Upgrade YouTube.js from 4.3 to 5.0.2 (fixes throttling)

[absidue]

Upgrade YouTube.js from 4.3 to 5.0.2 (fixes throttling)

Pull Request Type

• Bugfix

Related issue

closes #3457
closes #3475 (dependabot PR)

Description

Upgrades to YouTube.js 5.0.1, as YouTube.js 5 did a bunch of refactoring this PR also has changes for that in it.

The code changes in this PR in no particular order:

• Update jsdoc types, v5 moved files around so some of them broke, so I decided to go through all of them and use the exported types instead of referencing the paths
• Use the new endpoint builders and constants (I know that Endpoints.BrowseEndpoint.PATH is verbose but as they are ES6 namespace exports/imports, webpack optimises it to a direct reference to PATH at build time, even for the code inside YouTube.js) for the local API subscription fetching
• Some properties got renamed, the old names are deprecated and will be removed in a future YouTube.js update, so I switched to the new ones e.g. views -> view_count
• YouTube.js now parses the dialog in text runs, so we can simplify our code a bit (Fix video comment external link parsing for local API #3448)
• A change that YouTube.js made a while ago but doesn't seem to have affected us too much so far (only edge cases) is no longer returning "N/A" if a Text element is empty, it now has an Text#isEmpty() method that we can use instead, so I switched to that as well.

Testing

Everything with the local API without video proxying enabled

1. Check that videos load faster than before
2. Refresh subscriptions without RSS (if it works it's good)
3. Test case from Fix video comment external link parsing for local API #3448
4. Visit the @YouTube channel or https://youtube.com/@YouTube if you prefer a full URL and check that it doesn't spit out errors about not being able to call .replace() on undefined anymore.

Desktop

• OS: Windows
• OS Version: 10
• FreeTube version: 0e11fd0

[absidue]

The view count on the watch page is currently always NaN, created a PR in YouTube.js to fix it LuanRT/YouTube.js#393.

[PikachuEXE]

• Fix video comment external link parsing for local API #3448 tested
• Refresh subscriptions without RSS
  • Still forced to refresh via RSS with 125 channels
  • Works on 120 channels
  • Manually increased limit and still works with 125 channels (but slower than RSS)
• @YouTube and URL form works
• Tested example video https://www.youtube.com/watch?v=i3DwBAa1boY about loading speed
  • With & without AV1 it seems loading is faster

[PikachuEXE]

Screenshot for NaN views

[MarmadileManteater]

Most everything looks good. The buffering speed seems to be noticeably improved for me, but I did notice a warning out of youtube.js on the video: https://youtu.be/eBZ2vA9D2RA which didn't seem to be present before. It doesn't look like anything is breaking, but this warning does appear in the console when the video first loads before the comment section is opened, so I don't think this warning has do to with anything that FT uses since the comment section is only shown after an additional request is made for a comment continuation.

[ChunkyProgrammer]

I'm guessing this PR will be updated to 5.0.2?

[absidue]

Views are fixed now.

[absidue]

@MarmadileManteater that's probably fine, that just means there was a node/renderer in the Innertube response that YouTube.js doesn't know about. Unless something is broken (e.g. missing information, other errors) it's not something we need to worry about, if there is missing information it probably needs to be parsed from the new nodes in YouTube.js.

[Gorrrg]

$ node _scripts/build.js • electron-builder version=23.6.0 os=10.0.20348 • packaging platform=win32 arch=x64 electron=22.3.2 appOutDir=build\win-unpacked ⨯ cannot resolve https://github.com/electron/electron/releases/download/v22.3.2/electron-v22.3.2-win32-x64.zip: status code 503

Build script had trouble downloading a package.

[absidue]

You don't have to build it yourself anymore, as it's been merged you can just download the nightly build for it.

[Gorrrg]

Oh sorry, forgot to mention. That was from the nightly build log. Your win x86 builds failed. But the next nightly 2890 succeeded, so it's good.