fix: verify data-dir pack members by content

scripts/packverify.py

@@ -19,6 +19,7 @@ from ziptools import check_inside_zip
 from nativemode import digest_algorithm
 from validation import settle_mismatch
 from nativemode import hash_mismatch_excludes_file
+import functools
 import hashlib
 from common import filter_systems_by_target
 from common import load_emulator_profiles
@@ -35,6 +36,16 @@ import io
 from collections.abc import Callable
 
 
+@functools.lru_cache(maxsize=None)
+def _file_sha1(path: str) -> str:
+    """SHA-1 of a cached data-directory file, read once per run."""
+    digest = hashlib.sha1()
+    with open(path, "rb") as handle:
+        for chunk in iter(lambda: handle.read(1 << 20), b""):
+            digest.update(chunk)
+    return digest.hexdigest()
+
+
 def _members_are_held(data: bytes, by_md5: dict, held_inside: dict) -> bool:
     """Whether every member of an archive is a dump the collection holds."""
     try:
@@ -215,7 +226,10 @@ def verify_pack(
                 for _dp in _cands:
                     if not os.path.exists(_dp):
                         continue
-                    if os.path.getsize(_dp) == size:
+                    # The content, not the size: a member written corrupt
+                    # keeps its length, and a same-sized homonym from another
+                    # cache is another file.
+                    if os.path.getsize(_dp) == size and _file_sha1(_dp) == sha1:
                         status = "verified_data"
                         file_name = _bn
                         break

tests/test_pack_member_identity.py

@@ -43,6 +43,29 @@ class MembersMustBeKnown(unittest.TestCase):
         self.assertTrue(any("garbled.bin" in e for e in manifest["errors"]))
 
 
+class DataMembersAreCheckedByContent(unittest.TestCase):
+    """A data-directory member was verified by name and size: bytes altered
+    while the pack was written kept their length and passed."""
+
+    def test_same_size_other_bytes_is_an_error(self):
+        good = b"A" * 64
+        with tempfile.TemporaryDirectory(dir=REPO_ROOT / "tmp") as tmp:
+            cache = Path(tmp) / "cache"
+            (cache / "Sys").mkdir(parents=True)
+            (cache / "Sys" / "font.bin").write_bytes(good)
+            registry = {"demo": {"local_cache": str(cache)}}
+            db = {"files": {}, "indexes": {"by_md5": {}, "by_name": {}}}
+            pack = Path(tmp) / "P_BIOS_Pack.zip"
+            with zipfile.ZipFile(pack, "w") as zf:
+                zf.writestr("system/Sys/font.bin", good)
+                zf.writestr("system/Sys/other/font.bin", b"B" * 64)
+            ok, manifest = verify_pack(str(pack), db, registry)
+        statuses = {f["path"]: f["status"] for f in manifest["files"]}
+        self.assertEqual(statuses["system/Sys/font.bin"], "verified_data")
+        self.assertNotEqual(statuses["system/Sys/other/font.bin"], "verified_data")
+        self.assertFalse(ok)
+
+
 class SchemaAcceptsEveryStatus(unittest.TestCase):
     """verified_members reached every pack manifest and the schema refused it."""