Pathlib .owner() and .group() methods fail on broken links

[inigoserna]

BPO	21322
Nosy	@pitrou
Files	pathlib-34-owner_group_fixed.diff

^{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 2017-11-09.18:11:28.341>
created_at = <Date 2014-04-21.16:42:49.821>
labels = ['library', 'type-crash']
title = 'Pathlib .owner() and .group() methods fail on broken links'
updated_at = <Date 2017-11-09.18:11:28.340>
user = 'https://bugs.python.org/inigoserna'

bugs.python.org fields:

activity = <Date 2017-11-09.18:11:28.340>
actor = 'serhiy.storchaka'
assignee = 'none'
closed = True
closed_date = <Date 2017-11-09.18:11:28.341>
closer = 'serhiy.storchaka'
components = ['Library (Lib)']
creation = <Date 2014-04-21.16:42:49.821>
creator = 'inigoserna'
dependencies = []
files = ['34992']
hgrepos = []
issue_num = 21322
keywords = ['patch']
message_count = 4.0
messages = ['216950', '216971', '216995', '216996']
nosy_count = 2.0
nosy_names = ['pitrou', 'inigoserna']
pr_nums = []
priority = 'normal'
resolution = 'rejected'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'crash'
url = 'https://bugs.python.org/issue21322'
versions = ['Python 3.4']

[inigoserna]

Pathlib .owner() and .group() methods fail on broken symlinks.

They use:
return pwd.getpwuid(self.stat().st_uid).pw_name
and:
return grp.getgrgid(self.stat().st_gid).gr_name

It should be self.lstat().

Attached simple fix as unified diff.

[pitrou]

Most other Path methods operate on the link target, not the link itself, so I don't see why these methods would work otherwise.

[inigoserna]

Mainly, 2 reasons:

• It can make programs crash *unexpectedly*

• Pathlib should provide a complete and uniform API for dealing with all types of files. If not, users would need to use Pathlib for some kind of files and go to os and os.path for others, then why the reason to use Pathlib?

[pitrou]

• It can make programs crash *unexpectedly*

A broken link is an error, so it's normal to have an exception raised
here. An exception can always be caught if you were expecting the error.

• Pathlib should provide a complete and uniform API for dealing with
  all types of files. If not, users would need to use Pathlib for some
  kind of files and go to os and os.path for others, then why the reason
  to use Pathlib?

I sympathize with this. But you can already use Path.lstat() and inspect
the st_uid and st_gid fields if you are really interested in the link's
owner and group (rather than the target's).

The fact that most Path methods dereference symlinks reflects the
semantics of other common filesystem calls (such as those in the os
module, or the underlying functions of the POSIX API).