Eliminate most provider match checks to improve speed

[dullbananas]

No description provided.

[dessalines]

cc @jendrikw

[jendrikw]

Sorry for not responding earlier, I was a little busy. reviewing now

[dessalines]

bump

[jendrikw]

So I thought about this and it feels like writing our own little regex parser, which obviously sounds like a bad idea. Are you sure the regexes are a bottleneck for lemmy? If you reuse the UrlCleaner instance, it should not take a significant amount of time.

[dullbananas]

I'm pretty sure I didn't measure the impact on Lemmy's performance. But it's definitely not good to always loop through each regex to check for a match.

Possible alternatives:

• put everything in one regex::RegexSet, if the regex library can efficiently handle regexes that have duplicated parts
• for the clearurls json file, propose a different format that makes it easy to do filtering mostly with hashmaps/hashsets, instead of checking a lot of regexes every time