@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.
- package/LICENSE +21 -0
- package/README.md +458 -0
- package/dist/__tests__/helpers.js +27 -0
- package/dist/__tests__/helpers.js.map +1 -0
- package/dist/cli/commands/export.command.js +99 -0
- package/dist/cli/commands/export.command.js.map +1 -0
- package/dist/cli/commands/index.js +22 -0
- package/dist/cli/commands/index.js.map +1 -0
- package/dist/cli/commands/inspectMerges.command.js +84 -0
- package/dist/cli/commands/inspectMerges.command.js.map +1 -0
- package/dist/cli/commands/metrics.command.js +196 -0
- package/dist/cli/commands/metrics.command.js.map +1 -0
- package/dist/cli/commands/process.command.js +82 -0
- package/dist/cli/commands/process.command.js.map +1 -0
- package/dist/cli/commands/watch.command.js +91 -0
- package/dist/cli/commands/watch.command.js.map +1 -0
- package/dist/cli/index.js +269 -0
- package/dist/cli/index.js.map +1 -0
- package/dist/cli/optionsToConfig.js +160 -0
- package/dist/cli/optionsToConfig.js.map +1 -0
- package/dist/config/index.js +59 -0
- package/dist/config/index.js.map +1 -0
- package/dist/config/legacyHints.js +113 -0
- package/dist/config/legacyHints.js.map +1 -0
- package/dist/config/schema.js +803 -0
- package/dist/config/schema.js.map +1 -0
- package/dist/config/ui.js +221 -0
- package/dist/config/ui.js.map +1 -0
- package/dist/core/DirectoryProcessor.js +725 -0
- package/dist/core/DirectoryProcessor.js.map +1 -0
- package/dist/core/adapters/IStructuredAdapter.js +3 -0
- package/dist/core/adapters/IStructuredAdapter.js.map +1 -0
- package/dist/core/adapters/SqliteAdapter.js +267 -0
- package/dist/core/adapters/SqliteAdapter.js.map +1 -0
- package/dist/core/adapters/StructuredAdapterRegistry.js +31 -0
- package/dist/core/adapters/StructuredAdapterRegistry.js.map +1 -0
- package/dist/core/adapters/index.js +20 -0
- package/dist/core/adapters/index.js.map +1 -0
- package/dist/core/checkpoint/CheckpointService.js +188 -0
- package/dist/core/checkpoint/CheckpointService.js.map +1 -0
- package/dist/core/checkpoint/index.js +18 -0
- package/dist/core/checkpoint/index.js.map +1 -0
- package/dist/core/corpus/CorpusAnalyzer.js +266 -0
- package/dist/core/corpus/CorpusAnalyzer.js.map +1 -0
- package/dist/core/corpus/CorpusProfileStore.js +92 -0
- package/dist/core/corpus/CorpusProfileStore.js.map +1 -0
- package/dist/core/corpus/index.js +21 -0
- package/dist/core/corpus/index.js.map +1 -0
- package/dist/core/corpus/normalizeGlossary.js +60 -0
- package/dist/core/corpus/normalizeGlossary.js.map +1 -0
- package/dist/core/corpus/relPath.js +52 -0
- package/dist/core/corpus/relPath.js.map +1 -0
- package/dist/core/corpus/termFrequency.js +86 -0
- package/dist/core/corpus/termFrequency.js.map +1 -0
- package/dist/core/cost/CostMeter.js +235 -0
- package/dist/core/cost/CostMeter.js.map +1 -0
- package/dist/core/cost/index.js +19 -0
- package/dist/core/cost/index.js.map +1 -0
- package/dist/core/cost/prices.js +38 -0
- package/dist/core/cost/prices.js.map +1 -0
- package/dist/core/cv/ObjectDetectionService.js +119 -0
- package/dist/core/cv/ObjectDetectionService.js.map +1 -0
- package/dist/core/di/ContainerFactory.js +670 -0
- package/dist/core/di/ContainerFactory.js.map +1 -0
- package/dist/core/di/DIContainer.js +103 -0
- package/dist/core/di/DIContainer.js.map +1 -0
- package/dist/core/di/index.js +19 -0
- package/dist/core/di/index.js.map +1 -0
- package/dist/core/errors/CustomErrors.js +342 -0
- package/dist/core/errors/CustomErrors.js.map +1 -0
- package/dist/core/errors/index.js +18 -0
- package/dist/core/errors/index.js.map +1 -0
- package/dist/core/export/KnowledgeGraphExportService.js +56 -0
- package/dist/core/export/KnowledgeGraphExportService.js.map +1 -0
- package/dist/core/export/index.js +19 -0
- package/dist/core/export/index.js.map +1 -0
- package/dist/core/export/strategies/GraphitiExportStrategy.js +115 -0
- package/dist/core/export/strategies/GraphitiExportStrategy.js.map +1 -0
- package/dist/core/export/strategies/GraphvizDotExportStrategy.js +331 -0
- package/dist/core/export/strategies/GraphvizDotExportStrategy.js.map +1 -0
- package/dist/core/export/strategies/IExportStrategy.js +3 -0
- package/dist/core/export/strategies/IExportStrategy.js.map +1 -0
- package/dist/core/export/strategies/JsonExportStrategy.js +19 -0
- package/dist/core/export/strategies/JsonExportStrategy.js.map +1 -0
- package/dist/core/export/strategies/JsonlExportStrategy.js +69 -0
- package/dist/core/export/strategies/JsonlExportStrategy.js.map +1 -0
- package/dist/core/export/strategies/KblamExportStrategy.js +36 -0
- package/dist/core/export/strategies/KblamExportStrategy.js.map +1 -0
- package/dist/core/export/strategies/LoraExportStrategy.js +46 -0
- package/dist/core/export/strategies/LoraExportStrategy.js.map +1 -0
- package/dist/core/export/strategies/McpExportStrategy.js +67 -0
- package/dist/core/export/strategies/McpExportStrategy.js.map +1 -0
- package/dist/core/export/strategies/index.js +25 -0
- package/dist/core/export/strategies/index.js.map +1 -0
- package/dist/core/export/strategies/kbTriples.js +60 -0
- package/dist/core/export/strategies/kbTriples.js.map +1 -0
- package/dist/core/index.js +22 -0
- package/dist/core/index.js.map +1 -0
- package/dist/core/knowledge/KnowledgeGraphBuilder.js +627 -0
- package/dist/core/knowledge/KnowledgeGraphBuilder.js.map +1 -0
- package/dist/core/knowledge/MergeRecord.js +3 -0
- package/dist/core/knowledge/MergeRecord.js.map +1 -0
- package/dist/core/knowledge/canon/Canonicalizer.js +414 -0
- package/dist/core/knowledge/canon/Canonicalizer.js.map +1 -0
- package/dist/core/knowledge/canon/index.js +18 -0
- package/dist/core/knowledge/canon/index.js.map +1 -0
- package/dist/core/knowledge/contradiction/HeuristicContradictionChecker.js +92 -0
- package/dist/core/knowledge/contradiction/HeuristicContradictionChecker.js.map +1 -0
- package/dist/core/knowledge/contradiction/LlmContradictionChecker.js +52 -0
- package/dist/core/knowledge/contradiction/LlmContradictionChecker.js.map +1 -0
- package/dist/core/knowledge/contradiction/index.js +19 -0
- package/dist/core/knowledge/contradiction/index.js.map +1 -0
- package/dist/core/knowledge/grounding/KeywordGroundingChecker.js +33 -0
- package/dist/core/knowledge/grounding/KeywordGroundingChecker.js.map +1 -0
- package/dist/core/knowledge/grounding/MiniCheckGroundingChecker.js +82 -0
- package/dist/core/knowledge/grounding/MiniCheckGroundingChecker.js.map +1 -0
- package/dist/core/knowledge/grounding/index.js +20 -0
- package/dist/core/knowledge/grounding/index.js.map +1 -0
- package/dist/core/knowledge/grounding/verbalize.js +38 -0
- package/dist/core/knowledge/grounding/verbalize.js.map +1 -0
- package/dist/core/knowledge/images/imageMetaGraph.js +136 -0
- package/dist/core/knowledge/images/imageMetaGraph.js.map +1 -0
- package/dist/core/knowledge/index.js +20 -0
- package/dist/core/knowledge/index.js.map +1 -0
- package/dist/core/knowledge/merging/KnowledgeMerger.js +624 -0
- package/dist/core/knowledge/merging/KnowledgeMerger.js.map +1 -0
- package/dist/core/knowledge/references/ReferenceResolver.js +184 -0
- package/dist/core/knowledge/references/ReferenceResolver.js.map +1 -0
- package/dist/core/knowledge/references/citations/CitationEvidenceProcessor.js +401 -0
- package/dist/core/knowledge/references/citations/CitationEvidenceProcessor.js.map +1 -0
- package/dist/core/knowledge/references/citations/CitationResolver.js +95 -0
- package/dist/core/knowledge/references/citations/CitationResolver.js.map +1 -0
- package/dist/core/knowledge/references/citations/GrobidClient.js +143 -0
- package/dist/core/knowledge/references/citations/GrobidClient.js.map +1 -0
- package/dist/core/knowledge/references/citations/TitleIdResolver.js +101 -0
- package/dist/core/knowledge/references/citations/TitleIdResolver.js.map +1 -0
- package/dist/core/knowledge/references/web/FetchCacheService.js +114 -0
- package/dist/core/knowledge/references/web/FetchCacheService.js.map +1 -0
- package/dist/core/knowledge/references/web/GatedFetcher.js +228 -0
- package/dist/core/knowledge/references/web/GatedFetcher.js.map +1 -0
- package/dist/core/knowledge/references/web/WebReferenceProcessor.js +164 -0
- package/dist/core/knowledge/references/web/WebReferenceProcessor.js.map +1 -0
- package/dist/core/knowledge/search/KnowledgeGraphSearch.js +261 -0
- package/dist/core/knowledge/search/KnowledgeGraphSearch.js.map +1 -0
- package/dist/core/knowledge/vocabulary.js +162 -0
- package/dist/core/knowledge/vocabulary.js.map +1 -0
- package/dist/core/llm/EmbeddingService.js +113 -0
- package/dist/core/llm/EmbeddingService.js.map +1 -0
- package/dist/core/llm/OllamaService.js +146 -0
- package/dist/core/llm/OllamaService.js.map +1 -0
- package/dist/core/llm/OpenAICompatibleService.js +190 -0
- package/dist/core/llm/OpenAICompatibleService.js.map +1 -0
- package/dist/core/llm/OpenAIEmbeddingService.js +129 -0
- package/dist/core/llm/OpenAIEmbeddingService.js.map +1 -0
- package/dist/core/llm/embeddingUtils.js +25 -0
- package/dist/core/llm/embeddingUtils.js.map +1 -0
- package/dist/core/llm/index.js +23 -0
- package/dist/core/llm/index.js.map +1 -0
- package/dist/core/llm/prompts/PromptManager.js +388 -0
- package/dist/core/llm/prompts/PromptManager.js.map +1 -0
- package/dist/core/llm/prompts/PromptTemplateEngine.js +257 -0
- package/dist/core/llm/prompts/PromptTemplateEngine.js.map +1 -0
- package/dist/core/llm/prompts/templates/partials/examples/EXAMPLE_STYLE_GUIDE.md +84 -0
- package/dist/core/llm/prompts/templates/partials/examples/article.md +187 -0
- package/dist/core/llm/prompts/templates/partials/examples/code.md +229 -0
- package/dist/core/llm/prompts/templates/partials/examples/communication.md +205 -0
- package/dist/core/llm/prompts/templates/partials/examples/documentation.md +262 -0
- package/dist/core/llm/prompts/templates/partials/examples/financial.md +157 -0
- package/dist/core/llm/prompts/templates/partials/examples/legal.md +153 -0
- package/dist/core/llm/prompts/templates/partials/examples/logs.md +127 -0
- package/dist/core/llm/prompts/templates/partials/examples/medical.md +218 -0
- package/dist/core/llm/prompts/templates/partials/examples/notes.md +201 -0
- package/dist/core/llm/prompts/templates/partials/examples/research.md +208 -0
- package/dist/core/llm/prompts/templates/partials/examples/tabular.md +178 -0
- package/dist/core/llm/prompts/templates/partials/examples/transcript.md +204 -0
- package/dist/core/llm/prompts/templates/partials/retrieved-context.hbs +18 -0
- package/dist/core/llm/prompts/templates/v1/system.hbs +371 -0
- package/dist/core/llm/prompts/templates/v1/user.hbs +20 -0
- package/dist/core/llm/prompts/templates/v2/system.hbs +573 -0
- package/dist/core/llm/prompts/templates/v2/user.hbs +20 -0
- package/dist/core/llm/prompts/templates/v3/system.hbs +861 -0
- package/dist/core/llm/prompts/templates/v3/user.hbs +16 -0
- package/dist/core/llm/prompts/templates/v4/system.hbs +800 -0
- package/dist/core/llm/prompts/templates/v4/user.hbs +40 -0
- package/dist/core/llm/prompts/templates/v4.5/system.hbs +71 -0
- package/dist/core/llm/prompts/templates/v4.5/user.hbs +46 -0
- package/dist/core/llm/prompts/templates/v5/glossary/system.hbs +40 -0
- package/dist/core/llm/prompts/templates/v5/glossary/user.hbs +11 -0
- package/dist/core/llm/prompts/templates/v5/system.hbs +163 -0
- package/dist/core/llm/prompts/templates/v5/user.hbs +55 -0
- package/dist/core/pipeline/GroundingTransform.js +52 -0
- package/dist/core/pipeline/GroundingTransform.js.map +1 -0
- package/dist/core/pipeline/PipelineRunner.js +51 -0
- package/dist/core/pipeline/PipelineRunner.js.map +1 -0
- package/dist/core/pipeline/RelationFilterTransform.js +72 -0
- package/dist/core/pipeline/RelationFilterTransform.js.map +1 -0
- package/dist/core/pipeline/index.js +20 -0
- package/dist/core/pipeline/index.js.map +1 -0
- package/dist/core/processor/FileProcessor.js +184 -0
- package/dist/core/processor/FileProcessor.js.map +1 -0
- package/dist/core/processor/ProcessedRegistry.js +38 -0
- package/dist/core/processor/ProcessedRegistry.js.map +1 -0
- package/dist/core/processor/ast/AstSeedService.js +0 -0
- package/dist/core/processor/ast/AstSeedService.js.map +1 -0
- package/dist/core/processor/ast/AstSymbolStore.js +110 -0
- package/dist/core/processor/ast/AstSymbolStore.js.map +1 -0
- package/dist/core/processor/ast/index.js +19 -0
- package/dist/core/processor/ast/index.js.map +1 -0
- package/dist/core/processor/chunking/TextChunker.js +98 -0
- package/dist/core/processor/chunking/TextChunker.js.map +1 -0
- package/dist/core/processor/chunking/index.js +18 -0
- package/dist/core/processor/chunking/index.js.map +1 -0
- package/dist/core/processor/classifier/CONTENT_CLASSES.js +294 -0
- package/dist/core/processor/classifier/CONTENT_CLASSES.js.map +1 -0
- package/dist/core/processor/classifier/CascadeContentClassifier.js +107 -0
- package/dist/core/processor/classifier/CascadeContentClassifier.js.map +1 -0
- package/dist/core/processor/classifier/HeuristicContentClassifier.js +113 -0
- package/dist/core/processor/classifier/HeuristicContentClassifier.js.map +1 -0
- package/dist/core/processor/classifier/IContentTypeClassifier.js +3 -0
- package/dist/core/processor/classifier/IContentTypeClassifier.js.map +1 -0
- package/dist/core/processor/classifier/LlmContentClassifier.js +107 -0
- package/dist/core/processor/classifier/LlmContentClassifier.js.map +1 -0
- package/dist/core/processor/classifier/NER_DOMAIN_EXAMPLES.js +498 -0
- package/dist/core/processor/classifier/NER_DOMAIN_EXAMPLES.js.map +1 -0
- package/dist/core/processor/classifier/index.js +21 -0
- package/dist/core/processor/classifier/index.js.map +1 -0
- package/dist/core/processor/classifier/mergeClassifications.js +32 -0
- package/dist/core/processor/classifier/mergeClassifications.js.map +1 -0
- package/dist/core/processor/index.js +20 -0
- package/dist/core/processor/index.js.map +1 -0
- package/dist/core/processor/readers/AudioReader.js +462 -0
- package/dist/core/processor/readers/AudioReader.js.map +1 -0
- package/dist/core/processor/readers/BinaryReader.js +90 -0
- package/dist/core/processor/readers/BinaryReader.js.map +1 -0
- package/dist/core/processor/readers/ChandraPdfReader.js +187 -0
- package/dist/core/processor/readers/ChandraPdfReader.js.map +1 -0
- package/dist/core/processor/readers/ChatExportReader.js +365 -0
- package/dist/core/processor/readers/ChatExportReader.js.map +1 -0
- package/dist/core/processor/readers/DoclingReader.js +445 -0
- package/dist/core/processor/readers/DoclingReader.js.map +1 -0
- package/dist/core/processor/readers/EmailReader.js +259 -0
- package/dist/core/processor/readers/EmailReader.js.map +1 -0
- package/dist/core/processor/readers/EpubReader.js +175 -0
- package/dist/core/processor/readers/EpubReader.js.map +1 -0
- package/dist/core/processor/readers/FileReader.js +90 -0
- package/dist/core/processor/readers/FileReader.js.map +1 -0
- package/dist/core/processor/readers/FileReaderFactory.js +49 -0
- package/dist/core/processor/readers/FileReaderFactory.js.map +1 -0
- package/dist/core/processor/readers/HtmlReader.js +371 -0
- package/dist/core/processor/readers/HtmlReader.js.map +1 -0
- package/dist/core/processor/readers/ImageReader.js +162 -0
- package/dist/core/processor/readers/ImageReader.js.map +1 -0
- package/dist/core/processor/readers/JsonFileReader.js +232 -0
- package/dist/core/processor/readers/JsonFileReader.js.map +1 -0
- package/dist/core/processor/readers/JupyterReader.js +178 -0
- package/dist/core/processor/readers/JupyterReader.js.map +1 -0
- package/dist/core/processor/readers/LatexReader.js +176 -0
- package/dist/core/processor/readers/LatexReader.js.map +1 -0
- package/dist/core/processor/readers/MarkdownReader.js +289 -0
- package/dist/core/processor/readers/MarkdownReader.js.map +1 -0
- package/dist/core/processor/readers/MarkerPdfReader.js +193 -0
- package/dist/core/processor/readers/MarkerPdfReader.js.map +1 -0
- package/dist/core/processor/readers/MistralOcrReader.js +198 -0
- package/dist/core/processor/readers/MistralOcrReader.js.map +1 -0
- package/dist/core/processor/readers/OfficeReader.js +174 -0
- package/dist/core/processor/readers/OfficeReader.js.map +1 -0
- package/dist/core/processor/readers/PdfReader.js +116 -0
- package/dist/core/processor/readers/PdfReader.js.map +1 -0
- package/dist/core/processor/readers/RtfReader.js +107 -0
- package/dist/core/processor/readers/RtfReader.js.map +1 -0
- package/dist/core/processor/readers/SubtitleReader.js +145 -0
- package/dist/core/processor/readers/SubtitleReader.js.map +1 -0
- package/dist/core/processor/readers/TesseractPdfReader.js +183 -0
- package/dist/core/processor/readers/TesseractPdfReader.js.map +1 -0
- package/dist/core/processor/readers/TextReader.js +129 -0
- package/dist/core/processor/readers/TextReader.js.map +1 -0
- package/dist/core/processor/readers/TranscriptReader.js +234 -0
- package/dist/core/processor/readers/TranscriptReader.js.map +1 -0
- package/dist/core/processor/readers/image/imageMetadata.js +155 -0
- package/dist/core/processor/readers/image/imageMetadata.js.map +1 -0
- package/dist/core/processor/readers/index.js +41 -0
- package/dist/core/processor/readers/index.js.map +1 -0
- package/dist/core/processor/readers/referenceExtraction.js +198 -0
- package/dist/core/processor/readers/referenceExtraction.js.map +1 -0
- package/dist/core/processor/readers/stripReferences.js +59 -0
- package/dist/core/processor/readers/stripReferences.js.map +1 -0
- package/dist/core/processor/readers/transcript/turnPacking.js +81 -0
- package/dist/core/processor/readers/transcript/turnPacking.js.map +1 -0
- package/dist/core/progress/NdjsonProgressEmitter.js +30 -0
- package/dist/core/progress/NdjsonProgressEmitter.js.map +1 -0
- package/dist/core/progress/NoopProgressEmitter.js +15 -0
- package/dist/core/progress/NoopProgressEmitter.js.map +1 -0
- package/dist/core/progress/index.js +19 -0
- package/dist/core/progress/index.js.map +1 -0
- package/dist/core/trace/TraceWriter.js +100 -0
- package/dist/core/trace/TraceWriter.js.map +1 -0
- package/dist/core/trace/events.js +13 -0
- package/dist/core/trace/events.js.map +1 -0
- package/dist/core/trace/index.js +20 -0
- package/dist/core/trace/index.js.map +1 -0
- package/dist/core/trace/lineage.js +97 -0
- package/dist/core/trace/lineage.js.map +1 -0
- package/dist/evaluation/BenchmarkRunner.js +171 -0
- package/dist/evaluation/BenchmarkRunner.js.map +1 -0
- package/dist/evaluation/classifier/ClassifierAccuracy.js +185 -0
- package/dist/evaluation/classifier/ClassifierAccuracy.js.map +1 -0
- package/dist/evaluation/classifier/labeledSamples.js +379 -0
- package/dist/evaluation/classifier/labeledSamples.js.map +1 -0
- package/dist/evaluation/compare/goldCompare.js +126 -0
- package/dist/evaluation/compare/goldCompare.js.map +1 -0
- package/dist/evaluation/crossre/compareScoring.js +30 -0
- package/dist/evaluation/crossre/compareScoring.js.map +1 -0
- package/dist/evaluation/datasets/CrossREDataset.js +170 -0
- package/dist/evaluation/datasets/CrossREDataset.js.map +1 -0
- package/dist/evaluation/datasets/IDataset.js +3 -0
- package/dist/evaluation/datasets/IDataset.js.map +1 -0
- package/dist/evaluation/datasets/RebelDataset.js +117 -0
- package/dist/evaluation/datasets/RebelDataset.js.map +1 -0
- package/dist/evaluation/datasets/RedocredDataset.js +218 -0
- package/dist/evaluation/datasets/RedocredDataset.js.map +1 -0
- package/dist/evaluation/datasets/SemEval2010Dataset.js +150 -0
- package/dist/evaluation/datasets/SemEval2010Dataset.js.map +1 -0
- package/dist/evaluation/index.js +33 -0
- package/dist/evaluation/index.js.map +1 -0
- package/dist/evaluation/matching/ExactMatcher.js +75 -0
- package/dist/evaluation/matching/ExactMatcher.js.map +1 -0
- package/dist/evaluation/matching/SemanticMatcher.js +143 -0
- package/dist/evaluation/matching/SemanticMatcher.js.map +1 -0
- package/dist/evaluation/metrics/TripleMetrics.js +64 -0
- package/dist/evaluation/metrics/TripleMetrics.js.map +1 -0
- package/dist/evaluation/mine/MineCheckpoint.js +114 -0
- package/dist/evaluation/mine/MineCheckpoint.js.map +1 -0
- package/dist/evaluation/mine/MineDataset.js +208 -0
- package/dist/evaluation/mine/MineDataset.js.map +1 -0
- package/dist/evaluation/mine/MineReporter.js +98 -0
- package/dist/evaluation/mine/MineReporter.js.map +1 -0
- package/dist/evaluation/mine/MineRunner.js +148 -0
- package/dist/evaluation/mine/MineRunner.js.map +1 -0
- package/dist/evaluation/mine/MineScorer.js +127 -0
- package/dist/evaluation/mine/MineScorer.js.map +1 -0
- package/dist/evaluation/mine/types.js +12 -0
- package/dist/evaluation/mine/types.js.map +1 -0
- package/dist/evaluation/reporters/ConsoleReporter.js +55 -0
- package/dist/evaluation/reporters/ConsoleReporter.js.map +1 -0
- package/dist/evaluation/reporters/JsonReporter.js +50 -0
- package/dist/evaluation/reporters/JsonReporter.js.map +1 -0
- package/dist/index.js +28 -0
- package/dist/index.js.map +1 -0
- package/dist/quality/CompositeScore.js +61 -0
- package/dist/quality/CompositeScore.js.map +1 -0
- package/dist/quality/ConsistencyMetrics.js +70 -0
- package/dist/quality/ConsistencyMetrics.js.map +1 -0
- package/dist/quality/FactualMetrics.js +76 -0
- package/dist/quality/FactualMetrics.js.map +1 -0
- package/dist/quality/GraphHealthMetrics.js +68 -0
- package/dist/quality/GraphHealthMetrics.js.map +1 -0
- package/dist/quality/SemanticMetrics.js +102 -0
- package/dist/quality/SemanticMetrics.js.map +1 -0
- package/dist/quality/StructuralMetrics.js +60 -0
- package/dist/quality/StructuralMetrics.js.map +1 -0
- package/dist/quality/index.js +23 -0
- package/dist/quality/index.js.map +1 -0
- package/dist/shared/index.js +20 -0
- package/dist/shared/index.js.map +1 -0
- package/dist/shared/logger/Logger.js +3 -0
- package/dist/shared/logger/Logger.js.map +1 -0
- package/dist/shared/logger/LoggerFactory.js +75 -0
- package/dist/shared/logger/LoggerFactory.js.map +1 -0
- package/dist/shared/logger/index.js +19 -0
- package/dist/shared/logger/index.js.map +1 -0
- package/dist/shared/shutdown.js +30 -0
- package/dist/shared/shutdown.js.map +1 -0
- package/dist/shared/utils/agglomerativeCluster.js +269 -0
- package/dist/shared/utils/agglomerativeCluster.js.map +1 -0
- package/dist/shared/utils/astSymbols.js +69 -0
- package/dist/shared/utils/astSymbols.js.map +1 -0
- package/dist/shared/utils/cosineSimilarity.js +18 -0
- package/dist/shared/utils/cosineSimilarity.js.map +1 -0
- package/dist/shared/utils/directoryTree.js +184 -0
- package/dist/shared/utils/directoryTree.js.map +1 -0
- package/dist/shared/utils/documentOutline.js +74 -0
- package/dist/shared/utils/documentOutline.js.map +1 -0
- package/dist/shared/utils/index.js +24 -0
- package/dist/shared/utils/index.js.map +1 -0
- package/dist/shared/utils/jaroWinklerSimilarity.js +60 -0
- package/dist/shared/utils/jaroWinklerSimilarity.js.map +1 -0
- package/dist/shared/utils/parseJsonLenient.js +27 -0
- package/dist/shared/utils/parseJsonLenient.js.map +1 -0
- package/dist/shared/utils/readConfig.js +42 -0
- package/dist/shared/utils/readConfig.js.map +1 -0
- package/dist/shared/utils/readRtf.js +216 -0
- package/dist/shared/utils/readRtf.js.map +1 -0
- package/dist/shared/utils/softmax.js +26 -0
- package/dist/shared/utils/softmax.js.map +1 -0
- package/dist/types/ContentClass.js +3 -0
- package/dist/types/ContentClass.js.map +1 -0
- package/dist/types/CorpusProfile.js +3 -0
- package/dist/types/CorpusProfile.js.map +1 -0
- package/dist/types/IContradictionChecker.js +3 -0
- package/dist/types/IContradictionChecker.js.map +1 -0
- package/dist/types/ICorpusAnalyzer.js +3 -0
- package/dist/types/ICorpusAnalyzer.js.map +1 -0
- package/dist/types/IDirectoryProcessor.js +3 -0
- package/dist/types/IDirectoryProcessor.js.map +1 -0
- package/dist/types/IEmbeddingProvider.js +3 -0
- package/dist/types/IEmbeddingProvider.js.map +1 -0
- package/dist/types/IEmbeddingService.js +6 -0
- package/dist/types/IEmbeddingService.js.map +1 -0
- package/dist/types/IFileProcessor.js +3 -0
- package/dist/types/IFileProcessor.js.map +1 -0
- package/dist/types/IGroundingChecker.js +3 -0
- package/dist/types/IGroundingChecker.js.map +1 -0
- package/dist/types/IKnowledgeGraphBuilder.js +3 -0
- package/dist/types/IKnowledgeGraphBuilder.js.map +1 -0
- package/dist/types/IKnowledgeGraphExporter.js +3 -0
- package/dist/types/IKnowledgeGraphExporter.js.map +1 -0
- package/dist/types/IKnowledgeGraphMerger.js +3 -0
- package/dist/types/IKnowledgeGraphMerger.js.map +1 -0
- package/dist/types/IKnowledgeGraphSearch.js +3 -0
- package/dist/types/IKnowledgeGraphSearch.js.map +1 -0
- package/dist/types/ILLMProvider.js +3 -0
- package/dist/types/ILLMProvider.js.map +1 -0
- package/dist/types/ILLMService.js +3 -0
- package/dist/types/ILLMService.js.map +1 -0
- package/dist/types/IObjectDetector.js +3 -0
- package/dist/types/IObjectDetector.js.map +1 -0
- package/dist/types/IProcessingService.js +3 -0
- package/dist/types/IProcessingService.js.map +1 -0
- package/dist/types/IProgressEmitter.js +3 -0
- package/dist/types/IProgressEmitter.js.map +1 -0
- package/dist/types/IPromptManager.js +3 -0
- package/dist/types/IPromptManager.js.map +1 -0
- package/dist/types/KnowledgeGraph.js +3 -0
- package/dist/types/KnowledgeGraph.js.map +1 -0
- package/dist/types/MCPKnowledgeGraph.js +3 -0
- package/dist/types/MCPKnowledgeGraph.js.map +1 -0
- package/dist/types/Observation.js +21 -0
- package/dist/types/Observation.js.map +1 -0
- package/dist/types/ProcessingOptions.js +3 -0
- package/dist/types/ProcessingOptions.js.map +1 -0
- package/dist/types/index.js +40 -0
- package/dist/types/index.js.map +1 -0
- package/package.json +122 -0
|
@@ -0,0 +1,145 @@
|
|
|
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.SubtitleReader = void 0;
|
|
46
|
+
const fs = __importStar(require("fs"));
|
|
47
|
+
const FileReader_1 = require("./FileReader");
|
|
48
|
+
const turnPacking_1 = require("./transcript/turnPacking");
|
|
49
|
+
/**
|
|
50
|
+
* Reads subtitle/caption files (`.srt`, `.vtt`) into clean, **denoised** chunks:
|
|
51
|
+
* the index numbers, timecodes, and styling tags that would otherwise pollute
|
|
52
|
+
* extraction are stripped, leaving the spoken text. When the captions carry
|
|
53
|
+
* speaker info (VTT `<v Speaker>` voice tags) the cues become attributed
|
|
54
|
+
* `Turn`s via the shared `packTurns()` (same path as transcripts/email/chat);
|
|
55
|
+
* otherwise the caption text is concatenated and size-chunked as prose.
|
|
56
|
+
* `sourceAdapter:"subtitle"` is stamped centrally from `adapterId()`.
|
|
57
|
+
*
|
|
58
|
+
* NB (deviation from the data-sinks brief): a cue offset like `00:01:23` is a
|
|
59
|
+
* **media position, not wall-clock valid-time**, so it is NOT written to
|
|
60
|
+
* `occurredAt`/`validAt` (that would fabricate bitemporal data). A cue-time
|
|
61
|
+
* `locator` (à la PDF `p.<n>`) is the right home and is deferred to v2.
|
|
62
|
+
*
|
|
63
|
+
* v1 deferred: cue-time locator, multi-language tracks, karaoke timing tags.
|
|
64
|
+
*/
|
|
65
|
+
class SubtitleReader extends FileReader_1.FileReader {
|
|
66
|
+
constructor(chunker, logger, maxChunkSize) {
|
|
67
|
+
super([".srt", ".vtt"], chunker, logger);
|
|
68
|
+
this.maxChunkSize = maxChunkSize;
|
|
69
|
+
}
|
|
70
|
+
getName() {
|
|
71
|
+
return "SubtitleReader";
|
|
72
|
+
}
|
|
73
|
+
adapterId() {
|
|
74
|
+
return "subtitle";
|
|
75
|
+
}
|
|
76
|
+
read(filePath) {
|
|
77
|
+
return __awaiter(this, void 0, void 0, function* () {
|
|
78
|
+
const raw = yield fs.promises.readFile(filePath, "utf-8");
|
|
79
|
+
try {
|
|
80
|
+
const cues = this.parseCues(raw);
|
|
81
|
+
if (cues.length === 0)
|
|
82
|
+
return this.plainFallback(raw);
|
|
83
|
+
const format = filePath.toLowerCase().endsWith(".vtt") ? "vtt" : "srt";
|
|
84
|
+
// Speaker-attributed (VTT <v>) → the turn path; else denoised prose.
|
|
85
|
+
if (cues.some((c) => c.speaker)) {
|
|
86
|
+
const turns = cues.map((c) => ({ speaker: c.speaker || "speaker", text: c.text, conversation: 0 }));
|
|
87
|
+
const chunks = yield (0, turnPacking_1.packTurns)(turns, filePath, this.maxChunkSize, this.chunker);
|
|
88
|
+
return { chunks, metadata: { type: "subtitle", format, cues: cues.length } };
|
|
89
|
+
}
|
|
90
|
+
const parts = yield this.chunker.chunk(cues.map((c) => c.text).join(" "));
|
|
91
|
+
return { chunks: parts.map((p) => (Object.assign({}, p))), metadata: { type: "subtitle", format, cues: cues.length } };
|
|
92
|
+
}
|
|
93
|
+
catch (e) {
|
|
94
|
+
this.logger.warn(`SubtitleReader could not parse ${filePath}; falling back to plain chunking: ${e}`);
|
|
95
|
+
return this.plainFallback(raw);
|
|
96
|
+
}
|
|
97
|
+
});
|
|
98
|
+
}
|
|
99
|
+
/** SRT blocks and VTT cues → cleaned cue text (+ speaker), consecutive dups dropped. */
|
|
100
|
+
parseCues(raw) {
|
|
101
|
+
const blocks = raw.replace(/^/, "").split(/\r?\n\r?\n+/);
|
|
102
|
+
const cues = [];
|
|
103
|
+
let prev = "";
|
|
104
|
+
for (const block of blocks) {
|
|
105
|
+
const trimmed = block.trim();
|
|
106
|
+
if (!trimmed || /^WEBVTT/.test(trimmed) || /^(NOTE|STYLE|REGION)\b/.test(trimmed))
|
|
107
|
+
continue;
|
|
108
|
+
const lines = trimmed.split(/\r?\n/);
|
|
109
|
+
const tcIdx = lines.findIndex((l) => l.includes("-->"));
|
|
110
|
+
if (tcIdx === -1)
|
|
111
|
+
continue; // header / stray block, not a cue
|
|
112
|
+
const textLines = lines.slice(tcIdx + 1);
|
|
113
|
+
if (textLines.length === 0)
|
|
114
|
+
continue;
|
|
115
|
+
const { text, speaker } = this.cleanCue(textLines.join("\n"));
|
|
116
|
+
if (!text || text === prev)
|
|
117
|
+
continue; // drop empties + rolling-caption dups
|
|
118
|
+
prev = text;
|
|
119
|
+
cues.push({ text, speaker });
|
|
120
|
+
}
|
|
121
|
+
return cues;
|
|
122
|
+
}
|
|
123
|
+
/** Capture a VTT `<v Speaker>` voice tag, then strip all caption markup. */
|
|
124
|
+
cleanCue(raw) {
|
|
125
|
+
const v = raw.match(/<v(?:\.[^ >]+)?\s+([^>]+)>/i);
|
|
126
|
+
const speaker = v ? v[1].trim() : undefined;
|
|
127
|
+
const text = raw
|
|
128
|
+
.replace(/<[^>]+>/g, "") // <v>, <i>, <b>, <c.class>, inline <00:00:01.000> timestamps
|
|
129
|
+
.replace(/ /gi, " ")
|
|
130
|
+
.replace(/&/gi, "&")
|
|
131
|
+
.replace(/</gi, "<")
|
|
132
|
+
.replace(/>/gi, ">")
|
|
133
|
+
.replace(/[ \t]+/g, " ")
|
|
134
|
+
.trim();
|
|
135
|
+
return { text, speaker };
|
|
136
|
+
}
|
|
137
|
+
plainFallback(raw) {
|
|
138
|
+
return __awaiter(this, void 0, void 0, function* () {
|
|
139
|
+
const parts = yield this.chunker.chunk(raw);
|
|
140
|
+
return { chunks: parts.map((p) => (Object.assign({}, p))), metadata: { type: "subtitle-fallback" } };
|
|
141
|
+
});
|
|
142
|
+
}
|
|
143
|
+
}
|
|
144
|
+
exports.SubtitleReader = SubtitleReader;
|
|
145
|
+
//# sourceMappingURL=SubtitleReader.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"SubtitleReader.js","sourceRoot":"","sources":["../../../../src/core/processor/readers/SubtitleReader.ts"],"names":[],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAAA,uCAAyB;AACzB,6CAA0D;AAG1D,0DAA2D;AAO3D;;;;;;;;;;;;;;;GAeG;AACH,MAAa,cAAe,SAAQ,uBAAU;IAG5C,YAAY,OAAoB,EAAE,MAAc,EAAE,YAAoB;QACpE,KAAK,CAAC,CAAC,MAAM,EAAE,MAAM,CAAC,EAAE,OAAO,EAAE,MAAM,CAAC,CAAC;QACzC,IAAI,CAAC,YAAY,GAAG,YAAY,CAAC;IACnC,CAAC;IAED,OAAO;QACL,OAAO,gBAAgB,CAAC;IAC1B,CAAC;IAED,SAAS;QACP,OAAO,UAAU,CAAC;IACpB,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,IAAI,GAAG,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC;gBACjC,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;oBAAE,OAAO,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC,CAAC;gBACtD,MAAM,MAAM,GAAG,QAAQ,CAAC,WAAW,EAAE,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC;gBAEvE,qEAAqE;gBACrE,IAAI,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,CAAC,EAAE,CAAC;oBAChC,MAAM,KAAK,GAAW,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,OAAO,EAAE,CAAC,CAAC,OAAO,IAAI,SAAS,EAAE,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,YAAY,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC;oBAC5G,MAAM,MAAM,GAAG,MAAM,IAAA,uBAAS,EAAC,KAAK,EAAE,QAAQ,EAAE,IAAI,CAAC,YAAY,EAAE,IAAI,CAAC,OAAO,CAAC,CAAC;oBACjF,OAAO,EAAE,MAAM,EAAE,QAAQ,EAAE,EAAE,IAAI,EAAE,UAAU,EAAE,MAAM,EAAE,IAAI,EAAE,IAAI,CAAC,MAAM,EAAE,EAAE,CAAC;gBAC/E,CAAC;gBACD,MAAM,KAAK,GAAG,MAAM,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;gBAC1E,OAAO,EAAE,MAAM,EAAE,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,mBAAM,CAAC,EAAG,CAAC,EAAE,QAAQ,EAAE,EAAE,IAAI,EAAE,UAAU,EAAE,MAAM,EAAE,IAAI,EAAE,IAAI,CAAC,MAAM,EAAE,EAAE,CAAC;YAC7G,CAAC;YAAC,OAAO,CAAC,EAAE,CAAC;gBACX,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,kCAAkC,QAAQ,qCAAqC,CAAC,EAAE,CAAC,CAAC;gBACrG,OAAO,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC,CAAC;YACjC,CAAC;QACH,CAAC;KAAA;IAED,wFAAwF;IAChF,SAAS,CAAC,GAAW;QAC3B,MAAM,MAAM,GAAG,GAAG,CAAC,OAAO,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,KAAK,CAAC,aAAa,CAAC,CAAC;QAC1D,MAAM,IAAI,GAAU,EAAE,CAAC;QACvB,IAAI,IAAI,GAAG,EAAE,CAAC;QACd,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;YAC3B,MAAM,OAAO,GAAG,KAAK,CAAC,IAAI,EAAE,CAAC;YAC7B,IAAI,CAAC,OAAO,IAAI,SAAS,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,wBAAwB,CAAC,IAAI,CAAC,OAAO,CAAC;gBAAE,SAAS;YAC5F,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC;YACrC,MAAM,KAAK,GAAG,KAAK,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC;YACxD,IAAI,KAAK,KAAK,CAAC,CAAC;gBAAE,SAAS,CAAC,kCAAkC;YAC9D,MAAM,SAAS,GAAG,KAAK,CAAC,KAAK,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC;YACzC,IAAI,SAAS,CAAC,MAAM,KAAK,CAAC;gBAAE,SAAS;YACrC,MAAM,EAAE,IAAI,EAAE,OAAO,EAAE,GAAG,IAAI,CAAC,QAAQ,CAAC,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC;YAC9D,IAAI,CAAC,IAAI,IAAI,IAAI,KAAK,IAAI;gBAAE,SAAS,CAAC,sCAAsC;YAC5E,IAAI,GAAG,IAAI,CAAC;YACZ,IAAI,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,CAAC,CAAC;QAC/B,CAAC;QACD,OAAO,IAAI,CAAC;IACd,CAAC;IAED,4EAA4E;IACpE,QAAQ,CAAC,GAAW;QAC1B,MAAM,CAAC,GAAG,GAAG,CAAC,KAAK,CAAC,6BAA6B,CAAC,CAAC;QACnD,MAAM,OAAO,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,SAAS,CAAC;QAC5C,MAAM,IAAI,GAAG,GAAG;aACb,OAAO,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC,6DAA6D;aACrF,OAAO,CAAC,UAAU,EAAE,GAAG,CAAC;aACxB,OAAO,CAAC,SAAS,EAAE,GAAG,CAAC;aACvB,OAAO,CAAC,QAAQ,EAAE,GAAG,CAAC;aACtB,OAAO,CAAC,QAAQ,EAAE,GAAG,CAAC;aACtB,OAAO,CAAC,SAAS,EAAE,GAAG,CAAC;aACvB,IAAI,EAAE,CAAC;QACV,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,CAAC;IAC3B,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,mBAAmB,EAAE,EAAE,CAAC;QAC3F,CAAC;KAAA;CACF;AA7ED,wCA6EC"}
|
|
@@ -0,0 +1,183 @@
|
|
|
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.TesseractPdfReader = void 0;
|
|
49
|
+
const FileReader_1 = require("./FileReader");
|
|
50
|
+
const promises_1 = __importDefault(require("fs/promises"));
|
|
51
|
+
/**
|
|
52
|
+
* PDF reader backed by Tesseract OCR — the **light-local** rung of the engine
|
|
53
|
+
* ladder, for hardware with no GPU and no VLM. Pure-JS/WASM, zero system binaries:
|
|
54
|
+
* `pdf-to-png-converter` (pdf.js, no native/OS deps) rasterizes each page to a PNG
|
|
55
|
+
* buffer in memory, then `tesseract.js` (WASM) OCRs it. Claims `.pdf` only.
|
|
56
|
+
*
|
|
57
|
+
* OCR is slow, so per-page text caches to a `<pdf>.tesseract.json` sidecar reused
|
|
58
|
+
* when newer than the source (re-runs never re-OCR). Any failure — render error,
|
|
59
|
+
* missing language data, empty output — **degrades to the injected pdf2json
|
|
60
|
+
* fallback** so a run never dies on one PDF and the default path stays portable.
|
|
61
|
+
* Each page is its own chunk with an ECS `locator: "p.<n>"` (PdfReader/Mistral
|
|
62
|
+
* pattern). The heavy deps are lazy-imported so a default (pdf2json) run never
|
|
63
|
+
* pays for pdfjs/WASM at startup.
|
|
64
|
+
*
|
|
65
|
+
* NB: tesseract.js fetches the language traineddata (~10-15 MB/lang) from its CDN
|
|
66
|
+
* on first use and caches it; point `langPath` at a local mirror for a fully
|
|
67
|
+
* offline weak-hardware deployment.
|
|
68
|
+
*/
|
|
69
|
+
class TesseractPdfReader extends FileReader_1.FileReader {
|
|
70
|
+
constructor(opts, fallback, chunker, logger, injectedDeps) {
|
|
71
|
+
super([".pdf"], chunker, logger);
|
|
72
|
+
this.opts = opts;
|
|
73
|
+
this.fallback = fallback;
|
|
74
|
+
this.injectedDeps = injectedDeps;
|
|
75
|
+
}
|
|
76
|
+
getName() {
|
|
77
|
+
return "TesseractPdfReader";
|
|
78
|
+
}
|
|
79
|
+
adapterId() {
|
|
80
|
+
return "pdf:tesseract";
|
|
81
|
+
}
|
|
82
|
+
read(filePath) {
|
|
83
|
+
return __awaiter(this, void 0, void 0, function* () {
|
|
84
|
+
yield this.validateFile(filePath);
|
|
85
|
+
try {
|
|
86
|
+
const startTime = Date.now();
|
|
87
|
+
const { pages, cached } = yield this.ocr(filePath);
|
|
88
|
+
const usable = pages.filter((p) => p.text.trim());
|
|
89
|
+
if (usable.length === 0)
|
|
90
|
+
throw new Error("Tesseract produced no text");
|
|
91
|
+
const chunks = [];
|
|
92
|
+
let offset = 0;
|
|
93
|
+
usable.forEach((p) => {
|
|
94
|
+
const content = p.text;
|
|
95
|
+
chunks.push({
|
|
96
|
+
content,
|
|
97
|
+
index: p.index,
|
|
98
|
+
totalChunks: usable.length,
|
|
99
|
+
startOffset: offset,
|
|
100
|
+
endOffset: offset + content.length,
|
|
101
|
+
provenance: { locator: `p.${p.index}` }, // 1-based page number (ECS locator)
|
|
102
|
+
});
|
|
103
|
+
offset += content.length;
|
|
104
|
+
});
|
|
105
|
+
const stats = yield promises_1.default.stat(filePath);
|
|
106
|
+
return {
|
|
107
|
+
chunks,
|
|
108
|
+
metadata: {
|
|
109
|
+
type: "pdf",
|
|
110
|
+
fileName: filePath,
|
|
111
|
+
fileSize: stats.size,
|
|
112
|
+
pdfEngine: "tesseract",
|
|
113
|
+
tesseractLang: this.opts.lang,
|
|
114
|
+
tesseractCached: cached,
|
|
115
|
+
pageCount: usable.length,
|
|
116
|
+
processingTimeMs: Date.now() - startTime,
|
|
117
|
+
status: "success",
|
|
118
|
+
},
|
|
119
|
+
};
|
|
120
|
+
}
|
|
121
|
+
catch (error) {
|
|
122
|
+
this.logger.warn(`Tesseract OCR engine failed for ${filePath} (${error.message}); falling back to pdf2json`);
|
|
123
|
+
return this.fallback.read(filePath);
|
|
124
|
+
}
|
|
125
|
+
});
|
|
126
|
+
}
|
|
127
|
+
/** Rasterize + OCR every page (or reuse the sidecar). Throws on failure. */
|
|
128
|
+
ocr(filePath) {
|
|
129
|
+
return __awaiter(this, void 0, void 0, function* () {
|
|
130
|
+
var _a;
|
|
131
|
+
const sidecar = `${filePath}.tesseract.json`;
|
|
132
|
+
if (yield this.sidecarIsFresh(sidecar, filePath)) {
|
|
133
|
+
this.logger.debug(`Reusing Tesseract sidecar: ${sidecar}`);
|
|
134
|
+
return { pages: JSON.parse(yield promises_1.default.readFile(sidecar, "utf-8")), cached: true };
|
|
135
|
+
}
|
|
136
|
+
const { pdfToPng, createWorker } = yield this.loadDeps();
|
|
137
|
+
const pngPages = yield pdfToPng(filePath, {
|
|
138
|
+
returnPageContent: true, // in-memory PNG buffers, no temp files on disk
|
|
139
|
+
viewportScale: this.opts.scale,
|
|
140
|
+
});
|
|
141
|
+
const worker = yield createWorker(this.opts.lang, this.opts.oem, Object.assign({}, (this.opts.langPath ? { langPath: this.opts.langPath } : {})));
|
|
142
|
+
try {
|
|
143
|
+
if (typeof this.opts.psm === "number") {
|
|
144
|
+
yield worker.setParameters({ tessedit_pageseg_mode: this.opts.psm });
|
|
145
|
+
}
|
|
146
|
+
const pages = [];
|
|
147
|
+
for (const pg of pngPages) {
|
|
148
|
+
if (!pg.content)
|
|
149
|
+
continue;
|
|
150
|
+
const { data } = yield worker.recognize(pg.content);
|
|
151
|
+
pages.push({ index: pg.pageNumber, text: (_a = data.text) !== null && _a !== void 0 ? _a : "" });
|
|
152
|
+
}
|
|
153
|
+
yield promises_1.default.writeFile(sidecar, JSON.stringify(pages), "utf-8");
|
|
154
|
+
return { pages, cached: false };
|
|
155
|
+
}
|
|
156
|
+
finally {
|
|
157
|
+
yield worker.terminate().catch(() => undefined);
|
|
158
|
+
}
|
|
159
|
+
});
|
|
160
|
+
}
|
|
161
|
+
/** Lazy-load the heavy deps (or use the injected test doubles). */
|
|
162
|
+
loadDeps() {
|
|
163
|
+
return __awaiter(this, void 0, void 0, function* () {
|
|
164
|
+
var _a, _b, _c, _d;
|
|
165
|
+
const pdfToPng = (_b = (_a = this.injectedDeps) === null || _a === void 0 ? void 0 : _a.pdfToPng) !== null && _b !== void 0 ? _b : (yield Promise.resolve().then(() => __importStar(require("pdf-to-png-converter")))).pdfToPng;
|
|
166
|
+
const createWorker = (_d = (_c = this.injectedDeps) === null || _c === void 0 ? void 0 : _c.createWorker) !== null && _d !== void 0 ? _d : (yield Promise.resolve().then(() => __importStar(require("tesseract.js")))).createWorker;
|
|
167
|
+
return { pdfToPng, createWorker };
|
|
168
|
+
});
|
|
169
|
+
}
|
|
170
|
+
sidecarIsFresh(sidecar, filePath) {
|
|
171
|
+
return __awaiter(this, void 0, void 0, function* () {
|
|
172
|
+
try {
|
|
173
|
+
const [s, a] = yield Promise.all([promises_1.default.stat(sidecar), promises_1.default.stat(filePath)]);
|
|
174
|
+
return s.mtimeMs >= a.mtimeMs;
|
|
175
|
+
}
|
|
176
|
+
catch (_a) {
|
|
177
|
+
return false;
|
|
178
|
+
}
|
|
179
|
+
});
|
|
180
|
+
}
|
|
181
|
+
}
|
|
182
|
+
exports.TesseractPdfReader = TesseractPdfReader;
|
|
183
|
+
//# sourceMappingURL=TesseractPdfReader.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"TesseractPdfReader.js","sourceRoot":"","sources":["../../../../src/core/processor/readers/TesseractPdfReader.ts"],"names":[],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAAA,6CAAuE;AACvE,2DAA6B;AAsC7B;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAa,kBAAmB,SAAQ,uBAAU;IAChD,YACmB,IAAsB,EACtB,QAAoB,EACrC,OAAoB,EACpB,MAAc,EACG,YAAqC;QAEtD,KAAK,CAAC,CAAC,MAAM,CAAC,EAAE,OAAO,EAAE,MAAM,CAAC,CAAC;QANhB,SAAI,GAAJ,IAAI,CAAkB;QACtB,aAAQ,GAAR,QAAQ,CAAY;QAGpB,iBAAY,GAAZ,YAAY,CAAyB;IAGxD,CAAC;IAED,OAAO;QACL,OAAO,oBAAoB,CAAC;IAC9B,CAAC;IAED,SAAS;QACP,OAAO,eAAe,CAAC;IACzB,CAAC;IAEK,IAAI,CAAC,QAAgB;;YACzB,MAAM,IAAI,CAAC,YAAY,CAAC,QAAQ,CAAC,CAAC;YAClC,IAAI,CAAC;gBACH,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;gBAC7B,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,GAAG,MAAM,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC;gBACnD,MAAM,MAAM,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC;gBAClD,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;oBAAE,MAAM,IAAI,KAAK,CAAC,4BAA4B,CAAC,CAAC;gBAEvE,MAAM,MAAM,GAAkB,EAAE,CAAC;gBACjC,IAAI,MAAM,GAAG,CAAC,CAAC;gBACf,MAAM,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE;oBACnB,MAAM,OAAO,GAAG,CAAC,CAAC,IAAI,CAAC;oBACvB,MAAM,CAAC,IAAI,CAAC;wBACV,OAAO;wBACP,KAAK,EAAE,CAAC,CAAC,KAAK;wBACd,WAAW,EAAE,MAAM,CAAC,MAAM;wBAC1B,WAAW,EAAE,MAAM;wBACnB,SAAS,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM;wBAClC,UAAU,EAAE,EAAE,OAAO,EAAE,KAAK,CAAC,CAAC,KAAK,EAAE,EAAE,EAAE,oCAAoC;qBAC9E,CAAC,CAAC;oBACH,MAAM,IAAI,OAAO,CAAC,MAAM,CAAC;gBAC3B,CAAC,CAAC,CAAC;gBAEH,MAAM,KAAK,GAAG,MAAM,kBAAE,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;gBACtC,OAAO;oBACL,MAAM;oBACN,QAAQ,EAAE;wBACR,IAAI,EAAE,KAAK;wBACX,QAAQ,EAAE,QAAQ;wBAClB,QAAQ,EAAE,KAAK,CAAC,IAAI;wBACpB,SAAS,EAAE,WAAW;wBACtB,aAAa,EAAE,IAAI,CAAC,IAAI,CAAC,IAAI;wBAC7B,eAAe,EAAE,MAAM;wBACvB,SAAS,EAAE,MAAM,CAAC,MAAM;wBACxB,gBAAgB,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,SAAS;wBACxC,MAAM,EAAE,SAAS;qBAClB;iBACF,CAAC;YACJ,CAAC;YAAC,OAAO,KAAU,EAAE,CAAC;gBACpB,IAAI,CAAC,MAAM,CAAC,IAAI,CACd,mCAAmC,QAAQ,KAAK,KAAK,CAAC,OAAO,6BAA6B,CAC3F,CAAC;gBACF,OAAO,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;YACtC,CAAC;QACH,CAAC;KAAA;IAED,4EAA4E;IAC9D,GAAG,CAAC,QAAgB;;;YAChC,MAAM,OAAO,GAAG,GAAG,QAAQ,iBAAiB,CAAC;YAC7C,IAAI,MAAM,IAAI,CAAC,cAAc,CAAC,OAAO,EAAE,QAAQ,CAAC,EAAE,CAAC;gBACjD,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,8BAA8B,OAAO,EAAE,CAAC,CAAC;gBAC3D,OAAO,EAAE,KAAK,EAAE,IAAI,CAAC,KAAK,CAAC,MAAM,kBAAE,CAAC,QAAQ,CAAC,OAAO,EAAE,OAAO,CAAC,CAAc,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC;YAC/F,CAAC;YAED,MAAM,EAAE,QAAQ,EAAE,YAAY,EAAE,GAAG,MAAM,IAAI,CAAC,QAAQ,EAAE,CAAC;YAEzD,MAAM,QAAQ,GAAG,MAAM,QAAQ,CAAC,QAAQ,EAAE;gBACxC,iBAAiB,EAAE,IAAI,EAAE,+CAA+C;gBACxE,aAAa,EAAE,IAAI,CAAC,IAAI,CAAC,KAAK;aAC/B,CAAC,CAAC;YAEH,MAAM,MAAM,GAAG,MAAM,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,IAAI,CAAC,GAAG,oBAC1D,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,QAAQ,EAAE,IAAI,CAAC,IAAI,CAAC,QAAQ,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,EAC/D,CAAC;YACH,IAAI,CAAC;gBACH,IAAI,OAAO,IAAI,CAAC,IAAI,CAAC,GAAG,KAAK,QAAQ,EAAE,CAAC;oBACtC,MAAM,MAAM,CAAC,aAAa,CAAC,EAAE,qBAAqB,EAAE,IAAI,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC,CAAC;gBACvE,CAAC;gBACD,MAAM,KAAK,GAAc,EAAE,CAAC;gBAC5B,KAAK,MAAM,EAAE,IAAI,QAAQ,EAAE,CAAC;oBAC1B,IAAI,CAAC,EAAE,CAAC,OAAO;wBAAE,SAAS;oBAC1B,MAAM,EAAE,IAAI,EAAE,GAAG,MAAM,MAAM,CAAC,SAAS,CAAC,EAAE,CAAC,OAAO,CAAC,CAAC;oBACpD,KAAK,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,EAAE,CAAC,UAAU,EAAE,IAAI,EAAE,MAAA,IAAI,CAAC,IAAI,mCAAI,EAAE,EAAE,CAAC,CAAC;gBAC9D,CAAC;gBACD,MAAM,kBAAE,CAAC,SAAS,CAAC,OAAO,EAAE,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,EAAE,OAAO,CAAC,CAAC;gBAC5D,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC;YAClC,CAAC;oBAAS,CAAC;gBACT,MAAM,MAAM,CAAC,SAAS,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,CAAC;YAClD,CAAC;QACH,CAAC;KAAA;IAED,mEAAmE;IACrD,QAAQ;;;YACpB,MAAM,QAAQ,GACZ,MAAA,MAAA,IAAI,CAAC,YAAY,0CAAE,QAAQ,mCAAK,CAAC,wDAAa,sBAAsB,GAAC,CAAC,CAAC,QAAgB,CAAC;YAC1F,MAAM,YAAY,GAChB,MAAA,MAAA,IAAI,CAAC,YAAY,0CAAE,YAAY,mCAAK,CAAC,wDAAa,cAAc,GAAC,CAAC,CAAC,YAAoB,CAAC;YAC1F,OAAO,EAAE,QAAQ,EAAE,YAAY,EAAE,CAAC;QACpC,CAAC;KAAA;IAEa,cAAc,CAAC,OAAe,EAAE,QAAgB;;YAC5D,IAAI,CAAC;gBACH,MAAM,CAAC,CAAC,EAAE,CAAC,CAAC,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC,CAAC,kBAAE,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,kBAAE,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC;gBACxE,OAAO,CAAC,CAAC,OAAO,IAAI,CAAC,CAAC,OAAO,CAAC;YAChC,CAAC;YAAC,WAAM,CAAC;gBACP,OAAO,KAAK,CAAC;YACf,CAAC;QACH,CAAC;KAAA;CACF;AArHD,gDAqHC"}
|
|
@@ -0,0 +1,129 @@
|
|
|
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.TextReader = void 0;
|
|
46
|
+
const fs = __importStar(require("fs"));
|
|
47
|
+
const FileReader_1 = require("./FileReader");
|
|
48
|
+
/**
|
|
49
|
+
* Reader for plain text files
|
|
50
|
+
*/
|
|
51
|
+
class TextReader extends FileReader_1.FileReader {
|
|
52
|
+
constructor(chunker, logger) {
|
|
53
|
+
super([
|
|
54
|
+
'.txt', '.rst', '.asciidoc', '.csv',
|
|
55
|
+
'.ts', '.tsx', '.js', '.jsx', '.mjs', '.cjs',
|
|
56
|
+
'.py', '.pyw', '.pyi', '.pyc', '.pyd', '.pyo',
|
|
57
|
+
'.java', '.class', '.jar',
|
|
58
|
+
'.c', '.cpp', '.cc', '.cxx', '.h', '.hpp', '.hxx',
|
|
59
|
+
'.cs', '.vb', '.fs', '.fsx', '.fsi',
|
|
60
|
+
'.rb', '.rake', '.gemspec',
|
|
61
|
+
'.go', '.mod', '.sum',
|
|
62
|
+
'.rs', '.toml',
|
|
63
|
+
'.swift', '.kt', '.kts',
|
|
64
|
+
'.scala', '.sbt',
|
|
65
|
+
'.r', '.R', '.rmd', '.Rmd',
|
|
66
|
+
'.php', '.phtml',
|
|
67
|
+
'.l', '.log', '.lisp', '.lua',
|
|
68
|
+
'.vim', '.vimrc',
|
|
69
|
+
'.pl', '.pm', '.pod',
|
|
70
|
+
'.css', '.scss', '.sass', '.less',
|
|
71
|
+
'.html', '.htm', '.xhtml',
|
|
72
|
+
'.xml', '.xsl', '.xslt',
|
|
73
|
+
'.json', '.jsonl', '.geojson',
|
|
74
|
+
'.yaml', '.yml',
|
|
75
|
+
'.toml', '.ini', '.cfg', '.conf', '.config',
|
|
76
|
+
'.env', '.env.example', '.env.local', '.env.development', '.env.production',
|
|
77
|
+
'.sh', '.bash', '.zsh', '.fish', '.ksh',
|
|
78
|
+
'.ps1', '.psm1', '.psd1',
|
|
79
|
+
'.bat', '.cmd',
|
|
80
|
+
'.sql', '.pgsql', '.mysql',
|
|
81
|
+
'.dockerfile', '.dockerignore',
|
|
82
|
+
'.gitignore', '.gitattributes', '.gitmodules',
|
|
83
|
+
'.editorconfig', '.eslintrc', '.prettierrc',
|
|
84
|
+
'.npmrc', '.yarnrc', '.nvmrc',
|
|
85
|
+
// Common extensionless text files (matched by basename now that the
|
|
86
|
+
// universal '' entry is gone). Keep this list curated — genuinely
|
|
87
|
+
// unknown / extensionless files fall through to BinaryReader (skipped).
|
|
88
|
+
'LICENSE', 'LICENCE', 'README', 'CHANGELOG', 'AUTHORS', 'NOTICE', 'COPYING',
|
|
89
|
+
'Dockerfile', 'Containerfile',
|
|
90
|
+
'Makefile', 'makefile', 'GNUmakefile',
|
|
91
|
+
'Rakefile', 'Gemfile', 'Guardfile',
|
|
92
|
+
'Gruntfile', 'gulpfile',
|
|
93
|
+
'webpack.config', 'rollup.config', 'vite.config',
|
|
94
|
+
'tsconfig', 'jsconfig', 'package.json', 'package-lock.json',
|
|
95
|
+
'pom.xml', 'build.gradle', 'build.sbt',
|
|
96
|
+
'requirements.txt', 'setup.py', 'setup.cfg', 'pyproject.toml',
|
|
97
|
+
'Cargo.toml', 'Cargo.lock',
|
|
98
|
+
'go.mod', 'go.sum',
|
|
99
|
+
'.proto', '.graphql', '.gql'
|
|
100
|
+
], chunker, logger);
|
|
101
|
+
}
|
|
102
|
+
getName() {
|
|
103
|
+
return 'TextReader';
|
|
104
|
+
}
|
|
105
|
+
read(filePath) {
|
|
106
|
+
return __awaiter(this, void 0, void 0, function* () {
|
|
107
|
+
yield this.validateFile(filePath);
|
|
108
|
+
try {
|
|
109
|
+
this.logger.debug(`Reading text file: ${filePath}`);
|
|
110
|
+
const content = yield fs.promises.readFile(filePath, 'utf-8');
|
|
111
|
+
const chunks = yield this.chunker.chunk(content);
|
|
112
|
+
return {
|
|
113
|
+
chunks,
|
|
114
|
+
metadata: {
|
|
115
|
+
type: 'text',
|
|
116
|
+
encoding: 'utf-8',
|
|
117
|
+
size: content.length
|
|
118
|
+
}
|
|
119
|
+
};
|
|
120
|
+
}
|
|
121
|
+
catch (error) {
|
|
122
|
+
this.logger.error(`Failed to read text file ${filePath}: ${error}`);
|
|
123
|
+
throw new Error(`Failed to read text file: ${error}`);
|
|
124
|
+
}
|
|
125
|
+
});
|
|
126
|
+
}
|
|
127
|
+
}
|
|
128
|
+
exports.TextReader = TextReader;
|
|
129
|
+
//# sourceMappingURL=TextReader.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"TextReader.js","sourceRoot":"","sources":["../../../../src/core/processor/readers/TextReader.ts"],"names":[],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAAA,uCAAyB;AACzB,6CAA0D;AAI1D;;GAEG;AACH,MAAa,UAAW,SAAQ,uBAAU;IACxC,YAAY,OAAoB,EAAE,MAAc;QAC9C,KAAK,CAAC;YACJ,MAAM,EAAE,MAAM,EAAE,WAAW,EAAE,MAAM;YACnC,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM;YAC5C,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM;YAC7C,OAAO,EAAE,QAAQ,EAAE,MAAM;YACzB,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM;YACjD,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM;YACnC,KAAK,EAAE,OAAO,EAAE,UAAU;YAC1B,KAAK,EAAE,MAAM,EAAE,MAAM;YACrB,KAAK,EAAE,OAAO;YACd,QAAQ,EAAE,KAAK,EAAE,MAAM;YACvB,QAAQ,EAAE,MAAM;YAChB,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM;YAC1B,MAAM,EAAE,QAAQ;YAChB,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM;YAC7B,MAAM,EAAE,QAAQ;YAChB,KAAK,EAAE,KAAK,EAAE,MAAM;YACpB,MAAM,EAAE,OAAO,EAAE,OAAO,EAAE,OAAO;YACjC,OAAO,EAAE,MAAM,EAAE,QAAQ;YACzB,MAAM,EAAE,MAAM,EAAE,OAAO;YACvB,OAAO,EAAE,QAAQ,EAAE,UAAU;YAC7B,OAAO,EAAE,MAAM;YACf,OAAO,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO,EAAE,SAAS;YAC3C,MAAM,EAAE,cAAc,EAAE,YAAY,EAAE,kBAAkB,EAAE,iBAAiB;YAC3E,KAAK,EAAE,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM;YACvC,MAAM,EAAE,OAAO,EAAE,OAAO;YACxB,MAAM,EAAE,MAAM;YACd,MAAM,EAAE,QAAQ,EAAE,QAAQ;YAC1B,aAAa,EAAE,eAAe;YAC9B,YAAY,EAAE,gBAAgB,EAAE,aAAa;YAC7C,eAAe,EAAE,WAAW,EAAE,aAAa;YAC3C,QAAQ,EAAE,SAAS,EAAE,QAAQ;YAC7B,oEAAoE;YACpE,kEAAkE;YAClE,wEAAwE;YACxE,SAAS,EAAE,SAAS,EAAE,QAAQ,EAAE,WAAW,EAAE,SAAS,EAAE,QAAQ,EAAE,SAAS;YAC3E,YAAY,EAAE,eAAe;YAC7B,UAAU,EAAE,UAAU,EAAE,aAAa;YACrC,UAAU,EAAE,SAAS,EAAE,WAAW;YAClC,WAAW,EAAE,UAAU;YACvB,gBAAgB,EAAE,eAAe,EAAE,aAAa;YAChD,UAAU,EAAE,UAAU,EAAE,cAAc,EAAE,mBAAmB;YAC3D,SAAS,EAAE,cAAc,EAAE,WAAW;YACtC,kBAAkB,EAAE,UAAU,EAAE,WAAW,EAAE,gBAAgB;YAC7D,YAAY,EAAE,YAAY;YAC1B,QAAQ,EAAE,QAAQ;YAClB,QAAQ,EAAE,UAAU,EAAE,MAAM;SAC7B,EAAE,OAAO,EAAE,MAAM,CAAC,CAAC;IACtB,CAAC;IAED,OAAO;QACL,OAAO,YAAY,CAAC;IACtB,CAAC;IAEK,IAAI,CAAC,QAAgB;;YACzB,MAAM,IAAI,CAAC,YAAY,CAAC,QAAQ,CAAC,CAAC;YAElC,IAAI,CAAC;gBACH,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,sBAAsB,QAAQ,EAAE,CAAC,CAAC;gBACpD,MAAM,OAAO,GAAG,MAAM,EAAE,CAAC,QAAQ,CAAC,QAAQ,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC;gBAC9D,MAAM,MAAM,GAAG,MAAM,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC;gBAEjD,OAAO;oBACL,MAAM;oBACN,QAAQ,EAAE;wBACR,IAAI,EAAE,MAAM;wBACZ,QAAQ,EAAE,OAAO;wBACjB,IAAI,EAAE,OAAO,CAAC,MAAM;qBACrB;iBACF,CAAC;YACJ,CAAC;YAAC,OAAO,KAAK,EAAE,CAAC;gBACf,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,4BAA4B,QAAQ,KAAK,KAAK,EAAE,CAAC,CAAC;gBACpE,MAAM,IAAI,KAAK,CAAC,6BAA6B,KAAK,EAAE,CAAC,CAAC;YACxD,CAAC;QACH,CAAC;KAAA;CACF;AA7ED,gCA6EC"}
|