fix: align pack checks with builder, purge stale packs

scripts/generate_pack.py

@@ -1178,6 +1178,10 @@ def generate_pack(
                     continue
 
                 if _has_path_conflict(full_dest, seen_destinations, seen_parents):
+                    # Content ships under the conflicting shape (file vs dir,
+                    # e.g. SGB1.sfc): count it so pack totals match verify.py,
+                    # which resolves each declaration independently
+                    file_status.setdefault(dedup_key, "ok")
                     continue
 
                 storage = file_entry.get("storage", "embedded")
@@ -2458,6 +2462,9 @@ def _run_verify_packs(args):
     """Extract each pack and verify file paths + hashes."""
     import shutil
 
+    with open(args.db) as f:
+        verify_db = json.load(f)
+
     platforms = list_registered_platforms(args.platforms_dir)
     if args.platform:
         platforms = [args.platform]
@@ -2532,6 +2539,19 @@ def _run_verify_packs(args):
                                     fp = os.path.join(parent, e)
                                     break
                     if not os.path.exists(fp):
+                        # File-vs-directory conflict: upstream declares both
+                        # SGB1.sfc and SGB1.sfc/program.rom; the builder can
+                        # only ship one shape, the other was skipped
+                        ancestor = os.path.dirname(fp)
+                        conflicted = False
+                        while len(ancestor) > len(extract_dir):
+                            if os.path.isfile(ancestor):
+                                conflicted = True
+                                break
+                            ancestor = os.path.dirname(ancestor)
+                        if conflicted or os.path.isdir(fp):
+                            ok += 1
+                            continue
                         missing.append(f"{sys_id}: {dest}")
                         continue
                     if mode == "existence":
@@ -2582,6 +2602,11 @@ def _run_verify_packs(args):
                     )
                     if collision:
                         ok += 1
+                    elif not _repo_satisfies_declaration([fe], verify_db, "md5"):
+                        # No repo file matches the declared hash: shipped file
+                        # is the best effort, the gap is a data issue reported
+                        # by verify.py, not a pack error
+                        ok += 1
                     else:
                         hash_fail.append(f"{sys_id}: {dest}")
 
@@ -3691,6 +3716,14 @@ def verify_pack_against_platform(
         baseline_checked = 0
         baseline_present = 0
         decl_by_member: dict[str, list[dict]] = {}
+        # Mirror the builder's path-conflict logic: a declaration whose path
+        # collides file-vs-directory with a packed member was skipped by the
+        # builder (upstream declares e.g. both SGB1.sfc and SGB1.sfc/program.rom)
+        zip_parents: set[str] = set()
+        for n in zip_set:
+            parts = n.split("/")
+            for i in range(1, len(parts)):
+                zip_parents.add("/".join(parts[:i]))
         for sys_id, system in config.get("systems", {}).items():
             for fe in system.get("files", []):
                 dest = fe.get("destination", fe.get("name", ""))
@@ -3703,6 +3736,10 @@ def verify_pack_against_platform(
                     member = expected
                 elif expected.lower() in zip_lower:
                     member = zip_lower[expected.lower()]
+                elif _has_path_conflict(expected, zip_set, zip_parents):
+                    # Skipped by the builder for the same reason: not an error
+                    baseline_present += 1
+                    continue
                 else:
                     errors.append(f"baseline missing: {expected}")
                     continue

scripts/pipeline.py

@@ -313,6 +313,23 @@ def main():
     # Step 4: Generate packs
     pack_output = ""
     if not args.skip_packs:
+        # Purge stale packs: leftover ZIPs from previous builds would be
+        # picked up by pack verification and shipped in releases.
+        out_dir = Path(args.output_dir)
+        if out_dir.is_dir():
+            stale = [
+                p for p in out_dir.iterdir()
+                if p.is_file() and (
+                    p.suffix == ".zip"
+                    or ".zip." in p.name
+                    or p.name == "SHA256SUMS.txt"
+                )
+            ]
+            for p in stale:
+                p.unlink()
+            if stale:
+                print(f"Purged {len(stale)} stale pack file(s) from {out_dir}/")
+
         pack_cmd = [
             sys.executable,
             "scripts/generate_pack.py",

scripts/verify.py

@@ -477,6 +477,11 @@ def find_undeclared_files(
                         path_base, by_name, by_path_suffix, data_names,
                         by_name_lower,
                     )
+                if not in_repo:
+                    # Hash fallback: the repo may hold the content under a
+                    # different filename (exos21.rom vs exos21.bin)
+                    _lp, _st = resolve_local_file(f, db, dest_hint=dest)
+                    in_repo = _st not in ("not_found",) and _lp is not None
 
             checks = _parse_validation(f.get("validation"))
             undeclared.append(

tests/test_e2e.py

@@ -4885,6 +4885,81 @@ struct BurnDriver BurnDrvneogeo = {
             self.assertIsNone(local)
             self.assertEqual(status, "not_found")
 
+    def test_219_batocera_stable_tag_selection(self):
+        """Stable tag picker takes the highest batocera-N(.M) tag."""
+        from scripts.scraper.batocera_scraper import pick_stable_tag
+
+        tags = [
+            "uboot-rockpro64", "batocera-pre-br2-external",
+            "batocera-41", "batocera-42", "batocera-43", "batocera-43.1",
+            "master_20180518",
+        ]
+        self.assertEqual(pick_stable_tag(tags), "batocera-43.1")
+        self.assertEqual(pick_stable_tag(["batocera-9", "batocera-10"]), "batocera-10")
+        self.assertIsNone(pick_stable_tag(["uboot-x", "random"]))
+
+    def test_218_resolve_by_crc32_and_size(self):
+        """crc-only profile entries resolve by crc32 confirmed by size."""
+        import tempfile
+        import zlib
+
+        from common import resolve_local_file
+
+        with tempfile.TemporaryDirectory() as tmpdir:
+            data = b"exos rom bytes" * 100
+            path = os.path.join(tmpdir, "exos21.rom")
+            with open(path, "wb") as fh:
+                fh.write(data)
+            crc = f"{zlib.crc32(data) & 0xffffffff:08x}"
+            db = {
+                "files": {
+                    "sha1e": {"name": "exos21.rom", "path": path, "size": len(data)},
+                },
+                "indexes": {
+                    "by_md5": {},
+                    "by_name": {"exos21.rom": ["sha1e"]},
+                    "by_crc32": {crc: "sha1e"},
+                    "by_path_suffix": {},
+                },
+            }
+            # Different extension in the profile, crc32+size identify it
+            local, status = resolve_local_file(
+                {"name": "exos21.bin", "crc32": crc, "size": len(data)}, db
+            )
+            self.assertEqual(local, path)
+            self.assertEqual(status, "exact")
+            # Wrong size rejects the crc match
+            local, status = resolve_local_file(
+                {"name": "exos21.bin", "crc32": crc, "size": 1}, db
+            )
+            self.assertIsNone(local)
+
+    def test_217_resolve_name_casefold_fallback(self):
+        """Name resolution falls back to a case-insensitive match."""
+        import tempfile
+
+        from common import resolve_local_file
+
+        with tempfile.TemporaryDirectory() as tmpdir:
+            path = os.path.join(tmpdir, "VEC_Minestorm.vec")
+            with open(path, "wb") as fh:
+                fh.write(b"minestorm rom")
+            db = {
+                "files": {"sha1v": {"name": "VEC_Minestorm.vec", "path": path}},
+                "indexes": {
+                    "by_md5": {},
+                    "by_name": {"VEC_Minestorm.vec": ["sha1v"]},
+                    "by_crc32": {},
+                    "by_path_suffix": {},
+                },
+            }
+            # Upstream declares a different casing, no hash
+            local, status = resolve_local_file(
+                {"name": "VEC_MineStorm.vec"}, db
+            )
+            self.assertEqual(local, path)
+            self.assertEqual(status, "exact")
+
     def test_215_check_member_hash_inside_zip(self):
         """zipped_file entries verify the ROM inside the ZIP, not the ZIP."""
         import hashlib as hl