feat(player): move comments and the description onto InnerTube and bring the player surfaces onto Material 3 Expressive

[A-EDev]

Summary

Moves the video player's comments and description onto the native InnerTube schemas, and brings the player's sheets, dialogs and buttons onto Material 3 Expressive.

Comments now come from InnerTube instead of the extractor, so the section's own sort menu, its real total, creator hearts, author badges and per-comment timestamps are read rather than guessed. The description is built from the watch page's attributed text and header factoids, with a new transcript and creator card beside it. The settings pages, download dialogs, action row and queue dock were reworked on M3 Expressive components and the app's own theme tokens.

Related issue

Closes #791
Closes #794
Closes #956

Change type

• Bug fix
• Feature
• Refactor or maintenance
• Build, packaging, or CI
• Documentation

Validation

• ./gradlew :app:assembleGithubDebug
• ./gradlew :app:testGithubDebugUnitTest — 1,676 tests, 0 failures (was 1,626 on main; +50 new)
• ./gradlew :app:compileFossDebugKotlin, and Spotless applied to every touched file
• I manually tested the affected behavior on an Android device or emulator.

Test device and Android version: Xiaomi POCO X6 Pro, Android 16

Exercised on device: comments (sort chips, the timed filter, pinned and creator badges, reply threads), the transcript (segments, search, follow-along), the description (factoid cards, collapsed body, social links, creator card), the player settings pages, both download dialogs and the queue dock.

Risk and compatibility

Comments and description now depend on the InnerTube watch response. One next request per video is made lazily, the first time a surface asks for it, and cached — a video whose comments and description are never opened costs nothing, and nothing was added to the playback path. If the comment response cannot be parsed, the repository falls back to the extractor, so a schema change on YouTube's side degrades the section rather than emptying it.

The transcript reads the caption track the player already resolves for subtitles, asking for json3 and dropping the aAppend roll-up repeats. It replaces the fmt already on the URL rather than appending a second one. One request per track, cached.

• Sorting is server-side now. YouTube offers Top and Newest only; Oldest is kept and clearly re-orders the pages already loaded. Comment like counts are the abbreviated figures the response carries ("529K"), as YouTube itself shows them.

• A bounded run of three further comment pages is prefetched after the first, sequentially and cancelled on a sort change or video change, so the Timed filter has something to filter.

• The author-avatar repair fetches the extractor path needed are gone; InnerTube carries the avatar inline.

• Removes four hardcoded colours (Color.Gray, Color.DarkGray, the accent-coloured download badges, the queue dock's primary-tinted tile and outline).

• Visual changes by design: settings rows are Material 3 segmented groups, the description sheet is rebuilt on artwork-tinted cards, the subscribe button is the extra-small size in the player only, and the action row is on Expressive buttons with haptics.

• The dead thumbs-down and reply icons are removed from each comment rather than made to work; posting and liking comments needs an account session and is not part of this PR.

• Known, separate from this PR: YouTube.transcript() (the YouTube-subtitle lyrics provider) returns HTTP 400 for every video — the get_transcript endpoint it calls is gone. Untouched here.

• The change does not introduce secrets, private data, or unexpected telemetry.

• New user-facing text uses Android string resources.

• Dependency and lockfile changes are intentional and limited to this PR — no dependency changes.

• Room schema changes include the required version bump and migration, or this PR does not change the Room schema.

• Breaking changes and upgrade steps are clearly documented — none.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.