wikipedia: support query strings in Wikipedia URLs

[SnoopJ]

Description

This PR fixes #2412 by discarding the query string from the name of the article to be searched.

Checklist

• I have read CONTRIBUTING.md
• I can and do license this contribution under the EFLv2
• No issues are reported by make qa (runs make quality and make test)
• I have tested the functionality of the things this change touches

[dgw]

I am foolish for not looking more closely at this earlier. Why not use the parsed URL for the fragment, also? Are you trying to avoid touching too much? If you change the URL pattern to https?:\/\/([a-z]+(?:\.m)?\.wikipedia\.org)\/wiki\/(?!File\:)[^ ]+ it should be possible to offload fetching all the bits to urllib.parse instead of the regex.

(Not a review comment because I'm just trying to understand the intended scope; Exi's comment you referenced was about converting all of the link-matching logic to use urllib.parse, stripping the regex down to only what's necessary to make sure what's been given is a Wikipedia article of some kind.)

[dgw]

Since you're updating comments anyway...

[Exirel]

I think it's the right moment to push a little further the cleanup of the plugin and remove the use of sopel.tools.web.unquote when it's literally just a shortcut for the stdlib urllib.parse.unquote.

Edit: actually, same for sopel.tools.web.quote. Both are useless now.

[Exirel]

Alright, all that is left is a squash, since @dgw approved and I checked the changes. I don't think there is enough to require more than one commit, and I don't mind if you lose the co-author from my suggestions (I don't assume you know, but you might, all that to say it's up to you).

[dgw]

I would have put changing the existing imports from sopel.tools.web as out of scope for this, but it doesn't ultimately matter. The quote and unquote wrappers there have comments that we should consider deprecating them, anyway.