[Feature Request] New command-line argument to filter extractors.

[mattoz0]

Checklist

• I'm reporting a feature request
• I've verified that I'm running youtube-dl version 2021.03.03
• I've searched the bugtracker for similar feature requests including closed ones

Description

Would be nice if we could improve loading times by filtering the extractors that would be checked.

youtube-dl --extractors youtube url
youtube-dl --extractors youtube,soundcloud url

Often enough i would guess people know which service they are using! Extractor checking can take a lot of time on low powered devices. This would remove the need to make a custom build for a lot of usecases.

[pukkandan]

There is an experimental feature lazy_extractors that tries to speed up extractor loading.

[mmkhitaryan]

AFAIK youtube-dl can cause SSRF. So that feature would also increase backend security for web apps. I have done something like this by monkey patching youtube_dl.extractor._ALL_CLASSES
https://stackoverflow.com/a/65284695/7415288
I think it would be better if there was a feature that allows doing that without monkey patching.

[mattoz0]

Lazy extractors seems like a step in the right direction. However, my understanding is its still loading all the extractor's (if its going to load an extractor).

I propose filtering the __ALL_CLASSES array in youtube-dl/youtube_dl/extractor/init.py, based on the command line argument.

My assumption is a solution could be created, very similar to the link in the above monkey patching example.