@steipete/summarize-core 0.10.0 → 0.11.1
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/README.md +0 -1
- package/dist/esm/content/index.js +5 -5
- package/dist/esm/content/index.js.map +1 -1
- package/dist/esm/content/link-preview/client.js +20 -9
- package/dist/esm/content/link-preview/client.js.map +1 -1
- package/dist/esm/content/link-preview/content/article.js +84 -83
- package/dist/esm/content/link-preview/content/article.js.map +1 -1
- package/dist/esm/content/link-preview/content/cleaner.js +23 -20
- package/dist/esm/content/link-preview/content/cleaner.js.map +1 -1
- package/dist/esm/content/link-preview/content/constants.js.map +1 -1
- package/dist/esm/content/link-preview/content/fetcher.js +46 -40
- package/dist/esm/content/link-preview/content/fetcher.js.map +1 -1
- package/dist/esm/content/link-preview/content/firecrawl.js +16 -16
- package/dist/esm/content/link-preview/content/firecrawl.js.map +1 -1
- package/dist/esm/content/link-preview/content/html.js +29 -27
- package/dist/esm/content/link-preview/content/html.js.map +1 -1
- package/dist/esm/content/link-preview/content/index.js +141 -88
- package/dist/esm/content/link-preview/content/index.js.map +1 -1
- package/dist/esm/content/link-preview/content/jsonld.js +12 -12
- package/dist/esm/content/link-preview/content/jsonld.js.map +1 -1
- package/dist/esm/content/link-preview/content/parsers.js +20 -20
- package/dist/esm/content/link-preview/content/parsers.js.map +1 -1
- package/dist/esm/content/link-preview/content/podcast-utils.js +34 -34
- package/dist/esm/content/link-preview/content/podcast-utils.js.map +1 -1
- package/dist/esm/content/link-preview/content/readability.js +16 -15
- package/dist/esm/content/link-preview/content/readability.js.map +1 -1
- package/dist/esm/content/link-preview/content/twitter-utils.js +24 -11
- package/dist/esm/content/link-preview/content/twitter-utils.js.map +1 -1
- package/dist/esm/content/link-preview/content/types.js +1 -1
- package/dist/esm/content/link-preview/content/types.js.map +1 -1
- package/dist/esm/content/link-preview/content/utils.js +17 -17
- package/dist/esm/content/link-preview/content/utils.js.map +1 -1
- package/dist/esm/content/link-preview/content/video.js +19 -19
- package/dist/esm/content/link-preview/content/video.js.map +1 -1
- package/dist/esm/content/link-preview/content/visibility.js +121 -0
- package/dist/esm/content/link-preview/content/visibility.js.map +1 -0
- package/dist/esm/content/link-preview/content/youtube.js +10 -10
- package/dist/esm/content/link-preview/content/youtube.js.map +1 -1
- package/dist/esm/content/link-preview/deps.js +16 -16
- package/dist/esm/content/link-preview/deps.js.map +1 -1
- package/dist/esm/content/link-preview/fetch-with-timeout.js +4 -4
- package/dist/esm/content/link-preview/fetch-with-timeout.js.map +1 -1
- package/dist/esm/content/link-preview/types.js +1 -1
- package/dist/esm/content/link-preview/types.js.map +1 -1
- package/dist/esm/content/transcript/cache.js +22 -22
- package/dist/esm/content/transcript/cache.js.map +1 -1
- package/dist/esm/content/transcript/index.js +34 -24
- package/dist/esm/content/transcript/index.js.map +1 -1
- package/dist/esm/content/transcript/normalize.js +10 -10
- package/dist/esm/content/transcript/normalize.js.map +1 -1
- package/dist/esm/content/transcript/parse.js +31 -31
- package/dist/esm/content/transcript/parse.js.map +1 -1
- package/dist/esm/content/transcript/providers/generic.js +74 -78
- package/dist/esm/content/transcript/providers/generic.js.map +1 -1
- package/dist/esm/content/transcript/providers/podcast/apple-flow.js +36 -36
- package/dist/esm/content/transcript/providers/podcast/apple-flow.js.map +1 -1
- package/dist/esm/content/transcript/providers/podcast/apple.js +5 -5
- package/dist/esm/content/transcript/providers/podcast/apple.js.map +1 -1
- package/dist/esm/content/transcript/providers/podcast/constants.js +2 -2
- package/dist/esm/content/transcript/providers/podcast/constants.js.map +1 -1
- package/dist/esm/content/transcript/providers/podcast/itunes.js +44 -42
- package/dist/esm/content/transcript/providers/podcast/itunes.js.map +1 -1
- package/dist/esm/content/transcript/providers/podcast/json.js +4 -4
- package/dist/esm/content/transcript/providers/podcast/json.js.map +1 -1
- package/dist/esm/content/transcript/providers/podcast/media.js +58 -49
- package/dist/esm/content/transcript/providers/podcast/media.js.map +1 -1
- package/dist/esm/content/transcript/providers/podcast/results.js +2 -2
- package/dist/esm/content/transcript/providers/podcast/results.js.map +1 -1
- package/dist/esm/content/transcript/providers/podcast/rss.js +29 -29
- package/dist/esm/content/transcript/providers/podcast/rss.js.map +1 -1
- package/dist/esm/content/transcript/providers/podcast/spotify-flow.js +38 -38
- package/dist/esm/content/transcript/providers/podcast/spotify-flow.js.map +1 -1
- package/dist/esm/content/transcript/providers/podcast/spotify.js +32 -32
- package/dist/esm/content/transcript/providers/podcast/spotify.js.map +1 -1
- package/dist/esm/content/transcript/providers/podcast.js +43 -47
- package/dist/esm/content/transcript/providers/podcast.js.map +1 -1
- package/dist/esm/content/transcript/providers/transcription-start.js +59 -31
- package/dist/esm/content/transcript/providers/transcription-start.js.map +1 -1
- package/dist/esm/content/transcript/providers/youtube/api.js +56 -56
- package/dist/esm/content/transcript/providers/youtube/api.js.map +1 -1
- package/dist/esm/content/transcript/providers/youtube/apify.js +7 -7
- package/dist/esm/content/transcript/providers/youtube/apify.js.map +1 -1
- package/dist/esm/content/transcript/providers/youtube/captions.js +76 -76
- package/dist/esm/content/transcript/providers/youtube/captions.js.map +1 -1
- package/dist/esm/content/transcript/providers/youtube/yt-dlp.js +82 -75
- package/dist/esm/content/transcript/providers/youtube/yt-dlp.js.map +1 -1
- package/dist/esm/content/transcript/providers/youtube.js +84 -77
- package/dist/esm/content/transcript/providers/youtube.js.map +1 -1
- package/dist/esm/content/transcript/timestamps.js +8 -8
- package/dist/esm/content/transcript/timestamps.js.map +1 -1
- package/dist/esm/content/transcript/transcription-config.js +14 -0
- package/dist/esm/content/transcript/transcription-config.js.map +1 -0
- package/dist/esm/content/transcript/utils.js +35 -35
- package/dist/esm/content/transcript/utils.js.map +1 -1
- package/dist/esm/content/url.js +59 -28
- package/dist/esm/content/url.js.map +1 -1
- package/dist/esm/index.js +4 -3
- package/dist/esm/index.js.map +1 -1
- package/dist/esm/language.js +77 -77
- package/dist/esm/language.js.map +1 -1
- package/dist/esm/openai/base-url.js +35 -0
- package/dist/esm/openai/base-url.js.map +1 -0
- package/dist/esm/processes.js +16 -16
- package/dist/esm/processes.js.map +1 -1
- package/dist/esm/prompts/cli.js +17 -17
- package/dist/esm/prompts/cli.js.map +1 -1
- package/dist/esm/prompts/file.js +54 -54
- package/dist/esm/prompts/file.js.map +1 -1
- package/dist/esm/prompts/format.js +2 -2
- package/dist/esm/prompts/format.js.map +1 -1
- package/dist/esm/prompts/index.js +5 -5
- package/dist/esm/prompts/index.js.map +1 -1
- package/dist/esm/prompts/link-summary.js +65 -65
- package/dist/esm/prompts/link-summary.js.map +1 -1
- package/dist/esm/prompts/summary-lengths.js +10 -10
- package/dist/esm/prompts/summary-lengths.js.map +1 -1
- package/dist/esm/prompts/summary-system.js +9 -9
- package/dist/esm/prompts/summary-system.js.map +1 -1
- package/dist/esm/shared/contracts.js +1 -1
- package/dist/esm/shared/contracts.js.map +1 -1
- package/dist/esm/transcription/onnx-cli.js +69 -69
- package/dist/esm/transcription/onnx-cli.js.map +1 -1
- package/dist/esm/transcription/whisper/constants.js +3 -3
- package/dist/esm/transcription/whisper/constants.js.map +1 -1
- package/dist/esm/transcription/whisper/core.js +148 -59
- package/dist/esm/transcription/whisper/core.js.map +1 -1
- package/dist/esm/transcription/whisper/fal.js +14 -14
- package/dist/esm/transcription/whisper/fal.js.map +1 -1
- package/dist/esm/transcription/whisper/ffmpeg.js +106 -106
- package/dist/esm/transcription/whisper/ffmpeg.js.map +1 -1
- package/dist/esm/transcription/whisper/groq.js +46 -0
- package/dist/esm/transcription/whisper/groq.js.map +1 -0
- package/dist/esm/transcription/whisper/openai.js +19 -13
- package/dist/esm/transcription/whisper/openai.js.map +1 -1
- package/dist/esm/transcription/whisper/utils.js +19 -19
- package/dist/esm/transcription/whisper/utils.js.map +1 -1
- package/dist/esm/transcription/whisper/whisper-cpp.js +64 -64
- package/dist/esm/transcription/whisper/whisper-cpp.js.map +1 -1
- package/dist/esm/transcription/whisper.js +4 -4
- package/dist/esm/transcription/whisper.js.map +1 -1
- package/dist/types/content/cache/types.d.ts +1 -1
- package/dist/types/content/index.d.ts +7 -7
- package/dist/types/content/link-preview/client.d.ts +7 -4
- package/dist/types/content/link-preview/content/cleaner.d.ts +1 -0
- package/dist/types/content/link-preview/content/fetcher.d.ts +2 -2
- package/dist/types/content/link-preview/content/firecrawl.d.ts +7 -7
- package/dist/types/content/link-preview/content/html.d.ts +8 -8
- package/dist/types/content/link-preview/content/index.d.ts +3 -3
- package/dist/types/content/link-preview/content/twitter-utils.d.ts +1 -0
- package/dist/types/content/link-preview/content/types.d.ts +8 -8
- package/dist/types/content/link-preview/content/utils.d.ts +3 -3
- package/dist/types/content/link-preview/content/video.d.ts +1 -1
- package/dist/types/content/link-preview/content/visibility.d.ts +1 -0
- package/dist/types/content/link-preview/deps.d.ts +36 -33
- package/dist/types/content/link-preview/types.d.ts +4 -4
- package/dist/types/content/transcript/cache.d.ts +4 -4
- package/dist/types/content/transcript/index.d.ts +7 -7
- package/dist/types/content/transcript/parse.d.ts +1 -1
- package/dist/types/content/transcript/providers/generic.d.ts +1 -1
- package/dist/types/content/transcript/providers/podcast/apple-flow.d.ts +2 -2
- package/dist/types/content/transcript/providers/podcast/flow-context.d.ts +4 -4
- package/dist/types/content/transcript/providers/podcast/media.d.ts +9 -6
- package/dist/types/content/transcript/providers/podcast/results.d.ts +3 -3
- package/dist/types/content/transcript/providers/podcast/rss.d.ts +1 -1
- package/dist/types/content/transcript/providers/podcast/spotify-flow.d.ts +2 -2
- package/dist/types/content/transcript/providers/podcast/spotify.d.ts +2 -2
- package/dist/types/content/transcript/providers/podcast.d.ts +5 -5
- package/dist/types/content/transcript/providers/transcription-start.d.ts +14 -8
- package/dist/types/content/transcript/providers/youtube/api.d.ts +1 -1
- package/dist/types/content/transcript/providers/youtube/captions.d.ts +1 -1
- package/dist/types/content/transcript/providers/youtube/yt-dlp.d.ts +11 -8
- package/dist/types/content/transcript/providers/youtube.d.ts +1 -1
- package/dist/types/content/transcript/timestamps.d.ts +1 -1
- package/dist/types/content/transcript/transcription-config.d.ts +15 -0
- package/dist/types/content/transcript/types.d.ts +12 -9
- package/dist/types/content/transcript/utils.d.ts +1 -1
- package/dist/types/content/url.d.ts +5 -3
- package/dist/types/index.d.ts +5 -4
- package/dist/types/language.d.ts +4 -4
- package/dist/types/openai/base-url.d.ts +14 -0
- package/dist/types/processes.d.ts +2 -2
- package/dist/types/prompts/cli.d.ts +3 -3
- package/dist/types/prompts/file.d.ts +2 -2
- package/dist/types/prompts/index.d.ts +6 -6
- package/dist/types/prompts/link-summary.d.ts +3 -3
- package/dist/types/prompts/summary-lengths.d.ts +1 -1
- package/dist/types/transcription/onnx-cli.d.ts +3 -3
- package/dist/types/transcription/whisper/core.d.ts +6 -3
- package/dist/types/transcription/whisper/groq.d.ts +2 -0
- package/dist/types/transcription/whisper/openai.d.ts +6 -1
- package/dist/types/transcription/whisper/types.d.ts +1 -1
- package/dist/types/transcription/whisper/whisper-cpp.d.ts +1 -1
- package/dist/types/transcription/whisper.d.ts +5 -5
- package/package.json +13 -13
|
@@ -1,14 +1,15 @@
|
|
|
1
|
-
import { mapCachedSource, readTranscriptCache, writeTranscriptCache } from
|
|
2
|
-
import { canHandle as canHandleGeneric, fetchTranscript as fetchGeneric, } from
|
|
3
|
-
import { canHandle as canHandlePodcast, fetchTranscript as fetchPodcast, } from
|
|
4
|
-
import { canHandle as canHandleYoutube, fetchTranscript as fetchYoutube, } from
|
|
5
|
-
import {
|
|
1
|
+
import { mapCachedSource, readTranscriptCache, writeTranscriptCache } from "./cache.js";
|
|
2
|
+
import { canHandle as canHandleGeneric, fetchTranscript as fetchGeneric, } from "./providers/generic.js";
|
|
3
|
+
import { canHandle as canHandlePodcast, fetchTranscript as fetchPodcast, } from "./providers/podcast.js";
|
|
4
|
+
import { canHandle as canHandleYoutube, fetchTranscript as fetchYoutube, } from "./providers/youtube.js";
|
|
5
|
+
import { resolveTranscriptionConfig } from "./transcription-config.js";
|
|
6
|
+
import { extractEmbeddedYouTubeUrlFromHtml, extractYouTubeVideoId as extractYouTubeVideoIdInternal, isYouTubeUrl as isYouTubeUrlInternal, } from "./utils.js";
|
|
6
7
|
const PROVIDERS = [
|
|
7
|
-
{ id:
|
|
8
|
-
{ id:
|
|
9
|
-
{ id:
|
|
8
|
+
{ id: "youtube", canHandle: canHandleYoutube, fetchTranscript: fetchYoutube },
|
|
9
|
+
{ id: "podcast", canHandle: canHandlePodcast, fetchTranscript: fetchPodcast },
|
|
10
|
+
{ id: "generic", canHandle: canHandleGeneric, fetchTranscript: fetchGeneric },
|
|
10
11
|
];
|
|
11
|
-
const GENERIC_PROVIDER_ID =
|
|
12
|
+
const GENERIC_PROVIDER_ID = "generic";
|
|
12
13
|
export const resolveTranscriptForLink = async (url, html, deps, { youtubeTranscriptMode, mediaTranscriptMode, mediaKindHint, transcriptTimestamps, cacheMode: providedCacheMode, fileMtime, } = {}) => {
|
|
13
14
|
const normalizedUrl = url.trim();
|
|
14
15
|
const embeddedYoutubeUrl = !isYouTubeUrlInternal(normalizedUrl) && html
|
|
@@ -18,7 +19,7 @@ export const resolveTranscriptForLink = async (url, html, deps, { youtubeTranscr
|
|
|
18
19
|
const resourceKey = extractResourceKey(effectiveUrl);
|
|
19
20
|
const baseContext = { url: effectiveUrl, html, resourceKey };
|
|
20
21
|
const provider = selectProvider(baseContext);
|
|
21
|
-
const cacheMode = providedCacheMode ??
|
|
22
|
+
const cacheMode = providedCacheMode ?? "default";
|
|
22
23
|
const cacheOutcome = await readTranscriptCache({
|
|
23
24
|
url: normalizedUrl,
|
|
24
25
|
cacheMode,
|
|
@@ -40,36 +41,45 @@ export const resolveTranscriptForLink = async (url, html, deps, { youtubeTranscr
|
|
|
40
41
|
diagnostics,
|
|
41
42
|
};
|
|
42
43
|
}
|
|
43
|
-
const shouldReportProgress = provider.id ===
|
|
44
|
+
const shouldReportProgress = provider.id === "youtube" || provider.id === "podcast";
|
|
44
45
|
if (shouldReportProgress) {
|
|
45
46
|
deps.onProgress?.({
|
|
46
|
-
kind:
|
|
47
|
+
kind: "transcript-start",
|
|
47
48
|
url: normalizedUrl,
|
|
48
49
|
service: provider.id,
|
|
49
|
-
hint: provider.id ===
|
|
50
|
-
?
|
|
51
|
-
:
|
|
50
|
+
hint: provider.id === "youtube"
|
|
51
|
+
? "YouTube: resolving transcript"
|
|
52
|
+
: "Podcast: resolving transcript",
|
|
52
53
|
});
|
|
53
54
|
}
|
|
55
|
+
const transcription = resolveTranscriptionConfig({
|
|
56
|
+
env: deps.env,
|
|
57
|
+
transcription: deps.transcription ?? null,
|
|
58
|
+
falApiKey: deps.falApiKey,
|
|
59
|
+
groqApiKey: deps.groqApiKey,
|
|
60
|
+
openaiApiKey: deps.openaiApiKey,
|
|
61
|
+
});
|
|
54
62
|
const providerResult = await executeProvider(provider, baseContext, {
|
|
55
63
|
fetch: deps.fetch,
|
|
56
64
|
env: deps.env,
|
|
57
65
|
scrapeWithFirecrawl: deps.scrapeWithFirecrawl,
|
|
58
66
|
apifyApiToken: deps.apifyApiToken,
|
|
59
67
|
ytDlpPath: deps.ytDlpPath,
|
|
60
|
-
|
|
61
|
-
|
|
68
|
+
transcription,
|
|
69
|
+
falApiKey: transcription.falApiKey,
|
|
70
|
+
groqApiKey: transcription.groqApiKey,
|
|
71
|
+
openaiApiKey: transcription.openaiApiKey,
|
|
62
72
|
mediaCache: deps.mediaCache ?? null,
|
|
63
73
|
resolveTwitterCookies: deps.resolveTwitterCookies ?? null,
|
|
64
74
|
onProgress: deps.onProgress ?? null,
|
|
65
|
-
youtubeTranscriptMode: youtubeTranscriptMode ??
|
|
66
|
-
mediaTranscriptMode: mediaTranscriptMode ??
|
|
75
|
+
youtubeTranscriptMode: youtubeTranscriptMode ?? "auto",
|
|
76
|
+
mediaTranscriptMode: mediaTranscriptMode ?? "auto",
|
|
67
77
|
mediaKindHint: mediaKindHint ?? null,
|
|
68
78
|
transcriptTimestamps: transcriptTimestamps ?? false,
|
|
69
79
|
});
|
|
70
80
|
if (shouldReportProgress) {
|
|
71
81
|
deps.onProgress?.({
|
|
72
|
-
kind:
|
|
82
|
+
kind: "transcript-done",
|
|
73
83
|
url: normalizedUrl,
|
|
74
84
|
ok: Boolean(providerResult.text && providerResult.text.length > 0),
|
|
75
85
|
service: provider.id,
|
|
@@ -110,11 +120,11 @@ export const resolveTranscriptForLink = async (url, html, deps, { youtubeTranscr
|
|
|
110
120
|
fileMtime,
|
|
111
121
|
});
|
|
112
122
|
}
|
|
113
|
-
if (!providerResult.text && cacheOutcome.cached?.content && cacheMode !==
|
|
114
|
-
diagnostics.cacheStatus =
|
|
123
|
+
if (!providerResult.text && cacheOutcome.cached?.content && cacheMode !== "bypass") {
|
|
124
|
+
diagnostics.cacheStatus = "fallback";
|
|
115
125
|
diagnostics.provider = mapCachedSource(cacheOutcome.cached.source);
|
|
116
126
|
diagnostics.textProvided = Boolean(cacheOutcome.cached.content && cacheOutcome.cached.content.length > 0);
|
|
117
|
-
diagnostics.notes = appendNote(diagnostics.notes,
|
|
127
|
+
diagnostics.notes = appendNote(diagnostics.notes, "Falling back to cached transcript content after provider miss");
|
|
118
128
|
return {
|
|
119
129
|
text: cacheOutcome.cached.content,
|
|
120
130
|
source: diagnostics.provider,
|
|
@@ -148,7 +158,7 @@ const selectProvider = (context) => {
|
|
|
148
158
|
if (genericProviderModule) {
|
|
149
159
|
return genericProviderModule;
|
|
150
160
|
}
|
|
151
|
-
throw new Error(
|
|
161
|
+
throw new Error("Generic transcript provider is not registered");
|
|
152
162
|
};
|
|
153
163
|
const executeProvider = async (provider, context, options) => provider.fetchTranscript(context, options);
|
|
154
164
|
const appendNote = (existing, next) => {
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"index.js","sourceRoot":"","sources":["../../../../src/content/transcript/index.ts"],"names":[],"mappings":"
|
|
1
|
+
{"version":3,"file":"index.js","sourceRoot":"","sources":["../../../../src/content/transcript/index.ts"],"names":[],"mappings":"AAYA,OAAO,EAAE,eAAe,EAAE,mBAAmB,EAAE,oBAAoB,EAAE,MAAM,YAAY,CAAC;AACxF,OAAO,EACL,SAAS,IAAI,gBAAgB,EAC7B,eAAe,IAAI,YAAY,GAChC,MAAM,wBAAwB,CAAC;AAChC,OAAO,EACL,SAAS,IAAI,gBAAgB,EAC7B,eAAe,IAAI,YAAY,GAChC,MAAM,wBAAwB,CAAC;AAChC,OAAO,EACL,SAAS,IAAI,gBAAgB,EAC7B,eAAe,IAAI,YAAY,GAChC,MAAM,wBAAwB,CAAC;AAChC,OAAO,EAAE,0BAA0B,EAAE,MAAM,2BAA2B,CAAC;AACvE,OAAO,EACL,iCAAiC,EACjC,qBAAqB,IAAI,6BAA6B,EACtD,YAAY,IAAI,oBAAoB,GACrC,MAAM,YAAY,CAAC;AAWpB,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,CAAC;AACF,MAAM,mBAAmB,GAAG,SAAS,CAAC;AAEtC,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,CAAC;IACjC,MAAM,kBAAkB,GACtB,CAAC,oBAAoB,CAAC,aAAa,CAAC,IAAI,IAAI;QAC1C,CAAC,CAAC,MAAM,iCAAiC,CAAC,IAAI,CAAC;QAC/C,CAAC,CAAC,IAAI,CAAC;IACX,MAAM,YAAY,GAAG,kBAAkB,IAAI,aAAa,CAAC;IACzD,MAAM,WAAW,GAAG,kBAAkB,CAAC,YAAY,CAAC,CAAC;IACrD,MAAM,WAAW,GAAoB,EAAE,GAAG,EAAE,YAAY,EAAE,IAAI,EAAE,WAAW,EAAE,CAAC;IAC9E,MAAM,QAAQ,GAAmB,cAAc,CAAC,WAAW,CAAC,CAAC;IAC7D,MAAM,SAAS,GAAc,iBAAiB,IAAI,SAAS,CAAC;IAE5D,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,CAAC;IAEH,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,CAAC;IAEF,IAAI,YAAY,CAAC,UAAU,EAAE,CAAC;QAC5B,OAAO;YACL,GAAG,YAAY,CAAC,UAAU;YAC1B,WAAW;SACZ,CAAC;IACJ,CAAC;IAED,MAAM,oBAAoB,GAAG,QAAQ,CAAC,EAAE,KAAK,SAAS,IAAI,QAAQ,CAAC,EAAE,KAAK,SAAS,CAAC;IACpF,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,CAAC;IACL,CAAC;IAED,MAAM,aAAa,GAAG,0BAA0B,CAAC;QAC/C,GAAG,EAAE,IAAI,CAAC,GAAG;QACb,aAAa,EAAE,IAAI,CAAC,aAAa,IAAI,IAAI;QACzC,SAAS,EAAE,IAAI,CAAC,SAAS;QACzB,UAAU,EAAE,IAAI,CAAC,UAAU;QAC3B,YAAY,EAAE,IAAI,CAAC,YAAY;KAChC,CAAC,CAAC;IAEH,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,aAAa;QACb,SAAS,EAAE,aAAa,CAAC,SAAS;QAClC,UAAU,EAAE,aAAa,CAAC,UAAU;QACpC,YAAY,EAAE,aAAa,CAAC,YAAY;QACxC,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,CAAC;IAEH,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,CAAC;IACL,CAAC;IAED,WAAW,CAAC,QAAQ,GAAG,cAAc,CAAC,MAAM,CAAC;IAC7C,WAAW,CAAC,kBAAkB,GAAG,cAAc,CAAC,kBAAkB,CAAC;IACnE,WAAW,CAAC,YAAY,GAAG,OAAO,CAAC,cAAc,CAAC,IAAI,IAAI,cAAc,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;IAC1F,IAAI,cAAc,CAAC,KAAK,EAAE,CAAC;QACzB,WAAW,CAAC,KAAK,GAAG,UAAU,CAAC,WAAW,CAAC,KAAK,EAAE,cAAc,CAAC,KAAK,CAAC,CAAC;IAC1E,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,CAAC;YACxD,IAAI,cAAc,CAAC,QAAQ,IAAI,cAAc,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAClE,QAAQ,CAAC,UAAU,GAAG,IAAI,CAAC;gBAC3B,QAAQ,CAAC,QAAQ,GAAG,cAAc,CAAC,QAAQ,CAAC;YAC9C,CAAC;iBAAM,IAAI,QAAQ,CAAC,UAAU,IAAI,IAAI,EAAE,CAAC;gBACvC,QAAQ,CAAC,UAAU,GAAG,KAAK,CAAC;YAC9B,CAAC;YACD,cAAc,CAAC,QAAQ,GAAG,QAAQ,CAAC;QACrC,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,CAAC;QACJ,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,CAAC;IACL,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,CAAC;QACrC,WAAW,CAAC,QAAQ,GAAG,eAAe,CAAC,YAAY,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;QACnE,WAAW,CAAC,YAAY,GAAG,OAAO,CAChC,YAAY,CAAC,MAAM,CAAC,OAAO,IAAI,YAAY,CAAC,MAAM,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CACtE,CAAC;QACF,WAAW,CAAC,KAAK,GAAG,UAAU,CAC5B,WAAW,CAAC,KAAK,EACjB,+DAA+D,CAChE,CAAC;QAEF,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,CAAC;IACJ,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,CAAC;AACJ,CAAC,CAAC;AAEF,MAAM,kBAAkB,GAAG,CAAC,GAAW,EAAiB,EAAE;IACxD,IAAI,oBAAoB,CAAC,GAAG,CAAC,EAAE,CAAC;QAC9B,OAAO,6BAA6B,CAAC,GAAG,CAAC,CAAC;IAC5C,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC,CAAC;AAEF,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,CAAC;IAEhG,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,CAAC;IACF,IAAI,mBAAmB,EAAE,CAAC;QACxB,OAAO,mBAAmB,CAAC;IAC7B,CAAC;IAED,IAAI,qBAAqB,EAAE,CAAC;QAC1B,OAAO,qBAAqB,CAAC;IAC/B,CAAC;IAED,MAAM,IAAI,KAAK,CAAC,+CAA+C,CAAC,CAAC;AACnE,CAAC,CAAC;AAEF,MAAM,eAAe,GAAG,KAAK,EAC3B,QAAwB,EACxB,OAAwB,EACxB,OAA6B,EACJ,EAAE,CAAC,QAAQ,CAAC,eAAe,CAAC,OAAO,EAAE,OAAO,CAAC,CAAC;AAEzE,MAAM,UAAU,GAAG,CAAC,QAAmC,EAAE,IAAY,EAAU,EAAE;IAC/E,IAAI,CAAC,QAAQ,EAAE,CAAC;QACd,OAAO,IAAI,CAAC;IACd,CAAC;IACD,OAAO,GAAG,QAAQ,KAAK,IAAI,EAAE,CAAC;AAChC,CAAC,CAAC;AAEF,MAAM,2BAA2B,GAAG,CAAC,QAAyC,EAAE,EAAE;IAChF,IAAI,CAAC,QAAQ;QAAE,OAAO,IAAI,CAAC;IAC3B,MAAM,QAAQ,GAAI,QAAmC,CAAC,QAAQ,CAAC;IAC/D,OAAO,KAAK,CAAC,OAAO,CAAC,QAAQ,CAAC,IAAI,QAAQ,CAAC,MAAM,GAAG,CAAC;QACnD,CAAC,CAAE,QAA6C;QAChD,CAAC,CAAC,IAAI,CAAC;AACX,CAAC,CAAC"}
|
|
@@ -1,19 +1,19 @@
|
|
|
1
|
-
import { isRecord } from
|
|
1
|
+
import { isRecord } from "./utils.js";
|
|
2
2
|
export const normalizeTranscriptText = (input) => input
|
|
3
|
-
.replaceAll(
|
|
4
|
-
.replaceAll(/[\t ]+/g,
|
|
5
|
-
.replaceAll(/\s*\n\s*/g,
|
|
6
|
-
.replaceAll(/\n{3,}/g,
|
|
3
|
+
.replaceAll("\u00A0", " ")
|
|
4
|
+
.replaceAll(/[\t ]+/g, " ")
|
|
5
|
+
.replaceAll(/\s*\n\s*/g, "\n")
|
|
6
|
+
.replaceAll(/\n{3,}/g, "\n\n")
|
|
7
7
|
.trim();
|
|
8
8
|
export const normalizeTranscriptLines = (lines) => {
|
|
9
9
|
if (lines.length === 0) {
|
|
10
10
|
return null;
|
|
11
11
|
}
|
|
12
|
-
const normalized = normalizeTranscriptText(lines.join(
|
|
12
|
+
const normalized = normalizeTranscriptText(lines.join("\n"));
|
|
13
13
|
return normalized.length > 0 ? normalized : null;
|
|
14
14
|
};
|
|
15
15
|
export const normalizeApifyTranscript = (raw) => {
|
|
16
|
-
if (typeof raw ===
|
|
16
|
+
if (typeof raw === "string") {
|
|
17
17
|
const trimmed = normalizeTranscriptText(raw);
|
|
18
18
|
return trimmed.length > 0 ? trimmed : null;
|
|
19
19
|
}
|
|
@@ -21,10 +21,10 @@ export const normalizeApifyTranscript = (raw) => {
|
|
|
21
21
|
const lines = raw
|
|
22
22
|
.map((entry) => {
|
|
23
23
|
if (!isRecord(entry)) {
|
|
24
|
-
return
|
|
24
|
+
return "";
|
|
25
25
|
}
|
|
26
26
|
const textValue = entry.text;
|
|
27
|
-
return typeof textValue ===
|
|
27
|
+
return typeof textValue === "string" ? textValue.trim() : "";
|
|
28
28
|
})
|
|
29
29
|
.filter((line) => line.length > 0);
|
|
30
30
|
if (lines.length > 0) {
|
|
@@ -33,7 +33,7 @@ export const normalizeApifyTranscript = (raw) => {
|
|
|
33
33
|
}
|
|
34
34
|
if (isRecord(raw)) {
|
|
35
35
|
const singleText = raw.text;
|
|
36
|
-
if (typeof singleText ===
|
|
36
|
+
if (typeof singleText === "string") {
|
|
37
37
|
const trimmed = normalizeTranscriptText(singleText);
|
|
38
38
|
return trimmed.length > 0 ? trimmed : null;
|
|
39
39
|
}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"normalize.js","sourceRoot":"","sources":["../../../../src/content/transcript/normalize.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,MAAM,YAAY,
|
|
1
|
+
{"version":3,"file":"normalize.js","sourceRoot":"","sources":["../../../../src/content/transcript/normalize.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,MAAM,YAAY,CAAC;AAEtC,MAAM,CAAC,MAAM,uBAAuB,GAAG,CAAC,KAAa,EAAU,EAAE,CAC/D,KAAK;KACF,UAAU,CAAC,QAAQ,EAAE,GAAG,CAAC;KACzB,UAAU,CAAC,SAAS,EAAE,GAAG,CAAC;KAC1B,UAAU,CAAC,WAAW,EAAE,IAAI,CAAC;KAC7B,UAAU,CAAC,SAAS,EAAE,MAAM,CAAC;KAC7B,IAAI,EAAE,CAAC;AAEZ,MAAM,CAAC,MAAM,wBAAwB,GAAG,CAAC,KAAwB,EAAiB,EAAE;IAClF,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACvB,OAAO,IAAI,CAAC;IACd,CAAC;IACD,MAAM,UAAU,GAAG,uBAAuB,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;IAC7D,OAAO,UAAU,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,IAAI,CAAC;AACnD,CAAC,CAAC;AAEF,MAAM,CAAC,MAAM,wBAAwB,GAAG,CAAC,GAAY,EAAiB,EAAE;IACtE,IAAI,OAAO,GAAG,KAAK,QAAQ,EAAE,CAAC;QAC5B,MAAM,OAAO,GAAG,uBAAuB,CAAC,GAAG,CAAC,CAAC;QAC7C,OAAO,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC;IAC7C,CAAC;IAED,IAAI,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC;QAEvB,MAAM,KAAK,GAAG,GAAG;aACd,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE;YACb,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;gBACrB,OAAO,EAAE,CAAC;YACZ,CAAC;YACD,MAAM,SAAS,GAAI,KAAuB,CAAC,IAAI,CAAC;YAChD,OAAO,OAAO,SAAS,KAAK,QAAQ,CAAC,CAAC,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;QAC/D,CAAC,CAAC;aACD,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QACrC,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACrB,OAAO,wBAAwB,CAAC,KAAK,CAAC,IAAI,IAAI,CAAC;QACjD,CAAC;IACH,CAAC;IAED,IAAI,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC;QAClB,MAAM,UAAU,GAAI,GAA0B,CAAC,IAAI,CAAC;QACpD,IAAI,OAAO,UAAU,KAAK,QAAQ,EAAE,CAAC;YACnC,MAAM,OAAO,GAAG,uBAAuB,CAAC,UAAU,CAAC,CAAC;YACpD,OAAO,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC;QAC7C,CAAC;IACH,CAAC;IAED,OAAO,IAAI,CAAC;AACd,CAAC,CAAC"}
|
|
@@ -1,22 +1,22 @@
|
|
|
1
|
-
import { parseTimestampStringToMs, parseTimestampToMs } from
|
|
1
|
+
import { parseTimestampStringToMs, parseTimestampToMs } from "./timestamps.js";
|
|
2
2
|
export function vttToSegments(raw) {
|
|
3
|
-
const normalized = raw.replace(/\r\n/g,
|
|
4
|
-
const lines = normalized.split(
|
|
3
|
+
const normalized = raw.replace(/\r\n/g, "\n");
|
|
4
|
+
const lines = normalized.split("\n");
|
|
5
5
|
const segments = [];
|
|
6
6
|
let idx = 0;
|
|
7
7
|
while (idx < lines.length) {
|
|
8
|
-
const line = lines[idx]?.trim() ??
|
|
9
|
-
if (!line || line.toUpperCase() ===
|
|
8
|
+
const line = lines[idx]?.trim() ?? "";
|
|
9
|
+
if (!line || line.toUpperCase() === "WEBVTT" || /^(NOTE|STYLE|REGION)\b/i.test(line)) {
|
|
10
10
|
idx += 1;
|
|
11
11
|
continue;
|
|
12
12
|
}
|
|
13
|
-
if (!line.includes(
|
|
13
|
+
if (!line.includes("-->")) {
|
|
14
14
|
idx += 1;
|
|
15
15
|
continue;
|
|
16
16
|
}
|
|
17
|
-
const [startRaw, rest] = line.split(
|
|
18
|
-
const endRaw = rest?.trim().split(/\s+/)[0] ??
|
|
19
|
-
const startMs = parseTimestampStringToMs(startRaw?.trim() ??
|
|
17
|
+
const [startRaw, rest] = line.split("-->");
|
|
18
|
+
const endRaw = rest?.trim().split(/\s+/)[0] ?? "";
|
|
19
|
+
const startMs = parseTimestampStringToMs(startRaw?.trim() ?? "");
|
|
20
20
|
const endMs = parseTimestampStringToMs(endRaw);
|
|
21
21
|
idx += 1;
|
|
22
22
|
const textLines = [];
|
|
@@ -32,7 +32,7 @@ export function vttToSegments(raw) {
|
|
|
32
32
|
idx += 1;
|
|
33
33
|
if (startMs == null)
|
|
34
34
|
continue;
|
|
35
|
-
const text = textLines.join(
|
|
35
|
+
const text = textLines.join(" ").replace(/\s+/g, " ").trim();
|
|
36
36
|
if (!text)
|
|
37
37
|
continue;
|
|
38
38
|
segments.push({
|
|
@@ -48,29 +48,29 @@ export function vttToPlainText(raw) {
|
|
|
48
48
|
if (segments) {
|
|
49
49
|
return segments
|
|
50
50
|
.map((segment) => segment.text)
|
|
51
|
-
.join(
|
|
51
|
+
.join("\n")
|
|
52
52
|
.trim();
|
|
53
53
|
}
|
|
54
54
|
const lines = raw
|
|
55
|
-
.replace(/\r\n/g,
|
|
56
|
-
.split(
|
|
55
|
+
.replace(/\r\n/g, "\n")
|
|
56
|
+
.split("\n")
|
|
57
57
|
.map((line) => line.trim())
|
|
58
58
|
.filter((line) => line.length > 0)
|
|
59
|
-
.filter((line) => line.toUpperCase() !==
|
|
59
|
+
.filter((line) => line.toUpperCase() !== "WEBVTT")
|
|
60
60
|
.filter((line) => !/^\d{2}:\d{2}:\d{2}\.\d{3}\s+-->\s+\d{2}:\d{2}:\d{2}\.\d{3}/.test(line))
|
|
61
61
|
.filter((line) => !/^\d+$/.test(line))
|
|
62
62
|
.filter((line) => !/^(NOTE|STYLE|REGION)\b/i.test(line));
|
|
63
|
-
return lines.join(
|
|
63
|
+
return lines.join("\n").trim();
|
|
64
64
|
}
|
|
65
65
|
function parseSegmentsFromJsonArray(items) {
|
|
66
66
|
const segments = [];
|
|
67
67
|
for (const item of items) {
|
|
68
|
-
if (!item || typeof item !==
|
|
68
|
+
if (!item || typeof item !== "object")
|
|
69
69
|
continue;
|
|
70
70
|
const record = item;
|
|
71
|
-
const text = typeof record.text ===
|
|
71
|
+
const text = typeof record.text === "string"
|
|
72
72
|
? record.text
|
|
73
|
-
: typeof record.utf8 ===
|
|
73
|
+
: typeof record.utf8 === "string"
|
|
74
74
|
? record.utf8
|
|
75
75
|
: null;
|
|
76
76
|
if (!text)
|
|
@@ -86,7 +86,7 @@ function parseSegmentsFromJsonArray(items) {
|
|
|
86
86
|
segments.push({
|
|
87
87
|
startMs: start,
|
|
88
88
|
endMs: end ?? null,
|
|
89
|
-
text: text.replace(/\s+/g,
|
|
89
|
+
text: text.replace(/\s+/g, " ").trim(),
|
|
90
90
|
});
|
|
91
91
|
}
|
|
92
92
|
return segments;
|
|
@@ -96,7 +96,7 @@ export function jsonTranscriptToSegments(payload) {
|
|
|
96
96
|
const segments = parseSegmentsFromJsonArray(payload);
|
|
97
97
|
return segments.length > 0 ? segments : null;
|
|
98
98
|
}
|
|
99
|
-
if (payload && typeof payload ===
|
|
99
|
+
if (payload && typeof payload === "object") {
|
|
100
100
|
const record = payload;
|
|
101
101
|
const segmentsPayload = record.segments;
|
|
102
102
|
if (Array.isArray(segmentsPayload)) {
|
|
@@ -112,23 +112,23 @@ export function jsonTranscriptToPlainText(payload) {
|
|
|
112
112
|
if (segments) {
|
|
113
113
|
const text = segments
|
|
114
114
|
.map((segment) => segment.text)
|
|
115
|
-
.join(
|
|
115
|
+
.join("\n")
|
|
116
116
|
.trim();
|
|
117
117
|
return text.length > 0 ? text : null;
|
|
118
118
|
}
|
|
119
119
|
const parts = payload
|
|
120
|
-
.map((row) => (row && typeof row ===
|
|
121
|
-
.filter((t) => typeof t ===
|
|
120
|
+
.map((row) => (row && typeof row === "object" ? row.text : null))
|
|
121
|
+
.filter((t) => typeof t === "string")
|
|
122
122
|
.map((t) => t.trim())
|
|
123
123
|
.filter(Boolean);
|
|
124
|
-
const text = parts.join(
|
|
124
|
+
const text = parts.join("\n").trim();
|
|
125
125
|
return text.length > 0 ? text : null;
|
|
126
126
|
}
|
|
127
|
-
if (payload && typeof payload ===
|
|
127
|
+
if (payload && typeof payload === "object") {
|
|
128
128
|
const record = payload;
|
|
129
|
-
if (typeof record.transcript ===
|
|
129
|
+
if (typeof record.transcript === "string" && record.transcript.trim())
|
|
130
130
|
return record.transcript.trim();
|
|
131
|
-
if (typeof record.text ===
|
|
131
|
+
if (typeof record.text === "string" && record.text.trim())
|
|
132
132
|
return record.text.trim();
|
|
133
133
|
const segments = record.segments;
|
|
134
134
|
if (Array.isArray(segments)) {
|
|
@@ -136,16 +136,16 @@ export function jsonTranscriptToPlainText(payload) {
|
|
|
136
136
|
if (segmentText) {
|
|
137
137
|
const text = segmentText
|
|
138
138
|
.map((segment) => segment.text)
|
|
139
|
-
.join(
|
|
139
|
+
.join("\n")
|
|
140
140
|
.trim();
|
|
141
141
|
return text.length > 0 ? text : null;
|
|
142
142
|
}
|
|
143
143
|
const parts = segments
|
|
144
|
-
.map((row) => row && typeof row ===
|
|
145
|
-
.filter((t) => typeof t ===
|
|
144
|
+
.map((row) => row && typeof row === "object" ? row.text : null)
|
|
145
|
+
.filter((t) => typeof t === "string")
|
|
146
146
|
.map((t) => t.trim())
|
|
147
147
|
.filter(Boolean);
|
|
148
|
-
const text = parts.join(
|
|
148
|
+
const text = parts.join("\n").trim();
|
|
149
149
|
return text.length > 0 ? text : null;
|
|
150
150
|
}
|
|
151
151
|
}
|
|
@@ -1 +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,
|
|
1
|
+
{"version":3,"file":"parse.js","sourceRoot":"","sources":["../../../../src/content/transcript/parse.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,wBAAwB,EAAE,kBAAkB,EAAE,MAAM,iBAAiB,CAAC;AAO/E,MAAM,UAAU,aAAa,CAAC,GAAW;IACvC,MAAM,UAAU,GAAG,GAAG,CAAC,OAAO,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC;IAC9C,MAAM,KAAK,GAAG,UAAU,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IACrC,MAAM,QAAQ,GAAwB,EAAE,CAAC;IAEzC,IAAI,GAAG,GAAG,CAAC,CAAC;IACZ,OAAO,GAAG,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC;QAC1B,MAAM,IAAI,GAAG,KAAK,CAAC,GAAG,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC;QACtC,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,WAAW,EAAE,KAAK,QAAQ,IAAI,yBAAyB,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YACrF,GAAG,IAAI,CAAC,CAAC;YACT,SAAS;QACX,CAAC;QACD,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;YAC1B,GAAG,IAAI,CAAC,CAAC;YACT,SAAS;QACX,CAAC;QAED,MAAM,CAAC,QAAQ,EAAE,IAAI,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC;QAC3C,MAAM,MAAM,GAAG,IAAI,EAAE,IAAI,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;QAClD,MAAM,OAAO,GAAG,wBAAwB,CAAC,QAAQ,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC,CAAC;QACjE,MAAM,KAAK,GAAG,wBAAwB,CAAC,MAAM,CAAC,CAAC;QAC/C,GAAG,IAAI,CAAC,CAAC;QAET,MAAM,SAAS,GAAa,EAAE,CAAC;QAC/B,OAAO,GAAG,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC;YAC1B,MAAM,OAAO,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC;YAC3B,IAAI,CAAC,OAAO,IAAI,OAAO,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC;gBAAE,MAAM;YACnD,IAAI,CAAC,yBAAyB,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC;gBACpD,SAAS,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC;YACjC,CAAC;YACD,GAAG,IAAI,CAAC,CAAC;QACX,CAAC;QACD,GAAG,IAAI,CAAC,CAAC;QAET,IAAI,OAAO,IAAI,IAAI;YAAE,SAAS;QAC9B,MAAM,IAAI,GAAG,SAAS,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;QAC7D,IAAI,CAAC,IAAI;YAAE,SAAS;QACpB,QAAQ,CAAC,IAAI,CAAC;YACZ,OAAO;YACP,KAAK,EAAE,KAAK,IAAI,IAAI;YACpB,IAAI;SACL,CAAC,CAAC;IACL,CAAC;IAED,OAAO,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC;AAC/C,CAAC;AAED,MAAM,UAAU,cAAc,CAAC,GAAW;IACxC,MAAM,QAAQ,GAAG,aAAa,CAAC,GAAG,CAAC,CAAC;IACpC,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,CAAC;IACZ,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,CAAC;IAC3D,OAAO,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC;AACjC,CAAC;AAED,SAAS,0BAA0B,CAAC,KAAgB;IAClD,MAAM,QAAQ,GAAwB,EAAE,CAAC;IACzC,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,IAAI,CAAC,IAAI,IAAI,OAAO,IAAI,KAAK,QAAQ;YAAE,SAAS;QAChD,MAAM,MAAM,GAAG,IAA+B,CAAC;QAC/C,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,CAAC;QACb,IAAI,CAAC,IAAI;YAAE,SAAS;QACpB,MAAM,OAAO,GAAG,kBAAkB,CAAC,MAAM,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC;QAC1D,MAAM,KAAK,GAAG,kBAAkB,CAAC,MAAM,CAAC,KAAK,EAAE,KAAK,CAAC,CAAC;QACtD,MAAM,YAAY,GAAG,kBAAkB,CAAC,MAAM,CAAC,KAAK,EAAE,IAAI,CAAC,CAAC;QAC5D,MAAM,UAAU,GAAG,kBAAkB,CAAC,MAAM,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC;QACxD,MAAM,KAAK,GAAG,OAAO,IAAI,YAAY,CAAC;QACtC,MAAM,GAAG,GAAG,KAAK,IAAI,UAAU,CAAC;QAChC,IAAI,KAAK,IAAI,IAAI;YAAE,SAAS;QAC5B,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,CAAC;IACL,CAAC;IACD,OAAO,QAAQ,CAAC;AAClB,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,CAAC;QACrD,OAAO,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC;IAC/C,CAAC;IAED,IAAI,OAAO,IAAI,OAAO,OAAO,KAAK,QAAQ,EAAE,CAAC;QAC3C,MAAM,MAAM,GAAG,OAAkC,CAAC;QAClD,MAAM,eAAe,GAAG,MAAM,CAAC,QAAQ,CAAC;QACxC,IAAI,KAAK,CAAC,OAAO,CAAC,eAAe,CAAC,EAAE,CAAC;YACnC,MAAM,QAAQ,GAAG,0BAA0B,CAAC,eAAe,CAAC,CAAC;YAC7D,OAAO,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC;QAC/C,CAAC;IACH,CAAC;IAED,OAAO,IAAI,CAAC;AACd,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,CAAC;QACnD,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,CAAC;YACV,OAAO,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC;QACvC,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,CAAC;QACnB,MAAM,IAAI,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC;QACrC,OAAO,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC;IACvC,CAAC;IAED,IAAI,OAAO,IAAI,OAAO,OAAO,KAAK,QAAQ,EAAE,CAAC;QAC3C,MAAM,MAAM,GAAG,OAAkC,CAAC;QAClD,IAAI,OAAO,MAAM,CAAC,UAAU,KAAK,QAAQ,IAAI,MAAM,CAAC,UAAU,CAAC,IAAI,EAAE;YACnE,OAAO,MAAM,CAAC,UAAU,CAAC,IAAI,EAAE,CAAC;QAClC,IAAI,OAAO,MAAM,CAAC,IAAI,KAAK,QAAQ,IAAI,MAAM,CAAC,IAAI,CAAC,IAAI,EAAE;YAAE,OAAO,MAAM,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;QACrF,MAAM,QAAQ,GAAG,MAAM,CAAC,QAAQ,CAAC;QACjC,IAAI,KAAK,CAAC,OAAO,CAAC,QAAQ,CAAC,EAAE,CAAC;YAC5B,MAAM,WAAW,GAAG,wBAAwB,CAAC,MAAM,CAAC,CAAC;YACrD,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,CAAC;gBACV,OAAO,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC;YACvC,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,CAAC;YACnB,MAAM,IAAI,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC;YACrC,OAAO,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC;QACvC,CAAC;IACH,CAAC;IAED,OAAO,IAAI,CAAC;AACd,CAAC"}
|