmp, True))
+        bios = self.tmp / "bios" / "Arcade"
+        bios.mkdir(parents=True)
+        parent = bios / "parent.zip"
+        parent.write_bytes(b"PARENTSET")
+        (self.tmp / "_mame_clones.json").write_text(
+            json.dumps({"parent.zip": {"clones": ["clone.zip"]}})
+        )
+        sha1 = hashlib.sha1(parent.read_bytes()).hexdigest()
+        self.db = {
+            "bios_dir": str(self.tmp / "bios"),
+            "files": {sha1: {"path": str(parent), "name": "parent.zip",
+                             "sha1": sha1, "size": 9}},
+            "indexes": {"by_name": {"parent.zip": [sha1]}, "by_md5": {},
+                        "by_path_suffix": {}, "by_crc32": {}},
+        }
+
+    def resolve(self, **fields):
+        return resolve_local_file({"name": "clone.zip", **fields}, self.db)
+
+    def test_a_clone_resolves_to_its_parent(self):
+        self.assertEqual(self.resolve()[1], "mame_clone")
+
+    def test_an_unsourceable_entry_is_not_served_the_parent(self):
+        self.assertEqual(self.resolve(unsourceable="not held"), (None, "not_found"))
+
+    def test_a_size_the_parent_contradicts_is_not_served(self):
+        self.assertEqual(
+            self.resolve(size=5, validation=["size"]), (None, "not_found")
+        )