@steipete/summarize-core 0.7.0 → 0.7.1

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.
Files changed (123) hide show
  1. package/LICENSE +21 -0
  2. package/dist/cjs/content/index.js +14 -0
  3. package/dist/cjs/content/index.js.map +1 -0
  4. package/dist/cjs/content/link-preview/client.js +31 -0
  5. package/dist/cjs/content/link-preview/client.js.map +1 -0
  6. package/dist/cjs/content/link-preview/content/article.js +164 -0
  7. package/dist/cjs/content/link-preview/content/article.js.map +1 -0
  8. package/dist/cjs/content/link-preview/content/cleaner.js +63 -0
  9. package/dist/cjs/content/link-preview/content/cleaner.js.map +1 -0
  10. package/dist/cjs/content/link-preview/content/constants.js +10 -0
  11. package/dist/cjs/content/link-preview/content/constants.js.map +1 -0
  12. package/dist/cjs/content/link-preview/content/fetcher.js +128 -0
  13. package/dist/cjs/content/link-preview/content/fetcher.js.map +1 -0
  14. package/dist/cjs/content/link-preview/content/firecrawl.js +90 -0
  15. package/dist/cjs/content/link-preview/content/firecrawl.js.map +1 -0
  16. package/dist/cjs/content/link-preview/content/html.js +165 -0
  17. package/dist/cjs/content/link-preview/content/html.js.map +1 -0
  18. package/dist/cjs/content/link-preview/content/index.js +348 -0
  19. package/dist/cjs/content/link-preview/content/index.js.map +1 -0
  20. package/dist/cjs/content/link-preview/content/jsonld.js +80 -0
  21. package/dist/cjs/content/link-preview/content/jsonld.js.map +1 -0
  22. package/dist/cjs/content/link-preview/content/parsers.js +81 -0
  23. package/dist/cjs/content/link-preview/content/parsers.js.map +1 -0
  24. package/dist/cjs/content/link-preview/content/podcast-utils.js +85 -0
  25. package/dist/cjs/content/link-preview/content/podcast-utils.js.map +1 -0
  26. package/dist/cjs/content/link-preview/content/readability.js +90 -0
  27. package/dist/cjs/content/link-preview/content/readability.js.map +1 -0
  28. package/dist/cjs/content/link-preview/content/twitter-utils.js +74 -0
  29. package/dist/cjs/content/link-preview/content/twitter-utils.js.map +1 -0
  30. package/dist/cjs/content/link-preview/content/types.js +7 -0
  31. package/dist/cjs/content/link-preview/content/types.js.map +1 -0
  32. package/dist/cjs/content/link-preview/content/utils.js +177 -0
  33. package/dist/cjs/content/link-preview/content/utils.js.map +1 -0
  34. package/dist/cjs/content/link-preview/content/video.js +99 -0
  35. package/dist/cjs/content/link-preview/content/video.js.map +1 -0
  36. package/dist/cjs/content/link-preview/content/youtube.js +85 -0
  37. package/dist/cjs/content/link-preview/content/youtube.js.map +1 -0
  38. package/dist/cjs/content/link-preview/deps.js +23 -0
  39. package/dist/cjs/content/link-preview/deps.js.map +1 -0
  40. package/dist/cjs/content/link-preview/fetch-with-timeout.js +38 -0
  41. package/dist/cjs/content/link-preview/fetch-with-timeout.js.map +1 -0
  42. package/dist/cjs/content/link-preview/types.js +5 -0
  43. package/dist/cjs/content/link-preview/types.js.map +1 -0
  44. package/dist/cjs/content/transcript/cache.js +85 -0
  45. package/dist/cjs/content/transcript/cache.js.map +1 -0
  46. package/dist/cjs/content/transcript/index.js +134 -0
  47. package/dist/cjs/content/transcript/index.js.map +1 -0
  48. package/dist/cjs/content/transcript/normalize.js +49 -0
  49. package/dist/cjs/content/transcript/normalize.js.map +1 -0
  50. package/dist/cjs/content/transcript/providers/generic.js +16 -0
  51. package/dist/cjs/content/transcript/providers/generic.js.map +1 -0
  52. package/dist/cjs/content/transcript/providers/podcast/apple-flow.js +226 -0
  53. package/dist/cjs/content/transcript/providers/podcast/apple-flow.js.map +1 -0
  54. package/dist/cjs/content/transcript/providers/podcast/apple.js +43 -0
  55. package/dist/cjs/content/transcript/providers/podcast/apple.js.map +1 -0
  56. package/dist/cjs/content/transcript/providers/podcast/constants.js +11 -0
  57. package/dist/cjs/content/transcript/providers/podcast/constants.js.map +1 -0
  58. package/dist/cjs/content/transcript/providers/podcast/flow-context.js +3 -0
  59. package/dist/cjs/content/transcript/providers/podcast/flow-context.js.map +1 -0
  60. package/dist/cjs/content/transcript/providers/podcast/itunes.js +139 -0
  61. package/dist/cjs/content/transcript/providers/podcast/itunes.js.map +1 -0
  62. package/dist/cjs/content/transcript/providers/podcast/json.js +43 -0
  63. package/dist/cjs/content/transcript/providers/podcast/json.js.map +1 -0
  64. package/dist/cjs/content/transcript/providers/podcast/media.js +355 -0
  65. package/dist/cjs/content/transcript/providers/podcast/media.js.map +1 -0
  66. package/dist/cjs/content/transcript/providers/podcast/results.js +32 -0
  67. package/dist/cjs/content/transcript/providers/podcast/results.js.map +1 -0
  68. package/dist/cjs/content/transcript/providers/podcast/rss.js +262 -0
  69. package/dist/cjs/content/transcript/providers/podcast/rss.js.map +1 -0
  70. package/dist/cjs/content/transcript/providers/podcast/spotify-flow.js +221 -0
  71. package/dist/cjs/content/transcript/providers/podcast/spotify-flow.js.map +1 -0
  72. package/dist/cjs/content/transcript/providers/podcast/spotify.js +119 -0
  73. package/dist/cjs/content/transcript/providers/podcast/spotify.js.map +1 -0
  74. package/dist/cjs/content/transcript/providers/podcast.js +260 -0
  75. package/dist/cjs/content/transcript/providers/podcast.js.map +1 -0
  76. package/dist/cjs/content/transcript/providers/youtube/api.js +264 -0
  77. package/dist/cjs/content/transcript/providers/youtube/api.js.map +1 -0
  78. package/dist/cjs/content/transcript/providers/youtube/apify.js +59 -0
  79. package/dist/cjs/content/transcript/providers/youtube/apify.js.map +1 -0
  80. package/dist/cjs/content/transcript/providers/youtube/captions.js +413 -0
  81. package/dist/cjs/content/transcript/providers/youtube/captions.js.map +1 -0
  82. package/dist/cjs/content/transcript/providers/youtube/yt-dlp.js +170 -0
  83. package/dist/cjs/content/transcript/providers/youtube/yt-dlp.js.map +1 -0
  84. package/dist/cjs/content/transcript/providers/youtube.js +178 -0
  85. package/dist/cjs/content/transcript/providers/youtube.js.map +1 -0
  86. package/dist/cjs/content/transcript/types.js +3 -0
  87. package/dist/cjs/content/transcript/types.js.map +1 -0
  88. package/dist/cjs/content/transcript/utils.js +303 -0
  89. package/dist/cjs/content/transcript/utils.js.map +1 -0
  90. package/dist/cjs/index.js +22 -0
  91. package/dist/cjs/index.js.map +1 -0
  92. package/dist/cjs/language.js +132 -0
  93. package/dist/cjs/language.js.map +1 -0
  94. package/dist/cjs/package.json +3 -0
  95. package/dist/cjs/prompts/cli.js +23 -0
  96. package/dist/cjs/prompts/cli.js.map +1 -0
  97. package/dist/cjs/prompts/file.js +52 -0
  98. package/dist/cjs/prompts/file.js.map +1 -0
  99. package/dist/cjs/prompts/index.js +14 -0
  100. package/dist/cjs/prompts/index.js.map +1 -0
  101. package/dist/cjs/prompts/link-summary.js +122 -0
  102. package/dist/cjs/prompts/link-summary.js.map +1 -0
  103. package/dist/cjs/shared/contracts.js +5 -0
  104. package/dist/cjs/shared/contracts.js.map +1 -0
  105. package/dist/cjs/transcription/whisper/constants.js +11 -0
  106. package/dist/cjs/transcription/whisper/constants.js.map +1 -0
  107. package/dist/cjs/transcription/whisper/core.js +307 -0
  108. package/dist/cjs/transcription/whisper/core.js.map +1 -0
  109. package/dist/cjs/transcription/whisper/fal.js +44 -0
  110. package/dist/cjs/transcription/whisper/fal.js.map +1 -0
  111. package/dist/cjs/transcription/whisper/ffmpeg.js +187 -0
  112. package/dist/cjs/transcription/whisper/ffmpeg.js.map +1 -0
  113. package/dist/cjs/transcription/whisper/openai.js +51 -0
  114. package/dist/cjs/transcription/whisper/openai.js.map +1 -0
  115. package/dist/cjs/transcription/whisper/types.js +3 -0
  116. package/dist/cjs/transcription/whisper/types.js.map +1 -0
  117. package/dist/cjs/transcription/whisper/utils.js +70 -0
  118. package/dist/cjs/transcription/whisper/utils.js.map +1 -0
  119. package/dist/cjs/transcription/whisper/whisper-cpp.js +232 -0
  120. package/dist/cjs/transcription/whisper/whisper-cpp.js.map +1 -0
  121. package/dist/cjs/transcription/whisper.js +15 -0
  122. package/dist/cjs/transcription/whisper.js.map +1 -0
  123. package/package.json +15 -12
@@ -0,0 +1 @@
1
+ {"version":3,"file":"video.js","sourceRoot":"","sources":["../../../../../src/content/link-preview/content/video.ts"],"names":[],"mappings":";;AA+DA,gEA2CC;AA1GD,qCAA+C;AAO/C,MAAM,gBAAgB,GAAG,IAAI,GAAG,CAAC,CAAC,MAAM,EAAE,OAAO,EAAE,MAAM,EAAE,MAAM,CAAC,CAAC,CAAA;AAEnE,SAAS,kBAAkB,CAAC,SAAiB,EAAE,OAAe;IAC5D,MAAM,OAAO,GAAG,SAAS,CAAC,IAAI,EAAE,CAAA;IAChC,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,IAAI,CAAA;IACrC,IAAI,CAAC;QACH,OAAO,IAAI,GAAG,CAAC,OAAO,EAAE,OAAO,CAAC,CAAC,QAAQ,EAAE,CAAA;IAC7C,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,IAAI,CAAA;IACb,CAAC;AACH,CAAC;AAED,SAAS,gBAAgB,CAAC,GAAW;IACnC,IAAI,CAAC;QACH,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAA;QAC3B,MAAM,KAAK,GAAG,MAAM,CAAC,QAAQ,CAAC,WAAW,EAAE,CAAA;QAC3C,KAAK,MAAM,GAAG,IAAI,gBAAgB,EAAE,CAAC;YACnC,IAAI,KAAK,CAAC,QAAQ,CAAC,GAAG,CAAC;gBAAE,OAAO,IAAI,CAAA;QACtC,CAAC;QACD,OAAO,KAAK,CAAA;IACd,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,KAAK,CAAA;IACd,CAAC;AACH,CAAC;AAED,SAAS,iCAAiC,CAAC,GAAW;IACpD,IAAI,CAAC;QACH,MAAM,CAAC,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAA;QACtB,MAAM,IAAI,GAAG,CAAC,CAAC,QAAQ,CAAC,WAAW,EAAE,CAAC,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAA;QAC3D,IAAI,IAAI,KAAK,aAAa,IAAI,IAAI,CAAC,QAAQ,CAAC,cAAc,CAAC,EAAE,CAAC;YAC5D,MAAM,CAAC,GAAG,CAAC,CAAC,QAAQ,CAAC,KAAK,CAAC,8BAA8B,CAAC,CAAA;YAC1D,OAAO,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,IAAI,CAAA;QACvB,CAAC;QACD,IAAI,IAAI,KAAK,UAAU,EAAE,CAAC;YACxB,MAAM,EAAE,GAAG,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;YAC/C,OAAO,qBAAqB,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,CAAA;QACnD,CAAC;QACD,OAAO,IAAI,CAAA;IACb,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,IAAI,CAAA;IACb,CAAC;AACH,CAAC;AAED,SAAS,WAAW,CAClB,CAAa,EACb,SAAmE;IAEnE,KAAK,MAAM,GAAG,IAAI,SAAS,EAAE,CAAC;QAC5B,MAAM,IAAI,GAAG,CAAC,CAAC,QAAQ,GAAG,CAAC,SAAS,KAAK,GAAG,CAAC,KAAK,IAAI,CAAC,CAAC,KAAK,EAAE,CAAA;QAC/D,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;YAAE,SAAQ;QAC/B,MAAM,KAAK,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,IAAI,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;QACvE,IAAI,KAAK;YAAE,OAAO,KAAK,CAAA;IACzB,CAAC;IACD,OAAO,IAAI,CAAA;AACb,CAAC;AAED,SAAgB,0BAA0B,CAAC,IAAY,EAAE,GAAW;IAClE,MAAM,CAAC,GAAG,IAAA,cAAI,EAAC,IAAI,CAAC,CAAA;IAEpB,wCAAwC;IACxC,MAAM,SAAS,GACb,CAAC,CAAC,6DAA6D,CAAC,CAAC,KAAK,EAAE,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,IAAI,CAAA;IAC9F,IAAI,SAAS,EAAE,CAAC;QACd,MAAM,QAAQ,GAAG,kBAAkB,CAAC,SAAS,EAAE,GAAG,CAAC,CAAA;QACnD,MAAM,OAAO,GAAG,QAAQ,CAAC,CAAC,CAAC,iCAAiC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,IAAI,CAAA;QAC7E,IAAI,OAAO,EAAE,CAAC;YACZ,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,GAAG,EAAE,mCAAmC,OAAO,EAAE,EAAE,CAAA;QAC/E,CAAC;IACH,CAAC;IAED,qBAAqB;IACrB,MAAM,OAAO,GAAG,WAAW,CAAC,CAAC,EAAE;QAC7B,EAAE,SAAS,EAAE,UAAU,EAAE,KAAK,EAAE,UAAU,EAAE;QAC5C,EAAE,SAAS,EAAE,UAAU,EAAE,KAAK,EAAE,cAAc,EAAE;QAChD,EAAE,SAAS,EAAE,UAAU,EAAE,KAAK,EAAE,qBAAqB,EAAE;QACvD,EAAE,SAAS,EAAE,MAAM,EAAE,KAAK,EAAE,UAAU,EAAE;QACxC,EAAE,SAAS,EAAE,MAAM,EAAE,KAAK,EAAE,cAAc,EAAE;QAC5C,EAAE,SAAS,EAAE,MAAM,EAAE,KAAK,EAAE,qBAAqB,EAAE;KACpD,CAAC,CAAA;IACF,IAAI,OAAO,EAAE,CAAC;QACZ,MAAM,QAAQ,GAAG,kBAAkB,CAAC,OAAO,EAAE,GAAG,CAAC,CAAA;QACjD,IAAI,QAAQ,IAAI,gBAAgB,CAAC,QAAQ,CAAC,EAAE,CAAC;YAC3C,OAAO,EAAE,IAAI,EAAE,QAAQ,EAAE,GAAG,EAAE,QAAQ,EAAE,CAAA;QAC1C,CAAC;QACD,MAAM,IAAI,GAAG,QAAQ,CAAC,CAAC,CAAC,iCAAiC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,IAAI,CAAA;QAC1E,IAAI,IAAI;YAAE,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,GAAG,EAAE,mCAAmC,IAAI,EAAE,EAAE,CAAA;IACtF,CAAC;IAED,kBAAkB;IAClB,MAAM,QAAQ,GACZ,CAAC,CAAC,YAAY,CAAC,CAAC,KAAK,EAAE,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,mBAAmB,CAAC,CAAC,KAAK,EAAE,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,IAAI,CAAA;IAC3F,IAAI,QAAQ,EAAE,CAAC;QACb,MAAM,QAAQ,GAAG,kBAAkB,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAA;QAClD,IAAI,QAAQ,IAAI,gBAAgB,CAAC,QAAQ,CAAC,EAAE,CAAC;YAC3C,OAAO,EAAE,IAAI,EAAE,QAAQ,EAAE,GAAG,EAAE,QAAQ,EAAE,CAAA;QAC1C,CAAC;IACH,CAAC;IAED,OAAO,IAAI,CAAA;AACb,CAAC"}
@@ -0,0 +1,85 @@
1
+ "use strict";
2
+ Object.defineProperty(exports, "__esModule", { value: true });
3
+ exports.extractYouTubeShortDescription = extractYouTubeShortDescription;
4
+ const cleaner_js_1 = require("./cleaner.js");
5
+ function extractBalancedJsonObject(source, startAt) {
6
+ const start = source.indexOf('{', startAt);
7
+ if (start < 0) {
8
+ return null;
9
+ }
10
+ let depth = 0;
11
+ let inString = false;
12
+ let quote = null;
13
+ let escaping = false;
14
+ for (let i = start; i < source.length; i += 1) {
15
+ const ch = source[i];
16
+ if (!ch) {
17
+ continue;
18
+ }
19
+ if (inString) {
20
+ if (escaping) {
21
+ escaping = false;
22
+ continue;
23
+ }
24
+ if (ch === '\\') {
25
+ escaping = true;
26
+ continue;
27
+ }
28
+ if (quote && ch === quote) {
29
+ inString = false;
30
+ quote = null;
31
+ }
32
+ continue;
33
+ }
34
+ if (ch === '"' || ch === "'") {
35
+ inString = true;
36
+ quote = ch;
37
+ continue;
38
+ }
39
+ if (ch === '{') {
40
+ depth += 1;
41
+ continue;
42
+ }
43
+ if (ch === '}') {
44
+ depth -= 1;
45
+ if (depth === 0) {
46
+ return source.slice(start, i + 1);
47
+ }
48
+ }
49
+ }
50
+ return null;
51
+ }
52
+ function extractYouTubeShortDescription(html) {
53
+ const tokenIndex = html.indexOf('ytInitialPlayerResponse');
54
+ if (tokenIndex < 0) {
55
+ return null;
56
+ }
57
+ const assignmentIndex = html.indexOf('=', tokenIndex);
58
+ if (assignmentIndex < 0) {
59
+ return null;
60
+ }
61
+ const objectText = extractBalancedJsonObject(html, assignmentIndex);
62
+ if (!objectText) {
63
+ return null;
64
+ }
65
+ try {
66
+ const parsed = JSON.parse(objectText);
67
+ if (!parsed || typeof parsed !== 'object') {
68
+ return null;
69
+ }
70
+ const videoDetails = parsed.videoDetails;
71
+ if (!videoDetails || typeof videoDetails !== 'object') {
72
+ return null;
73
+ }
74
+ const description = videoDetails.shortDescription;
75
+ if (typeof description !== 'string') {
76
+ return null;
77
+ }
78
+ const normalized = (0, cleaner_js_1.normalizeWhitespace)(description);
79
+ return normalized && normalized.length > 0 ? normalized : null;
80
+ }
81
+ catch {
82
+ return null;
83
+ }
84
+ }
85
+ //# sourceMappingURL=youtube.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"youtube.js","sourceRoot":"","sources":["../../../../../src/content/link-preview/content/youtube.ts"],"names":[],"mappings":";;AAwDA,wEAiCC;AAzFD,6CAAkD;AAElD,SAAS,yBAAyB,CAAC,MAAc,EAAE,OAAe;IAChE,MAAM,KAAK,GAAG,MAAM,CAAC,OAAO,CAAC,GAAG,EAAE,OAAO,CAAC,CAAA;IAC1C,IAAI,KAAK,GAAG,CAAC,EAAE,CAAC;QACd,OAAO,IAAI,CAAA;IACb,CAAC;IAED,IAAI,KAAK,GAAG,CAAC,CAAA;IACb,IAAI,QAAQ,GAAG,KAAK,CAAA;IACpB,IAAI,KAAK,GAAqB,IAAI,CAAA;IAClC,IAAI,QAAQ,GAAG,KAAK,CAAA;IAEpB,KAAK,IAAI,CAAC,GAAG,KAAK,EAAE,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC,IAAI,CAAC,EAAE,CAAC;QAC9C,MAAM,EAAE,GAAG,MAAM,CAAC,CAAC,CAAC,CAAA;QACpB,IAAI,CAAC,EAAE,EAAE,CAAC;YACR,SAAQ;QACV,CAAC;QAED,IAAI,QAAQ,EAAE,CAAC;YACb,IAAI,QAAQ,EAAE,CAAC;gBACb,QAAQ,GAAG,KAAK,CAAA;gBAChB,SAAQ;YACV,CAAC;YACD,IAAI,EAAE,KAAK,IAAI,EAAE,CAAC;gBAChB,QAAQ,GAAG,IAAI,CAAA;gBACf,SAAQ;YACV,CAAC;YACD,IAAI,KAAK,IAAI,EAAE,KAAK,KAAK,EAAE,CAAC;gBAC1B,QAAQ,GAAG,KAAK,CAAA;gBAChB,KAAK,GAAG,IAAI,CAAA;YACd,CAAC;YACD,SAAQ;QACV,CAAC;QAED,IAAI,EAAE,KAAK,GAAG,IAAI,EAAE,KAAK,GAAG,EAAE,CAAC;YAC7B,QAAQ,GAAG,IAAI,CAAA;YACf,KAAK,GAAG,EAAE,CAAA;YACV,SAAQ;QACV,CAAC;QAED,IAAI,EAAE,KAAK,GAAG,EAAE,CAAC;YACf,KAAK,IAAI,CAAC,CAAA;YACV,SAAQ;QACV,CAAC;QACD,IAAI,EAAE,KAAK,GAAG,EAAE,CAAC;YACf,KAAK,IAAI,CAAC,CAAA;YACV,IAAI,KAAK,KAAK,CAAC,EAAE,CAAC;gBAChB,OAAO,MAAM,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC,GAAG,CAAC,CAAC,CAAA;YACnC,CAAC;QACH,CAAC;IACH,CAAC;IAED,OAAO,IAAI,CAAA;AACb,CAAC;AAED,SAAgB,8BAA8B,CAAC,IAAY;IACzD,MAAM,UAAU,GAAG,IAAI,CAAC,OAAO,CAAC,yBAAyB,CAAC,CAAA;IAC1D,IAAI,UAAU,GAAG,CAAC,EAAE,CAAC;QACnB,OAAO,IAAI,CAAA;IACb,CAAC;IACD,MAAM,eAAe,GAAG,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,UAAU,CAAC,CAAA;IACrD,IAAI,eAAe,GAAG,CAAC,EAAE,CAAC;QACxB,OAAO,IAAI,CAAA;IACb,CAAC;IACD,MAAM,UAAU,GAAG,yBAAyB,CAAC,IAAI,EAAE,eAAe,CAAC,CAAA;IACnE,IAAI,CAAC,UAAU,EAAE,CAAC;QAChB,OAAO,IAAI,CAAA;IACb,CAAC;IAED,IAAI,CAAC;QACH,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,UAAU,CAAY,CAAA;QAChD,IAAI,CAAC,MAAM,IAAI,OAAO,MAAM,KAAK,QAAQ,EAAE,CAAC;YAC1C,OAAO,IAAI,CAAA;QACb,CAAC;QACD,MAAM,YAAY,GAAI,MAAkC,CAAC,YAAY,CAAA;QACrE,IAAI,CAAC,YAAY,IAAI,OAAO,YAAY,KAAK,QAAQ,EAAE,CAAC;YACtD,OAAO,IAAI,CAAA;QACb,CAAC;QACD,MAAM,WAAW,GAAI,YAAwC,CAAC,gBAAgB,CAAA;QAC9E,IAAI,OAAO,WAAW,KAAK,QAAQ,EAAE,CAAC;YACpC,OAAO,IAAI,CAAA;QACb,CAAC;QAED,MAAM,UAAU,GAAG,IAAA,gCAAmB,EAAC,WAAW,CAAC,CAAA;QACnD,OAAO,UAAU,IAAI,UAAU,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,IAAI,CAAA;IAChE,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,IAAI,CAAA;IACb,CAAC;AACH,CAAC"}
@@ -0,0 +1,23 @@
1
+ "use strict";
2
+ Object.defineProperty(exports, "__esModule", { value: true });
3
+ exports.ProgressKind = void 0;
4
+ // Enum-like constants for progress kinds (keeps call sites typo-resistant without TS `enum` runtime quirks).
5
+ exports.ProgressKind = {
6
+ FetchHtmlStart: 'fetch-html-start',
7
+ FetchHtmlProgress: 'fetch-html-progress',
8
+ FetchHtmlDone: 'fetch-html-done',
9
+ TranscriptMediaDownloadStart: 'transcript-media-download-start',
10
+ TranscriptMediaDownloadProgress: 'transcript-media-download-progress',
11
+ TranscriptMediaDownloadDone: 'transcript-media-download-done',
12
+ TranscriptWhisperStart: 'transcript-whisper-start',
13
+ TranscriptWhisperProgress: 'transcript-whisper-progress',
14
+ TranscriptStart: 'transcript-start',
15
+ TranscriptDone: 'transcript-done',
16
+ FirecrawlStart: 'firecrawl-start',
17
+ FirecrawlDone: 'firecrawl-done',
18
+ NitterStart: 'nitter-start',
19
+ NitterDone: 'nitter-done',
20
+ BirdStart: 'bird-start',
21
+ BirdDone: 'bird-done',
22
+ };
23
+ //# sourceMappingURL=deps.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"deps.js","sourceRoot":"","sources":["../../../../src/content/link-preview/deps.ts"],"names":[],"mappings":";;;AAEA,6GAA6G;AAChG,QAAA,YAAY,GAAG;IAC1B,cAAc,EAAE,kBAAkB;IAClC,iBAAiB,EAAE,qBAAqB;IACxC,aAAa,EAAE,iBAAiB;IAEhC,4BAA4B,EAAE,iCAAiC;IAC/D,+BAA+B,EAAE,oCAAoC;IACrE,2BAA2B,EAAE,gCAAgC;IAE7D,sBAAsB,EAAE,0BAA0B;IAClD,yBAAyB,EAAE,6BAA6B;IAExD,eAAe,EAAE,kBAAkB;IACnC,cAAc,EAAE,iBAAiB;IAEjC,cAAc,EAAE,iBAAiB;IACjC,aAAa,EAAE,gBAAgB;IAE/B,WAAW,EAAE,cAAc;IAC3B,UAAU,EAAE,aAAa;IAEzB,SAAS,EAAE,YAAY;IACvB,QAAQ,EAAE,WAAW;CACb,CAAA"}
@@ -0,0 +1,38 @@
1
+ "use strict";
2
+ Object.defineProperty(exports, "__esModule", { value: true });
3
+ exports.fetchWithTimeout = fetchWithTimeout;
4
+ const DEFAULT_TIMEOUT_MS = 120_000;
5
+ async function fetchWithTimeout(fetchImpl, input, init, timeoutMs = DEFAULT_TIMEOUT_MS) {
6
+ if (init?.signal) {
7
+ return fetchImpl(input, init ?? {});
8
+ }
9
+ const controller = new AbortController();
10
+ const normalizedTimeoutMs = Number.isFinite(timeoutMs) ? timeoutMs : DEFAULT_TIMEOUT_MS;
11
+ const clampedTimeoutMs = Math.max(0, normalizedTimeoutMs);
12
+ const timer = setTimeout(() => {
13
+ if (typeof DOMException === 'function') {
14
+ controller.abort(new DOMException('Request timed out', 'AbortError'));
15
+ return;
16
+ }
17
+ controller.abort();
18
+ }, clampedTimeoutMs);
19
+ try {
20
+ const finalInit = {
21
+ ...init,
22
+ signal: controller.signal,
23
+ };
24
+ return await fetchImpl(input, finalInit);
25
+ }
26
+ catch (error) {
27
+ if (error instanceof Error && error.name === 'AbortError') {
28
+ const timeoutError = new Error(`Fetch aborted after ${clampedTimeoutMs}ms`);
29
+ timeoutError.name = 'FetchTimeoutError';
30
+ throw timeoutError;
31
+ }
32
+ throw error;
33
+ }
34
+ finally {
35
+ clearTimeout(timer);
36
+ }
37
+ }
38
+ //# sourceMappingURL=fetch-with-timeout.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"fetch-with-timeout.js","sourceRoot":"","sources":["../../../../src/content/link-preview/fetch-with-timeout.ts"],"names":[],"mappings":";;AAKA,4CAsCC;AA3CD,MAAM,kBAAkB,GAAG,OAAO,CAAA;AAK3B,KAAK,UAAU,gBAAgB,CACpC,SAAoB,EACpB,KAAwB,EACxB,IAAwB,EACxB,YAAoB,kBAAkB;IAEtC,IAAI,IAAI,EAAE,MAAM,EAAE,CAAC;QACjB,OAAO,SAAS,CAAC,KAAK,EAAE,IAAI,IAAI,EAAE,CAAC,CAAA;IACrC,CAAC;IAED,MAAM,UAAU,GAAG,IAAI,eAAe,EAAE,CAAA;IACxC,MAAM,mBAAmB,GAAG,MAAM,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,kBAAkB,CAAA;IACvF,MAAM,gBAAgB,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,mBAAmB,CAAC,CAAA;IAEzD,MAAM,KAAK,GAAG,UAAU,CAAC,GAAG,EAAE;QAC5B,IAAI,OAAO,YAAY,KAAK,UAAU,EAAE,CAAC;YACvC,UAAU,CAAC,KAAK,CAAC,IAAI,YAAY,CAAC,mBAAmB,EAAE,YAAY,CAAC,CAAC,CAAA;YACrE,OAAM;QACR,CAAC;QACD,UAAU,CAAC,KAAK,EAAE,CAAA;IACpB,CAAC,EAAE,gBAAgB,CAAC,CAAA;IAEpB,IAAI,CAAC;QACH,MAAM,SAAS,GAAgB;YAC7B,GAAG,IAAI;YACP,MAAM,EAAE,UAAU,CAAC,MAAM;SAC1B,CAAA;QACD,OAAO,MAAM,SAAS,CAAC,KAAK,EAAE,SAAS,CAAC,CAAA;IAC1C,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,IAAI,KAAK,YAAY,KAAK,IAAI,KAAK,CAAC,IAAI,KAAK,YAAY,EAAE,CAAC;YAC1D,MAAM,YAAY,GAAG,IAAI,KAAK,CAAC,uBAAuB,gBAAgB,IAAI,CAAC,CAAA;YAC3E,YAAY,CAAC,IAAI,GAAG,mBAAmB,CAAA;YACvC,MAAM,YAAY,CAAA;QACpB,CAAC;QACD,MAAM,KAAK,CAAA;IACb,CAAC;YAAS,CAAC;QACT,YAAY,CAAC,KAAK,CAAC,CAAA;IACrB,CAAC;AACH,CAAC"}
@@ -0,0 +1,5 @@
1
+ "use strict";
2
+ Object.defineProperty(exports, "__esModule", { value: true });
3
+ exports.CACHE_MODES = void 0;
4
+ exports.CACHE_MODES = ['default', 'bypass'];
5
+ //# sourceMappingURL=types.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"types.js","sourceRoot":"","sources":["../../../../src/content/link-preview/types.ts"],"names":[],"mappings":";;;AAWa,QAAA,WAAW,GAAG,CAAC,SAAS,EAAE,QAAQ,CAAU,CAAA"}
@@ -0,0 +1,85 @@
1
+ "use strict";
2
+ Object.defineProperty(exports, "__esModule", { value: true });
3
+ exports.writeTranscriptCache = exports.mapCachedSource = exports.readTranscriptCache = exports.NEGATIVE_TTL_MS = exports.DEFAULT_TTL_MS = void 0;
4
+ exports.DEFAULT_TTL_MS = 1000 * 60 * 60 * 24 * 7;
5
+ exports.NEGATIVE_TTL_MS = 1000 * 60 * 60 * 6;
6
+ const readTranscriptCache = async ({ url, cacheMode, transcriptCache, }) => {
7
+ const cached = transcriptCache ? await transcriptCache.get({ url }) : null;
8
+ const diagnostics = buildBaseDiagnostics(cacheMode);
9
+ if (!cached) {
10
+ return { cached: null, resolution: null, diagnostics };
11
+ }
12
+ const provider = (0, exports.mapCachedSource)(cached.source);
13
+ diagnostics.provider = provider;
14
+ diagnostics.attemptedProviders = provider ? [provider] : [];
15
+ diagnostics.textProvided = Boolean(cached.content && cached.content.length > 0);
16
+ if (cacheMode === 'bypass') {
17
+ diagnostics.notes = appendNote(diagnostics.notes, 'Cached transcript ignored due to bypass request');
18
+ return { cached, resolution: null, diagnostics };
19
+ }
20
+ if (cached.expired) {
21
+ diagnostics.cacheStatus = 'expired';
22
+ diagnostics.notes = appendNote(diagnostics.notes, 'Cached transcript expired; fetching fresh copy');
23
+ return { cached, resolution: null, diagnostics };
24
+ }
25
+ diagnostics.cacheStatus = 'hit';
26
+ diagnostics.notes = appendNote(diagnostics.notes, 'Served transcript from cache');
27
+ const resolution = {
28
+ text: cached.content,
29
+ source: provider,
30
+ metadata: cached.metadata ?? null,
31
+ };
32
+ return { cached, resolution, diagnostics };
33
+ };
34
+ exports.readTranscriptCache = readTranscriptCache;
35
+ const buildBaseDiagnostics = (cacheMode) => ({
36
+ cacheMode,
37
+ cacheStatus: cacheMode === 'bypass' ? 'bypassed' : 'miss',
38
+ provider: null,
39
+ attemptedProviders: [],
40
+ textProvided: false,
41
+ notes: cacheMode === 'bypass' ? 'Cache bypass requested' : null,
42
+ });
43
+ const appendNote = (existing, next) => {
44
+ if (!existing) {
45
+ return next;
46
+ }
47
+ return `${existing}; ${next}`;
48
+ };
49
+ const mapCachedSource = (source) => {
50
+ if (source === null)
51
+ return null;
52
+ if (source === 'youtubei' ||
53
+ source === 'captionTracks' ||
54
+ source === 'yt-dlp' ||
55
+ source === 'podcastTranscript' ||
56
+ source === 'whisper' ||
57
+ source === 'apify' ||
58
+ source === 'html' ||
59
+ source === 'unavailable') {
60
+ return source;
61
+ }
62
+ return 'unknown';
63
+ };
64
+ exports.mapCachedSource = mapCachedSource;
65
+ const writeTranscriptCache = async ({ url, service, resourceKey, result, transcriptCache, }) => {
66
+ if (!transcriptCache) {
67
+ return;
68
+ }
69
+ if (result.source === null && result.text === null) {
70
+ return;
71
+ }
72
+ const ttlMs = result.text ? exports.DEFAULT_TTL_MS : exports.NEGATIVE_TTL_MS;
73
+ const resolvedSource = result.source ?? (result.text ? 'unknown' : 'unavailable');
74
+ await transcriptCache.set({
75
+ url,
76
+ service,
77
+ resourceKey,
78
+ ttlMs,
79
+ content: result.text,
80
+ source: resolvedSource,
81
+ metadata: result.metadata ?? null,
82
+ });
83
+ };
84
+ exports.writeTranscriptCache = writeTranscriptCache;
85
+ //# sourceMappingURL=cache.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"cache.js","sourceRoot":"","sources":["../../../../src/content/transcript/cache.ts"],"names":[],"mappings":";;;AAQa,QAAA,cAAc,GAAG,IAAI,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AACxC,QAAA,eAAe,GAAG,IAAI,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AAmB1C,MAAM,mBAAmB,GAAG,KAAK,EAAE,EACxC,GAAG,EACH,SAAS,EACT,eAAe,GACI,EAAkC,EAAE;IACvD,MAAM,MAAM,GAAG,eAAe,CAAC,CAAC,CAAC,MAAM,eAAe,CAAC,GAAG,CAAC,EAAE,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,CAAA;IAC1E,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,IAAA,uBAAe,EAAC,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,UAAU,GAAyB;QACvC,IAAI,EAAE,MAAM,CAAC,OAAO;QACpB,MAAM,EAAE,QAAQ;QAChB,QAAQ,EAAE,MAAM,CAAC,QAAQ,IAAI,IAAI;KAClC,CAAA;IACD,OAAO,EAAE,MAAM,EAAE,UAAU,EAAE,WAAW,EAAE,CAAA;AAC5C,CAAC,CAAA;AA3CY,QAAA,mBAAmB,uBA2C/B;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;AAEM,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,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;AAfY,QAAA,eAAe,mBAe3B;AAEM,MAAM,oBAAoB,GAAG,KAAK,EAAE,EACzC,GAAG,EACH,OAAO,EACP,WAAW,EACX,MAAM,EACN,eAAe,GAWhB,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,sBAAc,CAAC,CAAC,CAAC,uBAAe,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;KAClC,CAAC,CAAA;AACJ,CAAC,CAAA;AArCY,QAAA,oBAAoB,wBAqChC"}
@@ -0,0 +1,134 @@
1
+ "use strict";
2
+ Object.defineProperty(exports, "__esModule", { value: true });
3
+ exports.resolveTranscriptForLink = void 0;
4
+ const cache_js_1 = require("./cache.js");
5
+ const generic_js_1 = require("./providers/generic.js");
6
+ const podcast_js_1 = require("./providers/podcast.js");
7
+ const youtube_js_1 = require("./providers/youtube.js");
8
+ const utils_js_1 = require("./utils.js");
9
+ const PROVIDERS = [
10
+ { id: 'youtube', canHandle: youtube_js_1.canHandle, fetchTranscript: youtube_js_1.fetchTranscript },
11
+ { id: 'podcast', canHandle: podcast_js_1.canHandle, fetchTranscript: podcast_js_1.fetchTranscript },
12
+ { id: 'generic', canHandle: generic_js_1.canHandle, fetchTranscript: generic_js_1.fetchTranscript },
13
+ ];
14
+ const GENERIC_PROVIDER_ID = 'generic';
15
+ const resolveTranscriptForLink = async (url, html, deps, { youtubeTranscriptMode, cacheMode: providedCacheMode } = {}) => {
16
+ const normalizedUrl = url.trim();
17
+ const embeddedYoutubeUrl = !(0, utils_js_1.isYouTubeUrl)(normalizedUrl) && html
18
+ ? await (0, utils_js_1.extractEmbeddedYouTubeUrlFromHtml)(html)
19
+ : null;
20
+ const effectiveUrl = embeddedYoutubeUrl ?? normalizedUrl;
21
+ const resourceKey = extractResourceKey(effectiveUrl);
22
+ const baseContext = { url: effectiveUrl, html, resourceKey };
23
+ const provider = selectProvider(baseContext);
24
+ const cacheMode = providedCacheMode ?? 'default';
25
+ const cacheOutcome = await (0, cache_js_1.readTranscriptCache)({
26
+ url: normalizedUrl,
27
+ cacheMode,
28
+ transcriptCache: deps.transcriptCache,
29
+ });
30
+ const diagnostics = {
31
+ cacheMode,
32
+ cacheStatus: cacheOutcome.diagnostics.cacheStatus,
33
+ textProvided: cacheOutcome.diagnostics.textProvided,
34
+ provider: cacheOutcome.diagnostics.provider,
35
+ attemptedProviders: [],
36
+ notes: cacheOutcome.diagnostics.notes ?? null,
37
+ };
38
+ if (cacheOutcome.resolution) {
39
+ return {
40
+ ...cacheOutcome.resolution,
41
+ diagnostics,
42
+ };
43
+ }
44
+ const shouldReportProgress = provider.id === 'youtube' || provider.id === 'podcast';
45
+ if (shouldReportProgress) {
46
+ deps.onProgress?.({
47
+ kind: 'transcript-start',
48
+ url: normalizedUrl,
49
+ service: provider.id,
50
+ hint: provider.id === 'youtube'
51
+ ? 'YouTube: resolving transcript'
52
+ : 'Podcast: resolving transcript',
53
+ });
54
+ }
55
+ const providerResult = await executeProvider(provider, baseContext, {
56
+ fetch: deps.fetch,
57
+ scrapeWithFirecrawl: deps.scrapeWithFirecrawl,
58
+ apifyApiToken: deps.apifyApiToken,
59
+ ytDlpPath: deps.ytDlpPath,
60
+ falApiKey: deps.falApiKey,
61
+ openaiApiKey: deps.openaiApiKey,
62
+ onProgress: deps.onProgress ?? null,
63
+ youtubeTranscriptMode: youtubeTranscriptMode ?? 'auto',
64
+ });
65
+ if (shouldReportProgress) {
66
+ deps.onProgress?.({
67
+ kind: 'transcript-done',
68
+ url: normalizedUrl,
69
+ ok: Boolean(providerResult.text && providerResult.text.length > 0),
70
+ service: provider.id,
71
+ source: providerResult.source,
72
+ hint: providerResult.source ? `${provider.id}/${providerResult.source}` : provider.id,
73
+ });
74
+ }
75
+ diagnostics.provider = providerResult.source;
76
+ diagnostics.attemptedProviders = providerResult.attemptedProviders;
77
+ diagnostics.textProvided = Boolean(providerResult.text && providerResult.text.length > 0);
78
+ if (providerResult.notes) {
79
+ diagnostics.notes = appendNote(diagnostics.notes, providerResult.notes);
80
+ }
81
+ if (providerResult.source !== null || providerResult.text !== null) {
82
+ await (0, cache_js_1.writeTranscriptCache)({
83
+ url: normalizedUrl,
84
+ service: provider.id,
85
+ resourceKey,
86
+ result: providerResult,
87
+ transcriptCache: deps.transcriptCache,
88
+ });
89
+ }
90
+ if (!providerResult.text && cacheOutcome.cached?.content && cacheMode !== 'bypass') {
91
+ diagnostics.cacheStatus = 'fallback';
92
+ diagnostics.provider = (0, cache_js_1.mapCachedSource)(cacheOutcome.cached.source);
93
+ diagnostics.textProvided = Boolean(cacheOutcome.cached.content && cacheOutcome.cached.content.length > 0);
94
+ diagnostics.notes = appendNote(diagnostics.notes, 'Falling back to cached transcript content after provider miss');
95
+ return {
96
+ text: cacheOutcome.cached.content,
97
+ source: diagnostics.provider,
98
+ metadata: cacheOutcome.cached.metadata ?? null,
99
+ diagnostics,
100
+ };
101
+ }
102
+ return {
103
+ text: providerResult.text,
104
+ source: providerResult.source,
105
+ metadata: providerResult.metadata ?? null,
106
+ diagnostics,
107
+ };
108
+ };
109
+ exports.resolveTranscriptForLink = resolveTranscriptForLink;
110
+ const extractResourceKey = (url) => {
111
+ if ((0, utils_js_1.isYouTubeUrl)(url)) {
112
+ return (0, utils_js_1.extractYouTubeVideoId)(url);
113
+ }
114
+ return null;
115
+ };
116
+ const selectProvider = (context) => {
117
+ const genericProviderModule = PROVIDERS.find((provider) => provider.id === GENERIC_PROVIDER_ID);
118
+ const specializedProvider = PROVIDERS.find((provider) => provider.id !== GENERIC_PROVIDER_ID && provider.canHandle(context));
119
+ if (specializedProvider) {
120
+ return specializedProvider;
121
+ }
122
+ if (genericProviderModule) {
123
+ return genericProviderModule;
124
+ }
125
+ throw new Error('Generic transcript provider is not registered');
126
+ };
127
+ const executeProvider = async (provider, context, options) => provider.fetchTranscript(context, options);
128
+ const appendNote = (existing, next) => {
129
+ if (!existing) {
130
+ return next;
131
+ }
132
+ return `${existing}; ${next}`;
133
+ };
134
+ //# sourceMappingURL=index.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"index.js","sourceRoot":"","sources":["../../../../src/content/transcript/index.ts"],"names":[],"mappings":";;;AAMA,yCAAuF;AACvF,uDAG+B;AAC/B,uDAG+B;AAC/B,uDAG+B;AAO/B,yCAImB;AAOnB,MAAM,SAAS,GAAqB;IAClC,EAAE,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,sBAAgB,EAAE,eAAe,EAAE,4BAAY,EAAE;IAC7E,EAAE,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,sBAAgB,EAAE,eAAe,EAAE,4BAAY,EAAE;IAC7E,EAAE,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,sBAAgB,EAAE,eAAe,EAAE,4BAAY,EAAE;CAC9E,CAAA;AACD,MAAM,mBAAmB,GAAG,SAAS,CAAA;AAE9B,MAAM,wBAAwB,GAAG,KAAK,EAC3C,GAAW,EACX,IAAmB,EACnB,IAAqB,EACrB,EAAE,qBAAqB,EAAE,SAAS,EAAE,iBAAiB,KAA+B,EAAE,EACvD,EAAE;IACjC,MAAM,aAAa,GAAG,GAAG,CAAC,IAAI,EAAE,CAAA;IAChC,MAAM,kBAAkB,GACtB,CAAC,IAAA,uBAAoB,EAAC,aAAa,CAAC,IAAI,IAAI;QAC1C,CAAC,CAAC,MAAM,IAAA,4CAAiC,EAAC,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,IAAA,8BAAmB,EAAC;QAC7C,GAAG,EAAE,aAAa;QAClB,SAAS;QACT,eAAe,EAAE,IAAI,CAAC,eAAe;KACtC,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,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,qBAAqB,IAAI,MAAM;KACvD,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,MAAM,IAAA,+BAAoB,EAAC;YACzB,GAAG,EAAE,aAAa;YAClB,OAAO,EAAE,QAAQ,CAAC,EAAE;YACpB,WAAW;YACX,MAAM,EAAE,cAAc;YACtB,eAAe,EAAE,IAAI,CAAC,eAAe;SACtC,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,IAAA,0BAAe,EAAC,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;SACZ,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;KACZ,CAAA;AACH,CAAC,CAAA;AApHY,QAAA,wBAAwB,4BAoHpC;AAED,MAAM,kBAAkB,GAAG,CAAC,GAAW,EAAiB,EAAE;IACxD,IAAI,IAAA,uBAAoB,EAAC,GAAG,CAAC,EAAE,CAAC;QAC9B,OAAO,IAAA,gCAA6B,EAAC,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"}
@@ -0,0 +1,49 @@
1
+ "use strict";
2
+ Object.defineProperty(exports, "__esModule", { value: true });
3
+ exports.normalizeApifyTranscript = exports.normalizeTranscriptLines = exports.normalizeTranscriptText = void 0;
4
+ const utils_js_1 = require("./utils.js");
5
+ const normalizeTranscriptText = (input) => input
6
+ .replaceAll('\u00A0', ' ')
7
+ .replaceAll(/[\t ]+/g, ' ')
8
+ .replaceAll(/\s*\n\s*/g, '\n')
9
+ .replaceAll(/\n{3,}/g, '\n\n')
10
+ .trim();
11
+ exports.normalizeTranscriptText = normalizeTranscriptText;
12
+ const normalizeTranscriptLines = (lines) => {
13
+ if (lines.length === 0) {
14
+ return null;
15
+ }
16
+ const normalized = (0, exports.normalizeTranscriptText)(lines.join('\n'));
17
+ return normalized.length > 0 ? normalized : null;
18
+ };
19
+ exports.normalizeTranscriptLines = normalizeTranscriptLines;
20
+ const normalizeApifyTranscript = (raw) => {
21
+ if (typeof raw === 'string') {
22
+ const trimmed = (0, exports.normalizeTranscriptText)(raw);
23
+ return trimmed.length > 0 ? trimmed : null;
24
+ }
25
+ if (Array.isArray(raw)) {
26
+ const lines = raw
27
+ .map((entry) => {
28
+ if (!(0, utils_js_1.isRecord)(entry)) {
29
+ return '';
30
+ }
31
+ const textValue = entry.text;
32
+ return typeof textValue === 'string' ? textValue.trim() : '';
33
+ })
34
+ .filter((line) => line.length > 0);
35
+ if (lines.length > 0) {
36
+ return (0, exports.normalizeTranscriptLines)(lines) ?? null;
37
+ }
38
+ }
39
+ if ((0, utils_js_1.isRecord)(raw)) {
40
+ const singleText = raw.text;
41
+ if (typeof singleText === 'string') {
42
+ const trimmed = (0, exports.normalizeTranscriptText)(singleText);
43
+ return trimmed.length > 0 ? trimmed : null;
44
+ }
45
+ }
46
+ return null;
47
+ };
48
+ exports.normalizeApifyTranscript = normalizeApifyTranscript;
49
+ //# sourceMappingURL=normalize.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"normalize.js","sourceRoot":"","sources":["../../../../src/content/transcript/normalize.ts"],"names":[],"mappings":";;;AAAA,yCAAqC;AAE9B,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,CAAA;AANE,QAAA,uBAAuB,2BAMzB;AAEJ,MAAM,wBAAwB,GAAG,CAAC,KAAwB,EAAiB,EAAE;IAClF,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QACvB,OAAO,IAAI,CAAA;IACb,CAAC;IACD,MAAM,UAAU,GAAG,IAAA,+BAAuB,EAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAA;IAC5D,OAAO,UAAU,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,IAAI,CAAA;AAClD,CAAC,CAAA;AANY,QAAA,wBAAwB,4BAMpC;AAEM,MAAM,wBAAwB,GAAG,CAAC,GAAY,EAAiB,EAAE;IACtE,IAAI,OAAO,GAAG,KAAK,QAAQ,EAAE,CAAC;QAC5B,MAAM,OAAO,GAAG,IAAA,+BAAuB,EAAC,GAAG,CAAC,CAAA;QAC5C,OAAO,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAA;IAC5C,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,IAAA,mBAAQ,EAAC,KAAK,CAAC,EAAE,CAAC;gBACrB,OAAO,EAAE,CAAA;YACX,CAAC;YACD,MAAM,SAAS,GAAI,KAAuB,CAAC,IAAI,CAAA;YAC/C,OAAO,OAAO,SAAS,KAAK,QAAQ,CAAC,CAAC,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,CAAA;QAC9D,CAAC,CAAC;aACD,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;QACpC,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACrB,OAAO,IAAA,gCAAwB,EAAC,KAAK,CAAC,IAAI,IAAI,CAAA;QAChD,CAAC;IACH,CAAC;IAED,IAAI,IAAA,mBAAQ,EAAC,GAAG,CAAC,EAAE,CAAC;QAClB,MAAM,UAAU,GAAI,GAA0B,CAAC,IAAI,CAAA;QACnD,IAAI,OAAO,UAAU,KAAK,QAAQ,EAAE,CAAC;YACnC,MAAM,OAAO,GAAG,IAAA,+BAAuB,EAAC,UAAU,CAAC,CAAA;YACnD,OAAO,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAA;QAC5C,CAAC;IACH,CAAC;IAED,OAAO,IAAI,CAAA;AACb,CAAC,CAAA;AA/BY,QAAA,wBAAwB,4BA+BpC"}
@@ -0,0 +1,16 @@
1
+ "use strict";
2
+ Object.defineProperty(exports, "__esModule", { value: true });
3
+ exports.fetchTranscript = exports.canHandle = void 0;
4
+ const canHandle = () => true;
5
+ exports.canHandle = canHandle;
6
+ const fetchTranscript = async (_context, _options) => {
7
+ await Promise.resolve();
8
+ return {
9
+ text: null,
10
+ source: null,
11
+ attemptedProviders: [],
12
+ metadata: { provider: 'generic', reason: 'not_implemented' },
13
+ };
14
+ };
15
+ exports.fetchTranscript = fetchTranscript;
16
+ //# sourceMappingURL=generic.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"generic.js","sourceRoot":"","sources":["../../../../../src/content/transcript/providers/generic.ts"],"names":[],"mappings":";;;AAEO,MAAM,SAAS,GAAG,GAAY,EAAE,CAAC,IAAI,CAAA;AAA/B,QAAA,SAAS,aAAsB;AAErC,MAAM,eAAe,GAAG,KAAK,EAClC,QAAyB,EACzB,QAA8B,EACL,EAAE;IAC3B,MAAM,OAAO,CAAC,OAAO,EAAE,CAAA;IACvB,OAAO;QACL,IAAI,EAAE,IAAI;QACV,MAAM,EAAE,IAAI;QACZ,kBAAkB,EAAE,EAAE;QACtB,QAAQ,EAAE,EAAE,QAAQ,EAAE,SAAS,EAAE,MAAM,EAAE,iBAAiB,EAAE;KAC7D,CAAA;AACH,CAAC,CAAA;AAXY,QAAA,eAAe,mBAW3B"}