Improve `uses_relative` in `urllib.parse.urljoin`

[alexpdp7]

Feature or enhancement

Proposal:

I was trying to use urllib.parse.urljoin with Gemini URLs and noticed it doesn't have the proper behavior for Gemini URLs (which are very similar to http/https URLs), because gemini is not in uses_relative.

Gemini has the behavior for uses_relative, uses_netloc, and uses_params.

Probably it is not desired to grow these list with uncommon protocols, but it would be nice to add optional arguments to functions such as urljoin to override behavior for certain protocols.

Has this already been discussed elsewhere?

This is a minor feature, which does not need previous discussion elsewhere

Links to previous discussion of this feature:

No response

[picnixz]

urljoin is meant to match RFC behaviors so I am not entirely sure we need to do that. This rather looks like an extension for a PyPI package.

[picnixz]

FTR you can add the gemini to the list of schemes yourself though I would instead write a dedicates function for creating Gemini URLs

[alexpdp7]

Ah, thank you so much, that works perfectly and I hadn't thought of it. Just in case someone arrives here via a web search:

$ python3
>>> import urllib.parse
>>> urllib.parse.uses_relative += ["gemini"]
>>> urllib.parse.uses_netloc += ["gemini"]
>>> urllib.parse.urljoin('gemini://example.com/foo/bar', '../qux.html')
'gemini://example.com/qux.html'

[picnixz]

While publicly named, I think we should have a proper API for registrring new schemes that are RFC like though so that they get properly recognized. I will do this in the next few weeks.