@wanshi-kg/wanshi 0.1.0

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.
Files changed (443) hide show
  1. package/LICENSE +21 -0
  2. package/README.md +458 -0
  3. package/dist/__tests__/helpers.js +27 -0
  4. package/dist/__tests__/helpers.js.map +1 -0
  5. package/dist/cli/commands/export.command.js +99 -0
  6. package/dist/cli/commands/export.command.js.map +1 -0
  7. package/dist/cli/commands/index.js +22 -0
  8. package/dist/cli/commands/index.js.map +1 -0
  9. package/dist/cli/commands/inspectMerges.command.js +84 -0
  10. package/dist/cli/commands/inspectMerges.command.js.map +1 -0
  11. package/dist/cli/commands/metrics.command.js +196 -0
  12. package/dist/cli/commands/metrics.command.js.map +1 -0
  13. package/dist/cli/commands/process.command.js +82 -0
  14. package/dist/cli/commands/process.command.js.map +1 -0
  15. package/dist/cli/commands/watch.command.js +91 -0
  16. package/dist/cli/commands/watch.command.js.map +1 -0
  17. package/dist/cli/index.js +269 -0
  18. package/dist/cli/index.js.map +1 -0
  19. package/dist/cli/optionsToConfig.js +160 -0
  20. package/dist/cli/optionsToConfig.js.map +1 -0
  21. package/dist/config/index.js +59 -0
  22. package/dist/config/index.js.map +1 -0
  23. package/dist/config/legacyHints.js +113 -0
  24. package/dist/config/legacyHints.js.map +1 -0
  25. package/dist/config/schema.js +803 -0
  26. package/dist/config/schema.js.map +1 -0
  27. package/dist/config/ui.js +221 -0
  28. package/dist/config/ui.js.map +1 -0
  29. package/dist/core/DirectoryProcessor.js +725 -0
  30. package/dist/core/DirectoryProcessor.js.map +1 -0
  31. package/dist/core/adapters/IStructuredAdapter.js +3 -0
  32. package/dist/core/adapters/IStructuredAdapter.js.map +1 -0
  33. package/dist/core/adapters/SqliteAdapter.js +267 -0
  34. package/dist/core/adapters/SqliteAdapter.js.map +1 -0
  35. package/dist/core/adapters/StructuredAdapterRegistry.js +31 -0
  36. package/dist/core/adapters/StructuredAdapterRegistry.js.map +1 -0
  37. package/dist/core/adapters/index.js +20 -0
  38. package/dist/core/adapters/index.js.map +1 -0
  39. package/dist/core/checkpoint/CheckpointService.js +188 -0
  40. package/dist/core/checkpoint/CheckpointService.js.map +1 -0
  41. package/dist/core/checkpoint/index.js +18 -0
  42. package/dist/core/checkpoint/index.js.map +1 -0
  43. package/dist/core/corpus/CorpusAnalyzer.js +266 -0
  44. package/dist/core/corpus/CorpusAnalyzer.js.map +1 -0
  45. package/dist/core/corpus/CorpusProfileStore.js +92 -0
  46. package/dist/core/corpus/CorpusProfileStore.js.map +1 -0
  47. package/dist/core/corpus/index.js +21 -0
  48. package/dist/core/corpus/index.js.map +1 -0
  49. package/dist/core/corpus/normalizeGlossary.js +60 -0
  50. package/dist/core/corpus/normalizeGlossary.js.map +1 -0
  51. package/dist/core/corpus/relPath.js +52 -0
  52. package/dist/core/corpus/relPath.js.map +1 -0
  53. package/dist/core/corpus/termFrequency.js +86 -0
  54. package/dist/core/corpus/termFrequency.js.map +1 -0
  55. package/dist/core/cost/CostMeter.js +235 -0
  56. package/dist/core/cost/CostMeter.js.map +1 -0
  57. package/dist/core/cost/index.js +19 -0
  58. package/dist/core/cost/index.js.map +1 -0
  59. package/dist/core/cost/prices.js +38 -0
  60. package/dist/core/cost/prices.js.map +1 -0
  61. package/dist/core/cv/ObjectDetectionService.js +119 -0
  62. package/dist/core/cv/ObjectDetectionService.js.map +1 -0
  63. package/dist/core/di/ContainerFactory.js +670 -0
  64. package/dist/core/di/ContainerFactory.js.map +1 -0
  65. package/dist/core/di/DIContainer.js +103 -0
  66. package/dist/core/di/DIContainer.js.map +1 -0
  67. package/dist/core/di/index.js +19 -0
  68. package/dist/core/di/index.js.map +1 -0
  69. package/dist/core/errors/CustomErrors.js +342 -0
  70. package/dist/core/errors/CustomErrors.js.map +1 -0
  71. package/dist/core/errors/index.js +18 -0
  72. package/dist/core/errors/index.js.map +1 -0
  73. package/dist/core/export/KnowledgeGraphExportService.js +56 -0
  74. package/dist/core/export/KnowledgeGraphExportService.js.map +1 -0
  75. package/dist/core/export/index.js +19 -0
  76. package/dist/core/export/index.js.map +1 -0
  77. package/dist/core/export/strategies/GraphitiExportStrategy.js +115 -0
  78. package/dist/core/export/strategies/GraphitiExportStrategy.js.map +1 -0
  79. package/dist/core/export/strategies/GraphvizDotExportStrategy.js +331 -0
  80. package/dist/core/export/strategies/GraphvizDotExportStrategy.js.map +1 -0
  81. package/dist/core/export/strategies/IExportStrategy.js +3 -0
  82. package/dist/core/export/strategies/IExportStrategy.js.map +1 -0
  83. package/dist/core/export/strategies/JsonExportStrategy.js +19 -0
  84. package/dist/core/export/strategies/JsonExportStrategy.js.map +1 -0
  85. package/dist/core/export/strategies/JsonlExportStrategy.js +69 -0
  86. package/dist/core/export/strategies/JsonlExportStrategy.js.map +1 -0
  87. package/dist/core/export/strategies/KblamExportStrategy.js +36 -0
  88. package/dist/core/export/strategies/KblamExportStrategy.js.map +1 -0
  89. package/dist/core/export/strategies/LoraExportStrategy.js +46 -0
  90. package/dist/core/export/strategies/LoraExportStrategy.js.map +1 -0
  91. package/dist/core/export/strategies/McpExportStrategy.js +67 -0
  92. package/dist/core/export/strategies/McpExportStrategy.js.map +1 -0
  93. package/dist/core/export/strategies/index.js +25 -0
  94. package/dist/core/export/strategies/index.js.map +1 -0
  95. package/dist/core/export/strategies/kbTriples.js +60 -0
  96. package/dist/core/export/strategies/kbTriples.js.map +1 -0
  97. package/dist/core/index.js +22 -0
  98. package/dist/core/index.js.map +1 -0
  99. package/dist/core/knowledge/KnowledgeGraphBuilder.js +627 -0
  100. package/dist/core/knowledge/KnowledgeGraphBuilder.js.map +1 -0
  101. package/dist/core/knowledge/MergeRecord.js +3 -0
  102. package/dist/core/knowledge/MergeRecord.js.map +1 -0
  103. package/dist/core/knowledge/canon/Canonicalizer.js +414 -0
  104. package/dist/core/knowledge/canon/Canonicalizer.js.map +1 -0
  105. package/dist/core/knowledge/canon/index.js +18 -0
  106. package/dist/core/knowledge/canon/index.js.map +1 -0
  107. package/dist/core/knowledge/contradiction/HeuristicContradictionChecker.js +92 -0
  108. package/dist/core/knowledge/contradiction/HeuristicContradictionChecker.js.map +1 -0
  109. package/dist/core/knowledge/contradiction/LlmContradictionChecker.js +52 -0
  110. package/dist/core/knowledge/contradiction/LlmContradictionChecker.js.map +1 -0
  111. package/dist/core/knowledge/contradiction/index.js +19 -0
  112. package/dist/core/knowledge/contradiction/index.js.map +1 -0
  113. package/dist/core/knowledge/grounding/KeywordGroundingChecker.js +33 -0
  114. package/dist/core/knowledge/grounding/KeywordGroundingChecker.js.map +1 -0
  115. package/dist/core/knowledge/grounding/MiniCheckGroundingChecker.js +82 -0
  116. package/dist/core/knowledge/grounding/MiniCheckGroundingChecker.js.map +1 -0
  117. package/dist/core/knowledge/grounding/index.js +20 -0
  118. package/dist/core/knowledge/grounding/index.js.map +1 -0
  119. package/dist/core/knowledge/grounding/verbalize.js +38 -0
  120. package/dist/core/knowledge/grounding/verbalize.js.map +1 -0
  121. package/dist/core/knowledge/images/imageMetaGraph.js +136 -0
  122. package/dist/core/knowledge/images/imageMetaGraph.js.map +1 -0
  123. package/dist/core/knowledge/index.js +20 -0
  124. package/dist/core/knowledge/index.js.map +1 -0
  125. package/dist/core/knowledge/merging/KnowledgeMerger.js +624 -0
  126. package/dist/core/knowledge/merging/KnowledgeMerger.js.map +1 -0
  127. package/dist/core/knowledge/references/ReferenceResolver.js +184 -0
  128. package/dist/core/knowledge/references/ReferenceResolver.js.map +1 -0
  129. package/dist/core/knowledge/references/citations/CitationEvidenceProcessor.js +401 -0
  130. package/dist/core/knowledge/references/citations/CitationEvidenceProcessor.js.map +1 -0
  131. package/dist/core/knowledge/references/citations/CitationResolver.js +95 -0
  132. package/dist/core/knowledge/references/citations/CitationResolver.js.map +1 -0
  133. package/dist/core/knowledge/references/citations/GrobidClient.js +143 -0
  134. package/dist/core/knowledge/references/citations/GrobidClient.js.map +1 -0
  135. package/dist/core/knowledge/references/citations/TitleIdResolver.js +101 -0
  136. package/dist/core/knowledge/references/citations/TitleIdResolver.js.map +1 -0
  137. package/dist/core/knowledge/references/web/FetchCacheService.js +114 -0
  138. package/dist/core/knowledge/references/web/FetchCacheService.js.map +1 -0
  139. package/dist/core/knowledge/references/web/GatedFetcher.js +228 -0
  140. package/dist/core/knowledge/references/web/GatedFetcher.js.map +1 -0
  141. package/dist/core/knowledge/references/web/WebReferenceProcessor.js +164 -0
  142. package/dist/core/knowledge/references/web/WebReferenceProcessor.js.map +1 -0
  143. package/dist/core/knowledge/search/KnowledgeGraphSearch.js +261 -0
  144. package/dist/core/knowledge/search/KnowledgeGraphSearch.js.map +1 -0
  145. package/dist/core/knowledge/vocabulary.js +162 -0
  146. package/dist/core/knowledge/vocabulary.js.map +1 -0
  147. package/dist/core/llm/EmbeddingService.js +113 -0
  148. package/dist/core/llm/EmbeddingService.js.map +1 -0
  149. package/dist/core/llm/OllamaService.js +146 -0
  150. package/dist/core/llm/OllamaService.js.map +1 -0
  151. package/dist/core/llm/OpenAICompatibleService.js +190 -0
  152. package/dist/core/llm/OpenAICompatibleService.js.map +1 -0
  153. package/dist/core/llm/OpenAIEmbeddingService.js +129 -0
  154. package/dist/core/llm/OpenAIEmbeddingService.js.map +1 -0
  155. package/dist/core/llm/embeddingUtils.js +25 -0
  156. package/dist/core/llm/embeddingUtils.js.map +1 -0
  157. package/dist/core/llm/index.js +23 -0
  158. package/dist/core/llm/index.js.map +1 -0
  159. package/dist/core/llm/prompts/PromptManager.js +388 -0
  160. package/dist/core/llm/prompts/PromptManager.js.map +1 -0
  161. package/dist/core/llm/prompts/PromptTemplateEngine.js +257 -0
  162. package/dist/core/llm/prompts/PromptTemplateEngine.js.map +1 -0
  163. package/dist/core/llm/prompts/templates/partials/examples/EXAMPLE_STYLE_GUIDE.md +84 -0
  164. package/dist/core/llm/prompts/templates/partials/examples/article.md +187 -0
  165. package/dist/core/llm/prompts/templates/partials/examples/code.md +229 -0
  166. package/dist/core/llm/prompts/templates/partials/examples/communication.md +205 -0
  167. package/dist/core/llm/prompts/templates/partials/examples/documentation.md +262 -0
  168. package/dist/core/llm/prompts/templates/partials/examples/financial.md +157 -0
  169. package/dist/core/llm/prompts/templates/partials/examples/legal.md +153 -0
  170. package/dist/core/llm/prompts/templates/partials/examples/logs.md +127 -0
  171. package/dist/core/llm/prompts/templates/partials/examples/medical.md +218 -0
  172. package/dist/core/llm/prompts/templates/partials/examples/notes.md +201 -0
  173. package/dist/core/llm/prompts/templates/partials/examples/research.md +208 -0
  174. package/dist/core/llm/prompts/templates/partials/examples/tabular.md +178 -0
  175. package/dist/core/llm/prompts/templates/partials/examples/transcript.md +204 -0
  176. package/dist/core/llm/prompts/templates/partials/retrieved-context.hbs +18 -0
  177. package/dist/core/llm/prompts/templates/v1/system.hbs +371 -0
  178. package/dist/core/llm/prompts/templates/v1/user.hbs +20 -0
  179. package/dist/core/llm/prompts/templates/v2/system.hbs +573 -0
  180. package/dist/core/llm/prompts/templates/v2/user.hbs +20 -0
  181. package/dist/core/llm/prompts/templates/v3/system.hbs +861 -0
  182. package/dist/core/llm/prompts/templates/v3/user.hbs +16 -0
  183. package/dist/core/llm/prompts/templates/v4/system.hbs +800 -0
  184. package/dist/core/llm/prompts/templates/v4/user.hbs +40 -0
  185. package/dist/core/llm/prompts/templates/v4.5/system.hbs +71 -0
  186. package/dist/core/llm/prompts/templates/v4.5/user.hbs +46 -0
  187. package/dist/core/llm/prompts/templates/v5/glossary/system.hbs +40 -0
  188. package/dist/core/llm/prompts/templates/v5/glossary/user.hbs +11 -0
  189. package/dist/core/llm/prompts/templates/v5/system.hbs +163 -0
  190. package/dist/core/llm/prompts/templates/v5/user.hbs +55 -0
  191. package/dist/core/pipeline/GroundingTransform.js +52 -0
  192. package/dist/core/pipeline/GroundingTransform.js.map +1 -0
  193. package/dist/core/pipeline/PipelineRunner.js +51 -0
  194. package/dist/core/pipeline/PipelineRunner.js.map +1 -0
  195. package/dist/core/pipeline/RelationFilterTransform.js +72 -0
  196. package/dist/core/pipeline/RelationFilterTransform.js.map +1 -0
  197. package/dist/core/pipeline/index.js +20 -0
  198. package/dist/core/pipeline/index.js.map +1 -0
  199. package/dist/core/processor/FileProcessor.js +184 -0
  200. package/dist/core/processor/FileProcessor.js.map +1 -0
  201. package/dist/core/processor/ProcessedRegistry.js +38 -0
  202. package/dist/core/processor/ProcessedRegistry.js.map +1 -0
  203. package/dist/core/processor/ast/AstSeedService.js +0 -0
  204. package/dist/core/processor/ast/AstSeedService.js.map +1 -0
  205. package/dist/core/processor/ast/AstSymbolStore.js +110 -0
  206. package/dist/core/processor/ast/AstSymbolStore.js.map +1 -0
  207. package/dist/core/processor/ast/index.js +19 -0
  208. package/dist/core/processor/ast/index.js.map +1 -0
  209. package/dist/core/processor/chunking/TextChunker.js +98 -0
  210. package/dist/core/processor/chunking/TextChunker.js.map +1 -0
  211. package/dist/core/processor/chunking/index.js +18 -0
  212. package/dist/core/processor/chunking/index.js.map +1 -0
  213. package/dist/core/processor/classifier/CONTENT_CLASSES.js +294 -0
  214. package/dist/core/processor/classifier/CONTENT_CLASSES.js.map +1 -0
  215. package/dist/core/processor/classifier/CascadeContentClassifier.js +107 -0
  216. package/dist/core/processor/classifier/CascadeContentClassifier.js.map +1 -0
  217. package/dist/core/processor/classifier/HeuristicContentClassifier.js +113 -0
  218. package/dist/core/processor/classifier/HeuristicContentClassifier.js.map +1 -0
  219. package/dist/core/processor/classifier/IContentTypeClassifier.js +3 -0
  220. package/dist/core/processor/classifier/IContentTypeClassifier.js.map +1 -0
  221. package/dist/core/processor/classifier/LlmContentClassifier.js +107 -0
  222. package/dist/core/processor/classifier/LlmContentClassifier.js.map +1 -0
  223. package/dist/core/processor/classifier/NER_DOMAIN_EXAMPLES.js +498 -0
  224. package/dist/core/processor/classifier/NER_DOMAIN_EXAMPLES.js.map +1 -0
  225. package/dist/core/processor/classifier/index.js +21 -0
  226. package/dist/core/processor/classifier/index.js.map +1 -0
  227. package/dist/core/processor/classifier/mergeClassifications.js +32 -0
  228. package/dist/core/processor/classifier/mergeClassifications.js.map +1 -0
  229. package/dist/core/processor/index.js +20 -0
  230. package/dist/core/processor/index.js.map +1 -0
  231. package/dist/core/processor/readers/AudioReader.js +462 -0
  232. package/dist/core/processor/readers/AudioReader.js.map +1 -0
  233. package/dist/core/processor/readers/BinaryReader.js +90 -0
  234. package/dist/core/processor/readers/BinaryReader.js.map +1 -0
  235. package/dist/core/processor/readers/ChandraPdfReader.js +187 -0
  236. package/dist/core/processor/readers/ChandraPdfReader.js.map +1 -0
  237. package/dist/core/processor/readers/ChatExportReader.js +365 -0
  238. package/dist/core/processor/readers/ChatExportReader.js.map +1 -0
  239. package/dist/core/processor/readers/DoclingReader.js +445 -0
  240. package/dist/core/processor/readers/DoclingReader.js.map +1 -0
  241. package/dist/core/processor/readers/EmailReader.js +259 -0
  242. package/dist/core/processor/readers/EmailReader.js.map +1 -0
  243. package/dist/core/processor/readers/EpubReader.js +175 -0
  244. package/dist/core/processor/readers/EpubReader.js.map +1 -0
  245. package/dist/core/processor/readers/FileReader.js +90 -0
  246. package/dist/core/processor/readers/FileReader.js.map +1 -0
  247. package/dist/core/processor/readers/FileReaderFactory.js +49 -0
  248. package/dist/core/processor/readers/FileReaderFactory.js.map +1 -0
  249. package/dist/core/processor/readers/HtmlReader.js +371 -0
  250. package/dist/core/processor/readers/HtmlReader.js.map +1 -0
  251. package/dist/core/processor/readers/ImageReader.js +162 -0
  252. package/dist/core/processor/readers/ImageReader.js.map +1 -0
  253. package/dist/core/processor/readers/JsonFileReader.js +232 -0
  254. package/dist/core/processor/readers/JsonFileReader.js.map +1 -0
  255. package/dist/core/processor/readers/JupyterReader.js +178 -0
  256. package/dist/core/processor/readers/JupyterReader.js.map +1 -0
  257. package/dist/core/processor/readers/LatexReader.js +176 -0
  258. package/dist/core/processor/readers/LatexReader.js.map +1 -0
  259. package/dist/core/processor/readers/MarkdownReader.js +289 -0
  260. package/dist/core/processor/readers/MarkdownReader.js.map +1 -0
  261. package/dist/core/processor/readers/MarkerPdfReader.js +193 -0
  262. package/dist/core/processor/readers/MarkerPdfReader.js.map +1 -0
  263. package/dist/core/processor/readers/MistralOcrReader.js +198 -0
  264. package/dist/core/processor/readers/MistralOcrReader.js.map +1 -0
  265. package/dist/core/processor/readers/OfficeReader.js +174 -0
  266. package/dist/core/processor/readers/OfficeReader.js.map +1 -0
  267. package/dist/core/processor/readers/PdfReader.js +116 -0
  268. package/dist/core/processor/readers/PdfReader.js.map +1 -0
  269. package/dist/core/processor/readers/RtfReader.js +107 -0
  270. package/dist/core/processor/readers/RtfReader.js.map +1 -0
  271. package/dist/core/processor/readers/SubtitleReader.js +145 -0
  272. package/dist/core/processor/readers/SubtitleReader.js.map +1 -0
  273. package/dist/core/processor/readers/TesseractPdfReader.js +183 -0
  274. package/dist/core/processor/readers/TesseractPdfReader.js.map +1 -0
  275. package/dist/core/processor/readers/TextReader.js +129 -0
  276. package/dist/core/processor/readers/TextReader.js.map +1 -0
  277. package/dist/core/processor/readers/TranscriptReader.js +234 -0
  278. package/dist/core/processor/readers/TranscriptReader.js.map +1 -0
  279. package/dist/core/processor/readers/image/imageMetadata.js +155 -0
  280. package/dist/core/processor/readers/image/imageMetadata.js.map +1 -0
  281. package/dist/core/processor/readers/index.js +41 -0
  282. package/dist/core/processor/readers/index.js.map +1 -0
  283. package/dist/core/processor/readers/referenceExtraction.js +198 -0
  284. package/dist/core/processor/readers/referenceExtraction.js.map +1 -0
  285. package/dist/core/processor/readers/stripReferences.js +59 -0
  286. package/dist/core/processor/readers/stripReferences.js.map +1 -0
  287. package/dist/core/processor/readers/transcript/turnPacking.js +81 -0
  288. package/dist/core/processor/readers/transcript/turnPacking.js.map +1 -0
  289. package/dist/core/progress/NdjsonProgressEmitter.js +30 -0
  290. package/dist/core/progress/NdjsonProgressEmitter.js.map +1 -0
  291. package/dist/core/progress/NoopProgressEmitter.js +15 -0
  292. package/dist/core/progress/NoopProgressEmitter.js.map +1 -0
  293. package/dist/core/progress/index.js +19 -0
  294. package/dist/core/progress/index.js.map +1 -0
  295. package/dist/core/trace/TraceWriter.js +100 -0
  296. package/dist/core/trace/TraceWriter.js.map +1 -0
  297. package/dist/core/trace/events.js +13 -0
  298. package/dist/core/trace/events.js.map +1 -0
  299. package/dist/core/trace/index.js +20 -0
  300. package/dist/core/trace/index.js.map +1 -0
  301. package/dist/core/trace/lineage.js +97 -0
  302. package/dist/core/trace/lineage.js.map +1 -0
  303. package/dist/evaluation/BenchmarkRunner.js +171 -0
  304. package/dist/evaluation/BenchmarkRunner.js.map +1 -0
  305. package/dist/evaluation/classifier/ClassifierAccuracy.js +185 -0
  306. package/dist/evaluation/classifier/ClassifierAccuracy.js.map +1 -0
  307. package/dist/evaluation/classifier/labeledSamples.js +379 -0
  308. package/dist/evaluation/classifier/labeledSamples.js.map +1 -0
  309. package/dist/evaluation/compare/goldCompare.js +126 -0
  310. package/dist/evaluation/compare/goldCompare.js.map +1 -0
  311. package/dist/evaluation/crossre/compareScoring.js +30 -0
  312. package/dist/evaluation/crossre/compareScoring.js.map +1 -0
  313. package/dist/evaluation/datasets/CrossREDataset.js +170 -0
  314. package/dist/evaluation/datasets/CrossREDataset.js.map +1 -0
  315. package/dist/evaluation/datasets/IDataset.js +3 -0
  316. package/dist/evaluation/datasets/IDataset.js.map +1 -0
  317. package/dist/evaluation/datasets/RebelDataset.js +117 -0
  318. package/dist/evaluation/datasets/RebelDataset.js.map +1 -0
  319. package/dist/evaluation/datasets/RedocredDataset.js +218 -0
  320. package/dist/evaluation/datasets/RedocredDataset.js.map +1 -0
  321. package/dist/evaluation/datasets/SemEval2010Dataset.js +150 -0
  322. package/dist/evaluation/datasets/SemEval2010Dataset.js.map +1 -0
  323. package/dist/evaluation/index.js +33 -0
  324. package/dist/evaluation/index.js.map +1 -0
  325. package/dist/evaluation/matching/ExactMatcher.js +75 -0
  326. package/dist/evaluation/matching/ExactMatcher.js.map +1 -0
  327. package/dist/evaluation/matching/SemanticMatcher.js +143 -0
  328. package/dist/evaluation/matching/SemanticMatcher.js.map +1 -0
  329. package/dist/evaluation/metrics/TripleMetrics.js +64 -0
  330. package/dist/evaluation/metrics/TripleMetrics.js.map +1 -0
  331. package/dist/evaluation/mine/MineCheckpoint.js +114 -0
  332. package/dist/evaluation/mine/MineCheckpoint.js.map +1 -0
  333. package/dist/evaluation/mine/MineDataset.js +208 -0
  334. package/dist/evaluation/mine/MineDataset.js.map +1 -0
  335. package/dist/evaluation/mine/MineReporter.js +98 -0
  336. package/dist/evaluation/mine/MineReporter.js.map +1 -0
  337. package/dist/evaluation/mine/MineRunner.js +148 -0
  338. package/dist/evaluation/mine/MineRunner.js.map +1 -0
  339. package/dist/evaluation/mine/MineScorer.js +127 -0
  340. package/dist/evaluation/mine/MineScorer.js.map +1 -0
  341. package/dist/evaluation/mine/types.js +12 -0
  342. package/dist/evaluation/mine/types.js.map +1 -0
  343. package/dist/evaluation/reporters/ConsoleReporter.js +55 -0
  344. package/dist/evaluation/reporters/ConsoleReporter.js.map +1 -0
  345. package/dist/evaluation/reporters/JsonReporter.js +50 -0
  346. package/dist/evaluation/reporters/JsonReporter.js.map +1 -0
  347. package/dist/index.js +28 -0
  348. package/dist/index.js.map +1 -0
  349. package/dist/quality/CompositeScore.js +61 -0
  350. package/dist/quality/CompositeScore.js.map +1 -0
  351. package/dist/quality/ConsistencyMetrics.js +70 -0
  352. package/dist/quality/ConsistencyMetrics.js.map +1 -0
  353. package/dist/quality/FactualMetrics.js +76 -0
  354. package/dist/quality/FactualMetrics.js.map +1 -0
  355. package/dist/quality/GraphHealthMetrics.js +68 -0
  356. package/dist/quality/GraphHealthMetrics.js.map +1 -0
  357. package/dist/quality/SemanticMetrics.js +102 -0
  358. package/dist/quality/SemanticMetrics.js.map +1 -0
  359. package/dist/quality/StructuralMetrics.js +60 -0
  360. package/dist/quality/StructuralMetrics.js.map +1 -0
  361. package/dist/quality/index.js +23 -0
  362. package/dist/quality/index.js.map +1 -0
  363. package/dist/shared/index.js +20 -0
  364. package/dist/shared/index.js.map +1 -0
  365. package/dist/shared/logger/Logger.js +3 -0
  366. package/dist/shared/logger/Logger.js.map +1 -0
  367. package/dist/shared/logger/LoggerFactory.js +75 -0
  368. package/dist/shared/logger/LoggerFactory.js.map +1 -0
  369. package/dist/shared/logger/index.js +19 -0
  370. package/dist/shared/logger/index.js.map +1 -0
  371. package/dist/shared/shutdown.js +30 -0
  372. package/dist/shared/shutdown.js.map +1 -0
  373. package/dist/shared/utils/agglomerativeCluster.js +269 -0
  374. package/dist/shared/utils/agglomerativeCluster.js.map +1 -0
  375. package/dist/shared/utils/astSymbols.js +69 -0
  376. package/dist/shared/utils/astSymbols.js.map +1 -0
  377. package/dist/shared/utils/cosineSimilarity.js +18 -0
  378. package/dist/shared/utils/cosineSimilarity.js.map +1 -0
  379. package/dist/shared/utils/directoryTree.js +184 -0
  380. package/dist/shared/utils/directoryTree.js.map +1 -0
  381. package/dist/shared/utils/documentOutline.js +74 -0
  382. package/dist/shared/utils/documentOutline.js.map +1 -0
  383. package/dist/shared/utils/index.js +24 -0
  384. package/dist/shared/utils/index.js.map +1 -0
  385. package/dist/shared/utils/jaroWinklerSimilarity.js +60 -0
  386. package/dist/shared/utils/jaroWinklerSimilarity.js.map +1 -0
  387. package/dist/shared/utils/parseJsonLenient.js +27 -0
  388. package/dist/shared/utils/parseJsonLenient.js.map +1 -0
  389. package/dist/shared/utils/readConfig.js +42 -0
  390. package/dist/shared/utils/readConfig.js.map +1 -0
  391. package/dist/shared/utils/readRtf.js +216 -0
  392. package/dist/shared/utils/readRtf.js.map +1 -0
  393. package/dist/shared/utils/softmax.js +26 -0
  394. package/dist/shared/utils/softmax.js.map +1 -0
  395. package/dist/types/ContentClass.js +3 -0
  396. package/dist/types/ContentClass.js.map +1 -0
  397. package/dist/types/CorpusProfile.js +3 -0
  398. package/dist/types/CorpusProfile.js.map +1 -0
  399. package/dist/types/IContradictionChecker.js +3 -0
  400. package/dist/types/IContradictionChecker.js.map +1 -0
  401. package/dist/types/ICorpusAnalyzer.js +3 -0
  402. package/dist/types/ICorpusAnalyzer.js.map +1 -0
  403. package/dist/types/IDirectoryProcessor.js +3 -0
  404. package/dist/types/IDirectoryProcessor.js.map +1 -0
  405. package/dist/types/IEmbeddingProvider.js +3 -0
  406. package/dist/types/IEmbeddingProvider.js.map +1 -0
  407. package/dist/types/IEmbeddingService.js +6 -0
  408. package/dist/types/IEmbeddingService.js.map +1 -0
  409. package/dist/types/IFileProcessor.js +3 -0
  410. package/dist/types/IFileProcessor.js.map +1 -0
  411. package/dist/types/IGroundingChecker.js +3 -0
  412. package/dist/types/IGroundingChecker.js.map +1 -0
  413. package/dist/types/IKnowledgeGraphBuilder.js +3 -0
  414. package/dist/types/IKnowledgeGraphBuilder.js.map +1 -0
  415. package/dist/types/IKnowledgeGraphExporter.js +3 -0
  416. package/dist/types/IKnowledgeGraphExporter.js.map +1 -0
  417. package/dist/types/IKnowledgeGraphMerger.js +3 -0
  418. package/dist/types/IKnowledgeGraphMerger.js.map +1 -0
  419. package/dist/types/IKnowledgeGraphSearch.js +3 -0
  420. package/dist/types/IKnowledgeGraphSearch.js.map +1 -0
  421. package/dist/types/ILLMProvider.js +3 -0
  422. package/dist/types/ILLMProvider.js.map +1 -0
  423. package/dist/types/ILLMService.js +3 -0
  424. package/dist/types/ILLMService.js.map +1 -0
  425. package/dist/types/IObjectDetector.js +3 -0
  426. package/dist/types/IObjectDetector.js.map +1 -0
  427. package/dist/types/IProcessingService.js +3 -0
  428. package/dist/types/IProcessingService.js.map +1 -0
  429. package/dist/types/IProgressEmitter.js +3 -0
  430. package/dist/types/IProgressEmitter.js.map +1 -0
  431. package/dist/types/IPromptManager.js +3 -0
  432. package/dist/types/IPromptManager.js.map +1 -0
  433. package/dist/types/KnowledgeGraph.js +3 -0
  434. package/dist/types/KnowledgeGraph.js.map +1 -0
  435. package/dist/types/MCPKnowledgeGraph.js +3 -0
  436. package/dist/types/MCPKnowledgeGraph.js.map +1 -0
  437. package/dist/types/Observation.js +21 -0
  438. package/dist/types/Observation.js.map +1 -0
  439. package/dist/types/ProcessingOptions.js +3 -0
  440. package/dist/types/ProcessingOptions.js.map +1 -0
  441. package/dist/types/index.js +40 -0
  442. package/dist/types/index.js.map +1 -0
  443. package/package.json +122 -0
@@ -0,0 +1,234 @@
1
+ "use strict";
2
+ var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
3
+ if (k2 === undefined) k2 = k;
4
+ var desc = Object.getOwnPropertyDescriptor(m, k);
5
+ if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
6
+ desc = { enumerable: true, get: function() { return m[k]; } };
7
+ }
8
+ Object.defineProperty(o, k2, desc);
9
+ }) : (function(o, m, k, k2) {
10
+ if (k2 === undefined) k2 = k;
11
+ o[k2] = m[k];
12
+ }));
13
+ var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
14
+ Object.defineProperty(o, "default", { enumerable: true, value: v });
15
+ }) : function(o, v) {
16
+ o["default"] = v;
17
+ });
18
+ var __importStar = (this && this.__importStar) || (function () {
19
+ var ownKeys = function(o) {
20
+ ownKeys = Object.getOwnPropertyNames || function (o) {
21
+ var ar = [];
22
+ for (var k in o) if (Object.prototype.hasOwnProperty.call(o, k)) ar[ar.length] = k;
23
+ return ar;
24
+ };
25
+ return ownKeys(o);
26
+ };
27
+ return function (mod) {
28
+ if (mod && mod.__esModule) return mod;
29
+ var result = {};
30
+ if (mod != null) for (var k = ownKeys(mod), i = 0; i < k.length; i++) if (k[i] !== "default") __createBinding(result, mod, k[i]);
31
+ __setModuleDefault(result, mod);
32
+ return result;
33
+ };
34
+ })();
35
+ var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, generator) {
36
+ function adopt(value) { return value instanceof P ? value : new P(function (resolve) { resolve(value); }); }
37
+ return new (P || (P = Promise))(function (resolve, reject) {
38
+ function fulfilled(value) { try { step(generator.next(value)); } catch (e) { reject(e); } }
39
+ function rejected(value) { try { step(generator["throw"](value)); } catch (e) { reject(e); } }
40
+ function step(result) { result.done ? resolve(result.value) : adopt(result.value).then(fulfilled, rejected); }
41
+ step((generator = generator.apply(thisArg, _arguments || [])).next());
42
+ });
43
+ };
44
+ Object.defineProperty(exports, "__esModule", { value: true });
45
+ exports.TranscriptReader = void 0;
46
+ const fs = __importStar(require("fs"));
47
+ const FileReader_1 = require("./FileReader");
48
+ const turnPacking_1 = require("./transcript/turnPacking");
49
+ /** Speaker-labeled plain-text transcripts (recua `.parakeet.txt`, etc.). */
50
+ const TEXT_SUFFIXES = [
51
+ ".parakeet.txt",
52
+ ".whisper.txt",
53
+ ".corrected.txt",
54
+ ".transcript.txt",
55
+ ];
56
+ /**
57
+ * Reads conversational transcripts into **size-packed** chunks (capped at
58
+ * `maxChunkSize`, tied to the global `chunkSize`) so a long dialogue becomes a
59
+ * handful of chunks instead of one-per-turn. Each turn is rendered inline as
60
+ * `speaker: text` so attribution stays visible to the model, and chunk
61
+ * provenance carries the speaker whenever a chunk happens to be single-speaker.
62
+ *
63
+ * Handles three real shapes:
64
+ * - recua speaker-labeled text (`SPEAKER_XX: …` blocks)
65
+ * - recua turns JSON (`[{ start, end, speaker, <backend>: text }]`)
66
+ * - Claude/ChatGPT chat export (`[{ chat_messages: [{ sender, created_at, … }] }]`)
67
+ *
68
+ * Registered before JsonFileReader/TextReader; `canRead` claims only files that
69
+ * sniff as transcripts, deferring everything else.
70
+ */
71
+ class TranscriptReader extends FileReader_1.FileReader {
72
+ constructor(chunker, logger, maxChunkSize = 4000) {
73
+ super([], chunker, logger); // extension list unused; canRead is overridden
74
+ this.maxChunkSize = maxChunkSize;
75
+ }
76
+ getName() {
77
+ return "TranscriptReader";
78
+ }
79
+ adapterId() {
80
+ return "transcript";
81
+ }
82
+ canRead(filePath) {
83
+ const lower = filePath.toLowerCase();
84
+ if (TEXT_SUFFIXES.some((s) => lower.endsWith(s)))
85
+ return true;
86
+ if (lower.endsWith(".json"))
87
+ return this.sniffJsonTranscript(filePath);
88
+ return false;
89
+ }
90
+ /** Cheap content sniff (first 8 KB) to claim only transcript-shaped JSON. */
91
+ sniffJsonTranscript(filePath) {
92
+ try {
93
+ const fd = fs.openSync(filePath, "r");
94
+ const buf = Buffer.alloc(8192);
95
+ const n = fs.readSync(fd, buf, 0, 8192, 0);
96
+ fs.closeSync(fd);
97
+ const head = buf.toString("utf8", 0, n);
98
+ // Claude/ChatGPT export
99
+ if (head.includes('"chat_messages"') && head.includes('"sender"'))
100
+ return true;
101
+ // recua turns array
102
+ if (/"speaker"\s*:/.test(head) &&
103
+ /"start"\s*:/.test(head) &&
104
+ /"end"\s*:/.test(head))
105
+ return true;
106
+ return false;
107
+ }
108
+ catch (_a) {
109
+ return false;
110
+ }
111
+ }
112
+ read(filePath) {
113
+ return __awaiter(this, void 0, void 0, function* () {
114
+ const raw = yield fs.promises.readFile(filePath, "utf-8");
115
+ let turns;
116
+ try {
117
+ turns = this.parse(filePath, raw);
118
+ }
119
+ catch (e) {
120
+ this.logger.warn(`TranscriptReader could not parse ${filePath}; falling back to plain chunking: ${e}`);
121
+ return this.plainFallback(raw);
122
+ }
123
+ if (turns.length === 0)
124
+ return this.plainFallback(raw);
125
+ const chunks = yield (0, turnPacking_1.packTurns)(turns, filePath, this.maxChunkSize, this.chunker);
126
+ return {
127
+ chunks,
128
+ metadata: { type: "transcript", source: filePath, turns: turns.length },
129
+ };
130
+ });
131
+ }
132
+ parse(filePath, raw) {
133
+ var _a, _b;
134
+ if (filePath.toLowerCase().endsWith(".json")) {
135
+ const data = JSON.parse(raw);
136
+ if (Array.isArray(data) && ((_a = data[0]) === null || _a === void 0 ? void 0 : _a.chat_messages) !== undefined) {
137
+ return this.parseChatExport(data);
138
+ }
139
+ if (Array.isArray(data) && ((_b = data[0]) === null || _b === void 0 ? void 0 : _b.speaker) !== undefined) {
140
+ return this.parseRecuaTurns(data);
141
+ }
142
+ throw new Error("unrecognized transcript JSON shape");
143
+ }
144
+ return this.parseSpeakerText(raw);
145
+ }
146
+ /** "SPEAKER_XX: text" blocks separated by blank lines. */
147
+ parseSpeakerText(raw) {
148
+ const labelRe = /^([A-Za-z0-9_][\w .\-]{0,40}?):\s+/;
149
+ const blocks = raw
150
+ .split(/\n\s*\n/)
151
+ .map((b) => b.trim())
152
+ .filter(Boolean);
153
+ const turns = [];
154
+ for (const block of blocks) {
155
+ const m = block.match(labelRe);
156
+ if (m) {
157
+ turns.push({ speaker: m[1].trim(), text: block.slice(m[0].length).trim() });
158
+ }
159
+ else if (turns.length > 0) {
160
+ turns[turns.length - 1].text += " " + block; // continuation
161
+ }
162
+ else {
163
+ turns.push({ speaker: "UNKNOWN", text: block });
164
+ }
165
+ }
166
+ return turns;
167
+ }
168
+ parseRecuaTurns(data) {
169
+ return data
170
+ .map((t) => {
171
+ var _a;
172
+ return ({
173
+ speaker: String((_a = t.speaker) !== null && _a !== void 0 ? _a : "UNKNOWN"),
174
+ text: this.pickRecuaText(t),
175
+ });
176
+ })
177
+ .filter((t) => t.text);
178
+ }
179
+ pickRecuaText(t) {
180
+ for (const k of ["corrected", "parakeet", "whisper"]) {
181
+ if (typeof t[k] === "string" && t[k].trim())
182
+ return t[k].trim();
183
+ }
184
+ for (const [k, v] of Object.entries(t)) {
185
+ if (k === "start" || k === "end" || k === "speaker")
186
+ continue;
187
+ if (typeof v === "string" && v.trim())
188
+ return v.trim();
189
+ }
190
+ return "";
191
+ }
192
+ parseChatExport(convs) {
193
+ const turns = [];
194
+ convs.forEach((conv, convIndex) => {
195
+ var _a, _b;
196
+ for (const msg of (_a = conv.chat_messages) !== null && _a !== void 0 ? _a : []) {
197
+ const text = this.chatMessageText(msg);
198
+ if (!text)
199
+ continue;
200
+ turns.push({
201
+ speaker: String((_b = msg.sender) !== null && _b !== void 0 ? _b : "unknown"),
202
+ text,
203
+ occurredAt: typeof msg.created_at === "string" ? msg.created_at : undefined,
204
+ // Tag each turn with its conversation so chunkTurns never packs two
205
+ // conversations together (cross-conversation fact bleed, KG-10).
206
+ conversation: convIndex,
207
+ });
208
+ }
209
+ });
210
+ return turns;
211
+ }
212
+ chatMessageText(msg) {
213
+ if (typeof msg.text === "string" && msg.text.trim())
214
+ return msg.text.trim();
215
+ if (Array.isArray(msg.content)) {
216
+ return msg.content
217
+ .map((c) => (typeof (c === null || c === void 0 ? void 0 : c.text) === "string" ? c.text : ""))
218
+ .join(" ")
219
+ .trim();
220
+ }
221
+ return "";
222
+ }
223
+ plainFallback(raw) {
224
+ return __awaiter(this, void 0, void 0, function* () {
225
+ const parts = yield this.chunker.chunk(raw);
226
+ return {
227
+ chunks: parts.map((p) => (Object.assign({}, p))),
228
+ metadata: { type: "transcript-fallback" },
229
+ };
230
+ });
231
+ }
232
+ }
233
+ exports.TranscriptReader = TranscriptReader;
234
+ //# sourceMappingURL=TranscriptReader.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"TranscriptReader.js","sourceRoot":"","sources":["../../../../src/core/processor/readers/TranscriptReader.ts"],"names":[],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAAA,uCAAyB;AACzB,6CAA0D;AAG1D,0DAA2D;AAE3D,4EAA4E;AAC5E,MAAM,aAAa,GAAG;IACpB,eAAe;IACf,cAAc;IACd,gBAAgB;IAChB,iBAAiB;CAClB,CAAC;AAEF;;;;;;;;;;;;;;GAcG;AACH,MAAa,gBAAiB,SAAQ,uBAAU;IAG9C,YAAY,OAAoB,EAAE,MAAc,EAAE,YAAY,GAAG,IAAI;QACnE,KAAK,CAAC,EAAE,EAAE,OAAO,EAAE,MAAM,CAAC,CAAC,CAAC,+CAA+C;QAC3E,IAAI,CAAC,YAAY,GAAG,YAAY,CAAC;IACnC,CAAC;IAED,OAAO;QACL,OAAO,kBAAkB,CAAC;IAC5B,CAAC;IAED,SAAS;QACP,OAAO,YAAY,CAAC;IACtB,CAAC;IAED,OAAO,CAAC,QAAgB;QACtB,MAAM,KAAK,GAAG,QAAQ,CAAC,WAAW,EAAE,CAAC;QACrC,IAAI,aAAa,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC;YAAE,OAAO,IAAI,CAAC;QAC9D,IAAI,KAAK,CAAC,QAAQ,CAAC,OAAO,CAAC;YAAE,OAAO,IAAI,CAAC,mBAAmB,CAAC,QAAQ,CAAC,CAAC;QACvE,OAAO,KAAK,CAAC;IACf,CAAC;IAED,6EAA6E;IACrE,mBAAmB,CAAC,QAAgB;QAC1C,IAAI,CAAC;YACH,MAAM,EAAE,GAAG,EAAE,CAAC,QAAQ,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC;YACtC,MAAM,GAAG,GAAG,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;YAC/B,MAAM,CAAC,GAAG,EAAE,CAAC,QAAQ,CAAC,EAAE,EAAE,GAAG,EAAE,CAAC,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC;YAC3C,EAAE,CAAC,SAAS,CAAC,EAAE,CAAC,CAAC;YACjB,MAAM,IAAI,GAAG,GAAG,CAAC,QAAQ,CAAC,MAAM,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC;YACxC,wBAAwB;YACxB,IAAI,IAAI,CAAC,QAAQ,CAAC,iBAAiB,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,UAAU,CAAC;gBAAE,OAAO,IAAI,CAAC;YAC/E,oBAAoB;YACpB,IACE,eAAe,CAAC,IAAI,CAAC,IAAI,CAAC;gBAC1B,aAAa,CAAC,IAAI,CAAC,IAAI,CAAC;gBACxB,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC;gBAEtB,OAAO,IAAI,CAAC;YACd,OAAO,KAAK,CAAC;QACf,CAAC;QAAC,WAAM,CAAC;YACP,OAAO,KAAK,CAAC;QACf,CAAC;IACH,CAAC;IAEK,IAAI,CAAC,QAAgB;;YACzB,MAAM,GAAG,GAAG,MAAM,EAAE,CAAC,QAAQ,CAAC,QAAQ,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC;YAC1D,IAAI,KAAa,CAAC;YAClB,IAAI,CAAC;gBACH,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC;YACpC,CAAC;YAAC,OAAO,CAAC,EAAE,CAAC;gBACX,IAAI,CAAC,MAAM,CAAC,IAAI,CACd,oCAAoC,QAAQ,qCAAqC,CAAC,EAAE,CACrF,CAAC;gBACF,OAAO,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC,CAAC;YACjC,CAAC;YACD,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;gBAAE,OAAO,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC,CAAC;YACvD,MAAM,MAAM,GAAG,MAAM,IAAA,uBAAS,EAAC,KAAK,EAAE,QAAQ,EAAE,IAAI,CAAC,YAAY,EAAE,IAAI,CAAC,OAAO,CAAC,CAAC;YACjF,OAAO;gBACL,MAAM;gBACN,QAAQ,EAAE,EAAE,IAAI,EAAE,YAAY,EAAE,MAAM,EAAE,QAAQ,EAAE,KAAK,EAAE,KAAK,CAAC,MAAM,EAAE;aACxE,CAAC;QACJ,CAAC;KAAA;IAEO,KAAK,CAAC,QAAgB,EAAE,GAAW;;QACzC,IAAI,QAAQ,CAAC,WAAW,EAAE,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;YAC7C,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;YAC7B,IAAI,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAA,MAAA,IAAI,CAAC,CAAC,CAAC,0CAAE,aAAa,MAAK,SAAS,EAAE,CAAC;gBAChE,OAAO,IAAI,CAAC,eAAe,CAAC,IAAI,CAAC,CAAC;YACpC,CAAC;YACD,IAAI,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAA,MAAA,IAAI,CAAC,CAAC,CAAC,0CAAE,OAAO,MAAK,SAAS,EAAE,CAAC;gBAC1D,OAAO,IAAI,CAAC,eAAe,CAAC,IAAI,CAAC,CAAC;YACpC,CAAC;YACD,MAAM,IAAI,KAAK,CAAC,oCAAoC,CAAC,CAAC;QACxD,CAAC;QACD,OAAO,IAAI,CAAC,gBAAgB,CAAC,GAAG,CAAC,CAAC;IACpC,CAAC;IAED,0DAA0D;IAClD,gBAAgB,CAAC,GAAW;QAClC,MAAM,OAAO,GAAG,oCAAoC,CAAC;QACrD,MAAM,MAAM,GAAG,GAAG;aACf,KAAK,CAAC,SAAS,CAAC;aAChB,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;aACpB,MAAM,CAAC,OAAO,CAAC,CAAC;QACnB,MAAM,KAAK,GAAW,EAAE,CAAC;QACzB,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;YAC3B,MAAM,CAAC,GAAG,KAAK,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC;YAC/B,IAAI,CAAC,EAAE,CAAC;gBACN,KAAK,CAAC,IAAI,CAAC,EAAE,OAAO,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,EAAE,IAAI,EAAE,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC;YAC9E,CAAC;iBAAM,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAC5B,KAAK,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,IAAI,IAAI,GAAG,GAAG,KAAK,CAAC,CAAC,eAAe;YAC9D,CAAC;iBAAM,CAAC;gBACN,KAAK,CAAC,IAAI,CAAC,EAAE,OAAO,EAAE,SAAS,EAAE,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC;YAClD,CAAC;QACH,CAAC;QACD,OAAO,KAAK,CAAC;IACf,CAAC;IAEO,eAAe,CAAC,IAAW;QACjC,OAAO,IAAI;aACR,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE;;YAAC,OAAA,CAAC;gBACX,OAAO,EAAE,MAAM,CAAC,MAAA,CAAC,CAAC,OAAO,mCAAI,SAAS,CAAC;gBACvC,IAAI,EAAE,IAAI,CAAC,aAAa,CAAC,CAAC,CAAC;aAC5B,CAAC,CAAA;SAAA,CAAC;aACF,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;IAC3B,CAAC;IAEO,aAAa,CAAC,CAAM;QAC1B,KAAK,MAAM,CAAC,IAAI,CAAC,WAAW,EAAE,UAAU,EAAE,SAAS,CAAC,EAAE,CAAC;YACrD,IAAI,OAAO,CAAC,CAAC,CAAC,CAAC,KAAK,QAAQ,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE;gBAAE,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;QAClE,CAAC;QACD,KAAK,MAAM,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC;YACvC,IAAI,CAAC,KAAK,OAAO,IAAI,CAAC,KAAK,KAAK,IAAI,CAAC,KAAK,SAAS;gBAAE,SAAS;YAC9D,IAAI,OAAO,CAAC,KAAK,QAAQ,IAAI,CAAC,CAAC,IAAI,EAAE;gBAAE,OAAO,CAAC,CAAC,IAAI,EAAE,CAAC;QACzD,CAAC;QACD,OAAO,EAAE,CAAC;IACZ,CAAC;IAEO,eAAe,CAAC,KAAY;QAClC,MAAM,KAAK,GAAW,EAAE,CAAC;QACzB,KAAK,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,SAAS,EAAE,EAAE;;YAChC,KAAK,MAAM,GAAG,IAAI,MAAA,IAAI,CAAC,aAAa,mCAAI,EAAE,EAAE,CAAC;gBAC3C,MAAM,IAAI,GAAG,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC,CAAC;gBACvC,IAAI,CAAC,IAAI;oBAAE,SAAS;gBACpB,KAAK,CAAC,IAAI,CAAC;oBACT,OAAO,EAAE,MAAM,CAAC,MAAA,GAAG,CAAC,MAAM,mCAAI,SAAS,CAAC;oBACxC,IAAI;oBACJ,UAAU,EAAE,OAAO,GAAG,CAAC,UAAU,KAAK,QAAQ,CAAC,CAAC,CAAC,GAAG,CAAC,UAAU,CAAC,CAAC,CAAC,SAAS;oBAC3E,oEAAoE;oBACpE,iEAAiE;oBACjE,YAAY,EAAE,SAAS;iBACxB,CAAC,CAAC;YACL,CAAC;QACH,CAAC,CAAC,CAAC;QACH,OAAO,KAAK,CAAC;IACf,CAAC;IAEO,eAAe,CAAC,GAAQ;QAC9B,IAAI,OAAO,GAAG,CAAC,IAAI,KAAK,QAAQ,IAAI,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE;YAAE,OAAO,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;QAC5E,IAAI,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,CAAC;YAC/B,OAAO,GAAG,CAAC,OAAO;iBACf,GAAG,CAAC,CAAC,CAAM,EAAE,EAAE,CAAC,CAAC,OAAO,CAAA,CAAC,aAAD,CAAC,uBAAD,CAAC,CAAE,IAAI,CAAA,KAAK,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;iBAC5D,IAAI,CAAC,GAAG,CAAC;iBACT,IAAI,EAAE,CAAC;QACZ,CAAC;QACD,OAAO,EAAE,CAAC;IACZ,CAAC;IAEa,aAAa,CAAC,GAAW;;YACrC,MAAM,KAAK,GAAG,MAAM,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;YAC5C,OAAO;gBACL,MAAM,EAAE,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,mBAAM,CAAC,EAAG,CAAC;gBACpC,QAAQ,EAAE,EAAE,IAAI,EAAE,qBAAqB,EAAE;aAC1C,CAAC;QACJ,CAAC;KAAA;CACF;AA7JD,4CA6JC"}
@@ -0,0 +1,155 @@
1
+ "use strict";
2
+ var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
3
+ if (k2 === undefined) k2 = k;
4
+ var desc = Object.getOwnPropertyDescriptor(m, k);
5
+ if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
6
+ desc = { enumerable: true, get: function() { return m[k]; } };
7
+ }
8
+ Object.defineProperty(o, k2, desc);
9
+ }) : (function(o, m, k, k2) {
10
+ if (k2 === undefined) k2 = k;
11
+ o[k2] = m[k];
12
+ }));
13
+ var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
14
+ Object.defineProperty(o, "default", { enumerable: true, value: v });
15
+ }) : function(o, v) {
16
+ o["default"] = v;
17
+ });
18
+ var __importStar = (this && this.__importStar) || (function () {
19
+ var ownKeys = function(o) {
20
+ ownKeys = Object.getOwnPropertyNames || function (o) {
21
+ var ar = [];
22
+ for (var k in o) if (Object.prototype.hasOwnProperty.call(o, k)) ar[ar.length] = k;
23
+ return ar;
24
+ };
25
+ return ownKeys(o);
26
+ };
27
+ return function (mod) {
28
+ if (mod && mod.__esModule) return mod;
29
+ var result = {};
30
+ if (mod != null) for (var k = ownKeys(mod), i = 0; i < k.length; i++) if (k[i] !== "default") __createBinding(result, mod, k[i]);
31
+ __setModuleDefault(result, mod);
32
+ return result;
33
+ };
34
+ })();
35
+ var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, generator) {
36
+ function adopt(value) { return value instanceof P ? value : new P(function (resolve) { resolve(value); }); }
37
+ return new (P || (P = Promise))(function (resolve, reject) {
38
+ function fulfilled(value) { try { step(generator.next(value)); } catch (e) { reject(e); } }
39
+ function rejected(value) { try { step(generator["throw"](value)); } catch (e) { reject(e); } }
40
+ function step(result) { result.done ? resolve(result.value) : adopt(result.value).then(fulfilled, rejected); }
41
+ step((generator = generator.apply(thisArg, _arguments || [])).next());
42
+ });
43
+ };
44
+ Object.defineProperty(exports, "__esModule", { value: true });
45
+ exports.readExif = readExif;
46
+ exports.readC2pa = readC2pa;
47
+ const child_process_1 = require("child_process");
48
+ /**
49
+ * Read EXIF metadata from an image via `exifr` (pure-JS). Returns `undefined`
50
+ * when the image has no usable EXIF (e.g. a PNG screenshot) or on any parse
51
+ * error — extraction is an enhancement, never a failure mode. Lazy-imports
52
+ * `exifr` so a run that never touches an image pays nothing.
53
+ */
54
+ function readExif(filePath, logger) {
55
+ return __awaiter(this, void 0, void 0, function* () {
56
+ var _a, _b, _c, _d;
57
+ try {
58
+ const exifr = yield Promise.resolve().then(() => __importStar(require("exifr")));
59
+ const parse = (_a = exifr.parse) !== null && _a !== void 0 ? _a : (_b = exifr.default) === null || _b === void 0 ? void 0 : _b.parse;
60
+ // Block selectors: tiff (IFD0: Make/Model/Artist/Software) + exif (DateTimeOriginal)
61
+ // + gps (so exifr computes decimal latitude/longitude).
62
+ const o = yield parse(filePath, { tiff: true, exif: true, gps: true });
63
+ if (!o)
64
+ return undefined;
65
+ const meta = {};
66
+ if (typeof o.latitude === "number" && typeof o.longitude === "number") {
67
+ meta.gps = { lat: o.latitude, lng: o.longitude };
68
+ }
69
+ const dt = (_d = (_c = o.DateTimeOriginal) !== null && _c !== void 0 ? _c : o.CreateDate) !== null && _d !== void 0 ? _d : o.ModifyDate;
70
+ if (dt)
71
+ meta.dateTaken = dt instanceof Date ? dt.toISOString() : String(dt);
72
+ const make = typeof o.Make === "string" ? o.Make.trim() : undefined;
73
+ const model = typeof o.Model === "string" ? o.Model.trim() : undefined;
74
+ if (make || model)
75
+ meta.camera = Object.assign(Object.assign({}, (make ? { make } : {})), (model ? { model } : {}));
76
+ if (typeof o.Artist === "string" && o.Artist.trim())
77
+ meta.author = o.Artist.trim();
78
+ if (typeof o.Software === "string" && o.Software.trim())
79
+ meta.software = o.Software.trim();
80
+ return Object.keys(meta).length ? meta : undefined;
81
+ }
82
+ catch (e) {
83
+ logger === null || logger === void 0 ? void 0 : logger.debug(`EXIF read failed for ${filePath}: ${e}`);
84
+ return undefined;
85
+ }
86
+ });
87
+ }
88
+ /**
89
+ * Read C2PA Content Credentials by shelling the official Adobe/CAI `c2patool`
90
+ * (reference-grade cryptographic validation; degrade-if-absent like marker). Never
91
+ * throws — a missing binary / parse error yields `{present:false, unavailable:true}`.
92
+ */
93
+ function readC2pa(filePath_1) {
94
+ return __awaiter(this, arguments, void 0, function* (filePath, command = "c2patool", logger) {
95
+ try {
96
+ const { code, stdout, stderr } = yield runCommand(command, [filePath], 30000);
97
+ if (code !== 0) {
98
+ // c2patool exits non-zero with "No claim found" when the asset has no
99
+ // credential — a real fact (absent), distinct from the tool being missing.
100
+ if (/no (claim|manifest|provenance|c2pa)/i.test(`${stderr}\n${stdout}`))
101
+ return { present: false };
102
+ logger === null || logger === void 0 ? void 0 : logger.debug(`c2patool exited ${code} for ${filePath}: ${stderr.trim().slice(-200)}`);
103
+ return { present: false, unavailable: true };
104
+ }
105
+ return parseC2pa(JSON.parse(stdout));
106
+ }
107
+ catch (e) {
108
+ // ENOENT (binary not installed), timeout, or unparseable output → we could not
109
+ // read provenance. Distinct from "no credential present".
110
+ logger === null || logger === void 0 ? void 0 : logger.debug(`C2PA read unavailable for ${filePath}: ${e}`);
111
+ return { present: false, unavailable: true };
112
+ }
113
+ });
114
+ }
115
+ /** Map a c2patool manifest-store JSON report to the trust facts we record. */
116
+ function parseC2pa(json) {
117
+ var _a, _b;
118
+ const manifests = (_a = json === null || json === void 0 ? void 0 : json.manifests) !== null && _a !== void 0 ? _a : {};
119
+ const active = json === null || json === void 0 ? void 0 : json.active_manifest;
120
+ const present = !!active || Object.keys(manifests).length > 0;
121
+ if (!present)
122
+ return { present: false };
123
+ const m = (active && manifests[active]) || Object.values(manifests)[0] || {};
124
+ const sigInfo = (_b = m === null || m === void 0 ? void 0 : m.signature_info) !== null && _b !== void 0 ? _b : {};
125
+ const signer = typeof sigInfo.issuer === "string" ? sigInfo.issuer : typeof sigInfo.common_name === "string" ? sigInfo.common_name : undefined;
126
+ const vstatus = Array.isArray(json === null || json === void 0 ? void 0 : json.validation_status) ? json.validation_status : [];
127
+ const valid = vstatus.every((v) => { var _a; return !/error|invalid|fail/i.test(String((_a = v === null || v === void 0 ? void 0 : v.code) !== null && _a !== void 0 ? _a : "")); });
128
+ // C2PA's AI-generation indicator (digitalSourceType …trainedAlgorithmicMedia).
129
+ const aiGenerated = /trainedAlgorithmicMedia/i.test(JSON.stringify(json));
130
+ return { present: true, valid, signer, aiGenerated };
131
+ }
132
+ /** Spawn a CLI with captured output + timeout; rejects on launch error (ENOENT). */
133
+ function runCommand(command, args, timeoutMs) {
134
+ return new Promise((resolve, reject) => {
135
+ var _a, _b;
136
+ const child = (0, child_process_1.spawn)(command, args, { stdio: ["ignore", "pipe", "pipe"] });
137
+ let stdout = "";
138
+ let stderr = "";
139
+ (_a = child.stdout) === null || _a === void 0 ? void 0 : _a.on("data", (d) => (stdout += d.toString()));
140
+ (_b = child.stderr) === null || _b === void 0 ? void 0 : _b.on("data", (d) => (stderr += d.toString()));
141
+ const timer = setTimeout(() => {
142
+ child.kill("SIGTERM");
143
+ reject(new Error(`${command} timed out after ${timeoutMs}ms`));
144
+ }, timeoutMs);
145
+ child.on("error", (e) => {
146
+ clearTimeout(timer);
147
+ reject(e);
148
+ });
149
+ child.on("close", (code) => {
150
+ clearTimeout(timer);
151
+ resolve({ code: code !== null && code !== void 0 ? code : 0, stdout, stderr });
152
+ });
153
+ });
154
+ }
155
+ //# sourceMappingURL=imageMetadata.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"imageMetadata.js","sourceRoot":"","sources":["../../../../../src/core/processor/readers/image/imageMetadata.ts"],"names":[],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AA6BA,4BA0BC;AAsBD,4BAqBC;AAlGD,iDAAsC;AAuBtC;;;;;GAKG;AACH,SAAsB,QAAQ,CAAC,QAAgB,EAAE,MAAe;;;QAC9D,IAAI,CAAC;YACH,MAAM,KAAK,GAAQ,wDAAa,OAAO,GAAC,CAAC;YACzC,MAAM,KAAK,GAAG,MAAA,KAAK,CAAC,KAAK,mCAAI,MAAA,KAAK,CAAC,OAAO,0CAAE,KAAK,CAAC;YAClD,qFAAqF;YACrF,wDAAwD;YACxD,MAAM,CAAC,GAAQ,MAAM,KAAK,CAAC,QAAQ,EAAE,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,EAAE,IAAI,EAAE,CAAC,CAAC;YAC5E,IAAI,CAAC,CAAC;gBAAE,OAAO,SAAS,CAAC;YAEzB,MAAM,IAAI,GAAiB,EAAE,CAAC;YAC9B,IAAI,OAAO,CAAC,CAAC,QAAQ,KAAK,QAAQ,IAAI,OAAO,CAAC,CAAC,SAAS,KAAK,QAAQ,EAAE,CAAC;gBACtE,IAAI,CAAC,GAAG,GAAG,EAAE,GAAG,EAAE,CAAC,CAAC,QAAQ,EAAE,GAAG,EAAE,CAAC,CAAC,SAAS,EAAE,CAAC;YACnD,CAAC;YACD,MAAM,EAAE,GAAG,MAAA,MAAA,CAAC,CAAC,gBAAgB,mCAAI,CAAC,CAAC,UAAU,mCAAI,CAAC,CAAC,UAAU,CAAC;YAC9D,IAAI,EAAE;gBAAE,IAAI,CAAC,SAAS,GAAG,EAAE,YAAY,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC;YAC5E,MAAM,IAAI,GAAG,OAAO,CAAC,CAAC,IAAI,KAAK,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,SAAS,CAAC;YACpE,MAAM,KAAK,GAAG,OAAO,CAAC,CAAC,KAAK,KAAK,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,SAAS,CAAC;YACvE,IAAI,IAAI,IAAI,KAAK;gBAAE,IAAI,CAAC,MAAM,mCAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,GAAK,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAE,CAAC;YAC5F,IAAI,OAAO,CAAC,CAAC,MAAM,KAAK,QAAQ,IAAI,CAAC,CAAC,MAAM,CAAC,IAAI,EAAE;gBAAE,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC;YACnF,IAAI,OAAO,CAAC,CAAC,QAAQ,KAAK,QAAQ,IAAI,CAAC,CAAC,QAAQ,CAAC,IAAI,EAAE;gBAAE,IAAI,CAAC,QAAQ,GAAG,CAAC,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAC;YAE3F,OAAO,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,SAAS,CAAC;QACrD,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACX,MAAM,aAAN,MAAM,uBAAN,MAAM,CAAE,KAAK,CAAC,wBAAwB,QAAQ,KAAK,CAAC,EAAE,CAAC,CAAC;YACxD,OAAO,SAAS,CAAC;QACnB,CAAC;IACH,CAAC;CAAA;AAiBD;;;;GAIG;AACH,SAAsB,QAAQ;yDAC5B,QAAgB,EAChB,OAAO,GAAG,UAAU,EACpB,MAAe;QAEf,IAAI,CAAC;YACH,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,GAAG,MAAM,UAAU,CAAC,OAAO,EAAE,CAAC,QAAQ,CAAC,EAAE,KAAM,CAAC,CAAC;YAC/E,IAAI,IAAI,KAAK,CAAC,EAAE,CAAC;gBACf,sEAAsE;gBACtE,2EAA2E;gBAC3E,IAAI,sCAAsC,CAAC,IAAI,CAAC,GAAG,MAAM,KAAK,MAAM,EAAE,CAAC;oBAAE,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,CAAC;gBACnG,MAAM,aAAN,MAAM,uBAAN,MAAM,CAAE,KAAK,CAAC,mBAAmB,IAAI,QAAQ,QAAQ,KAAK,MAAM,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;gBACvF,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,WAAW,EAAE,IAAI,EAAE,CAAC;YAC/C,CAAC;YACD,OAAO,SAAS,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC;QACvC,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACX,+EAA+E;YAC/E,0DAA0D;YAC1D,MAAM,aAAN,MAAM,uBAAN,MAAM,CAAE,KAAK,CAAC,6BAA6B,QAAQ,KAAK,CAAC,EAAE,CAAC,CAAC;YAC7D,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,WAAW,EAAE,IAAI,EAAE,CAAC;QAC/C,CAAC;IACH,CAAC;CAAA;AAED,8EAA8E;AAC9E,SAAS,SAAS,CAAC,IAAS;;IAC1B,MAAM,SAAS,GAAG,MAAA,IAAI,aAAJ,IAAI,uBAAJ,IAAI,CAAE,SAAS,mCAAI,EAAE,CAAC;IACxC,MAAM,MAAM,GAAG,IAAI,aAAJ,IAAI,uBAAJ,IAAI,CAAE,eAAe,CAAC;IACrC,MAAM,OAAO,GAAG,CAAC,CAAC,MAAM,IAAI,MAAM,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC;IAC9D,IAAI,CAAC,OAAO;QAAE,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,CAAC;IAExC,MAAM,CAAC,GAAG,CAAC,MAAM,IAAI,SAAS,CAAC,MAAM,CAAC,CAAC,IAAI,MAAM,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;IAC7E,MAAM,OAAO,GAAG,MAAC,CAAS,aAAT,CAAC,uBAAD,CAAC,CAAU,cAAc,mCAAI,EAAE,CAAC;IACjD,MAAM,MAAM,GAAG,OAAO,OAAO,CAAC,MAAM,KAAK,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,OAAO,OAAO,CAAC,WAAW,KAAK,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,WAAW,CAAC,CAAC,CAAC,SAAS,CAAC;IAC/I,MAAM,OAAO,GAAU,KAAK,CAAC,OAAO,CAAC,IAAI,aAAJ,IAAI,uBAAJ,IAAI,CAAE,iBAAiB,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,iBAAiB,CAAC,CAAC,CAAC,EAAE,CAAC;IAC5F,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE,WAAC,OAAA,CAAC,qBAAqB,CAAC,IAAI,CAAC,MAAM,CAAC,MAAA,CAAC,aAAD,CAAC,uBAAD,CAAC,CAAE,IAAI,mCAAI,EAAE,CAAC,CAAC,CAAA,EAAA,CAAC,CAAC;IACvF,+EAA+E;IAC/E,MAAM,WAAW,GAAG,0BAA0B,CAAC,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC;IAC1E,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE,MAAM,EAAE,WAAW,EAAE,CAAC;AACvD,CAAC;AAED,oFAAoF;AACpF,SAAS,UAAU,CACjB,OAAe,EACf,IAAc,EACd,SAAiB;IAEjB,OAAO,IAAI,OAAO,CAAC,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE;;QACrC,MAAM,KAAK,GAAG,IAAA,qBAAK,EAAC,OAAO,EAAE,IAAI,EAAE,EAAE,KAAK,EAAE,CAAC,QAAQ,EAAE,MAAM,EAAE,MAAM,CAAC,EAAE,CAAC,CAAC;QAC1E,IAAI,MAAM,GAAG,EAAE,CAAC;QAChB,IAAI,MAAM,GAAG,EAAE,CAAC;QAChB,MAAA,KAAK,CAAC,MAAM,0CAAE,EAAE,CAAC,MAAM,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,MAAM,IAAI,CAAC,CAAC,QAAQ,EAAE,CAAC,CAAC,CAAC;QAC1D,MAAA,KAAK,CAAC,MAAM,0CAAE,EAAE,CAAC,MAAM,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,MAAM,IAAI,CAAC,CAAC,QAAQ,EAAE,CAAC,CAAC,CAAC;QAC1D,MAAM,KAAK,GAAG,UAAU,CAAC,GAAG,EAAE;YAC5B,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;YACtB,MAAM,CAAC,IAAI,KAAK,CAAC,GAAG,OAAO,oBAAoB,SAAS,IAAI,CAAC,CAAC,CAAC;QACjE,CAAC,EAAE,SAAS,CAAC,CAAC;QACd,KAAK,CAAC,EAAE,CAAC,OAAO,EAAE,CAAC,CAAC,EAAE,EAAE;YACtB,YAAY,CAAC,KAAK,CAAC,CAAC;YACpB,MAAM,CAAC,CAAC,CAAC,CAAC;QACZ,CAAC,CAAC,CAAC;QACH,KAAK,CAAC,EAAE,CAAC,OAAO,EAAE,CAAC,IAAI,EAAE,EAAE;YACzB,YAAY,CAAC,KAAK,CAAC,CAAC;YACpB,OAAO,CAAC,EAAE,IAAI,EAAE,IAAI,aAAJ,IAAI,cAAJ,IAAI,GAAI,CAAC,EAAE,MAAM,EAAE,MAAM,EAAE,CAAC,CAAC;QAC/C,CAAC,CAAC,CAAC;IACL,CAAC,CAAC,CAAC;AACL,CAAC"}
@@ -0,0 +1,41 @@
1
+ "use strict";
2
+ var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
3
+ if (k2 === undefined) k2 = k;
4
+ var desc = Object.getOwnPropertyDescriptor(m, k);
5
+ if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
6
+ desc = { enumerable: true, get: function() { return m[k]; } };
7
+ }
8
+ Object.defineProperty(o, k2, desc);
9
+ }) : (function(o, m, k, k2) {
10
+ if (k2 === undefined) k2 = k;
11
+ o[k2] = m[k];
12
+ }));
13
+ var __exportStar = (this && this.__exportStar) || function(m, exports) {
14
+ for (var p in m) if (p !== "default" && !Object.prototype.hasOwnProperty.call(exports, p)) __createBinding(exports, m, p);
15
+ };
16
+ Object.defineProperty(exports, "__esModule", { value: true });
17
+ __exportStar(require("./AudioReader"), exports);
18
+ __exportStar(require("./BinaryReader"), exports);
19
+ __exportStar(require("./ChandraPdfReader"), exports);
20
+ __exportStar(require("./ChatExportReader"), exports);
21
+ __exportStar(require("./DoclingReader"), exports);
22
+ __exportStar(require("./EmailReader"), exports);
23
+ __exportStar(require("./EpubReader"), exports);
24
+ __exportStar(require("./FileReader"), exports);
25
+ __exportStar(require("./FileReaderFactory"), exports);
26
+ __exportStar(require("./HtmlReader"), exports);
27
+ __exportStar(require("./ImageReader"), exports);
28
+ __exportStar(require("./JsonFileReader"), exports);
29
+ __exportStar(require("./JupyterReader"), exports);
30
+ __exportStar(require("./LatexReader"), exports);
31
+ __exportStar(require("./MarkdownReader"), exports);
32
+ __exportStar(require("./MarkerPdfReader"), exports);
33
+ __exportStar(require("./MistralOcrReader"), exports);
34
+ __exportStar(require("./OfficeReader"), exports);
35
+ __exportStar(require("./PdfReader"), exports);
36
+ __exportStar(require("./RtfReader"), exports);
37
+ __exportStar(require("./SubtitleReader"), exports);
38
+ __exportStar(require("./TesseractPdfReader"), exports);
39
+ __exportStar(require("./TextReader"), exports);
40
+ __exportStar(require("./TranscriptReader"), exports);
41
+ //# sourceMappingURL=index.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"index.js","sourceRoot":"","sources":["../../../../src/core/processor/readers/index.ts"],"names":[],"mappings":";;;;;;;;;;;;;;;;AAAA,gDAA8B;AAC9B,iDAA+B;AAC/B,qDAAmC;AACnC,qDAAmC;AACnC,kDAAgC;AAChC,gDAA8B;AAC9B,+CAA6B;AAC7B,+CAA6B;AAC7B,sDAAoC;AACpC,+CAA6B;AAC7B,gDAA8B;AAC9B,mDAAiC;AACjC,kDAAgC;AAChC,gDAA8B;AAC9B,mDAAiC;AACjC,oDAAkC;AAClC,qDAAmC;AACnC,iDAA+B;AAC/B,8CAA4B;AAC5B,8CAA4B;AAC5B,mDAAiC;AACjC,uDAAqC;AACrC,+CAA6B;AAC7B,qDAAmC"}
@@ -0,0 +1,198 @@
1
+ "use strict";
2
+ /**
3
+ * Reader-side raw reference extraction (Phase 0, network-free).
4
+ *
5
+ * Pure functions that pull the *references a document already contains* out of
6
+ * its source text — internal links (markdown `[t](u)`, `[[wikilinks]]`, HTML
7
+ * `href`) and citations (a trailing bibliography block + inline arXiv/DOI/PMID).
8
+ * Nothing here resolves or fetches: readers stash the raw result on
9
+ * `metadata.references` and the corpus-level `ReferenceResolver` turns it into
10
+ * `links_to` / `cites` edges once the full file set is known.
11
+ *
12
+ * Citation parsing is hybrid: Citation.js handles structured BibTeX/`.bib`
13
+ * blocks; a regex fallback handles prose bibliographies (the common case) and
14
+ * inline ids. Citation.js throwing on prose is expected and caught.
15
+ */
16
+ Object.defineProperty(exports, "__esModule", { value: true });
17
+ exports.isExternalTarget = isExternalTarget;
18
+ exports.extractMarkdownLinks = extractMarkdownLinks;
19
+ exports.extractHtmlLinks = extractHtmlLinks;
20
+ exports.extractBareUrls = extractBareUrls;
21
+ exports.extractCitations = extractCitations;
22
+ let citeCtor;
23
+ function loadCite() {
24
+ if (citeCtor !== undefined)
25
+ return citeCtor;
26
+ try {
27
+ const core = require("@citation-js/core");
28
+ require("@citation-js/plugin-bibtex"); // side-effect: registers the @biblatex input
29
+ citeCtor = core.Cite;
30
+ }
31
+ catch (_a) {
32
+ citeCtor = null;
33
+ }
34
+ return citeCtor;
35
+ }
36
+ // --- ID patterns (global; reused for block + inline scans) --------------------
37
+ const ARXIV_RE = /arxiv:\s*(\d{4}\.\d{4,5}(?:v\d+)?)/gi;
38
+ const DOI_RE = /\b10\.\d{4,9}\/[-._;()/:a-z0-9]+/gi;
39
+ const PMID_RE = /\bpmid:\s*(\d{1,9})\b/gi;
40
+ const REF_HEADING_RE = /^(?:#{1,6}\s*)?(?:[\divxlc]+\.?\s+)?(references|bibliography|works cited)\s*:?\s*$/i;
41
+ /** Strip trailing punctuation a DOI/URL regex commonly over-captures. */
42
+ function trimTrailingPunct(s) {
43
+ return s.replace(/[.,;:)\]]+$/, "");
44
+ }
45
+ /** http(s)/protocol-relative/mailto/tel/data targets — external, not a corpus file. */
46
+ function isExternalTarget(target) {
47
+ return /^(?:[a-z][a-z0-9+.-]*:|\/\/)/i.test(target);
48
+ }
49
+ // --- Internal links -----------------------------------------------------------
50
+ /** Markdown inline links `[text](target)` (excluding `![images]`) + `[[wikilinks]]`. */
51
+ function extractMarkdownLinks(text) {
52
+ var _a;
53
+ const links = [];
54
+ // `[text](target ...)` but not image `![text](...)`; ignore pure `#fragment`.
55
+ const inline = /(?<!!)\[([^\]]*)\]\(\s*([^)\s]+)[^)]*\)/g;
56
+ for (const m of text.matchAll(inline)) {
57
+ const target = m[2].trim();
58
+ if (!target || target.startsWith("#"))
59
+ continue;
60
+ links.push({ target, text: m[1].trim() || undefined, kind: "markdown" });
61
+ }
62
+ // `[[target]]` or `[[target|alias]]`.
63
+ const wiki = /\[\[([^\]|]+?)(?:\|([^\]]+))?\]\]/g;
64
+ for (const m of text.matchAll(wiki)) {
65
+ const target = m[1].trim();
66
+ if (!target)
67
+ continue;
68
+ links.push({ target, text: ((_a = m[2]) === null || _a === void 0 ? void 0 : _a.trim()) || undefined, kind: "wikilink" });
69
+ }
70
+ return links;
71
+ }
72
+ /** Anchor `href`s from raw HTML; ignores pure `#fragment` links. */
73
+ function extractHtmlLinks(html) {
74
+ const links = [];
75
+ const anchor = /<a\b[^>]*\bhref\s*=\s*["']([^"']+)["'][^>]*>([\s\S]*?)<\/a>/gi;
76
+ for (const m of html.matchAll(anchor)) {
77
+ const target = m[1].trim();
78
+ if (!target || target.startsWith("#"))
79
+ continue;
80
+ const text = m[2].replace(/<[^>]+>/g, "").trim();
81
+ links.push({ target, text: text || undefined, kind: "html" });
82
+ }
83
+ return links;
84
+ }
85
+ /**
86
+ * Bare absolute URLs anywhere in the text — `> source: https://…` web-clip
87
+ * frontmatter, `Source: …` lines, and inline bare links the markdown-link regex
88
+ * misses (web clips often flatten `[t](u)` to plain text + a source header).
89
+ * These are external by construction; the Phase-1 fetcher consumes them (gated
90
+ * by the allowlist), the Phase-0 resolver ignores them.
91
+ */
92
+ function extractBareUrls(text) {
93
+ const seen = new Set();
94
+ const out = [];
95
+ for (const m of text.matchAll(/\bhttps?:\/\/[^\s)<>"'\]}|]+/gi)) {
96
+ const target = m[0].replace(/[.,;:]+$/, "");
97
+ if (seen.has(target))
98
+ continue;
99
+ seen.add(target);
100
+ out.push({ target, kind: "url" });
101
+ }
102
+ return out;
103
+ }
104
+ // --- Citations ----------------------------------------------------------------
105
+ function citationKey(c) {
106
+ var _a, _b, _c, _d, _e, _f, _g;
107
+ return ((_g = (_e = (_d = (_b = (_a = c.arxivId) === null || _a === void 0 ? void 0 : _a.toLowerCase()) !== null && _b !== void 0 ? _b : (_c = c.doi) === null || _c === void 0 ? void 0 : _c.toLowerCase()) !== null && _d !== void 0 ? _d : (c.pmid ? `pmid:${c.pmid}` : undefined)) !== null && _e !== void 0 ? _e : (_f = c.title) === null || _f === void 0 ? void 0 : _f.toLowerCase()) !== null && _g !== void 0 ? _g : c.raw.toLowerCase());
108
+ }
109
+ /** Pull arXiv-id / DOI / PMID out of a single reference string. */
110
+ function extractIds(s) {
111
+ const arxiv = new RegExp(ARXIV_RE.source, "i").exec(s);
112
+ const doi = new RegExp(DOI_RE.source, "i").exec(s);
113
+ const pmid = new RegExp(PMID_RE.source, "i").exec(s);
114
+ return Object.assign(Object.assign(Object.assign({}, (arxiv ? { arxivId: arxiv[1] } : {})), (doi ? { doi: trimTrailingPunct(doi[0]) } : {})), (pmid ? { pmid: pmid[1] } : {}));
115
+ }
116
+ /** Structured parse via Citation.js; returns [] (not throw) on non-BibTeX input. */
117
+ function parseStructured(block) {
118
+ var _a;
119
+ if (!/@\w+\s*\{/.test(block))
120
+ return []; // cheap guard: looks like BibTeX?
121
+ const Cite = loadCite();
122
+ if (!Cite)
123
+ return [];
124
+ try {
125
+ const entries = (_a = new Cite(block).data) !== null && _a !== void 0 ? _a : [];
126
+ return entries
127
+ .map((e) => {
128
+ var _a;
129
+ const raw = ((_a = e.title) !== null && _a !== void 0 ? _a : "").toString().trim();
130
+ return Object.assign(Object.assign({ raw: raw || JSON.stringify(e).slice(0, 200) }, (e.title ? { title: String(e.title) } : {})), (e.DOI ? { doi: String(e.DOI) } : {}));
131
+ })
132
+ .filter((c) => c.raw.length > 0);
133
+ }
134
+ catch (_b) {
135
+ return [];
136
+ }
137
+ }
138
+ /** Split a prose bibliography block into entries, marker-aware with line fallback. */
139
+ function splitProseEntries(block) {
140
+ const lines = block
141
+ .split("\n")
142
+ .filter((l, i) => !(i === 0 && REF_HEADING_RE.test(l.trim()))) // drop a leading heading
143
+ .map((l) => l.trim());
144
+ const marker = /^\s*(?:\[\d+\]|\(\d+\)|\d+[.)])\s+/;
145
+ const hasMarkers = lines.some((l) => marker.test(l));
146
+ if (hasMarkers) {
147
+ const entries = [];
148
+ for (const line of lines) {
149
+ if (!line)
150
+ continue;
151
+ if (marker.test(line))
152
+ entries.push(line.replace(marker, ""));
153
+ else if (entries.length)
154
+ entries[entries.length - 1] += ` ${line}`;
155
+ else
156
+ entries.push(line);
157
+ }
158
+ return entries;
159
+ }
160
+ // No numbering: one entry per non-empty line.
161
+ return lines.filter(Boolean);
162
+ }
163
+ /**
164
+ * Extract citations from a (possibly undefined) trailing references block plus
165
+ * inline arXiv/DOI ids found anywhere in the body. Deduped by id/title.
166
+ */
167
+ function extractCitations(referencesBlock, bodyText) {
168
+ const out = new Map();
169
+ const add = (c) => {
170
+ const key = citationKey(c);
171
+ if (key && !out.has(key))
172
+ out.set(key, c);
173
+ };
174
+ if (referencesBlock && referencesBlock.trim()) {
175
+ const structured = parseStructured(referencesBlock);
176
+ if (structured.length) {
177
+ structured.forEach((c) => { var _a; return add(Object.assign(Object.assign({}, c), extractIds(c.raw + " " + ((_a = c.doi) !== null && _a !== void 0 ? _a : "")))); });
178
+ }
179
+ else {
180
+ for (const entry of splitProseEntries(referencesBlock)) {
181
+ const ids = extractIds(entry);
182
+ add(Object.assign({ raw: entry.slice(0, 300) }, ids));
183
+ }
184
+ }
185
+ }
186
+ // Inline ids in the body (e.g. "see arXiv:2001.12345") that aren't in the list.
187
+ for (const m of bodyText.matchAll(new RegExp(ARXIV_RE.source, "gi"))) {
188
+ add({ raw: m[0], arxivId: m[1] });
189
+ }
190
+ for (const m of bodyText.matchAll(new RegExp(DOI_RE.source, "gi"))) {
191
+ add({ raw: m[0], doi: trimTrailingPunct(m[0]) });
192
+ }
193
+ for (const m of bodyText.matchAll(new RegExp(PMID_RE.source, "gi"))) {
194
+ add({ raw: m[0], pmid: m[1] });
195
+ }
196
+ return Array.from(out.values());
197
+ }
198
+ //# sourceMappingURL=referenceExtraction.js.map