Windows pathlib.Path.glob(pattern) fixed part of the pattern changed to lowercase whereas it should be unchanged.

@pfmoore

BPO	31202
Nosy	@pfmoore, @tjguk, @zware, @serhiy-storchaka, @zooba, @brice-gros, @miss-islington, @iritkatriel
PRs	bpo-31202: Windows pathlib.Path.glob(pattern) fixed part of the pattern changed to lowercase whereas it should be unchanged. #3087 bpo-31202: Preserve case of literal parts in Path.glob() on Windows. #16860 [3.7] bpo-31202: Preserve case of literal parts in Path.glob() on Windows. (GH-16860) #16874 [3.8] bpo-31202: Preserve case of literal parts in Path.glob() on Windows. (GH-16860) #16875

^{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 = None
closed_at = <Date 2020-10-17.00:11:35.555>
created_at = <Date 2017-08-14.13:57:31.993>
labels = ['type-bug', '3.8', '3.9', '3.7', 'library', 'OS-windows']
title = 'Windows pathlib.Path.glob(pattern) fixed part of the pattern changed to lowercase whereas it should be unchanged.'
updated_at = <Date 2020-10-17.00:11:35.552>
user = 'https://github.com/brice-gros'

bugs.python.org fields:

activity = <Date 2020-10-17.00:11:35.552>
actor = 'steve.dower'
assignee = 'none'
closed = True
closed_date = <Date 2020-10-17.00:11:35.555>
closer = 'steve.dower'
components = ['Library (Lib)', 'Windows']
creation = <Date 2017-08-14.13:57:31.993>
creator = 'brice.gros'
dependencies = []
files = []
hgrepos = []
issue_num = 31202
keywords = ['patch']
message_count = 8.0
messages = ['300245', '354976', '355087', '355093', '355095', '378786', '378792', '378794']
nosy_count = 8.0
nosy_names = ['paul.moore', 'tim.golden', 'zach.ware', 'serhiy.storchaka', 'steve.dower', 'brice.gros', 'miss-islington', 'iritkatriel']
pr_nums = ['3087', '16860', '16874', '16875']
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue31202'
versions = ['Python 3.7', 'Python 3.8', 'Python 3.9']

Windows pathlib.Path.glob(pattern) fixed part of the pattern to lowercase whereas it should be unchanged.
Note that this issue is different from http://bugs.python.org/issue26655 : "pathlib glob case sensitivity issue on Windows" where it was asked to get the actual case of the folder from the file system.

Assuming a directory contains a folder named 'Folder'.
On Windows, calling pathlib.Path().glob('Folder') gives 'folder', but 'Folde?' will return 'Folder'
This is an issue for instance if trying to glob files to put them in an archive to be sent to a case sensitive platform.
glob.glob() does behave properly though, Windows pathlib.Path is the only platform which has such a behavior.

I would expect Path.glob to output the same as glob.glob() for each platform.
From comments on http://bugs.python.org/issue19718 : "Path.glob() on case-insensitive Posix filesystems" it sounds that it should even match the native shell behavior.
And it looks like it is the case for linux and macOS, I tested that with the following script, whose results on win32, darwin and linux platforms follow:

#!/usr/bin/env python3.6
# Let's say this path exists : ./Folder/file
from pathlib import Path
import glob
import os
import sys
import subprocess

def ls(pattern):
    if sys.platform in ('win32', 'win64'):
        process = subprocess.run(['powershell', '-Command', f'dir {pattern}'], stdout=subprocess.PIPE, stderr=subprocess.STDOUT, cwd=os.getcwd(), encoding='utf-8')
        if process.returncode:
            return []
        # expected output:
        # 
        # 
        #     Directory: C:\path_to\Folder
        # 
        # 
        # Mode                LastWriteTime         Length Name
        # ----                -------------         ------ ----
        # -a----       2017-08-14     10:16              0 file
        lines = process.stdout.splitlines()
        folder = os.path.basename(lines[2].split()[-1])
        file = lines[7].split()[-1]
        result = f"{folder}{os.path.sep}{file}"
        return [result]
    else:
        cmd = ['ls', f'{pattern}']
        process = subprocess.run(' '.join(cmd), stdout=subprocess.PIPE, stderr=subprocess.PIPE, cwd=os.getcwd(), encoding='utf-8', shell=True)
        if process.returncode:
            return []
        return [process.stdout.strip()]

def main():
    print(sys.platform)
    p = Path('.')
    tests = ['Folder/*', 'FOlder/*', 'F?lder/*', 'FOlde?/*', 'folder/*', 'f?lder/*']
    for t in tests:
        print(f'{t}:')
        print(f'    Path.glob():  {[str(f) for f in p.glob(t)]}')
        print(f'    glob.glob():  {[f for f in glob.glob(str(p/t))]}')
        print(f'    shell:        {ls(str(p/t))}')

if __name__ == '__main__':
    main()

Output:
        win32                                            darwin                                        linux
1:      Folder/*:                                        Folder/*:                                     Folder/*:
1a:         Path.glob():  ['folder\\file']                   Path.glob():  ['Folder/file']                 Path.glob():  ['Folder/file']
1b:         glob.glob():  ['Folder\\file']                   glob.glob():  ['Folder/file']                 glob.glob():  ['Folder/file']
1c:         shell:        ['Folder\\file']                   shell:        ['Folder/file']                 shell:        ['Folder/file']
2:      FOlder/*:                                        FOlder/*:                                     FOlder/*:
2a:         Path.glob():  ['folder\\file']                   Path.glob():  ['FOlder/file']                 Path.glob():  []
2b:         glob.glob():  ['FOlder\\file']                   glob.glob():  ['FOlder/file']                 glob.glob():  []
2c:         shell:        ['FOlder\\file']                   shell:        ['FOlder/file']                 shell:        []
3:      F?lder/*:                                        F?lder/*:                                     F?lder/*:
3a:         Path.glob():  ['Folder\\file']                   Path.glob():  ['Folder/file']                 Path.glob():  ['Folder/file']
3b:         glob.glob():  ['Folder\\file']                   glob.glob():  ['Folder/file']                 glob.glob():  ['Folder/file']
3c:         shell:        ['Folder\\file']                   shell:        ['Folder/file']                 shell:        ['Folder/file']
4:      FOlde?/*:                                        FOlde?/*:                                     FOlde?/*:
4a:         Path.glob():  ['Folder\\file']                   Path.glob():  []                              Path.glob():  []
4b:         glob.glob():  ['Folder\\file']                   glob.glob():  []                              glob.glob():  []
4c:         shell:        ['Folder\\file']                   shell:        []                              shell:        []
5:      folder/*:                                        folder/*:                                     folder/*:
5a:         Path.glob():  ['folder\\file']                   Path.glob():  ['folder/file']                 Path.glob():  []
5b:         glob.glob():  ['folder\\file']                   glob.glob():  ['folder/file']                 glob.glob():  []
5c:         shell:        ['folder\\file']                   shell:        ['folder/file']                 shell:        []
6:      f?lder/*:                                        f?lder/*:                                     f?lder/*:
6a:         Path.glob():  ['Folder\\file']                   Path.glob():  []                              Path.glob():  []
6b:         glob.glob():  ['Folder\\file']                   glob.glob():  []                              glob.glob():  []
6c:         shell:        ['Folder\\file']                   shell:        []                              shell:        []

Bumping this - it's bitten me a couple of times as one of the build/release scripts relies on Path.glob().

New changeset 10ecbad by Serhiy Storchaka in branch 'master':
bpo-31202: Preserve case of literal parts in Path.glob() on Windows. (GH-16860)
10ecbad

New changeset 175abcc by Miss Skeleton (bot) in branch '3.7':
bpo-31202: Preserve case of literal parts in Path.glob() on Windows. (GH-16860)
175abcc

New changeset 2f8d4f0 by Miss Skeleton (bot) in branch '3.8':
bpo-31202: Preserve case of literal parts in Path.glob() on Windows. (GH-16860)
2f8d4f0

Can this be closed?

I think so, the original PR has been ported to 3.9, 3.8, 3.7, and all the PRs were merged.

Yeah, looks done.

Navigation Menu

Repository navigation

Windows pathlib.Path.glob(pattern) fixed part of the pattern changed to lowercase whereas it should be unchanged. #75385

Metadata

Assignees

Labels

Projects

Milestone

Relationships

Development

Windows pathlib.Path.glob(pattern) fixed part of the pattern changed to lowercase whereas it should be unchanged. #75385

Description

Activity

brice-gros commented on Aug 14, 2017

zooba commented on Oct 19, 2019

serhiy-storchaka commented on Oct 21, 2019

miss-islington commented on Oct 21, 2019

miss-islington commented on Oct 21, 2019

iritkatriel commented on Oct 16, 2020

brice-gros commented on Oct 16, 2020

zooba commented on Oct 17, 2020

Metadata

Metadata

Assignees

Labels

Projects

Milestone

Relationships

Development

Issue actions