KoalaDev
36e1291d2c
The weighted score summed incomparable units, so size could outvote traits that disqualify an element outright. Measured on a real page: a display:none preload reports its full 1080p intrinsic size and scored 2073600, beating a visible unmuted player at 509920. Selection now compares an ordered list of signals, highest priority first: has a source, is rendered, is not a silent background loop, rendered size bucket, is playing, has controls, duration. Rendered size replaces intrinsic resolution, and mute state is gone from the ranking entirely: it is a viewer preference, not evidence about which element is the player. It stays a ranking rather than a filter, so a page of only bad candidates still yields one and findVideo never returns null where a video exists. The new tests/e2e suite runs the shipped finder against real fixture pages and drives the packed extension for injection, reinjection and remote play/pause/seek into a first-party frame. All five scoring scenarios fail against the previous implementation. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Extension E2E smoke tests
Browser-level tests for the parts that unit tests cannot reach: which <video>
the extension picks on a real page, and whether the packed extension gets far
enough to control it.
npm run test:e2e:install # once, downloads the browsers
npm run build:extension # extension.spec.mjs loads dist/chrome
npm run test:e2e
Layout
| Path | Purpose |
|---|---|
detection.spec.mjs |
Runs the shipped findVideo() against the fixture pages |
extension.spec.mjs |
Loads dist/chrome, injects into a tab, applies remote play/pause/seek |
fixture-server.mjs |
Static server for the fixtures, with byte-range support for media |
fixtures/pages/ |
One page per scenario |
fixtures/media/ |
Small generated clips (see below) |
helpers/content-source.mjs |
Lifts the real finder out of extension/content.js |
Two rules worth keeping
The specs run the shipped source, not a copy. helpers/content-source.mjs
extracts findVideo and its ranking helpers straight out of
extension/content.js. A fixture that passes against a reimplementation would
prove nothing about the extension. If you split the finder into more functions,
add them to VIDEO_FINDER_EXPORTS there and to VIDEO_FINDER_PARTS in
scripts/test-content-video-finder.cjs, or the extraction fails loudly.
Each fixture marks its own answer. The element that must win carries
data-expected; videos that have to be playing carry data-autoplay, and
ready.js holds the page back until metadata and playback have settled. The
specs assert those preconditions before judging the finder, so a broken fixture
reads as a broken fixture instead of a scoring regression.
Scenarios
| Fixture | What it pins down |
|---|---|
simple-player.html |
The ordinary case |
iframe-player.html |
Player inside a same-origin frame, empty top document |
late-frame.html |
Player frame attached after the page settled |
shadow-player.html |
Player in a shadow root, tiny teaser in the light DOM |
muted-player.html |
Mute must not disqualify the only player |
hidden-preload.html |
A display:none preload still reports 1080p; it must lose |
ad-frame.html |
1080p asset in a 300x250 ad slot must lose to the real player |
background-loop.html |
Silent looping hero must lose despite being the largest |
multi-player.html |
Between equal players, the playing one wins |
sourceless.html |
A large <video> with no source can never be the player |
Regenerating the media
Solid-colour clips, a few KB each, committed so the suite needs no network:
ffmpeg -y -f lavfi -i "color=c=blue:s=1920x1080:d=30:r=10" -c:v libx264 -preset veryfast -crf 45 -pix_fmt yuv420p -movflags +faststart fixtures/media/player-1080p-30s.mp4
Same command with green/854x480/12, red/640x360/5 and gray/1280x720/3 for
the other three.