brandonschabell / urlpath Goto Github PK

Python3.12 removed _Flavour and _PosixFlavour from pathlib.

import pathlib

ImportError: cannot import name '_PosixFlavour' from 'pathlib'

That would be nice to have a relative_to method for URL that would not catch the PurePath.relative_to method.

An implementation:

ralative_path = Path(base_url.path).relative_to(Path(child_path.path))

I'll try to take some time one day to help you for this nice package.

Seen behaviour:

URL('a') / "/b"
URL('/b')

Expected behaviour:

URL('a') / "/b"
URL('a/b')

I have a url in string that has white spaces enconded as '%20'.
url = 'https://somepath.com/test/Test%20path/my%20test%20file.txt'

when I pass it to URL I get the following:

>>> from urlpath import URL
>>> URL("https://somepath.com/test/Test%20path/my%20test%20file.txt")
URL('https://somepath.com/test/Test%2520path/my%20test%20file.txt')

apparently only happens when the spaces are in the midle of the path.

URL('http://www.example.com/abc:def.html') → URL('http://www.example.com/def.html')

The urlpath.URL class for some reason is being classified as type Any by mypy and hence not being checked properly. I thought this might be a mypy bug, but the same issue occurs with pytype as well.

Minimal example to demonstrate:

from pathlib import Path

from urlpath import URL

def do_something(path: Path, url: URL) -> str:

    return url / path

do_something(Path('file.ext'), URL('example.com')) # ok
do_something('file.ext', URL('example.com')) # error
do_something(Path('file.ext'), 'example.com') # ok ?!?

reveal_type(Path('file.ext')) # Revealed type is pathlib.Path
reveal_type(URL('example.com')) # Revealed type is "Any" ?!?

Any idea why this is? I just found this library today, already use pathlib a lot in my project and really like the look of this, but this behaviour is wierd and makes it less usable in a type hinted enviroment.