From 53625156d6b6e7c763f11e23427095e7f5bb7bf5 Mon Sep 17 00:00:00 2001
From: Abdessamad Derraz <3028866+Abdess@users.noreply.github.com>
Date: Sat, 17 Oct 2026 08:52:21 +0200
Subject: [PATCH] fix: key the zip member index on all its inputs

---
 scripts/ziptools.py             | 10 ++++++++--
 tests/test_audit_regressions.py | 23 +++++++++++++++++++++++
 2 files changed, 31 insertions(+), 2 deletions(-)

diff --git a/scripts/ziptools.py b/scripts/ziptools.py
index fb848988..c5a8cc5c 100644
--- a/scripts/ziptools.py
+++ b/scripts/ziptools.py
@@ -57,7 +57,7 @@ def zip_member_crcs(path: str) -> frozenset[str]:
     return _member_crcs(path, mtime)
 
 
-_zip_contents_cache: tuple[frozenset[tuple[str, float]], dict] | None = None
+_zip_contents_cache: tuple[tuple, dict] | None = None
 
 def build_zip_contents_index(db: dict, max_entry_size: int = 512 * 1024 * 1024) -> dict:
     """Build {inner_rom_md5: zip_file_sha1} for ROMs inside ZIP files.
@@ -74,7 +74,13 @@ def build_zip_contents_index(db: dict, max_entry_size: int = 512 * 1024 * 1024)
         if path.endswith(".zip") and os.path.exists(path):
             zip_entries.append((path, sha1))
 
-    fingerprint = frozenset((path, os.path.getmtime(path)) for path, _ in zip_entries)
+    # The bound and the sha1 each path carries are part of the answer: a
+    # second call with a smaller bound got members the first one admitted,
+    # and a regenerated database got the sha1 of the old one.
+    fingerprint = (
+        max_entry_size,
+        frozenset((path, os.path.getmtime(path), sha1) for path, sha1 in zip_entries),
+    )
     if _zip_contents_cache is not None and _zip_contents_cache[0] == fingerprint:
         return _zip_contents_cache[1]
 
diff --git a/tests/test_audit_regressions.py b/tests/test_audit_regressions.py
index a0da649d..9e19e73c 100644
--- a/tests/test_audit_regressions.py
+++ b/tests/test_audit_regressions.py
@@ -2448,6 +2448,29 @@ class AnUnreadableDatabaseStopsTheScan(unittest.TestCase):
                 )
 
 
+class TheZipIndexCacheKnowsItsInputs(unittest.TestCase):
+    """build_zip_contents_index cached on (path, mtime) alone: a smaller
+    max_entry_size, or a database giving the same archive another sha1,
+    got the first call's index back."""
+
+    def test_the_bound_and_the_sha1_are_part_of_the_key(self):
+        from scripts import ziptools
+
+        with tempfile.TemporaryDirectory(dir=TMP_ROOT) as directory:
+            path = Path(directory) / "a.zip"
+            with zipfile.ZipFile(path, "w") as archive:
+                archive.writestr("big.bin", b"x" * 4096)
+            ziptools._zip_contents_cache = None
+            db = {"files": {"a" * 40: {"path": str(path)}}}
+            self.assertEqual(len(ziptools.build_zip_contents_index(db)), 1)
+            self.assertEqual(ziptools.build_zip_contents_index(db, max_entry_size=10), {})
+            other = {"files": {"b" * 40: {"path": str(path)}}}
+            self.assertEqual(
+                set(ziptools.build_zip_contents_index(other).values()), {"b" * 40}
+            )
+            ziptools._zip_contents_cache = None
+
+
 class TheDatabaseCacheIsWrittenWhole(unittest.TestCase):
     """A crash while serializing left a truncated db_cache.json behind, and
     the next run rehashed the whole collection from scratch."""