Document path converter regexes and limitation

[UnknownPlatypus]

I noticed while testing on a personal project that the url slug rewriter was matching the [-a-zA-Z0-9_]+ but not the equivalent [\w-] or [-\w].

I thought maybe we could support that too ?

[adamchainz]

Django’s SlugConverter uses that regex:

https://github.com/django/django/blob/f88fc72da4eb76f2d464edb4874ef6046f8a8658/django/urls/converters.py#L35-L36

\w matches more characters, since it matches all Unicode letter chars:

In [1]: import re

In [2]: re.match(r'\w', 'α')
Out[2]: <re.Match object; span=(0, 1), match='α'>

So unfortunately we can't automatically convert here. But I understand this might be quite common. Perhaps we could add some flag or extensibility here to allow saying that "in this project \w really only means to match latin letters" ?

[UnknownPlatypus]

Oh indeed, the greek letters.

Maybe a flag is a great idea if this case seems to be quite common ?
I'm not sure it actually is thought. Maybe we should keep the current behaviour after-all ?

[adamchainz]

I think at least we could document the regexs, and maybe add a sentence that \w is sometimes used for slugs but it's not converted due to potential incompatibility. Would you like to take a crack at that?

[UnknownPlatypus]

Looks like a good idea, I'll redo this PR tomorrow to document the slug regexes and why we don't support \w

[UnknownPlatypus]

Took me more time than expected to get back to this, I'm in the middle of moving to a new place.
Anyway, let me know if it's looking good for you!

[adamchainz]

Looking good, I just made a few tweaks to improve readability.

First commit merged here at the Djangocon Europe sprints 👍👍👍