Rework youtube link parsing

[Ivan-Shaml]

What I aim to do with this PR:

1. Rework the front-end parsing of YouTube links, to make the code cleaner, easier to follow, and reusable.

• Long regular expressions for non-straightforward tasks, bring more complexity and make the code hard to follow.
  • Before: with one long regex we tried to handle youtu.be and youtube.com URLs, which are structurally different.
  • Now: for each URL type, we handle it accordingly.

2. Added Scala change, where the URL parsing of in the markdown is fixed.

• Before: if protocol is omitted (www.youtube.com) the link goes to e.g. https://lichess.org/forum/general-chess-discussion/training-your-chess-intuition-and-board-awareness/www.youtube.com
• Now: goes to https://www.youtube.com where http:// is added by default, if URL protocol is missing.

Closes #18780

[fitztrev]

Could you add a test for the url structure mentioned in #18780 to ensure it doesn't try to embed?

And then would you mind providing a copy+paste-able block of sample URLs that we can put in a blog or forum post to test the behavior locally?

[Ivan-Shaml]

Links to test with (rick roll not intended)

• Should embed (you can try variations with http:// in the beginning instead of https:// or remove the protocol it altogether to test the Scala change):

  • https://www.youtube.com/watch?v=dQw4w9WgXcQ
  • https://www.youtube.com/watch?v=dQw4w9WgXcQ&t=10s
  • https://www.youtube.com/watch?v=dQw4w9WgXcQ&t=1m3s
  • https://youtu.be/dQw4w9WgXcQ
  • https://youtu.be/dQw4w9WgXcQ?t=10
  • https://youtu.be/dQw4w9WgXcQ?t=1m10s
  • https://www.youtube.com/embed/dQw4w9WgXcQ
  • https://www.youtube.com/embed/dQw4w9WgXcQ?start=15

• Should not embed:

  • Channels:
    • https://www.youtube.com/channel/UCuAXFkgsw1L7xaCfnd5JJOw
    • https://www.youtube.com/@HangingPawn

[Ivan-Shaml]

LGTM

[SergioGlorias]

we can check also:
https://www.youtube.com/live/dQw4w9WgXcQ (URL for livestreams)
https://youtube.com/shorts/dQw4w9WgXcQ (URL for shorts)

[SergioGlorias]

also can add playlist:
https://www.youtube.com/playlist?list=PLe5ZNOR8Ttm1HHLAo64Hv4XGa_XU5-a7t
-> embed to https://www.youtube-nocookie.com/embed/videoseries?list=PLe5ZNOR8Ttm1HHLAo64Hv4XGa_XU5-a7t

[Ivan-Shaml]

also can add playlist:

Hmm, there is an edge-case, if user sends video URL like https://www.youtube.com/watch?v=Nf7oA1Q42Pw&list=PLe5ZNOR8Ttm1HHLAo64Hv4XGa_XU5-a7t&index=1

It will be parsed as regular watch, instead of playlist + watch. Should we handle that ?
P.S: Youtube embed API does not support this, we will treat it as regular video.

[ornicar]

probably should be https here?

[Ivan-Shaml]

Yeah, I did it like this initially, but thought what if the server is http only. If its https and correctly configured it will HTTP 301 to the secure url. We don't know what link they will paste in.

[ornicar]

In 2025 http:// is fully dead, I haven't seen such a link for a long time.