backend: Handle corrupt zip files

Corrupt zip files would already make hircine throw an error, but
depending on the exact problem it would not report which file (or even
which entry in a zip file) is affected. Use ZipFile.testzip() to catch
common problems and make sure to re-raise any exception within as a
BadZipFile exception. This makes sure to also report decompression
problems that are raised as a zlib.error exception, for example.

1 files changed, 6 insertions, 1 deletions

diff --git a/src/hircine/scanner.py b/src/hircine/scanner.py
index 29ae04f..6e3fafb 100644
--- a/src/hircine/scanner.py
+++ b/src/hircine/scanner.py
@@ -8,7 +8,7 @@ from datetime import datetime, timezone
 from enum import Enum
 from hashlib import file_digest
 from typing import NamedTuple
-from zipfile import ZipFile, is_zipfile
+from zipfile import BadZipFile, ZipFile, is_zipfile
 
 from blake3 import blake3
 from natsort import natsorted, ns
@@ -286,6 +286,11 @@ class Scanner:
         hash = blake3()
 
         with ZipFile(path, mode="r") as z:
+            try:
+                z.testzip()
+            except Exception as e:
+                raise BadZipFile(f"Corrupt zip file {path}") from e
+
             input = [(path, info.filename) for info in z.infolist()]
 
         loop = asyncio.get_event_loop()