https://github.com/python/cpython/commit/fc2f9137fe913cc5600b913e3d6afa3344a8baa2 commit: fc2f9137fe913cc5600b913e3d6afa3344a8baa2 branch: 3.14 author: Miss Islington (bot) <[email protected]> committer: ambv <[email protected]> date: 2026-11-13T17:31:43-07:00 summary:
[3.14] gh-152845: Preserve ZIP encoding flags when appending (gh-161950) (gh-161961) (cherry picked from commit 5151e7f2a3dabf74941f2c3857e17886d87cb5dd) Co-authored-by: Kasra Ghodsi <[email protected]> files: A Misc/NEWS.d/next/Library/2026-11-13-09-49-13.gh-issue-152845.ZipFlags.rst M Lib/test/test_zipfile/test_core.py M Lib/zipfile/__init__.py diff --git a/Lib/test/test_zipfile/test_core.py b/Lib/test/test_zipfile/test_core.py index ee93e27f2bff8f..08ec019c12eb20 100644 --- a/Lib/test/test_zipfile/test_core.py +++ b/Lib/test/test_zipfile/test_core.py @@ -2371,6 +2371,66 @@ def test_empty_zipfile(self): self.fail("Unable to create empty ZIP file in 'a' mode") zipf.close() + def test_append_preserves_ascii_utf8_flag(self): + comments = (b'', 'caf\N{LATIN SMALL LETTER E WITH ACUTE}'.encode('utf-8')) + updates = ( + ('append', lambda zf: zf.writestr('added.txt', b'added')), + ('comment', lambda zf: setattr(zf, 'comment', b'archive comment')), + ) + for comment, (operation, update) in itertools.product(comments, updates): + with self.subTest(comment=comment, operation=operation): + stream = io.BytesIO() + with zipfile.ZipFile(stream, 'w') as archive: + entry = zipfile.ZipInfo('file.txt') + entry.comment = comment + archive.writestr(entry, b'payload') + original = bytearray(stream.getvalue()) + central_offset = original.index(b'PK\x01\x02') + struct.pack_into('<H', original, 6, 1 << 11) + struct.pack_into('<H', original, central_offset + 8, 1 << 11) + stream = io.BytesIO(original) + with zipfile.ZipFile(stream, 'a') as archive: + update(archive) + with zipfile.ZipFile(stream) as archive: + entry = archive.getinfo('file.txt') + self.assertEqual(entry.flag_bits, 1 << 11) + self.assertEqual(entry.comment, comment) + self.assertEqual(archive.read('file.txt'), b'payload') + self.assertIsNone(archive.testzip()) + self.assertEqual(stream.getvalue()[:central_offset], + original[:central_offset]) + + def test_new_entry_filename_encoding_flags(self): + cases = ( + ('file.txt', b'\x82', 0, 'cp437', 0, + '\N{LATIN SMALL LETTER E WITH ACUTE}'), + ('caf\N{LATIN SMALL LETTER E WITH ACUTE}.txt', b'\xc3\xa9', + 0, 'utf-8', 1 << 11, + '\N{LATIN SMALL LETTER E WITH ACUTE}'), + ('file.txt', b'\xc3\xa9', 1 << 11, 'utf-8', 1 << 11, + '\N{LATIN SMALL LETTER E WITH ACUTE}'), + ('file.txt', b'', 1 << 11, 'utf-8', 1 << 11, ''), + ('file.txt', b'ascii', 1 << 11, 'utf-8', 1 << 11, 'ascii'), + ) + for (name, comment, initial_flags, encoding, expected_flags, + expected_comment) in cases: + with self.subTest(name=name, comment=comment, + initial_flags=initial_flags): + stream = io.BytesIO() + with zipfile.ZipFile(stream, 'w') as archive: + entry = zipfile.ZipInfo(name) + entry.flag_bits = initial_flags + entry.comment = comment + archive.writestr(entry, b'payload') + with zipfile.ZipFile(stream) as archive: + entry = archive.getinfo(name) + self.assertEqual(entry.flag_bits, expected_flags) + self.assertEqual(entry.comment, comment) + self.assertEqual(entry.comment.decode(encoding), + expected_comment) + self.assertEqual(archive.read(name), b'payload') + self.assertIsNone(archive.testzip()) + def test_open_empty_file(self): # Issue 1710703: Check that opening a file with less than 22 bytes # raises a BadZipFile exception (rather than the previously unhelpful diff --git a/Lib/zipfile/__init__.py b/Lib/zipfile/__init__.py index d4b104056115f8..27e2deceb3eb82 100644 --- a/Lib/zipfile/__init__.py +++ b/Lib/zipfile/__init__.py @@ -568,11 +568,11 @@ def FileHeader(self, zip64=None): def _encodeFilenameFlags(self): if self.flag_bits & _MASK_UTF_FILENAME: - encoding = 'ascii' + encoding = 'utf-8' else: encoding = 'cp437' try: - return self.filename.encode(encoding), self.flag_bits & ~_MASK_UTF_FILENAME + return self.filename.encode(encoding), self.flag_bits except UnicodeEncodeError: return self.filename.encode('utf-8'), self.flag_bits | _MASK_UTF_FILENAME @@ -1842,7 +1842,9 @@ def _open_to_write(self, zinfo, force_zip64=False): zinfo.compress_size = 0 zinfo.CRC = 0 - zinfo.flag_bits = _MASK_UTF_FILENAME + zinfo.flag_bits &= _MASK_UTF_FILENAME + if not zinfo.filename.isascii(): + zinfo.flag_bits |= _MASK_UTF_FILENAME if zinfo.compress_type == ZIP_LZMA: # Compressed data includes an end-of-stream (EOS) marker zinfo.flag_bits |= _MASK_COMPRESS_OPTION_1 diff --git a/Misc/NEWS.d/next/Library/2026-11-13-09-49-13.gh-issue-152845.ZipFlags.rst b/Misc/NEWS.d/next/Library/2026-11-13-09-49-13.gh-issue-152845.ZipFlags.rst new file mode 100644 index 00000000000000..70cf5134d3fe9f --- /dev/null +++ b/Misc/NEWS.d/next/Library/2026-11-13-09-49-13.gh-issue-152845.ZipFlags.rst @@ -0,0 +1,3 @@ +Fix :mod:`zipfile` dropping the UTF-8 (EFS) flag for existing entries +when rewriting an archive in append mode, and honor an explicitly set +flag on new entries. Patch by Kasra Ghodsi. _______________________________________________ Python-checkins mailing list -- [email protected] To unsubscribe send an email to [email protected] https://mail.python.org/mailman3//lists/python-checkins.python.org Member address: [email protected]
