pathlib's Path.mkdir() should allow for "mkdir -p" functionality

[garrison]

BPO	21539
Nosy	@warsaw, @pitrou, @berkerpeksag
Files	issue21539.diff issue21539_v2.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 = 'https://github.com/warsaw'
closed_at = <Date 2014-08-05.15:29:26.112>
created_at = <Date 2014-05-20.03:53:56.770>
labels = ['type-feature', 'library']
title = 'pathlib\'s Path.mkdir() should allow for "mkdir -p" functionality'
updated_at = <Date 2014-08-05.18:29:38.923>
user = 'https://bugs.python.org/garrison'

bugs.python.org fields:

activity = <Date 2014-08-05.18:29:38.923>
actor = 'berker.peksag'
assignee = 'barry'
closed = True
closed_date = <Date 2014-08-05.15:29:26.112>
closer = 'barry'
components = ['Library (Lib)']
creation = <Date 2014-05-20.03:53:56.770>
creator = 'garrison'
dependencies = []
files = ['35303', '35421']
hgrepos = []
issue_num = 21539
keywords = ['patch']
message_count = 8.0
messages = ['218832', '218842', '219233', '219238', '219239', '219468', '224839', '224842']
nosy_count = 5.0
nosy_names = ['barry', 'pitrou', 'python-dev', 'berker.peksag', 'garrison']
pr_nums = []
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'enhancement'
url = 'https://bugs.python.org/issue21539'
versions = ['Python 3.5']

[garrison]

As of Python 3.2, os.makedirs() takes exist_ok as a keyword argument. If set to true, the function does not raise an error if the directory already exists. This makes the function's behavior similar to mkdir -p on the UNIX commandline. (See https://docs.python.org/3/library/os.html#os.makedirs)

However, this functionality is missing in Python's new pathlib. In fact, the documentation for Path.mkdir() even mentions that if parent=True is passed, then it should behave like mkdir -p. This is accurate in that it indeed recursively makes any parent directories necessary, but it is inaccurate in that it raises an error if the directory already exists.

I propose that either parents=True should imply that there is no error if the directory already exists (which would be a backwards-incompatible change); or, there could be a new exist_ok=True, which could suppress the error for an existing directory. Either way, it ought to be possible to get the mkdir -p functionality easily from pathlib. And the documentation for Path.mkdir() should be updated to explain how to get this mkdir -p functionality, with the existing (inaccurate) mkdir -p mention removed.

[pitrou]

I think we can a new exist_ok parameter indeed. It should probably be False by default, though.

[warsaw]

I've just been writing some new code to use pathlib and ran into this one myself. An exist_ok=False is fine, although it's a slight shame that for backward compatibility we can't adopt os.makedirs() signature exactly.

[pitrou]

Le 27/05/2014 20:32, Barry A. Warsaw a écrit :

I've just been writing some new code to use pathlib and ran into
this
one myself. An exist_ok=False is fine, although it's a slight shame that
for backward compatibility we can't adopt os.makedirs() signature exactly.

What do you mean by that? The os.makedirs() signature is
os.makedirs(name, mode=0o777, exist_ok=False)

[warsaw]

On May 27, 2014, at 08:48 PM, Antoine Pitrou wrote:

What do you mean by that? The os.makedirs() signature is
os.makedirs(name, mode=0o777, exist_ok=False)

Right, but this is Path.mkdir's signature:

Path.mkdir(mode=0o777, parents=False)

so it's too late to make exist_ok=False the second argument (i.e. the one
after mode). Oh well.

[berkerpeksag]

Updated patch. Thank you Barry and Antoine.

[warsaw]

I like it. Thanks for the contribution! Assigning to myself since I plan on working to land this in 3.5.

[python-dev]

New changeset 8af80d68bcdc by Barry Warsaw in branch 'default':

• Issue bpo-21539: Add a exists_ok argument to Pathlib.mkdir() to mimic
  http://hg.python.org/cpython/rev/8af80d68bcdc