pathlib.Path.exists() on non-existent drive raises WinError instead of returning False

[JordanHueckstaedt]

BPO	35692
Nosy	@pfmoore, @pitrou, @tjguk, @zware, @eryksun, @zooba, @miss-islington
PRs	bpo-35692: pathlib no longer raises when checking file and directory existence #11746 bpo-35692: pathlib no longer raises when checking file and directory existence #11746 bpo-35692: pathlib no longer raises when checking file and directory existence #11746 [3.7] bpo-35692: pathlib no longer raises when checking file and directory existence on drives that are not ready (GH-11746) #11752 [3.7] bpo-35692: pathlib no longer raises when checking file and directory existence on drives that are not ready (GH-11746) #11752 [3.7] bpo-35692: pathlib no longer raises when checking file and directory existence on drives that are not ready (GH-11746) #11752

^{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/zooba'
closed_at = <Date 2019-02-04.07:33:25.544>
created_at = <Date 2019-01-09.03:00:15.061>
labels = ['3.7', '3.8', 'type-bug', 'library', 'OS-windows']
title = 'pathlib.Path.exists() on non-existent drive raises WinError instead of returning False'
updated_at = <Date 2019-02-04.07:33:25.543>
user = 'https://bugs.python.org/JordanHueckstaedt'

bugs.python.org fields:

activity = <Date 2019-02-04.07:33:25.543>
actor = 'steve.dower'
assignee = 'steve.dower'
closed = True
closed_date = <Date 2019-02-04.07:33:25.544>
closer = 'steve.dower'
components = ['Library (Lib)', 'Windows']
creation = <Date 2019-01-09.03:00:15.061>
creator = 'Jordan Hueckstaedt'
dependencies = []
files = []
hgrepos = []
issue_num = 35692
keywords = ['patch', 'patch', 'patch']
message_count = 6.0
messages = ['333275', '333724', '333729', '333748', '334811', '334814']
nosy_count = 8.0
nosy_names = ['paul.moore', 'pitrou', 'tim.golden', 'zach.ware', 'eryksun', 'steve.dower', 'miss-islington', 'Jordan Hueckstaedt']
pr_nums = ['11746', '11746', '11746', '11752', '11752', '11752']
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue35692'
versions = ['Python 3.7', 'Python 3.8']

[JordanHueckstaedt]

Tested in 3.7.0 on windows 10.
This looks related to https://bugs.python.org/issue22759

>>> import pathlib
>>> pathlib.Path(r"E:\Whatever\blah.txt").exists()  # This drive doesn't exist
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "C:\Users\jordanhu\AppData\Local\Continuum\anaconda2\envs\py3\lib\pathlib.py", line 1318, in exists
    self.stat()
  File "C:\Users\jordanhu\AppData\Local\Continuum\anaconda2\envs\py3\lib\pathlib.py", line 1140, in stat
    return self._accessor.stat(self)
PermissionError: [WinError 21] The device is not ready: 'E:\\Whatever\\blah.txt'
>>> pathlib.Path(r"C:\Whatever\blah.txt").exists() # This drive exists
False

[zooba]

In bpo-22759 there was some logic applied for which errors to forward rather than hide.

I'm inclined to agree that this one should be hidden, but it may have to be done by checking the winerror field rather than the exception type, since other PermissionErrors may mean the file is guaranteed to exist (but you can't touch it) or that the path exists up to the point where you are not allowed to see.

I'd happily argue that since these permissions indicate that the file does not exist *for the current user* and so they should be swallowed more broadly, but I'll let Antoine make the call.

[pitrou]

I think exists() should simply return False here. There's no reason a non-existing drive should fail differently than a non-existing parent directory.

[eryksun]

There's no reason a non-existing drive should fail differently than
a non-existing parent directory.

The drive exists (or should) if we're getting ERROR_NOT_READY (21). It's likely a removable media device, such as an optical disc or card reader, and there's no media in the device.

If a logical drive isn't defined at all, we should get ERROR_PATH_NOT_FOUND (from the NT status value STATUS_OBJECT_PATH_NOT_FOUND). This gets mapped to the errno value ENOENT, which is already handled. For example:

    >>> os.stat('Q:/')
    Traceback (most recent call last):
      File "<stdin>", line 1, in <module>
    FileNotFoundError: [WinError 3] The system cannot find the path specified: 'Q:/'

    >>> pathlib.Path('Q:/whatever/blah.txt').exists()
    False

Similarly if a UNC 'drive' doesn't exist, we should get ERROR_BAD_NET_NAME (from NT STATUS_BAD_NETWORK_NAME), which is also mapped to ENOENT:

    >>> os.stat('//some/where')
    Traceback (most recent call last):
      File "<stdin>", line 1, in <module>
    FileNotFoundError: [WinError 67] The network name cannot be found: '//some/where'

    >>> pathlib.Path('//some/where/whatever/blah.txt').exists()
    False

[zooba]

New changeset 2f6fae6 by Steve Dower in branch 'master':
bpo-35692: pathlib no longer raises when checking file and directory existence on drives that are not ready (GH-11746)
2f6fae6

[miss-islington]

New changeset 69af439 by Miss Islington (bot) in branch '3.7':
bpo-35692: pathlib no longer raises when checking file and directory existence on drives that are not ready (GH-11746)
69af439