Issue
Whenever I want to work with files or directories, the approach of today is to use pathlib
.
pathlib
is great, convenient, easy to use, and OS independent.
However, some libraries still expect me to use os.path
.
How can I transform my Path
to the old-school version?
What should I do with my p
in the following example?
from pathlib import Path
p = Path.cwd()
some_old_function(p) # <---- error
Solution
The phrase some libraries still expect me to use os.path
seem to confuse you. os.path
is a module that simply treats paths as strings - as opposed to pathlib
which treats them as specialized objects in the OOP approach.
To create an easy and well-defined transition between the two representations, PEP 519 was introduced in Python 3.6. It introduces a new abstract class - PathLike
- and a new os
function - fspath
. Any API/library that wishes to implement a path object (pathlib
included), should inherit PathLike
and implement the __fspath__
method (which will be called by os.fspath
).
The main rationale for this is that using str(path)
is error-prone - many objects have a string representation (str(None)
gives "None"
), but they should not all be treated as paths.
Specifically for pathlib
(at the time of writing this), the implementation of __fspath__
is basically just a wrapper over str()
, but you should still arguably prefer to use it in case it ever changes in the future.
Note that you don't need to actually call os.fspath
when passing an argument to most built-in functions. Most os.path
functions were changed to accept PathLike
s instead of strings and already use os.fspath
on it (can be seen in the docs as Changed in version 3.6: Accepts a path-like object. under most functions). So you can simply pass a Path
object as-is to most of them.
Answered By - Tomerikoo
0 comments:
Post a Comment
Note: Only a member of this blog may post a comment.