On Windows os.listdir('') -> cwd and os.listdir(u'') -> C:\

[ezio-melotti]

BPO	5913
Nosy	@loewis, @pitrou, @ezio-melotti, @bitdancer
Files	listdir_on_empty_path.patch

^{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 2009-05-04.05:59:55.596>
created_at = <Date 2009-05-03.13:44:59.874>
labels = ['type-bug', 'library', 'OS-windows']
title = "On Windows os.listdir('') -> cwd and os.listdir(u'') -> C:\\"
updated_at = <Date 2009-05-04.06:21:53.091>
user = 'https://github.com/ezio-melotti'

bugs.python.org fields:

activity = <Date 2009-05-04.06:21:53.091>
actor = 'ezio.melotti'
assignee = 'ocean-city'
closed = True
closed_date = <Date 2009-05-04.05:59:55.596>
closer = 'ocean-city'
components = ['Library (Lib)', 'Windows']
creation = <Date 2009-05-03.13:44:59.874>
creator = 'ezio.melotti'
dependencies = []
files = ['13861']
hgrepos = []
issue_num = 5913
keywords = ['patch']
message_count = 9.0
messages = ['87045', '87054', '87056', '87080', '87094', '87095', '87097', '87099', '87101']
nosy_count = 5.0
nosy_names = ['loewis', 'pitrou', 'ocean-city', 'ezio.melotti', 'r.david.murray']
pr_nums = []
priority = 'high'
resolution = 'fixed'
stage = 'test needed'
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue5913'
versions = ['Python 3.1', 'Python 2.7']

[ezio-melotti]

On Windows, with Python2/Python3, os.listdir('')/os.listdir(b'') list
the content of the current working directory and
os.listdir(u'')/os.listdir('') the content of C:\.

On Linux the error "OSError: [Errno 2] No such file or directory: ''" is
raised.

I also noticed that os.listdir('C:') (without the \) lists the content
of the cwd too.

[pitrou]

As for the "C:" behaviour, I think it is normal: you are not specifying
the path itself, only the drive letter, so it uses the current path in
the specified drive (which /is/ cwd if you are already running from C:,
but may be something else if running from another drive).

As for '' and b'', I think that ambiguity is bad and these values should
be prohibited by Python (ValueError comes to mind).

[loewis]

Before anything is changed, I would first like to understand where the
difference comes from.

If it gives ENOENT on Unix, it should also give ENOENT on Windows
(rather than giving ValueError). In addition, it should also give what
open("")/open(b"") gives - as it really is the request to open the
directory named "". (disclaimer: I haven done any tests or code review yet)

[bitdancer]

On Sun, 3 May 2009 at 17:36, Antoine Pitrou wrote:

As for the "C:" behaviour, I think it is normal: you are not specifying
the path itself, only the drive letter, so it uses the current path in
the specified drive (which /is/ cwd if you are already running from C:,
but may be something else if running from another drive).

It is normal behavior for Windows. Try it with the 'dir' command
in a command window. (That said, I have no idea, not being a
Windows user, how Windows decides what the 'current' directory
is on any given drive from any given context).

[ocean-city]

I hope attached patch works.

>>> import os
>>> os.listdir("")
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
WindowsError: [Error 3] 指定されたパスが見つかりません。: ''
[36200 refs]
>>> os.listdir(u"")
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
WindowsError: [Error 3] 指定されたパスが見つかりません。: u''

# Error message says "Cannot find specified path"

[ocean-city]

For u"", os.listdir calls FindFirstFileW with u"\\*.*",
For "", os.listdir calls FindFirstFileA with "*.*".

The code before FindFirstFile[AW] is slightly different for empty path.

[loewis]

Hirokazu, the patch looks fine, please apply to 2.7 and 3k.

I wouldn't backport it to 2.6/3.0, since it may break existing code.

[ocean-city]

Committed in r72273(trunk) and r72274(py3k).

[ezio-melotti]

It is normal behavior for Windows. Try it with the 'dir' command
in a command window. (That said, I have no idea, not being a
Windows user, how Windows decides what the 'current' directory
is on any given drive from any given context).

Yes, I tried to run cmd.exe and I did:

C:\> cd D:\foo
C:\> cd E:\foo
C:\> python

os.listdir('D:') and os.listdir('E:') then listed the content of D:\foo
and E:\bar. This is the correct behaviour, even if it's not documented.
The CWD is the one where Python has been launched, and apparently on the
other drives is the root folder by default, if it's not changed with the
cd command.

Thanks Hirokazu for the patch.