hiro protagonist hiro protagonist - 2 months ago 16
Python Question

normalize non-existing path using pathlib only

python has recently added the pathlib module (which i like a lot!).

there is just one thing i'm struggling with: is it possible to normalize a path to a file or directory that does not exist? i can do that perfectly well with

. but wouldn't it be absurd to have to use something other than the library that should take care of path related stuff?

the functionality i would like to have is this:

from os.path import normpath
from pathlib import Path
pth = Path('/tmp/some_directory/../i_do_not_exist.txt')
pth = Path(normpath(str(pth)))
# -> /tmp/i_do_not_exist.txt

but without having to resort to
and without having to type-cast to
and back to
. also
does not work for non-existing files.

is there a simple way to do that with just


No, there's not.

PEP 0428 states:

Path resolution

The resolve() method makes a path absolute, resolving any symlink on the way (like the POSIX realpath() call). It is the only operation which will remove " .. " path components. On Windows, this method will also take care to return the canonical path (with the right casing).

Since resolve() is the only operation to remove the ".." components, and it fails when the file doesn't exist, there won't be a simple means using just pathlib.

Also, the pathlib documentation gives a hint as to why:

Spurious slashes and single dots are collapsed, but double dots ('..') are not, since this would change the meaning of a path in the face of symbolic links:

PurePath('foo//bar') produces PurePosixPath('foo/bar')

PurePath('foo/./bar') produces PurePosixPath('foo/bar')

PurePath('foo/../bar') produces PurePosixPath('foo/../bar')

(a naïve approach would make PurePosixPath('foo/../bar') equivalent to PurePosixPath('bar'), which is wrong if foo is a symbolic link to another directory)

All that said, you could create a 0 byte file at the location of your path, and then it'd be possible to resolve the path (thus eliminating the ..). I'm not sure that's any simpler than your normpath approach, though.