Per-frame source and subtitle lists grow without bound on long-lived pages, each addition costing a page round trip and a broadcast to every tab #234

Severity: medium

Where: chrome/background/background.mjs (addSource: frame.getSources().push({url, mode, headers, time}); onSourceRecieved runs scrapeCaptionsTags(frame) and sendSourcesToMainFramePlayers(frame) (a tabs.query({}) plus a SOURCES message carrying the whole list to every player-page tab) for every detected source; onHeadersReceived runs on every tab, On or Off)

What happens

Nothing trims trackedSources/trackedSubtitles except a navigation (FRAME_ADDED reset) or collectSources(frame, true) when a player asks. tests/e2e/ext-specs/source-length.e2e.mjs confirms that media segments named .mp4 (Shaka Packager style) are detected as MP4 sources of their own; HLS WebVTT subtitle segments (seg-N.vtt) go through isSubtitles the same way, one entry per unique URL.

A tab left Off (the normal state) plays a DASH/HLS title whose segments are seg-123.mp4/seg-123.vtt for two hours: about 3,600 sources, each holding the full requestHeaders array of its request (Cookie included), plus one SCRAPE_CAPTIONS round trip to the page and one chrome.tabs.query({}) + a SOURCES message carrying the whole growing list to every open player-page tab, per segment (O(n²) traffic). A single-page site that never navigates keeps this for the session.

Suggested fix

Cap each frame's lists (keep the newest N, e.g. 200, by time), and skip tracking a file whose StreamLengths.shape already has a tracked entry (the pieces of one stream need one representative, not all). Move scrapeCaptionsTags and sendSourcesToMainFramePlayers behind frame.tab.isOn / "a player tab exists" checks, or rate-limit the broadcast.

Confidence: high.

From a line-by-line code review of the tree at 41363fd (2026-10-02). Each claim above was checked against the source before filing; the reproduction steps and line references are as of that commit.

Generated by Claude Code