@hazeljs/rag 0.2.0-alpha.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/LICENSE +192 -0
- package/README.md +504 -0
- package/dist/__tests__/graph/community-detector.test.d.ts +2 -0
- package/dist/__tests__/graph/community-detector.test.d.ts.map +1 -0
- package/dist/__tests__/graph/community-detector.test.js +87 -0
- package/dist/__tests__/graph/community-detector.test.js.map +1 -0
- package/dist/__tests__/graph/community-summarizer.test.d.ts +2 -0
- package/dist/__tests__/graph/community-summarizer.test.d.ts.map +1 -0
- package/dist/__tests__/graph/community-summarizer.test.js +131 -0
- package/dist/__tests__/graph/community-summarizer.test.js.map +1 -0
- package/dist/__tests__/graph/entity-extractor.test.d.ts +2 -0
- package/dist/__tests__/graph/entity-extractor.test.d.ts.map +1 -0
- package/dist/__tests__/graph/entity-extractor.test.js +129 -0
- package/dist/__tests__/graph/entity-extractor.test.js.map +1 -0
- package/dist/__tests__/graph/graph-rag-pipeline.test.d.ts +2 -0
- package/dist/__tests__/graph/graph-rag-pipeline.test.d.ts.map +1 -0
- package/dist/__tests__/graph/graph-rag-pipeline.test.js +158 -0
- package/dist/__tests__/graph/graph-rag-pipeline.test.js.map +1 -0
- package/dist/__tests__/graph/knowledge-graph.test.d.ts +2 -0
- package/dist/__tests__/graph/knowledge-graph.test.d.ts.map +1 -0
- package/dist/__tests__/graph/knowledge-graph.test.js +208 -0
- package/dist/__tests__/graph/knowledge-graph.test.js.map +1 -0
- package/dist/__tests__/loaders/base.loader.test.d.ts +2 -0
- package/dist/__tests__/loaders/base.loader.test.d.ts.map +1 -0
- package/dist/__tests__/loaders/base.loader.test.js +114 -0
- package/dist/__tests__/loaders/base.loader.test.js.map +1 -0
- package/dist/__tests__/loaders/csv-file.loader.test.d.ts +2 -0
- package/dist/__tests__/loaders/csv-file.loader.test.d.ts.map +1 -0
- package/dist/__tests__/loaders/csv-file.loader.test.js +98 -0
- package/dist/__tests__/loaders/csv-file.loader.test.js.map +1 -0
- package/dist/__tests__/loaders/directory.loader.test.d.ts +2 -0
- package/dist/__tests__/loaders/directory.loader.test.d.ts.map +1 -0
- package/dist/__tests__/loaders/directory.loader.test.js +154 -0
- package/dist/__tests__/loaders/directory.loader.test.js.map +1 -0
- package/dist/__tests__/loaders/html-file.loader.test.d.ts +2 -0
- package/dist/__tests__/loaders/html-file.loader.test.d.ts.map +1 -0
- package/dist/__tests__/loaders/html-file.loader.test.js +93 -0
- package/dist/__tests__/loaders/html-file.loader.test.js.map +1 -0
- package/dist/__tests__/loaders/json-file.loader.test.d.ts +2 -0
- package/dist/__tests__/loaders/json-file.loader.test.d.ts.map +1 -0
- package/dist/__tests__/loaders/json-file.loader.test.js +84 -0
- package/dist/__tests__/loaders/json-file.loader.test.js.map +1 -0
- package/dist/__tests__/loaders/markdown-file.loader.test.d.ts +2 -0
- package/dist/__tests__/loaders/markdown-file.loader.test.d.ts.map +1 -0
- package/dist/__tests__/loaders/markdown-file.loader.test.js +83 -0
- package/dist/__tests__/loaders/markdown-file.loader.test.js.map +1 -0
- package/dist/__tests__/loaders/text-file.loader.test.d.ts +2 -0
- package/dist/__tests__/loaders/text-file.loader.test.d.ts.map +1 -0
- package/dist/__tests__/loaders/text-file.loader.test.js +50 -0
- package/dist/__tests__/loaders/text-file.loader.test.js.map +1 -0
- package/dist/__tests__/rag-pipeline.test.d.ts +2 -0
- package/dist/__tests__/rag-pipeline.test.d.ts.map +1 -0
- package/dist/__tests__/rag-pipeline.test.js +210 -0
- package/dist/__tests__/rag-pipeline.test.js.map +1 -0
- package/dist/__tests__/retrieval/bm25.test.d.ts +2 -0
- package/dist/__tests__/retrieval/bm25.test.d.ts.map +1 -0
- package/dist/__tests__/retrieval/bm25.test.js +86 -0
- package/dist/__tests__/retrieval/bm25.test.js.map +1 -0
- package/dist/__tests__/retrieval/hybrid-search.test.d.ts +2 -0
- package/dist/__tests__/retrieval/hybrid-search.test.d.ts.map +1 -0
- package/dist/__tests__/retrieval/hybrid-search.test.js +85 -0
- package/dist/__tests__/retrieval/hybrid-search.test.js.map +1 -0
- package/dist/__tests__/retrieval/multi-query.test.d.ts +2 -0
- package/dist/__tests__/retrieval/multi-query.test.d.ts.map +1 -0
- package/dist/__tests__/retrieval/multi-query.test.js +90 -0
- package/dist/__tests__/retrieval/multi-query.test.js.map +1 -0
- package/dist/__tests__/text-splitters/recursive-text-splitter.test.d.ts +2 -0
- package/dist/__tests__/text-splitters/recursive-text-splitter.test.d.ts.map +1 -0
- package/dist/__tests__/text-splitters/recursive-text-splitter.test.js +97 -0
- package/dist/__tests__/text-splitters/recursive-text-splitter.test.js.map +1 -0
- package/dist/__tests__/utils/similarity.test.d.ts +2 -0
- package/dist/__tests__/utils/similarity.test.d.ts.map +1 -0
- package/dist/__tests__/utils/similarity.test.js +47 -0
- package/dist/__tests__/utils/similarity.test.js.map +1 -0
- package/dist/agentic/agentic-rag.service.d.ts +49 -0
- package/dist/agentic/agentic-rag.service.d.ts.map +1 -0
- package/dist/agentic/agentic-rag.service.js +149 -0
- package/dist/agentic/agentic-rag.service.js.map +1 -0
- package/dist/agentic/decorators/active-learning.decorator.d.ts +19 -0
- package/dist/agentic/decorators/active-learning.decorator.d.ts.map +1 -0
- package/dist/agentic/decorators/active-learning.decorator.js +98 -0
- package/dist/agentic/decorators/active-learning.decorator.js.map +1 -0
- package/dist/agentic/decorators/adaptive-retrieval.decorator.d.ts +17 -0
- package/dist/agentic/decorators/adaptive-retrieval.decorator.d.ts.map +1 -0
- package/dist/agentic/decorators/adaptive-retrieval.decorator.js +103 -0
- package/dist/agentic/decorators/adaptive-retrieval.decorator.js.map +1 -0
- package/dist/agentic/decorators/cached.decorator.d.ts +18 -0
- package/dist/agentic/decorators/cached.decorator.d.ts.map +1 -0
- package/dist/agentic/decorators/cached.decorator.js +93 -0
- package/dist/agentic/decorators/cached.decorator.js.map +1 -0
- package/dist/agentic/decorators/context-aware.decorator.d.ts +16 -0
- package/dist/agentic/decorators/context-aware.decorator.d.ts.map +1 -0
- package/dist/agentic/decorators/context-aware.decorator.js +169 -0
- package/dist/agentic/decorators/context-aware.decorator.js.map +1 -0
- package/dist/agentic/decorators/corrective-rag.decorator.d.ts +16 -0
- package/dist/agentic/decorators/corrective-rag.decorator.d.ts.map +1 -0
- package/dist/agentic/decorators/corrective-rag.decorator.js +142 -0
- package/dist/agentic/decorators/corrective-rag.decorator.js.map +1 -0
- package/dist/agentic/decorators/hyde.decorator.d.ts +15 -0
- package/dist/agentic/decorators/hyde.decorator.d.ts.map +1 -0
- package/dist/agentic/decorators/hyde.decorator.js +91 -0
- package/dist/agentic/decorators/hyde.decorator.js.map +1 -0
- package/dist/agentic/decorators/index.d.ts +16 -0
- package/dist/agentic/decorators/index.d.ts.map +1 -0
- package/dist/agentic/decorators/index.js +32 -0
- package/dist/agentic/decorators/index.js.map +1 -0
- package/dist/agentic/decorators/multi-hop.decorator.d.ts +15 -0
- package/dist/agentic/decorators/multi-hop.decorator.d.ts.map +1 -0
- package/dist/agentic/decorators/multi-hop.decorator.js +109 -0
- package/dist/agentic/decorators/multi-hop.decorator.js.map +1 -0
- package/dist/agentic/decorators/query-planner.decorator.d.ts +20 -0
- package/dist/agentic/decorators/query-planner.decorator.d.ts.map +1 -0
- package/dist/agentic/decorators/query-planner.decorator.js +213 -0
- package/dist/agentic/decorators/query-planner.decorator.js.map +1 -0
- package/dist/agentic/decorators/query-rewriter.decorator.d.ts +16 -0
- package/dist/agentic/decorators/query-rewriter.decorator.d.ts.map +1 -0
- package/dist/agentic/decorators/query-rewriter.decorator.js +143 -0
- package/dist/agentic/decorators/query-rewriter.decorator.js.map +1 -0
- package/dist/agentic/decorators/self-reflective.decorator.d.ts +20 -0
- package/dist/agentic/decorators/self-reflective.decorator.d.ts.map +1 -0
- package/dist/agentic/decorators/self-reflective.decorator.js +189 -0
- package/dist/agentic/decorators/self-reflective.decorator.js.map +1 -0
- package/dist/agentic/decorators/source-verification.decorator.d.ts +15 -0
- package/dist/agentic/decorators/source-verification.decorator.d.ts.map +1 -0
- package/dist/agentic/decorators/source-verification.decorator.js +121 -0
- package/dist/agentic/decorators/source-verification.decorator.js.map +1 -0
- package/dist/agentic/index.d.ts +9 -0
- package/dist/agentic/index.d.ts.map +1 -0
- package/dist/agentic/index.js +25 -0
- package/dist/agentic/index.js.map +1 -0
- package/dist/agentic/types.d.ts +210 -0
- package/dist/agentic/types.d.ts.map +1 -0
- package/dist/agentic/types.js +7 -0
- package/dist/agentic/types.js.map +1 -0
- package/dist/decorators/embeddable.decorator.d.ts +31 -0
- package/dist/decorators/embeddable.decorator.d.ts.map +1 -0
- package/dist/decorators/embeddable.decorator.js +44 -0
- package/dist/decorators/embeddable.decorator.js.map +1 -0
- package/dist/decorators/rag.decorator.d.ts +58 -0
- package/dist/decorators/rag.decorator.d.ts.map +1 -0
- package/dist/decorators/rag.decorator.js +78 -0
- package/dist/decorators/rag.decorator.js.map +1 -0
- package/dist/decorators/semantic-search.decorator.d.ts +69 -0
- package/dist/decorators/semantic-search.decorator.d.ts.map +1 -0
- package/dist/decorators/semantic-search.decorator.js +116 -0
- package/dist/decorators/semantic-search.decorator.js.map +1 -0
- package/dist/embeddings/cohere-embeddings.d.ts +33 -0
- package/dist/embeddings/cohere-embeddings.d.ts.map +1 -0
- package/dist/embeddings/cohere-embeddings.js +91 -0
- package/dist/embeddings/cohere-embeddings.js.map +1 -0
- package/dist/embeddings/openai-embeddings.d.ts +21 -0
- package/dist/embeddings/openai-embeddings.d.ts.map +1 -0
- package/dist/embeddings/openai-embeddings.js +53 -0
- package/dist/embeddings/openai-embeddings.js.map +1 -0
- package/dist/graph/community-detector.d.ts +45 -0
- package/dist/graph/community-detector.d.ts.map +1 -0
- package/dist/graph/community-detector.js +153 -0
- package/dist/graph/community-detector.js.map +1 -0
- package/dist/graph/community-summarizer.d.ts +41 -0
- package/dist/graph/community-summarizer.d.ts.map +1 -0
- package/dist/graph/community-summarizer.js +119 -0
- package/dist/graph/community-summarizer.js.map +1 -0
- package/dist/graph/entity-extractor.d.ts +47 -0
- package/dist/graph/entity-extractor.d.ts.map +1 -0
- package/dist/graph/entity-extractor.js +224 -0
- package/dist/graph/entity-extractor.js.map +1 -0
- package/dist/graph/graph-rag-pipeline.d.ts +83 -0
- package/dist/graph/graph-rag-pipeline.d.ts.map +1 -0
- package/dist/graph/graph-rag-pipeline.js +390 -0
- package/dist/graph/graph-rag-pipeline.js.map +1 -0
- package/dist/graph/graph.types.d.ts +186 -0
- package/dist/graph/graph.types.d.ts.map +1 -0
- package/dist/graph/graph.types.js +20 -0
- package/dist/graph/graph.types.js.map +1 -0
- package/dist/graph/index.d.ts +15 -0
- package/dist/graph/index.d.ts.map +1 -0
- package/dist/graph/index.js +31 -0
- package/dist/graph/index.js.map +1 -0
- package/dist/graph/knowledge-graph.d.ts +57 -0
- package/dist/graph/knowledge-graph.d.ts.map +1 -0
- package/dist/graph/knowledge-graph.js +198 -0
- package/dist/graph/knowledge-graph.js.map +1 -0
- package/dist/index.d.ts +29 -0
- package/dist/index.d.ts.map +1 -0
- package/dist/index.js +58 -0
- package/dist/index.js.map +1 -0
- package/dist/loaders/base.loader.d.ts +108 -0
- package/dist/loaders/base.loader.d.ts.map +1 -0
- package/dist/loaders/base.loader.js +123 -0
- package/dist/loaders/base.loader.js.map +1 -0
- package/dist/loaders/csv-file.loader.d.ts +61 -0
- package/dist/loaders/csv-file.loader.d.ts.map +1 -0
- package/dist/loaders/csv-file.loader.js +162 -0
- package/dist/loaders/csv-file.loader.js.map +1 -0
- package/dist/loaders/directory.loader.d.ts +67 -0
- package/dist/loaders/directory.loader.d.ts.map +1 -0
- package/dist/loaders/directory.loader.js +163 -0
- package/dist/loaders/directory.loader.js.map +1 -0
- package/dist/loaders/docx.loader.d.ts +52 -0
- package/dist/loaders/docx.loader.d.ts.map +1 -0
- package/dist/loaders/docx.loader.js +110 -0
- package/dist/loaders/docx.loader.js.map +1 -0
- package/dist/loaders/github.loader.d.ts +114 -0
- package/dist/loaders/github.loader.d.ts.map +1 -0
- package/dist/loaders/github.loader.js +217 -0
- package/dist/loaders/github.loader.js.map +1 -0
- package/dist/loaders/html-file.loader.d.ts +55 -0
- package/dist/loaders/html-file.loader.d.ts.map +1 -0
- package/dist/loaders/html-file.loader.js +170 -0
- package/dist/loaders/html-file.loader.js.map +1 -0
- package/dist/loaders/index.d.ts +52 -0
- package/dist/loaders/index.d.ts.map +1 -0
- package/dist/loaders/index.js +61 -0
- package/dist/loaders/index.js.map +1 -0
- package/dist/loaders/json-file.loader.d.ts +51 -0
- package/dist/loaders/json-file.loader.d.ts.map +1 -0
- package/dist/loaders/json-file.loader.js +100 -0
- package/dist/loaders/json-file.loader.js.map +1 -0
- package/dist/loaders/markdown-file.loader.d.ts +61 -0
- package/dist/loaders/markdown-file.loader.d.ts.map +1 -0
- package/dist/loaders/markdown-file.loader.js +148 -0
- package/dist/loaders/markdown-file.loader.js.map +1 -0
- package/dist/loaders/pdf.loader.d.ts +64 -0
- package/dist/loaders/pdf.loader.d.ts.map +1 -0
- package/dist/loaders/pdf.loader.js +163 -0
- package/dist/loaders/pdf.loader.js.map +1 -0
- package/dist/loaders/text-file.loader.d.ts +39 -0
- package/dist/loaders/text-file.loader.d.ts.map +1 -0
- package/dist/loaders/text-file.loader.js +69 -0
- package/dist/loaders/text-file.loader.js.map +1 -0
- package/dist/loaders/web.loader.d.ts +87 -0
- package/dist/loaders/web.loader.d.ts.map +1 -0
- package/dist/loaders/web.loader.js +194 -0
- package/dist/loaders/web.loader.js.map +1 -0
- package/dist/loaders/youtube-transcript.loader.d.ts +92 -0
- package/dist/loaders/youtube-transcript.loader.d.ts.map +1 -0
- package/dist/loaders/youtube-transcript.loader.js +254 -0
- package/dist/loaders/youtube-transcript.loader.js.map +1 -0
- package/dist/memory/index.d.ts +11 -0
- package/dist/memory/index.d.ts.map +1 -0
- package/dist/memory/index.js +31 -0
- package/dist/memory/index.js.map +1 -0
- package/dist/memory/memory-manager.d.ts +96 -0
- package/dist/memory/memory-manager.d.ts.map +1 -0
- package/dist/memory/memory-manager.js +369 -0
- package/dist/memory/memory-manager.js.map +1 -0
- package/dist/memory/memory-store.interface.d.ts +73 -0
- package/dist/memory/memory-store.interface.d.ts.map +1 -0
- package/dist/memory/memory-store.interface.js +6 -0
- package/dist/memory/memory-store.interface.js.map +1 -0
- package/dist/memory/stores/buffer-memory.d.ts +47 -0
- package/dist/memory/stores/buffer-memory.d.ts.map +1 -0
- package/dist/memory/stores/buffer-memory.js +280 -0
- package/dist/memory/stores/buffer-memory.js.map +1 -0
- package/dist/memory/stores/hybrid-memory.d.ts +49 -0
- package/dist/memory/stores/hybrid-memory.d.ts.map +1 -0
- package/dist/memory/stores/hybrid-memory.js +194 -0
- package/dist/memory/stores/hybrid-memory.js.map +1 -0
- package/dist/memory/stores/vector-memory.d.ts +48 -0
- package/dist/memory/stores/vector-memory.d.ts.map +1 -0
- package/dist/memory/stores/vector-memory.js +312 -0
- package/dist/memory/stores/vector-memory.js.map +1 -0
- package/dist/memory/types.d.ts +119 -0
- package/dist/memory/types.d.ts.map +1 -0
- package/dist/memory/types.js +18 -0
- package/dist/memory/types.js.map +1 -0
- package/dist/prompts/agentic/adaptive-retrieval.prompt.d.ts +8 -0
- package/dist/prompts/agentic/adaptive-retrieval.prompt.d.ts.map +1 -0
- package/dist/prompts/agentic/adaptive-retrieval.prompt.js +27 -0
- package/dist/prompts/agentic/adaptive-retrieval.prompt.js.map +1 -0
- package/dist/prompts/agentic/corrective-rag.prompt.d.ts +9 -0
- package/dist/prompts/agentic/corrective-rag.prompt.d.ts.map +1 -0
- package/dist/prompts/agentic/corrective-rag.prompt.js +23 -0
- package/dist/prompts/agentic/corrective-rag.prompt.js.map +1 -0
- package/dist/prompts/agentic/hyde.prompt.d.ts +9 -0
- package/dist/prompts/agentic/hyde.prompt.d.ts.map +1 -0
- package/dist/prompts/agentic/hyde.prompt.js +18 -0
- package/dist/prompts/agentic/hyde.prompt.js.map +1 -0
- package/dist/prompts/agentic/multi-hop.prompt.d.ts +15 -0
- package/dist/prompts/agentic/multi-hop.prompt.d.ts.map +1 -0
- package/dist/prompts/agentic/multi-hop.prompt.js +38 -0
- package/dist/prompts/agentic/multi-hop.prompt.js.map +1 -0
- package/dist/prompts/agentic/query-planner.prompt.d.ts +8 -0
- package/dist/prompts/agentic/query-planner.prompt.d.ts.map +1 -0
- package/dist/prompts/agentic/query-planner.prompt.js +30 -0
- package/dist/prompts/agentic/query-planner.prompt.js.map +1 -0
- package/dist/prompts/agentic/query-rewriter.prompt.d.ts +10 -0
- package/dist/prompts/agentic/query-rewriter.prompt.d.ts.map +1 -0
- package/dist/prompts/agentic/query-rewriter.prompt.js +17 -0
- package/dist/prompts/agentic/query-rewriter.prompt.js.map +1 -0
- package/dist/prompts/agentic/self-reflective-improve.prompt.d.ts +10 -0
- package/dist/prompts/agentic/self-reflective-improve.prompt.d.ts.map +1 -0
- package/dist/prompts/agentic/self-reflective-improve.prompt.js +24 -0
- package/dist/prompts/agentic/self-reflective-improve.prompt.js.map +1 -0
- package/dist/prompts/agentic/self-reflective.prompt.d.ts +9 -0
- package/dist/prompts/agentic/self-reflective.prompt.d.ts.map +1 -0
- package/dist/prompts/agentic/self-reflective.prompt.js +32 -0
- package/dist/prompts/agentic/self-reflective.prompt.js.map +1 -0
- package/dist/prompts/community-summary.prompt.d.ts +9 -0
- package/dist/prompts/community-summary.prompt.d.ts.map +1 -0
- package/dist/prompts/community-summary.prompt.js +30 -0
- package/dist/prompts/community-summary.prompt.js.map +1 -0
- package/dist/prompts/entity-extraction.prompt.d.ts +10 -0
- package/dist/prompts/entity-extraction.prompt.d.ts.map +1 -0
- package/dist/prompts/entity-extraction.prompt.js +39 -0
- package/dist/prompts/entity-extraction.prompt.js.map +1 -0
- package/dist/prompts/graph-search.prompt.d.ts +10 -0
- package/dist/prompts/graph-search.prompt.d.ts.map +1 -0
- package/dist/prompts/graph-search.prompt.js +23 -0
- package/dist/prompts/graph-search.prompt.js.map +1 -0
- package/dist/prompts/index.d.ts +13 -0
- package/dist/prompts/index.d.ts.map +1 -0
- package/dist/prompts/index.js +29 -0
- package/dist/prompts/index.js.map +1 -0
- package/dist/prompts/rag-answer.prompt.d.ts +9 -0
- package/dist/prompts/rag-answer.prompt.d.ts.map +1 -0
- package/dist/prompts/rag-answer.prompt.js +20 -0
- package/dist/prompts/rag-answer.prompt.js.map +1 -0
- package/dist/rag-pipeline-with-memory.d.ts +68 -0
- package/dist/rag-pipeline-with-memory.d.ts.map +1 -0
- package/dist/rag-pipeline-with-memory.js +186 -0
- package/dist/rag-pipeline-with-memory.js.map +1 -0
- package/dist/rag-pipeline.d.ts +59 -0
- package/dist/rag-pipeline.d.ts.map +1 -0
- package/dist/rag-pipeline.js +181 -0
- package/dist/rag-pipeline.js.map +1 -0
- package/dist/rag.module.d.ts +26 -0
- package/dist/rag.module.d.ts.map +1 -0
- package/dist/rag.module.js +40 -0
- package/dist/rag.module.js.map +1 -0
- package/dist/rag.service.d.ts +96 -0
- package/dist/rag.service.d.ts.map +1 -0
- package/dist/rag.service.js +173 -0
- package/dist/rag.service.js.map +1 -0
- package/dist/retrieval/bm25.d.ts +57 -0
- package/dist/retrieval/bm25.d.ts.map +1 -0
- package/dist/retrieval/bm25.js +106 -0
- package/dist/retrieval/bm25.js.map +1 -0
- package/dist/retrieval/hybrid-search.d.ts +48 -0
- package/dist/retrieval/hybrid-search.d.ts.map +1 -0
- package/dist/retrieval/hybrid-search.js +123 -0
- package/dist/retrieval/hybrid-search.js.map +1 -0
- package/dist/retrieval/multi-query.d.ts +38 -0
- package/dist/retrieval/multi-query.d.ts.map +1 -0
- package/dist/retrieval/multi-query.js +135 -0
- package/dist/retrieval/multi-query.js.map +1 -0
- package/dist/text-splitters/recursive-text-splitter.d.ts +21 -0
- package/dist/text-splitters/recursive-text-splitter.d.ts.map +1 -0
- package/dist/text-splitters/recursive-text-splitter.js +95 -0
- package/dist/text-splitters/recursive-text-splitter.js.map +1 -0
- package/dist/types/index.d.ts +144 -0
- package/dist/types/index.d.ts.map +1 -0
- package/dist/types/index.js +16 -0
- package/dist/types/index.js.map +1 -0
- package/dist/utils/similarity.d.ts +16 -0
- package/dist/utils/similarity.d.ts.map +1 -0
- package/dist/utils/similarity.js +58 -0
- package/dist/utils/similarity.js.map +1 -0
- package/dist/vector-stores/chroma.store.d.ts +42 -0
- package/dist/vector-stores/chroma.store.d.ts.map +1 -0
- package/dist/vector-stores/chroma.store.js +242 -0
- package/dist/vector-stores/chroma.store.js.map +1 -0
- package/dist/vector-stores/memory-vector-store.d.ts +20 -0
- package/dist/vector-stores/memory-vector-store.d.ts.map +1 -0
- package/dist/vector-stores/memory-vector-store.js +94 -0
- package/dist/vector-stores/memory-vector-store.js.map +1 -0
- package/dist/vector-stores/pinecone.store.d.ts +34 -0
- package/dist/vector-stores/pinecone.store.d.ts.map +1 -0
- package/dist/vector-stores/pinecone.store.js +146 -0
- package/dist/vector-stores/pinecone.store.js.map +1 -0
- package/dist/vector-stores/qdrant.store.d.ts +33 -0
- package/dist/vector-stores/qdrant.store.d.ts.map +1 -0
- package/dist/vector-stores/qdrant.store.js +174 -0
- package/dist/vector-stores/qdrant.store.js.map +1 -0
- package/dist/vector-stores/weaviate.store.d.ts +37 -0
- package/dist/vector-stores/weaviate.store.d.ts.map +1 -0
- package/dist/vector-stores/weaviate.store.js +226 -0
- package/dist/vector-stores/weaviate.store.js.map +1 -0
- package/package.json +146 -0
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"directory.loader.test.js","sourceRoot":"","sources":["../../../src/__tests__/loaders/directory.loader.test.ts"],"names":[],"mappings":";;AAAA,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,GAAG,EAAE,CAAC,CAAC;IAC9B,OAAO,EAAE,IAAI,CAAC,EAAE,EAAE;IAClB,IAAI,EAAE,IAAI,CAAC,EAAE,EAAE;IACf,QAAQ,EAAE,IAAI,CAAC,EAAE,EAAE;CACpB,CAAC,CAAC,CAAC;AAEJ,0CAAsD;AACtD,qEAAiE;AAEjE,MAAM,WAAW,GAAG,kBAA8C,CAAC;AACnE,MAAM,QAAQ,GAAG,eAAwC,CAAC;AAC1D,MAAM,YAAY,GAAG,mBAAgD,CAAC;AAEtE,SAAS,QAAQ,CAAC,KAAc;IAC9B,OAAO,EAAE,WAAW,EAAE,GAAG,EAAE,CAAC,KAAK,EAAsC,CAAC;AAC1E,CAAC;AAED,QAAQ,CAAC,iBAAiB,EAAE,GAAG,EAAE;IAC/B,SAAS,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,aAAa,EAAE,CAAC,CAAC;IAEtC,EAAE,CAAC,wCAAwC,EAAE,KAAK,IAAI,EAAE;QACtD,WAAW,CAAC,qBAAqB,CAAC,CAAC,WAAW,CAAU,CAAC,CAAC;QAC1D,QAAQ,CAAC,qBAAqB,CAAC,QAAQ,CAAC,KAAK,CAAU,CAAC,CAAC;QACzD,YAAY,CAAC,qBAAqB,CAAC,cAAuB,CAAC,CAAC;QAE5D,MAAM,MAAM,GAAG,IAAI,kCAAe,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,CAAC,CAAC;QAC1D,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,eAAe,CAAC,CAAC,CAAC,CAAC;QACvC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC;QAC7C,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,aAAa,CAAC,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC;IAC5D,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,qDAAqD,EAAE,KAAK,IAAI,EAAE;QACnE,WAAW;aACR,qBAAqB,CAAC,CAAC,QAAQ,CAAU,CAAC;aAC1C,qBAAqB,CAAC,CAAC,SAAS,CAAU,CAAC,CAAC;QAC/C,QAAQ;aACL,qBAAqB,CAAC,QAAQ,CAAC,IAAI,CAAU,CAAC;aAC9C,qBAAqB,CAAC,QAAQ,CAAC,KAAK,CAAU,CAAC,CAAC;QACnD,YAAY,CAAC,qBAAqB,CAAC,gBAAyB,CAAC,CAAC;QAE9D,MAAM,MAAM,GAAG,IAAI,kCAAe,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAC;QAC3E,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,eAAe,CAAC,CAAC,CAAC,CAAC;IACzC,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,0CAA0C,EAAE,KAAK,IAAI,EAAE;QACxD,WAAW,CAAC,qBAAqB,CAAC,CAAC,QAAQ,CAAU,CAAC,CAAC;QACvD,QAAQ,CAAC,qBAAqB,CAAC,QAAQ,CAAC,IAAI,CAAU,CAAC,CAAC;QAExD,MAAM,MAAM,GAAG,IAAI,kCAAe,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,SAAS,EAAE,KAAK,EAAE,CAAC,CAAC;QAC5E,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC;IAC/B,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,gCAAgC,EAAE,KAAK,IAAI,EAAE;QAC9C,WAAW,CAAC,qBAAqB,CAAC,CAAC,SAAS,EAAE,aAAa,CAAU,CAAC,CAAC;QACvE,QAAQ,CAAC,qBAAqB,CAAC,QAAQ,CAAC,KAAK,CAAU,CAAC,CAAC;QACzD,YAAY,CAAC,qBAAqB,CAAC,SAAkB,CAAC,CAAC;QAEvD,MAAM,MAAM,GAAG,IAAI,kCAAe,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,CAAC,CAAC;QAC1D,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC;QAC7B,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;IAC1C,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,mCAAmC,EAAE,KAAK,IAAI,EAAE;QACjD,WAAW,CAAC,qBAAqB,CAAC,CAAC,SAAS,EAAE,WAAW,CAAU,CAAC,CAAC;QACrE,QAAQ;aACL,qBAAqB,CAAC,QAAQ,CAAC,KAAK,CAAU,CAAC;aAC/C,qBAAqB,CAAC,QAAQ,CAAC,KAAK,CAAU,CAAC,CAAC;QACnD,YAAY,CAAC,qBAAqB,CAAC,cAAuB,CAAC,CAAC;QAE5D,MAAM,MAAM,GAAG,IAAI,kCAAe,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,iBAAiB,EAAE,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC;QACvF,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC;IAC/B,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,mCAAmC,EAAE,KAAK,IAAI,EAAE;QACjD,WAAW,CAAC,qBAAqB,CAAC,CAAC,SAAS,EAAE,WAAW,CAAU,CAAC,CAAC;QACrE,QAAQ;aACL,qBAAqB,CAAC,QAAQ,CAAC,KAAK,CAAU,CAAC;aAC/C,qBAAqB,CAAC,QAAQ,CAAC,KAAK,CAAU,CAAC,CAAC;QACnD,YAAY,CAAC,qBAAqB,CAAC,MAAe,CAAC,CAAC;QAEpD,MAAM,MAAM,GAAG,IAAI,kCAAe,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,iBAAiB,EAAE,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC;QACxF,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC;IAC/B,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,yBAAyB,EAAE,KAAK,IAAI,EAAE;QACvC,WAAW,CAAC,qBAAqB,CAAC,CAAC,OAAO,EAAE,OAAO,EAAE,OAAO,CAAU,CAAC,CAAC;QACxE,QAAQ;aACL,qBAAqB,CAAC,QAAQ,CAAC,KAAK,CAAU,CAAC;aAC/C,qBAAqB,CAAC,QAAQ,CAAC,KAAK,CAAU,CAAC;aAC/C,qBAAqB,CAAC,QAAQ,CAAC,KAAK,CAAU,CAAC,CAAC;QACnD,YAAY,CAAC,qBAAqB,CAAC,GAAY,CAAC,CAAC,qBAAqB,CAAC,GAAY,CAAC,CAAC;QAErF,MAAM,MAAM,GAAG,IAAI,kCAAe,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC,CAAC;QACvE,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,mBAAmB,CAAC,CAAC,CAAC,CAAC;IAC7C,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,mCAAmC,EAAE,KAAK,IAAI,EAAE;QACjD,WAAW,CAAC,qBAAqB,CAAC,CAAC,SAAS,CAAU,CAAC,CAAC;QACxD,QAAQ,CAAC,qBAAqB,CAAC,QAAQ,CAAC,KAAK,CAAU,CAAC,CAAC;QAEzD,MAAM,QAAQ,GAAG,CAAC,EAAE,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,EAAE,EAAE,CAAC,CAAC;QACvD,MAAM,OAAO,GAAG,IAAI,CAAC,EAAE,EAAE,CAAC,eAAe,CAAC,EAAE,IAAI,EAAE,IAAI,CAAC,EAAE,EAAE,CAAC,iBAAiB,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAC;QAE3F,MAAM,MAAM,GAAG,IAAI,kCAAe,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,OAAO,EAAE,EAAE,MAAM,EAAE,OAAO,EAAE,EAAE,CAAC,CAAC;QACxF,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC;QAC7B,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;QACvC,MAAM,CAAC,OAAO,CAAC,CAAC,oBAAoB,CAAC,MAAM,CAAC,gBAAgB,CAAC,SAAS,CAAC,CAAC,CAAC;IAC3E,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,wBAAwB,EAAE,KAAK,IAAI,EAAE;QACtC,WAAW,CAAC,qBAAqB,CAAC,CAAC,WAAW,CAAU,CAAC,CAAC;QAC1D,QAAQ,CAAC,qBAAqB,CAAC,QAAQ,CAAC,KAAK,CAAU,CAAC,CAAC;QACzD,YAAY,CAAC,qBAAqB,CAAC,SAAkB,CAAC,CAAC;QAEvD,MAAM,MAAM,GAAG,IAAI,kCAAe,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,CAAC,CAAC;QAC1D,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,eAAe,CAAC,CAAC,CAAC,CAAC;IACzC,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,0BAA0B,EAAE,KAAK,IAAI,EAAE;QACxC,WAAW,CAAC,qBAAqB,CAAC,CAAC,WAAW,CAAU,CAAC,CAAC;QAC1D,QAAQ,CAAC,qBAAqB,CAAC,QAAQ,CAAC,KAAK,CAAU,CAAC,CAAC;QACzD,YAAY,CAAC,qBAAqB,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,CAAC,CAAU,CAAC,CAAC;QAEjF,MAAM,MAAM,GAAG,IAAI,kCAAe,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,CAAC,CAAC;QAC1D,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,eAAe,CAAC,CAAC,CAAC,CAAC;IACzC,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,yBAAyB,EAAE,KAAK,IAAI,EAAE;QACvC,WAAW,CAAC,qBAAqB,CAAC,CAAC,UAAU,CAAU,CAAC,CAAC;QACzD,QAAQ,CAAC,qBAAqB,CAAC,QAAQ,CAAC,KAAK,CAAU,CAAC,CAAC;QACzD,YAAY,CAAC,qBAAqB,CAAC,UAAmB,CAAC,CAAC;QAExD,MAAM,MAAM,GAAG,IAAI,kCAAe,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,CAAC,CAAC;QAC1D,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,eAAe,CAAC,CAAC,CAAC,CAAC;IACzC,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,0BAA0B,EAAE,KAAK,IAAI,EAAE;QACxC,WAAW,CAAC,qBAAqB,CAAC,CAAC,WAAW,CAAU,CAAC,CAAC;QAC1D,QAAQ,CAAC,qBAAqB,CAAC,QAAQ,CAAC,KAAK,CAAU,CAAC,CAAC;QACzD,YAAY,CAAC,qBAAqB,CAAC,cAAuB,CAAC,CAAC;QAE5D,MAAM,MAAM,GAAG,IAAI,kCAAe,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,CAAC,CAAC;QAC1D,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,eAAe,CAAC,CAAC,CAAC,CAAC;IACzC,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,gDAAgD,EAAE,KAAK,IAAI,EAAE;QAC9D,WAAW,CAAC,qBAAqB,CAAC,CAAC,YAAY,CAAU,CAAC,CAAC;QAC3D,QAAQ,CAAC,qBAAqB,CAAC,QAAQ,CAAC,KAAK,CAAU,CAAC,CAAC;QAEzD,MAAM,MAAM,GAAG,IAAI,kCAAe,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,CAAC,CAAC;QAC1D,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC;IAC/B,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,8CAA8C,EAAE,KAAK,IAAI,EAAE;QAC5D,WAAW,CAAC,qBAAqB,CAAC,CAAC,SAAS,CAAU,CAAC,CAAC;QACxD,QAAQ,CAAC,qBAAqB,CAAC,QAAQ,CAAC,KAAK,CAAU,CAAC,CAAC;QACzD,YAAY,CAAC,qBAAqB,CAAC,IAAI,KAAK,CAAC,QAAQ,CAAU,CAAC,CAAC;QAEjE,MAAM,MAAM,GAAG,IAAI,kCAAe,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,CAAC,CAAC;QAC1D,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC;IAC/B,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,sCAAsC,EAAE,KAAK,IAAI,EAAE;QACpD,WAAW,CAAC,qBAAqB,CAAC,CAAC,WAAW,CAAU,CAAC,CAAC;QAC1D,QAAQ,CAAC,qBAAqB,CAAC,QAAQ,CAAC,KAAK,CAAU,CAAC,CAAC;QACzD,YAAY,CAAC,qBAAqB,CAAC,SAAkB,CAAC,CAAC;QAEvD,MAAM,MAAM,GAAG,IAAI,kCAAe,CAAC,EAAE,IAAI,EAAE,WAAW,EAAE,QAAQ,EAAE,EAAE,GAAG,EAAE,OAAO,EAAE,EAAE,CAAC,CAAC;QACtF,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IAC9C,CAAC,CAAC,CAAC;AACL,CAAC,CAAC,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"html-file.loader.test.d.ts","sourceRoot":"","sources":["../../../src/__tests__/loaders/html-file.loader.test.ts"],"names":[],"mappings":""}
|
|
@@ -0,0 +1,93 @@
|
|
|
1
|
+
"use strict";
|
|
2
|
+
Object.defineProperty(exports, "__esModule", { value: true });
|
|
3
|
+
jest.mock('fs/promises', () => ({
|
|
4
|
+
readFile: jest.fn(),
|
|
5
|
+
}));
|
|
6
|
+
const promises_1 = require("fs/promises");
|
|
7
|
+
const html_file_loader_1 = require("../../loaders/html-file.loader");
|
|
8
|
+
const mockReadFile = promises_1.readFile;
|
|
9
|
+
const HTML_BASIC = `<html><head><title>My Page</title></head><body><p>Hello world</p></body></html>`;
|
|
10
|
+
const HTML_WITH_SCRIPT = `<html><body><script>var x=1;</script><p>Visible text</p><style>.hide{}</style></body></html>`;
|
|
11
|
+
const HTML_ENTITIES = `<p>& < > " … A</p>`;
|
|
12
|
+
const HTML_MULTI_BLANK = `<p>Line 1</p>\n\n\n\n<p>Line 2</p>`;
|
|
13
|
+
describe('HTMLFileLoader', () => {
|
|
14
|
+
afterEach(() => jest.clearAllMocks());
|
|
15
|
+
it('throws when no path is provided', () => {
|
|
16
|
+
expect(() => new html_file_loader_1.HTMLFileLoader({})).toThrow('provide at least one path');
|
|
17
|
+
});
|
|
18
|
+
it('loads HTML and strips tags', async () => {
|
|
19
|
+
mockReadFile.mockResolvedValueOnce(HTML_BASIC);
|
|
20
|
+
const loader = new html_file_loader_1.HTMLFileLoader({ path: '/tmp/page.html' });
|
|
21
|
+
const docs = await loader.load();
|
|
22
|
+
expect(docs).toHaveLength(1);
|
|
23
|
+
expect(docs[0].content).toContain('Hello world');
|
|
24
|
+
expect(docs[0].content).not.toContain('<p>');
|
|
25
|
+
expect(docs[0].metadata?.loaderType).toBe('html');
|
|
26
|
+
expect(docs[0].metadata?.source).toBe('page.html');
|
|
27
|
+
});
|
|
28
|
+
it('extracts title from <title> tag', async () => {
|
|
29
|
+
mockReadFile.mockResolvedValueOnce(HTML_BASIC);
|
|
30
|
+
const loader = new html_file_loader_1.HTMLFileLoader({ path: '/tmp/page.html' });
|
|
31
|
+
const docs = await loader.load();
|
|
32
|
+
expect(docs[0].metadata?.title).toBe('My Page');
|
|
33
|
+
});
|
|
34
|
+
it('strips <script> and <style> blocks', async () => {
|
|
35
|
+
mockReadFile.mockResolvedValueOnce(HTML_WITH_SCRIPT);
|
|
36
|
+
const loader = new html_file_loader_1.HTMLFileLoader({ path: '/tmp/page.html', stripScripts: true });
|
|
37
|
+
const docs = await loader.load();
|
|
38
|
+
expect(docs[0].content).not.toContain('var x=1');
|
|
39
|
+
expect(docs[0].content).not.toContain('.hide');
|
|
40
|
+
expect(docs[0].content).toContain('Visible text');
|
|
41
|
+
});
|
|
42
|
+
it('does not strip scripts when stripScripts is false', async () => {
|
|
43
|
+
mockReadFile.mockResolvedValueOnce(HTML_WITH_SCRIPT);
|
|
44
|
+
const loader = new html_file_loader_1.HTMLFileLoader({ path: '/tmp/page.html', stripScripts: false });
|
|
45
|
+
const docs = await loader.load();
|
|
46
|
+
expect(docs[0].content).toContain('var x=1');
|
|
47
|
+
});
|
|
48
|
+
it('decodes HTML entities', async () => {
|
|
49
|
+
mockReadFile.mockResolvedValueOnce(HTML_ENTITIES);
|
|
50
|
+
const loader = new html_file_loader_1.HTMLFileLoader({ path: '/tmp/page.html' });
|
|
51
|
+
const docs = await loader.load();
|
|
52
|
+
expect(docs[0].content).toContain('&');
|
|
53
|
+
expect(docs[0].content).toContain('<');
|
|
54
|
+
expect(docs[0].content).toContain('>');
|
|
55
|
+
expect(docs[0].content).toContain('"');
|
|
56
|
+
expect(docs[0].content).toContain('...');
|
|
57
|
+
expect(docs[0].content).toContain('A'); // A
|
|
58
|
+
});
|
|
59
|
+
it('collapses multiple blank lines', async () => {
|
|
60
|
+
mockReadFile.mockResolvedValueOnce(HTML_MULTI_BLANK);
|
|
61
|
+
const loader = new html_file_loader_1.HTMLFileLoader({ path: '/tmp/page.html' });
|
|
62
|
+
const docs = await loader.load();
|
|
63
|
+
expect(docs[0].content).not.toMatch(/\n{3,}/);
|
|
64
|
+
});
|
|
65
|
+
it('loads multiple files', async () => {
|
|
66
|
+
mockReadFile
|
|
67
|
+
.mockResolvedValueOnce('<p>one</p>')
|
|
68
|
+
.mockResolvedValueOnce('<p>two</p>');
|
|
69
|
+
const loader = new html_file_loader_1.HTMLFileLoader({ paths: ['/tmp/a.html', '/tmp/b.html'] });
|
|
70
|
+
const docs = await loader.load();
|
|
71
|
+
expect(docs).toHaveLength(2);
|
|
72
|
+
});
|
|
73
|
+
it('merges extra metadata', async () => {
|
|
74
|
+
mockReadFile.mockResolvedValueOnce(HTML_BASIC);
|
|
75
|
+
const loader = new html_file_loader_1.HTMLFileLoader({ path: '/tmp/page.html', metadata: { tag: 'v1' } });
|
|
76
|
+
const docs = await loader.load();
|
|
77
|
+
expect(docs[0].metadata?.tag).toBe('v1');
|
|
78
|
+
});
|
|
79
|
+
it('handles HTML without a title tag gracefully', async () => {
|
|
80
|
+
mockReadFile.mockResolvedValueOnce('<p>no title here</p>');
|
|
81
|
+
const loader = new html_file_loader_1.HTMLFileLoader({ path: '/tmp/page.html' });
|
|
82
|
+
const docs = await loader.load();
|
|
83
|
+
expect(docs[0].metadata?.title).toBeUndefined();
|
|
84
|
+
});
|
|
85
|
+
it('does not collapse multiple blank lines when collapseWhitespace is false', async () => {
|
|
86
|
+
mockReadFile.mockResolvedValueOnce('<p>line1</p>\n\n\n\n<p>line2</p>');
|
|
87
|
+
const loader = new html_file_loader_1.HTMLFileLoader({ path: '/tmp/page.html', collapseWhitespace: false });
|
|
88
|
+
const docs = await loader.load();
|
|
89
|
+
// When collapseWhitespace is false, multiple blank lines are preserved
|
|
90
|
+
expect(docs[0].content).toMatch(/\n{3,}/);
|
|
91
|
+
});
|
|
92
|
+
});
|
|
93
|
+
//# sourceMappingURL=html-file.loader.test.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"html-file.loader.test.js","sourceRoot":"","sources":["../../../src/__tests__/loaders/html-file.loader.test.ts"],"names":[],"mappings":";;AAAA,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,GAAG,EAAE,CAAC,CAAC;IAC9B,QAAQ,EAAE,IAAI,CAAC,EAAE,EAAE;CACpB,CAAC,CAAC,CAAC;AAEJ,0CAAuC;AACvC,qEAAgE;AAEhE,MAAM,YAAY,GAAG,mBAAgD,CAAC;AAEtE,MAAM,UAAU,GAAG,iFAAiF,CAAC;AACrG,MAAM,gBAAgB,GAAG,8FAA8F,CAAC;AACxH,MAAM,aAAa,GAAG,qDAAqD,CAAC;AAC5E,MAAM,gBAAgB,GAAG,oCAAoC,CAAC;AAE9D,QAAQ,CAAC,gBAAgB,EAAE,GAAG,EAAE;IAC9B,SAAS,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,aAAa,EAAE,CAAC,CAAC;IAEtC,EAAE,CAAC,iCAAiC,EAAE,GAAG,EAAE;QACzC,MAAM,CAAC,GAAG,EAAE,CAAC,IAAI,iCAAc,CAAC,EAAE,CAAC,CAAC,CAAC,OAAO,CAAC,2BAA2B,CAAC,CAAC;IAC5E,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,4BAA4B,EAAE,KAAK,IAAI,EAAE;QAC1C,YAAY,CAAC,qBAAqB,CAAC,UAAmB,CAAC,CAAC;QACxD,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC,EAAE,IAAI,EAAE,gBAAgB,EAAE,CAAC,CAAC;QAC9D,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC;QAC7B,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,SAAS,CAAC,aAAa,CAAC,CAAC;QACjD,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,GAAG,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;QAC7C,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,UAAU,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QAClD,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC;IACrD,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,iCAAiC,EAAE,KAAK,IAAI,EAAE;QAC/C,YAAY,CAAC,qBAAqB,CAAC,UAAmB,CAAC,CAAC;QACxD,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC,EAAE,IAAI,EAAE,gBAAgB,EAAE,CAAC,CAAC;QAC9D,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;IAClD,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,oCAAoC,EAAE,KAAK,IAAI,EAAE;QAClD,YAAY,CAAC,qBAAqB,CAAC,gBAAyB,CAAC,CAAC;QAC9D,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC,EAAE,IAAI,EAAE,gBAAgB,EAAE,YAAY,EAAE,IAAI,EAAE,CAAC,CAAC;QAClF,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,GAAG,CAAC,SAAS,CAAC,SAAS,CAAC,CAAC;QACjD,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,GAAG,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC;QAC/C,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,SAAS,CAAC,cAAc,CAAC,CAAC;IACpD,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,mDAAmD,EAAE,KAAK,IAAI,EAAE;QACjE,YAAY,CAAC,qBAAqB,CAAC,gBAAyB,CAAC,CAAC;QAC9D,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC,EAAE,IAAI,EAAE,gBAAgB,EAAE,YAAY,EAAE,KAAK,EAAE,CAAC,CAAC;QACnF,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,SAAS,CAAC,SAAS,CAAC,CAAC;IAC/C,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,uBAAuB,EAAE,KAAK,IAAI,EAAE;QACrC,YAAY,CAAC,qBAAqB,CAAC,aAAsB,CAAC,CAAC;QAC3D,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC,EAAE,IAAI,EAAE,gBAAgB,EAAE,CAAC,CAAC;QAC9D,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC;QACvC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC;QACvC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC;QACvC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC;QACvC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;QACzC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC,CAAC,QAAQ;IAClD,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,gCAAgC,EAAE,KAAK,IAAI,EAAE;QAC9C,YAAY,CAAC,qBAAqB,CAAC,gBAAyB,CAAC,CAAC;QAC9D,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC,EAAE,IAAI,EAAE,gBAAgB,EAAE,CAAC,CAAC;QAC9D,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC;IAChD,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,sBAAsB,EAAE,KAAK,IAAI,EAAE;QACpC,YAAY;aACT,qBAAqB,CAAC,YAAqB,CAAC;aAC5C,qBAAqB,CAAC,YAAqB,CAAC,CAAC;QAChD,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC,EAAE,KAAK,EAAE,CAAC,aAAa,EAAE,aAAa,CAAC,EAAE,CAAC,CAAC;QAC7E,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC;IAC/B,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,uBAAuB,EAAE,KAAK,IAAI,EAAE;QACrC,YAAY,CAAC,qBAAqB,CAAC,UAAmB,CAAC,CAAC;QACxD,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC,EAAE,IAAI,EAAE,gBAAgB,EAAE,QAAQ,EAAE,EAAE,GAAG,EAAE,IAAI,EAAE,EAAE,CAAC,CAAC;QACvF,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IAC3C,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,6CAA6C,EAAE,KAAK,IAAI,EAAE;QAC3D,YAAY,CAAC,qBAAqB,CAAC,sBAA+B,CAAC,CAAC;QACpE,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC,EAAE,IAAI,EAAE,gBAAgB,EAAE,CAAC,CAAC;QAC9D,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,aAAa,EAAE,CAAC;IAClD,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,yEAAyE,EAAE,KAAK,IAAI,EAAE;QACvF,YAAY,CAAC,qBAAqB,CAAC,kCAA2C,CAAC,CAAC;QAChF,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC,EAAE,IAAI,EAAE,gBAAgB,EAAE,kBAAkB,EAAE,KAAK,EAAE,CAAC,CAAC;QACzF,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,uEAAuE;QACvE,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC;IAC5C,CAAC,CAAC,CAAC;AACL,CAAC,CAAC,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"json-file.loader.test.d.ts","sourceRoot":"","sources":["../../../src/__tests__/loaders/json-file.loader.test.ts"],"names":[],"mappings":""}
|
|
@@ -0,0 +1,84 @@
|
|
|
1
|
+
"use strict";
|
|
2
|
+
Object.defineProperty(exports, "__esModule", { value: true });
|
|
3
|
+
jest.mock('fs/promises', () => ({
|
|
4
|
+
readFile: jest.fn(),
|
|
5
|
+
}));
|
|
6
|
+
const promises_1 = require("fs/promises");
|
|
7
|
+
const json_file_loader_1 = require("../../loaders/json-file.loader");
|
|
8
|
+
const mockReadFile = promises_1.readFile;
|
|
9
|
+
describe('JSONFileLoader', () => {
|
|
10
|
+
afterEach(() => jest.clearAllMocks());
|
|
11
|
+
it('loads an array JSON file — one doc per element', async () => {
|
|
12
|
+
const data = JSON.stringify([
|
|
13
|
+
{ title: 'A', body: 'Text A' },
|
|
14
|
+
{ title: 'B', body: 'Text B' },
|
|
15
|
+
]);
|
|
16
|
+
mockReadFile.mockResolvedValueOnce(data);
|
|
17
|
+
const loader = new json_file_loader_1.JSONFileLoader({ path: '/tmp/items.json', textKey: 'body' });
|
|
18
|
+
const docs = await loader.load();
|
|
19
|
+
expect(docs).toHaveLength(2);
|
|
20
|
+
expect(docs[0].content).toBe('Text A');
|
|
21
|
+
expect(docs[0].metadata?.source).toBe('items.json');
|
|
22
|
+
expect(docs[0].metadata?.loaderType).toBe('json');
|
|
23
|
+
expect(docs[0].metadata?.index).toBe(0);
|
|
24
|
+
});
|
|
25
|
+
it('loads a single root object as one document', async () => {
|
|
26
|
+
mockReadFile.mockResolvedValueOnce(JSON.stringify({ message: 'hello' }));
|
|
27
|
+
const loader = new json_file_loader_1.JSONFileLoader({ path: '/tmp/obj.json', textKey: 'message' });
|
|
28
|
+
const docs = await loader.load();
|
|
29
|
+
expect(docs).toHaveLength(1);
|
|
30
|
+
expect(docs[0].content).toBe('hello');
|
|
31
|
+
});
|
|
32
|
+
it('stringifies element when textKey is absent', async () => {
|
|
33
|
+
mockReadFile.mockResolvedValueOnce(JSON.stringify([{ key: 'val' }]));
|
|
34
|
+
const loader = new json_file_loader_1.JSONFileLoader({ path: '/tmp/items.json' });
|
|
35
|
+
const docs = await loader.load();
|
|
36
|
+
expect(docs[0].content).toContain('"key"');
|
|
37
|
+
});
|
|
38
|
+
it('extracts metadataKeys into metadata', async () => {
|
|
39
|
+
const data = JSON.stringify([{ id: '1', title: 'My Title', body: 'content' }]);
|
|
40
|
+
mockReadFile.mockResolvedValueOnce(data);
|
|
41
|
+
const loader = new json_file_loader_1.JSONFileLoader({
|
|
42
|
+
path: '/tmp/items.json',
|
|
43
|
+
textKey: 'body',
|
|
44
|
+
metadataKeys: ['id', 'title'],
|
|
45
|
+
});
|
|
46
|
+
const docs = await loader.load();
|
|
47
|
+
expect(docs[0].metadata?.id).toBe('1');
|
|
48
|
+
expect(docs[0].metadata?.title).toBe('My Title');
|
|
49
|
+
});
|
|
50
|
+
it('navigates nested arrays via jsonPointer', async () => {
|
|
51
|
+
const data = JSON.stringify({ data: { results: [{ text: 'nested' }] } });
|
|
52
|
+
mockReadFile.mockResolvedValueOnce(data);
|
|
53
|
+
const loader = new json_file_loader_1.JSONFileLoader({
|
|
54
|
+
path: '/tmp/data.json',
|
|
55
|
+
textKey: 'text',
|
|
56
|
+
jsonPointer: 'data.results',
|
|
57
|
+
});
|
|
58
|
+
const docs = await loader.load();
|
|
59
|
+
expect(docs).toHaveLength(1);
|
|
60
|
+
expect(docs[0].content).toBe('nested');
|
|
61
|
+
});
|
|
62
|
+
it('throws when jsonPointer key is not found', async () => {
|
|
63
|
+
mockReadFile.mockResolvedValueOnce(JSON.stringify({ foo: 'bar' }));
|
|
64
|
+
const loader = new json_file_loader_1.JSONFileLoader({ path: '/tmp/data.json', jsonPointer: 'missing.key' });
|
|
65
|
+
await expect(loader.load()).rejects.toThrow('not found');
|
|
66
|
+
});
|
|
67
|
+
it('merges extra metadata', async () => {
|
|
68
|
+
mockReadFile.mockResolvedValueOnce(JSON.stringify([{ text: 'hi' }]));
|
|
69
|
+
const loader = new json_file_loader_1.JSONFileLoader({
|
|
70
|
+
path: '/tmp/data.json',
|
|
71
|
+
textKey: 'text',
|
|
72
|
+
metadata: { env: 'prod' },
|
|
73
|
+
});
|
|
74
|
+
const docs = await loader.load();
|
|
75
|
+
expect(docs[0].metadata?.env).toBe('prod');
|
|
76
|
+
});
|
|
77
|
+
it('stringifies non-object primitives in array', async () => {
|
|
78
|
+
mockReadFile.mockResolvedValueOnce(JSON.stringify(['hello', 'world']));
|
|
79
|
+
const loader = new json_file_loader_1.JSONFileLoader({ path: '/tmp/data.json' });
|
|
80
|
+
const docs = await loader.load();
|
|
81
|
+
expect(docs[0].content).toContain('hello');
|
|
82
|
+
});
|
|
83
|
+
});
|
|
84
|
+
//# sourceMappingURL=json-file.loader.test.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"json-file.loader.test.js","sourceRoot":"","sources":["../../../src/__tests__/loaders/json-file.loader.test.ts"],"names":[],"mappings":";;AAAA,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,GAAG,EAAE,CAAC,CAAC;IAC9B,QAAQ,EAAE,IAAI,CAAC,EAAE,EAAE;CACpB,CAAC,CAAC,CAAC;AAEJ,0CAAuC;AACvC,qEAAgE;AAEhE,MAAM,YAAY,GAAG,mBAAgD,CAAC;AAEtE,QAAQ,CAAC,gBAAgB,EAAE,GAAG,EAAE;IAC9B,SAAS,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,aAAa,EAAE,CAAC,CAAC;IAEtC,EAAE,CAAC,gDAAgD,EAAE,KAAK,IAAI,EAAE;QAC9D,MAAM,IAAI,GAAG,IAAI,CAAC,SAAS,CAAC;YAC1B,EAAE,KAAK,EAAE,GAAG,EAAE,IAAI,EAAE,QAAQ,EAAE;YAC9B,EAAE,KAAK,EAAE,GAAG,EAAE,IAAI,EAAE,QAAQ,EAAE;SAC/B,CAAC,CAAC;QACH,YAAY,CAAC,qBAAqB,CAAC,IAAa,CAAC,CAAC;QAClD,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC,EAAE,IAAI,EAAE,iBAAiB,EAAE,OAAO,EAAE,MAAM,EAAE,CAAC,CAAC;QAChF,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC;QAC7B,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;QACvC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAC,IAAI,CAAC,YAAY,CAAC,CAAC;QACpD,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,UAAU,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QAClD,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;IAC1C,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,4CAA4C,EAAE,KAAK,IAAI,EAAE;QAC1D,YAAY,CAAC,qBAAqB,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,OAAO,EAAE,OAAO,EAAE,CAAU,CAAC,CAAC;QAClF,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC,EAAE,IAAI,EAAE,eAAe,EAAE,OAAO,EAAE,SAAS,EAAE,CAAC,CAAC;QACjF,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC;QAC7B,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IACxC,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,4CAA4C,EAAE,KAAK,IAAI,EAAE;QAC1D,YAAY,CAAC,qBAAqB,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,CAAC,CAAU,CAAC,CAAC;QAC9E,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC,EAAE,IAAI,EAAE,iBAAiB,EAAE,CAAC,CAAC;QAC/D,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC;IAC7C,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,qCAAqC,EAAE,KAAK,IAAI,EAAE;QACnD,MAAM,IAAI,GAAG,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,EAAE,EAAE,GAAG,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,SAAS,EAAE,CAAC,CAAC,CAAC;QAC/E,YAAY,CAAC,qBAAqB,CAAC,IAAa,CAAC,CAAC;QAClD,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC;YAChC,IAAI,EAAE,iBAAiB;YACvB,OAAO,EAAE,MAAM;YACf,YAAY,EAAE,CAAC,IAAI,EAAE,OAAO,CAAC;SAC9B,CAAC,CAAC;QACH,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QACvC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;IACnD,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,yCAAyC,EAAE,KAAK,IAAI,EAAE;QACvD,MAAM,IAAI,GAAG,IAAI,CAAC,SAAS,CAAC,EAAE,IAAI,EAAE,EAAE,OAAO,EAAE,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC;QACzE,YAAY,CAAC,qBAAqB,CAAC,IAAa,CAAC,CAAC;QAClD,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC;YAChC,IAAI,EAAE,gBAAgB;YACtB,OAAO,EAAE,MAAM;YACf,WAAW,EAAE,cAAc;SAC5B,CAAC,CAAC;QACH,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC;QAC7B,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;IACzC,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,0CAA0C,EAAE,KAAK,IAAI,EAAE;QACxD,YAAY,CAAC,qBAAqB,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,CAAU,CAAC,CAAC;QAC5E,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC,EAAE,IAAI,EAAE,gBAAgB,EAAE,WAAW,EAAE,aAAa,EAAE,CAAC,CAAC;QAC1F,MAAM,MAAM,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC,CAAC,OAAO,CAAC,OAAO,CAAC,WAAW,CAAC,CAAC;IAC3D,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,uBAAuB,EAAE,KAAK,IAAI,EAAE;QACrC,YAAY,CAAC,qBAAqB,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC,CAAU,CAAC,CAAC;QAC9E,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC;YAChC,IAAI,EAAE,gBAAgB;YACtB,OAAO,EAAE,MAAM;YACf,QAAQ,EAAE,EAAE,GAAG,EAAE,MAAM,EAAE;SAC1B,CAAC,CAAC;QACH,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IAC7C,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,4CAA4C,EAAE,KAAK,IAAI,EAAE;QAC1D,YAAY,CAAC,qBAAqB,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,OAAO,EAAE,OAAO,CAAC,CAAU,CAAC,CAAC;QAChF,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC,EAAE,IAAI,EAAE,gBAAgB,EAAE,CAAC,CAAC;QAC9D,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC;IAC7C,CAAC,CAAC,CAAC;AACL,CAAC,CAAC,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"markdown-file.loader.test.d.ts","sourceRoot":"","sources":["../../../src/__tests__/loaders/markdown-file.loader.test.ts"],"names":[],"mappings":""}
|
|
@@ -0,0 +1,83 @@
|
|
|
1
|
+
"use strict";
|
|
2
|
+
Object.defineProperty(exports, "__esModule", { value: true });
|
|
3
|
+
jest.mock('fs/promises', () => ({
|
|
4
|
+
readFile: jest.fn(),
|
|
5
|
+
}));
|
|
6
|
+
const promises_1 = require("fs/promises");
|
|
7
|
+
const markdown_file_loader_1 = require("../../loaders/markdown-file.loader");
|
|
8
|
+
const mockReadFile = promises_1.readFile;
|
|
9
|
+
const MD_SIMPLE = `# Title\n\nSome content here.`;
|
|
10
|
+
const MD_WITH_FRONTMATTER = `---\ntitle: My Doc\nauthor: Alice\n---\n# Heading\n\nBody text.`;
|
|
11
|
+
const MD_H2_SPLIT = `## Section One\n\nContent one.\n\n## Section Two\n\nContent two.`;
|
|
12
|
+
const MD_H1_SPLIT = `# Chapter One\n\nText one.\n\n# Chapter Two\n\nText two.`;
|
|
13
|
+
describe('MarkdownFileLoader', () => {
|
|
14
|
+
afterEach(() => jest.clearAllMocks());
|
|
15
|
+
it('throws when no path is provided', () => {
|
|
16
|
+
expect(() => new markdown_file_loader_1.MarkdownFileLoader({})).toThrow('provide at least one path');
|
|
17
|
+
});
|
|
18
|
+
it('loads a single markdown file as one document', async () => {
|
|
19
|
+
mockReadFile.mockResolvedValueOnce(MD_SIMPLE);
|
|
20
|
+
const loader = new markdown_file_loader_1.MarkdownFileLoader({ path: '/tmp/readme.md' });
|
|
21
|
+
const docs = await loader.load();
|
|
22
|
+
expect(docs).toHaveLength(1);
|
|
23
|
+
expect(docs[0].metadata?.source).toBe('readme.md');
|
|
24
|
+
expect(docs[0].metadata?.loaderType).toBe('markdown');
|
|
25
|
+
expect(docs[0].content).toContain('Some content');
|
|
26
|
+
});
|
|
27
|
+
it('parses YAML front-matter into metadata', async () => {
|
|
28
|
+
mockReadFile.mockResolvedValueOnce(MD_WITH_FRONTMATTER);
|
|
29
|
+
const loader = new markdown_file_loader_1.MarkdownFileLoader({ path: '/tmp/doc.md' });
|
|
30
|
+
const docs = await loader.load();
|
|
31
|
+
expect(docs[0].metadata?.title).toBe('My Doc');
|
|
32
|
+
expect(docs[0].metadata?.author).toBe('Alice');
|
|
33
|
+
expect(docs[0].content).not.toContain('---');
|
|
34
|
+
});
|
|
35
|
+
it('skips front-matter parsing when parseFrontMatter is false', async () => {
|
|
36
|
+
mockReadFile.mockResolvedValueOnce(MD_WITH_FRONTMATTER);
|
|
37
|
+
const loader = new markdown_file_loader_1.MarkdownFileLoader({ path: '/tmp/doc.md', parseFrontMatter: false });
|
|
38
|
+
const docs = await loader.load();
|
|
39
|
+
expect(docs[0].metadata?.title).toBeUndefined();
|
|
40
|
+
expect(docs[0].content).toContain('---');
|
|
41
|
+
});
|
|
42
|
+
it('splits on h2 headings', async () => {
|
|
43
|
+
mockReadFile.mockResolvedValueOnce(MD_H2_SPLIT);
|
|
44
|
+
const loader = new markdown_file_loader_1.MarkdownFileLoader({ path: '/tmp/doc.md', splitByHeading: 'h2' });
|
|
45
|
+
const docs = await loader.load();
|
|
46
|
+
expect(docs).toHaveLength(2);
|
|
47
|
+
expect(docs[0].metadata?.heading).toBe('Section One');
|
|
48
|
+
expect(docs[1].metadata?.heading).toBe('Section Two');
|
|
49
|
+
expect(docs[0].content).toContain('Content one');
|
|
50
|
+
expect(docs[1].content).toContain('Content two');
|
|
51
|
+
});
|
|
52
|
+
it('splits on h1 headings', async () => {
|
|
53
|
+
mockReadFile.mockResolvedValueOnce(MD_H1_SPLIT);
|
|
54
|
+
const loader = new markdown_file_loader_1.MarkdownFileLoader({ path: '/tmp/doc.md', splitByHeading: 'h1' });
|
|
55
|
+
const docs = await loader.load();
|
|
56
|
+
expect(docs).toHaveLength(2);
|
|
57
|
+
expect(docs[0].metadata?.heading).toBe('Chapter One');
|
|
58
|
+
});
|
|
59
|
+
it('loads multiple files', async () => {
|
|
60
|
+
mockReadFile
|
|
61
|
+
.mockResolvedValueOnce('file one content')
|
|
62
|
+
.mockResolvedValueOnce('file two content');
|
|
63
|
+
const loader = new markdown_file_loader_1.MarkdownFileLoader({ paths: ['/tmp/a.md', '/tmp/b.md'] });
|
|
64
|
+
const docs = await loader.load();
|
|
65
|
+
expect(docs).toHaveLength(2);
|
|
66
|
+
expect(docs[0].metadata?.source).toBe('a.md');
|
|
67
|
+
expect(docs[1].metadata?.source).toBe('b.md');
|
|
68
|
+
});
|
|
69
|
+
it('merges extra metadata', async () => {
|
|
70
|
+
mockReadFile.mockResolvedValueOnce(MD_SIMPLE);
|
|
71
|
+
const loader = new markdown_file_loader_1.MarkdownFileLoader({ path: '/tmp/doc.md', metadata: { env: 'test' } });
|
|
72
|
+
const docs = await loader.load();
|
|
73
|
+
expect(docs[0].metadata?.env).toBe('test');
|
|
74
|
+
});
|
|
75
|
+
it('handles front-matter with quoted values', async () => {
|
|
76
|
+
const md = `---\ntitle: "Quoted Title"\n---\nContent`;
|
|
77
|
+
mockReadFile.mockResolvedValueOnce(md);
|
|
78
|
+
const loader = new markdown_file_loader_1.MarkdownFileLoader({ path: '/tmp/doc.md' });
|
|
79
|
+
const docs = await loader.load();
|
|
80
|
+
expect(docs[0].metadata?.title).toBe('Quoted Title');
|
|
81
|
+
});
|
|
82
|
+
});
|
|
83
|
+
//# sourceMappingURL=markdown-file.loader.test.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"markdown-file.loader.test.js","sourceRoot":"","sources":["../../../src/__tests__/loaders/markdown-file.loader.test.ts"],"names":[],"mappings":";;AAAA,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,GAAG,EAAE,CAAC,CAAC;IAC9B,QAAQ,EAAE,IAAI,CAAC,EAAE,EAAE;CACpB,CAAC,CAAC,CAAC;AAEJ,0CAAuC;AACvC,6EAAwE;AAExE,MAAM,YAAY,GAAG,mBAAgD,CAAC;AAEtE,MAAM,SAAS,GAAG,+BAA+B,CAAC;AAClD,MAAM,mBAAmB,GAAG,iEAAiE,CAAC;AAC9F,MAAM,WAAW,GAAG,kEAAkE,CAAC;AACvF,MAAM,WAAW,GAAG,0DAA0D,CAAC;AAE/E,QAAQ,CAAC,oBAAoB,EAAE,GAAG,EAAE;IAClC,SAAS,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,aAAa,EAAE,CAAC,CAAC;IAEtC,EAAE,CAAC,iCAAiC,EAAE,GAAG,EAAE;QACzC,MAAM,CAAC,GAAG,EAAE,CAAC,IAAI,yCAAkB,CAAC,EAAE,CAAC,CAAC,CAAC,OAAO,CAAC,2BAA2B,CAAC,CAAC;IAChF,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,8CAA8C,EAAE,KAAK,IAAI,EAAE;QAC5D,YAAY,CAAC,qBAAqB,CAAC,SAAkB,CAAC,CAAC;QACvD,MAAM,MAAM,GAAG,IAAI,yCAAkB,CAAC,EAAE,IAAI,EAAE,gBAAgB,EAAE,CAAC,CAAC;QAClE,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC;QAC7B,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC;QACnD,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,UAAU,CAAC,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;QACtD,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,SAAS,CAAC,cAAc,CAAC,CAAC;IACpD,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,wCAAwC,EAAE,KAAK,IAAI,EAAE;QACtD,YAAY,CAAC,qBAAqB,CAAC,mBAA4B,CAAC,CAAC;QACjE,MAAM,MAAM,GAAG,IAAI,yCAAkB,CAAC,EAAE,IAAI,EAAE,aAAa,EAAE,CAAC,CAAC;QAC/D,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;QAC/C,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;QAC/C,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,GAAG,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;IAC/C,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,2DAA2D,EAAE,KAAK,IAAI,EAAE;QACzE,YAAY,CAAC,qBAAqB,CAAC,mBAA4B,CAAC,CAAC;QACjE,MAAM,MAAM,GAAG,IAAI,yCAAkB,CAAC,EAAE,IAAI,EAAE,aAAa,EAAE,gBAAgB,EAAE,KAAK,EAAE,CAAC,CAAC;QACxF,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,aAAa,EAAE,CAAC;QAChD,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;IAC3C,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,uBAAuB,EAAE,KAAK,IAAI,EAAE;QACrC,YAAY,CAAC,qBAAqB,CAAC,WAAoB,CAAC,CAAC;QACzD,MAAM,MAAM,GAAG,IAAI,yCAAkB,CAAC,EAAE,IAAI,EAAE,aAAa,EAAE,cAAc,EAAE,IAAI,EAAE,CAAC,CAAC;QACrF,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC;QAC7B,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC,aAAa,CAAC,CAAC;QACtD,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC,aAAa,CAAC,CAAC;QACtD,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,SAAS,CAAC,aAAa,CAAC,CAAC;QACjD,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,SAAS,CAAC,aAAa,CAAC,CAAC;IACnD,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,uBAAuB,EAAE,KAAK,IAAI,EAAE;QACrC,YAAY,CAAC,qBAAqB,CAAC,WAAoB,CAAC,CAAC;QACzD,MAAM,MAAM,GAAG,IAAI,yCAAkB,CAAC,EAAE,IAAI,EAAE,aAAa,EAAE,cAAc,EAAE,IAAI,EAAE,CAAC,CAAC;QACrF,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC;QAC7B,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC,aAAa,CAAC,CAAC;IACxD,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,sBAAsB,EAAE,KAAK,IAAI,EAAE;QACpC,YAAY;aACT,qBAAqB,CAAC,kBAA2B,CAAC;aAClD,qBAAqB,CAAC,kBAA2B,CAAC,CAAC;QACtD,MAAM,MAAM,GAAG,IAAI,yCAAkB,CAAC,EAAE,KAAK,EAAE,CAAC,WAAW,EAAE,WAAW,CAAC,EAAE,CAAC,CAAC;QAC7E,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC;QAC7B,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QAC9C,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IAChD,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,uBAAuB,EAAE,KAAK,IAAI,EAAE;QACrC,YAAY,CAAC,qBAAqB,CAAC,SAAkB,CAAC,CAAC;QACvD,MAAM,MAAM,GAAG,IAAI,yCAAkB,CAAC,EAAE,IAAI,EAAE,aAAa,EAAE,QAAQ,EAAE,EAAE,GAAG,EAAE,MAAM,EAAE,EAAE,CAAC,CAAC;QAC1F,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IAC7C,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,yCAAyC,EAAE,KAAK,IAAI,EAAE;QACvD,MAAM,EAAE,GAAG,0CAA0C,CAAC;QACtD,YAAY,CAAC,qBAAqB,CAAC,EAAW,CAAC,CAAC;QAChD,MAAM,MAAM,GAAG,IAAI,yCAAkB,CAAC,EAAE,IAAI,EAAE,aAAa,EAAE,CAAC,CAAC;QAC/D,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC;IACvD,CAAC,CAAC,CAAC;AACL,CAAC,CAAC,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"text-file.loader.test.d.ts","sourceRoot":"","sources":["../../../src/__tests__/loaders/text-file.loader.test.ts"],"names":[],"mappings":""}
|
|
@@ -0,0 +1,50 @@
|
|
|
1
|
+
"use strict";
|
|
2
|
+
Object.defineProperty(exports, "__esModule", { value: true });
|
|
3
|
+
jest.mock('fs/promises', () => ({
|
|
4
|
+
readFile: jest.fn(),
|
|
5
|
+
}));
|
|
6
|
+
const promises_1 = require("fs/promises");
|
|
7
|
+
const text_file_loader_1 = require("../../loaders/text-file.loader");
|
|
8
|
+
const mockReadFile = promises_1.readFile;
|
|
9
|
+
describe('TextFileLoader', () => {
|
|
10
|
+
afterEach(() => jest.clearAllMocks());
|
|
11
|
+
it('throws when no path is provided', () => {
|
|
12
|
+
expect(() => new text_file_loader_1.TextFileLoader({})).toThrow('provide at least one file path');
|
|
13
|
+
});
|
|
14
|
+
it('loads a single file', async () => {
|
|
15
|
+
mockReadFile.mockResolvedValueOnce('hello world');
|
|
16
|
+
const loader = new text_file_loader_1.TextFileLoader({ path: '/tmp/notes.txt' });
|
|
17
|
+
const docs = await loader.load();
|
|
18
|
+
expect(docs).toHaveLength(1);
|
|
19
|
+
expect(docs[0].content).toBe('hello world');
|
|
20
|
+
expect(docs[0].metadata?.source).toBe('notes.txt');
|
|
21
|
+
expect(docs[0].metadata?.loaderType).toBe('text');
|
|
22
|
+
expect(docs[0].metadata?.filePath).toBe('/tmp/notes.txt');
|
|
23
|
+
});
|
|
24
|
+
it('loads multiple files', async () => {
|
|
25
|
+
mockReadFile
|
|
26
|
+
.mockResolvedValueOnce('file one')
|
|
27
|
+
.mockResolvedValueOnce('file two');
|
|
28
|
+
const loader = new text_file_loader_1.TextFileLoader({ paths: ['/tmp/a.txt', '/tmp/b.txt'] });
|
|
29
|
+
const docs = await loader.load();
|
|
30
|
+
expect(docs).toHaveLength(2);
|
|
31
|
+
expect(docs[0].content).toBe('file one');
|
|
32
|
+
expect(docs[1].content).toBe('file two');
|
|
33
|
+
});
|
|
34
|
+
it('merges extra metadata into each document', async () => {
|
|
35
|
+
mockReadFile.mockResolvedValueOnce('content');
|
|
36
|
+
const loader = new text_file_loader_1.TextFileLoader({
|
|
37
|
+
path: '/tmp/doc.txt',
|
|
38
|
+
metadata: { project: 'hazeljs' },
|
|
39
|
+
});
|
|
40
|
+
const docs = await loader.load();
|
|
41
|
+
expect(docs[0].metadata?.project).toBe('hazeljs');
|
|
42
|
+
});
|
|
43
|
+
it('trims whitespace from content', async () => {
|
|
44
|
+
mockReadFile.mockResolvedValueOnce(' trimmed ');
|
|
45
|
+
const loader = new text_file_loader_1.TextFileLoader({ path: '/tmp/f.txt' });
|
|
46
|
+
const docs = await loader.load();
|
|
47
|
+
expect(docs[0].content).toBe('trimmed');
|
|
48
|
+
});
|
|
49
|
+
});
|
|
50
|
+
//# sourceMappingURL=text-file.loader.test.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"text-file.loader.test.js","sourceRoot":"","sources":["../../../src/__tests__/loaders/text-file.loader.test.ts"],"names":[],"mappings":";;AAAA,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,GAAG,EAAE,CAAC,CAAC;IAC9B,QAAQ,EAAE,IAAI,CAAC,EAAE,EAAE;CACpB,CAAC,CAAC,CAAC;AAEJ,0CAAuC;AACvC,qEAAgE;AAEhE,MAAM,YAAY,GAAG,mBAAgD,CAAC;AAEtE,QAAQ,CAAC,gBAAgB,EAAE,GAAG,EAAE;IAC9B,SAAS,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,aAAa,EAAE,CAAC,CAAC;IAEtC,EAAE,CAAC,iCAAiC,EAAE,GAAG,EAAE;QACzC,MAAM,CAAC,GAAG,EAAE,CAAC,IAAI,iCAAc,CAAC,EAAE,CAAC,CAAC,CAAC,OAAO,CAAC,gCAAgC,CAAC,CAAC;IACjF,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,qBAAqB,EAAE,KAAK,IAAI,EAAE;QACnC,YAAY,CAAC,qBAAqB,CAAC,aAAsB,CAAC,CAAC;QAC3D,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC,EAAE,IAAI,EAAE,gBAAgB,EAAE,CAAC,CAAC;QAC9D,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC;QAC7B,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,aAAa,CAAC,CAAC;QAC5C,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC;QACnD,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,UAAU,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QAClD,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,QAAQ,CAAC,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAC;IAC5D,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,sBAAsB,EAAE,KAAK,IAAI,EAAE;QACpC,YAAY;aACT,qBAAqB,CAAC,UAAmB,CAAC;aAC1C,qBAAqB,CAAC,UAAmB,CAAC,CAAC;QAC9C,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC,EAAE,KAAK,EAAE,CAAC,YAAY,EAAE,YAAY,CAAC,EAAE,CAAC,CAAC;QAC3E,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC;QAC7B,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;QACzC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;IAC3C,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,0CAA0C,EAAE,KAAK,IAAI,EAAE;QACxD,YAAY,CAAC,qBAAqB,CAAC,SAAkB,CAAC,CAAC;QACvD,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC;YAChC,IAAI,EAAE,cAAc;YACpB,QAAQ,EAAE,EAAE,OAAO,EAAE,SAAS,EAAE;SACjC,CAAC,CAAC;QACH,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;IACpD,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,+BAA+B,EAAE,KAAK,IAAI,EAAE;QAC7C,YAAY,CAAC,qBAAqB,CAAC,aAAsB,CAAC,CAAC;QAC3D,MAAM,MAAM,GAAG,IAAI,iCAAc,CAAC,EAAE,IAAI,EAAE,YAAY,EAAE,CAAC,CAAC;QAC1D,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,IAAI,EAAE,CAAC;QACjC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;IAC1C,CAAC,CAAC,CAAC;AACL,CAAC,CAAC,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"rag-pipeline.test.d.ts","sourceRoot":"","sources":["../../src/__tests__/rag-pipeline.test.ts"],"names":[],"mappings":""}
|
|
@@ -0,0 +1,210 @@
|
|
|
1
|
+
"use strict";
|
|
2
|
+
Object.defineProperty(exports, "__esModule", { value: true });
|
|
3
|
+
const rag_pipeline_1 = require("../rag-pipeline");
|
|
4
|
+
const types_1 = require("../types");
|
|
5
|
+
function makeSearchResult(id, content, score = 0.9) {
|
|
6
|
+
return { id, content, score, metadata: { source: `${id}.txt` }, embedding: [0.1, 0.2, 0.3] };
|
|
7
|
+
}
|
|
8
|
+
function makeVectorStore(results = []) {
|
|
9
|
+
return {
|
|
10
|
+
initialize: jest.fn().mockResolvedValue(undefined),
|
|
11
|
+
addDocuments: jest.fn().mockResolvedValue(['id1']),
|
|
12
|
+
search: jest.fn().mockResolvedValue(results),
|
|
13
|
+
searchByVector: jest.fn().mockResolvedValue(results),
|
|
14
|
+
deleteDocuments: jest.fn().mockResolvedValue(undefined),
|
|
15
|
+
updateDocument: jest.fn().mockResolvedValue(undefined),
|
|
16
|
+
getDocument: jest.fn().mockResolvedValue(undefined),
|
|
17
|
+
clear: jest.fn().mockResolvedValue(undefined),
|
|
18
|
+
};
|
|
19
|
+
}
|
|
20
|
+
function makeEmbeddingProvider() {
|
|
21
|
+
return {
|
|
22
|
+
embed: jest.fn().mockResolvedValue([0.1, 0.2, 0.3]),
|
|
23
|
+
embedBatch: jest.fn().mockResolvedValue([[0.1, 0.2, 0.3]]),
|
|
24
|
+
getDimension: jest.fn().mockReturnValue(3),
|
|
25
|
+
};
|
|
26
|
+
}
|
|
27
|
+
function makeTextSplitter() {
|
|
28
|
+
return {
|
|
29
|
+
split: jest.fn().mockReturnValue(['chunk 1', 'chunk 2']),
|
|
30
|
+
splitDocuments: jest
|
|
31
|
+
.fn()
|
|
32
|
+
.mockImplementation((docs) => docs.map((d) => ({ ...d, content: d.content + ' (split)' }))),
|
|
33
|
+
};
|
|
34
|
+
}
|
|
35
|
+
describe('RAGPipeline', () => {
|
|
36
|
+
describe('initialize', () => {
|
|
37
|
+
it('calls vectorStore.initialize', async () => {
|
|
38
|
+
const vs = makeVectorStore();
|
|
39
|
+
const pipeline = new rag_pipeline_1.RAGPipeline({
|
|
40
|
+
vectorStore: vs,
|
|
41
|
+
embeddingProvider: makeEmbeddingProvider(),
|
|
42
|
+
});
|
|
43
|
+
await pipeline.initialize();
|
|
44
|
+
expect(vs.initialize).toHaveBeenCalled();
|
|
45
|
+
});
|
|
46
|
+
});
|
|
47
|
+
describe('addDocuments', () => {
|
|
48
|
+
it('adds documents to the vector store', async () => {
|
|
49
|
+
const vs = makeVectorStore();
|
|
50
|
+
const pipeline = new rag_pipeline_1.RAGPipeline({
|
|
51
|
+
vectorStore: vs,
|
|
52
|
+
embeddingProvider: makeEmbeddingProvider(),
|
|
53
|
+
});
|
|
54
|
+
const docs = [{ id: 'd1', content: 'Hello world', metadata: {} }];
|
|
55
|
+
await pipeline.addDocuments(docs);
|
|
56
|
+
expect(vs.addDocuments).toHaveBeenCalledWith(docs);
|
|
57
|
+
});
|
|
58
|
+
it('splits documents when textSplitter is configured', async () => {
|
|
59
|
+
const vs = makeVectorStore();
|
|
60
|
+
const splitter = makeTextSplitter();
|
|
61
|
+
const pipeline = new rag_pipeline_1.RAGPipeline({
|
|
62
|
+
vectorStore: vs,
|
|
63
|
+
embeddingProvider: makeEmbeddingProvider(),
|
|
64
|
+
textSplitter: splitter,
|
|
65
|
+
});
|
|
66
|
+
const docs = [{ content: 'Hello world', metadata: {} }];
|
|
67
|
+
await pipeline.addDocuments(docs);
|
|
68
|
+
expect(splitter.splitDocuments).toHaveBeenCalledWith(docs);
|
|
69
|
+
expect(vs.addDocuments).toHaveBeenCalledWith(expect.arrayContaining([
|
|
70
|
+
expect.objectContaining({ content: expect.stringContaining('(split)') }),
|
|
71
|
+
]));
|
|
72
|
+
});
|
|
73
|
+
});
|
|
74
|
+
describe('query', () => {
|
|
75
|
+
it('returns context when no LLM is configured', async () => {
|
|
76
|
+
const vs = makeVectorStore([makeSearchResult('d1', 'TypeScript is great')]);
|
|
77
|
+
const pipeline = new rag_pipeline_1.RAGPipeline({
|
|
78
|
+
vectorStore: vs,
|
|
79
|
+
embeddingProvider: makeEmbeddingProvider(),
|
|
80
|
+
});
|
|
81
|
+
const response = await pipeline.query('What is TypeScript?');
|
|
82
|
+
expect(response.answer).toContain('TypeScript is great');
|
|
83
|
+
expect(response.sources).toHaveLength(1);
|
|
84
|
+
expect(response.context).toContain('TypeScript');
|
|
85
|
+
});
|
|
86
|
+
it('calls LLM when llmFunction and llmPrompt are both set', async () => {
|
|
87
|
+
const vs = makeVectorStore([makeSearchResult('d1', 'TypeScript content')]);
|
|
88
|
+
const llm = jest.fn().mockResolvedValue('Generated answer');
|
|
89
|
+
const pipeline = new rag_pipeline_1.RAGPipeline({ vectorStore: vs, embeddingProvider: makeEmbeddingProvider() }, llm);
|
|
90
|
+
const response = await pipeline.query('query', {
|
|
91
|
+
llmPrompt: 'Context: {context}\nQ: {query}',
|
|
92
|
+
});
|
|
93
|
+
expect(llm).toHaveBeenCalled();
|
|
94
|
+
expect(response.answer).toBe('Generated answer');
|
|
95
|
+
});
|
|
96
|
+
it('does not call LLM when llmPrompt is missing', async () => {
|
|
97
|
+
const vs = makeVectorStore([makeSearchResult('d1', 'content')]);
|
|
98
|
+
const llm = jest.fn().mockResolvedValue('answer');
|
|
99
|
+
const pipeline = new rag_pipeline_1.RAGPipeline({ vectorStore: vs, embeddingProvider: makeEmbeddingProvider() }, llm);
|
|
100
|
+
await pipeline.query('query');
|
|
101
|
+
expect(llm).not.toHaveBeenCalled();
|
|
102
|
+
});
|
|
103
|
+
it('excludes context when includeContext is false', async () => {
|
|
104
|
+
const vs = makeVectorStore([makeSearchResult('d1', 'content')]);
|
|
105
|
+
const pipeline = new rag_pipeline_1.RAGPipeline({
|
|
106
|
+
vectorStore: vs,
|
|
107
|
+
embeddingProvider: makeEmbeddingProvider(),
|
|
108
|
+
});
|
|
109
|
+
const response = await pipeline.query('query', { includeContext: false });
|
|
110
|
+
expect(response.context).toBe('');
|
|
111
|
+
});
|
|
112
|
+
it('uses config.topK when options.topK is omitted', async () => {
|
|
113
|
+
const vs = makeVectorStore([]);
|
|
114
|
+
const pipeline = new rag_pipeline_1.RAGPipeline({
|
|
115
|
+
vectorStore: vs,
|
|
116
|
+
embeddingProvider: makeEmbeddingProvider(),
|
|
117
|
+
topK: 3,
|
|
118
|
+
});
|
|
119
|
+
await pipeline.query('query');
|
|
120
|
+
expect(vs.search).toHaveBeenCalledWith('query', expect.objectContaining({ topK: 3 }));
|
|
121
|
+
});
|
|
122
|
+
});
|
|
123
|
+
describe('retrieve strategies', () => {
|
|
124
|
+
it('SIMILARITY strategy calls vectorStore.search', async () => {
|
|
125
|
+
const vs = makeVectorStore([]);
|
|
126
|
+
const pipeline = new rag_pipeline_1.RAGPipeline({
|
|
127
|
+
vectorStore: vs,
|
|
128
|
+
embeddingProvider: makeEmbeddingProvider(),
|
|
129
|
+
});
|
|
130
|
+
await pipeline.retrieve('query', {}, types_1.RetrievalStrategy.SIMILARITY);
|
|
131
|
+
expect(vs.search).toHaveBeenCalled();
|
|
132
|
+
});
|
|
133
|
+
it('HYBRID strategy falls back to vectorStore.search', async () => {
|
|
134
|
+
const vs = makeVectorStore([]);
|
|
135
|
+
const pipeline = new rag_pipeline_1.RAGPipeline({
|
|
136
|
+
vectorStore: vs,
|
|
137
|
+
embeddingProvider: makeEmbeddingProvider(),
|
|
138
|
+
});
|
|
139
|
+
await pipeline.retrieve('query', {}, types_1.RetrievalStrategy.HYBRID);
|
|
140
|
+
expect(vs.search).toHaveBeenCalled();
|
|
141
|
+
});
|
|
142
|
+
it('MMR strategy gets 3x candidates then runs MMR algorithm', async () => {
|
|
143
|
+
const results = [
|
|
144
|
+
makeSearchResult('d1', 'content 1'),
|
|
145
|
+
makeSearchResult('d2', 'content 2'),
|
|
146
|
+
makeSearchResult('d3', 'content 3'),
|
|
147
|
+
];
|
|
148
|
+
const vs = makeVectorStore(results);
|
|
149
|
+
const embedder = makeEmbeddingProvider();
|
|
150
|
+
const pipeline = new rag_pipeline_1.RAGPipeline({ vectorStore: vs, embeddingProvider: embedder });
|
|
151
|
+
const selected = await pipeline.retrieve('query', { topK: 2 }, types_1.RetrievalStrategy.MMR);
|
|
152
|
+
expect(vs.search).toHaveBeenCalledWith('query', expect.objectContaining({ topK: 6, includeEmbedding: true }));
|
|
153
|
+
expect(selected.length).toBeLessThanOrEqual(2);
|
|
154
|
+
});
|
|
155
|
+
it('MMR returns empty array when no candidates', async () => {
|
|
156
|
+
const vs = makeVectorStore([]);
|
|
157
|
+
const pipeline = new rag_pipeline_1.RAGPipeline({
|
|
158
|
+
vectorStore: vs,
|
|
159
|
+
embeddingProvider: makeEmbeddingProvider(),
|
|
160
|
+
});
|
|
161
|
+
const selected = await pipeline.retrieve('query', {}, types_1.RetrievalStrategy.MMR);
|
|
162
|
+
expect(selected).toHaveLength(0);
|
|
163
|
+
});
|
|
164
|
+
it('MMR skips candidates without embeddings', async () => {
|
|
165
|
+
const results = [
|
|
166
|
+
{ id: 'd1', content: 'no embedding', score: 0.9, metadata: {} }, // no embedding
|
|
167
|
+
];
|
|
168
|
+
const vs = makeVectorStore(results);
|
|
169
|
+
const pipeline = new rag_pipeline_1.RAGPipeline({
|
|
170
|
+
vectorStore: vs,
|
|
171
|
+
embeddingProvider: makeEmbeddingProvider(),
|
|
172
|
+
});
|
|
173
|
+
const selected = await pipeline.retrieve('query', { topK: 1 }, types_1.RetrievalStrategy.MMR);
|
|
174
|
+
expect(Array.isArray(selected)).toBe(true);
|
|
175
|
+
});
|
|
176
|
+
});
|
|
177
|
+
describe('deleteDocuments & clear', () => {
|
|
178
|
+
it('deleteDocuments delegates to vectorStore', async () => {
|
|
179
|
+
const vs = makeVectorStore();
|
|
180
|
+
const pipeline = new rag_pipeline_1.RAGPipeline({
|
|
181
|
+
vectorStore: vs,
|
|
182
|
+
embeddingProvider: makeEmbeddingProvider(),
|
|
183
|
+
});
|
|
184
|
+
await pipeline.deleteDocuments(['d1', 'd2']);
|
|
185
|
+
expect(vs.deleteDocuments).toHaveBeenCalledWith(['d1', 'd2']);
|
|
186
|
+
});
|
|
187
|
+
it('clear delegates to vectorStore', async () => {
|
|
188
|
+
const vs = makeVectorStore();
|
|
189
|
+
const pipeline = new rag_pipeline_1.RAGPipeline({
|
|
190
|
+
vectorStore: vs,
|
|
191
|
+
embeddingProvider: makeEmbeddingProvider(),
|
|
192
|
+
});
|
|
193
|
+
await pipeline.clear();
|
|
194
|
+
expect(vs.clear).toHaveBeenCalled();
|
|
195
|
+
});
|
|
196
|
+
});
|
|
197
|
+
describe('buildContext', () => {
|
|
198
|
+
it('includes metadata in context string', async () => {
|
|
199
|
+
const vs = makeVectorStore([makeSearchResult('d1', 'Some content')]);
|
|
200
|
+
const pipeline = new rag_pipeline_1.RAGPipeline({
|
|
201
|
+
vectorStore: vs,
|
|
202
|
+
embeddingProvider: makeEmbeddingProvider(),
|
|
203
|
+
});
|
|
204
|
+
const response = await pipeline.query('query');
|
|
205
|
+
expect(response.context).toContain('[1]');
|
|
206
|
+
expect(response.context).toContain('Some content');
|
|
207
|
+
});
|
|
208
|
+
});
|
|
209
|
+
});
|
|
210
|
+
//# sourceMappingURL=rag-pipeline.test.js.map
|