@steipete/summarize-core 0.8.2 → 0.10.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/LICENSE +1 -1
- package/dist/esm/content/index.js.map +1 -1
- package/dist/esm/content/link-preview/client.js +4 -0
- package/dist/esm/content/link-preview/client.js.map +1 -1
- package/dist/esm/content/link-preview/content/fetcher.js +3 -2
- package/dist/esm/content/link-preview/content/fetcher.js.map +1 -1
- package/dist/esm/content/link-preview/content/firecrawl.js +3 -1
- package/dist/esm/content/link-preview/content/firecrawl.js.map +1 -1
- package/dist/esm/content/link-preview/content/html.js +3 -1
- package/dist/esm/content/link-preview/content/html.js.map +1 -1
- package/dist/esm/content/link-preview/content/index.js +101 -19
- package/dist/esm/content/link-preview/content/index.js.map +1 -1
- package/dist/esm/content/link-preview/content/types.js.map +1 -1
- package/dist/esm/content/link-preview/content/utils.js +15 -7
- package/dist/esm/content/link-preview/content/utils.js.map +1 -1
- package/dist/esm/content/link-preview/types.js.map +1 -1
- package/dist/esm/content/transcript/cache.js +29 -3
- package/dist/esm/content/transcript/cache.js.map +1 -1
- package/dist/esm/content/transcript/index.js +38 -1
- package/dist/esm/content/transcript/index.js.map +1 -1
- package/dist/esm/content/transcript/parse.js +154 -0
- package/dist/esm/content/transcript/parse.js.map +1 -0
- package/dist/esm/content/transcript/providers/generic.js +235 -4
- package/dist/esm/content/transcript/providers/generic.js.map +1 -1
- package/dist/esm/content/transcript/providers/podcast/apple-flow.js +2 -0
- package/dist/esm/content/transcript/providers/podcast/apple-flow.js.map +1 -1
- package/dist/esm/content/transcript/providers/podcast/media.js +21 -20
- package/dist/esm/content/transcript/providers/podcast/media.js.map +1 -1
- package/dist/esm/content/transcript/providers/podcast/rss.js +16 -42
- package/dist/esm/content/transcript/providers/podcast/rss.js.map +1 -1
- package/dist/esm/content/transcript/providers/podcast/spotify-flow.js +1 -0
- package/dist/esm/content/transcript/providers/podcast/spotify-flow.js.map +1 -1
- package/dist/esm/content/transcript/providers/podcast.js +17 -4
- package/dist/esm/content/transcript/providers/podcast.js.map +1 -1
- package/dist/esm/content/transcript/providers/transcription-start.js +50 -0
- package/dist/esm/content/transcript/providers/transcription-start.js.map +1 -0
- package/dist/esm/content/transcript/providers/youtube/api.js +24 -7
- package/dist/esm/content/transcript/providers/youtube/api.js.map +1 -1
- package/dist/esm/content/transcript/providers/youtube/captions.js +132 -4
- package/dist/esm/content/transcript/providers/youtube/captions.js.map +1 -1
- package/dist/esm/content/transcript/providers/youtube/yt-dlp.js +153 -50
- package/dist/esm/content/transcript/providers/youtube/yt-dlp.js.map +1 -1
- package/dist/esm/content/transcript/providers/youtube.js +60 -26
- package/dist/esm/content/transcript/providers/youtube.js.map +1 -1
- package/dist/esm/content/transcript/timestamps.js +79 -0
- package/dist/esm/content/transcript/timestamps.js.map +1 -0
- package/dist/esm/language.js +1 -1
- package/dist/esm/language.js.map +1 -1
- package/dist/esm/processes.js +121 -0
- package/dist/esm/processes.js.map +1 -0
- package/dist/esm/prompts/cli.js +9 -1
- package/dist/esm/prompts/cli.js.map +1 -1
- package/dist/esm/prompts/file.js +34 -0
- package/dist/esm/prompts/file.js.map +1 -1
- package/dist/esm/prompts/index.js +2 -0
- package/dist/esm/prompts/index.js.map +1 -1
- package/dist/esm/prompts/link-summary.js +68 -47
- package/dist/esm/prompts/link-summary.js.map +1 -1
- package/dist/esm/prompts/summary-lengths.js +56 -0
- package/dist/esm/prompts/summary-lengths.js.map +1 -0
- package/dist/esm/prompts/summary-system.js +13 -0
- package/dist/esm/prompts/summary-system.js.map +1 -0
- package/dist/esm/transcription/onnx-cli.js +319 -0
- package/dist/esm/transcription/onnx-cli.js.map +1 -0
- package/dist/esm/transcription/whisper/core.js +72 -4
- package/dist/esm/transcription/whisper/core.js.map +1 -1
- package/dist/esm/transcription/whisper/ffmpeg.js +57 -5
- package/dist/esm/transcription/whisper/ffmpeg.js.map +1 -1
- package/dist/esm/transcription/whisper/whisper-cpp.js +13 -3
- package/dist/esm/transcription/whisper/whisper-cpp.js.map +1 -1
- package/dist/types/content/cache/types.d.ts +24 -0
- package/dist/types/content/index.d.ts +1 -1
- package/dist/types/content/link-preview/client.d.ts +3 -1
- package/dist/types/content/link-preview/content/fetcher.d.ts +5 -1
- package/dist/types/content/link-preview/content/firecrawl.d.ts +3 -1
- package/dist/types/content/link-preview/content/html.d.ts +3 -1
- package/dist/types/content/link-preview/content/types.d.ts +8 -1
- package/dist/types/content/link-preview/content/utils.d.ts +1 -1
- package/dist/types/content/link-preview/deps.d.ts +15 -2
- package/dist/types/content/link-preview/types.d.ts +7 -1
- package/dist/types/content/transcript/cache.d.ts +5 -2
- package/dist/types/content/transcript/index.d.ts +5 -1
- package/dist/types/content/transcript/parse.d.ts +9 -0
- package/dist/types/content/transcript/providers/podcast/media.d.ts +2 -1
- package/dist/types/content/transcript/providers/podcast/rss.d.ts +2 -0
- package/dist/types/content/transcript/providers/transcription-start.d.ts +26 -0
- package/dist/types/content/transcript/providers/youtube/api.d.ts +7 -2
- package/dist/types/content/transcript/providers/youtube/captions.d.ts +11 -1
- package/dist/types/content/transcript/providers/youtube/yt-dlp.d.ts +10 -1
- package/dist/types/content/transcript/timestamps.d.ts +5 -0
- package/dist/types/content/transcript/types.d.ts +7 -1
- package/dist/types/processes.d.ts +50 -0
- package/dist/types/prompts/cli.d.ts +1 -1
- package/dist/types/prompts/file.d.ts +1 -1
- package/dist/types/prompts/index.d.ts +2 -0
- package/dist/types/prompts/link-summary.d.ts +8 -2
- package/dist/types/prompts/summary-lengths.d.ts +15 -0
- package/dist/types/prompts/summary-system.d.ts +1 -0
- package/dist/types/transcription/onnx-cli.d.ts +25 -0
- package/dist/types/transcription/whisper/core.d.ts +6 -2
- package/dist/types/transcription/whisper/ffmpeg.d.ts +5 -0
- package/dist/types/transcription/whisper/types.d.ts +1 -1
- package/package.json +11 -6
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"types.js","sourceRoot":"","sources":["../../../../src/content/link-preview/types.ts"],"names":[],"mappings":"
|
|
1
|
+
{"version":3,"file":"types.js","sourceRoot":"","sources":["../../../../src/content/link-preview/types.ts"],"names":[],"mappings":"AAkBA,MAAM,CAAC,MAAM,WAAW,GAAG,CAAC,SAAS,EAAE,QAAQ,CAAU,CAAA"}
|
|
@@ -1,7 +1,9 @@
|
|
|
1
1
|
export const DEFAULT_TTL_MS = 1000 * 60 * 60 * 24 * 7;
|
|
2
2
|
export const NEGATIVE_TTL_MS = 1000 * 60 * 60 * 6;
|
|
3
|
-
export const readTranscriptCache = async ({ url, cacheMode, transcriptCache, }) => {
|
|
4
|
-
const cached = transcriptCache
|
|
3
|
+
export const readTranscriptCache = async ({ url, cacheMode, transcriptCache, transcriptTimestamps = false, fileMtime, }) => {
|
|
4
|
+
const cached = transcriptCache
|
|
5
|
+
? await transcriptCache.get({ url, fileMtime: fileMtime ?? null })
|
|
6
|
+
: null;
|
|
5
7
|
const diagnostics = buildBaseDiagnostics(cacheMode);
|
|
6
8
|
if (!cached) {
|
|
7
9
|
return { cached: null, resolution: null, diagnostics };
|
|
@@ -21,10 +23,23 @@ export const readTranscriptCache = async ({ url, cacheMode, transcriptCache, })
|
|
|
21
23
|
}
|
|
22
24
|
diagnostics.cacheStatus = 'hit';
|
|
23
25
|
diagnostics.notes = appendNote(diagnostics.notes, 'Served transcript from cache');
|
|
26
|
+
const cachedSegments = extractSegments(cached.metadata);
|
|
27
|
+
const hasSegments = Boolean(cachedSegments && cachedSegments.length > 0);
|
|
28
|
+
const timestampsFlag = cached.metadata?.timestamps;
|
|
29
|
+
if (transcriptTimestamps &&
|
|
30
|
+
timestampsFlag !== false &&
|
|
31
|
+
(cachedSegments == null || cachedSegments.length === 0)) {
|
|
32
|
+
diagnostics.notes = appendNote(diagnostics.notes, 'Cached transcript missing timestamps; fetching fresh copy');
|
|
33
|
+
return { cached, resolution: null, diagnostics };
|
|
34
|
+
}
|
|
35
|
+
if (transcriptTimestamps && timestampsFlag === false) {
|
|
36
|
+
diagnostics.notes = appendNote(diagnostics.notes, 'Transcript timestamps unavailable');
|
|
37
|
+
}
|
|
24
38
|
const resolution = {
|
|
25
39
|
text: cached.content,
|
|
26
40
|
source: provider,
|
|
27
41
|
metadata: cached.metadata ?? null,
|
|
42
|
+
segments: transcriptTimestamps && hasSegments ? cachedSegments : null,
|
|
28
43
|
};
|
|
29
44
|
return { cached, resolution, diagnostics };
|
|
30
45
|
};
|
|
@@ -47,6 +62,7 @@ export const mapCachedSource = (source) => {
|
|
|
47
62
|
return null;
|
|
48
63
|
if (source === 'youtubei' ||
|
|
49
64
|
source === 'captionTracks' ||
|
|
65
|
+
source === 'embedded' ||
|
|
50
66
|
source === 'yt-dlp' ||
|
|
51
67
|
source === 'podcastTranscript' ||
|
|
52
68
|
source === 'whisper' ||
|
|
@@ -57,7 +73,7 @@ export const mapCachedSource = (source) => {
|
|
|
57
73
|
}
|
|
58
74
|
return 'unknown';
|
|
59
75
|
};
|
|
60
|
-
export const writeTranscriptCache = async ({ url, service, resourceKey, result, transcriptCache, }) => {
|
|
76
|
+
export const writeTranscriptCache = async ({ url, service, resourceKey, result, transcriptCache, fileMtime, }) => {
|
|
61
77
|
if (!transcriptCache) {
|
|
62
78
|
return;
|
|
63
79
|
}
|
|
@@ -74,6 +90,16 @@ export const writeTranscriptCache = async ({ url, service, resourceKey, result,
|
|
|
74
90
|
content: result.text,
|
|
75
91
|
source: resolvedSource,
|
|
76
92
|
metadata: result.metadata ?? null,
|
|
93
|
+
fileMtime,
|
|
77
94
|
});
|
|
78
95
|
};
|
|
96
|
+
function extractSegments(metadata) {
|
|
97
|
+
if (!metadata)
|
|
98
|
+
return null;
|
|
99
|
+
const segments = metadata.segments;
|
|
100
|
+
if (!Array.isArray(segments))
|
|
101
|
+
return null;
|
|
102
|
+
const normalized = segments.filter((segment) => segment && typeof segment === 'object');
|
|
103
|
+
return normalized.length > 0 ? normalized : null;
|
|
104
|
+
}
|
|
79
105
|
//# sourceMappingURL=cache.js.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"cache.js","sourceRoot":"","sources":["../../../../src/content/transcript/cache.ts"],"names":[],"mappings":"AAQA,MAAM,CAAC,MAAM,cAAc,GAAG,IAAI,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AACrD,MAAM,CAAC,MAAM,eAAe,GAAG,IAAI,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;
|
|
1
|
+
{"version":3,"file":"cache.js","sourceRoot":"","sources":["../../../../src/content/transcript/cache.ts"],"names":[],"mappings":"AAQA,MAAM,CAAC,MAAM,cAAc,GAAG,IAAI,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AACrD,MAAM,CAAC,MAAM,eAAe,GAAG,IAAI,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AAqBjD,MAAM,CAAC,MAAM,mBAAmB,GAAG,KAAK,EAAE,EACxC,GAAG,EACH,SAAS,EACT,eAAe,EACf,oBAAoB,GAAG,KAAK,EAC5B,SAAS,GACU,EAAkC,EAAE;IACvD,MAAM,MAAM,GAAG,eAAe;QAC5B,CAAC,CAAC,MAAM,eAAe,CAAC,GAAG,CAAC,EAAE,GAAG,EAAE,SAAS,EAAE,SAAS,IAAI,IAAI,EAAE,CAAC;QAClE,CAAC,CAAC,IAAI,CAAA;IACR,MAAM,WAAW,GAAG,oBAAoB,CAAC,SAAS,CAAC,CAAA;IAEnD,IAAI,CAAC,MAAM,EAAE,CAAC;QACZ,OAAO,EAAE,MAAM,EAAE,IAAI,EAAE,UAAU,EAAE,IAAI,EAAE,WAAW,EAAE,CAAA;IACxD,CAAC;IAED,MAAM,QAAQ,GAAG,eAAe,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA;IAC/C,WAAW,CAAC,QAAQ,GAAG,QAAQ,CAAA;IAC/B,WAAW,CAAC,kBAAkB,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,EAAE,CAAA;IAC3D,WAAW,CAAC,YAAY,GAAG,OAAO,CAAC,MAAM,CAAC,OAAO,IAAI,MAAM,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IAE/E,IAAI,SAAS,KAAK,QAAQ,EAAE,CAAC;QAC3B,WAAW,CAAC,KAAK,GAAG,UAAU,CAC5B,WAAW,CAAC,KAAK,EACjB,iDAAiD,CAClD,CAAA;QACD,OAAO,EAAE,MAAM,EAAE,UAAU,EAAE,IAAI,EAAE,WAAW,EAAE,CAAA;IAClD,CAAC;IAED,IAAI,MAAM,CAAC,OAAO,EAAE,CAAC;QACnB,WAAW,CAAC,WAAW,GAAG,SAAS,CAAA;QACnC,WAAW,CAAC,KAAK,GAAG,UAAU,CAC5B,WAAW,CAAC,KAAK,EACjB,gDAAgD,CACjD,CAAA;QACD,OAAO,EAAE,MAAM,EAAE,UAAU,EAAE,IAAI,EAAE,WAAW,EAAE,CAAA;IAClD,CAAC;IAED,WAAW,CAAC,WAAW,GAAG,KAAK,CAAA;IAC/B,WAAW,CAAC,KAAK,GAAG,UAAU,CAAC,WAAW,CAAC,KAAK,EAAE,8BAA8B,CAAC,CAAA;IAEjF,MAAM,cAAc,GAAG,eAAe,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAA;IACvD,MAAM,WAAW,GAAG,OAAO,CAAC,cAAc,IAAI,cAAc,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IACxE,MAAM,cAAc,GAAG,MAAM,CAAC,QAAQ,EAAE,UAAU,CAAA;IAClD,IACE,oBAAoB;QACpB,cAAc,KAAK,KAAK;QACxB,CAAC,cAAc,IAAI,IAAI,IAAI,cAAc,CAAC,MAAM,KAAK,CAAC,CAAC,EACvD,CAAC;QACD,WAAW,CAAC,KAAK,GAAG,UAAU,CAC5B,WAAW,CAAC,KAAK,EACjB,2DAA2D,CAC5D,CAAA;QACD,OAAO,EAAE,MAAM,EAAE,UAAU,EAAE,IAAI,EAAE,WAAW,EAAE,CAAA;IAClD,CAAC;IACD,IAAI,oBAAoB,IAAI,cAAc,KAAK,KAAK,EAAE,CAAC;QACrD,WAAW,CAAC,KAAK,GAAG,UAAU,CAAC,WAAW,CAAC,KAAK,EAAE,mCAAmC,CAAC,CAAA;IACxF,CAAC;IAED,MAAM,UAAU,GAAyB;QACvC,IAAI,EAAE,MAAM,CAAC,OAAO;QACpB,MAAM,EAAE,QAAQ;QAChB,QAAQ,EAAE,MAAM,CAAC,QAAQ,IAAI,IAAI;QACjC,QAAQ,EAAE,oBAAoB,IAAI,WAAW,CAAC,CAAC,CAAC,cAAc,CAAC,CAAC,CAAC,IAAI;KACtE,CAAA;IACD,OAAO,EAAE,MAAM,EAAE,UAAU,EAAE,WAAW,EAAE,CAAA;AAC5C,CAAC,CAAA;AAED,MAAM,oBAAoB,GAAG,CAAC,SAAoB,EAAoB,EAAE,CAAC,CAAC;IACxE,SAAS;IACT,WAAW,EAAE,SAAS,KAAK,QAAQ,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,MAAM;IACzD,QAAQ,EAAE,IAAI;IACd,kBAAkB,EAAE,EAAE;IACtB,YAAY,EAAE,KAAK;IACnB,KAAK,EAAE,SAAS,KAAK,QAAQ,CAAC,CAAC,CAAC,wBAAwB,CAAC,CAAC,CAAC,IAAI;CAChE,CAAC,CAAA;AAEF,MAAM,UAAU,GAAG,CAAC,QAAmC,EAAE,IAAY,EAAU,EAAE;IAC/E,IAAI,CAAC,QAAQ,EAAE,CAAC;QACd,OAAO,IAAI,CAAA;IACb,CAAC;IACD,OAAO,GAAG,QAAQ,KAAK,IAAI,EAAE,CAAA;AAC/B,CAAC,CAAA;AAED,MAAM,CAAC,MAAM,eAAe,GAAG,CAAC,MAAqB,EAA2B,EAAE;IAChF,IAAI,MAAM,KAAK,IAAI;QAAE,OAAO,IAAI,CAAA;IAChC,IACE,MAAM,KAAK,UAAU;QACrB,MAAM,KAAK,eAAe;QAC1B,MAAM,KAAK,UAAU;QACrB,MAAM,KAAK,QAAQ;QACnB,MAAM,KAAK,mBAAmB;QAC9B,MAAM,KAAK,SAAS;QACpB,MAAM,KAAK,OAAO;QAClB,MAAM,KAAK,MAAM;QACjB,MAAM,KAAK,aAAa,EACxB,CAAC;QACD,OAAO,MAAM,CAAA;IACf,CAAC;IACD,OAAO,SAAS,CAAA;AAClB,CAAC,CAAA;AAED,MAAM,CAAC,MAAM,oBAAoB,GAAG,KAAK,EAAE,EACzC,GAAG,EACH,OAAO,EACP,WAAW,EACX,MAAM,EACN,eAAe,EACf,SAAS,GAYV,EAAiB,EAAE;IAClB,IAAI,CAAC,eAAe,EAAE,CAAC;QACrB,OAAM;IACR,CAAC;IAED,IAAI,MAAM,CAAC,MAAM,KAAK,IAAI,IAAI,MAAM,CAAC,IAAI,KAAK,IAAI,EAAE,CAAC;QACnD,OAAM;IACR,CAAC;IAED,MAAM,KAAK,GAAG,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,cAAc,CAAC,CAAC,CAAC,eAAe,CAAA;IAC5D,MAAM,cAAc,GAAG,MAAM,CAAC,MAAM,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,aAAa,CAAC,CAAA;IAEjF,MAAM,eAAe,CAAC,GAAG,CAAC;QACxB,GAAG;QACH,OAAO;QACP,WAAW;QACX,KAAK;QACL,OAAO,EAAE,MAAM,CAAC,IAAI;QACpB,MAAM,EAAE,cAAc;QACtB,QAAQ,EAAE,MAAM,CAAC,QAAQ,IAAI,IAAI;QACjC,SAAS;KACV,CAAC,CAAA;AACJ,CAAC,CAAA;AAED,SAAS,eAAe,CAAC,QAAoD;IAC3E,IAAI,CAAC,QAAQ;QAAE,OAAO,IAAI,CAAA;IAC1B,MAAM,QAAQ,GAAI,QAAmC,CAAC,QAAQ,CAAA;IAC9D,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,QAAQ,CAAC;QAAE,OAAO,IAAI,CAAA;IACzC,MAAM,UAAU,GAAG,QAAQ,CAAC,MAAM,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,IAAI,OAAO,OAAO,KAAK,QAAQ,CAAC,CAAA;IACvF,OAAO,UAAU,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAE,UAA+C,CAAC,CAAC,CAAC,IAAI,CAAA;AACxF,CAAC"}
|
|
@@ -9,7 +9,7 @@ const PROVIDERS = [
|
|
|
9
9
|
{ id: 'generic', canHandle: canHandleGeneric, fetchTranscript: fetchGeneric },
|
|
10
10
|
];
|
|
11
11
|
const GENERIC_PROVIDER_ID = 'generic';
|
|
12
|
-
export const resolveTranscriptForLink = async (url, html, deps, { youtubeTranscriptMode, cacheMode: providedCacheMode } = {}) => {
|
|
12
|
+
export const resolveTranscriptForLink = async (url, html, deps, { youtubeTranscriptMode, mediaTranscriptMode, mediaKindHint, transcriptTimestamps, cacheMode: providedCacheMode, fileMtime, } = {}) => {
|
|
13
13
|
const normalizedUrl = url.trim();
|
|
14
14
|
const embeddedYoutubeUrl = !isYouTubeUrlInternal(normalizedUrl) && html
|
|
15
15
|
? await extractEmbeddedYouTubeUrlFromHtml(html)
|
|
@@ -23,6 +23,8 @@ export const resolveTranscriptForLink = async (url, html, deps, { youtubeTranscr
|
|
|
23
23
|
url: normalizedUrl,
|
|
24
24
|
cacheMode,
|
|
25
25
|
transcriptCache: deps.transcriptCache,
|
|
26
|
+
transcriptTimestamps: Boolean(transcriptTimestamps),
|
|
27
|
+
fileMtime: fileMtime ?? null,
|
|
26
28
|
});
|
|
27
29
|
const diagnostics = {
|
|
28
30
|
cacheMode,
|
|
@@ -51,14 +53,19 @@ export const resolveTranscriptForLink = async (url, html, deps, { youtubeTranscr
|
|
|
51
53
|
}
|
|
52
54
|
const providerResult = await executeProvider(provider, baseContext, {
|
|
53
55
|
fetch: deps.fetch,
|
|
56
|
+
env: deps.env,
|
|
54
57
|
scrapeWithFirecrawl: deps.scrapeWithFirecrawl,
|
|
55
58
|
apifyApiToken: deps.apifyApiToken,
|
|
56
59
|
ytDlpPath: deps.ytDlpPath,
|
|
57
60
|
falApiKey: deps.falApiKey,
|
|
58
61
|
openaiApiKey: deps.openaiApiKey,
|
|
62
|
+
mediaCache: deps.mediaCache ?? null,
|
|
59
63
|
resolveTwitterCookies: deps.resolveTwitterCookies ?? null,
|
|
60
64
|
onProgress: deps.onProgress ?? null,
|
|
61
65
|
youtubeTranscriptMode: youtubeTranscriptMode ?? 'auto',
|
|
66
|
+
mediaTranscriptMode: mediaTranscriptMode ?? 'auto',
|
|
67
|
+
mediaKindHint: mediaKindHint ?? null,
|
|
68
|
+
transcriptTimestamps: transcriptTimestamps ?? false,
|
|
62
69
|
});
|
|
63
70
|
if (shouldReportProgress) {
|
|
64
71
|
deps.onProgress?.({
|
|
@@ -77,12 +84,30 @@ export const resolveTranscriptForLink = async (url, html, deps, { youtubeTranscr
|
|
|
77
84
|
diagnostics.notes = appendNote(diagnostics.notes, providerResult.notes);
|
|
78
85
|
}
|
|
79
86
|
if (providerResult.source !== null || providerResult.text !== null) {
|
|
87
|
+
if (transcriptTimestamps) {
|
|
88
|
+
const nextMeta = { ...(providerResult.metadata ?? {}) };
|
|
89
|
+
if (providerResult.segments && providerResult.segments.length > 0) {
|
|
90
|
+
nextMeta.timestamps = true;
|
|
91
|
+
nextMeta.segments = providerResult.segments;
|
|
92
|
+
}
|
|
93
|
+
else if (nextMeta.timestamps == null) {
|
|
94
|
+
nextMeta.timestamps = false;
|
|
95
|
+
}
|
|
96
|
+
providerResult.metadata = nextMeta;
|
|
97
|
+
}
|
|
98
|
+
else if (providerResult.segments && providerResult.segments.length > 0) {
|
|
99
|
+
providerResult.metadata = {
|
|
100
|
+
...(providerResult.metadata ?? {}),
|
|
101
|
+
segments: providerResult.segments,
|
|
102
|
+
};
|
|
103
|
+
}
|
|
80
104
|
await writeTranscriptCache({
|
|
81
105
|
url: normalizedUrl,
|
|
82
106
|
service: provider.id,
|
|
83
107
|
resourceKey,
|
|
84
108
|
result: providerResult,
|
|
85
109
|
transcriptCache: deps.transcriptCache,
|
|
110
|
+
fileMtime,
|
|
86
111
|
});
|
|
87
112
|
}
|
|
88
113
|
if (!providerResult.text && cacheOutcome.cached?.content && cacheMode !== 'bypass') {
|
|
@@ -95,6 +120,9 @@ export const resolveTranscriptForLink = async (url, html, deps, { youtubeTranscr
|
|
|
95
120
|
source: diagnostics.provider,
|
|
96
121
|
metadata: cacheOutcome.cached.metadata ?? null,
|
|
97
122
|
diagnostics,
|
|
123
|
+
segments: transcriptTimestamps
|
|
124
|
+
? resolveSegmentsFromMetadata(cacheOutcome.cached.metadata)
|
|
125
|
+
: null,
|
|
98
126
|
};
|
|
99
127
|
}
|
|
100
128
|
return {
|
|
@@ -102,6 +130,7 @@ export const resolveTranscriptForLink = async (url, html, deps, { youtubeTranscr
|
|
|
102
130
|
source: providerResult.source,
|
|
103
131
|
metadata: providerResult.metadata ?? null,
|
|
104
132
|
diagnostics,
|
|
133
|
+
segments: transcriptTimestamps ? (providerResult.segments ?? null) : null,
|
|
105
134
|
};
|
|
106
135
|
};
|
|
107
136
|
const extractResourceKey = (url) => {
|
|
@@ -128,4 +157,12 @@ const appendNote = (existing, next) => {
|
|
|
128
157
|
}
|
|
129
158
|
return `${existing}; ${next}`;
|
|
130
159
|
};
|
|
160
|
+
const resolveSegmentsFromMetadata = (metadata) => {
|
|
161
|
+
if (!metadata)
|
|
162
|
+
return null;
|
|
163
|
+
const segments = metadata.segments;
|
|
164
|
+
return Array.isArray(segments) && segments.length > 0
|
|
165
|
+
? segments
|
|
166
|
+
: null;
|
|
167
|
+
};
|
|
131
168
|
//# sourceMappingURL=index.js.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"index.js","sourceRoot":"","sources":["../../../../src/content/transcript/index.ts"],"names":[],"mappings":"AAMA,OAAO,EAAE,eAAe,EAAE,mBAAmB,EAAE,oBAAoB,EAAE,MAAM,YAAY,CAAA;AACvF,OAAO,EACL,SAAS,IAAI,gBAAgB,EAC7B,eAAe,IAAI,YAAY,GAChC,MAAM,wBAAwB,CAAA;AAC/B,OAAO,EACL,SAAS,IAAI,gBAAgB,EAC7B,eAAe,IAAI,YAAY,GAChC,MAAM,wBAAwB,CAAA;AAC/B,OAAO,EACL,SAAS,IAAI,gBAAgB,EAC7B,eAAe,IAAI,YAAY,GAChC,MAAM,wBAAwB,CAAA;AAO/B,OAAO,EACL,iCAAiC,EACjC,qBAAqB,IAAI,6BAA6B,EACtD,YAAY,IAAI,oBAAoB,GACrC,MAAM,YAAY,CAAA;
|
|
1
|
+
{"version":3,"file":"index.js","sourceRoot":"","sources":["../../../../src/content/transcript/index.ts"],"names":[],"mappings":"AAMA,OAAO,EAAE,eAAe,EAAE,mBAAmB,EAAE,oBAAoB,EAAE,MAAM,YAAY,CAAA;AACvF,OAAO,EACL,SAAS,IAAI,gBAAgB,EAC7B,eAAe,IAAI,YAAY,GAChC,MAAM,wBAAwB,CAAA;AAC/B,OAAO,EACL,SAAS,IAAI,gBAAgB,EAC7B,eAAe,IAAI,YAAY,GAChC,MAAM,wBAAwB,CAAA;AAC/B,OAAO,EACL,SAAS,IAAI,gBAAgB,EAC7B,eAAe,IAAI,YAAY,GAChC,MAAM,wBAAwB,CAAA;AAO/B,OAAO,EACL,iCAAiC,EACjC,qBAAqB,IAAI,6BAA6B,EACtD,YAAY,IAAI,oBAAoB,GACrC,MAAM,YAAY,CAAA;AAWnB,MAAM,SAAS,GAAqB;IAClC,EAAE,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,gBAAgB,EAAE,eAAe,EAAE,YAAY,EAAE;IAC7E,EAAE,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,gBAAgB,EAAE,eAAe,EAAE,YAAY,EAAE;IAC7E,EAAE,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,gBAAgB,EAAE,eAAe,EAAE,YAAY,EAAE;CAC9E,CAAA;AACD,MAAM,mBAAmB,GAAG,SAAS,CAAA;AAErC,MAAM,CAAC,MAAM,wBAAwB,GAAG,KAAK,EAC3C,GAAW,EACX,IAAmB,EACnB,IAAqB,EACrB,EACE,qBAAqB,EACrB,mBAAmB,EACnB,aAAa,EACb,oBAAoB,EACpB,SAAS,EAAE,iBAAiB,EAC5B,SAAS,MACmB,EAAE,EACD,EAAE;IACjC,MAAM,aAAa,GAAG,GAAG,CAAC,IAAI,EAAE,CAAA;IAChC,MAAM,kBAAkB,GACtB,CAAC,oBAAoB,CAAC,aAAa,CAAC,IAAI,IAAI;QAC1C,CAAC,CAAC,MAAM,iCAAiC,CAAC,IAAI,CAAC;QAC/C,CAAC,CAAC,IAAI,CAAA;IACV,MAAM,YAAY,GAAG,kBAAkB,IAAI,aAAa,CAAA;IACxD,MAAM,WAAW,GAAG,kBAAkB,CAAC,YAAY,CAAC,CAAA;IACpD,MAAM,WAAW,GAAoB,EAAE,GAAG,EAAE,YAAY,EAAE,IAAI,EAAE,WAAW,EAAE,CAAA;IAC7E,MAAM,QAAQ,GAAmB,cAAc,CAAC,WAAW,CAAC,CAAA;IAC5D,MAAM,SAAS,GAAc,iBAAiB,IAAI,SAAS,CAAA;IAE3D,MAAM,YAAY,GAAG,MAAM,mBAAmB,CAAC;QAC7C,GAAG,EAAE,aAAa;QAClB,SAAS;QACT,eAAe,EAAE,IAAI,CAAC,eAAe;QACrC,oBAAoB,EAAE,OAAO,CAAC,oBAAoB,CAAC;QACnD,SAAS,EAAE,SAAS,IAAI,IAAI;KAC7B,CAAC,CAAA;IAEF,MAAM,WAAW,GAA0B;QACzC,SAAS;QACT,WAAW,EAAE,YAAY,CAAC,WAAW,CAAC,WAAW;QACjD,YAAY,EAAE,YAAY,CAAC,WAAW,CAAC,YAAY;QACnD,QAAQ,EAAE,YAAY,CAAC,WAAW,CAAC,QAAQ;QAC3C,kBAAkB,EAAE,EAAE;QACtB,KAAK,EAAE,YAAY,CAAC,WAAW,CAAC,KAAK,IAAI,IAAI;KAC9C,CAAA;IAED,IAAI,YAAY,CAAC,UAAU,EAAE,CAAC;QAC5B,OAAO;YACL,GAAG,YAAY,CAAC,UAAU;YAC1B,WAAW;SACZ,CAAA;IACH,CAAC;IAED,MAAM,oBAAoB,GAAG,QAAQ,CAAC,EAAE,KAAK,SAAS,IAAI,QAAQ,CAAC,EAAE,KAAK,SAAS,CAAA;IACnF,IAAI,oBAAoB,EAAE,CAAC;QACzB,IAAI,CAAC,UAAU,EAAE,CAAC;YAChB,IAAI,EAAE,kBAAkB;YACxB,GAAG,EAAE,aAAa;YAClB,OAAO,EAAE,QAAQ,CAAC,EAAE;YACpB,IAAI,EACF,QAAQ,CAAC,EAAE,KAAK,SAAS;gBACvB,CAAC,CAAC,+BAA+B;gBACjC,CAAC,CAAC,+BAA+B;SACtC,CAAC,CAAA;IACJ,CAAC;IAED,MAAM,cAAc,GAAG,MAAM,eAAe,CAAC,QAAQ,EAAE,WAAW,EAAE;QAClE,KAAK,EAAE,IAAI,CAAC,KAAK;QACjB,GAAG,EAAE,IAAI,CAAC,GAAG;QACb,mBAAmB,EAAE,IAAI,CAAC,mBAAmB;QAC7C,aAAa,EAAE,IAAI,CAAC,aAAa;QACjC,SAAS,EAAE,IAAI,CAAC,SAAS;QACzB,SAAS,EAAE,IAAI,CAAC,SAAS;QACzB,YAAY,EAAE,IAAI,CAAC,YAAY;QAC/B,UAAU,EAAE,IAAI,CAAC,UAAU,IAAI,IAAI;QACnC,qBAAqB,EAAE,IAAI,CAAC,qBAAqB,IAAI,IAAI;QACzD,UAAU,EAAE,IAAI,CAAC,UAAU,IAAI,IAAI;QACnC,qBAAqB,EAAE,qBAAqB,IAAI,MAAM;QACtD,mBAAmB,EAAE,mBAAmB,IAAI,MAAM;QAClD,aAAa,EAAE,aAAa,IAAI,IAAI;QACpC,oBAAoB,EAAE,oBAAoB,IAAI,KAAK;KACpD,CAAC,CAAA;IAEF,IAAI,oBAAoB,EAAE,CAAC;QACzB,IAAI,CAAC,UAAU,EAAE,CAAC;YAChB,IAAI,EAAE,iBAAiB;YACvB,GAAG,EAAE,aAAa;YAClB,EAAE,EAAE,OAAO,CAAC,cAAc,CAAC,IAAI,IAAI,cAAc,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC;YAClE,OAAO,EAAE,QAAQ,CAAC,EAAE;YACpB,MAAM,EAAE,cAAc,CAAC,MAAM;YAC7B,IAAI,EAAE,cAAc,CAAC,MAAM,CAAC,CAAC,CAAC,GAAG,QAAQ,CAAC,EAAE,IAAI,cAAc,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC,QAAQ,CAAC,EAAE;SACtF,CAAC,CAAA;IACJ,CAAC;IAED,WAAW,CAAC,QAAQ,GAAG,cAAc,CAAC,MAAM,CAAA;IAC5C,WAAW,CAAC,kBAAkB,GAAG,cAAc,CAAC,kBAAkB,CAAA;IAClE,WAAW,CAAC,YAAY,GAAG,OAAO,CAAC,cAAc,CAAC,IAAI,IAAI,cAAc,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IACzF,IAAI,cAAc,CAAC,KAAK,EAAE,CAAC;QACzB,WAAW,CAAC,KAAK,GAAG,UAAU,CAAC,WAAW,CAAC,KAAK,EAAE,cAAc,CAAC,KAAK,CAAC,CAAA;IACzE,CAAC;IAED,IAAI,cAAc,CAAC,MAAM,KAAK,IAAI,IAAI,cAAc,CAAC,IAAI,KAAK,IAAI,EAAE,CAAC;QACnE,IAAI,oBAAoB,EAAE,CAAC;YACzB,MAAM,QAAQ,GAAG,EAAE,GAAG,CAAC,cAAc,CAAC,QAAQ,IAAI,EAAE,CAAC,EAAE,CAAA;YACvD,IAAI,cAAc,CAAC,QAAQ,IAAI,cAAc,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAClE,QAAQ,CAAC,UAAU,GAAG,IAAI,CAAA;gBAC1B,QAAQ,CAAC,QAAQ,GAAG,cAAc,CAAC,QAAQ,CAAA;YAC7C,CAAC;iBAAM,IAAI,QAAQ,CAAC,UAAU,IAAI,IAAI,EAAE,CAAC;gBACvC,QAAQ,CAAC,UAAU,GAAG,KAAK,CAAA;YAC7B,CAAC;YACD,cAAc,CAAC,QAAQ,GAAG,QAAQ,CAAA;QACpC,CAAC;aAAM,IAAI,cAAc,CAAC,QAAQ,IAAI,cAAc,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACzE,cAAc,CAAC,QAAQ,GAAG;gBACxB,GAAG,CAAC,cAAc,CAAC,QAAQ,IAAI,EAAE,CAAC;gBAClC,QAAQ,EAAE,cAAc,CAAC,QAAQ;aAClC,CAAA;QACH,CAAC;QACD,MAAM,oBAAoB,CAAC;YACzB,GAAG,EAAE,aAAa;YAClB,OAAO,EAAE,QAAQ,CAAC,EAAE;YACpB,WAAW;YACX,MAAM,EAAE,cAAc;YACtB,eAAe,EAAE,IAAI,CAAC,eAAe;YACrC,SAAS;SACV,CAAC,CAAA;IACJ,CAAC;IAED,IAAI,CAAC,cAAc,CAAC,IAAI,IAAI,YAAY,CAAC,MAAM,EAAE,OAAO,IAAI,SAAS,KAAK,QAAQ,EAAE,CAAC;QACnF,WAAW,CAAC,WAAW,GAAG,UAAU,CAAA;QACpC,WAAW,CAAC,QAAQ,GAAG,eAAe,CAAC,YAAY,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA;QAClE,WAAW,CAAC,YAAY,GAAG,OAAO,CAChC,YAAY,CAAC,MAAM,CAAC,OAAO,IAAI,YAAY,CAAC,MAAM,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CACtE,CAAA;QACD,WAAW,CAAC,KAAK,GAAG,UAAU,CAC5B,WAAW,CAAC,KAAK,EACjB,+DAA+D,CAChE,CAAA;QAED,OAAO;YACL,IAAI,EAAE,YAAY,CAAC,MAAM,CAAC,OAAO;YACjC,MAAM,EAAE,WAAW,CAAC,QAAQ;YAC5B,QAAQ,EAAE,YAAY,CAAC,MAAM,CAAC,QAAQ,IAAI,IAAI;YAC9C,WAAW;YACX,QAAQ,EAAE,oBAAoB;gBAC5B,CAAC,CAAC,2BAA2B,CAAC,YAAY,CAAC,MAAM,CAAC,QAAQ,CAAC;gBAC3D,CAAC,CAAC,IAAI;SACT,CAAA;IACH,CAAC;IAED,OAAO;QACL,IAAI,EAAE,cAAc,CAAC,IAAI;QACzB,MAAM,EAAE,cAAc,CAAC,MAAM;QAC7B,QAAQ,EAAE,cAAc,CAAC,QAAQ,IAAI,IAAI;QACzC,WAAW;QACX,QAAQ,EAAE,oBAAoB,CAAC,CAAC,CAAC,CAAC,cAAc,CAAC,QAAQ,IAAI,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI;KAC1E,CAAA;AACH,CAAC,CAAA;AAED,MAAM,kBAAkB,GAAG,CAAC,GAAW,EAAiB,EAAE;IACxD,IAAI,oBAAoB,CAAC,GAAG,CAAC,EAAE,CAAC;QAC9B,OAAO,6BAA6B,CAAC,GAAG,CAAC,CAAA;IAC3C,CAAC;IACD,OAAO,IAAI,CAAA;AACb,CAAC,CAAA;AAED,MAAM,cAAc,GAAG,CAAC,OAAwB,EAAkB,EAAE;IAClE,MAAM,qBAAqB,GAAG,SAAS,CAAC,IAAI,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,QAAQ,CAAC,EAAE,KAAK,mBAAmB,CAAC,CAAA;IAE/F,MAAM,mBAAmB,GAAG,SAAS,CAAC,IAAI,CACxC,CAAC,QAAQ,EAAE,EAAE,CAAC,QAAQ,CAAC,EAAE,KAAK,mBAAmB,IAAI,QAAQ,CAAC,SAAS,CAAC,OAAO,CAAC,CACjF,CAAA;IACD,IAAI,mBAAmB,EAAE,CAAC;QACxB,OAAO,mBAAmB,CAAA;IAC5B,CAAC;IAED,IAAI,qBAAqB,EAAE,CAAC;QAC1B,OAAO,qBAAqB,CAAA;IAC9B,CAAC;IAED,MAAM,IAAI,KAAK,CAAC,+CAA+C,CAAC,CAAA;AAClE,CAAC,CAAA;AAED,MAAM,eAAe,GAAG,KAAK,EAC3B,QAAwB,EACxB,OAAwB,EACxB,OAA6B,EACJ,EAAE,CAAC,QAAQ,CAAC,eAAe,CAAC,OAAO,EAAE,OAAO,CAAC,CAAA;AAExE,MAAM,UAAU,GAAG,CAAC,QAAmC,EAAE,IAAY,EAAU,EAAE;IAC/E,IAAI,CAAC,QAAQ,EAAE,CAAC;QACd,OAAO,IAAI,CAAA;IACb,CAAC;IACD,OAAO,GAAG,QAAQ,KAAK,IAAI,EAAE,CAAA;AAC/B,CAAC,CAAA;AAED,MAAM,2BAA2B,GAAG,CAAC,QAAyC,EAAE,EAAE;IAChF,IAAI,CAAC,QAAQ;QAAE,OAAO,IAAI,CAAA;IAC1B,MAAM,QAAQ,GAAI,QAAmC,CAAC,QAAQ,CAAA;IAC9D,OAAO,KAAK,CAAC,OAAO,CAAC,QAAQ,CAAC,IAAI,QAAQ,CAAC,MAAM,GAAG,CAAC;QACnD,CAAC,CAAE,QAA6C;QAChD,CAAC,CAAC,IAAI,CAAA;AACV,CAAC,CAAA"}
|
|
@@ -0,0 +1,154 @@
|
|
|
1
|
+
import { parseTimestampStringToMs, parseTimestampToMs } from './timestamps.js';
|
|
2
|
+
export function vttToSegments(raw) {
|
|
3
|
+
const normalized = raw.replace(/\r\n/g, '\n');
|
|
4
|
+
const lines = normalized.split('\n');
|
|
5
|
+
const segments = [];
|
|
6
|
+
let idx = 0;
|
|
7
|
+
while (idx < lines.length) {
|
|
8
|
+
const line = lines[idx]?.trim() ?? '';
|
|
9
|
+
if (!line || line.toUpperCase() === 'WEBVTT' || /^(NOTE|STYLE|REGION)\b/i.test(line)) {
|
|
10
|
+
idx += 1;
|
|
11
|
+
continue;
|
|
12
|
+
}
|
|
13
|
+
if (!line.includes('-->')) {
|
|
14
|
+
idx += 1;
|
|
15
|
+
continue;
|
|
16
|
+
}
|
|
17
|
+
const [startRaw, rest] = line.split('-->');
|
|
18
|
+
const endRaw = rest?.trim().split(/\s+/)[0] ?? '';
|
|
19
|
+
const startMs = parseTimestampStringToMs(startRaw?.trim() ?? '');
|
|
20
|
+
const endMs = parseTimestampStringToMs(endRaw);
|
|
21
|
+
idx += 1;
|
|
22
|
+
const textLines = [];
|
|
23
|
+
while (idx < lines.length) {
|
|
24
|
+
const cueLine = lines[idx];
|
|
25
|
+
if (!cueLine || cueLine.trim().length === 0)
|
|
26
|
+
break;
|
|
27
|
+
if (!/^(NOTE|STYLE|REGION)\b/i.test(cueLine.trim())) {
|
|
28
|
+
textLines.push(cueLine.trim());
|
|
29
|
+
}
|
|
30
|
+
idx += 1;
|
|
31
|
+
}
|
|
32
|
+
idx += 1;
|
|
33
|
+
if (startMs == null)
|
|
34
|
+
continue;
|
|
35
|
+
const text = textLines.join(' ').replace(/\s+/g, ' ').trim();
|
|
36
|
+
if (!text)
|
|
37
|
+
continue;
|
|
38
|
+
segments.push({
|
|
39
|
+
startMs,
|
|
40
|
+
endMs: endMs ?? null,
|
|
41
|
+
text,
|
|
42
|
+
});
|
|
43
|
+
}
|
|
44
|
+
return segments.length > 0 ? segments : null;
|
|
45
|
+
}
|
|
46
|
+
export function vttToPlainText(raw) {
|
|
47
|
+
const segments = vttToSegments(raw);
|
|
48
|
+
if (segments) {
|
|
49
|
+
return segments
|
|
50
|
+
.map((segment) => segment.text)
|
|
51
|
+
.join('\n')
|
|
52
|
+
.trim();
|
|
53
|
+
}
|
|
54
|
+
const lines = raw
|
|
55
|
+
.replace(/\r\n/g, '\n')
|
|
56
|
+
.split('\n')
|
|
57
|
+
.map((line) => line.trim())
|
|
58
|
+
.filter((line) => line.length > 0)
|
|
59
|
+
.filter((line) => line.toUpperCase() !== 'WEBVTT')
|
|
60
|
+
.filter((line) => !/^\d{2}:\d{2}:\d{2}\.\d{3}\s+-->\s+\d{2}:\d{2}:\d{2}\.\d{3}/.test(line))
|
|
61
|
+
.filter((line) => !/^\d+$/.test(line))
|
|
62
|
+
.filter((line) => !/^(NOTE|STYLE|REGION)\b/i.test(line));
|
|
63
|
+
return lines.join('\n').trim();
|
|
64
|
+
}
|
|
65
|
+
function parseSegmentsFromJsonArray(items) {
|
|
66
|
+
const segments = [];
|
|
67
|
+
for (const item of items) {
|
|
68
|
+
if (!item || typeof item !== 'object')
|
|
69
|
+
continue;
|
|
70
|
+
const record = item;
|
|
71
|
+
const text = typeof record.text === 'string'
|
|
72
|
+
? record.text
|
|
73
|
+
: typeof record.utf8 === 'string'
|
|
74
|
+
? record.utf8
|
|
75
|
+
: null;
|
|
76
|
+
if (!text)
|
|
77
|
+
continue;
|
|
78
|
+
const startMs = parseTimestampToMs(record.startMs, false);
|
|
79
|
+
const endMs = parseTimestampToMs(record.endMs, false);
|
|
80
|
+
const startSeconds = parseTimestampToMs(record.start, true);
|
|
81
|
+
const endSeconds = parseTimestampToMs(record.end, true);
|
|
82
|
+
const start = startMs ?? startSeconds;
|
|
83
|
+
const end = endMs ?? endSeconds;
|
|
84
|
+
if (start == null)
|
|
85
|
+
continue;
|
|
86
|
+
segments.push({
|
|
87
|
+
startMs: start,
|
|
88
|
+
endMs: end ?? null,
|
|
89
|
+
text: text.replace(/\s+/g, ' ').trim(),
|
|
90
|
+
});
|
|
91
|
+
}
|
|
92
|
+
return segments;
|
|
93
|
+
}
|
|
94
|
+
export function jsonTranscriptToSegments(payload) {
|
|
95
|
+
if (Array.isArray(payload)) {
|
|
96
|
+
const segments = parseSegmentsFromJsonArray(payload);
|
|
97
|
+
return segments.length > 0 ? segments : null;
|
|
98
|
+
}
|
|
99
|
+
if (payload && typeof payload === 'object') {
|
|
100
|
+
const record = payload;
|
|
101
|
+
const segmentsPayload = record.segments;
|
|
102
|
+
if (Array.isArray(segmentsPayload)) {
|
|
103
|
+
const segments = parseSegmentsFromJsonArray(segmentsPayload);
|
|
104
|
+
return segments.length > 0 ? segments : null;
|
|
105
|
+
}
|
|
106
|
+
}
|
|
107
|
+
return null;
|
|
108
|
+
}
|
|
109
|
+
export function jsonTranscriptToPlainText(payload) {
|
|
110
|
+
if (Array.isArray(payload)) {
|
|
111
|
+
const segments = jsonTranscriptToSegments(payload);
|
|
112
|
+
if (segments) {
|
|
113
|
+
const text = segments
|
|
114
|
+
.map((segment) => segment.text)
|
|
115
|
+
.join('\n')
|
|
116
|
+
.trim();
|
|
117
|
+
return text.length > 0 ? text : null;
|
|
118
|
+
}
|
|
119
|
+
const parts = payload
|
|
120
|
+
.map((row) => (row && typeof row === 'object' ? row.text : null))
|
|
121
|
+
.filter((t) => typeof t === 'string')
|
|
122
|
+
.map((t) => t.trim())
|
|
123
|
+
.filter(Boolean);
|
|
124
|
+
const text = parts.join('\n').trim();
|
|
125
|
+
return text.length > 0 ? text : null;
|
|
126
|
+
}
|
|
127
|
+
if (payload && typeof payload === 'object') {
|
|
128
|
+
const record = payload;
|
|
129
|
+
if (typeof record.transcript === 'string' && record.transcript.trim())
|
|
130
|
+
return record.transcript.trim();
|
|
131
|
+
if (typeof record.text === 'string' && record.text.trim())
|
|
132
|
+
return record.text.trim();
|
|
133
|
+
const segments = record.segments;
|
|
134
|
+
if (Array.isArray(segments)) {
|
|
135
|
+
const segmentText = jsonTranscriptToSegments(record);
|
|
136
|
+
if (segmentText) {
|
|
137
|
+
const text = segmentText
|
|
138
|
+
.map((segment) => segment.text)
|
|
139
|
+
.join('\n')
|
|
140
|
+
.trim();
|
|
141
|
+
return text.length > 0 ? text : null;
|
|
142
|
+
}
|
|
143
|
+
const parts = segments
|
|
144
|
+
.map((row) => row && typeof row === 'object' ? row.text : null)
|
|
145
|
+
.filter((t) => typeof t === 'string')
|
|
146
|
+
.map((t) => t.trim())
|
|
147
|
+
.filter(Boolean);
|
|
148
|
+
const text = parts.join('\n').trim();
|
|
149
|
+
return text.length > 0 ? text : null;
|
|
150
|
+
}
|
|
151
|
+
}
|
|
152
|
+
return null;
|
|
153
|
+
}
|
|
154
|
+
//# sourceMappingURL=parse.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"parse.js","sourceRoot":"","sources":["../../../../src/content/transcript/parse.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,wBAAwB,EAAE,kBAAkB,EAAE,MAAM,iBAAiB,CAAA;AAO9E,MAAM,UAAU,aAAa,CAAC,GAAW;IACvC,MAAM,UAAU,GAAG,GAAG,CAAC,OAAO,CAAC,OAAO,EAAE,IAAI,CAAC,CAAA;IAC7C,MAAM,KAAK,GAAG,UAAU,CAAC,KAAK,CAAC,IAAI,CAAC,CAAA;IACpC,MAAM,QAAQ,GAAwB,EAAE,CAAA;IAExC,IAAI,GAAG,GAAG,CAAC,CAAA;IACX,OAAO,GAAG,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC;QAC1B,MAAM,IAAI,GAAG,KAAK,CAAC,GAAG,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,CAAA;QACrC,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,WAAW,EAAE,KAAK,QAAQ,IAAI,yBAAyB,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YACrF,GAAG,IAAI,CAAC,CAAA;YACR,SAAQ;QACV,CAAC;QACD,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;YAC1B,GAAG,IAAI,CAAC,CAAA;YACR,SAAQ;QACV,CAAC;QAED,MAAM,CAAC,QAAQ,EAAE,IAAI,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAA;QAC1C,MAAM,MAAM,GAAG,IAAI,EAAE,IAAI,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAA;QACjD,MAAM,OAAO,GAAG,wBAAwB,CAAC,QAAQ,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC,CAAA;QAChE,MAAM,KAAK,GAAG,wBAAwB,CAAC,MAAM,CAAC,CAAA;QAC9C,GAAG,IAAI,CAAC,CAAA;QAER,MAAM,SAAS,GAAa,EAAE,CAAA;QAC9B,OAAO,GAAG,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC;YAC1B,MAAM,OAAO,GAAG,KAAK,CAAC,GAAG,CAAC,CAAA;YAC1B,IAAI,CAAC,OAAO,IAAI,OAAO,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC;gBAAE,MAAK;YAClD,IAAI,CAAC,yBAAyB,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC;gBACpD,SAAS,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAA;YAChC,CAAC;YACD,GAAG,IAAI,CAAC,CAAA;QACV,CAAC;QACD,GAAG,IAAI,CAAC,CAAA;QAER,IAAI,OAAO,IAAI,IAAI;YAAE,SAAQ;QAC7B,MAAM,IAAI,GAAG,SAAS,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAA;QAC5D,IAAI,CAAC,IAAI;YAAE,SAAQ;QACnB,QAAQ,CAAC,IAAI,CAAC;YACZ,OAAO;YACP,KAAK,EAAE,KAAK,IAAI,IAAI;YACpB,IAAI;SACL,CAAC,CAAA;IACJ,CAAC;IAED,OAAO,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAA;AAC9C,CAAC;AAED,MAAM,UAAU,cAAc,CAAC,GAAW;IACxC,MAAM,QAAQ,GAAG,aAAa,CAAC,GAAG,CAAC,CAAA;IACnC,IAAI,QAAQ,EAAE,CAAC;QACb,OAAO,QAAQ;aACZ,GAAG,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC;aAC9B,IAAI,CAAC,IAAI,CAAC;aACV,IAAI,EAAE,CAAA;IACX,CAAC;IAED,MAAM,KAAK,GAAG,GAAG;SACd,OAAO,CAAC,OAAO,EAAE,IAAI,CAAC;SACtB,KAAK,CAAC,IAAI,CAAC;SACX,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;SAC1B,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC;SACjC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,WAAW,EAAE,KAAK,QAAQ,CAAC;SACjD,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,4DAA4D,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;SAC1F,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;SACrC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,yBAAyB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAA;IAC1D,OAAO,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAA;AAChC,CAAC;AAED,SAAS,0BAA0B,CAAC,KAAgB;IAClD,MAAM,QAAQ,GAAwB,EAAE,CAAA;IACxC,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,IAAI,CAAC,IAAI,IAAI,OAAO,IAAI,KAAK,QAAQ;YAAE,SAAQ;QAC/C,MAAM,MAAM,GAAG,IAA+B,CAAA;QAC9C,MAAM,IAAI,GACR,OAAO,MAAM,CAAC,IAAI,KAAK,QAAQ;YAC7B,CAAC,CAAC,MAAM,CAAC,IAAI;YACb,CAAC,CAAC,OAAO,MAAM,CAAC,IAAI,KAAK,QAAQ;gBAC/B,CAAC,CAAC,MAAM,CAAC,IAAI;gBACb,CAAC,CAAC,IAAI,CAAA;QACZ,IAAI,CAAC,IAAI;YAAE,SAAQ;QACnB,MAAM,OAAO,GAAG,kBAAkB,CAAC,MAAM,CAAC,OAAO,EAAE,KAAK,CAAC,CAAA;QACzD,MAAM,KAAK,GAAG,kBAAkB,CAAC,MAAM,CAAC,KAAK,EAAE,KAAK,CAAC,CAAA;QACrD,MAAM,YAAY,GAAG,kBAAkB,CAAC,MAAM,CAAC,KAAK,EAAE,IAAI,CAAC,CAAA;QAC3D,MAAM,UAAU,GAAG,kBAAkB,CAAC,MAAM,CAAC,GAAG,EAAE,IAAI,CAAC,CAAA;QACvD,MAAM,KAAK,GAAG,OAAO,IAAI,YAAY,CAAA;QACrC,MAAM,GAAG,GAAG,KAAK,IAAI,UAAU,CAAA;QAC/B,IAAI,KAAK,IAAI,IAAI;YAAE,SAAQ;QAC3B,QAAQ,CAAC,IAAI,CAAC;YACZ,OAAO,EAAE,KAAK;YACd,KAAK,EAAE,GAAG,IAAI,IAAI;YAClB,IAAI,EAAE,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE;SACvC,CAAC,CAAA;IACJ,CAAC;IACD,OAAO,QAAQ,CAAA;AACjB,CAAC;AAED,MAAM,UAAU,wBAAwB,CAAC,OAAgB;IACvD,IAAI,KAAK,CAAC,OAAO,CAAC,OAAO,CAAC,EAAE,CAAC;QAC3B,MAAM,QAAQ,GAAG,0BAA0B,CAAC,OAAO,CAAC,CAAA;QACpD,OAAO,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAA;IAC9C,CAAC;IAED,IAAI,OAAO,IAAI,OAAO,OAAO,KAAK,QAAQ,EAAE,CAAC;QAC3C,MAAM,MAAM,GAAG,OAAkC,CAAA;QACjD,MAAM,eAAe,GAAG,MAAM,CAAC,QAAQ,CAAA;QACvC,IAAI,KAAK,CAAC,OAAO,CAAC,eAAe,CAAC,EAAE,CAAC;YACnC,MAAM,QAAQ,GAAG,0BAA0B,CAAC,eAAe,CAAC,CAAA;YAC5D,OAAO,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAA;QAC9C,CAAC;IACH,CAAC;IAED,OAAO,IAAI,CAAA;AACb,CAAC;AAED,MAAM,UAAU,yBAAyB,CAAC,OAAgB;IACxD,IAAI,KAAK,CAAC,OAAO,CAAC,OAAO,CAAC,EAAE,CAAC;QAC3B,MAAM,QAAQ,GAAG,wBAAwB,CAAC,OAAO,CAAC,CAAA;QAClD,IAAI,QAAQ,EAAE,CAAC;YACb,MAAM,IAAI,GAAG,QAAQ;iBAClB,GAAG,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC;iBAC9B,IAAI,CAAC,IAAI,CAAC;iBACV,IAAI,EAAE,CAAA;YACT,OAAO,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAA;QACtC,CAAC;QACD,MAAM,KAAK,GAAG,OAAO;aAClB,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC,GAAG,IAAI,OAAO,GAAG,KAAK,QAAQ,CAAC,CAAC,CAAE,GAA+B,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;aAC7F,MAAM,CAAC,CAAC,CAAC,EAAe,EAAE,CAAC,OAAO,CAAC,KAAK,QAAQ,CAAC;aACjD,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;aACpB,MAAM,CAAC,OAAO,CAAC,CAAA;QAClB,MAAM,IAAI,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAA;QACpC,OAAO,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAA;IACtC,CAAC;IAED,IAAI,OAAO,IAAI,OAAO,OAAO,KAAK,QAAQ,EAAE,CAAC;QAC3C,MAAM,MAAM,GAAG,OAAkC,CAAA;QACjD,IAAI,OAAO,MAAM,CAAC,UAAU,KAAK,QAAQ,IAAI,MAAM,CAAC,UAAU,CAAC,IAAI,EAAE;YACnE,OAAO,MAAM,CAAC,UAAU,CAAC,IAAI,EAAE,CAAA;QACjC,IAAI,OAAO,MAAM,CAAC,IAAI,KAAK,QAAQ,IAAI,MAAM,CAAC,IAAI,CAAC,IAAI,EAAE;YAAE,OAAO,MAAM,CAAC,IAAI,CAAC,IAAI,EAAE,CAAA;QACpF,MAAM,QAAQ,GAAG,MAAM,CAAC,QAAQ,CAAA;QAChC,IAAI,KAAK,CAAC,OAAO,CAAC,QAAQ,CAAC,EAAE,CAAC;YAC5B,MAAM,WAAW,GAAG,wBAAwB,CAAC,MAAM,CAAC,CAAA;YACpD,IAAI,WAAW,EAAE,CAAC;gBAChB,MAAM,IAAI,GAAG,WAAW;qBACrB,GAAG,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAC,IAAI,CAAC;qBAC9B,IAAI,CAAC,IAAI,CAAC;qBACV,IAAI,EAAE,CAAA;gBACT,OAAO,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAA;YACtC,CAAC;YACD,MAAM,KAAK,GAAG,QAAQ;iBACnB,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CACX,GAAG,IAAI,OAAO,GAAG,KAAK,QAAQ,CAAC,CAAC,CAAE,GAA+B,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAC9E;iBACA,MAAM,CAAC,CAAC,CAAC,EAAe,EAAE,CAAC,OAAO,CAAC,KAAK,QAAQ,CAAC;iBACjD,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;iBACpB,MAAM,CAAC,OAAO,CAAC,CAAA;YAClB,MAAM,IAAI,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAA;YACpC,OAAO,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAA;QACtC,CAAC;IACH,CAAC;IAED,OAAO,IAAI,CAAA;AACb,CAAC"}
|
|
@@ -1,16 +1,69 @@
|
|
|
1
|
-
import {
|
|
1
|
+
import { load } from 'cheerio';
|
|
2
2
|
import { isTwitterStatusUrl } from '../../link-preview/content/twitter-utils.js';
|
|
3
|
+
import { isDirectMediaUrl } from '../../url.js';
|
|
3
4
|
import { normalizeTranscriptText } from '../normalize.js';
|
|
5
|
+
import { jsonTranscriptToPlainText, jsonTranscriptToSegments, vttToPlainText, vttToSegments, } from '../parse.js';
|
|
6
|
+
import { resolveTranscriptionAvailability } from './transcription-start.js';
|
|
4
7
|
export const canHandle = () => true;
|
|
5
8
|
export const fetchTranscript = async (context, options) => {
|
|
6
9
|
const attemptedProviders = [];
|
|
7
10
|
const notes = [];
|
|
11
|
+
const embedded = context.html ? detectEmbeddedMedia(context.html, context.url) : null;
|
|
12
|
+
const twitterStatus = isTwitterStatusUrl(context.url);
|
|
13
|
+
const hasEmbeddedMedia = Boolean(embedded?.mediaUrl || embedded?.kind);
|
|
14
|
+
const mediaKindHint = options.mediaKindHint ?? embedded?.kind ?? null;
|
|
15
|
+
if (embedded?.track) {
|
|
16
|
+
attemptedProviders.push('embedded');
|
|
17
|
+
const caption = await fetchCaptionTrack(options.fetch, embedded.track, notes, Boolean(options.transcriptTimestamps));
|
|
18
|
+
if (caption?.text) {
|
|
19
|
+
return {
|
|
20
|
+
text: normalizeTranscriptText(caption.text),
|
|
21
|
+
source: 'embedded',
|
|
22
|
+
segments: options.transcriptTimestamps ? (caption.segments ?? null) : null,
|
|
23
|
+
attemptedProviders,
|
|
24
|
+
metadata: {
|
|
25
|
+
provider: 'embedded',
|
|
26
|
+
kind: embedded.kind,
|
|
27
|
+
trackUrl: embedded.track.url,
|
|
28
|
+
trackType: embedded.track.type,
|
|
29
|
+
trackLanguage: embedded.track.language,
|
|
30
|
+
},
|
|
31
|
+
notes: notes.length > 0 ? notes.join('; ') : null,
|
|
32
|
+
};
|
|
33
|
+
}
|
|
34
|
+
}
|
|
35
|
+
const shouldAttemptMediaTranscript = options.mediaTranscriptMode === 'prefer' || (twitterStatus && hasEmbeddedMedia);
|
|
36
|
+
const mediaUrl = shouldAttemptMediaTranscript
|
|
37
|
+
? (embedded?.mediaUrl ?? (isDirectMediaUrl(context.url) ? context.url : null))
|
|
38
|
+
: null;
|
|
39
|
+
if (shouldAttemptMediaTranscript &&
|
|
40
|
+
(mediaUrl || embedded?.kind || isDirectMediaUrl(context.url))) {
|
|
41
|
+
const result = await fetchDirectMediaTranscript({
|
|
42
|
+
url: mediaUrl ?? context.url,
|
|
43
|
+
options,
|
|
44
|
+
notes,
|
|
45
|
+
attemptedProviders,
|
|
46
|
+
kind: embedded?.kind ?? null,
|
|
47
|
+
});
|
|
48
|
+
if (result)
|
|
49
|
+
return result;
|
|
50
|
+
}
|
|
51
|
+
if (twitterStatus && options.mediaTranscriptMode !== 'prefer' && !hasEmbeddedMedia) {
|
|
52
|
+
return {
|
|
53
|
+
text: null,
|
|
54
|
+
source: null,
|
|
55
|
+
attemptedProviders,
|
|
56
|
+
metadata: { provider: 'generic', kind: 'twitter', reason: 'media_mode_auto' },
|
|
57
|
+
notes: 'Twitter transcript skipped (media transcript mode is auto; enable --video-mode transcript to force audio).',
|
|
58
|
+
};
|
|
59
|
+
}
|
|
8
60
|
if (!isTwitterStatusUrl(context.url)) {
|
|
9
61
|
return {
|
|
10
62
|
text: null,
|
|
11
63
|
source: null,
|
|
12
64
|
attemptedProviders,
|
|
13
65
|
metadata: { provider: 'generic', reason: 'not_implemented' },
|
|
66
|
+
notes: notes.length > 0 ? notes.join('; ') : null,
|
|
14
67
|
};
|
|
15
68
|
}
|
|
16
69
|
if (!options.ytDlpPath) {
|
|
@@ -22,9 +75,12 @@ export const fetchTranscript = async (context, options) => {
|
|
|
22
75
|
notes: 'yt-dlp is not configured (set YT_DLP_PATH or ensure yt-dlp is on PATH)',
|
|
23
76
|
};
|
|
24
77
|
}
|
|
25
|
-
const
|
|
26
|
-
|
|
27
|
-
|
|
78
|
+
const transcriptionAvailability = await resolveTranscriptionAvailability({
|
|
79
|
+
env: options.env,
|
|
80
|
+
openaiApiKey: options.openaiApiKey,
|
|
81
|
+
falApiKey: options.falApiKey,
|
|
82
|
+
});
|
|
83
|
+
if (!transcriptionAvailability.hasAnyProvider) {
|
|
28
84
|
return {
|
|
29
85
|
text: null,
|
|
30
86
|
source: null,
|
|
@@ -48,12 +104,15 @@ export const fetchTranscript = async (context, options) => {
|
|
|
48
104
|
const mod = await import('./youtube/yt-dlp.js');
|
|
49
105
|
const ytdlpResult = await mod.fetchTranscriptWithYtDlp({
|
|
50
106
|
ytDlpPath: options.ytDlpPath,
|
|
107
|
+
env: options.env,
|
|
51
108
|
openaiApiKey: options.openaiApiKey,
|
|
52
109
|
falApiKey: options.falApiKey,
|
|
110
|
+
mediaCache: options.mediaCache ?? null,
|
|
53
111
|
url: context.url,
|
|
54
112
|
onProgress: options.onProgress ?? null,
|
|
55
113
|
service: 'generic',
|
|
56
114
|
extraArgs: extraArgs.length > 0 ? extraArgs : undefined,
|
|
115
|
+
mediaKind: mediaKindHint,
|
|
57
116
|
});
|
|
58
117
|
if (ytdlpResult.notes.length > 0) {
|
|
59
118
|
notes.push(...ytdlpResult.notes);
|
|
@@ -88,4 +147,176 @@ export const fetchTranscript = async (context, options) => {
|
|
|
88
147
|
notes: notes.length > 0 ? notes.join('; ') : null,
|
|
89
148
|
};
|
|
90
149
|
};
|
|
150
|
+
function detectEmbeddedMedia(html, baseUrl) {
|
|
151
|
+
const $ = load(html);
|
|
152
|
+
const trackCandidates = [];
|
|
153
|
+
$('track[kind="captions"], track[kind="subtitles"]').each((_idx, el) => {
|
|
154
|
+
const src = $(el).attr('src')?.trim();
|
|
155
|
+
if (!src)
|
|
156
|
+
return;
|
|
157
|
+
const url = resolveAbsoluteUrl(src, baseUrl);
|
|
158
|
+
if (!url)
|
|
159
|
+
return;
|
|
160
|
+
const type = $(el).attr('type')?.trim() ?? null;
|
|
161
|
+
const language = $(el).attr('srclang')?.trim() ?? $(el).attr('lang')?.trim() ?? null;
|
|
162
|
+
trackCandidates.push({ url, type, language });
|
|
163
|
+
});
|
|
164
|
+
const track = selectPreferredTrack(trackCandidates);
|
|
165
|
+
const videoUrl = resolveFirstMediaUrl($, baseUrl, 'video');
|
|
166
|
+
const audioUrl = resolveFirstMediaUrl($, baseUrl, 'audio');
|
|
167
|
+
const ogVideo = resolveOgMediaUrl($, baseUrl, 'video');
|
|
168
|
+
const ogAudio = resolveOgMediaUrl($, baseUrl, 'audio');
|
|
169
|
+
if (videoUrl || ogVideo) {
|
|
170
|
+
const mediaUrl = pickMediaUrl([videoUrl, ogVideo]);
|
|
171
|
+
return { kind: 'video', mediaUrl, track };
|
|
172
|
+
}
|
|
173
|
+
if (audioUrl || ogAudio) {
|
|
174
|
+
const mediaUrl = pickMediaUrl([audioUrl, ogAudio]);
|
|
175
|
+
return { kind: 'audio', mediaUrl, track };
|
|
176
|
+
}
|
|
177
|
+
const hasVideoTag = $('video').length > 0;
|
|
178
|
+
const hasAudioTag = !hasVideoTag && $('audio').length > 0;
|
|
179
|
+
if (track || hasVideoTag || hasAudioTag) {
|
|
180
|
+
return { kind: hasAudioTag ? 'audio' : 'video', mediaUrl: null, track };
|
|
181
|
+
}
|
|
182
|
+
return null;
|
|
183
|
+
}
|
|
184
|
+
function selectPreferredTrack(tracks) {
|
|
185
|
+
if (tracks.length === 0)
|
|
186
|
+
return null;
|
|
187
|
+
const normalized = tracks.map((track) => ({
|
|
188
|
+
...track,
|
|
189
|
+
language: track.language?.toLowerCase() ?? null,
|
|
190
|
+
}));
|
|
191
|
+
const english = normalized.find((track) => track.language?.startsWith('en'));
|
|
192
|
+
return english ?? normalized[0] ?? null;
|
|
193
|
+
}
|
|
194
|
+
function resolveFirstMediaUrl($, baseUrl, tag) {
|
|
195
|
+
const direct = $(`${tag}[src]`).first().attr('src') ?? $(`${tag} source[src]`).first().attr('src') ?? null;
|
|
196
|
+
if (!direct)
|
|
197
|
+
return null;
|
|
198
|
+
return resolveAbsoluteUrl(direct, baseUrl);
|
|
199
|
+
}
|
|
200
|
+
function resolveOgMediaUrl($, baseUrl, kind) {
|
|
201
|
+
const meta = $(`meta[property="og:${kind}"], meta[property="og:${kind}:url"], meta[property="og:${kind}:secure_url"], meta[name="og:${kind}"], meta[name="og:${kind}:url"], meta[name="og:${kind}:secure_url"]`)
|
|
202
|
+
.first()
|
|
203
|
+
.attr('content');
|
|
204
|
+
if (!meta)
|
|
205
|
+
return null;
|
|
206
|
+
return resolveAbsoluteUrl(meta, baseUrl);
|
|
207
|
+
}
|
|
208
|
+
function resolveAbsoluteUrl(candidate, baseUrl) {
|
|
209
|
+
const trimmed = candidate.trim();
|
|
210
|
+
if (trimmed.length === 0)
|
|
211
|
+
return null;
|
|
212
|
+
try {
|
|
213
|
+
return new URL(trimmed, baseUrl).toString();
|
|
214
|
+
}
|
|
215
|
+
catch {
|
|
216
|
+
return null;
|
|
217
|
+
}
|
|
218
|
+
}
|
|
219
|
+
function pickMediaUrl(candidates) {
|
|
220
|
+
let fallback = null;
|
|
221
|
+
for (const candidate of candidates) {
|
|
222
|
+
if (!candidate)
|
|
223
|
+
continue;
|
|
224
|
+
if (isDirectMediaUrl(candidate))
|
|
225
|
+
return candidate;
|
|
226
|
+
if (!fallback)
|
|
227
|
+
fallback = candidate;
|
|
228
|
+
}
|
|
229
|
+
return fallback;
|
|
230
|
+
}
|
|
231
|
+
async function fetchCaptionTrack(fetchImpl, track, notes, includeSegments) {
|
|
232
|
+
try {
|
|
233
|
+
const res = await fetchImpl(track.url, {
|
|
234
|
+
headers: { accept: 'text/vtt,text/plain,application/json;q=0.9,*/*;q=0.8' },
|
|
235
|
+
});
|
|
236
|
+
if (!res.ok) {
|
|
237
|
+
notes.push(`Embedded captions fetch failed (${res.status})`);
|
|
238
|
+
return null;
|
|
239
|
+
}
|
|
240
|
+
const body = await res.text();
|
|
241
|
+
const contentType = res.headers.get('content-type')?.toLowerCase() ?? '';
|
|
242
|
+
const type = track.type?.toLowerCase() ?? '';
|
|
243
|
+
if (type.includes('application/json') || contentType.includes('application/json')) {
|
|
244
|
+
try {
|
|
245
|
+
const parsed = JSON.parse(body);
|
|
246
|
+
const text = jsonTranscriptToPlainText(parsed);
|
|
247
|
+
if (!text)
|
|
248
|
+
return null;
|
|
249
|
+
const segments = includeSegments ? jsonTranscriptToSegments(parsed) : null;
|
|
250
|
+
return { text, segments };
|
|
251
|
+
}
|
|
252
|
+
catch {
|
|
253
|
+
notes.push('Embedded captions JSON parse failed');
|
|
254
|
+
return null;
|
|
255
|
+
}
|
|
256
|
+
}
|
|
257
|
+
if (type.includes('text/vtt') ||
|
|
258
|
+
contentType.includes('text/vtt') ||
|
|
259
|
+
track.url.toLowerCase().endsWith('.vtt')) {
|
|
260
|
+
const plain = vttToPlainText(body);
|
|
261
|
+
if (plain.length === 0)
|
|
262
|
+
return null;
|
|
263
|
+
const segments = includeSegments ? vttToSegments(body) : null;
|
|
264
|
+
return { text: plain, segments };
|
|
265
|
+
}
|
|
266
|
+
const trimmed = body.trim();
|
|
267
|
+
return trimmed.length > 0 ? { text: trimmed, segments: null } : null;
|
|
268
|
+
}
|
|
269
|
+
catch (error) {
|
|
270
|
+
notes.push(`Embedded captions fetch failed: ${error instanceof Error ? error.message : error}`);
|
|
271
|
+
return null;
|
|
272
|
+
}
|
|
273
|
+
}
|
|
274
|
+
async function fetchDirectMediaTranscript({ url, options, notes, attemptedProviders, kind, }) {
|
|
275
|
+
if (!options.ytDlpPath) {
|
|
276
|
+
notes.push('yt-dlp is not configured (set YT_DLP_PATH or ensure yt-dlp is on PATH)');
|
|
277
|
+
return null;
|
|
278
|
+
}
|
|
279
|
+
const transcriptionAvailability = await resolveTranscriptionAvailability({
|
|
280
|
+
env: options.env,
|
|
281
|
+
openaiApiKey: options.openaiApiKey,
|
|
282
|
+
falApiKey: options.falApiKey,
|
|
283
|
+
});
|
|
284
|
+
if (!transcriptionAvailability.hasAnyProvider) {
|
|
285
|
+
notes.push('Missing transcription provider (install whisper-cpp or set OPENAI_API_KEY/FAL_KEY)');
|
|
286
|
+
return null;
|
|
287
|
+
}
|
|
288
|
+
attemptedProviders.push('yt-dlp');
|
|
289
|
+
const mod = await import('./youtube/yt-dlp.js');
|
|
290
|
+
const ytdlpResult = await mod.fetchTranscriptWithYtDlp({
|
|
291
|
+
ytDlpPath: options.ytDlpPath,
|
|
292
|
+
env: options.env,
|
|
293
|
+
openaiApiKey: options.openaiApiKey,
|
|
294
|
+
falApiKey: options.falApiKey,
|
|
295
|
+
mediaCache: options.mediaCache ?? null,
|
|
296
|
+
url,
|
|
297
|
+
onProgress: options.onProgress ?? null,
|
|
298
|
+
service: 'generic',
|
|
299
|
+
mediaKind: kind ?? options.mediaKindHint ?? null,
|
|
300
|
+
});
|
|
301
|
+
if (ytdlpResult.notes.length > 0) {
|
|
302
|
+
notes.push(...ytdlpResult.notes);
|
|
303
|
+
}
|
|
304
|
+
if (ytdlpResult.text) {
|
|
305
|
+
return {
|
|
306
|
+
text: normalizeTranscriptText(ytdlpResult.text),
|
|
307
|
+
source: 'yt-dlp',
|
|
308
|
+
attemptedProviders,
|
|
309
|
+
metadata: {
|
|
310
|
+
provider: 'generic',
|
|
311
|
+
kind: kind ?? 'media',
|
|
312
|
+
transcriptionProvider: ytdlpResult.provider,
|
|
313
|
+
},
|
|
314
|
+
notes: notes.length > 0 ? notes.join('; ') : null,
|
|
315
|
+
};
|
|
316
|
+
}
|
|
317
|
+
if (ytdlpResult.error) {
|
|
318
|
+
notes.push(`yt-dlp transcription failed: ${ytdlpResult.error.message}`);
|
|
319
|
+
}
|
|
320
|
+
return null;
|
|
321
|
+
}
|
|
91
322
|
//# sourceMappingURL=generic.js.map
|