fix: hash recalbox archives by their members

install/recalbox.json

@@ -238,9 +238,9 @@
     }
   ],
   "total_files": 3378,
-  "total_size": 4262186697,
+  "total_size": 4263070113,
   "pack_files": 3380,
-  "pack_size": 4262186697,
+  "pack_size": 4263070113,
   "total_omitted": 1,
   "omitted_files": [
     {
@@ -5318,10 +5318,10 @@
     },
     {
       "dest": "dc/hod2bios.zip",
-      "sha1": "0b5cc968644c936d4819c1eb5f623f5f3f83eda1",
-      "sha256": "8ae5a99e4ffc15b92e4d590c9e14558743e8f2068947a23e942dfc038ca326c6",
-      "size": 1465991,
-      "repo_path": "bios/Arcade/Arcade/.variants/hod2bios.zip.0b5cc968",
+      "sha1": "9b9c9f0af3de914de677ba8e19f4190673142c1b",
+      "sha256": "c90c754f36d4ab0c927da1f79883cb322a4f94b7cebe9cc7f401151a1b8f1a5a",
+      "size": 1057881,
+      "repo_path": "bios/Arcade/Arcade/hod2bios.zip",
       "cores": null,
       "systems": [
         "sega-dreamcast-arcade"
@@ -5340,10 +5340,10 @@
     },
     {
       "dest": "dc/naomigd.zip",
-      "sha1": "f9ad4a4c6b0bbbe39ba358690a48f763ecbd98f0",
-      "sha256": "1280f4185ac4f7352ab85e49d71db1ca0b3e679c6430bd2c8545088042117eb1",
-      "size": 3371765,
-      "repo_path": "bios/Arcade/Arcade/.variants/naomigd.zip",
+      "sha1": "4b92936f396c8199f745d3ed0ff588a324b8dd70",
+      "sha256": "793db2a859be2d814d334ab511811c42c6727a16841e7627fa7c33171c3de441",
+      "size": 4972898,
+      "repo_path": "bios/Arcade/Arcade/naomigd.zip",
       "cores": null,
       "systems": [
         "sega-dreamcast-arcade"
@@ -5892,10 +5892,10 @@
     },
     {
       "dest": "neocdz.zip",
-      "sha1": "7bb26d1e5d1e930515219cb18bcde5b7b23e2eda",
-      "sha256": "2e93af5848080ea04d17a7841b742f009330d30e4ff40c3410547581d921c892",
-      "size": 524288,
-      "repo_path": "bios/SNK/Neo Geo CD/neocd.bin",
+      "sha1": "4f28af31ca0defdd73d80edec2fa296908e624dc",
+      "sha256": "6aa507a5dc33c45d2592bf2bc77eb3e497762221ddd9225f18583a5c55d38667",
+      "size": 214681,
+      "repo_path": "bios/Arcade/Arcade/.variants/neocdz.zip.4f28af31",
       "cores": null,
       "systems": [
         "snk-neogeo-cd"

platforms/_registry.yml

@@ -415,6 +415,8 @@ platforms:
     source_url: https://gitlab.com/recalbox/recalbox/-/raw/master/board/recalbox/fsoverlay/recalbox/share_init/system/.emulationstation/es_bios.xml
     source_format: xml
     hash_type: md5
+    # Bios.cpp:64-68 hashes a .zip with Zip::Md5Composite, never its bytes
+    zip_digest: composite
     schedule: monthly
     target_scraper: null
     target_source: null

scripts/common.py

@@ -197,6 +197,22 @@ def load_platform_config(platform_name: str, platforms_dir: str = "platforms") -
             if key != "cores" and key not in config:
                 config[key] = val
 
+    # Recalbox hashes a .zip by its members, so a plain md5 never answers one
+    # of its archive declarations: the resolver reads the mark per entry.
+    if config.get("zip_digest") == "composite":
+        config["systems"] = {
+            sys_id: {
+                **system,
+                "files": [
+                    {**f, "zip_digest": "composite"}
+                    if str(f.get("name", "")).lower().endswith(".zip")
+                    else f
+                    for f in system.get("files", [])
+                ],
+            }
+            for sys_id, system in config.get("systems", {}).items()
+        }
+
     _platform_config_cache[cache_key] = config
     return config
 
@@ -848,7 +864,12 @@ def resolve_local_file(
         # .variants/ pattern: filename like "neogeo.zip.fc398ab4"
         return any(bn.startswith(n + ".") for n in _name_set)
 
-    if md5_list and not zipped_file:
+    # A frontend that hashes an archive by its members (Recalbox) never reads
+    # the container's own md5: neocd.bin's md5 is also the composite of a
+    # one-member neocdz.zip, and the raw ROM shipped under the archive's name.
+    composite_only = file_entry.get("zip_digest") == "composite"
+
+    if md5_list and not zipped_file and not composite_only:
         for md5_candidate in md5_list:
             sha1_match = by_md5.get(md5_candidate)
             if sha1_match and sha1_match in files_db:
@@ -1014,6 +1035,8 @@ def resolve_local_file(
 
     if candidates:
         if md5_set and not (sha1_candidates or sha256_candidates or crc_candidates):
+            if composite_only:
+                candidates.sort(key=lambda c: "/.variants/" in c[0])
             for path, db_md5, _size in candidates:
                 if ".zip" in os.path.basename(path):
                     try:
@@ -1022,7 +1045,7 @@ def resolve_local_file(
                             return path, "md5_composite_exact"
                     except (zipfile.BadZipFile, OSError):
                         pass
-                if db_md5.lower() in md5_set:
+                if db_md5.lower() in md5_set and not composite_only:
                     return path, "md5_exact"
         # When zipped_file is set, only accept candidates that contain it
         if zipped_file:

tests/test_archive_choice.py

@@ -218,5 +218,56 @@ class TheArchiveHoldingTheCoreMembers(unittest.TestCase):
                 self.assertNotIn('{"name": archive', source)
 
 
+class AnArchiveHashedByItsMembers(unittest.TestCase):
+    """Recalbox hashes a .zip with Zip::Md5Composite (Bios.cpp:64-68). Its
+    neocdz.zip pin is neocd.bin's md5, which is also the composite of a
+    one-member archive, and the md5 step served the raw ROM under the
+    archive's name: Recalbox cannot open it and reports the BIOS red."""
+
+    def setUp(self):
+        directory = Path(tempfile.mkdtemp(dir=REPO_ROOT / "tmp"))
+        self.addCleanup(shutil.rmtree, directory, True)
+        rom = b"bios" * 8
+        raw = directory / "bios/neocd.bin"
+        archive = directory / "bios/.variants/neocdz.zip.4f28af31"
+        archive.parent.mkdir(parents=True)
+        raw.write_bytes(rom)
+        with zipfile.ZipFile(archive, "w") as zf:
+            zf.writestr("neocd.bin", rom)
+        files, by_md5, by_name = {}, {}, {}
+        for path, name in ((raw, "neocd.bin"), (archive, "neocdz.zip")):
+            data = path.read_bytes()
+            sha1, md5 = hashlib.sha1(data).hexdigest(), hashlib.md5(data).hexdigest()
+            files[sha1] = {"path": str(path), "name": name, "sha1": sha1, "md5": md5}
+            by_md5[md5] = sha1
+            by_name.setdefault(name, []).append(sha1)
+        self.db = {"files": files, "indexes": {"by_name": by_name, "by_md5": by_md5,
+                                               "by_path_suffix": {}, "by_crc32": {}}}
+        self.entry = {"name": "neocdz.zip", "md5": hashlib.md5(rom).hexdigest()}
+        self.archive = str(archive)
+
+    def test_the_archive_answers_a_composite_frontend(self):
+        from common import resolve_local_file
+
+        path, status = resolve_local_file({**self.entry, "zip_digest": "composite"}, self.db)
+        self.assertEqual((path, status), (self.archive, "md5_composite_exact"))
+
+    def test_a_plain_digest_still_takes_the_md5_match(self):
+        from common import resolve_local_file
+
+        path, _status = resolve_local_file(self.entry, self.db)
+        self.assertTrue(path.endswith("neocd.bin"))
+
+    def test_the_registry_marks_recalbox_archives(self):
+        from common import load_platform_config
+
+        config = load_platform_config("recalbox", str(REPO_ROOT / "platforms"))
+        marks = {
+            (f["name"].lower().endswith(".zip"), f.get("zip_digest"))
+            for system in config["systems"].values() for f in system["files"]
+        }
+        self.assertEqual(marks, {(True, "composite"), (False, None)})
+
+
 if __name__ == "__main__":
     unittest.main()