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 @@
|
|
|
1
|
+
{"version":3,"file":"file-manager.d.ts","sourceRoot":"","sources":["../../../src/services/ocr/file-manager.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAYH,UAAU,gBAAgB;IACxB,MAAM,EAAE,MAAM,CAAC;IACf,SAAS,EAAE,MAAM,GAAG,IAAI,CAAC;IACzB,QAAQ,EAAE,MAAM,CAAC;IACjB,QAAQ,EAAE,MAAM,CAAC;IACjB,QAAQ,EAAE,MAAM,CAAC;IACjB,WAAW,EAAE,MAAM,CAAC;IACpB,MAAM,EAAE,UAAU,GAAG,QAAQ,CAAC;IAC9B,KAAK,EAAE,MAAM,GAAG,IAAI,CAAC;IACrB,oBAAoB,EAAE,MAAM,CAAC;CAC9B;AAED,UAAU,QAAQ;IAChB,MAAM,EAAE,MAAM,CAAC;IACf,QAAQ,EAAE,MAAM,GAAG,IAAI,CAAC;IACxB,QAAQ,EAAE,MAAM,GAAG,IAAI,CAAC;IACxB,WAAW,EAAE,MAAM,GAAG,IAAI,CAAC;IAC3B,SAAS,EAAE,MAAM,GAAG,IAAI,CAAC;IACzB,SAAS,EAAE,MAAM,GAAG,IAAI,CAAC;IACzB,MAAM,EAAE,MAAM,GAAG,IAAI,CAAC;CACvB;AAED,UAAU,cAAc;IACtB,KAAK,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,EAAE,CAAC;IACjC,KAAK,EAAE,MAAM,CAAC;CACf;AAmCD,UAAU,uBAAuB;IAC/B,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,OAAO,CAAC,EAAE,MAAM,CAAC;CAClB;AAED,qBAAa,iBAAiB;IAC5B,OAAO,CAAC,QAAQ,CAAC,UAAU,CAAqB;IAChD,OAAO,CAAC,QAAQ,CAAC,UAAU,CAAS;IACpC,OAAO,CAAC,QAAQ,CAAC,OAAO,CAAS;gBAErB,MAAM,GAAE,uBAA4B;IAOhD;;;;OAIG;IACG,UAAU,CAAC,QAAQ,EAAE,MAAM,GAAG,OAAO,CAAC,gBAAgB,CAAC;IAgB7D;;OAEG;IACG,SAAS,CAAC,KAAK,GAAE,MAAW,EAAE,MAAM,GAAE,MAAU,GAAG,OAAO,CAAC,cAAc,CAAC;IAShF;;OAEG;IACG,OAAO,CAAC,MAAM,EAAE,MAAM,GAAG,OAAO,CAAC,QAAQ,CAAC;IAchD;;;;;OAKG;IACG,cAAc,CAAC,MAAM,EAAE,MAAM,EAAE,SAAS,GAAE,MAAa,GAAG,OAAO,CAAC;QAAE,WAAW,EAAE,MAAM,CAAC;QAAC,SAAS,EAAE,MAAM,CAAC;QAAC,MAAM,EAAE,MAAM,CAAA;KAAE,CAAC;IAUnI;;OAEG;IACG,UAAU,CAAC,MAAM,EAAE,MAAM,GAAG,OAAO,CAAC,OAAO,CAAC;IAMlD;;OAEG;IACH,OAAO,CAAC,SAAS;IAwKjB,OAAO,CAAC,eAAe;CAQxB"}
|
|
@@ -0,0 +1,238 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* Datalab File Manager Bridge
|
|
3
|
+
*
|
|
4
|
+
* Invokes python/file_manager_worker.py via python-shell.
|
|
5
|
+
* NO MOCKS, NO FALLBACKS - real API calls only.
|
|
6
|
+
*/
|
|
7
|
+
import { PythonShell } from 'python-shell';
|
|
8
|
+
import { resolve, dirname } from 'path';
|
|
9
|
+
import { fileURLToPath } from 'url';
|
|
10
|
+
import { OCRError, mapPythonError } from './errors.js';
|
|
11
|
+
const __dirname = dirname(fileURLToPath(import.meta.url));
|
|
12
|
+
/** Max stderr accumulation: 10KB (matches datalab.ts pattern) */
|
|
13
|
+
const MAX_STDERR_LENGTH = 10_240;
|
|
14
|
+
export class FileManagerClient {
|
|
15
|
+
pythonPath;
|
|
16
|
+
workerPath;
|
|
17
|
+
timeout;
|
|
18
|
+
constructor(config = {}) {
|
|
19
|
+
this.pythonPath = config.pythonPath;
|
|
20
|
+
this.workerPath = resolve(__dirname, '../../../python/file_manager_worker.py');
|
|
21
|
+
const parsedTimeout = parseInt(process.env.DATALAB_TIMEOUT || '330000');
|
|
22
|
+
this.timeout = config.timeout ?? (Number.isNaN(parsedTimeout) ? 330000 : parsedTimeout);
|
|
23
|
+
}
|
|
24
|
+
/**
|
|
25
|
+
* Upload a file to Datalab cloud storage
|
|
26
|
+
*
|
|
27
|
+
* FAIL-FAST: Throws on any error, no fallbacks
|
|
28
|
+
*/
|
|
29
|
+
async uploadFile(filePath) {
|
|
30
|
+
const args = ['--action', 'upload', '--file', filePath];
|
|
31
|
+
const response = await this.runWorker(args);
|
|
32
|
+
return {
|
|
33
|
+
fileId: String(response.file_id),
|
|
34
|
+
reference: response.reference,
|
|
35
|
+
fileName: response.file_name,
|
|
36
|
+
fileHash: response.file_hash,
|
|
37
|
+
fileSize: response.file_size,
|
|
38
|
+
contentType: response.content_type,
|
|
39
|
+
status: response.status,
|
|
40
|
+
error: response.error,
|
|
41
|
+
processingDurationMs: response.processing_duration_ms ?? 0,
|
|
42
|
+
};
|
|
43
|
+
}
|
|
44
|
+
/**
|
|
45
|
+
* List files in Datalab cloud storage
|
|
46
|
+
*/
|
|
47
|
+
async listFiles(limit = 50, offset = 0) {
|
|
48
|
+
const args = ['--action', 'list', '--limit', String(limit), '--offset', String(offset)];
|
|
49
|
+
const response = await this.runWorker(args);
|
|
50
|
+
return {
|
|
51
|
+
files: response.files ?? [],
|
|
52
|
+
total: response.total ?? 0,
|
|
53
|
+
};
|
|
54
|
+
}
|
|
55
|
+
/**
|
|
56
|
+
* Get file metadata by ID
|
|
57
|
+
*/
|
|
58
|
+
async getFile(fileId) {
|
|
59
|
+
const args = ['--action', 'get', '--file-id', fileId];
|
|
60
|
+
const response = await this.runWorker(args);
|
|
61
|
+
return {
|
|
62
|
+
fileId: String(response.file_id),
|
|
63
|
+
fileName: response.file_name,
|
|
64
|
+
fileSize: response.file_size,
|
|
65
|
+
contentType: response.content_type,
|
|
66
|
+
createdAt: response.created_at,
|
|
67
|
+
reference: response.reference,
|
|
68
|
+
status: response.status,
|
|
69
|
+
};
|
|
70
|
+
}
|
|
71
|
+
/**
|
|
72
|
+
* Get download URL for a file
|
|
73
|
+
*
|
|
74
|
+
* @param fileId - Datalab file ID
|
|
75
|
+
* @param expiresIn - URL expiry time in seconds (default: 3600, min: 60, max: 86400)
|
|
76
|
+
*/
|
|
77
|
+
async getDownloadUrl(fileId, expiresIn = 3600) {
|
|
78
|
+
const args = ['--action', 'download-url', '--file-id', fileId, '--expires-in', String(expiresIn)];
|
|
79
|
+
const response = await this.runWorker(args);
|
|
80
|
+
return {
|
|
81
|
+
downloadUrl: response.download_url,
|
|
82
|
+
expiresIn: response.expires_in,
|
|
83
|
+
fileId: String(response.file_id),
|
|
84
|
+
};
|
|
85
|
+
}
|
|
86
|
+
/**
|
|
87
|
+
* Delete a file from Datalab cloud storage
|
|
88
|
+
*/
|
|
89
|
+
async deleteFile(fileId) {
|
|
90
|
+
const args = ['--action', 'delete', '--file-id', fileId];
|
|
91
|
+
const response = await this.runWorker(args);
|
|
92
|
+
return response.deleted;
|
|
93
|
+
}
|
|
94
|
+
/**
|
|
95
|
+
* Run the Python worker with given args and parse JSON output
|
|
96
|
+
*/
|
|
97
|
+
runWorker(args) {
|
|
98
|
+
const shellOptions = {
|
|
99
|
+
mode: 'text',
|
|
100
|
+
pythonPath: this.pythonPath,
|
|
101
|
+
pythonOptions: ['-u'],
|
|
102
|
+
args,
|
|
103
|
+
};
|
|
104
|
+
return new Promise((promiseResolve, reject) => {
|
|
105
|
+
let settled = false;
|
|
106
|
+
const shell = new PythonShell(this.workerPath, shellOptions);
|
|
107
|
+
const outputChunks = [];
|
|
108
|
+
let stderr = '';
|
|
109
|
+
let sigkillTimer = null;
|
|
110
|
+
const cleanup = () => {
|
|
111
|
+
if (sigkillTimer) {
|
|
112
|
+
clearTimeout(sigkillTimer);
|
|
113
|
+
sigkillTimer = null;
|
|
114
|
+
}
|
|
115
|
+
};
|
|
116
|
+
const timer = setTimeout(() => {
|
|
117
|
+
if (settled)
|
|
118
|
+
return;
|
|
119
|
+
try {
|
|
120
|
+
shell.kill();
|
|
121
|
+
}
|
|
122
|
+
catch (error) {
|
|
123
|
+
console.error('[FileManager] Failed to kill shell on timeout:', error instanceof Error ? error.message : String(error));
|
|
124
|
+
/* ignore */
|
|
125
|
+
}
|
|
126
|
+
// M-6: SIGKILL escalation if SIGTERM doesn't exit within 5s
|
|
127
|
+
sigkillTimer = setTimeout(() => {
|
|
128
|
+
if (!settled) {
|
|
129
|
+
console.error(`[FileManager] Process did not exit after SIGTERM, sending SIGKILL (pid: ${shell.childProcess?.pid})`);
|
|
130
|
+
try {
|
|
131
|
+
shell.childProcess?.kill('SIGKILL');
|
|
132
|
+
}
|
|
133
|
+
catch (error) {
|
|
134
|
+
console.error('[FileManager] Failed to SIGKILL process (may already be gone):', error instanceof Error ? error.message : String(error));
|
|
135
|
+
}
|
|
136
|
+
}
|
|
137
|
+
if (!settled) {
|
|
138
|
+
settled = true;
|
|
139
|
+
reject(new OCRError(`File manager timeout after ${this.timeout}ms (SIGKILL after 5s grace)`, 'OCR_TIMEOUT'));
|
|
140
|
+
}
|
|
141
|
+
}, 5000);
|
|
142
|
+
}, this.timeout);
|
|
143
|
+
shell.on('message', (msg) => {
|
|
144
|
+
outputChunks.push(msg);
|
|
145
|
+
});
|
|
146
|
+
shell.on('stderr', (err) => {
|
|
147
|
+
if (stderr.length < MAX_STDERR_LENGTH) {
|
|
148
|
+
stderr += err + '\n';
|
|
149
|
+
}
|
|
150
|
+
});
|
|
151
|
+
shell.end((err) => {
|
|
152
|
+
clearTimeout(timer);
|
|
153
|
+
cleanup();
|
|
154
|
+
if (settled)
|
|
155
|
+
return;
|
|
156
|
+
settled = true;
|
|
157
|
+
// Capture exit code/signal for diagnostics
|
|
158
|
+
const exitCode = shell.exitCode ?? null;
|
|
159
|
+
const exitSignal = shell.exitSignal ?? null;
|
|
160
|
+
const output = outputChunks.join('\n');
|
|
161
|
+
outputChunks.length = 0;
|
|
162
|
+
// python-shell bug: when process is killed by signal (SIGTERM/SIGKILL),
|
|
163
|
+
// exitCode is null and exitSignal is set, but err is NOT provided.
|
|
164
|
+
if (!err && exitSignal) {
|
|
165
|
+
const signalDetail = stderr.trim()
|
|
166
|
+
? `Process killed by ${exitSignal}. Python stderr:\n${stderr.trim()}`
|
|
167
|
+
: `Process killed by ${exitSignal} with no stderr output.`;
|
|
168
|
+
reject(new OCRError(`File manager worker killed by signal ${exitSignal}. ${signalDetail}`, exitSignal === 'SIGTERM' || exitSignal === 'SIGALRM' ? 'OCR_TIMEOUT' : 'FILE_MANAGER_API_ERROR'));
|
|
169
|
+
return;
|
|
170
|
+
}
|
|
171
|
+
if (err) {
|
|
172
|
+
const lines = output
|
|
173
|
+
.trim()
|
|
174
|
+
.split('\n')
|
|
175
|
+
.filter((l) => l.trim());
|
|
176
|
+
for (let i = lines.length - 1; i >= 0; i--) {
|
|
177
|
+
try {
|
|
178
|
+
const parsed = JSON.parse(lines[i].trim());
|
|
179
|
+
if (this.isErrorResponse(parsed)) {
|
|
180
|
+
reject(mapPythonError(parsed.category, parsed.error, parsed.details ?? {}));
|
|
181
|
+
return;
|
|
182
|
+
}
|
|
183
|
+
}
|
|
184
|
+
catch (error) {
|
|
185
|
+
console.error('[FileManager] JSON parse failed for error output line, skipping:', error instanceof Error ? error.message : String(error));
|
|
186
|
+
/* not JSON, skip */
|
|
187
|
+
}
|
|
188
|
+
}
|
|
189
|
+
const detail = stderr ? `${err.message}\nPython stderr:\n${stderr}` : err.message;
|
|
190
|
+
reject(new OCRError(`File manager worker failed: ${detail}`, 'OCR_API_ERROR'));
|
|
191
|
+
return;
|
|
192
|
+
}
|
|
193
|
+
const lines = output
|
|
194
|
+
.trim()
|
|
195
|
+
.split('\n')
|
|
196
|
+
.filter((l) => l.trim());
|
|
197
|
+
if (lines.length === 0) {
|
|
198
|
+
const diagnostics = [
|
|
199
|
+
'No output from file manager worker.',
|
|
200
|
+
`Exit code: ${exitCode}, signal: ${exitSignal}.`,
|
|
201
|
+
];
|
|
202
|
+
if (stderr.trim()) {
|
|
203
|
+
diagnostics.push(`Python stderr: ${stderr.trim().substring(0, 500)}`);
|
|
204
|
+
}
|
|
205
|
+
reject(new OCRError(diagnostics.join(' '), 'FILE_MANAGER_API_ERROR'));
|
|
206
|
+
return;
|
|
207
|
+
}
|
|
208
|
+
let response;
|
|
209
|
+
for (let i = lines.length - 1; i >= 0; i--) {
|
|
210
|
+
try {
|
|
211
|
+
response = JSON.parse(lines[i].trim());
|
|
212
|
+
break;
|
|
213
|
+
}
|
|
214
|
+
catch (error) {
|
|
215
|
+
console.error('[FileManager] JSON parse failed for output line, trying previous:', error instanceof Error ? error.message : String(error));
|
|
216
|
+
/* not JSON, try previous line */
|
|
217
|
+
}
|
|
218
|
+
}
|
|
219
|
+
if (!response) {
|
|
220
|
+
reject(new OCRError(`Failed to parse file manager output as JSON. Last line: ${lines[lines.length - 1]?.substring(0, 200)}`, 'OCR_API_ERROR'));
|
|
221
|
+
return;
|
|
222
|
+
}
|
|
223
|
+
if (this.isErrorResponse(response)) {
|
|
224
|
+
reject(mapPythonError(response.category, response.error, response.details ?? {}));
|
|
225
|
+
return;
|
|
226
|
+
}
|
|
227
|
+
promiseResolve(response);
|
|
228
|
+
});
|
|
229
|
+
});
|
|
230
|
+
}
|
|
231
|
+
isErrorResponse(response) {
|
|
232
|
+
return (typeof response === 'object' &&
|
|
233
|
+
response !== null &&
|
|
234
|
+
'error' in response &&
|
|
235
|
+
'category' in response);
|
|
236
|
+
}
|
|
237
|
+
}
|
|
238
|
+
//# sourceMappingURL=file-manager.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"file-manager.js","sourceRoot":"","sources":["../../../src/services/ocr/file-manager.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO,EAAE,WAAW,EAAW,MAAM,cAAc,CAAC;AACpD,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,MAAM,MAAM,CAAC;AACxC,OAAO,EAAE,aAAa,EAAE,MAAM,KAAK,CAAC;AACpC,OAAO,EAAE,QAAQ,EAAE,cAAc,EAAE,MAAM,aAAa,CAAC;AAEvD,MAAM,SAAS,GAAG,OAAO,CAAC,aAAa,CAAC,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;AAE1D,iEAAiE;AACjE,MAAM,iBAAiB,GAAG,MAAM,CAAC;AAmEjC,MAAM,OAAO,iBAAiB;IACX,UAAU,CAAqB;IAC/B,UAAU,CAAS;IACnB,OAAO,CAAS;IAEjC,YAAY,SAAkC,EAAE;QAC9C,IAAI,CAAC,UAAU,GAAG,MAAM,CAAC,UAAU,CAAC;QACpC,IAAI,CAAC,UAAU,GAAG,OAAO,CAAC,SAAS,EAAE,wCAAwC,CAAC,CAAC;QAC/E,MAAM,aAAa,GAAG,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,eAAe,IAAI,QAAQ,CAAC,CAAC;QACxE,IAAI,CAAC,OAAO,GAAG,MAAM,CAAC,OAAO,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,aAAa,CAAC,CAAC;IAC1F,CAAC;IAED;;;;OAIG;IACH,KAAK,CAAC,UAAU,CAAC,QAAgB;QAC/B,MAAM,IAAI,GAAG,CAAC,UAAU,EAAE,QAAQ,EAAE,QAAQ,EAAE,QAAQ,CAAC,CAAC;QACxD,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,SAAS,CAAuB,IAAI,CAAC,CAAC;QAClE,OAAO;YACL,MAAM,EAAE,MAAM,CAAC,QAAQ,CAAC,OAAO,CAAC;YAChC,SAAS,EAAE,QAAQ,CAAC,SAAS;YAC7B,QAAQ,EAAE,QAAQ,CAAC,SAAS;YAC5B,QAAQ,EAAE,QAAQ,CAAC,SAAS;YAC5B,QAAQ,EAAE,QAAQ,CAAC,SAAS;YAC5B,WAAW,EAAE,QAAQ,CAAC,YAAY;YAClC,MAAM,EAAE,QAAQ,CAAC,MAA+B;YAChD,KAAK,EAAE,QAAQ,CAAC,KAAK;YACrB,oBAAoB,EAAE,QAAQ,CAAC,sBAAsB,IAAI,CAAC;SAC3D,CAAC;IACJ,CAAC;IAED;;OAEG;IACH,KAAK,CAAC,SAAS,CAAC,QAAgB,EAAE,EAAE,SAAiB,CAAC;QACpD,MAAM,IAAI,GAAG,CAAC,UAAU,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,CAAC,KAAK,CAAC,EAAE,UAAU,EAAE,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC;QACxF,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,SAAS,CAAyB,IAAI,CAAC,CAAC;QACpE,OAAO;YACL,KAAK,EAAE,QAAQ,CAAC,KAAK,IAAI,EAAE;YAC3B,KAAK,EAAE,QAAQ,CAAC,KAAK,IAAI,CAAC;SAC3B,CAAC;IACJ,CAAC;IAED;;OAEG;IACH,KAAK,CAAC,OAAO,CAAC,MAAc;QAC1B,MAAM,IAAI,GAAG,CAAC,UAAU,EAAE,KAAK,EAAE,WAAW,EAAE,MAAM,CAAC,CAAC;QACtD,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,SAAS,CAAyB,IAAI,CAAC,CAAC;QACpE,OAAO;YACL,MAAM,EAAE,MAAM,CAAC,QAAQ,CAAC,OAAO,CAAC;YAChC,QAAQ,EAAE,QAAQ,CAAC,SAAS;YAC5B,QAAQ,EAAE,QAAQ,CAAC,SAAS;YAC5B,WAAW,EAAE,QAAQ,CAAC,YAAY;YAClC,SAAS,EAAE,QAAQ,CAAC,UAAU;YAC9B,SAAS,EAAE,QAAQ,CAAC,SAAS;YAC7B,MAAM,EAAE,QAAQ,CAAC,MAAM;SACxB,CAAC;IACJ,CAAC;IAED;;;;;OAKG;IACH,KAAK,CAAC,cAAc,CAAC,MAAc,EAAE,YAAoB,IAAI;QAC3D,MAAM,IAAI,GAAG,CAAC,UAAU,EAAE,cAAc,EAAE,WAAW,EAAE,MAAM,EAAE,cAAc,EAAE,MAAM,CAAC,SAAS,CAAC,CAAC,CAAC;QAClG,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,SAAS,CAAgE,IAAI,CAAC,CAAC;QAC3G,OAAO;YACL,WAAW,EAAE,QAAQ,CAAC,YAAY;YAClC,SAAS,EAAE,QAAQ,CAAC,UAAU;YAC9B,MAAM,EAAE,MAAM,CAAC,QAAQ,CAAC,OAAO,CAAC;SACjC,CAAC;IACJ,CAAC;IAED;;OAEG;IACH,KAAK,CAAC,UAAU,CAAC,MAAc;QAC7B,MAAM,IAAI,GAAG,CAAC,UAAU,EAAE,QAAQ,EAAE,WAAW,EAAE,MAAM,CAAC,CAAC;QACzD,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,SAAS,CAAuB,IAAI,CAAC,CAAC;QAClE,OAAO,QAAQ,CAAC,OAAO,CAAC;IAC1B,CAAC;IAED;;OAEG;IACK,SAAS,CAAI,IAAc;QACjC,MAAM,YAAY,GAAY;YAC5B,IAAI,EAAE,MAAM;YACZ,UAAU,EAAE,IAAI,CAAC,UAAU;YAC3B,aAAa,EAAE,CAAC,IAAI,CAAC;YACrB,IAAI;SACL,CAAC;QAEF,OAAO,IAAI,OAAO,CAAC,CAAC,cAAc,EAAE,MAAM,EAAE,EAAE;YAC5C,IAAI,OAAO,GAAG,KAAK,CAAC;YACpB,MAAM,KAAK,GAAG,IAAI,WAAW,CAAC,IAAI,CAAC,UAAU,EAAE,YAAY,CAAC,CAAC;YAC7D,MAAM,YAAY,GAAa,EAAE,CAAC;YAClC,IAAI,MAAM,GAAG,EAAE,CAAC;YAChB,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,MAAM,KAAK,GAAG,UAAU,CAAC,GAAG,EAAE;gBAC5B,IAAI,OAAO;oBAAE,OAAO;gBACpB,IAAI,CAAC;oBACH,KAAK,CAAC,IAAI,EAAE,CAAC;gBACf,CAAC;gBAAC,OAAO,KAAK,EAAE,CAAC;oBACf,OAAO,CAAC,KAAK,CACX,gDAAgD,EAChD,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CACvD,CAAC;oBACF,YAAY;gBACd,CAAC;gBACD,4DAA4D;gBAC5D,YAAY,GAAG,UAAU,CAAC,GAAG,EAAE;oBAC7B,IAAI,CAAC,OAAO,EAAE,CAAC;wBACb,OAAO,CAAC,KAAK,CACX,2EAA2E,KAAK,CAAC,YAAY,EAAE,GAAG,GAAG,CACtG,CAAC;wBACF,IAAI,CAAC;4BACH,KAAK,CAAC,YAAY,EAAE,IAAI,CAAC,SAAS,CAAC,CAAC;wBACtC,CAAC;wBAAC,OAAO,KAAK,EAAE,CAAC;4BACf,OAAO,CAAC,KAAK,CACX,gEAAgE,EAChE,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CACvD,CAAC;wBACJ,CAAC;oBACH,CAAC;oBACD,IAAI,CAAC,OAAO,EAAE,CAAC;wBACb,OAAO,GAAG,IAAI,CAAC;wBACf,MAAM,CAAC,IAAI,QAAQ,CAAC,8BAA8B,IAAI,CAAC,OAAO,6BAA6B,EAAE,aAAa,CAAC,CAAC,CAAC;oBAC/G,CAAC;gBACH,CAAC,EAAE,IAAI,CAAC,CAAC;YACX,CAAC,EAAE,IAAI,CAAC,OAAO,CAAC,CAAC;YAEjB,KAAK,CAAC,EAAE,CAAC,SAAS,EAAE,CAAC,GAAW,EAAE,EAAE;gBAClC,YAAY,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;YACzB,CAAC,CAAC,CAAC;YAEH,KAAK,CAAC,EAAE,CAAC,QAAQ,EAAE,CAAC,GAAW,EAAE,EAAE;gBACjC,IAAI,MAAM,CAAC,MAAM,GAAG,iBAAiB,EAAE,CAAC;oBACtC,MAAM,IAAI,GAAG,GAAG,IAAI,CAAC;gBACvB,CAAC;YACH,CAAC,CAAC,CAAC;YAEH,KAAK,CAAC,GAAG,CAAC,CAAC,GAAW,EAAE,EAAE;gBACxB,YAAY,CAAC,KAAK,CAAC,CAAC;gBACpB,OAAO,EAAE,CAAC;gBACV,IAAI,OAAO;oBAAE,OAAO;gBACpB,OAAO,GAAG,IAAI,CAAC;gBAEf,2CAA2C;gBAC3C,MAAM,QAAQ,GAAG,KAAK,CAAC,QAAQ,IAAI,IAAI,CAAC;gBACxC,MAAM,UAAU,GAAG,KAAK,CAAC,UAAU,IAAI,IAAI,CAAC;gBAE5C,MAAM,MAAM,GAAG,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;gBACvC,YAAY,CAAC,MAAM,GAAG,CAAC,CAAC;gBAExB,wEAAwE;gBACxE,mEAAmE;gBACnE,IAAI,CAAC,GAAG,IAAI,UAAU,EAAE,CAAC;oBACvB,MAAM,YAAY,GAAG,MAAM,CAAC,IAAI,EAAE;wBAChC,CAAC,CAAC,qBAAqB,UAAU,qBAAqB,MAAM,CAAC,IAAI,EAAE,EAAE;wBACrE,CAAC,CAAC,qBAAqB,UAAU,yBAAyB,CAAC;oBAC7D,MAAM,CACJ,IAAI,QAAQ,CACV,wCAAwC,UAAU,KAAK,YAAY,EAAE,EACrE,UAAU,KAAK,SAAS,IAAI,UAAU,KAAK,SAAS,CAAC,CAAC,CAAC,aAAa,CAAC,CAAC,CAAC,wBAAwB,CAChG,CACF,CAAC;oBACF,OAAO;gBACT,CAAC;gBAED,IAAI,GAAG,EAAE,CAAC;oBACR,MAAM,KAAK,GAAG,MAAM;yBACjB,IAAI,EAAE;yBACN,KAAK,CAAC,IAAI,CAAC;yBACX,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;oBAC3B,KAAK,IAAI,CAAC,GAAG,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;wBAC3C,IAAI,CAAC;4BACH,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAY,CAAC;4BACtD,IAAI,IAAI,CAAC,eAAe,CAAC,MAAM,CAAC,EAAE,CAAC;gCACjC,MAAM,CAAC,cAAc,CAAC,MAAM,CAAC,QAAQ,EAAE,MAAM,CAAC,KAAK,EAAE,MAAM,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,CAAC;gCAC5E,OAAO;4BACT,CAAC;wBACH,CAAC;wBAAC,OAAO,KAAK,EAAE,CAAC;4BACf,OAAO,CAAC,KAAK,CACX,kEAAkE,EAClE,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CACvD,CAAC;4BACF,oBAAoB;wBACtB,CAAC;oBACH,CAAC;oBACD,MAAM,MAAM,GAAG,MAAM,CAAC,CAAC,CAAC,GAAG,GAAG,CAAC,OAAO,qBAAqB,MAAM,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC;oBAClF,MAAM,CAAC,IAAI,QAAQ,CAAC,+BAA+B,MAAM,EAAE,EAAE,eAAe,CAAC,CAAC,CAAC;oBAC/E,OAAO;gBACT,CAAC;gBAED,MAAM,KAAK,GAAG,MAAM;qBACjB,IAAI,EAAE;qBACN,KAAK,CAAC,IAAI,CAAC;qBACX,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;gBAC3B,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;oBACvB,MAAM,WAAW,GAAG;wBAClB,qCAAqC;wBACrC,cAAc,QAAQ,aAAa,UAAU,GAAG;qBACjD,CAAC;oBACF,IAAI,MAAM,CAAC,IAAI,EAAE,EAAE,CAAC;wBAClB,WAAW,CAAC,IAAI,CAAC,kBAAkB,MAAM,CAAC,IAAI,EAAE,CAAC,SAAS,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,CAAC,CAAC;oBACxE,CAAC;oBACD,MAAM,CAAC,IAAI,QAAQ,CAAC,WAAW,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,wBAAwB,CAAC,CAAC,CAAC;oBACtE,OAAO;gBACT,CAAC;gBAED,IAAI,QAAiB,CAAC;gBACtB,KAAK,IAAI,CAAC,GAAG,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;oBAC3C,IAAI,CAAC;wBACH,QAAQ,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;wBACvC,MAAM;oBACR,CAAC;oBAAC,OAAO,KAAK,EAAE,CAAC;wBACf,OAAO,CAAC,KAAK,CACX,mEAAmE,EACnE,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CACvD,CAAC;wBACF,iCAAiC;oBACnC,CAAC;gBACH,CAAC;gBAED,IAAI,CAAC,QAAQ,EAAE,CAAC;oBACd,MAAM,CACJ,IAAI,QAAQ,CACV,2DAA2D,KAAK,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,EACvG,eAAe,CAChB,CACF,CAAC;oBACF,OAAO;gBACT,CAAC;gBAED,IAAI,IAAI,CAAC,eAAe,CAAC,QAAQ,CAAC,EAAE,CAAC;oBACnC,MAAM,CAAC,cAAc,CAAC,QAAQ,CAAC,QAAQ,EAAE,QAAQ,CAAC,KAAK,EAAE,QAAQ,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,CAAC;oBAClF,OAAO;gBACT,CAAC;gBAED,cAAc,CAAC,QAAa,CAAC,CAAC;YAChC,CAAC,CAAC,CAAC;QACL,CAAC,CAAC,CAAC;IACL,CAAC;IAEO,eAAe,CAAC,QAAiB;QACvC,OAAO,CACL,OAAO,QAAQ,KAAK,QAAQ;YAC5B,QAAQ,KAAK,IAAI;YACjB,OAAO,IAAI,QAAQ;YACnB,UAAU,IAAI,QAAQ,CACvB,CAAC;IACJ,CAAC;CACF"}
|
|
@@ -0,0 +1,48 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* Datalab Form Fill Bridge
|
|
3
|
+
*
|
|
4
|
+
* Invokes python/form_fill_worker.py via python-shell.
|
|
5
|
+
* NO MOCKS, NO FALLBACKS - real API calls only.
|
|
6
|
+
*/
|
|
7
|
+
interface FormFillOptions {
|
|
8
|
+
fieldData: Record<string, {
|
|
9
|
+
value: string;
|
|
10
|
+
description?: string;
|
|
11
|
+
}>;
|
|
12
|
+
context?: string;
|
|
13
|
+
confidenceThreshold?: number;
|
|
14
|
+
pageRange?: string;
|
|
15
|
+
}
|
|
16
|
+
interface FormFillResult {
|
|
17
|
+
id: string;
|
|
18
|
+
sourceFilePath: string;
|
|
19
|
+
sourceFileHash: string;
|
|
20
|
+
outputBase64: string | null;
|
|
21
|
+
fieldsFilled: string[];
|
|
22
|
+
fieldsNotFound: string[];
|
|
23
|
+
pageCount: number | null;
|
|
24
|
+
costCents: number | null;
|
|
25
|
+
status: 'complete' | 'failed';
|
|
26
|
+
error: string | null;
|
|
27
|
+
processingDurationMs: number;
|
|
28
|
+
}
|
|
29
|
+
interface FormFillClientConfig {
|
|
30
|
+
pythonPath?: string;
|
|
31
|
+
timeout?: number;
|
|
32
|
+
}
|
|
33
|
+
export declare class FormFillClient {
|
|
34
|
+
private readonly pythonPath;
|
|
35
|
+
private readonly workerPath;
|
|
36
|
+
private readonly timeout;
|
|
37
|
+
constructor(config?: FormFillClientConfig);
|
|
38
|
+
/**
|
|
39
|
+
* Fill a form document through Datalab API
|
|
40
|
+
*
|
|
41
|
+
* FAIL-FAST: Throws on any error, no fallbacks
|
|
42
|
+
*/
|
|
43
|
+
fillForm(filePath: string, options: FormFillOptions): Promise<FormFillResult>;
|
|
44
|
+
private isErrorResponse;
|
|
45
|
+
private toFormFillResult;
|
|
46
|
+
}
|
|
47
|
+
export {};
|
|
48
|
+
//# sourceMappingURL=form-fill.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"form-fill.d.ts","sourceRoot":"","sources":["../../../src/services/ocr/form-fill.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAYH,UAAU,eAAe;IACvB,SAAS,EAAE,MAAM,CAAC,MAAM,EAAE;QAAE,KAAK,EAAE,MAAM,CAAC;QAAC,WAAW,CAAC,EAAE,MAAM,CAAA;KAAE,CAAC,CAAC;IACnE,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,mBAAmB,CAAC,EAAE,MAAM,CAAC;IAC7B,SAAS,CAAC,EAAE,MAAM,CAAC;CACpB;AAED,UAAU,cAAc;IACtB,EAAE,EAAE,MAAM,CAAC;IACX,cAAc,EAAE,MAAM,CAAC;IACvB,cAAc,EAAE,MAAM,CAAC;IACvB,YAAY,EAAE,MAAM,GAAG,IAAI,CAAC;IAC5B,YAAY,EAAE,MAAM,EAAE,CAAC;IACvB,cAAc,EAAE,MAAM,EAAE,CAAC;IACzB,SAAS,EAAE,MAAM,GAAG,IAAI,CAAC;IACzB,SAAS,EAAE,MAAM,GAAG,IAAI,CAAC;IACzB,MAAM,EAAE,UAAU,GAAG,QAAQ,CAAC;IAC9B,KAAK,EAAE,MAAM,GAAG,IAAI,CAAC;IACrB,oBAAoB,EAAE,MAAM,CAAC;CAC9B;AAsBD,UAAU,oBAAoB;IAC5B,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,OAAO,CAAC,EAAE,MAAM,CAAC;CAClB;AAED,qBAAa,cAAc;IACzB,OAAO,CAAC,QAAQ,CAAC,UAAU,CAAqB;IAChD,OAAO,CAAC,QAAQ,CAAC,UAAU,CAAS;IACpC,OAAO,CAAC,QAAQ,CAAC,OAAO,CAAS;gBAErB,MAAM,GAAE,oBAAyB;IAO7C;;;;OAIG;IACG,QAAQ,CAAC,QAAQ,EAAE,MAAM,EAAE,OAAO,EAAE,eAAe,GAAG,OAAO,CAAC,cAAc,CAAC;IAsMnF,OAAO,CAAC,eAAe;IASvB,OAAO,CAAC,gBAAgB;CAezB"}
|
|
@@ -0,0 +1,213 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* Datalab Form Fill Bridge
|
|
3
|
+
*
|
|
4
|
+
* Invokes python/form_fill_worker.py via python-shell.
|
|
5
|
+
* NO MOCKS, NO FALLBACKS - real API calls only.
|
|
6
|
+
*/
|
|
7
|
+
import { PythonShell } from 'python-shell';
|
|
8
|
+
import { resolve, dirname } from 'path';
|
|
9
|
+
import { fileURLToPath } from 'url';
|
|
10
|
+
import { OCRError, mapPythonError } from './errors.js';
|
|
11
|
+
const __dirname = dirname(fileURLToPath(import.meta.url));
|
|
12
|
+
/** Max stderr accumulation: 10KB (matches datalab.ts pattern) */
|
|
13
|
+
const MAX_STDERR_LENGTH = 10_240;
|
|
14
|
+
export class FormFillClient {
|
|
15
|
+
pythonPath;
|
|
16
|
+
workerPath;
|
|
17
|
+
timeout;
|
|
18
|
+
constructor(config = {}) {
|
|
19
|
+
this.pythonPath = config.pythonPath;
|
|
20
|
+
this.workerPath = resolve(__dirname, '../../../python/form_fill_worker.py');
|
|
21
|
+
const parsedTimeout = parseInt(process.env.DATALAB_TIMEOUT || '330000');
|
|
22
|
+
this.timeout = config.timeout ?? (Number.isNaN(parsedTimeout) ? 330000 : parsedTimeout);
|
|
23
|
+
}
|
|
24
|
+
/**
|
|
25
|
+
* Fill a form document through Datalab API
|
|
26
|
+
*
|
|
27
|
+
* FAIL-FAST: Throws on any error, no fallbacks
|
|
28
|
+
*/
|
|
29
|
+
async fillForm(filePath, options) {
|
|
30
|
+
const args = ['--file', filePath, '--field-data', JSON.stringify(options.fieldData), '--json'];
|
|
31
|
+
if (options.context)
|
|
32
|
+
args.push('--context', options.context);
|
|
33
|
+
if (options.confidenceThreshold !== undefined)
|
|
34
|
+
args.push('--confidence-threshold', String(options.confidenceThreshold));
|
|
35
|
+
if (options.pageRange)
|
|
36
|
+
args.push('--page-range', options.pageRange);
|
|
37
|
+
const shellOptions = {
|
|
38
|
+
mode: 'text',
|
|
39
|
+
pythonPath: this.pythonPath,
|
|
40
|
+
pythonOptions: ['-u'],
|
|
41
|
+
args,
|
|
42
|
+
};
|
|
43
|
+
return new Promise((promiseResolve, reject) => {
|
|
44
|
+
let settled = false;
|
|
45
|
+
const shell = new PythonShell(this.workerPath, shellOptions);
|
|
46
|
+
const outputChunks = [];
|
|
47
|
+
let stderr = '';
|
|
48
|
+
let sigkillTimer = null;
|
|
49
|
+
const cleanup = () => {
|
|
50
|
+
if (sigkillTimer) {
|
|
51
|
+
clearTimeout(sigkillTimer);
|
|
52
|
+
sigkillTimer = null;
|
|
53
|
+
}
|
|
54
|
+
};
|
|
55
|
+
const timer = setTimeout(() => {
|
|
56
|
+
if (settled)
|
|
57
|
+
return;
|
|
58
|
+
try {
|
|
59
|
+
shell.kill();
|
|
60
|
+
}
|
|
61
|
+
catch (error) {
|
|
62
|
+
console.error('[FormFill] Failed to kill shell on timeout:', error instanceof Error ? error.message : String(error));
|
|
63
|
+
/* ignore */
|
|
64
|
+
}
|
|
65
|
+
// M-6: SIGKILL escalation if SIGTERM doesn't exit within 5s
|
|
66
|
+
sigkillTimer = setTimeout(() => {
|
|
67
|
+
if (!settled) {
|
|
68
|
+
console.error(`[FormFill] Process did not exit after SIGTERM, sending SIGKILL (pid: ${shell.childProcess?.pid})`);
|
|
69
|
+
try {
|
|
70
|
+
shell.childProcess?.kill('SIGKILL');
|
|
71
|
+
}
|
|
72
|
+
catch (error) {
|
|
73
|
+
console.error('[FormFill] Failed to SIGKILL process (may already be gone):', error instanceof Error ? error.message : String(error));
|
|
74
|
+
}
|
|
75
|
+
}
|
|
76
|
+
if (!settled) {
|
|
77
|
+
settled = true;
|
|
78
|
+
reject(new OCRError(`Form fill timeout after ${this.timeout}ms (SIGKILL after 5s grace)`, 'FORM_FILL_TIMEOUT'));
|
|
79
|
+
}
|
|
80
|
+
}, 5000);
|
|
81
|
+
}, this.timeout);
|
|
82
|
+
shell.on('message', (msg) => {
|
|
83
|
+
outputChunks.push(msg);
|
|
84
|
+
});
|
|
85
|
+
shell.on('stderr', (err) => {
|
|
86
|
+
if (stderr.length < MAX_STDERR_LENGTH) {
|
|
87
|
+
stderr += err + '\n';
|
|
88
|
+
}
|
|
89
|
+
});
|
|
90
|
+
shell.end((err) => {
|
|
91
|
+
clearTimeout(timer);
|
|
92
|
+
cleanup();
|
|
93
|
+
if (settled)
|
|
94
|
+
return;
|
|
95
|
+
settled = true;
|
|
96
|
+
// Capture exit code/signal for diagnostics
|
|
97
|
+
const exitCode = shell.exitCode ?? null;
|
|
98
|
+
const exitSignal = shell.exitSignal ?? null;
|
|
99
|
+
// Join chunks once instead of repeated string concatenation
|
|
100
|
+
const output = outputChunks.join('\n');
|
|
101
|
+
// Allow early GC of chunk array
|
|
102
|
+
outputChunks.length = 0;
|
|
103
|
+
// python-shell bug: when process is killed by signal (SIGTERM/SIGKILL),
|
|
104
|
+
// exitCode is null and exitSignal is set, but err is NOT provided.
|
|
105
|
+
if (!err && exitSignal) {
|
|
106
|
+
const signalDetail = stderr.trim()
|
|
107
|
+
? `Process killed by ${exitSignal}. Python stderr:\n${stderr.trim()}`
|
|
108
|
+
: `Process killed by ${exitSignal} with no stderr output.`;
|
|
109
|
+
reject(new OCRError(`Form fill worker killed by signal ${exitSignal} (file: ${filePath}). ${signalDetail}`, exitSignal === 'SIGTERM' || exitSignal === 'SIGALRM' ? 'FORM_FILL_TIMEOUT' : 'FORM_FILL_API_ERROR'));
|
|
110
|
+
return;
|
|
111
|
+
}
|
|
112
|
+
if (err) {
|
|
113
|
+
// Try to parse JSON from output for structured error
|
|
114
|
+
const lines = output
|
|
115
|
+
.trim()
|
|
116
|
+
.split('\n')
|
|
117
|
+
.filter((l) => l.trim());
|
|
118
|
+
for (let i = lines.length - 1; i >= 0; i--) {
|
|
119
|
+
try {
|
|
120
|
+
const parsed = JSON.parse(lines[i].trim());
|
|
121
|
+
if (this.isErrorResponse(parsed)) {
|
|
122
|
+
reject(mapPythonError(parsed.category, parsed.error, parsed.details ?? {}));
|
|
123
|
+
return;
|
|
124
|
+
}
|
|
125
|
+
}
|
|
126
|
+
catch (error) {
|
|
127
|
+
console.error('[FormFill] JSON parse failed for error output line, skipping:', error instanceof Error ? error.message : String(error));
|
|
128
|
+
/* not JSON, skip */
|
|
129
|
+
}
|
|
130
|
+
}
|
|
131
|
+
const detail = stderr ? `${err.message}\nPython stderr:\n${stderr}` : err.message;
|
|
132
|
+
reject(new OCRError(`Form fill worker failed: ${detail}`, 'FORM_FILL_API_ERROR'));
|
|
133
|
+
return;
|
|
134
|
+
}
|
|
135
|
+
// Parse the last JSON line from stdout (Python may output non-JSON logging)
|
|
136
|
+
const lines = output
|
|
137
|
+
.trim()
|
|
138
|
+
.split('\n')
|
|
139
|
+
.filter((l) => l.trim());
|
|
140
|
+
if (lines.length === 0) {
|
|
141
|
+
const diagnostics = [
|
|
142
|
+
'No output from form fill worker.',
|
|
143
|
+
`Exit code: ${exitCode}, signal: ${exitSignal}.`,
|
|
144
|
+
`File: ${filePath}.`,
|
|
145
|
+
];
|
|
146
|
+
if (stderr.trim()) {
|
|
147
|
+
diagnostics.push(`Python stderr: ${stderr.trim().substring(0, 500)}`);
|
|
148
|
+
}
|
|
149
|
+
reject(new OCRError(diagnostics.join(' '), 'FORM_FILL_API_ERROR'));
|
|
150
|
+
return;
|
|
151
|
+
}
|
|
152
|
+
let response;
|
|
153
|
+
for (let i = lines.length - 1; i >= 0; i--) {
|
|
154
|
+
try {
|
|
155
|
+
response = JSON.parse(lines[i].trim());
|
|
156
|
+
break;
|
|
157
|
+
}
|
|
158
|
+
catch (error) {
|
|
159
|
+
console.error('[FormFill] JSON parse failed for output line, trying previous:', error instanceof Error ? error.message : String(error));
|
|
160
|
+
/* not JSON, try previous line */
|
|
161
|
+
}
|
|
162
|
+
}
|
|
163
|
+
if (!response) {
|
|
164
|
+
reject(new OCRError(`Failed to parse form fill output as JSON. Last line: ${lines[lines.length - 1]?.substring(0, 200)}`, 'FORM_FILL_API_ERROR'));
|
|
165
|
+
return;
|
|
166
|
+
}
|
|
167
|
+
// Check for error response
|
|
168
|
+
if (this.isErrorResponse(response)) {
|
|
169
|
+
const resp = response;
|
|
170
|
+
const category = resp.category ?? 'FORM_FILL_API_ERROR';
|
|
171
|
+
const error = resp.error ?? 'Form fill failed';
|
|
172
|
+
reject(mapPythonError(category, error, resp.details ?? {}));
|
|
173
|
+
return;
|
|
174
|
+
}
|
|
175
|
+
// Validate required fields before casting
|
|
176
|
+
const resp = response;
|
|
177
|
+
if (typeof resp.id !== 'string' ||
|
|
178
|
+
typeof resp.source_file_path !== 'string' ||
|
|
179
|
+
typeof resp.status !== 'string') {
|
|
180
|
+
reject(new OCRError(`Form fill worker response missing required fields (id, source_file_path, status). Got keys: ${Object.keys(resp).join(', ')}`, 'FORM_FILL_API_ERROR'));
|
|
181
|
+
return;
|
|
182
|
+
}
|
|
183
|
+
promiseResolve(this.toFormFillResult(response));
|
|
184
|
+
});
|
|
185
|
+
});
|
|
186
|
+
}
|
|
187
|
+
isErrorResponse(response) {
|
|
188
|
+
if (typeof response !== 'object' || response === null)
|
|
189
|
+
return false;
|
|
190
|
+
// Standard error response with error + category (from Python worker error paths)
|
|
191
|
+
if ('error' in response && 'category' in response)
|
|
192
|
+
return true;
|
|
193
|
+
// NOTE: Do NOT treat status='failed' FormFillResult as an error — it is a valid
|
|
194
|
+
// partial-failure response containing fields_filled/fields_not_found data.
|
|
195
|
+
return false;
|
|
196
|
+
}
|
|
197
|
+
toFormFillResult(r) {
|
|
198
|
+
return {
|
|
199
|
+
id: r.id,
|
|
200
|
+
sourceFilePath: r.source_file_path,
|
|
201
|
+
sourceFileHash: r.source_file_hash,
|
|
202
|
+
outputBase64: r.output_base64,
|
|
203
|
+
fieldsFilled: r.fields_filled ?? [],
|
|
204
|
+
fieldsNotFound: r.fields_not_found ?? [],
|
|
205
|
+
pageCount: r.page_count,
|
|
206
|
+
costCents: r.cost_cents,
|
|
207
|
+
status: r.status,
|
|
208
|
+
error: r.error,
|
|
209
|
+
processingDurationMs: r.processing_duration_ms ?? 0,
|
|
210
|
+
};
|
|
211
|
+
}
|
|
212
|
+
}
|
|
213
|
+
//# sourceMappingURL=form-fill.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"form-fill.js","sourceRoot":"","sources":["../../../src/services/ocr/form-fill.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO,EAAE,WAAW,EAAW,MAAM,cAAc,CAAC;AACpD,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,MAAM,MAAM,CAAC;AACxC,OAAO,EAAE,aAAa,EAAE,MAAM,KAAK,CAAC;AACpC,OAAO,EAAE,QAAQ,EAAE,cAAc,EAAE,MAAM,aAAa,CAAC;AAEvD,MAAM,SAAS,GAAG,OAAO,CAAC,aAAa,CAAC,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;AAE1D,iEAAiE;AACjE,MAAM,iBAAiB,GAAG,MAAM,CAAC;AAgDjC,MAAM,OAAO,cAAc;IACR,UAAU,CAAqB;IAC/B,UAAU,CAAS;IACnB,OAAO,CAAS;IAEjC,YAAY,SAA+B,EAAE;QAC3C,IAAI,CAAC,UAAU,GAAG,MAAM,CAAC,UAAU,CAAC;QACpC,IAAI,CAAC,UAAU,GAAG,OAAO,CAAC,SAAS,EAAE,qCAAqC,CAAC,CAAC;QAC5E,MAAM,aAAa,GAAG,QAAQ,CAAC,OAAO,CAAC,GAAG,CAAC,eAAe,IAAI,QAAQ,CAAC,CAAC;QACxE,IAAI,CAAC,OAAO,GAAG,MAAM,CAAC,OAAO,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,aAAa,CAAC,CAAC;IAC1F,CAAC;IAED;;;;OAIG;IACH,KAAK,CAAC,QAAQ,CAAC,QAAgB,EAAE,OAAwB;QACvD,MAAM,IAAI,GAAG,CAAC,QAAQ,EAAE,QAAQ,EAAE,cAAc,EAAE,IAAI,CAAC,SAAS,CAAC,OAAO,CAAC,SAAS,CAAC,EAAE,QAAQ,CAAC,CAAC;QAC/F,IAAI,OAAO,CAAC,OAAO;YAAE,IAAI,CAAC,IAAI,CAAC,WAAW,EAAE,OAAO,CAAC,OAAO,CAAC,CAAC;QAC7D,IAAI,OAAO,CAAC,mBAAmB,KAAK,SAAS;YAC3C,IAAI,CAAC,IAAI,CAAC,wBAAwB,EAAE,MAAM,CAAC,OAAO,CAAC,mBAAmB,CAAC,CAAC,CAAC;QAC3E,IAAI,OAAO,CAAC,SAAS;YAAE,IAAI,CAAC,IAAI,CAAC,cAAc,EAAE,OAAO,CAAC,SAAS,CAAC,CAAC;QAEpE,MAAM,YAAY,GAAY;YAC5B,IAAI,EAAE,MAAM;YACZ,UAAU,EAAE,IAAI,CAAC,UAAU;YAC3B,aAAa,EAAE,CAAC,IAAI,CAAC;YACrB,IAAI;SACL,CAAC;QAEF,OAAO,IAAI,OAAO,CAAC,CAAC,cAAc,EAAE,MAAM,EAAE,EAAE;YAC5C,IAAI,OAAO,GAAG,KAAK,CAAC;YACpB,MAAM,KAAK,GAAG,IAAI,WAAW,CAAC,IAAI,CAAC,UAAU,EAAE,YAAY,CAAC,CAAC;YAC7D,MAAM,YAAY,GAAa,EAAE,CAAC;YAClC,IAAI,MAAM,GAAG,EAAE,CAAC;YAChB,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,MAAM,KAAK,GAAG,UAAU,CAAC,GAAG,EAAE;gBAC5B,IAAI,OAAO;oBAAE,OAAO;gBACpB,IAAI,CAAC;oBACH,KAAK,CAAC,IAAI,EAAE,CAAC;gBACf,CAAC;gBAAC,OAAO,KAAK,EAAE,CAAC;oBACf,OAAO,CAAC,KAAK,CACX,6CAA6C,EAC7C,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CACvD,CAAC;oBACF,YAAY;gBACd,CAAC;gBACD,4DAA4D;gBAC5D,YAAY,GAAG,UAAU,CAAC,GAAG,EAAE;oBAC7B,IAAI,CAAC,OAAO,EAAE,CAAC;wBACb,OAAO,CAAC,KAAK,CACX,wEAAwE,KAAK,CAAC,YAAY,EAAE,GAAG,GAAG,CACnG,CAAC;wBACF,IAAI,CAAC;4BACH,KAAK,CAAC,YAAY,EAAE,IAAI,CAAC,SAAS,CAAC,CAAC;wBACtC,CAAC;wBAAC,OAAO,KAAK,EAAE,CAAC;4BACf,OAAO,CAAC,KAAK,CACX,6DAA6D,EAC7D,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CACvD,CAAC;wBACJ,CAAC;oBACH,CAAC;oBACD,IAAI,CAAC,OAAO,EAAE,CAAC;wBACb,OAAO,GAAG,IAAI,CAAC;wBACf,MAAM,CAAC,IAAI,QAAQ,CAAC,2BAA2B,IAAI,CAAC,OAAO,6BAA6B,EAAE,mBAAmB,CAAC,CAAC,CAAC;oBAClH,CAAC;gBACH,CAAC,EAAE,IAAI,CAAC,CAAC;YACX,CAAC,EAAE,IAAI,CAAC,OAAO,CAAC,CAAC;YAEjB,KAAK,CAAC,EAAE,CAAC,SAAS,EAAE,CAAC,GAAW,EAAE,EAAE;gBAClC,YAAY,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;YACzB,CAAC,CAAC,CAAC;YAEH,KAAK,CAAC,EAAE,CAAC,QAAQ,EAAE,CAAC,GAAW,EAAE,EAAE;gBACjC,IAAI,MAAM,CAAC,MAAM,GAAG,iBAAiB,EAAE,CAAC;oBACtC,MAAM,IAAI,GAAG,GAAG,IAAI,CAAC;gBACvB,CAAC;YACH,CAAC,CAAC,CAAC;YAEH,KAAK,CAAC,GAAG,CAAC,CAAC,GAAW,EAAE,EAAE;gBACxB,YAAY,CAAC,KAAK,CAAC,CAAC;gBACpB,OAAO,EAAE,CAAC;gBACV,IAAI,OAAO;oBAAE,OAAO;gBACpB,OAAO,GAAG,IAAI,CAAC;gBAEf,2CAA2C;gBAC3C,MAAM,QAAQ,GAAG,KAAK,CAAC,QAAQ,IAAI,IAAI,CAAC;gBACxC,MAAM,UAAU,GAAG,KAAK,CAAC,UAAU,IAAI,IAAI,CAAC;gBAE5C,4DAA4D;gBAC5D,MAAM,MAAM,GAAG,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;gBACvC,gCAAgC;gBAChC,YAAY,CAAC,MAAM,GAAG,CAAC,CAAC;gBAExB,wEAAwE;gBACxE,mEAAmE;gBACnE,IAAI,CAAC,GAAG,IAAI,UAAU,EAAE,CAAC;oBACvB,MAAM,YAAY,GAAG,MAAM,CAAC,IAAI,EAAE;wBAChC,CAAC,CAAC,qBAAqB,UAAU,qBAAqB,MAAM,CAAC,IAAI,EAAE,EAAE;wBACrE,CAAC,CAAC,qBAAqB,UAAU,yBAAyB,CAAC;oBAC7D,MAAM,CACJ,IAAI,QAAQ,CACV,qCAAqC,UAAU,WAAW,QAAQ,MAAM,YAAY,EAAE,EACtF,UAAU,KAAK,SAAS,IAAI,UAAU,KAAK,SAAS,CAAC,CAAC,CAAC,mBAAmB,CAAC,CAAC,CAAC,qBAAqB,CACnG,CACF,CAAC;oBACF,OAAO;gBACT,CAAC;gBAED,IAAI,GAAG,EAAE,CAAC;oBACR,qDAAqD;oBACrD,MAAM,KAAK,GAAG,MAAM;yBACjB,IAAI,EAAE;yBACN,KAAK,CAAC,IAAI,CAAC;yBACX,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;oBAC3B,KAAK,IAAI,CAAC,GAAG,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;wBAC3C,IAAI,CAAC;4BACH,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAY,CAAC;4BACtD,IAAI,IAAI,CAAC,eAAe,CAAC,MAAM,CAAC,EAAE,CAAC;gCACjC,MAAM,CAAC,cAAc,CAAC,MAAM,CAAC,QAAQ,EAAE,MAAM,CAAC,KAAK,EAAE,MAAM,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,CAAC;gCAC5E,OAAO;4BACT,CAAC;wBACH,CAAC;wBAAC,OAAO,KAAK,EAAE,CAAC;4BACf,OAAO,CAAC,KAAK,CACX,+DAA+D,EAC/D,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CACvD,CAAC;4BACF,oBAAoB;wBACtB,CAAC;oBACH,CAAC;oBACD,MAAM,MAAM,GAAG,MAAM,CAAC,CAAC,CAAC,GAAG,GAAG,CAAC,OAAO,qBAAqB,MAAM,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC;oBAClF,MAAM,CAAC,IAAI,QAAQ,CAAC,4BAA4B,MAAM,EAAE,EAAE,qBAAqB,CAAC,CAAC,CAAC;oBAClF,OAAO;gBACT,CAAC;gBAED,4EAA4E;gBAC5E,MAAM,KAAK,GAAG,MAAM;qBACjB,IAAI,EAAE;qBACN,KAAK,CAAC,IAAI,CAAC;qBACX,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;gBAC3B,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;oBACvB,MAAM,WAAW,GAAG;wBAClB,kCAAkC;wBAClC,cAAc,QAAQ,aAAa,UAAU,GAAG;wBAChD,SAAS,QAAQ,GAAG;qBACrB,CAAC;oBACF,IAAI,MAAM,CAAC,IAAI,EAAE,EAAE,CAAC;wBAClB,WAAW,CAAC,IAAI,CAAC,kBAAkB,MAAM,CAAC,IAAI,EAAE,CAAC,SAAS,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,CAAC,CAAC;oBACxE,CAAC;oBACD,MAAM,CAAC,IAAI,QAAQ,CAAC,WAAW,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,qBAAqB,CAAC,CAAC,CAAC;oBACnE,OAAO;gBACT,CAAC;gBAED,IAAI,QAAiB,CAAC;gBACtB,KAAK,IAAI,CAAC,GAAG,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;oBAC3C,IAAI,CAAC;wBACH,QAAQ,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;wBACvC,MAAM;oBACR,CAAC;oBAAC,OAAO,KAAK,EAAE,CAAC;wBACf,OAAO,CAAC,KAAK,CACX,gEAAgE,EAChE,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CACvD,CAAC;wBACF,iCAAiC;oBACnC,CAAC;gBACH,CAAC;gBAED,IAAI,CAAC,QAAQ,EAAE,CAAC;oBACd,MAAM,CACJ,IAAI,QAAQ,CACV,wDAAwD,KAAK,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,EACpG,qBAAqB,CACtB,CACF,CAAC;oBACF,OAAO;gBACT,CAAC;gBAED,2BAA2B;gBAC3B,IAAI,IAAI,CAAC,eAAe,CAAC,QAAQ,CAAC,EAAE,CAAC;oBACnC,MAAM,IAAI,GAAG,QAA8C,CAAC;oBAC5D,MAAM,QAAQ,GAAI,IAAI,CAAC,QAAmB,IAAI,qBAAqB,CAAC;oBACpE,MAAM,KAAK,GAAI,IAAI,CAAC,KAAgB,IAAI,kBAAkB,CAAC;oBAC3D,MAAM,CAAC,cAAc,CAAC,QAAQ,EAAE,KAAK,EAAG,IAAI,CAAC,OAAmC,IAAI,EAAE,CAAC,CAAC,CAAC;oBACzF,OAAO;gBACT,CAAC;gBAED,0CAA0C;gBAC1C,MAAM,IAAI,GAAG,QAAmC,CAAC;gBACjD,IACE,OAAO,IAAI,CAAC,EAAE,KAAK,QAAQ;oBAC3B,OAAO,IAAI,CAAC,gBAAgB,KAAK,QAAQ;oBACzC,OAAO,IAAI,CAAC,MAAM,KAAK,QAAQ,EAC/B,CAAC;oBACD,MAAM,CACJ,IAAI,QAAQ,CACV,+FAA+F,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,EAC7H,qBAAqB,CACtB,CACF,CAAC;oBACF,OAAO;gBACT,CAAC;gBACD,cAAc,CAAC,IAAI,CAAC,gBAAgB,CAAC,QAAkC,CAAC,CAAC,CAAC;YAC5E,CAAC,CAAC,CAAC;QACL,CAAC,CAAC,CAAC;IACL,CAAC;IAEO,eAAe,CAAC,QAAiB;QACvC,IAAI,OAAO,QAAQ,KAAK,QAAQ,IAAI,QAAQ,KAAK,IAAI;YAAE,OAAO,KAAK,CAAC;QACpE,iFAAiF;QACjF,IAAI,OAAO,IAAI,QAAQ,IAAI,UAAU,IAAI,QAAQ;YAAE,OAAO,IAAI,CAAC;QAC/D,gFAAgF;QAChF,2EAA2E;QAC3E,OAAO,KAAK,CAAC;IACf,CAAC;IAEO,gBAAgB,CAAC,CAAyB;QAChD,OAAO;YACL,EAAE,EAAE,CAAC,CAAC,EAAE;YACR,cAAc,EAAE,CAAC,CAAC,gBAAgB;YAClC,cAAc,EAAE,CAAC,CAAC,gBAAgB;YAClC,YAAY,EAAE,CAAC,CAAC,aAAa;YAC7B,YAAY,EAAE,CAAC,CAAC,aAAa,IAAI,EAAE;YACnC,cAAc,EAAE,CAAC,CAAC,gBAAgB,IAAI,EAAE;YACxC,SAAS,EAAE,CAAC,CAAC,UAAU;YACvB,SAAS,EAAE,CAAC,CAAC,UAAU;YACvB,MAAM,EAAE,CAAC,CAAC,MAA+B;YACzC,KAAK,EAAE,CAAC,CAAC,KAAK;YACd,oBAAoB,EAAE,CAAC,CAAC,sBAAsB,IAAI,CAAC;SACpD,CAAC;IACJ,CAAC;CACF"}
|
|
@@ -0,0 +1,95 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* OCR Processing Orchestrator
|
|
3
|
+
*
|
|
4
|
+
* Complete pipeline: Document -> OCR -> Provenance -> Store -> Status Update
|
|
5
|
+
* FAIL-FAST: No fallbacks, errors propagate immediately
|
|
6
|
+
*/
|
|
7
|
+
import { type DatalabClientConfig } from './datalab.js';
|
|
8
|
+
import { DatabaseService } from '../storage/database/index.js';
|
|
9
|
+
import type { PageOffset } from '../../models/document.js';
|
|
10
|
+
interface ProcessorConfig extends DatalabClientConfig {
|
|
11
|
+
maxConcurrent?: number;
|
|
12
|
+
defaultMode?: 'fast' | 'balanced' | 'accurate';
|
|
13
|
+
}
|
|
14
|
+
export interface ProcessResult {
|
|
15
|
+
success: boolean;
|
|
16
|
+
documentId: string;
|
|
17
|
+
ocrResultId?: string;
|
|
18
|
+
provenanceId?: string;
|
|
19
|
+
pageCount?: number;
|
|
20
|
+
textLength?: number;
|
|
21
|
+
durationMs?: number;
|
|
22
|
+
error?: string;
|
|
23
|
+
/** Images extracted by Datalab: {filename: base64_data} */
|
|
24
|
+
images?: Record<string, string>;
|
|
25
|
+
/** JSON block hierarchy from Datalab (when output_format includes 'json') */
|
|
26
|
+
jsonBlocks?: Record<string, unknown> | null;
|
|
27
|
+
/** Datalab metadata (page_stats, block_counts, etc.) */
|
|
28
|
+
metadata?: Record<string, unknown> | null;
|
|
29
|
+
/** Page character offsets from Datalab for page-aware chunking */
|
|
30
|
+
pageOffsets?: PageOffset[];
|
|
31
|
+
/** Structured extraction result from page_schema */
|
|
32
|
+
extractionJson?: Record<string, unknown> | unknown[] | null;
|
|
33
|
+
/** Document title from metadata */
|
|
34
|
+
docTitle?: string | null;
|
|
35
|
+
/** Document author from metadata */
|
|
36
|
+
docAuthor?: string | null;
|
|
37
|
+
/** Document subject from metadata */
|
|
38
|
+
docSubject?: string | null;
|
|
39
|
+
}
|
|
40
|
+
export interface BatchResult {
|
|
41
|
+
processed: number;
|
|
42
|
+
failed: number;
|
|
43
|
+
remaining: number;
|
|
44
|
+
totalDurationMs: number;
|
|
45
|
+
results: ProcessResult[];
|
|
46
|
+
}
|
|
47
|
+
export declare class OCRProcessor {
|
|
48
|
+
private readonly client;
|
|
49
|
+
private readonly db;
|
|
50
|
+
private readonly maxConcurrent;
|
|
51
|
+
private readonly defaultMode;
|
|
52
|
+
constructor(db: DatabaseService, config?: ProcessorConfig);
|
|
53
|
+
/**
|
|
54
|
+
* Process single document through OCR
|
|
55
|
+
*
|
|
56
|
+
* Pipeline:
|
|
57
|
+
* 1. Get document from database (FAIL if not found)
|
|
58
|
+
* 2. Update status to 'processing'
|
|
59
|
+
* 3. Call Datalab OCR via Python worker
|
|
60
|
+
* 4. Create OCR_RESULT provenance record
|
|
61
|
+
* 5. Store OCR result in database
|
|
62
|
+
* 6. Update document status to 'complete'
|
|
63
|
+
*
|
|
64
|
+
* On failure: Update status to 'failed' with error message
|
|
65
|
+
*/
|
|
66
|
+
processDocument(documentId: string, mode?: 'fast' | 'balanced' | 'accurate', ocrOptions?: {
|
|
67
|
+
maxPages?: number;
|
|
68
|
+
pageRange?: string;
|
|
69
|
+
skipCache?: boolean;
|
|
70
|
+
disableImageExtraction?: boolean;
|
|
71
|
+
extras?: string[];
|
|
72
|
+
pageSchema?: string;
|
|
73
|
+
additionalConfig?: Record<string, unknown>;
|
|
74
|
+
}): Promise<ProcessResult>;
|
|
75
|
+
/**
|
|
76
|
+
* Process all pending documents.
|
|
77
|
+
*
|
|
78
|
+
* H-2 fix: Before processing, recover any documents stuck in 'processing'
|
|
79
|
+
* status for longer than 30 minutes (indicates a prior server crash).
|
|
80
|
+
* These are reset to 'pending' so they get re-processed.
|
|
81
|
+
*/
|
|
82
|
+
processPending(mode?: 'fast' | 'balanced' | 'accurate'): Promise<BatchResult>;
|
|
83
|
+
/**
|
|
84
|
+
* H-2: Recover documents stuck in 'processing' status after a server crash.
|
|
85
|
+
* Any document that has been 'processing' for longer than 30 minutes is
|
|
86
|
+
* assumed to be orphaned from a crash and is reset to 'pending'.
|
|
87
|
+
*/
|
|
88
|
+
private recoverStaleProcessingDocuments;
|
|
89
|
+
/**
|
|
90
|
+
* Create OCR_RESULT provenance record
|
|
91
|
+
*/
|
|
92
|
+
private createOCRProvenance;
|
|
93
|
+
}
|
|
94
|
+
export {};
|
|
95
|
+
//# sourceMappingURL=processor.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"processor.d.ts","sourceRoot":"","sources":["../../../src/services/ocr/processor.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAGH,OAAO,EAAiB,KAAK,mBAAmB,EAAE,MAAM,cAAc,CAAC;AAGvE,OAAO,EAAE,eAAe,EAAE,MAAM,8BAA8B,CAAC;AAC/D,OAAO,KAAK,EAAuB,UAAU,EAAE,MAAM,0BAA0B,CAAC;AAGhF,UAAU,eAAgB,SAAQ,mBAAmB;IACnD,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB,WAAW,CAAC,EAAE,MAAM,GAAG,UAAU,GAAG,UAAU,CAAC;CAChD;AAED,MAAM,WAAW,aAAa;IAC5B,OAAO,EAAE,OAAO,CAAC;IACjB,UAAU,EAAE,MAAM,CAAC;IACnB,WAAW,CAAC,EAAE,MAAM,CAAC;IACrB,YAAY,CAAC,EAAE,MAAM,CAAC;IACtB,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,2DAA2D;IAC3D,MAAM,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;IAChC,6EAA6E;IAC7E,UAAU,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,IAAI,CAAC;IAC5C,wDAAwD;IACxD,QAAQ,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,IAAI,CAAC;IAC1C,kEAAkE;IAClE,WAAW,CAAC,EAAE,UAAU,EAAE,CAAC;IAC3B,oDAAoD;IACpD,cAAc,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,OAAO,EAAE,GAAG,IAAI,CAAC;IAC5D,mCAAmC;IACnC,QAAQ,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;IACzB,oCAAoC;IACpC,SAAS,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;IAC1B,qCAAqC;IACrC,UAAU,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;CAC5B;AAED,MAAM,WAAW,WAAW;IAC1B,SAAS,EAAE,MAAM,CAAC;IAClB,MAAM,EAAE,MAAM,CAAC;IACf,SAAS,EAAE,MAAM,CAAC;IAClB,eAAe,EAAE,MAAM,CAAC;IACxB,OAAO,EAAE,aAAa,EAAE,CAAC;CAC1B;AAiBD,qBAAa,YAAY;IACvB,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAgB;IACvC,OAAO,CAAC,QAAQ,CAAC,EAAE,CAAkB;IACrC,OAAO,CAAC,QAAQ,CAAC,aAAa,CAAS;IACvC,OAAO,CAAC,QAAQ,CAAC,WAAW,CAAmC;gBAEnD,EAAE,EAAE,eAAe,EAAE,MAAM,GAAE,eAAoB;IAO7D;;;;;;;;;;;;OAYG;IACG,eAAe,CACnB,UAAU,EAAE,MAAM,EAClB,IAAI,CAAC,EAAE,MAAM,GAAG,UAAU,GAAG,UAAU,EACvC,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;KAC5C,GACA,OAAO,CAAC,aAAa,CAAC;IA2HzB;;;;;;OAMG;IACG,cAAc,CAAC,IAAI,CAAC,EAAE,MAAM,GAAG,UAAU,GAAG,UAAU,GAAG,OAAO,CAAC,WAAW,CAAC;IAwDnF;;;;OAIG;IACH,OAAO,CAAC,+BAA+B;IAuBvC;;OAEG;IACH,OAAO,CAAC,mBAAmB;CAuC5B"}
|