Bug report

Bug description:

When calling ZipFile.mkdir(), the user provided mode does not have the file format bits cleared, which may cause the created entry be recognized as an incorrect or unknown file format.

import io
import stat
import zipfile

with zipfile.ZipFile(io.BytesIO(), 'w') as zh:
    zh.mkdir('foo/', 0o107777)  # regular file
    zinfo = zh.getinfo('foo/')
    print(oct(stat.S_IFMT(zinfo.external_attr >> 16)))  # 0o140000 (socket file)

with zipfile.ZipFile(io.BytesIO(), 'w') as zh:
    zh.mkdir('foo/', 0o127777)  # symbolic link
    zinfo = zh.getinfo('foo/')
    print(oct(stat.S_IFMT(zinfo.external_attr >> 16)))  # 0o160000 (unknown)

Additionally, the current source code and doc for ZipFile.mkdir()'s default mode value are 511, which is identical to 0o777 but less intuitive.

Suggestion

Instead of the current sanitization:

zinfo.external_attr = ((0o40000 | mode) & 0xFFFF) << 16

we should probably sanitize the provided mode as what stat.S_IMODE does, i.e.:

zinfo.external_attr = (0o40000 | (mode & 0o7777)) << 16

We should probably also revise the source code and doc for ZipFile.mkdir()'s default mode value to 0o777, to be consistent with os.mkdir() and more intuitive.

CPython versions tested on:

3.14

Operating systems tested on:

No response

Linked PRs

• gh-154490: Only keep permission bits of the mode argument in ZipFile.mkdir() #154509