pathlib.Path.glob() does not list dangling symlink when pattern is the exact filename

[raek]

BPO	45606
Nosy	@Fidget-Spinner, @akulakov, @raek
PRs	GH-89769: Pathlib - do not follow links when checking for precise glob match #29655

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = 'https://github.com/akulakov'
closed_at = None
created_at = <Date 2021-10-25.19:20:55.106>
labels = ['type-bug', 'library', '3.9', '3.10', '3.11']
title = 'pathlib.Path.glob() does not list dangling symlink when pattern is the exact filename'
updated_at = <Date 2022-01-23.06:57:12.913>
user = 'https://github.com/raek'

bugs.python.org fields:

activity = <Date 2022-01-23.06:57:12.913>
actor = 'andrei.avk'
assignee = 'andrei.avk'
closed = False
closed_date = None
closer = None
components = ['Library (Lib)']
creation = <Date 2021-10-25.19:20:55.106>
creator = 'raek'
dependencies = []
files = []
hgrepos = []
issue_num = 45606
keywords = ['patch']
message_count = 4.0
messages = ['404996', '406643', '406644', '406645']
nosy_count = 3.0
nosy_names = ['kj', 'andrei.avk', 'raek']
pr_nums = ['29655']
priority = 'normal'
resolution = None
stage = 'patch review'
status = 'open'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue45606'
versions = ['Python 3.9', 'Python 3.10', 'Python 3.11']

[raek]

Create a symlink that points to file that doesn't exist:

ln -s /nonexisting_file my_symlink

Then try to glob for that symlink from Python using pathlib:

    python3
    >>> import pathlib
    >>> list(pathlib.Path(".").glob("my_symlink"))
    []
    >>> list(pathlib.Path(".").glob("my_symlink*"))
    [PosixPath('my_symlink')]

I'm a bit surprised that these two globs do not return the same results. Personally I would expect both to find the symlink.

Is this behaviour a bug or is it intentional?

[akulakov]

The issue is that _PreciseSelector follows the symlink when it checks if a path exists before yielding it as a result.

I've put up a PR with a fix; I've also added a follow_symlinks arg to exists() method because it seems more logical to be able to test if a path exists via the same method rather than having to also remember and to check for it being a symlink.

I will add docs and news a bit later today or tomorrow.

[akulakov]

Rasmus: thanks for the report, it does seem like a bug to me.

[akulakov]

By the way note that path.glob('**/my_symlink') also does return the dangling symlink match. And glob.glob('my_symlink') also returns a dangling symlink.

[barneygale]

Fixed in 3.12: #29655 / af886ff