list supported domains

[kokoko3k]

I'm writing a gui to mpv.
When using youtube-dl via mpv, trying to play a network stream without knowing in advance if it supported by youtube-dl or not , and so activating the ytdl hook on mpv may slow down the process (expecially on slow machines) of starting the play.
Is there a way to show a list of domains supported by youtube-dl so that a simple string match could be done in advance and choosing to use or not the ytdl hook within mpv?
--list-extractors does not show domains.

[dstftw]

That is not possible and here is why.

[kokoko3k]

As i see through the code, every extractor has a _VALID_URL variable.
Why listing that should not give a list of supported sites?
I did not talk about an an exhaustive list.

[kokoko3k]

Brute example of 10 supported sites (if not, bugreport them!):

#grep -R "VALID_URL"  ./|cut -d "=" -f 2|cut -d \' -f 2|grep '^http\:\|^https\:\|^https\?'|head -n 10
https?://(?:www|m)\.nuvid\.com/video/(?P<id>[0-9]+)
https?://www\.wayofthemaster\.com/([^/?#]*/)*(?P<id>[^/?#]+)\.s?html(?:$|[?#])
https?://videos\.toypics\.net/view/(?P<id>[0-9]+)/.*
http://videos\.toypics\.net/(?P<username>[^/?]+)(?:$|[?#])
https?://video(?P<team>\.[^.]*)?\.nhl\.com/videocenter/(?:console|embed)?(?:\?(?:.*?[?&])?)(?:id|hlg|playlist)
https?://(?:.+?\.)?nhl\.com/(?:ice|club)/news\.html?(?:\?(?:.*?[?&])?)id
https?://video\.(?P<team>[^.]*)\.nhl\.com/videocenter/(console\?[^(id
http://.*?\.defense\.gouv\.fr/layout/set/ligthboxvideo/base-de-medias/webtv/(?P<id>[^/?#]*)
https?://?(?:www\.)?escapistmagazine\.com/videos/view/[^/?#]+/(?P<id>[0-9]+)-[^/?#]*(?:$|[?#])
https?://(?:www\.)?gameone\.de/tv/(?P<id>\d+)

?

[phihag]

If you want any incomplete list, why not just hardcode those 10 lines?

I'm assuming that you really do not want any incomplete list, but instead, want to test whether a URL could be supported by youtube-dl. That is quite simple.

[kokoko3k]

I pasted 10 lines just to not flood the message.
I'll try to explain better my problem.

Use case is:
my mpv frontend trying to play a generic url on a slow netbook.
url is so generic that i don't know in advance if it is playable by plain mpv or needs to be parsed by mpv's ytdl hook.
I can let mpv do all of the work; it will try to use ytdl first, and if it fails it will fallback to ffmpeg.
It will work, but it is very, very slow (atom n270 here).

So i tried to do the opposite; using ffmpeg first, and if it failed, then fallback to ytdl.
The problem here, is that sometimes ffmpeg tries too hard to play webpages, spending a lot of time before failing.

The only way i could think to really speed up things (at least most of the times) is to know in advance if a site is (at least probably) supported by ytdl, so that on those sites i can first try ytdl, then ffmpeg; and do the opposite if i think the site is not supported.

By now i managed to create an incredibly brute and hacky command full of grep,sed,sort,awk that scans the youtube-dl code and creates a list of supported domains, and yes, i've hardcoded it into my mpv frontend code.

See the mess:
http://pastebin.com/b0pbvrMs

..it even works, but i'll have to regenerate the list everytime youtube-dl is updated with new extractors.

[arderyp]

here's the answer: #4503