@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 +1,259 @@
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.EmailReader = void 0;
46
+ const fs = __importStar(require("fs"));
47
+ const mailparser_1 = require("mailparser");
48
+ const FileReader_1 = require("./FileReader");
49
+ const turnPacking_1 = require("./transcript/turnPacking");
50
+ /**
51
+ * html-to-text profile for email bodies — rescued from the kg-mail-assistant
52
+ * prototype's `MailListener`: preserve document structure (headings, lists,
53
+ * blockquotes, tables) while dropping boilerplate (script/style/nav/footer/ads).
54
+ */
55
+ const EMAIL_HTML_OPTIONS = {
56
+ wordwrap: false,
57
+ selectors: [
58
+ // uppercase:false — preserve original case so entity names aren't SHOUTED.
59
+ { selector: "h1", format: "heading", options: { uppercase: false, leadingLineBreaks: 2, trailingLineBreaks: 1 } },
60
+ { selector: "h2", format: "heading", options: { uppercase: false, leadingLineBreaks: 2, trailingLineBreaks: 1 } },
61
+ { selector: "h3", format: "heading", options: { uppercase: false, leadingLineBreaks: 2, trailingLineBreaks: 1 } },
62
+ { selector: "h4", format: "heading", options: { uppercase: false, leadingLineBreaks: 1, trailingLineBreaks: 1 } },
63
+ { selector: "p", format: "paragraph", options: { leadingLineBreaks: 1, trailingLineBreaks: 1 } },
64
+ { selector: "br", format: "lineBreak" },
65
+ { selector: "ul", format: "unorderedList", options: { leadingLineBreaks: 1, trailingLineBreaks: 1 } },
66
+ { selector: "ol", format: "orderedList", options: { leadingLineBreaks: 1, trailingLineBreaks: 1 } },
67
+ { selector: "li", format: "listItem", options: { leadingLineBreaks: 1 } },
68
+ { selector: "blockquote", format: "blockquote", options: { leadingLineBreaks: 1, trailingLineBreaks: 1 } },
69
+ { selector: "table", format: "table" },
70
+ { selector: "a", options: { ignoreHref: true } },
71
+ { selector: "img", format: "skip" },
72
+ { selector: "script", format: "skip" },
73
+ { selector: "style", format: "skip" },
74
+ { selector: "nav", format: "skip" },
75
+ { selector: "header", format: "skip" },
76
+ { selector: "footer", format: "skip" },
77
+ { selector: ".navigation", format: "skip" },
78
+ { selector: ".sidebar", format: "skip" },
79
+ { selector: ".footer", format: "skip" },
80
+ { selector: ".header", format: "skip" },
81
+ { selector: ".menu", format: "skip" },
82
+ { selector: ".advertisement", format: "skip" },
83
+ { selector: ".ads", format: "skip" },
84
+ ],
85
+ };
86
+ /**
87
+ * Reads email (`.eml`, `.mbox`) into **size-packed** chunks by treating each
88
+ * message as a conversational turn — the same path as transcripts/chat exports.
89
+ * Sender → `speaker`, the `Date:` header → `occurredAt` (the observation's
90
+ * bitemporal `validAt`), so an email thread becomes a provenance-rich
91
+ * conversation graph with no email-specific logic leaking downstream.
92
+ *
93
+ * A `.mbox` is split into its messages (`From ` envelope lines); messages from
94
+ * different threads (`References`/`In-Reply-To` root) are tagged with distinct
95
+ * `conversation` ids so `packTurns` never packs two threads into one chunk
96
+ * (KG-10). HTML-only bodies are decoded with a structure-preserving,
97
+ * boilerplate-stripping html-to-text profile; quoted reply chains are stripped
98
+ * so a reply contributes its new content, not N copies of the thread.
99
+ *
100
+ * Registered before TextReader/BinaryReader so `.eml`/`.mbox` (otherwise
101
+ * unclaimed → skipped as binary) route here. `sourceAdapter:"email"` is stamped
102
+ * centrally by FileProcessor from `adapterId()`.
103
+ *
104
+ * v1 scope: `.eml` + `.mbox`. Deferred — `.msg` (binary Outlook); explicit
105
+ * `In-Reply-To → reply_to` edges (a packed chunk can span messages, so a single
106
+ * per-chunk `locator:msgid` is ambiguous too); attachment extraction.
107
+ */
108
+ class EmailReader extends FileReader_1.FileReader {
109
+ constructor(chunker, logger, maxChunkSize, opts) {
110
+ super([".eml", ".mbox"], chunker, logger);
111
+ this.maxChunkSize = maxChunkSize;
112
+ this.maxMessages = opts.maxMessages;
113
+ this.stripQuotes = opts.stripQuotes;
114
+ }
115
+ getName() {
116
+ return "EmailReader";
117
+ }
118
+ adapterId() {
119
+ return "email";
120
+ }
121
+ read(filePath) {
122
+ return __awaiter(this, void 0, void 0, function* () {
123
+ const raw = yield fs.promises.readFile(filePath, "utf-8");
124
+ try {
125
+ const blocks = filePath.toLowerCase().endsWith(".mbox") ? this.splitMbox(raw) : [raw];
126
+ if (blocks.length === 0)
127
+ return this.plainFallback(raw);
128
+ let limited = blocks;
129
+ if (blocks.length > this.maxMessages) {
130
+ this.logger.warn(`EmailReader: ${filePath} has ${blocks.length} messages; truncating to maxMessages=${this.maxMessages}`);
131
+ limited = blocks.slice(0, this.maxMessages);
132
+ }
133
+ const threadIndex = new Map();
134
+ const turns = [];
135
+ for (const block of limited) {
136
+ const turn = yield this.messageToTurn(block, threadIndex);
137
+ if (turn)
138
+ turns.push(turn);
139
+ }
140
+ if (turns.length === 0)
141
+ return this.plainFallback(raw);
142
+ const chunks = yield (0, turnPacking_1.packTurns)(turns, filePath, this.maxChunkSize, this.chunker);
143
+ return { chunks, metadata: { type: "email", source: filePath, messages: turns.length } };
144
+ }
145
+ catch (e) {
146
+ this.logger.warn(`EmailReader could not parse ${filePath}; falling back to plain chunking: ${e}`);
147
+ return this.plainFallback(raw);
148
+ }
149
+ });
150
+ }
151
+ /** Parse one RFC822 message into a conversational turn (null if empty). */
152
+ messageToTurn(block, threadIndex) {
153
+ return __awaiter(this, void 0, void 0, function* () {
154
+ var _a, _b;
155
+ const parsed = yield (0, mailparser_1.simpleParser)(block);
156
+ // Prefer the HTML part (decoded with our boilerplate-stripping profile) over
157
+ // mailparser's auto-generated `text` — for HTML-only mail mailparser populates
158
+ // `text` itself, but without skipping nav/footer/ads, so our selectors win.
159
+ let body = parsed.html
160
+ ? (yield this.htmlToText(parsed.html)).trim()
161
+ : ((_a = parsed.text) !== null && _a !== void 0 ? _a : "").trim();
162
+ if (this.stripQuotes)
163
+ body = this.stripQuotedReplies(body).trim();
164
+ const subject = ((_b = parsed.subject) !== null && _b !== void 0 ? _b : "").trim();
165
+ if (!body && !subject)
166
+ return null;
167
+ const text = subject ? `Subject: ${subject}\n${body}`.trim() : body;
168
+ const occurredAt = parsed.date instanceof Date && !Number.isNaN(parsed.date.getTime())
169
+ ? parsed.date.toISOString()
170
+ : undefined;
171
+ return {
172
+ speaker: this.senderName(parsed.from),
173
+ text,
174
+ occurredAt,
175
+ conversation: this.threadIndexFor(parsed, threadIndex),
176
+ };
177
+ });
178
+ }
179
+ /** Display name from `From:` (`Name <addr>` → Name, else the address). */
180
+ senderName(from) {
181
+ var _a;
182
+ const v = (_a = from === null || from === void 0 ? void 0 : from.value) === null || _a === void 0 ? void 0 : _a[0];
183
+ if ((v === null || v === void 0 ? void 0 : v.name) && v.name.trim())
184
+ return v.name.trim();
185
+ if ((v === null || v === void 0 ? void 0 : v.address) && v.address.trim())
186
+ return v.address.trim();
187
+ if ((from === null || from === void 0 ? void 0 : from.text) && from.text.trim())
188
+ return from.text.trim();
189
+ return "unknown";
190
+ }
191
+ /**
192
+ * Map a message to a 0-based thread index by its conversation root
193
+ * (`References[0]` ?? `In-Reply-To` ?? own `Message-ID`), so distinct threads
194
+ * in one `.mbox` get distinct `conversation` ids and never share a chunk.
195
+ */
196
+ threadIndexFor(parsed, map) {
197
+ var _a, _b;
198
+ const refs = parsed.references;
199
+ const root = Array.isArray(refs)
200
+ ? refs[0]
201
+ : typeof refs === "string"
202
+ ? refs.trim().split(/\s+/)[0]
203
+ : undefined;
204
+ const key = ((_b = (_a = root !== null && root !== void 0 ? root : parsed.inReplyTo) !== null && _a !== void 0 ? _a : parsed.messageId) !== null && _b !== void 0 ? _b : `__${map.size}`).trim();
205
+ if (!map.has(key))
206
+ map.set(key, map.size);
207
+ return map.get(key);
208
+ }
209
+ /** Split an mbox into RFC822 message blocks on `From ` envelope lines. */
210
+ splitMbox(raw) {
211
+ const fromLine = /^From .+\b(?:19|20)\d{2}\b/;
212
+ const blocks = [];
213
+ let cur = [];
214
+ for (const line of raw.split(/\r?\n/)) {
215
+ if (fromLine.test(line)) {
216
+ if (cur.length > 0) {
217
+ blocks.push(cur.join("\n"));
218
+ cur = [];
219
+ }
220
+ continue; // drop the envelope line itself
221
+ }
222
+ // Unescape mbox quoting: ">From " → "From ", ">>From " → ">From ".
223
+ cur.push(line.replace(/^>(>*From )/, "$1"));
224
+ }
225
+ if (cur.length > 0)
226
+ blocks.push(cur.join("\n"));
227
+ return blocks.map((b) => b.trim()).filter(Boolean);
228
+ }
229
+ htmlToText(html) {
230
+ return __awaiter(this, void 0, void 0, function* () {
231
+ const { convert } = yield Promise.resolve().then(() => __importStar(require("html-to-text")));
232
+ return convert(html, EMAIL_HTML_OPTIONS);
233
+ });
234
+ }
235
+ /**
236
+ * Light, best-effort quote stripping: cut the body at the first quote-chain
237
+ * marker ("On … wrote:" / "----- Original Message -----") and drop surviving
238
+ * leading-`>` lines. Not a full reply parser — keeps each message's own prose.
239
+ */
240
+ stripQuotedReplies(body) {
241
+ const onWrote = /^\s*On\b.*\bwrote:\s*$/i;
242
+ const origMsg = /^\s*-{2,}\s*Original Message\s*-{2,}/i;
243
+ const out = [];
244
+ for (const line of body.split(/\r?\n/)) {
245
+ if (onWrote.test(line) || origMsg.test(line))
246
+ break;
247
+ out.push(line);
248
+ }
249
+ return out.filter((l) => !/^\s*>/.test(l)).join("\n");
250
+ }
251
+ plainFallback(raw) {
252
+ return __awaiter(this, void 0, void 0, function* () {
253
+ const parts = yield this.chunker.chunk(raw);
254
+ return { chunks: parts.map((p) => (Object.assign({}, p))), metadata: { type: "email-fallback" } };
255
+ });
256
+ }
257
+ }
258
+ exports.EmailReader = EmailReader;
259
+ //# sourceMappingURL=EmailReader.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"EmailReader.js","sourceRoot":"","sources":["../../../../src/core/processor/readers/EmailReader.ts"],"names":[],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAAA,uCAAyB;AACzB,2CAAsD;AAEtD,6CAA0D;AAG1D,0DAA2D;AAU3D;;;;GAIG;AACH,MAAM,kBAAkB,GAAsB;IAC5C,QAAQ,EAAE,KAAK;IACf,SAAS,EAAE;QACT,2EAA2E;QAC3E,EAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,EAAE,SAAS,EAAE,KAAK,EAAE,iBAAiB,EAAE,CAAC,EAAE,kBAAkB,EAAE,CAAC,EAAE,EAAE;QACjH,EAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,EAAE,SAAS,EAAE,KAAK,EAAE,iBAAiB,EAAE,CAAC,EAAE,kBAAkB,EAAE,CAAC,EAAE,EAAE;QACjH,EAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,EAAE,SAAS,EAAE,KAAK,EAAE,iBAAiB,EAAE,CAAC,EAAE,kBAAkB,EAAE,CAAC,EAAE,EAAE;QACjH,EAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,EAAE,SAAS,EAAE,KAAK,EAAE,iBAAiB,EAAE,CAAC,EAAE,kBAAkB,EAAE,CAAC,EAAE,EAAE;QACjH,EAAE,QAAQ,EAAE,GAAG,EAAE,MAAM,EAAE,WAAW,EAAE,OAAO,EAAE,EAAE,iBAAiB,EAAE,CAAC,EAAE,kBAAkB,EAAE,CAAC,EAAE,EAAE;QAChG,EAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,EAAE,WAAW,EAAE;QACvC,EAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,EAAE,eAAe,EAAE,OAAO,EAAE,EAAE,iBAAiB,EAAE,CAAC,EAAE,kBAAkB,EAAE,CAAC,EAAE,EAAE;QACrG,EAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,EAAE,aAAa,EAAE,OAAO,EAAE,EAAE,iBAAiB,EAAE,CAAC,EAAE,kBAAkB,EAAE,CAAC,EAAE,EAAE;QACnG,EAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,EAAE,UAAU,EAAE,OAAO,EAAE,EAAE,iBAAiB,EAAE,CAAC,EAAE,EAAE;QACzE,EAAE,QAAQ,EAAE,YAAY,EAAE,MAAM,EAAE,YAAY,EAAE,OAAO,EAAE,EAAE,iBAAiB,EAAE,CAAC,EAAE,kBAAkB,EAAE,CAAC,EAAE,EAAE;QAC1G,EAAE,QAAQ,EAAE,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE;QACtC,EAAE,QAAQ,EAAE,GAAG,EAAE,OAAO,EAAE,EAAE,UAAU,EAAE,IAAI,EAAE,EAAE;QAChD,EAAE,QAAQ,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE;QACnC,EAAE,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE,MAAM,EAAE;QACtC,EAAE,QAAQ,EAAE,OAAO,EAAE,MAAM,EAAE,MAAM,EAAE;QACrC,EAAE,QAAQ,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE;QACnC,EAAE,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE,MAAM,EAAE;QACtC,EAAE,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE,MAAM,EAAE;QACtC,EAAE,QAAQ,EAAE,aAAa,EAAE,MAAM,EAAE,MAAM,EAAE;QAC3C,EAAE,QAAQ,EAAE,UAAU,EAAE,MAAM,EAAE,MAAM,EAAE;QACxC,EAAE,QAAQ,EAAE,SAAS,EAAE,MAAM,EAAE,MAAM,EAAE;QACvC,EAAE,QAAQ,EAAE,SAAS,EAAE,MAAM,EAAE,MAAM,EAAE;QACvC,EAAE,QAAQ,EAAE,OAAO,EAAE,MAAM,EAAE,MAAM,EAAE;QACrC,EAAE,QAAQ,EAAE,gBAAgB,EAAE,MAAM,EAAE,MAAM,EAAE;QAC9C,EAAE,QAAQ,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE;KACrC;CACF,CAAC;AAEF;;;;;;;;;;;;;;;;;;;;;GAqBG;AACH,MAAa,WAAY,SAAQ,uBAAU;IAKzC,YAAY,OAAoB,EAAE,MAAc,EAAE,YAAoB,EAAE,IAAwB;QAC9F,KAAK,CAAC,CAAC,MAAM,EAAE,OAAO,CAAC,EAAE,OAAO,EAAE,MAAM,CAAC,CAAC;QAC1C,IAAI,CAAC,YAAY,GAAG,YAAY,CAAC;QACjC,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC,WAAW,CAAC;QACpC,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC,WAAW,CAAC;IACtC,CAAC;IAED,OAAO;QACL,OAAO,aAAa,CAAC;IACvB,CAAC;IAED,SAAS;QACP,OAAO,OAAO,CAAC;IACjB,CAAC;IAEK,IAAI,CAAC,QAAgB;;YACzB,MAAM,GAAG,GAAG,MAAM,EAAE,CAAC,QAAQ,CAAC,QAAQ,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC;YAC1D,IAAI,CAAC;gBACH,MAAM,MAAM,GAAG,QAAQ,CAAC,WAAW,EAAE,CAAC,QAAQ,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC;gBACtF,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;oBAAE,OAAO,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC,CAAC;gBAExD,IAAI,OAAO,GAAG,MAAM,CAAC;gBACrB,IAAI,MAAM,CAAC,MAAM,GAAG,IAAI,CAAC,WAAW,EAAE,CAAC;oBACrC,IAAI,CAAC,MAAM,CAAC,IAAI,CACd,gBAAgB,QAAQ,QAAQ,MAAM,CAAC,MAAM,wCAAwC,IAAI,CAAC,WAAW,EAAE,CACxG,CAAC;oBACF,OAAO,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,IAAI,CAAC,WAAW,CAAC,CAAC;gBAC9C,CAAC;gBAED,MAAM,WAAW,GAAG,IAAI,GAAG,EAAkB,CAAC;gBAC9C,MAAM,KAAK,GAAW,EAAE,CAAC;gBACzB,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;oBAC5B,MAAM,IAAI,GAAG,MAAM,IAAI,CAAC,aAAa,CAAC,KAAK,EAAE,WAAW,CAAC,CAAC;oBAC1D,IAAI,IAAI;wBAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;gBAC7B,CAAC;gBACD,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;oBAAE,OAAO,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC,CAAC;gBAEvD,MAAM,MAAM,GAAG,MAAM,IAAA,uBAAS,EAAC,KAAK,EAAE,QAAQ,EAAE,IAAI,CAAC,YAAY,EAAE,IAAI,CAAC,OAAO,CAAC,CAAC;gBACjF,OAAO,EAAE,MAAM,EAAE,QAAQ,EAAE,EAAE,IAAI,EAAE,OAAO,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,EAAE,KAAK,CAAC,MAAM,EAAE,EAAE,CAAC;YAC3F,CAAC;YAAC,OAAO,CAAC,EAAE,CAAC;gBACX,IAAI,CAAC,MAAM,CAAC,IAAI,CACd,+BAA+B,QAAQ,qCAAqC,CAAC,EAAE,CAChF,CAAC;gBACF,OAAO,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC,CAAC;YACjC,CAAC;QACH,CAAC;KAAA;IAED,2EAA2E;IAC7D,aAAa,CAAC,KAAa,EAAE,WAAgC;;;YACzE,MAAM,MAAM,GAAG,MAAM,IAAA,yBAAY,EAAC,KAAK,CAAC,CAAC;YAEzC,6EAA6E;YAC7E,+EAA+E;YAC/E,4EAA4E;YAC5E,IAAI,IAAI,GAAG,MAAM,CAAC,IAAI;gBACpB,CAAC,CAAC,CAAC,MAAM,IAAI,CAAC,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,EAAE;gBAC7C,CAAC,CAAC,CAAC,MAAA,MAAM,CAAC,IAAI,mCAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;YAC/B,IAAI,IAAI,CAAC,WAAW;gBAAE,IAAI,GAAG,IAAI,CAAC,kBAAkB,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC;YAElE,MAAM,OAAO,GAAG,CAAC,MAAA,MAAM,CAAC,OAAO,mCAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;YAC9C,IAAI,CAAC,IAAI,IAAI,CAAC,OAAO;gBAAE,OAAO,IAAI,CAAC;YAEnC,MAAM,IAAI,GAAG,OAAO,CAAC,CAAC,CAAC,YAAY,OAAO,KAAK,IAAI,EAAE,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC;YACpE,MAAM,UAAU,GACd,MAAM,CAAC,IAAI,YAAY,IAAI,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,IAAI,CAAC,OAAO,EAAE,CAAC;gBACjE,CAAC,CAAC,MAAM,CAAC,IAAI,CAAC,WAAW,EAAE;gBAC3B,CAAC,CAAC,SAAS,CAAC;YAEhB,OAAO;gBACL,OAAO,EAAE,IAAI,CAAC,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC;gBACrC,IAAI;gBACJ,UAAU;gBACV,YAAY,EAAE,IAAI,CAAC,cAAc,CAAC,MAAM,EAAE,WAAW,CAAC;aACvD,CAAC;QACJ,CAAC;KAAA;IAED,0EAA0E;IAClE,UAAU,CAAC,IAAwB;;QACzC,MAAM,CAAC,GAAG,MAAA,IAAI,aAAJ,IAAI,uBAAJ,IAAI,CAAE,KAAK,0CAAG,CAAC,CAAC,CAAC;QAC3B,IAAI,CAAA,CAAC,aAAD,CAAC,uBAAD,CAAC,CAAE,IAAI,KAAI,CAAC,CAAC,IAAI,CAAC,IAAI,EAAE;YAAE,OAAO,CAAC,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;QACnD,IAAI,CAAA,CAAC,aAAD,CAAC,uBAAD,CAAC,CAAE,OAAO,KAAI,CAAC,CAAC,OAAO,CAAC,IAAI,EAAE;YAAE,OAAO,CAAC,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC;QAC5D,IAAI,CAAA,IAAI,aAAJ,IAAI,uBAAJ,IAAI,CAAE,IAAI,KAAI,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE;YAAE,OAAO,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;QAC5D,OAAO,SAAS,CAAC;IACnB,CAAC;IAED;;;;OAIG;IACK,cAAc,CAAC,MAAkB,EAAE,GAAwB;;QACjE,MAAM,IAAI,GAAG,MAAM,CAAC,UAAU,CAAC;QAC/B,MAAM,IAAI,GAAG,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC;YAC9B,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;YACT,CAAC,CAAC,OAAO,IAAI,KAAK,QAAQ;gBACxB,CAAC,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;gBAC7B,CAAC,CAAC,SAAS,CAAC;QAChB,MAAM,GAAG,GAAG,CAAC,MAAA,MAAA,IAAI,aAAJ,IAAI,cAAJ,IAAI,GAAI,MAAM,CAAC,SAAS,mCAAI,MAAM,CAAC,SAAS,mCAAI,KAAK,GAAG,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;QACrF,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC;YAAE,GAAG,CAAC,GAAG,CAAC,GAAG,EAAE,GAAG,CAAC,IAAI,CAAC,CAAC;QAC1C,OAAO,GAAG,CAAC,GAAG,CAAC,GAAG,CAAE,CAAC;IACvB,CAAC;IAED,0EAA0E;IAClE,SAAS,CAAC,GAAW;QAC3B,MAAM,QAAQ,GAAG,4BAA4B,CAAC;QAC9C,MAAM,MAAM,GAAa,EAAE,CAAC;QAC5B,IAAI,GAAG,GAAa,EAAE,CAAC;QACvB,KAAK,MAAM,IAAI,IAAI,GAAG,CAAC,KAAK,CAAC,OAAO,CAAC,EAAE,CAAC;YACtC,IAAI,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;gBACxB,IAAI,GAAG,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;oBACnB,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;oBAC5B,GAAG,GAAG,EAAE,CAAC;gBACX,CAAC;gBACD,SAAS,CAAC,gCAAgC;YAC5C,CAAC;YACD,mEAAmE;YACnE,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,aAAa,EAAE,IAAI,CAAC,CAAC,CAAC;QAC9C,CAAC;QACD,IAAI,GAAG,CAAC,MAAM,GAAG,CAAC;YAAE,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;QAChD,OAAO,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC;IACrD,CAAC;IAEa,UAAU,CAAC,IAAY;;YACnC,MAAM,EAAE,OAAO,EAAE,GAAG,wDAAa,cAAc,GAAC,CAAC;YACjD,OAAO,OAAO,CAAC,IAAI,EAAE,kBAAkB,CAAC,CAAC;QAC3C,CAAC;KAAA;IAED;;;;OAIG;IACK,kBAAkB,CAAC,IAAY;QACrC,MAAM,OAAO,GAAG,yBAAyB,CAAC;QAC1C,MAAM,OAAO,GAAG,uCAAuC,CAAC;QACxD,MAAM,GAAG,GAAa,EAAE,CAAC;QACzB,KAAK,MAAM,IAAI,IAAI,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,EAAE,CAAC;YACvC,IAAI,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC;gBAAE,MAAM;YACpD,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACjB,CAAC;QACD,OAAO,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IACxD,CAAC;IAEa,aAAa,CAAC,GAAW;;YACrC,MAAM,KAAK,GAAG,MAAM,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;YAC5C,OAAO,EAAE,MAAM,EAAE,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,mBAAM,CAAC,EAAG,CAAC,EAAE,QAAQ,EAAE,EAAE,IAAI,EAAE,gBAAgB,EAAE,EAAE,CAAC;QACxF,CAAC;KAAA;CACF;AAxJD,kCAwJC"}
@@ -0,0 +1,175 @@
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
+ var __importDefault = (this && this.__importDefault) || function (mod) {
45
+ return (mod && mod.__esModule) ? mod : { "default": mod };
46
+ };
47
+ Object.defineProperty(exports, "__esModule", { value: true });
48
+ exports.EpubReader = void 0;
49
+ const path = __importStar(require("path"));
50
+ const adm_zip_1 = __importDefault(require("adm-zip"));
51
+ const cheerio = __importStar(require("cheerio"));
52
+ const FileReader_1 = require("./FileReader");
53
+ /** html-to-text profile for EPUB chapter XHTML — keep structure, drop chrome. */
54
+ const EPUB_HTML_OPTIONS = {
55
+ wordwrap: false,
56
+ selectors: [
57
+ { selector: "h1", format: "heading", options: { uppercase: false, leadingLineBreaks: 2, trailingLineBreaks: 1 } },
58
+ { selector: "h2", format: "heading", options: { uppercase: false, leadingLineBreaks: 2, trailingLineBreaks: 1 } },
59
+ { selector: "h3", format: "heading", options: { uppercase: false, leadingLineBreaks: 1, trailingLineBreaks: 1 } },
60
+ { selector: "p", format: "paragraph", options: { leadingLineBreaks: 1, trailingLineBreaks: 1 } },
61
+ { selector: "ul", format: "unorderedList", options: { leadingLineBreaks: 1, trailingLineBreaks: 1 } },
62
+ { selector: "ol", format: "orderedList", options: { leadingLineBreaks: 1, trailingLineBreaks: 1 } },
63
+ { selector: "li", format: "listItem" },
64
+ { selector: "blockquote", format: "blockquote", options: { leadingLineBreaks: 1, trailingLineBreaks: 1 } },
65
+ { selector: "a", options: { ignoreHref: true } },
66
+ { selector: "img", format: "skip" },
67
+ { selector: "script", format: "skip" },
68
+ { selector: "style", format: "skip" },
69
+ { selector: "nav", format: "skip" },
70
+ ],
71
+ };
72
+ /**
73
+ * Reads EPUB (`.epub`) into **chapter-boundary** chunks. An EPUB is a ZIP of
74
+ * XHTML documents ordered by the OPF spine; this reader unzips (adm-zip), reads
75
+ * the spine (cheerio xmlMode — the `GrobidClient` precedent, no new XML dep),
76
+ * converts each chapter's XHTML to text (html-to-text), and chunks **each
77
+ * chapter separately** (re-indexed across all) so a chunk never spans two
78
+ * chapters. `sourceAdapter:"epub"` is stamped centrally from `adapterId()`.
79
+ *
80
+ * v1 deferred: TOC/nav-driven chapter titles, embedded images, EPUB3 nav specifics.
81
+ */
82
+ class EpubReader extends FileReader_1.FileReader {
83
+ constructor(chunker, logger, maxChunkSize) {
84
+ super([".epub"], chunker, logger);
85
+ this.maxChunkSize = maxChunkSize;
86
+ }
87
+ getName() {
88
+ return "EpubReader";
89
+ }
90
+ adapterId() {
91
+ return "epub";
92
+ }
93
+ read(filePath) {
94
+ return __awaiter(this, void 0, void 0, function* () {
95
+ try {
96
+ const zip = new adm_zip_1.default(filePath);
97
+ const opfPath = this.findOpfPath(zip);
98
+ if (!opfPath) {
99
+ this.logger.warn(`EpubReader: no OPF rootfile in ${filePath}`);
100
+ return { chunks: [], metadata: { type: "epub-error" } };
101
+ }
102
+ const chapters = yield this.readChapters(zip, opfPath);
103
+ if (chapters.length === 0)
104
+ return { chunks: [], metadata: { type: "epub-empty" } };
105
+ // Chunk each chapter independently, then re-index so no chunk spans two.
106
+ const all = [];
107
+ for (const ch of chapters) {
108
+ const text = (ch.title ? `# ${ch.title}\n\n` : "") + ch.text;
109
+ if (!text.trim())
110
+ continue;
111
+ for (const p of yield this.chunker.chunk(text))
112
+ all.push(Object.assign({}, p));
113
+ }
114
+ all.forEach((c, i) => {
115
+ c.index = i + 1;
116
+ c.totalChunks = all.length;
117
+ });
118
+ return { chunks: all, metadata: { type: "epub", chapters: chapters.length } };
119
+ }
120
+ catch (e) {
121
+ this.logger.warn(`EpubReader could not parse ${filePath}: ${e}`);
122
+ return { chunks: [], metadata: { type: "epub-error" } };
123
+ }
124
+ });
125
+ }
126
+ /** META-INF/container.xml → the OPF package path. */
127
+ findOpfPath(zip) {
128
+ const container = zip.getEntry("META-INF/container.xml");
129
+ if (!container)
130
+ return undefined;
131
+ const $ = cheerio.load(zip.readAsText(container), { xmlMode: true });
132
+ return $("rootfile").first().attr("full-path") || undefined;
133
+ }
134
+ /** OPF manifest+spine → ordered chapter {title, text}. */
135
+ readChapters(zip, opfPath) {
136
+ return __awaiter(this, void 0, void 0, function* () {
137
+ var _a;
138
+ const $ = cheerio.load(zip.readAsText(opfPath), { xmlMode: true });
139
+ const hrefById = new Map();
140
+ $("manifest item").each((_, el) => {
141
+ const id = $(el).attr("id");
142
+ const href = $(el).attr("href");
143
+ if (id && href)
144
+ hrefById.set(id, href);
145
+ });
146
+ const opfDir = path.posix.dirname(opfPath.split(path.sep).join("/"));
147
+ const chapters = [];
148
+ const spine = $("spine itemref").toArray();
149
+ for (const el of spine) {
150
+ const href = hrefById.get($(el).attr("idref") || "");
151
+ if (!href)
152
+ continue;
153
+ const entryName = opfDir === "." ? href : `${opfDir}/${href}`;
154
+ const entry = (_a = zip.getEntry(entryName)) !== null && _a !== void 0 ? _a : zip.getEntry(decodeURIComponent(entryName));
155
+ if (!entry)
156
+ continue;
157
+ const xhtml = zip.readAsText(entry);
158
+ const { title, text } = yield this.xhtmlToChapter(xhtml);
159
+ if (text.trim())
160
+ chapters.push({ title, text });
161
+ }
162
+ return chapters;
163
+ });
164
+ }
165
+ xhtmlToChapter(xhtml) {
166
+ return __awaiter(this, void 0, void 0, function* () {
167
+ const $ = cheerio.load(xhtml);
168
+ const title = ($("title").first().text() || $("h1").first().text() || "").trim() || undefined;
169
+ const { convert } = yield Promise.resolve().then(() => __importStar(require("html-to-text")));
170
+ return { title, text: convert(xhtml, EPUB_HTML_OPTIONS) };
171
+ });
172
+ }
173
+ }
174
+ exports.EpubReader = EpubReader;
175
+ //# sourceMappingURL=EpubReader.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"EpubReader.js","sourceRoot":"","sources":["../../../../src/core/processor/readers/EpubReader.ts"],"names":[],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAAA,2CAA6B;AAC7B,sDAA6B;AAC7B,iDAAmC;AAEnC,6CAAuE;AAIvE,iFAAiF;AACjF,MAAM,iBAAiB,GAAsB;IAC3C,QAAQ,EAAE,KAAK;IACf,SAAS,EAAE;QACT,EAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,EAAE,SAAS,EAAE,KAAK,EAAE,iBAAiB,EAAE,CAAC,EAAE,kBAAkB,EAAE,CAAC,EAAE,EAAE;QACjH,EAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,EAAE,SAAS,EAAE,KAAK,EAAE,iBAAiB,EAAE,CAAC,EAAE,kBAAkB,EAAE,CAAC,EAAE,EAAE;QACjH,EAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,EAAE,SAAS,EAAE,KAAK,EAAE,iBAAiB,EAAE,CAAC,EAAE,kBAAkB,EAAE,CAAC,EAAE,EAAE;QACjH,EAAE,QAAQ,EAAE,GAAG,EAAE,MAAM,EAAE,WAAW,EAAE,OAAO,EAAE,EAAE,iBAAiB,EAAE,CAAC,EAAE,kBAAkB,EAAE,CAAC,EAAE,EAAE;QAChG,EAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,EAAE,eAAe,EAAE,OAAO,EAAE,EAAE,iBAAiB,EAAE,CAAC,EAAE,kBAAkB,EAAE,CAAC,EAAE,EAAE;QACrG,EAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,EAAE,aAAa,EAAE,OAAO,EAAE,EAAE,iBAAiB,EAAE,CAAC,EAAE,kBAAkB,EAAE,CAAC,EAAE,EAAE;QACnG,EAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,EAAE,UAAU,EAAE;QACtC,EAAE,QAAQ,EAAE,YAAY,EAAE,MAAM,EAAE,YAAY,EAAE,OAAO,EAAE,EAAE,iBAAiB,EAAE,CAAC,EAAE,kBAAkB,EAAE,CAAC,EAAE,EAAE;QAC1G,EAAE,QAAQ,EAAE,GAAG,EAAE,OAAO,EAAE,EAAE,UAAU,EAAE,IAAI,EAAE,EAAE;QAChD,EAAE,QAAQ,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE;QACnC,EAAE,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE,MAAM,EAAE;QACtC,EAAE,QAAQ,EAAE,OAAO,EAAE,MAAM,EAAE,MAAM,EAAE;QACrC,EAAE,QAAQ,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE;KACpC;CACF,CAAC;AAEF;;;;;;;;;GASG;AACH,MAAa,UAAW,SAAQ,uBAAU;IAGxC,YAAY,OAAoB,EAAE,MAAc,EAAE,YAAoB;QACpE,KAAK,CAAC,CAAC,OAAO,CAAC,EAAE,OAAO,EAAE,MAAM,CAAC,CAAC;QAClC,IAAI,CAAC,YAAY,GAAG,YAAY,CAAC;IACnC,CAAC;IAED,OAAO;QACL,OAAO,YAAY,CAAC;IACtB,CAAC;IAED,SAAS;QACP,OAAO,MAAM,CAAC;IAChB,CAAC;IAEK,IAAI,CAAC,QAAgB;;YACzB,IAAI,CAAC;gBACH,MAAM,GAAG,GAAG,IAAI,iBAAM,CAAC,QAAQ,CAAC,CAAC;gBACjC,MAAM,OAAO,GAAG,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC;gBACtC,IAAI,CAAC,OAAO,EAAE,CAAC;oBACb,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,kCAAkC,QAAQ,EAAE,CAAC,CAAC;oBAC/D,OAAO,EAAE,MAAM,EAAE,EAAE,EAAE,QAAQ,EAAE,EAAE,IAAI,EAAE,YAAY,EAAE,EAAE,CAAC;gBAC1D,CAAC;gBACD,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,YAAY,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC;gBACvD,IAAI,QAAQ,CAAC,MAAM,KAAK,CAAC;oBAAE,OAAO,EAAE,MAAM,EAAE,EAAE,EAAE,QAAQ,EAAE,EAAE,IAAI,EAAE,YAAY,EAAE,EAAE,CAAC;gBAEnF,yEAAyE;gBACzE,MAAM,GAAG,GAAkB,EAAE,CAAC;gBAC9B,KAAK,MAAM,EAAE,IAAI,QAAQ,EAAE,CAAC;oBAC1B,MAAM,IAAI,GAAG,CAAC,EAAE,CAAC,KAAK,CAAC,CAAC,CAAC,KAAK,EAAE,CAAC,KAAK,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC;oBAC7D,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE;wBAAE,SAAS;oBAC3B,KAAK,MAAM,CAAC,IAAI,MAAM,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC;wBAAE,GAAG,CAAC,IAAI,mBAAM,CAAC,EAAG,CAAC;gBACrE,CAAC;gBACD,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE;oBACnB,CAAC,CAAC,KAAK,GAAG,CAAC,GAAG,CAAC,CAAC;oBAChB,CAAC,CAAC,WAAW,GAAG,GAAG,CAAC,MAAM,CAAC;gBAC7B,CAAC,CAAC,CAAC;gBACH,OAAO,EAAE,MAAM,EAAE,GAAG,EAAE,QAAQ,EAAE,EAAE,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,CAAC,MAAM,EAAE,EAAE,CAAC;YAChF,CAAC;YAAC,OAAO,CAAC,EAAE,CAAC;gBACX,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,8BAA8B,QAAQ,KAAK,CAAC,EAAE,CAAC,CAAC;gBACjE,OAAO,EAAE,MAAM,EAAE,EAAE,EAAE,QAAQ,EAAE,EAAE,IAAI,EAAE,YAAY,EAAE,EAAE,CAAC;YAC1D,CAAC;QACH,CAAC;KAAA;IAED,qDAAqD;IAC7C,WAAW,CAAC,GAAW;QAC7B,MAAM,SAAS,GAAG,GAAG,CAAC,QAAQ,CAAC,wBAAwB,CAAC,CAAC;QACzD,IAAI,CAAC,SAAS;YAAE,OAAO,SAAS,CAAC;QACjC,MAAM,CAAC,GAAG,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,UAAU,CAAC,SAAS,CAAC,EAAE,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC;QACrE,OAAO,CAAC,CAAC,UAAU,CAAC,CAAC,KAAK,EAAE,CAAC,IAAI,CAAC,WAAW,CAAC,IAAI,SAAS,CAAC;IAC9D,CAAC;IAED,0DAA0D;IAC5C,YAAY,CAAC,GAAW,EAAE,OAAe;;;YACrD,MAAM,CAAC,GAAG,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,UAAU,CAAC,OAAO,CAAC,EAAE,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC;YACnE,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAkB,CAAC;YAC3C,CAAC,CAAC,eAAe,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,EAAE,EAAE;gBAChC,MAAM,EAAE,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;gBAC5B,MAAM,IAAI,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;gBAChC,IAAI,EAAE,IAAI,IAAI;oBAAE,QAAQ,CAAC,GAAG,CAAC,EAAE,EAAE,IAAI,CAAC,CAAC;YACzC,CAAC,CAAC,CAAC;YACH,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;YAErE,MAAM,QAAQ,GAAuC,EAAE,CAAC;YACxD,MAAM,KAAK,GAAG,CAAC,CAAC,eAAe,CAAC,CAAC,OAAO,EAAE,CAAC;YAC3C,KAAK,MAAM,EAAE,IAAI,KAAK,EAAE,CAAC;gBACvB,MAAM,IAAI,GAAG,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC;gBACrD,IAAI,CAAC,IAAI;oBAAE,SAAS;gBACpB,MAAM,SAAS,GAAG,MAAM,KAAK,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,MAAM,IAAI,IAAI,EAAE,CAAC;gBAC9D,MAAM,KAAK,GAAG,MAAA,GAAG,CAAC,QAAQ,CAAC,SAAS,CAAC,mCAAI,GAAG,CAAC,QAAQ,CAAC,kBAAkB,CAAC,SAAS,CAAC,CAAC,CAAC;gBACrF,IAAI,CAAC,KAAK;oBAAE,SAAS;gBACrB,MAAM,KAAK,GAAG,GAAG,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC;gBACpC,MAAM,EAAE,KAAK,EAAE,IAAI,EAAE,GAAG,MAAM,IAAI,CAAC,cAAc,CAAC,KAAK,CAAC,CAAC;gBACzD,IAAI,IAAI,CAAC,IAAI,EAAE;oBAAE,QAAQ,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC,CAAC;YAClD,CAAC;YACD,OAAO,QAAQ,CAAC;QAClB,CAAC;KAAA;IAEa,cAAc,CAAC,KAAa;;YACxC,MAAM,CAAC,GAAG,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;YAC9B,MAAM,KAAK,GAAG,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,KAAK,EAAE,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,CAAC,IAAI,EAAE,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,IAAI,SAAS,CAAC;YAC9F,MAAM,EAAE,OAAO,EAAE,GAAG,wDAAa,cAAc,GAAC,CAAC;YACjD,OAAO,EAAE,KAAK,EAAE,IAAI,EAAE,OAAO,CAAC,KAAK,EAAE,iBAAiB,CAAC,EAAE,CAAC;QAC5D,CAAC;KAAA;CACF;AArFD,gCAqFC"}
@@ -0,0 +1,90 @@
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.FileReader = void 0;
46
+ const path = __importStar(require("path"));
47
+ /**
48
+ * Abstract base class for file readers
49
+ * Each file type should implement its own reader
50
+ */
51
+ class FileReader {
52
+ constructor(supportedExtensions, chunker, logger) {
53
+ this.logger = logger;
54
+ this.chunker = chunker;
55
+ this.supportedExtensions = supportedExtensions.map(ext => ext.toLowerCase());
56
+ }
57
+ /**
58
+ * Check if this reader supports the given file
59
+ */
60
+ canRead(filePath) {
61
+ const ext = path.extname(filePath).toLowerCase();
62
+ const base = path.basename(filePath).toLowerCase();
63
+ // Exact match on the extension (".ts") or, for extensionless / named files
64
+ // (Makefile, Dockerfile, LICENSE), the full basename. Never a prefix match:
65
+ // `startsWith` plus an empty-string entry once let TextReader claim every
66
+ // file (so .mp3/.wav read as UTF-8 and the ASR path was unreachable).
67
+ return this.supportedExtensions.some(e => e !== '' && (ext === e || base === e));
68
+ }
69
+ /**
70
+ * Stable, clean source-adapter id stamped onto every fact's provenance
71
+ * (`Observation.sourceAdapter`) so origin is queryable (ECS source-tagging).
72
+ * Defaults to the reader name minus the `Reader` suffix; readers with a more
73
+ * meaningful id (e.g. the PDF engines → "pdf:mistral") override this.
74
+ */
75
+ adapterId() {
76
+ return this.getName().replace(/Reader$/, "").toLowerCase();
77
+ }
78
+ /**
79
+ * Validate file before reading
80
+ */
81
+ validateFile(filePath) {
82
+ return __awaiter(this, void 0, void 0, function* () {
83
+ if (!this.canRead(filePath)) {
84
+ throw new Error(`${this.getName()} does not support file extension: ${path.extname(filePath)}, supported extensions are ${this.supportedExtensions.join(', ')}`);
85
+ }
86
+ });
87
+ }
88
+ }
89
+ exports.FileReader = FileReader;
90
+ //# sourceMappingURL=FileReader.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"FileReader.js","sourceRoot":"","sources":["../../../../src/core/processor/readers/FileReader.ts"],"names":[],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AACA,2CAA6B;AA8B7B;;;GAGG;AACH,MAAsB,UAAU;IAK9B,YAAY,mBAA6B,EAAE,OAAoB,EAAE,MAAc;QAC7E,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC;QACrB,IAAI,CAAC,OAAO,GAAG,OAAO,CAAC;QACvB,IAAI,CAAC,mBAAmB,GAAG,mBAAmB,CAAC,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,WAAW,EAAE,CAAC,CAAC;IAC/E,CAAC;IAED;;OAEG;IACH,OAAO,CAAC,QAAgB;QACtB,MAAM,GAAG,GAAG,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC,WAAW,EAAE,CAAC;QACjD,MAAM,IAAI,GAAG,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC,WAAW,EAAE,CAAC;QACnD,2EAA2E;QAC3E,4EAA4E;QAC5E,0EAA0E;QAC1E,sEAAsE;QACtE,OAAO,IAAI,CAAC,mBAAmB,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,KAAK,EAAE,IAAI,CAAC,GAAG,KAAK,CAAC,IAAI,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC;IACnF,CAAC;IAYD;;;;;OAKG;IACH,SAAS;QACP,OAAO,IAAI,CAAC,OAAO,EAAE,CAAC,OAAO,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC,WAAW,EAAE,CAAC;IAC7D,CAAC;IAED;;OAEG;IACa,YAAY,CAAC,QAAgB;;YAC3C,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,EAAE,CAAC;gBAC5B,MAAM,IAAI,KAAK,CACb,GAAG,IAAI,CAAC,OAAO,EAAE,qCAAqC,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,8BAA8B,IAAI,CAAC,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAChJ,CAAC;YACJ,CAAC;QACH,CAAC;KAAA;CACF;AAtDD,gCAsDC"}
@@ -0,0 +1,49 @@
1
+ "use strict";
2
+ Object.defineProperty(exports, "__esModule", { value: true });
3
+ exports.FileReaderFactory = void 0;
4
+ /**
5
+ * Factory for creating appropriate file readers based on file type
6
+ */
7
+ class FileReaderFactory {
8
+ constructor(logger) {
9
+ this.logger = logger;
10
+ this.readers = [];
11
+ }
12
+ /**
13
+ * Get appropriate reader for a file
14
+ * @param filePath Path to the file
15
+ * @returns FileReader instance or null if no reader supports the file
16
+ */
17
+ getReader(filePath) {
18
+ for (const reader of this.readers) {
19
+ if (reader.canRead(filePath)) {
20
+ this.logger.debug(`Using ${reader.getName()} for file: ${filePath}`);
21
+ return reader;
22
+ }
23
+ }
24
+ this.logger.warn(`No reader found for file: ${filePath}`);
25
+ return null;
26
+ }
27
+ /**
28
+ * Register a custom reader
29
+ * @param reader Custom FileReader implementation
30
+ */
31
+ registerReader(reader) {
32
+ this.readers.push(reader);
33
+ this.logger.info(`Registered custom reader: ${reader.getName()}`);
34
+ }
35
+ /**
36
+ * Get all registered readers
37
+ */
38
+ getReaders() {
39
+ return [...this.readers];
40
+ }
41
+ /**
42
+ * Check if any reader supports the file
43
+ */
44
+ canRead(filePath) {
45
+ return this.getReader(filePath) !== null;
46
+ }
47
+ }
48
+ exports.FileReaderFactory = FileReaderFactory;
49
+ //# sourceMappingURL=FileReaderFactory.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"FileReaderFactory.js","sourceRoot":"","sources":["../../../../src/core/processor/readers/FileReaderFactory.ts"],"names":[],"mappings":";;;AAGA;;GAEG;AACH,MAAa,iBAAiB;IAG5B,YAAoB,MAAc;QAAd,WAAM,GAAN,MAAM,CAAQ;QAF1B,YAAO,GAAiB,EAAE,CAAC;IAEG,CAAC;IAEvC;;;;OAIG;IACH,SAAS,CAAC,QAAgB;QACxB,KAAK,MAAM,MAAM,IAAI,IAAI,CAAC,OAAO,EAAE,CAAC;YAClC,IAAI,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,EAAE,CAAC;gBAC7B,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,SAAS,MAAM,CAAC,OAAO,EAAE,cAAc,QAAQ,EAAE,CAAC,CAAC;gBACrE,OAAO,MAAM,CAAC;YAChB,CAAC;QACH,CAAC;QAED,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,6BAA6B,QAAQ,EAAE,CAAC,CAAC;QAC1D,OAAO,IAAI,CAAC;IACd,CAAC;IAED;;;OAGG;IACH,cAAc,CAAC,MAAkB;QAC/B,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QAC1B,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,6BAA6B,MAAM,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC;IACpE,CAAC;IAED;;OAEG;IACH,UAAU;QACR,OAAO,CAAC,GAAG,IAAI,CAAC,OAAO,CAAC,CAAC;IAC3B,CAAC;IAED;;OAEG;IACH,OAAO,CAAC,QAAgB;QACtB,OAAO,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,KAAK,IAAI,CAAC;IAC3C,CAAC;CACF;AA5CD,8CA4CC"}