caption_extractor — video-site caption extraction

Add caption support for a new video site to the browser extension. The host sends the enabled extractor scripts to the extension; the extension runs on every web page, and only pages matched by match(url) have their captions captured.

Declaration

json

{
  "id": "example.caption.extractor",
  "name": "示例：自定义视频站字幕提取",
  "version": "1.0.0",
  "target_app_version": "0.2",
  "type": "caption_extractor",
  "entrypoint": "extractor.js",
  "enabled": false,
  "tags": ["字幕", "示例"]
}

entrypoint points at an extractor.js. Defaults to enabled: false.

Script skeleton

javascript

// Copy this file, change the id and match, and you can provide captions for a new video site.
TransBoxExtractors.register({
  id: "my_site",
  match(url) { return /example\.com/.test(String(url || "")); },
  videoId(url) { return new URL(url).pathname; },

  // Option A: use when the site has a full timed track file
  async fetchManifest(videoId, ctx) {
    return {
      src_lang: ctx.preferredLang || "en",
      cues: [{ start: 0, end: 1.2, text: "Hello" }],
      trackName: "my site",
    };
  },

  // Option B: use when captions only exist on screen and there is no timed track file
  collectLines(doc) {
    return TransBoxCaptionHost.linesFromSelector(".player-caption");
  },
});

(fetchManifest and collectLines are either/or — or provide both.)

Interface

Member	Description
`id`	Extractor identifier (required)
`match(url)`	Only enabled on a matching site; return falsy to stay out of the way
`videoId(url)`	Extract the video ID from the URL (used for dedup / change detection)
`fetchManifest(videoId, ctx)`	Returns `{src_lang, cues: [{start, end, text}], trackName}`; return `null` when there is no timed track
`collectLines(doc)`	Returns each line currently visible on screen; omit it and the built-in YouTube / Bilibili selectors are used instead

ctx.preferredLang is the source language the user currently prefers.

Available globals

The script runs in the browser extension's isolated world and may only rely on these two objects (do not rely on page globals):

Object	Methods
`TransBoxExtractors`	`register({...})`
`TransBoxCaptionHost`	`cleanCaptionText(raw)`, `parseRawTimedText(raw, lang)`, `preferredLang()`, `linesFromSelector(css)`

How it takes effect

After enabling the extractor, refresh the page that is playing; the desktop app hands the script to the extension during the handshake.

For the chain the captions take after reaching the desktop app, see Captions & translation. For the end-user side of the browser extension, see Browser native captions.

caption_bridge is the web caption bridge itself (plugin community_web_caption_bridge, plugin ID community.caption.browser); it ships through the Steam Workshop and is not bundled into the official installer. Third parties must not write a second bridge — to add a site, use the caption_extractor on this page.

caption_extractor — video-site caption extraction ​

Declaration ​

Script skeleton ​

Interface ​

Available globals ​

How it takes effect ​