ocr-provenance-mcp 1.0.0
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.
Potentially problematic release.
This version of ocr-provenance-mcp might be problematic. Click here for more details.
- package/.env.example +55 -0
- package/LICENSE +78 -0
- package/README.md +1154 -0
- package/dist/bin-http.d.ts +24 -0
- package/dist/bin-http.d.ts.map +1 -0
- package/dist/bin-http.js +275 -0
- package/dist/bin-http.js.map +1 -0
- package/dist/bin-setup.d.ts +11 -0
- package/dist/bin-setup.d.ts.map +1 -0
- package/dist/bin-setup.js +610 -0
- package/dist/bin-setup.js.map +1 -0
- package/dist/bin.d.ts +16 -0
- package/dist/bin.d.ts.map +1 -0
- package/dist/bin.js +16 -0
- package/dist/bin.js.map +1 -0
- package/dist/index.d.ts +13 -0
- package/dist/index.d.ts.map +1 -0
- package/dist/index.js +90 -0
- package/dist/index.js.map +1 -0
- package/dist/models/chunk.d.ts +136 -0
- package/dist/models/chunk.d.ts.map +1 -0
- package/dist/models/chunk.js +27 -0
- package/dist/models/chunk.js.map +1 -0
- package/dist/models/cluster.d.ts +79 -0
- package/dist/models/cluster.d.ts.map +1 -0
- package/dist/models/cluster.js +10 -0
- package/dist/models/cluster.js.map +1 -0
- package/dist/models/comparison.d.ts +62 -0
- package/dist/models/comparison.d.ts.map +1 -0
- package/dist/models/comparison.js +8 -0
- package/dist/models/comparison.js.map +1 -0
- package/dist/models/document.d.ts +104 -0
- package/dist/models/document.d.ts.map +1 -0
- package/dist/models/document.js +15 -0
- package/dist/models/document.js.map +1 -0
- package/dist/models/embedding.d.ts +87 -0
- package/dist/models/embedding.d.ts.map +1 -0
- package/dist/models/embedding.js +23 -0
- package/dist/models/embedding.js.map +1 -0
- package/dist/models/extraction.d.ts +15 -0
- package/dist/models/extraction.d.ts.map +1 -0
- package/dist/models/extraction.js +2 -0
- package/dist/models/extraction.js.map +1 -0
- package/dist/models/form-fill.d.ts +23 -0
- package/dist/models/form-fill.d.ts.map +1 -0
- package/dist/models/form-fill.js +2 -0
- package/dist/models/form-fill.js.map +1 -0
- package/dist/models/image.d.ts +177 -0
- package/dist/models/image.d.ts.map +1 -0
- package/dist/models/image.js +8 -0
- package/dist/models/image.js.map +1 -0
- package/dist/models/index.d.ts +14 -0
- package/dist/models/index.d.ts.map +1 -0
- package/dist/models/index.js +22 -0
- package/dist/models/index.js.map +1 -0
- package/dist/models/provenance.d.ts +174 -0
- package/dist/models/provenance.d.ts.map +1 -0
- package/dist/models/provenance.js +53 -0
- package/dist/models/provenance.js.map +1 -0
- package/dist/models/uploaded-file.d.ts +20 -0
- package/dist/models/uploaded-file.d.ts.map +1 -0
- package/dist/models/uploaded-file.js +2 -0
- package/dist/models/uploaded-file.js.map +1 -0
- package/dist/server/errors.d.ts +93 -0
- package/dist/server/errors.d.ts.map +1 -0
- package/dist/server/errors.js +256 -0
- package/dist/server/errors.js.map +1 -0
- package/dist/server/events.d.ts +36 -0
- package/dist/server/events.d.ts.map +1 -0
- package/dist/server/events.js +48 -0
- package/dist/server/events.js.map +1 -0
- package/dist/server/permissions.d.ts +26 -0
- package/dist/server/permissions.d.ts.map +1 -0
- package/dist/server/permissions.js +194 -0
- package/dist/server/permissions.js.map +1 -0
- package/dist/server/register-tools.d.ts +25 -0
- package/dist/server/register-tools.d.ts.map +1 -0
- package/dist/server/register-tools.js +102 -0
- package/dist/server/register-tools.js.map +1 -0
- package/dist/server/startup.d.ts +16 -0
- package/dist/server/startup.d.ts.map +1 -0
- package/dist/server/startup.js +37 -0
- package/dist/server/startup.js.map +1 -0
- package/dist/server/state.d.ts +166 -0
- package/dist/server/state.d.ts.map +1 -0
- package/dist/server/state.js +424 -0
- package/dist/server/state.js.map +1 -0
- package/dist/server/transports/http-transport.d.ts +37 -0
- package/dist/server/transports/http-transport.d.ts.map +1 -0
- package/dist/server/transports/http-transport.js +204 -0
- package/dist/server/transports/http-transport.js.map +1 -0
- package/dist/server/transports/index.d.ts +9 -0
- package/dist/server/transports/index.d.ts.map +1 -0
- package/dist/server/transports/index.js +9 -0
- package/dist/server/transports/index.js.map +1 -0
- package/dist/server/transports/session-manager.d.ts +40 -0
- package/dist/server/transports/session-manager.d.ts.map +1 -0
- package/dist/server/transports/session-manager.js +74 -0
- package/dist/server/transports/session-manager.js.map +1 -0
- package/dist/server/types.d.ts +82 -0
- package/dist/server/types.d.ts.map +1 -0
- package/dist/server/types.js +14 -0
- package/dist/server/types.js.map +1 -0
- package/dist/services/audit.d.ts +26 -0
- package/dist/services/audit.d.ts.map +1 -0
- package/dist/services/audit.js +43 -0
- package/dist/services/audit.js.map +1 -0
- package/dist/services/chunking/chunk-deduplicator.d.ts +33 -0
- package/dist/services/chunking/chunk-deduplicator.d.ts.map +1 -0
- package/dist/services/chunking/chunk-deduplicator.js +46 -0
- package/dist/services/chunking/chunk-deduplicator.js.map +1 -0
- package/dist/services/chunking/chunk-merger.d.ts +26 -0
- package/dist/services/chunking/chunk-merger.d.ts.map +1 -0
- package/dist/services/chunking/chunk-merger.js +94 -0
- package/dist/services/chunking/chunk-merger.js.map +1 -0
- package/dist/services/chunking/chunker.d.ts +62 -0
- package/dist/services/chunking/chunker.d.ts.map +1 -0
- package/dist/services/chunking/chunker.js +566 -0
- package/dist/services/chunking/chunker.js.map +1 -0
- package/dist/services/chunking/heading-normalizer.d.ts +33 -0
- package/dist/services/chunking/heading-normalizer.d.ts.map +1 -0
- package/dist/services/chunking/heading-normalizer.js +101 -0
- package/dist/services/chunking/heading-normalizer.js.map +1 -0
- package/dist/services/chunking/json-block-analyzer.d.ts +163 -0
- package/dist/services/chunking/json-block-analyzer.d.ts.map +1 -0
- package/dist/services/chunking/json-block-analyzer.js +1033 -0
- package/dist/services/chunking/json-block-analyzer.js.map +1 -0
- package/dist/services/chunking/markdown-parser.d.ts +75 -0
- package/dist/services/chunking/markdown-parser.d.ts.map +1 -0
- package/dist/services/chunking/markdown-parser.js +428 -0
- package/dist/services/chunking/markdown-parser.js.map +1 -0
- package/dist/services/chunking/text-normalizer.d.ts +20 -0
- package/dist/services/chunking/text-normalizer.d.ts.map +1 -0
- package/dist/services/chunking/text-normalizer.js +36 -0
- package/dist/services/chunking/text-normalizer.js.map +1 -0
- package/dist/services/clm/contract-schemas.d.ts +36 -0
- package/dist/services/clm/contract-schemas.d.ts.map +1 -0
- package/dist/services/clm/contract-schemas.js +92 -0
- package/dist/services/clm/contract-schemas.js.map +1 -0
- package/dist/services/clm/summarization.d.ts +46 -0
- package/dist/services/clm/summarization.d.ts.map +1 -0
- package/dist/services/clm/summarization.js +61 -0
- package/dist/services/clm/summarization.js.map +1 -0
- package/dist/services/clustering/clustering-service.d.ts +58 -0
- package/dist/services/clustering/clustering-service.d.ts.map +1 -0
- package/dist/services/clustering/clustering-service.js +467 -0
- package/dist/services/clustering/clustering-service.js.map +1 -0
- package/dist/services/comparison/diff-service.d.ts +41 -0
- package/dist/services/comparison/diff-service.d.ts.map +1 -0
- package/dist/services/comparison/diff-service.js +120 -0
- package/dist/services/comparison/diff-service.js.map +1 -0
- package/dist/services/embedding/embedder.d.ts +55 -0
- package/dist/services/embedding/embedder.d.ts.map +1 -0
- package/dist/services/embedding/embedder.js +202 -0
- package/dist/services/embedding/embedder.js.map +1 -0
- package/dist/services/embedding/nomic.d.ts +67 -0
- package/dist/services/embedding/nomic.d.ts.map +1 -0
- package/dist/services/embedding/nomic.js +280 -0
- package/dist/services/embedding/nomic.js.map +1 -0
- package/dist/services/gemini/circuit-breaker.d.ts +106 -0
- package/dist/services/gemini/circuit-breaker.d.ts.map +1 -0
- package/dist/services/gemini/circuit-breaker.js +237 -0
- package/dist/services/gemini/circuit-breaker.js.map +1 -0
- package/dist/services/gemini/client.d.ts +173 -0
- package/dist/services/gemini/client.d.ts.map +1 -0
- package/dist/services/gemini/client.js +483 -0
- package/dist/services/gemini/client.js.map +1 -0
- package/dist/services/gemini/config.d.ts +116 -0
- package/dist/services/gemini/config.d.ts.map +1 -0
- package/dist/services/gemini/config.js +118 -0
- package/dist/services/gemini/config.js.map +1 -0
- package/dist/services/gemini/index.d.ts +9 -0
- package/dist/services/gemini/index.d.ts.map +1 -0
- package/dist/services/gemini/index.js +13 -0
- package/dist/services/gemini/index.js.map +1 -0
- package/dist/services/gemini/rate-limiter.d.ts +62 -0
- package/dist/services/gemini/rate-limiter.d.ts.map +1 -0
- package/dist/services/gemini/rate-limiter.js +120 -0
- package/dist/services/gemini/rate-limiter.js.map +1 -0
- package/dist/services/images/extractor.d.ts +88 -0
- package/dist/services/images/extractor.d.ts.map +1 -0
- package/dist/services/images/extractor.js +340 -0
- package/dist/services/images/extractor.js.map +1 -0
- package/dist/services/images/optimizer.d.ts +130 -0
- package/dist/services/images/optimizer.d.ts.map +1 -0
- package/dist/services/images/optimizer.js +228 -0
- package/dist/services/images/optimizer.js.map +1 -0
- package/dist/services/ocr/datalab.d.ts +64 -0
- package/dist/services/ocr/datalab.d.ts.map +1 -0
- package/dist/services/ocr/datalab.js +425 -0
- package/dist/services/ocr/datalab.js.map +1 -0
- package/dist/services/ocr/errors.d.ts +38 -0
- package/dist/services/ocr/errors.d.ts.map +1 -0
- package/dist/services/ocr/errors.js +83 -0
- package/dist/services/ocr/errors.js.map +1 -0
- package/dist/services/ocr/file-manager.d.ts +76 -0
- package/dist/services/ocr/file-manager.d.ts.map +1 -0
- package/dist/services/ocr/file-manager.js +238 -0
- package/dist/services/ocr/file-manager.js.map +1 -0
- package/dist/services/ocr/form-fill.d.ts +48 -0
- package/dist/services/ocr/form-fill.d.ts.map +1 -0
- package/dist/services/ocr/form-fill.js +213 -0
- package/dist/services/ocr/form-fill.js.map +1 -0
- package/dist/services/ocr/processor.d.ts +95 -0
- package/dist/services/ocr/processor.d.ts.map +1 -0
- package/dist/services/ocr/processor.js +259 -0
- package/dist/services/ocr/processor.js.map +1 -0
- package/dist/services/provenance/agent-metadata.d.ts +82 -0
- package/dist/services/provenance/agent-metadata.d.ts.map +1 -0
- package/dist/services/provenance/agent-metadata.js +106 -0
- package/dist/services/provenance/agent-metadata.js.map +1 -0
- package/dist/services/provenance/chain-hash.d.ts +57 -0
- package/dist/services/provenance/chain-hash.d.ts.map +1 -0
- package/dist/services/provenance/chain-hash.js +131 -0
- package/dist/services/provenance/chain-hash.js.map +1 -0
- package/dist/services/provenance/exporter.d.ts +202 -0
- package/dist/services/provenance/exporter.d.ts.map +1 -0
- package/dist/services/provenance/exporter.js +457 -0
- package/dist/services/provenance/exporter.js.map +1 -0
- package/dist/services/provenance/index.d.ts +15 -0
- package/dist/services/provenance/index.d.ts.map +1 -0
- package/dist/services/provenance/index.js +17 -0
- package/dist/services/provenance/index.js.map +1 -0
- package/dist/services/provenance/tracker.d.ts +138 -0
- package/dist/services/provenance/tracker.d.ts.map +1 -0
- package/dist/services/provenance/tracker.js +293 -0
- package/dist/services/provenance/tracker.js.map +1 -0
- package/dist/services/provenance/verifier.d.ts +153 -0
- package/dist/services/provenance/verifier.d.ts.map +1 -0
- package/dist/services/provenance/verifier.js +536 -0
- package/dist/services/provenance/verifier.js.map +1 -0
- package/dist/services/python-pool.d.ts +70 -0
- package/dist/services/python-pool.d.ts.map +1 -0
- package/dist/services/python-pool.js +265 -0
- package/dist/services/python-pool.js.map +1 -0
- package/dist/services/search/bm25.d.ts +180 -0
- package/dist/services/search/bm25.d.ts.map +1 -0
- package/dist/services/search/bm25.js +656 -0
- package/dist/services/search/bm25.js.map +1 -0
- package/dist/services/search/fusion.d.ts +103 -0
- package/dist/services/search/fusion.d.ts.map +1 -0
- package/dist/services/search/fusion.js +122 -0
- package/dist/services/search/fusion.js.map +1 -0
- package/dist/services/search/local-reranker.d.ts +30 -0
- package/dist/services/search/local-reranker.d.ts.map +1 -0
- package/dist/services/search/local-reranker.js +123 -0
- package/dist/services/search/local-reranker.js.map +1 -0
- package/dist/services/search/quality.d.ts +11 -0
- package/dist/services/search/quality.d.ts.map +1 -0
- package/dist/services/search/quality.js +17 -0
- package/dist/services/search/quality.js.map +1 -0
- package/dist/services/search/query-classifier.d.ts +34 -0
- package/dist/services/search/query-classifier.d.ts.map +1 -0
- package/dist/services/search/query-classifier.js +114 -0
- package/dist/services/search/query-classifier.js.map +1 -0
- package/dist/services/search/query-expander.d.ts +73 -0
- package/dist/services/search/query-expander.d.ts.map +1 -0
- package/dist/services/search/query-expander.js +281 -0
- package/dist/services/search/query-expander.js.map +1 -0
- package/dist/services/search/reranker.d.ts +44 -0
- package/dist/services/search/reranker.d.ts.map +1 -0
- package/dist/services/search/reranker.js +101 -0
- package/dist/services/search/reranker.js.map +1 -0
- package/dist/services/storage/database/annotation-operations.d.ts +113 -0
- package/dist/services/storage/database/annotation-operations.d.ts.map +1 -0
- package/dist/services/storage/database/annotation-operations.js +177 -0
- package/dist/services/storage/database/annotation-operations.js.map +1 -0
- package/dist/services/storage/database/approval-operations.d.ts +132 -0
- package/dist/services/storage/database/approval-operations.d.ts.map +1 -0
- package/dist/services/storage/database/approval-operations.js +206 -0
- package/dist/services/storage/database/approval-operations.js.map +1 -0
- package/dist/services/storage/database/chunk-operations.d.ts +132 -0
- package/dist/services/storage/database/chunk-operations.d.ts.map +1 -0
- package/dist/services/storage/database/chunk-operations.js +306 -0
- package/dist/services/storage/database/chunk-operations.js.map +1 -0
- package/dist/services/storage/database/cluster-operations.d.ts +97 -0
- package/dist/services/storage/database/cluster-operations.d.ts.map +1 -0
- package/dist/services/storage/database/cluster-operations.js +258 -0
- package/dist/services/storage/database/cluster-operations.js.map +1 -0
- package/dist/services/storage/database/comparison-operations.d.ts +41 -0
- package/dist/services/storage/database/comparison-operations.d.ts.map +1 -0
- package/dist/services/storage/database/comparison-operations.js +65 -0
- package/dist/services/storage/database/comparison-operations.js.map +1 -0
- package/dist/services/storage/database/converters.d.ts +36 -0
- package/dist/services/storage/database/converters.d.ts.map +1 -0
- package/dist/services/storage/database/converters.js +244 -0
- package/dist/services/storage/database/converters.js.map +1 -0
- package/dist/services/storage/database/document-operations.d.ts +145 -0
- package/dist/services/storage/database/document-operations.d.ts.map +1 -0
- package/dist/services/storage/database/document-operations.js +498 -0
- package/dist/services/storage/database/document-operations.js.map +1 -0
- package/dist/services/storage/database/embedding-operations.d.ts +130 -0
- package/dist/services/storage/database/embedding-operations.d.ts.map +1 -0
- package/dist/services/storage/database/embedding-operations.js +315 -0
- package/dist/services/storage/database/embedding-operations.js.map +1 -0
- package/dist/services/storage/database/extraction-operations.d.ts +47 -0
- package/dist/services/storage/database/extraction-operations.d.ts.map +1 -0
- package/dist/services/storage/database/extraction-operations.js +85 -0
- package/dist/services/storage/database/extraction-operations.js.map +1 -0
- package/dist/services/storage/database/form-fill-operations.d.ts +58 -0
- package/dist/services/storage/database/form-fill-operations.d.ts.map +1 -0
- package/dist/services/storage/database/form-fill-operations.js +116 -0
- package/dist/services/storage/database/form-fill-operations.js.map +1 -0
- package/dist/services/storage/database/helpers.d.ts +29 -0
- package/dist/services/storage/database/helpers.d.ts.map +1 -0
- package/dist/services/storage/database/helpers.js +55 -0
- package/dist/services/storage/database/helpers.js.map +1 -0
- package/dist/services/storage/database/image-operations.d.ts +202 -0
- package/dist/services/storage/database/image-operations.d.ts.map +1 -0
- package/dist/services/storage/database/image-operations.js +484 -0
- package/dist/services/storage/database/image-operations.js.map +1 -0
- package/dist/services/storage/database/index.d.ts +13 -0
- package/dist/services/storage/database/index.d.ts.map +1 -0
- package/dist/services/storage/database/index.js +16 -0
- package/dist/services/storage/database/index.js.map +1 -0
- package/dist/services/storage/database/lock-operations.d.ts +59 -0
- package/dist/services/storage/database/lock-operations.d.ts.map +1 -0
- package/dist/services/storage/database/lock-operations.js +89 -0
- package/dist/services/storage/database/lock-operations.js.map +1 -0
- package/dist/services/storage/database/obligation-operations.d.ts +88 -0
- package/dist/services/storage/database/obligation-operations.d.ts.map +1 -0
- package/dist/services/storage/database/obligation-operations.js +206 -0
- package/dist/services/storage/database/obligation-operations.js.map +1 -0
- package/dist/services/storage/database/ocr-operations.d.ts +33 -0
- package/dist/services/storage/database/ocr-operations.d.ts.map +1 -0
- package/dist/services/storage/database/ocr-operations.js +70 -0
- package/dist/services/storage/database/ocr-operations.js.map +1 -0
- package/dist/services/storage/database/playbook-operations.d.ts +72 -0
- package/dist/services/storage/database/playbook-operations.d.ts.map +1 -0
- package/dist/services/storage/database/playbook-operations.js +247 -0
- package/dist/services/storage/database/playbook-operations.js.map +1 -0
- package/dist/services/storage/database/provenance-operations.d.ts +112 -0
- package/dist/services/storage/database/provenance-operations.d.ts.map +1 -0
- package/dist/services/storage/database/provenance-operations.js +251 -0
- package/dist/services/storage/database/provenance-operations.js.map +1 -0
- package/dist/services/storage/database/service.d.ts +142 -0
- package/dist/services/storage/database/service.d.ts.map +1 -0
- package/dist/services/storage/database/service.js +310 -0
- package/dist/services/storage/database/service.js.map +1 -0
- package/dist/services/storage/database/static-operations.d.ts +30 -0
- package/dist/services/storage/database/static-operations.d.ts.map +1 -0
- package/dist/services/storage/database/static-operations.js +218 -0
- package/dist/services/storage/database/static-operations.js.map +1 -0
- package/dist/services/storage/database/stats-operations.d.ts +101 -0
- package/dist/services/storage/database/stats-operations.d.ts.map +1 -0
- package/dist/services/storage/database/stats-operations.js +394 -0
- package/dist/services/storage/database/stats-operations.js.map +1 -0
- package/dist/services/storage/database/tag-operations.d.ts +76 -0
- package/dist/services/storage/database/tag-operations.d.ts.map +1 -0
- package/dist/services/storage/database/tag-operations.js +178 -0
- package/dist/services/storage/database/tag-operations.js.map +1 -0
- package/dist/services/storage/database/types.d.ts +286 -0
- package/dist/services/storage/database/types.d.ts.map +1 -0
- package/dist/services/storage/database/types.js +39 -0
- package/dist/services/storage/database/types.js.map +1 -0
- package/dist/services/storage/database/upload-operations.d.ts +71 -0
- package/dist/services/storage/database/upload-operations.d.ts.map +1 -0
- package/dist/services/storage/database/upload-operations.js +124 -0
- package/dist/services/storage/database/upload-operations.js.map +1 -0
- package/dist/services/storage/database/user-operations.d.ts +102 -0
- package/dist/services/storage/database/user-operations.d.ts.map +1 -0
- package/dist/services/storage/database/user-operations.js +151 -0
- package/dist/services/storage/database/user-operations.js.map +1 -0
- package/dist/services/storage/database/workflow-operations.d.ts +98 -0
- package/dist/services/storage/database/workflow-operations.d.ts.map +1 -0
- package/dist/services/storage/database/workflow-operations.js +157 -0
- package/dist/services/storage/database/workflow-operations.js.map +1 -0
- package/dist/services/storage/database.d.ts +16 -0
- package/dist/services/storage/database.d.ts.map +1 -0
- package/dist/services/storage/database.js +15 -0
- package/dist/services/storage/database.js.map +1 -0
- package/dist/services/storage/index.d.ts +10 -0
- package/dist/services/storage/index.d.ts.map +1 -0
- package/dist/services/storage/index.js +10 -0
- package/dist/services/storage/index.js.map +1 -0
- package/dist/services/storage/migrations/index.d.ts +16 -0
- package/dist/services/storage/migrations/index.d.ts.map +1 -0
- package/dist/services/storage/migrations/index.js +20 -0
- package/dist/services/storage/migrations/index.js.map +1 -0
- package/dist/services/storage/migrations/operations.d.ts +40 -0
- package/dist/services/storage/migrations/operations.d.ts.map +1 -0
- package/dist/services/storage/migrations/operations.js +2910 -0
- package/dist/services/storage/migrations/operations.js.map +1 -0
- package/dist/services/storage/migrations/schema-definitions.d.ts +306 -0
- package/dist/services/storage/migrations/schema-definitions.d.ts.map +1 -0
- package/dist/services/storage/migrations/schema-definitions.js +1006 -0
- package/dist/services/storage/migrations/schema-definitions.js.map +1 -0
- package/dist/services/storage/migrations/schema-helpers.d.ts +50 -0
- package/dist/services/storage/migrations/schema-helpers.d.ts.map +1 -0
- package/dist/services/storage/migrations/schema-helpers.js +176 -0
- package/dist/services/storage/migrations/schema-helpers.js.map +1 -0
- package/dist/services/storage/migrations/types.d.ts +15 -0
- package/dist/services/storage/migrations/types.d.ts.map +1 -0
- package/dist/services/storage/migrations/types.js +21 -0
- package/dist/services/storage/migrations/types.js.map +1 -0
- package/dist/services/storage/migrations/verification.d.ts +20 -0
- package/dist/services/storage/migrations/verification.d.ts.map +1 -0
- package/dist/services/storage/migrations/verification.js +78 -0
- package/dist/services/storage/migrations/verification.js.map +1 -0
- package/dist/services/storage/migrations.d.ts +16 -0
- package/dist/services/storage/migrations.d.ts.map +1 -0
- package/dist/services/storage/migrations.js +17 -0
- package/dist/services/storage/migrations.js.map +1 -0
- package/dist/services/storage/types.d.ts +12 -0
- package/dist/services/storage/types.d.ts.map +1 -0
- package/dist/services/storage/types.js +5 -0
- package/dist/services/storage/types.js.map +1 -0
- package/dist/services/storage/vector.d.ts +208 -0
- package/dist/services/storage/vector.d.ts.map +1 -0
- package/dist/services/storage/vector.js +526 -0
- package/dist/services/storage/vector.js.map +1 -0
- package/dist/services/vlm/pipeline.d.ts +194 -0
- package/dist/services/vlm/pipeline.d.ts.map +1 -0
- package/dist/services/vlm/pipeline.js +800 -0
- package/dist/services/vlm/pipeline.js.map +1 -0
- package/dist/services/vlm/prompts.d.ts +171 -0
- package/dist/services/vlm/prompts.d.ts.map +1 -0
- package/dist/services/vlm/prompts.js +229 -0
- package/dist/services/vlm/prompts.js.map +1 -0
- package/dist/services/vlm/service.d.ts +174 -0
- package/dist/services/vlm/service.d.ts.map +1 -0
- package/dist/services/vlm/service.js +256 -0
- package/dist/services/vlm/service.js.map +1 -0
- package/dist/services/webhook-delivery.d.ts +4 -0
- package/dist/services/webhook-delivery.d.ts.map +1 -0
- package/dist/services/webhook-delivery.js +140 -0
- package/dist/services/webhook-delivery.js.map +1 -0
- package/dist/tools/chunks.d.ts +19 -0
- package/dist/tools/chunks.d.ts.map +1 -0
- package/dist/tools/chunks.js +392 -0
- package/dist/tools/chunks.js.map +1 -0
- package/dist/tools/clm.d.ts +16 -0
- package/dist/tools/clm.d.ts.map +1 -0
- package/dist/tools/clm.js +668 -0
- package/dist/tools/clm.js.map +1 -0
- package/dist/tools/clustering.d.ts +13 -0
- package/dist/tools/clustering.d.ts.map +1 -0
- package/dist/tools/clustering.js +498 -0
- package/dist/tools/clustering.js.map +1 -0
- package/dist/tools/collaboration.d.ts +15 -0
- package/dist/tools/collaboration.d.ts.map +1 -0
- package/dist/tools/collaboration.js +516 -0
- package/dist/tools/collaboration.js.map +1 -0
- package/dist/tools/comparison.d.ts +13 -0
- package/dist/tools/comparison.d.ts.map +1 -0
- package/dist/tools/comparison.js +735 -0
- package/dist/tools/comparison.js.map +1 -0
- package/dist/tools/compliance.d.ts +15 -0
- package/dist/tools/compliance.d.ts.map +1 -0
- package/dist/tools/compliance.js +640 -0
- package/dist/tools/compliance.js.map +1 -0
- package/dist/tools/config.d.ts +19 -0
- package/dist/tools/config.d.ts.map +1 -0
- package/dist/tools/config.js +213 -0
- package/dist/tools/config.js.map +1 -0
- package/dist/tools/database.d.ts +62 -0
- package/dist/tools/database.d.ts.map +1 -0
- package/dist/tools/database.js +288 -0
- package/dist/tools/database.js.map +1 -0
- package/dist/tools/documents.d.ts +61 -0
- package/dist/tools/documents.d.ts.map +1 -0
- package/dist/tools/documents.js +1624 -0
- package/dist/tools/documents.js.map +1 -0
- package/dist/tools/embeddings.d.ts +14 -0
- package/dist/tools/embeddings.d.ts.map +1 -0
- package/dist/tools/embeddings.js +626 -0
- package/dist/tools/embeddings.js.map +1 -0
- package/dist/tools/evaluation.d.ts +25 -0
- package/dist/tools/evaluation.d.ts.map +1 -0
- package/dist/tools/evaluation.js +523 -0
- package/dist/tools/evaluation.js.map +1 -0
- package/dist/tools/events.d.ts +16 -0
- package/dist/tools/events.d.ts.map +1 -0
- package/dist/tools/events.js +493 -0
- package/dist/tools/events.js.map +1 -0
- package/dist/tools/extraction-structured.d.ts +13 -0
- package/dist/tools/extraction-structured.d.ts.map +1 -0
- package/dist/tools/extraction-structured.js +390 -0
- package/dist/tools/extraction-structured.js.map +1 -0
- package/dist/tools/extraction.d.ts +24 -0
- package/dist/tools/extraction.d.ts.map +1 -0
- package/dist/tools/extraction.js +424 -0
- package/dist/tools/extraction.js.map +1 -0
- package/dist/tools/file-management.d.ts +14 -0
- package/dist/tools/file-management.d.ts.map +1 -0
- package/dist/tools/file-management.js +523 -0
- package/dist/tools/file-management.js.map +1 -0
- package/dist/tools/form-fill.d.ts +13 -0
- package/dist/tools/form-fill.d.ts.map +1 -0
- package/dist/tools/form-fill.js +250 -0
- package/dist/tools/form-fill.js.map +1 -0
- package/dist/tools/health.d.ts +19 -0
- package/dist/tools/health.d.ts.map +1 -0
- package/dist/tools/health.js +229 -0
- package/dist/tools/health.js.map +1 -0
- package/dist/tools/images.d.ts +54 -0
- package/dist/tools/images.d.ts.map +1 -0
- package/dist/tools/images.js +787 -0
- package/dist/tools/images.js.map +1 -0
- package/dist/tools/ingestion.d.ts +94 -0
- package/dist/tools/ingestion.d.ts.map +1 -0
- package/dist/tools/ingestion.js +1659 -0
- package/dist/tools/ingestion.js.map +1 -0
- package/dist/tools/intelligence.d.ts +18 -0
- package/dist/tools/intelligence.d.ts.map +1 -0
- package/dist/tools/intelligence.js +1039 -0
- package/dist/tools/intelligence.js.map +1 -0
- package/dist/tools/provenance.d.ts +51 -0
- package/dist/tools/provenance.d.ts.map +1 -0
- package/dist/tools/provenance.js +691 -0
- package/dist/tools/provenance.js.map +1 -0
- package/dist/tools/reports.d.ts +41 -0
- package/dist/tools/reports.d.ts.map +1 -0
- package/dist/tools/reports.js +1394 -0
- package/dist/tools/reports.js.map +1 -0
- package/dist/tools/search.d.ts +35 -0
- package/dist/tools/search.d.ts.map +1 -0
- package/dist/tools/search.js +2528 -0
- package/dist/tools/search.js.map +1 -0
- package/dist/tools/shared.d.ts +52 -0
- package/dist/tools/shared.d.ts.map +1 -0
- package/dist/tools/shared.js +54 -0
- package/dist/tools/shared.js.map +1 -0
- package/dist/tools/tags.d.ts +15 -0
- package/dist/tools/tags.d.ts.map +1 -0
- package/dist/tools/tags.js +287 -0
- package/dist/tools/tags.js.map +1 -0
- package/dist/tools/timeline.d.ts +15 -0
- package/dist/tools/timeline.d.ts.map +1 -0
- package/dist/tools/timeline.js +14 -0
- package/dist/tools/timeline.js.map +1 -0
- package/dist/tools/users.d.ts +14 -0
- package/dist/tools/users.d.ts.map +1 -0
- package/dist/tools/users.js +257 -0
- package/dist/tools/users.js.map +1 -0
- package/dist/tools/vlm.d.ts +40 -0
- package/dist/tools/vlm.d.ts.map +1 -0
- package/dist/tools/vlm.js +475 -0
- package/dist/tools/vlm.js.map +1 -0
- package/dist/tools/workflow.d.ts +16 -0
- package/dist/tools/workflow.d.ts.map +1 -0
- package/dist/tools/workflow.js +495 -0
- package/dist/tools/workflow.js.map +1 -0
- package/dist/utils/backoff.d.ts +53 -0
- package/dist/utils/backoff.d.ts.map +1 -0
- package/dist/utils/backoff.js +78 -0
- package/dist/utils/backoff.js.map +1 -0
- package/dist/utils/config-persistence.d.ts +33 -0
- package/dist/utils/config-persistence.d.ts.map +1 -0
- package/dist/utils/config-persistence.js +61 -0
- package/dist/utils/config-persistence.js.map +1 -0
- package/dist/utils/hash.d.ts +65 -0
- package/dist/utils/hash.d.ts.map +1 -0
- package/dist/utils/hash.js +146 -0
- package/dist/utils/hash.js.map +1 -0
- package/dist/utils/math.d.ts +21 -0
- package/dist/utils/math.d.ts.map +1 -0
- package/dist/utils/math.js +39 -0
- package/dist/utils/math.js.map +1 -0
- package/dist/utils/validation.d.ts +697 -0
- package/dist/utils/validation.d.ts.map +1 -0
- package/dist/utils/validation.js +529 -0
- package/dist/utils/validation.js.map +1 -0
- package/package.json +96 -0
- package/python/.gitkeep +0 -0
- package/python/__init__.py +104 -0
- package/python/clustering_worker.py +440 -0
- package/python/docx_image_extractor.py +524 -0
- package/python/embedding_worker.py +552 -0
- package/python/file_manager_worker.py +564 -0
- package/python/form_fill_worker.py +399 -0
- package/python/gpu_utils.py +582 -0
- package/python/image_extractor.py +317 -0
- package/python/image_optimizer.py +444 -0
- package/python/ocr_worker.py +712 -0
- package/python/pyproject.toml +76 -0
- package/python/requirements.txt +51 -0
- package/python/reranker_worker.py +87 -0
|
@@ -0,0 +1,228 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* Image Optimizer Service
|
|
3
|
+
*
|
|
4
|
+
* TypeScript wrapper for Python image optimizer providing:
|
|
5
|
+
* 1. Resize for OCR (max 4800px width for Datalab API)
|
|
6
|
+
* 2. Resize for VLM (optimize token usage, max 2048px)
|
|
7
|
+
* 3. Relevance analysis to filter logos, icons, and decorative elements
|
|
8
|
+
*
|
|
9
|
+
* The relevance analysis uses multi-layer heuristics:
|
|
10
|
+
* - Size filtering (tiny images are likely icons)
|
|
11
|
+
* - Aspect ratio analysis (extreme ratios = banners/logos)
|
|
12
|
+
* - Color diversity (low color count = likely logo/icon)
|
|
13
|
+
*/
|
|
14
|
+
import { spawn } from 'child_process';
|
|
15
|
+
import * as path from 'path';
|
|
16
|
+
import * as fs from 'fs';
|
|
17
|
+
import * as os from 'os';
|
|
18
|
+
import { fileURLToPath } from 'url';
|
|
19
|
+
const __dirname = path.dirname(fileURLToPath(import.meta.url));
|
|
20
|
+
/** Max stderr accumulation: 10KB (matches nomic.ts pattern) */
|
|
21
|
+
const MAX_STDERR_LENGTH = 10_240;
|
|
22
|
+
const DEFAULT_CONFIG = {
|
|
23
|
+
pythonPath: process.platform === 'win32' ? 'python' : 'python3',
|
|
24
|
+
timeout: 60000, // 1 minute
|
|
25
|
+
vlmMaxDimension: 2048,
|
|
26
|
+
vlmSkipBelowSize: 50,
|
|
27
|
+
minRelevanceScore: 0.3,
|
|
28
|
+
};
|
|
29
|
+
/**
|
|
30
|
+
* Service for optimizing images for OCR and VLM processing
|
|
31
|
+
*/
|
|
32
|
+
export class ImageOptimizer {
|
|
33
|
+
config;
|
|
34
|
+
scriptPath;
|
|
35
|
+
constructor(config = {}) {
|
|
36
|
+
this.config = { ...DEFAULT_CONFIG, ...config };
|
|
37
|
+
this.scriptPath = path.resolve(__dirname, '../../../python/image_optimizer.py');
|
|
38
|
+
}
|
|
39
|
+
/**
|
|
40
|
+
* Analyze an image to determine if it should be processed by VLM.
|
|
41
|
+
*
|
|
42
|
+
* Uses multi-layer heuristics:
|
|
43
|
+
* 1. Size (tiny images = skip)
|
|
44
|
+
* 2. Aspect ratio (extreme ratios = skip)
|
|
45
|
+
* 3. Color diversity (low = likely logo/icon)
|
|
46
|
+
* 4. Category prediction
|
|
47
|
+
*
|
|
48
|
+
* @param imagePath - Path to the image file
|
|
49
|
+
* @returns Analysis result with should_vlm recommendation
|
|
50
|
+
*/
|
|
51
|
+
async analyzeImage(imagePath) {
|
|
52
|
+
return this.runPython(['--analyze', imagePath]);
|
|
53
|
+
}
|
|
54
|
+
/**
|
|
55
|
+
* Resize an image for VLM processing (Gemini).
|
|
56
|
+
*
|
|
57
|
+
* @param inputPath - Path to input image
|
|
58
|
+
* @param outputPath - Path for output (optional, creates temp file if not provided)
|
|
59
|
+
* @returns Resize result or skip result if too small
|
|
60
|
+
*/
|
|
61
|
+
async resizeForVLM(inputPath, outputPath) {
|
|
62
|
+
const output = outputPath ?? this.createTempPath(inputPath, 'vlm');
|
|
63
|
+
return this.runPython([
|
|
64
|
+
'--resize-for-vlm',
|
|
65
|
+
inputPath,
|
|
66
|
+
'--output',
|
|
67
|
+
output,
|
|
68
|
+
'--max-dimension',
|
|
69
|
+
String(this.config.vlmMaxDimension),
|
|
70
|
+
]);
|
|
71
|
+
}
|
|
72
|
+
/**
|
|
73
|
+
* Check if the Python optimizer script exists.
|
|
74
|
+
*/
|
|
75
|
+
isAvailable() {
|
|
76
|
+
return fs.existsSync(this.scriptPath);
|
|
77
|
+
}
|
|
78
|
+
/**
|
|
79
|
+
* Create a temporary file path for resized output.
|
|
80
|
+
*/
|
|
81
|
+
createTempPath(inputPath, suffix) {
|
|
82
|
+
const ext = path.extname(inputPath);
|
|
83
|
+
const base = path.basename(inputPath, ext);
|
|
84
|
+
const tmpDir = os.tmpdir();
|
|
85
|
+
return path.join(tmpDir, `${base}_${suffix}_${Date.now()}${ext}`);
|
|
86
|
+
}
|
|
87
|
+
/**
|
|
88
|
+
* Run the Python optimizer script.
|
|
89
|
+
* M-14: Sends SIGKILL after 5s if SIGTERM from Node.js timeout didn't terminate the process.
|
|
90
|
+
*/
|
|
91
|
+
runPython(args) {
|
|
92
|
+
return new Promise((resolve) => {
|
|
93
|
+
// Validate script exists
|
|
94
|
+
if (!fs.existsSync(this.scriptPath)) {
|
|
95
|
+
resolve({
|
|
96
|
+
success: false,
|
|
97
|
+
error: `Image optimizer script not found: ${this.scriptPath}`,
|
|
98
|
+
});
|
|
99
|
+
return;
|
|
100
|
+
}
|
|
101
|
+
const proc = spawn(this.config.pythonPath, [this.scriptPath, ...args], {
|
|
102
|
+
timeout: this.config.timeout,
|
|
103
|
+
});
|
|
104
|
+
let stdout = '';
|
|
105
|
+
let stderr = '';
|
|
106
|
+
let settled = false;
|
|
107
|
+
let sigkillTimer = null;
|
|
108
|
+
const cleanup = () => {
|
|
109
|
+
if (sigkillTimer) {
|
|
110
|
+
clearTimeout(sigkillTimer);
|
|
111
|
+
sigkillTimer = null;
|
|
112
|
+
}
|
|
113
|
+
};
|
|
114
|
+
proc.stdout.on('data', (data) => {
|
|
115
|
+
stdout += data.toString();
|
|
116
|
+
});
|
|
117
|
+
// H-9: Cap stderr accumulation to prevent unbounded memory growth
|
|
118
|
+
proc.stderr.on('data', (data) => {
|
|
119
|
+
if (stderr.length < MAX_STDERR_LENGTH) {
|
|
120
|
+
stderr += data.toString();
|
|
121
|
+
}
|
|
122
|
+
});
|
|
123
|
+
proc.on('error', (err) => {
|
|
124
|
+
cleanup();
|
|
125
|
+
if (settled)
|
|
126
|
+
return;
|
|
127
|
+
settled = true;
|
|
128
|
+
// The bottom-of-function SIGKILL timer at timeout+5000ms already covers
|
|
129
|
+
// the case where SIGTERM fails to terminate the process. No need for a
|
|
130
|
+
// second timer here (which would leak since cleanup() already ran).
|
|
131
|
+
resolve({
|
|
132
|
+
success: false,
|
|
133
|
+
error: `Failed to start Python process: ${err.message}`,
|
|
134
|
+
});
|
|
135
|
+
});
|
|
136
|
+
proc.on('close', (code, signal) => {
|
|
137
|
+
cleanup();
|
|
138
|
+
if (settled)
|
|
139
|
+
return;
|
|
140
|
+
settled = true;
|
|
141
|
+
if (stderr) {
|
|
142
|
+
console.error(`[ImageOptimizer] stderr: ${stderr}`);
|
|
143
|
+
}
|
|
144
|
+
if (signal === 'SIGTERM' || signal === 'SIGKILL') {
|
|
145
|
+
resolve({
|
|
146
|
+
success: false,
|
|
147
|
+
error: `Python process killed by ${signal} (timeout: ${this.config.timeout}ms)`,
|
|
148
|
+
});
|
|
149
|
+
return;
|
|
150
|
+
}
|
|
151
|
+
try {
|
|
152
|
+
// Python may output debug/warning lines before the JSON. Find the last
|
|
153
|
+
// valid JSON line (starts with '{' or '[') to handle multi-line output.
|
|
154
|
+
const lines = stdout.trim().split('\n');
|
|
155
|
+
let parsed;
|
|
156
|
+
for (let i = lines.length - 1; i >= 0; i--) {
|
|
157
|
+
const line = lines[i].trim();
|
|
158
|
+
if (line.startsWith('{') || line.startsWith('[')) {
|
|
159
|
+
try {
|
|
160
|
+
parsed = JSON.parse(line);
|
|
161
|
+
break;
|
|
162
|
+
}
|
|
163
|
+
catch (error) {
|
|
164
|
+
console.error('[ImageOptimizer] JSON parse failed for output line, trying previous:', error instanceof Error ? error.message : String(error));
|
|
165
|
+
/* not valid JSON, try previous line */
|
|
166
|
+
}
|
|
167
|
+
}
|
|
168
|
+
}
|
|
169
|
+
if (parsed === undefined) {
|
|
170
|
+
// Fallback: try parsing the entire stdout as one JSON blob
|
|
171
|
+
parsed = JSON.parse(stdout);
|
|
172
|
+
}
|
|
173
|
+
resolve(parsed);
|
|
174
|
+
}
|
|
175
|
+
catch (parseError) {
|
|
176
|
+
if (code !== 0) {
|
|
177
|
+
resolve({
|
|
178
|
+
success: false,
|
|
179
|
+
error: `Python script exited with code ${code}: ${(stderr || stdout).substring(0, 2000)}`,
|
|
180
|
+
});
|
|
181
|
+
}
|
|
182
|
+
else {
|
|
183
|
+
resolve({
|
|
184
|
+
success: false,
|
|
185
|
+
error: `Failed to parse result: ${parseError}`,
|
|
186
|
+
});
|
|
187
|
+
}
|
|
188
|
+
}
|
|
189
|
+
});
|
|
190
|
+
// SIGKILL fallback: if Node.js timeout sends SIGTERM and process doesn't exit within 5s, SIGKILL it
|
|
191
|
+
// The spawn timeout option sends SIGTERM. We set a timer to escalate to SIGKILL.
|
|
192
|
+
if (this.config.timeout > 0) {
|
|
193
|
+
sigkillTimer = setTimeout(() => {
|
|
194
|
+
if (!settled && !proc.killed) {
|
|
195
|
+
console.error(`[ImageOptimizer] Process did not exit after SIGTERM, sending SIGKILL (pid: ${proc.pid})`);
|
|
196
|
+
try {
|
|
197
|
+
proc.kill('SIGKILL');
|
|
198
|
+
}
|
|
199
|
+
catch (error) {
|
|
200
|
+
console.error('[ImageOptimizer] Failed to SIGKILL process (may already be gone):', error instanceof Error ? error.message : String(error));
|
|
201
|
+
// Process may already be gone
|
|
202
|
+
}
|
|
203
|
+
}
|
|
204
|
+
}, this.config.timeout + 5000);
|
|
205
|
+
}
|
|
206
|
+
});
|
|
207
|
+
}
|
|
208
|
+
}
|
|
209
|
+
/**
|
|
210
|
+
* Cached default optimizer instance (used when no config is provided)
|
|
211
|
+
*/
|
|
212
|
+
let cachedOptimizer = null;
|
|
213
|
+
/**
|
|
214
|
+
* Get an optimizer instance.
|
|
215
|
+
* When config is provided, always creates a new instance and updates the cache.
|
|
216
|
+
* When no config is provided, returns the cached instance (creating one if needed).
|
|
217
|
+
*/
|
|
218
|
+
export function getImageOptimizer(config) {
|
|
219
|
+
if (config) {
|
|
220
|
+
cachedOptimizer = new ImageOptimizer(config);
|
|
221
|
+
return cachedOptimizer;
|
|
222
|
+
}
|
|
223
|
+
if (!cachedOptimizer) {
|
|
224
|
+
cachedOptimizer = new ImageOptimizer();
|
|
225
|
+
}
|
|
226
|
+
return cachedOptimizer;
|
|
227
|
+
}
|
|
228
|
+
//# sourceMappingURL=optimizer.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"optimizer.js","sourceRoot":"","sources":["../../../src/services/images/optimizer.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;GAYG;AAEH,OAAO,EAAE,KAAK,EAAE,MAAM,eAAe,CAAC;AACtC,OAAO,KAAK,IAAI,MAAM,MAAM,CAAC;AAC7B,OAAO,KAAK,EAAE,MAAM,IAAI,CAAC;AACzB,OAAO,KAAK,EAAE,MAAM,IAAI,CAAC;AACzB,OAAO,EAAE,aAAa,EAAE,MAAM,KAAK,CAAC;AAEpC,MAAM,SAAS,GAAG,IAAI,CAAC,OAAO,CAAC,aAAa,CAAC,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;AA2E/D,+DAA+D;AAC/D,MAAM,iBAAiB,GAAG,MAAM,CAAC;AAEjC,MAAM,cAAc,GAAyB;IAC3C,UAAU,EAAE,OAAO,CAAC,QAAQ,KAAK,OAAO,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,SAAS;IAC/D,OAAO,EAAE,KAAK,EAAE,WAAW;IAC3B,eAAe,EAAE,IAAI;IACrB,gBAAgB,EAAE,EAAE;IACpB,iBAAiB,EAAE,GAAG;CACvB,CAAC;AAEF;;GAEG;AACH,MAAM,OAAO,cAAc;IACR,MAAM,CAAuB;IAC7B,UAAU,CAAS;IAEpC,YAAY,SAAwC,EAAE;QACpD,IAAI,CAAC,MAAM,GAAG,EAAE,GAAG,cAAc,EAAE,GAAG,MAAM,EAAE,CAAC;QAC/C,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC,OAAO,CAAC,SAAS,EAAE,oCAAoC,CAAC,CAAC;IAClF,CAAC;IAED;;;;;;;;;;;OAWG;IACH,KAAK,CAAC,YAAY,CAAC,SAAiB;QAClC,OAAO,IAAI,CAAC,SAAS,CAAC,CAAC,WAAW,EAAE,SAAS,CAAC,CAAC,CAAC;IAClD,CAAC;IAED;;;;;;OAMG;IACH,KAAK,CAAC,YAAY,CAChB,SAAiB,EACjB,UAAmB;QAEnB,MAAM,MAAM,GAAG,UAAU,IAAI,IAAI,CAAC,cAAc,CAAC,SAAS,EAAE,KAAK,CAAC,CAAC;QACnE,OAAO,IAAI,CAAC,SAAS,CAAC;YACpB,kBAAkB;YAClB,SAAS;YACT,UAAU;YACV,MAAM;YACN,iBAAiB;YACjB,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,eAAe,CAAC;SACpC,CAAC,CAAC;IACL,CAAC;IAED;;OAEG;IACH,WAAW;QACT,OAAO,EAAE,CAAC,UAAU,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;IACxC,CAAC;IAED;;OAEG;IACK,cAAc,CAAC,SAAiB,EAAE,MAAc;QACtD,MAAM,GAAG,GAAG,IAAI,CAAC,OAAO,CAAC,SAAS,CAAC,CAAC;QACpC,MAAM,IAAI,GAAG,IAAI,CAAC,QAAQ,CAAC,SAAS,EAAE,GAAG,CAAC,CAAC;QAC3C,MAAM,MAAM,GAAG,EAAE,CAAC,MAAM,EAAE,CAAC;QAC3B,OAAO,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE,GAAG,IAAI,IAAI,MAAM,IAAI,IAAI,CAAC,GAAG,EAAE,GAAG,GAAG,EAAE,CAAC,CAAC;IACpE,CAAC;IAED;;;OAGG;IACK,SAAS,CAAI,IAAc;QACjC,OAAO,IAAI,OAAO,CAAC,CAAC,OAAO,EAAE,EAAE;YAC7B,yBAAyB;YACzB,IAAI,CAAC,EAAE,CAAC,UAAU,CAAC,IAAI,CAAC,UAAU,CAAC,EAAE,CAAC;gBACpC,OAAO,CAAC;oBACN,OAAO,EAAE,KAAK;oBACd,KAAK,EAAE,qCAAqC,IAAI,CAAC,UAAU,EAAE;iBACzD,CAAC,CAAC;gBACR,OAAO;YACT,CAAC;YAED,MAAM,IAAI,GAAG,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,UAAU,EAAE,CAAC,IAAI,CAAC,UAAU,EAAE,GAAG,IAAI,CAAC,EAAE;gBACrE,OAAO,EAAE,IAAI,CAAC,MAAM,CAAC,OAAO;aAC7B,CAAC,CAAC;YAEH,IAAI,MAAM,GAAG,EAAE,CAAC;YAChB,IAAI,MAAM,GAAG,EAAE,CAAC;YAChB,IAAI,OAAO,GAAG,KAAK,CAAC;YACpB,IAAI,YAAY,GAAyC,IAAI,CAAC;YAE9D,MAAM,OAAO,GAAG,GAAG,EAAE;gBACnB,IAAI,YAAY,EAAE,CAAC;oBACjB,YAAY,CAAC,YAAY,CAAC,CAAC;oBAC3B,YAAY,GAAG,IAAI,CAAC;gBACtB,CAAC;YACH,CAAC,CAAC;YAEF,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,IAAI,EAAE,EAAE;gBAC9B,MAAM,IAAI,IAAI,CAAC,QAAQ,EAAE,CAAC;YAC5B,CAAC,CAAC,CAAC;YAEH,kEAAkE;YAClE,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,IAAI,EAAE,EAAE;gBAC9B,IAAI,MAAM,CAAC,MAAM,GAAG,iBAAiB,EAAE,CAAC;oBACtC,MAAM,IAAI,IAAI,CAAC,QAAQ,EAAE,CAAC;gBAC5B,CAAC;YACH,CAAC,CAAC,CAAC;YAEH,IAAI,CAAC,EAAE,CAAC,OAAO,EAAE,CAAC,GAAG,EAAE,EAAE;gBACvB,OAAO,EAAE,CAAC;gBACV,IAAI,OAAO;oBAAE,OAAO;gBACpB,OAAO,GAAG,IAAI,CAAC;gBAEf,wEAAwE;gBACxE,uEAAuE;gBACvE,oEAAoE;gBAEpE,OAAO,CAAC;oBACN,OAAO,EAAE,KAAK;oBACd,KAAK,EAAE,mCAAmC,GAAG,CAAC,OAAO,EAAE;iBACnD,CAAC,CAAC;YACV,CAAC,CAAC,CAAC;YAEH,IAAI,CAAC,EAAE,CAAC,OAAO,EAAE,CAAC,IAAI,EAAE,MAAM,EAAE,EAAE;gBAChC,OAAO,EAAE,CAAC;gBACV,IAAI,OAAO;oBAAE,OAAO;gBACpB,OAAO,GAAG,IAAI,CAAC;gBAEf,IAAI,MAAM,EAAE,CAAC;oBACX,OAAO,CAAC,KAAK,CAAC,4BAA4B,MAAM,EAAE,CAAC,CAAC;gBACtD,CAAC;gBAED,IAAI,MAAM,KAAK,SAAS,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;oBACjD,OAAO,CAAC;wBACN,OAAO,EAAE,KAAK;wBACd,KAAK,EAAE,4BAA4B,MAAM,cAAc,IAAI,CAAC,MAAM,CAAC,OAAO,KAAK;qBAC3E,CAAC,CAAC;oBACR,OAAO;gBACT,CAAC;gBAED,IAAI,CAAC;oBACH,uEAAuE;oBACvE,wEAAwE;oBACxE,MAAM,KAAK,GAAG,MAAM,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;oBACxC,IAAI,MAAqB,CAAC;oBAC1B,KAAK,IAAI,CAAC,GAAG,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;wBAC3C,MAAM,IAAI,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;wBAC7B,IAAI,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,IAAI,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;4BACjD,IAAI,CAAC;gCACH,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAM,CAAC;gCAC/B,MAAM;4BACR,CAAC;4BAAC,OAAO,KAAK,EAAE,CAAC;gCACf,OAAO,CAAC,KAAK,CACX,sEAAsE,EACtE,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CACvD,CAAC;gCACF,uCAAuC;4BACzC,CAAC;wBACH,CAAC;oBACH,CAAC;oBACD,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;wBACzB,2DAA2D;wBAC3D,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,MAAM,CAAM,CAAC;oBACnC,CAAC;oBACD,OAAO,CAAC,MAAM,CAAC,CAAC;gBAClB,CAAC;gBAAC,OAAO,UAAU,EAAE,CAAC;oBACpB,IAAI,IAAI,KAAK,CAAC,EAAE,CAAC;wBACf,OAAO,CAAC;4BACN,OAAO,EAAE,KAAK;4BACd,KAAK,EAAE,kCAAkC,IAAI,KAAK,CAAC,MAAM,IAAI,MAAM,CAAC,CAAC,SAAS,CAAC,CAAC,EAAE,IAAI,CAAC,EAAE;yBACrF,CAAC,CAAC;oBACV,CAAC;yBAAM,CAAC;wBACN,OAAO,CAAC;4BACN,OAAO,EAAE,KAAK;4BACd,KAAK,EAAE,2BAA2B,UAAU,EAAE;yBAC1C,CAAC,CAAC;oBACV,CAAC;gBACH,CAAC;YACH,CAAC,CAAC,CAAC;YAEH,oGAAoG;YACpG,iFAAiF;YACjF,IAAI,IAAI,CAAC,MAAM,CAAC,OAAO,GAAG,CAAC,EAAE,CAAC;gBAC5B,YAAY,GAAG,UAAU,CAAC,GAAG,EAAE;oBAC7B,IAAI,CAAC,OAAO,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE,CAAC;wBAC7B,OAAO,CAAC,KAAK,CACX,8EAA8E,IAAI,CAAC,GAAG,GAAG,CAC1F,CAAC;wBACF,IAAI,CAAC;4BACH,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;wBACvB,CAAC;wBAAC,OAAO,KAAK,EAAE,CAAC;4BACf,OAAO,CAAC,KAAK,CACX,mEAAmE,EACnE,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CACvD,CAAC;4BACF,8BAA8B;wBAChC,CAAC;oBACH,CAAC;gBACH,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,OAAO,GAAG,IAAI,CAAC,CAAC;YACjC,CAAC;QACH,CAAC,CAAC,CAAC;IACL,CAAC;CACF;AAED;;GAEG;AACH,IAAI,eAAe,GAA0B,IAAI,CAAC;AAElD;;;;GAIG;AACH,MAAM,UAAU,iBAAiB,CAAC,MAAsC;IACtE,IAAI,MAAM,EAAE,CAAC;QACX,eAAe,GAAG,IAAI,cAAc,CAAC,MAAM,CAAC,CAAC;QAC7C,OAAO,eAAe,CAAC;IACzB,CAAC;IACD,IAAI,CAAC,eAAe,EAAE,CAAC;QACrB,eAAe,GAAG,IAAI,cAAc,EAAE,CAAC;IACzC,CAAC;IACD,OAAO,eAAe,CAAC;AACzB,CAAC"}
|
|
@@ -0,0 +1,64 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* Datalab OCR Bridge
|
|
3
|
+
*
|
|
4
|
+
* Invokes python/ocr_worker.py via python-shell.
|
|
5
|
+
* NO MOCKS, NO FALLBACKS - real API calls only.
|
|
6
|
+
*/
|
|
7
|
+
import type { OCRResult, PageOffset } from '../../models/document.js';
|
|
8
|
+
export interface DatalabClientConfig {
|
|
9
|
+
pythonPath?: string;
|
|
10
|
+
timeout?: number;
|
|
11
|
+
}
|
|
12
|
+
export declare class DatalabClient {
|
|
13
|
+
private readonly pythonPath;
|
|
14
|
+
private readonly workerPath;
|
|
15
|
+
private readonly timeout;
|
|
16
|
+
constructor(config?: DatalabClientConfig);
|
|
17
|
+
/**
|
|
18
|
+
* Process document through Datalab OCR
|
|
19
|
+
*
|
|
20
|
+
* FAIL-FAST: Throws on any error, no fallbacks
|
|
21
|
+
*/
|
|
22
|
+
processDocument(filePath: string, documentId: string, provenanceId: string, mode?: 'fast' | 'balanced' | 'accurate', ocrOptions?: {
|
|
23
|
+
maxPages?: number;
|
|
24
|
+
pageRange?: string;
|
|
25
|
+
skipCache?: boolean;
|
|
26
|
+
disableImageExtraction?: boolean;
|
|
27
|
+
extras?: string[];
|
|
28
|
+
pageSchema?: string;
|
|
29
|
+
additionalConfig?: Record<string, unknown>;
|
|
30
|
+
fileUrl?: string;
|
|
31
|
+
}): Promise<{
|
|
32
|
+
result: OCRResult;
|
|
33
|
+
pageOffsets: PageOffset[];
|
|
34
|
+
images: Record<string, string>;
|
|
35
|
+
jsonBlocks: Record<string, unknown> | null;
|
|
36
|
+
metadata: Record<string, unknown> | null;
|
|
37
|
+
extractionJson: Record<string, unknown> | unknown[] | null;
|
|
38
|
+
docTitle: string | null;
|
|
39
|
+
docAuthor: string | null;
|
|
40
|
+
docSubject: string | null;
|
|
41
|
+
}>;
|
|
42
|
+
/**
|
|
43
|
+
* Process a file through Datalab OCR and return raw results without storing in DB.
|
|
44
|
+
* Used by ocr_convert_raw tool for quick one-off conversions.
|
|
45
|
+
*
|
|
46
|
+
* FAIL-FAST: Throws on any error, no fallbacks
|
|
47
|
+
*/
|
|
48
|
+
processRaw(filePath: string, mode?: 'fast' | 'balanced' | 'accurate', ocrOptions?: {
|
|
49
|
+
maxPages?: number;
|
|
50
|
+
pageRange?: string;
|
|
51
|
+
fileUrl?: string;
|
|
52
|
+
}): Promise<{
|
|
53
|
+
markdown: string;
|
|
54
|
+
pageCount: number;
|
|
55
|
+
qualityScore: number | null;
|
|
56
|
+
costCents: number | null;
|
|
57
|
+
durationMs: number;
|
|
58
|
+
metadata: Record<string, unknown> | null;
|
|
59
|
+
}>;
|
|
60
|
+
private isErrorResponse;
|
|
61
|
+
private toOCRResult;
|
|
62
|
+
private toPageOffsets;
|
|
63
|
+
}
|
|
64
|
+
//# sourceMappingURL=datalab.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"datalab.d.ts","sourceRoot":"","sources":["../../../src/services/ocr/datalab.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAKH,OAAO,KAAK,EAAE,SAAS,EAAE,UAAU,EAAE,MAAM,0BAA0B,CAAC;AAqDtE,MAAM,WAAW,mBAAmB;IAClC,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,OAAO,CAAC,EAAE,MAAM,CAAC;CAClB;AAED,qBAAa,aAAa;IACxB,OAAO,CAAC,QAAQ,CAAC,UAAU,CAAqB;IAChD,OAAO,CAAC,QAAQ,CAAC,UAAU,CAAS;IACpC,OAAO,CAAC,QAAQ,CAAC,OAAO,CAAS;gBAErB,MAAM,GAAE,mBAAwB;IAO5C;;;;OAIG;IACG,eAAe,CACnB,QAAQ,EAAE,MAAM,EAChB,UAAU,EAAE,MAAM,EAClB,YAAY,EAAE,MAAM,EACpB,IAAI,GAAE,MAAM,GAAG,UAAU,GAAG,UAAuB,EACnD,UAAU,CAAC,EAAE;QACX,QAAQ,CAAC,EAAE,MAAM,CAAC;QAClB,SAAS,CAAC,EAAE,MAAM,CAAC;QACnB,SAAS,CAAC,EAAE,OAAO,CAAC;QACpB,sBAAsB,CAAC,EAAE,OAAO,CAAC;QACjC,MAAM,CAAC,EAAE,MAAM,EAAE,CAAC;QAClB,UAAU,CAAC,EAAE,MAAM,CAAC;QACpB,gBAAgB,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;QAC3C,OAAO,CAAC,EAAE,MAAM,CAAC;KAClB,GACA,OAAO,CAAC;QACT,MAAM,EAAE,SAAS,CAAC;QAClB,WAAW,EAAE,UAAU,EAAE,CAAC;QAC1B,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;QAC/B,UAAU,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,IAAI,CAAC;QAC3C,QAAQ,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,IAAI,CAAC;QACzC,cAAc,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,OAAO,EAAE,GAAG,IAAI,CAAC;QAC3D,QAAQ,EAAE,MAAM,GAAG,IAAI,CAAC;QACxB,SAAS,EAAE,MAAM,GAAG,IAAI,CAAC;QACzB,UAAU,EAAE,MAAM,GAAG,IAAI,CAAC;KAC3B,CAAC;IA+NF;;;;;OAKG;IACG,UAAU,CACd,QAAQ,EAAE,MAAM,EAChB,IAAI,GAAE,MAAM,GAAG,UAAU,GAAG,UAAuB,EACnD,UAAU,CAAC,EAAE;QACX,QAAQ,CAAC,EAAE,MAAM,CAAC;QAClB,SAAS,CAAC,EAAE,MAAM,CAAC;QACnB,OAAO,CAAC,EAAE,MAAM,CAAC;KAClB,GACA,OAAO,CAAC;QACT,QAAQ,EAAE,MAAM,CAAC;QACjB,SAAS,EAAE,MAAM,CAAC;QAClB,YAAY,EAAE,MAAM,GAAG,IAAI,CAAC;QAC5B,SAAS,EAAE,MAAM,GAAG,IAAI,CAAC;QACzB,UAAU,EAAE,MAAM,CAAC;QACnB,QAAQ,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,IAAI,CAAC;KAC1C,CAAC;IAoMF,OAAO,CAAC,eAAe;IASvB,OAAO,CAAC,WAAW;IA2BnB,OAAO,CAAC,aAAa;CAMtB"}
|