Support file: ?

[pprados]

It's sympathic to use cloudpathlib with url file:.

[pjbull]

I'd be open to at least a simple shortcut implementation that assumes anything after file:// you can hand off to pathlib.

One question is: if you start with a file URI, do you want a file URI back from all of the methods or is a Path object enough (and consumers can call the Path.as_uri() method if they need to)?

Here's a way to go to a Path from a file that is reported as generally working.
python/cpython#91504 (comment)

That said, to my eye there do seem to be some idiosyncrasies about the specification for file URIs that may break. See: https://en.wikipedia.org/wiki/File_URI_scheme

Resources:

• related discussion with Python devs
• related Python issue with potential implementations

[pprados]

The idea is, with an environment variable only, I can choose the URL.
file: for test, s3: or other for real files. So, the entire API of cloudpathlib must be implemented, with .read_text() etc.
The same code must be run.

[pjbull]

@pprados, you can now use file: with AnyPath in the latest version of cloudpathlib on PyPI (0.18.0). See the docs here:
https://cloudpathlib.drivendata.org/stable/anypath-polymorphism/#file-uri-scheme