Adjust _REG_NAME_PAT to reject invalid characters

[kenballus]

Addresses #2802.

[kenballus]

It might be preferable to percent-encode invalid characters instead of rejecting them. Feedback welcome!

[jayaddison]

If you have time @kenballus, there's another potential regex simplification that I've been wondering about within the same Python module. Perhaps we could trade each other some code review/inspection time?

[kenballus]

If you have time @kenballus, there's another potential regex simplification that I've been wondering about within the same Python module. Perhaps we could trade each other some code review/inspection time?

I am definitely up for that. Shoot me an email and we can set up a time.
(My email is listed at the bottom of this page)

[jayaddison]

Despite my initial enthusiasm I haven't found the time to arrange a time yet, my apologies. A couple of questions that did occur to me in the meantime:

• Could/should the rfc3986 library provide much of the validation here (perhaps even instead of some/all of the regexes currently in use)?
• Is support for WHATWG URLs required here?

[kenballus]

Could/should the rfc3986 library provide much of the validation here (perhaps even instead of some/all of the regexes currently in use)?

Unfortunately, that library has URL parsing/validating issues as well. See this issue.

Is support for WHATWG URLs required here?

Some parts of WHATWG are supported, like '\' to '/' translation, but my understanding is that the RFCs are prioritized over WHATWG in this library.

[jayaddison]

Thanks @kenballus, and sorry for the delays responding. I think I'm beyond my level of understanding on this one, so I'm going to step aside.

[sethmlarson]

Closing this with justification here: #2802 (comment)