Open davidovski opened 12 months ago
Honestly just laziness. I don't feel like coding that in... Also Invidious works fine I guess...
It is a good idea, however.
Extra lazy way: Do what we're doing now, rely on one instance.
Lazy way: Rotate official clearnet Invidious instances. You can get this list from here and parse the results to remove hidden services.
Proper way: The Invidious API parses directly from this list.
Actual proper way: Just scrape YouTube itself.
Also, https://invidious.snopyta.org appears to redirect to https://redirect.invidious.io/, which requires JavaScript, which thereby makes video results broken, I think.
yt.ahwx.org is up I guess? Although I'd rather not be responsible for uptime of a huge project as most people will not change the default.
Currently LibreY gets its video results from a configured invidious instance. However I am concerned that most instance maintainers don't update this to point to a different instance, which may cause unecessary extra load on that instance.
Other than the fact that invidious provides a clean API for video search, I don't see any reason why we shouldn't fetch video results from youtube.com directly, as we are already proxying requests to google.