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 +1,425 @@
|
|
|
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 { 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 nomic.ts pattern) */
|
|
13
|
+
const MAX_STDERR_LENGTH = 10_240;
|
|
14
|
+
export class DatalabClient {
|
|
15
|
+
pythonPath;
|
|
16
|
+
workerPath;
|
|
17
|
+
timeout;
|
|
18
|
+
constructor(config = {}) {
|
|
19
|
+
this.pythonPath = config.pythonPath;
|
|
20
|
+
this.workerPath = resolve(__dirname, '../../../python/ocr_worker.py');
|
|
21
|
+
const parsedTimeout = parseInt(process.env.DATALAB_TIMEOUT || '330000');
|
|
22
|
+
this.timeout = config.timeout ?? (Number.isNaN(parsedTimeout) ? 330000 : parsedTimeout); // 5.5 min (FIX-P3-1)
|
|
23
|
+
}
|
|
24
|
+
/**
|
|
25
|
+
* Process document through Datalab OCR
|
|
26
|
+
*
|
|
27
|
+
* FAIL-FAST: Throws on any error, no fallbacks
|
|
28
|
+
*/
|
|
29
|
+
async processDocument(filePath, documentId, provenanceId, mode = 'accurate', ocrOptions) {
|
|
30
|
+
const args = [
|
|
31
|
+
'--file',
|
|
32
|
+
filePath,
|
|
33
|
+
'--mode',
|
|
34
|
+
mode,
|
|
35
|
+
'--doc-id',
|
|
36
|
+
documentId,
|
|
37
|
+
'--prov-id',
|
|
38
|
+
provenanceId,
|
|
39
|
+
'--json',
|
|
40
|
+
];
|
|
41
|
+
if (ocrOptions?.maxPages)
|
|
42
|
+
args.push('--max-pages', String(ocrOptions.maxPages));
|
|
43
|
+
if (ocrOptions?.pageRange)
|
|
44
|
+
args.push('--page-range', ocrOptions.pageRange);
|
|
45
|
+
if (ocrOptions?.skipCache)
|
|
46
|
+
args.push('--skip-cache');
|
|
47
|
+
if (ocrOptions?.disableImageExtraction)
|
|
48
|
+
args.push('--disable-image-extraction');
|
|
49
|
+
if (ocrOptions?.extras?.length)
|
|
50
|
+
args.push('--extras', ocrOptions.extras.join(','));
|
|
51
|
+
if (ocrOptions?.pageSchema)
|
|
52
|
+
args.push('--page-schema', ocrOptions.pageSchema);
|
|
53
|
+
if (ocrOptions?.additionalConfig)
|
|
54
|
+
args.push('--additional-config', JSON.stringify(ocrOptions.additionalConfig));
|
|
55
|
+
if (ocrOptions?.fileUrl)
|
|
56
|
+
args.push('--file-url', ocrOptions.fileUrl);
|
|
57
|
+
const options = {
|
|
58
|
+
mode: 'text',
|
|
59
|
+
pythonPath: this.pythonPath,
|
|
60
|
+
pythonOptions: ['-u'],
|
|
61
|
+
args,
|
|
62
|
+
};
|
|
63
|
+
return new Promise((resolve, reject) => {
|
|
64
|
+
let settled = false;
|
|
65
|
+
const shell = new PythonShell(this.workerPath, options);
|
|
66
|
+
const outputChunks = [];
|
|
67
|
+
let stderr = '';
|
|
68
|
+
let sigkillTimer = null;
|
|
69
|
+
const cleanup = () => {
|
|
70
|
+
if (sigkillTimer) {
|
|
71
|
+
clearTimeout(sigkillTimer);
|
|
72
|
+
sigkillTimer = null;
|
|
73
|
+
}
|
|
74
|
+
};
|
|
75
|
+
const timer = setTimeout(() => {
|
|
76
|
+
if (settled)
|
|
77
|
+
return;
|
|
78
|
+
// Kill the Python process to prevent orphans (SIGTERM)
|
|
79
|
+
try {
|
|
80
|
+
shell.kill();
|
|
81
|
+
}
|
|
82
|
+
catch (error) {
|
|
83
|
+
console.error('[DatalabOCR] Failed to kill shell on timeout:', error instanceof Error ? error.message : String(error));
|
|
84
|
+
/* ignore */
|
|
85
|
+
}
|
|
86
|
+
// M-6: SIGKILL escalation if SIGTERM doesn't exit within 5s
|
|
87
|
+
sigkillTimer = setTimeout(() => {
|
|
88
|
+
if (!settled) {
|
|
89
|
+
console.error(`[DatalabOCR] Process did not exit after SIGTERM, sending SIGKILL (pid: ${shell.childProcess?.pid})`);
|
|
90
|
+
try {
|
|
91
|
+
shell.childProcess?.kill('SIGKILL');
|
|
92
|
+
}
|
|
93
|
+
catch (error) {
|
|
94
|
+
console.error('[DatalabOCR] Failed to SIGKILL process (may already be gone):', error instanceof Error ? error.message : String(error));
|
|
95
|
+
}
|
|
96
|
+
}
|
|
97
|
+
// Settle the promise if close event hasn't fired yet (zombie prevention)
|
|
98
|
+
if (!settled) {
|
|
99
|
+
settled = true;
|
|
100
|
+
reject(new OCRError(`OCR timeout after ${this.timeout}ms (SIGKILL after 5s grace)`, 'OCR_TIMEOUT'));
|
|
101
|
+
}
|
|
102
|
+
}, 5000);
|
|
103
|
+
}, this.timeout);
|
|
104
|
+
shell.on('message', (msg) => {
|
|
105
|
+
outputChunks.push(msg);
|
|
106
|
+
});
|
|
107
|
+
shell.on('stderr', (err) => {
|
|
108
|
+
if (stderr.length < MAX_STDERR_LENGTH) {
|
|
109
|
+
stderr += err + '\n';
|
|
110
|
+
}
|
|
111
|
+
});
|
|
112
|
+
shell.end((err) => {
|
|
113
|
+
clearTimeout(timer);
|
|
114
|
+
cleanup();
|
|
115
|
+
if (settled)
|
|
116
|
+
return;
|
|
117
|
+
settled = true;
|
|
118
|
+
// Capture exit code/signal for diagnostics
|
|
119
|
+
const exitCode = shell.exitCode ?? null;
|
|
120
|
+
const exitSignal = shell.exitSignal ?? null;
|
|
121
|
+
// H-3: Join chunks once instead of repeated string concatenation
|
|
122
|
+
const output = outputChunks.join('\n');
|
|
123
|
+
// M-12: Allow early GC of chunk array
|
|
124
|
+
outputChunks.length = 0;
|
|
125
|
+
// python-shell bug: when process is killed by signal (SIGTERM/SIGKILL),
|
|
126
|
+
// exitCode is null and exitSignal is set, but err is NOT provided.
|
|
127
|
+
// Detect this and treat signal-killed processes as errors.
|
|
128
|
+
if (!err && exitSignal) {
|
|
129
|
+
const signalDetail = stderr.trim()
|
|
130
|
+
? `Process killed by ${exitSignal}. Python stderr:\n${stderr.trim()}`
|
|
131
|
+
: `Process killed by ${exitSignal} with no stderr output.`;
|
|
132
|
+
reject(new OCRError(`OCR worker killed by signal ${exitSignal} (file: ${filePath}). ${signalDetail}`, exitSignal === 'SIGTERM' || exitSignal === 'SIGALRM' ? 'OCR_TIMEOUT' : 'OCR_API_ERROR'));
|
|
133
|
+
return;
|
|
134
|
+
}
|
|
135
|
+
if (err) {
|
|
136
|
+
// Try to parse JSON from output for structured error
|
|
137
|
+
const lines = output
|
|
138
|
+
.trim()
|
|
139
|
+
.split('\n')
|
|
140
|
+
.filter((l) => l.trim());
|
|
141
|
+
for (let i = lines.length - 1; i >= 0; i--) {
|
|
142
|
+
try {
|
|
143
|
+
const parsed = JSON.parse(lines[i].trim());
|
|
144
|
+
if (this.isErrorResponse(parsed)) {
|
|
145
|
+
reject(mapPythonError(parsed.category, parsed.error, parsed.details));
|
|
146
|
+
return;
|
|
147
|
+
}
|
|
148
|
+
}
|
|
149
|
+
catch (error) {
|
|
150
|
+
console.error('[DatalabOCR] JSON parse failed for error output line, skipping:', error instanceof Error ? error.message : String(error));
|
|
151
|
+
/* not JSON, skip */
|
|
152
|
+
}
|
|
153
|
+
}
|
|
154
|
+
const detail = stderr ? `${err.message}\nPython stderr:\n${stderr}` : err.message;
|
|
155
|
+
reject(new OCRError(`Python worker failed: ${detail}`, 'OCR_API_ERROR'));
|
|
156
|
+
return;
|
|
157
|
+
}
|
|
158
|
+
// Parse the last JSON line from stdout (Python may output non-JSON logging)
|
|
159
|
+
const lines = output
|
|
160
|
+
.trim()
|
|
161
|
+
.split('\n')
|
|
162
|
+
.filter((l) => l.trim());
|
|
163
|
+
if (lines.length === 0) {
|
|
164
|
+
const diagnostics = [
|
|
165
|
+
'No output from OCR worker.',
|
|
166
|
+
`Exit code: ${exitCode}, signal: ${exitSignal}.`,
|
|
167
|
+
`File: ${filePath}.`,
|
|
168
|
+
];
|
|
169
|
+
if (stderr.trim()) {
|
|
170
|
+
diagnostics.push(`Python stderr: ${stderr.trim().substring(0, 500)}`);
|
|
171
|
+
}
|
|
172
|
+
else {
|
|
173
|
+
diagnostics.push('Python stderr was empty (logging suppressed in --json mode).');
|
|
174
|
+
}
|
|
175
|
+
reject(new OCRError(diagnostics.join(' '), 'OCR_API_ERROR'));
|
|
176
|
+
return;
|
|
177
|
+
}
|
|
178
|
+
let response;
|
|
179
|
+
for (let i = lines.length - 1; i >= 0; i--) {
|
|
180
|
+
try {
|
|
181
|
+
response = JSON.parse(lines[i].trim());
|
|
182
|
+
break;
|
|
183
|
+
}
|
|
184
|
+
catch (error) {
|
|
185
|
+
console.error('[DatalabOCR] JSON parse failed for output line, trying previous:', error instanceof Error ? error.message : String(error));
|
|
186
|
+
/* not JSON, try previous line */
|
|
187
|
+
}
|
|
188
|
+
}
|
|
189
|
+
if (!response) {
|
|
190
|
+
reject(new OCRError(`Failed to parse OCR worker output as JSON. Last line: ${lines[lines.length - 1]?.substring(0, 200)}`, 'OCR_API_ERROR'));
|
|
191
|
+
return;
|
|
192
|
+
}
|
|
193
|
+
// Check for error response
|
|
194
|
+
if (this.isErrorResponse(response)) {
|
|
195
|
+
reject(mapPythonError(response.category, response.error, response.details));
|
|
196
|
+
return;
|
|
197
|
+
}
|
|
198
|
+
const ocrResponse = response;
|
|
199
|
+
// Verify required fields exist
|
|
200
|
+
if (!ocrResponse.id || !ocrResponse.content_hash || !ocrResponse.extracted_text) {
|
|
201
|
+
reject(new OCRError(`Invalid OCR response: missing required fields. Got: ${JSON.stringify(Object.keys(ocrResponse))}`, 'OCR_API_ERROR'));
|
|
202
|
+
return;
|
|
203
|
+
}
|
|
204
|
+
resolve({
|
|
205
|
+
result: this.toOCRResult(ocrResponse),
|
|
206
|
+
pageOffsets: this.toPageOffsets(ocrResponse.page_offsets),
|
|
207
|
+
images: ocrResponse.images ?? {},
|
|
208
|
+
jsonBlocks: ocrResponse.json_blocks ?? null,
|
|
209
|
+
metadata: ocrResponse.metadata ?? null,
|
|
210
|
+
extractionJson: ocrResponse.extraction_json ?? null,
|
|
211
|
+
docTitle: ocrResponse.doc_title ?? null,
|
|
212
|
+
docAuthor: ocrResponse.doc_author ?? null,
|
|
213
|
+
docSubject: ocrResponse.doc_subject ?? null,
|
|
214
|
+
});
|
|
215
|
+
});
|
|
216
|
+
});
|
|
217
|
+
}
|
|
218
|
+
/**
|
|
219
|
+
* Process a file through Datalab OCR and return raw results without storing in DB.
|
|
220
|
+
* Used by ocr_convert_raw tool for quick one-off conversions.
|
|
221
|
+
*
|
|
222
|
+
* FAIL-FAST: Throws on any error, no fallbacks
|
|
223
|
+
*/
|
|
224
|
+
async processRaw(filePath, mode = 'balanced', ocrOptions) {
|
|
225
|
+
// Use dummy IDs since we are not storing in DB
|
|
226
|
+
const dummyDocId = 'raw-convert-' + Date.now();
|
|
227
|
+
const dummyProvId = 'raw-convert-prov-' + Date.now();
|
|
228
|
+
const args = [];
|
|
229
|
+
if (ocrOptions?.fileUrl) {
|
|
230
|
+
args.push('--file-url', ocrOptions.fileUrl);
|
|
231
|
+
}
|
|
232
|
+
else {
|
|
233
|
+
args.push('--file', filePath);
|
|
234
|
+
}
|
|
235
|
+
args.push('--mode', mode, '--doc-id', dummyDocId, '--prov-id', dummyProvId, '--json');
|
|
236
|
+
if (ocrOptions?.maxPages)
|
|
237
|
+
args.push('--max-pages', String(ocrOptions.maxPages));
|
|
238
|
+
if (ocrOptions?.pageRange)
|
|
239
|
+
args.push('--page-range', ocrOptions.pageRange);
|
|
240
|
+
const options = {
|
|
241
|
+
mode: 'text',
|
|
242
|
+
pythonPath: this.pythonPath,
|
|
243
|
+
pythonOptions: ['-u'],
|
|
244
|
+
args,
|
|
245
|
+
};
|
|
246
|
+
return new Promise((resolve, reject) => {
|
|
247
|
+
let settled = false;
|
|
248
|
+
const shell = new PythonShell(this.workerPath, options);
|
|
249
|
+
const outputChunks = [];
|
|
250
|
+
let stderr = '';
|
|
251
|
+
let sigkillTimer = null;
|
|
252
|
+
const cleanup = () => {
|
|
253
|
+
if (sigkillTimer) {
|
|
254
|
+
clearTimeout(sigkillTimer);
|
|
255
|
+
sigkillTimer = null;
|
|
256
|
+
}
|
|
257
|
+
};
|
|
258
|
+
const timer = setTimeout(() => {
|
|
259
|
+
if (settled)
|
|
260
|
+
return;
|
|
261
|
+
try {
|
|
262
|
+
shell.kill();
|
|
263
|
+
}
|
|
264
|
+
catch (error) {
|
|
265
|
+
console.error('[DatalabOCR] Failed to kill shell on processRaw timeout:', error instanceof Error ? error.message : String(error));
|
|
266
|
+
/* ignore */
|
|
267
|
+
}
|
|
268
|
+
// M-6: SIGKILL escalation if SIGTERM doesn't exit within 5s
|
|
269
|
+
sigkillTimer = setTimeout(() => {
|
|
270
|
+
if (!settled) {
|
|
271
|
+
console.error(`[DatalabOCR] processRaw process did not exit after SIGTERM, sending SIGKILL (pid: ${shell.childProcess?.pid})`);
|
|
272
|
+
try {
|
|
273
|
+
shell.childProcess?.kill('SIGKILL');
|
|
274
|
+
}
|
|
275
|
+
catch (error) {
|
|
276
|
+
console.error('[DatalabOCR] Failed to SIGKILL processRaw process (may already be gone):', error instanceof Error ? error.message : String(error));
|
|
277
|
+
}
|
|
278
|
+
}
|
|
279
|
+
if (!settled) {
|
|
280
|
+
settled = true;
|
|
281
|
+
reject(new OCRError(`OCR timeout after ${this.timeout}ms (SIGKILL after 5s grace)`, 'OCR_TIMEOUT'));
|
|
282
|
+
}
|
|
283
|
+
}, 5000);
|
|
284
|
+
}, this.timeout);
|
|
285
|
+
shell.on('message', (msg) => {
|
|
286
|
+
outputChunks.push(msg);
|
|
287
|
+
});
|
|
288
|
+
shell.on('stderr', (err) => {
|
|
289
|
+
if (stderr.length < MAX_STDERR_LENGTH) {
|
|
290
|
+
stderr += err + '\n';
|
|
291
|
+
}
|
|
292
|
+
});
|
|
293
|
+
shell.end((err) => {
|
|
294
|
+
clearTimeout(timer);
|
|
295
|
+
cleanup();
|
|
296
|
+
if (settled)
|
|
297
|
+
return;
|
|
298
|
+
settled = true;
|
|
299
|
+
// Capture exit code/signal for diagnostics
|
|
300
|
+
const exitCode = shell.exitCode ?? null;
|
|
301
|
+
const exitSignal = shell.exitSignal ?? null;
|
|
302
|
+
const output = outputChunks.join('\n');
|
|
303
|
+
outputChunks.length = 0;
|
|
304
|
+
// python-shell bug: when process is killed by signal (SIGTERM/SIGKILL),
|
|
305
|
+
// exitCode is null and exitSignal is set, but err is NOT provided.
|
|
306
|
+
// Detect this and treat signal-killed processes as errors.
|
|
307
|
+
const rawFilePath = ocrOptions?.fileUrl || filePath;
|
|
308
|
+
if (!err && exitSignal) {
|
|
309
|
+
const signalDetail = stderr.trim()
|
|
310
|
+
? `Process killed by ${exitSignal}. Python stderr:\n${stderr.trim()}`
|
|
311
|
+
: `Process killed by ${exitSignal} with no stderr output.`;
|
|
312
|
+
reject(new OCRError(`OCR worker killed by signal ${exitSignal} (file: ${rawFilePath}). ${signalDetail}`, exitSignal === 'SIGTERM' || exitSignal === 'SIGALRM' ? 'OCR_TIMEOUT' : 'OCR_API_ERROR'));
|
|
313
|
+
return;
|
|
314
|
+
}
|
|
315
|
+
if (err) {
|
|
316
|
+
const lines = output
|
|
317
|
+
.trim()
|
|
318
|
+
.split('\n')
|
|
319
|
+
.filter((l) => l.trim());
|
|
320
|
+
for (let i = lines.length - 1; i >= 0; i--) {
|
|
321
|
+
try {
|
|
322
|
+
const parsed = JSON.parse(lines[i].trim());
|
|
323
|
+
if (this.isErrorResponse(parsed)) {
|
|
324
|
+
reject(mapPythonError(parsed.category, parsed.error, parsed.details));
|
|
325
|
+
return;
|
|
326
|
+
}
|
|
327
|
+
}
|
|
328
|
+
catch (error) {
|
|
329
|
+
console.error('[DatalabOCR] JSON parse failed for processRaw error output line, skipping:', error instanceof Error ? error.message : String(error));
|
|
330
|
+
/* not JSON, skip */
|
|
331
|
+
}
|
|
332
|
+
}
|
|
333
|
+
const detail = stderr ? `${err.message}\nPython stderr:\n${stderr}` : err.message;
|
|
334
|
+
reject(new OCRError(`Python worker failed: ${detail}`, 'OCR_API_ERROR'));
|
|
335
|
+
return;
|
|
336
|
+
}
|
|
337
|
+
const lines = output
|
|
338
|
+
.trim()
|
|
339
|
+
.split('\n')
|
|
340
|
+
.filter((l) => l.trim());
|
|
341
|
+
if (lines.length === 0) {
|
|
342
|
+
const diagnostics = [
|
|
343
|
+
'No output from OCR worker.',
|
|
344
|
+
`Exit code: ${exitCode}, signal: ${exitSignal}.`,
|
|
345
|
+
`File: ${rawFilePath}.`,
|
|
346
|
+
];
|
|
347
|
+
if (stderr.trim()) {
|
|
348
|
+
diagnostics.push(`Python stderr: ${stderr.trim().substring(0, 500)}`);
|
|
349
|
+
}
|
|
350
|
+
else {
|
|
351
|
+
diagnostics.push('Python stderr was empty (logging suppressed in --json mode).');
|
|
352
|
+
}
|
|
353
|
+
reject(new OCRError(diagnostics.join(' '), 'OCR_API_ERROR'));
|
|
354
|
+
return;
|
|
355
|
+
}
|
|
356
|
+
let response;
|
|
357
|
+
for (let i = lines.length - 1; i >= 0; i--) {
|
|
358
|
+
try {
|
|
359
|
+
response = JSON.parse(lines[i].trim());
|
|
360
|
+
break;
|
|
361
|
+
}
|
|
362
|
+
catch (error) {
|
|
363
|
+
console.error('[DatalabOCR] JSON parse failed for processRaw output line, trying previous:', error instanceof Error ? error.message : String(error));
|
|
364
|
+
/* not JSON, try previous line */
|
|
365
|
+
}
|
|
366
|
+
}
|
|
367
|
+
if (!response) {
|
|
368
|
+
reject(new OCRError(`Failed to parse OCR worker output as JSON. Last line: ${lines[lines.length - 1]?.substring(0, 200)}`, 'OCR_API_ERROR'));
|
|
369
|
+
return;
|
|
370
|
+
}
|
|
371
|
+
if (this.isErrorResponse(response)) {
|
|
372
|
+
reject(mapPythonError(response.category, response.error, response.details));
|
|
373
|
+
return;
|
|
374
|
+
}
|
|
375
|
+
const ocrResponse = response;
|
|
376
|
+
resolve({
|
|
377
|
+
markdown: ocrResponse.extracted_text ?? '',
|
|
378
|
+
pageCount: ocrResponse.page_count ?? 0,
|
|
379
|
+
qualityScore: ocrResponse.parse_quality_score ?? null,
|
|
380
|
+
costCents: ocrResponse.cost_cents ?? null,
|
|
381
|
+
durationMs: ocrResponse.processing_duration_ms ?? 0,
|
|
382
|
+
metadata: ocrResponse.metadata ?? null,
|
|
383
|
+
});
|
|
384
|
+
});
|
|
385
|
+
});
|
|
386
|
+
}
|
|
387
|
+
isErrorResponse(response) {
|
|
388
|
+
return (typeof response === 'object' &&
|
|
389
|
+
response !== null &&
|
|
390
|
+
'error' in response &&
|
|
391
|
+
'category' in response);
|
|
392
|
+
}
|
|
393
|
+
toOCRResult(r) {
|
|
394
|
+
// Direct field mapping - Python snake_case matches TS interface
|
|
395
|
+
// Build extras_json from metadata + cost_breakdown
|
|
396
|
+
const extras = {};
|
|
397
|
+
if (r.metadata)
|
|
398
|
+
extras.metadata = r.metadata;
|
|
399
|
+
if (r.cost_breakdown_full)
|
|
400
|
+
extras.cost_breakdown = r.cost_breakdown_full;
|
|
401
|
+
return {
|
|
402
|
+
id: r.id,
|
|
403
|
+
provenance_id: r.provenance_id,
|
|
404
|
+
document_id: r.document_id,
|
|
405
|
+
extracted_text: r.extracted_text,
|
|
406
|
+
text_length: r.text_length,
|
|
407
|
+
datalab_request_id: r.datalab_request_id,
|
|
408
|
+
datalab_mode: r.datalab_mode,
|
|
409
|
+
parse_quality_score: r.parse_quality_score,
|
|
410
|
+
page_count: r.page_count,
|
|
411
|
+
cost_cents: r.cost_cents,
|
|
412
|
+
content_hash: r.content_hash,
|
|
413
|
+
processing_started_at: r.processing_started_at,
|
|
414
|
+
processing_completed_at: r.processing_completed_at,
|
|
415
|
+
processing_duration_ms: r.processing_duration_ms,
|
|
416
|
+
json_blocks: r.json_blocks ? JSON.stringify(r.json_blocks) : null,
|
|
417
|
+
extras_json: Object.keys(extras).length > 0 ? JSON.stringify(extras) : null,
|
|
418
|
+
};
|
|
419
|
+
}
|
|
420
|
+
toPageOffsets(offsets) {
|
|
421
|
+
// Convert Python snake_case to TS camelCase
|
|
422
|
+
return offsets.map((o) => ({ page: o.page, charStart: o.char_start, charEnd: o.char_end }));
|
|
423
|
+
}
|
|
424
|
+
}
|
|
425
|
+
//# sourceMappingURL=datalab.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"datalab.js","sourceRoot":"","sources":["../../../src/services/ocr/datalab.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;AAEpC,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,+DAA+D;AAC/D,MAAM,iBAAiB,GAAG,MAAM,CAAC;AAoDjC,MAAM,OAAO,aAAa;IACP,UAAU,CAAqB;IAC/B,UAAU,CAAS;IACnB,OAAO,CAAS;IAEjC,YAAY,SAA8B,EAAE;QAC1C,IAAI,CAAC,UAAU,GAAG,MAAM,CAAC,UAAU,CAAC;QACpC,IAAI,CAAC,UAAU,GAAG,OAAO,CAAC,SAAS,EAAE,+BAA+B,CAAC,CAAC;QACtE,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,CAAC,qBAAqB;IAChH,CAAC;IAED;;;;OAIG;IACH,KAAK,CAAC,eAAe,CACnB,QAAgB,EAChB,UAAkB,EAClB,YAAoB,EACpB,OAAyC,UAAU,EACnD,UASC;QAYD,MAAM,IAAI,GAAG;YACX,QAAQ;YACR,QAAQ;YACR,QAAQ;YACR,IAAI;YACJ,UAAU;YACV,UAAU;YACV,WAAW;YACX,YAAY;YACZ,QAAQ;SACT,CAAC;QACF,IAAI,UAAU,EAAE,QAAQ;YAAE,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,MAAM,CAAC,UAAU,CAAC,QAAQ,CAAC,CAAC,CAAC;QAChF,IAAI,UAAU,EAAE,SAAS;YAAE,IAAI,CAAC,IAAI,CAAC,cAAc,EAAE,UAAU,CAAC,SAAS,CAAC,CAAC;QAC3E,IAAI,UAAU,EAAE,SAAS;YAAE,IAAI,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC;QACrD,IAAI,UAAU,EAAE,sBAAsB;YAAE,IAAI,CAAC,IAAI,CAAC,4BAA4B,CAAC,CAAC;QAChF,IAAI,UAAU,EAAE,MAAM,EAAE,MAAM;YAAE,IAAI,CAAC,IAAI,CAAC,UAAU,EAAE,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;QACnF,IAAI,UAAU,EAAE,UAAU;YAAE,IAAI,CAAC,IAAI,CAAC,eAAe,EAAE,UAAU,CAAC,UAAU,CAAC,CAAC;QAC9E,IAAI,UAAU,EAAE,gBAAgB;YAC9B,IAAI,CAAC,IAAI,CAAC,qBAAqB,EAAE,IAAI,CAAC,SAAS,CAAC,UAAU,CAAC,gBAAgB,CAAC,CAAC,CAAC;QAChF,IAAI,UAAU,EAAE,OAAO;YAAE,IAAI,CAAC,IAAI,CAAC,YAAY,EAAE,UAAU,CAAC,OAAO,CAAC,CAAC;QAErE,MAAM,OAAO,GAAY;YACvB,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,OAAO,EAAE,MAAM,EAAE,EAAE;YACrC,IAAI,OAAO,GAAG,KAAK,CAAC;YACpB,MAAM,KAAK,GAAG,IAAI,WAAW,CAAC,IAAI,CAAC,UAAU,EAAE,OAAO,CAAC,CAAC;YACxD,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,uDAAuD;gBACvD,IAAI,CAAC;oBACH,KAAK,CAAC,IAAI,EAAE,CAAC;gBACf,CAAC;gBAAC,OAAO,KAAK,EAAE,CAAC;oBACf,OAAO,CAAC,KAAK,CACX,+CAA+C,EAC/C,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,0EAA0E,KAAK,CAAC,YAAY,EAAE,GAAG,GAAG,CACrG,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,+DAA+D,EAC/D,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CACvD,CAAC;wBACJ,CAAC;oBACH,CAAC;oBACD,yEAAyE;oBACzE,IAAI,CAAC,OAAO,EAAE,CAAC;wBACb,OAAO,GAAG,IAAI,CAAC;wBACf,MAAM,CAAC,IAAI,QAAQ,CAAC,qBAAqB,IAAI,CAAC,OAAO,6BAA6B,EAAE,aAAa,CAAC,CAAC,CAAC;oBACtG,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,iEAAiE;gBACjE,MAAM,MAAM,GAAG,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;gBACvC,sCAAsC;gBACtC,YAAY,CAAC,MAAM,GAAG,CAAC,CAAC;gBAExB,wEAAwE;gBACxE,mEAAmE;gBACnE,2DAA2D;gBAC3D,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,+BAA+B,UAAU,WAAW,QAAQ,MAAM,YAAY,EAAE,EAChF,UAAU,KAAK,SAAS,IAAI,UAAU,KAAK,SAAS,CAAC,CAAC,CAAC,aAAa,CAAC,CAAC,CAAC,eAAe,CACvF,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,CAAC,CAAC,CAAC;gCACtE,OAAO;4BACT,CAAC;wBACH,CAAC;wBAAC,OAAO,KAAK,EAAE,CAAC;4BACf,OAAO,CAAC,KAAK,CACX,iEAAiE,EACjE,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,yBAAyB,MAAM,EAAE,EAAE,eAAe,CAAC,CAAC,CAAC;oBACzE,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,4BAA4B;wBAC5B,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;yBAAM,CAAC;wBACN,WAAW,CAAC,IAAI,CAAC,8DAA8D,CAAC,CAAC;oBACnF,CAAC;oBACD,MAAM,CAAC,IAAI,QAAQ,CAAC,WAAW,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,eAAe,CAAC,CAAC,CAAC;oBAC7D,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,kEAAkE,EAClE,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,yDAAyD,KAAK,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,EACrG,eAAe,CAChB,CACF,CAAC;oBACF,OAAO;gBACT,CAAC;gBAED,2BAA2B;gBAC3B,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,CAAC,CAAC,CAAC;oBAC5E,OAAO;gBACT,CAAC;gBAED,MAAM,WAAW,GAAG,QAA6B,CAAC;gBAElD,+BAA+B;gBAC/B,IAAI,CAAC,WAAW,CAAC,EAAE,IAAI,CAAC,WAAW,CAAC,YAAY,IAAI,CAAC,WAAW,CAAC,cAAc,EAAE,CAAC;oBAChF,MAAM,CACJ,IAAI,QAAQ,CACV,uDAAuD,IAAI,CAAC,SAAS,CAAC,MAAM,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC,EAAE,EACjG,eAAe,CAChB,CACF,CAAC;oBACF,OAAO;gBACT,CAAC;gBAED,OAAO,CAAC;oBACN,MAAM,EAAE,IAAI,CAAC,WAAW,CAAC,WAAW,CAAC;oBACrC,WAAW,EAAE,IAAI,CAAC,aAAa,CAAC,WAAW,CAAC,YAAY,CAAC;oBACzD,MAAM,EAAE,WAAW,CAAC,MAAM,IAAI,EAAE;oBAChC,UAAU,EAAE,WAAW,CAAC,WAAW,IAAI,IAAI;oBAC3C,QAAQ,EAAE,WAAW,CAAC,QAAQ,IAAI,IAAI;oBACtC,cAAc,EAAE,WAAW,CAAC,eAAe,IAAI,IAAI;oBACnD,QAAQ,EAAE,WAAW,CAAC,SAAS,IAAI,IAAI;oBACvC,SAAS,EAAE,WAAW,CAAC,UAAU,IAAI,IAAI;oBACzC,UAAU,EAAE,WAAW,CAAC,WAAW,IAAI,IAAI;iBAC5C,CAAC,CAAC;YACL,CAAC,CAAC,CAAC;QACL,CAAC,CAAC,CAAC;IACL,CAAC;IAED;;;;;OAKG;IACH,KAAK,CAAC,UAAU,CACd,QAAgB,EAChB,OAAyC,UAAU,EACnD,UAIC;QASD,+CAA+C;QAC/C,MAAM,UAAU,GAAG,cAAc,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAC/C,MAAM,WAAW,GAAG,mBAAmB,GAAG,IAAI,CAAC,GAAG,EAAE,CAAC;QAErD,MAAM,IAAI,GAAa,EAAE,CAAC;QAC1B,IAAI,UAAU,EAAE,OAAO,EAAE,CAAC;YACxB,IAAI,CAAC,IAAI,CAAC,YAAY,EAAE,UAAU,CAAC,OAAO,CAAC,CAAC;QAC9C,CAAC;aAAM,CAAC;YACN,IAAI,CAAC,IAAI,CAAC,QAAQ,EAAE,QAAQ,CAAC,CAAC;QAChC,CAAC;QACD,IAAI,CAAC,IAAI,CAAC,QAAQ,EAAE,IAAI,EAAE,UAAU,EAAE,UAAU,EAAE,WAAW,EAAE,WAAW,EAAE,QAAQ,CAAC,CAAC;QACtF,IAAI,UAAU,EAAE,QAAQ;YAAE,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,MAAM,CAAC,UAAU,CAAC,QAAQ,CAAC,CAAC,CAAC;QAChF,IAAI,UAAU,EAAE,SAAS;YAAE,IAAI,CAAC,IAAI,CAAC,cAAc,EAAE,UAAU,CAAC,SAAS,CAAC,CAAC;QAE3E,MAAM,OAAO,GAAY;YACvB,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,OAAO,EAAE,MAAM,EAAE,EAAE;YACrC,IAAI,OAAO,GAAG,KAAK,CAAC;YACpB,MAAM,KAAK,GAAG,IAAI,WAAW,CAAC,IAAI,CAAC,UAAU,EAAE,OAAO,CAAC,CAAC;YACxD,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,0DAA0D,EAC1D,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,qFAAqF,KAAK,CAAC,YAAY,EAAE,GAAG,GAAG,CAChH,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,0EAA0E,EAC1E,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,qBAAqB,IAAI,CAAC,OAAO,6BAA6B,EAAE,aAAa,CAAC,CAAC,CAAC;oBACtG,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,2DAA2D;gBAC3D,MAAM,WAAW,GAAG,UAAU,EAAE,OAAO,IAAI,QAAQ,CAAC;gBACpD,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,+BAA+B,UAAU,WAAW,WAAW,MAAM,YAAY,EAAE,EACnF,UAAU,KAAK,SAAS,IAAI,UAAU,KAAK,SAAS,CAAC,CAAC,CAAC,aAAa,CAAC,CAAC,CAAC,eAAe,CACvF,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,CAAC,CAAC,CAAC;gCACtE,OAAO;4BACT,CAAC;wBACH,CAAC;wBAAC,OAAO,KAAK,EAAE,CAAC;4BACf,OAAO,CAAC,KAAK,CACX,4EAA4E,EAC5E,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,yBAAyB,MAAM,EAAE,EAAE,eAAe,CAAC,CAAC,CAAC;oBACzE,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,4BAA4B;wBAC5B,cAAc,QAAQ,aAAa,UAAU,GAAG;wBAChD,SAAS,WAAW,GAAG;qBACxB,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;yBAAM,CAAC;wBACN,WAAW,CAAC,IAAI,CAAC,8DAA8D,CAAC,CAAC;oBACnF,CAAC;oBACD,MAAM,CAAC,IAAI,QAAQ,CAAC,WAAW,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,eAAe,CAAC,CAAC,CAAC;oBAC7D,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,6EAA6E,EAC7E,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,yDAAyD,KAAK,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC,EAAE,GAAG,CAAC,EAAE,EACrG,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,CAAC,CAAC,CAAC;oBAC5E,OAAO;gBACT,CAAC;gBAED,MAAM,WAAW,GAAG,QAA6B,CAAC;gBAElD,OAAO,CAAC;oBACN,QAAQ,EAAE,WAAW,CAAC,cAAc,IAAI,EAAE;oBAC1C,SAAS,EAAE,WAAW,CAAC,UAAU,IAAI,CAAC;oBACtC,YAAY,EAAE,WAAW,CAAC,mBAAmB,IAAI,IAAI;oBACrD,SAAS,EAAE,WAAW,CAAC,UAAU,IAAI,IAAI;oBACzC,UAAU,EAAE,WAAW,CAAC,sBAAsB,IAAI,CAAC;oBACnD,QAAQ,EAAE,WAAW,CAAC,QAAQ,IAAI,IAAI;iBACvC,CAAC,CAAC;YACL,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;IAEO,WAAW,CAAC,CAAoB;QACtC,gEAAgE;QAChE,mDAAmD;QACnD,MAAM,MAAM,GAA4B,EAAE,CAAC;QAC3C,IAAI,CAAC,CAAC,QAAQ;YAAE,MAAM,CAAC,QAAQ,GAAG,CAAC,CAAC,QAAQ,CAAC;QAC7C,IAAI,CAAC,CAAC,mBAAmB;YAAE,MAAM,CAAC,cAAc,GAAG,CAAC,CAAC,mBAAmB,CAAC;QAEzE,OAAO;YACL,EAAE,EAAE,CAAC,CAAC,EAAE;YACR,aAAa,EAAE,CAAC,CAAC,aAAa;YAC9B,WAAW,EAAE,CAAC,CAAC,WAAW;YAC1B,cAAc,EAAE,CAAC,CAAC,cAAc;YAChC,WAAW,EAAE,CAAC,CAAC,WAAW;YAC1B,kBAAkB,EAAE,CAAC,CAAC,kBAAkB;YACxC,YAAY,EAAE,CAAC,CAAC,YAAY;YAC5B,mBAAmB,EAAE,CAAC,CAAC,mBAAmB;YAC1C,UAAU,EAAE,CAAC,CAAC,UAAU;YACxB,UAAU,EAAE,CAAC,CAAC,UAAU;YACxB,YAAY,EAAE,CAAC,CAAC,YAAY;YAC5B,qBAAqB,EAAE,CAAC,CAAC,qBAAqB;YAC9C,uBAAuB,EAAE,CAAC,CAAC,uBAAuB;YAClD,sBAAsB,EAAE,CAAC,CAAC,sBAAsB;YAChD,WAAW,EAAE,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,IAAI;YACjE,WAAW,EAAE,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,IAAI;SAC5E,CAAC;IACJ,CAAC;IAEO,aAAa,CACnB,OAAsE;QAEtE,4CAA4C;QAC5C,OAAO,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,SAAS,EAAE,CAAC,CAAC,UAAU,EAAE,OAAO,EAAE,CAAC,CAAC,QAAQ,EAAE,CAAC,CAAC,CAAC;IAC9F,CAAC;CACF"}
|
|
@@ -0,0 +1,38 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* OCR Error Classes
|
|
3
|
+
*
|
|
4
|
+
* FAIL-FAST: These errors propagate immediately, no fallbacks.
|
|
5
|
+
* Must match Python error categories in python/ocr_worker.py lines 34-94.
|
|
6
|
+
*/
|
|
7
|
+
type OCRErrorCategory = 'OCR_API_ERROR' | 'OCR_SERVER_ERROR' | 'OCR_RATE_LIMIT' | 'OCR_TIMEOUT' | 'OCR_FILE_ERROR' | 'OCR_AUTHENTICATION_ERROR' | 'FORM_FILL_API_ERROR' | 'FORM_FILL_SERVER_ERROR' | 'FORM_FILL_FILE_ERROR' | 'FORM_FILL_TIMEOUT' | 'FILE_MANAGER_API_ERROR' | 'FILE_MANAGER_SERVER_ERROR' | 'FILE_MANAGER_FILE_ERROR';
|
|
8
|
+
export declare class OCRError extends Error {
|
|
9
|
+
readonly category: OCRErrorCategory;
|
|
10
|
+
readonly requestId?: string | undefined;
|
|
11
|
+
constructor(message: string, category: OCRErrorCategory, requestId?: string | undefined);
|
|
12
|
+
}
|
|
13
|
+
export declare class OCRAPIError extends OCRError {
|
|
14
|
+
readonly statusCode: number;
|
|
15
|
+
constructor(message: string, statusCode: number, requestId?: string);
|
|
16
|
+
}
|
|
17
|
+
export declare class OCRRateLimitError extends OCRError {
|
|
18
|
+
readonly retryAfter: number;
|
|
19
|
+
constructor(message?: string, retryAfter?: number);
|
|
20
|
+
}
|
|
21
|
+
export declare class OCRTimeoutError extends OCRError {
|
|
22
|
+
constructor(message: string, requestId?: string);
|
|
23
|
+
}
|
|
24
|
+
export declare class OCRFileError extends OCRError {
|
|
25
|
+
readonly filePath: string;
|
|
26
|
+
constructor(message: string, filePath: string);
|
|
27
|
+
}
|
|
28
|
+
export declare class OCRAuthenticationError extends OCRError {
|
|
29
|
+
readonly statusCode: number;
|
|
30
|
+
constructor(message: string, statusCode: number);
|
|
31
|
+
}
|
|
32
|
+
/**
|
|
33
|
+
* Map Python error JSON to TypeScript error class
|
|
34
|
+
* FAIL-FAST: Unknown categories throw OCRError
|
|
35
|
+
*/
|
|
36
|
+
export declare function mapPythonError(category: string, message: string, details: Record<string, unknown>): OCRError;
|
|
37
|
+
export {};
|
|
38
|
+
//# sourceMappingURL=errors.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"errors.d.ts","sourceRoot":"","sources":["../../../src/services/ocr/errors.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,KAAK,gBAAgB,GACjB,eAAe,GACf,kBAAkB,GAClB,gBAAgB,GAChB,aAAa,GACb,gBAAgB,GAChB,0BAA0B,GAC1B,qBAAqB,GACrB,wBAAwB,GACxB,sBAAsB,GACtB,mBAAmB,GACnB,wBAAwB,GACxB,2BAA2B,GAC3B,yBAAyB,CAAC;AAE9B,qBAAa,QAAS,SAAQ,KAAK;aAGf,QAAQ,EAAE,gBAAgB;aAC1B,SAAS,CAAC,EAAE,MAAM;gBAFlC,OAAO,EAAE,MAAM,EACC,QAAQ,EAAE,gBAAgB,EAC1B,SAAS,CAAC,EAAE,MAAM,YAAA;CAKrC;AAED,qBAAa,WAAY,SAAQ,QAAQ;aAGrB,UAAU,EAAE,MAAM;gBADlC,OAAO,EAAE,MAAM,EACC,UAAU,EAAE,MAAM,EAClC,SAAS,CAAC,EAAE,MAAM;CAKrB;AAED,qBAAa,iBAAkB,SAAQ,QAAQ;aAG3B,UAAU,EAAE,MAAM;gBADlC,OAAO,GAAE,MAA8B,EACvB,UAAU,GAAE,MAAW;CAK1C;AAED,qBAAa,eAAgB,SAAQ,QAAQ;gBAC/B,OAAO,EAAE,MAAM,EAAE,SAAS,CAAC,EAAE,MAAM;CAIhD;AAED,qBAAa,YAAa,SAAQ,QAAQ;aAGtB,QAAQ,EAAE,MAAM;gBADhC,OAAO,EAAE,MAAM,EACC,QAAQ,EAAE,MAAM;CAKnC;AAED,qBAAa,sBAAuB,SAAQ,QAAQ;aAGhC,UAAU,EAAE,MAAM;gBADlC,OAAO,EAAE,MAAM,EACC,UAAU,EAAE,MAAM;CAKrC;AAED;;;GAGG;AACH,wBAAgB,cAAc,CAC5B,QAAQ,EAAE,MAAM,EAChB,OAAO,EAAE,MAAM,EACf,OAAO,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAC/B,QAAQ,CA8BV"}
|
|
@@ -0,0 +1,83 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* OCR Error Classes
|
|
3
|
+
*
|
|
4
|
+
* FAIL-FAST: These errors propagate immediately, no fallbacks.
|
|
5
|
+
* Must match Python error categories in python/ocr_worker.py lines 34-94.
|
|
6
|
+
*/
|
|
7
|
+
export class OCRError extends Error {
|
|
8
|
+
category;
|
|
9
|
+
requestId;
|
|
10
|
+
constructor(message, category, requestId) {
|
|
11
|
+
super(message);
|
|
12
|
+
this.category = category;
|
|
13
|
+
this.requestId = requestId;
|
|
14
|
+
this.name = 'OCRError';
|
|
15
|
+
}
|
|
16
|
+
}
|
|
17
|
+
export class OCRAPIError extends OCRError {
|
|
18
|
+
statusCode;
|
|
19
|
+
constructor(message, statusCode, requestId) {
|
|
20
|
+
super(message, statusCode >= 500 ? 'OCR_SERVER_ERROR' : 'OCR_API_ERROR', requestId);
|
|
21
|
+
this.statusCode = statusCode;
|
|
22
|
+
this.name = 'OCRAPIError';
|
|
23
|
+
}
|
|
24
|
+
}
|
|
25
|
+
export class OCRRateLimitError extends OCRError {
|
|
26
|
+
retryAfter;
|
|
27
|
+
constructor(message = 'Rate limit exceeded', retryAfter = 60) {
|
|
28
|
+
super(message, 'OCR_RATE_LIMIT');
|
|
29
|
+
this.retryAfter = retryAfter;
|
|
30
|
+
this.name = 'OCRRateLimitError';
|
|
31
|
+
}
|
|
32
|
+
}
|
|
33
|
+
export class OCRTimeoutError extends OCRError {
|
|
34
|
+
constructor(message, requestId) {
|
|
35
|
+
super(message, 'OCR_TIMEOUT', requestId);
|
|
36
|
+
this.name = 'OCRTimeoutError';
|
|
37
|
+
}
|
|
38
|
+
}
|
|
39
|
+
export class OCRFileError extends OCRError {
|
|
40
|
+
filePath;
|
|
41
|
+
constructor(message, filePath) {
|
|
42
|
+
super(message, 'OCR_FILE_ERROR');
|
|
43
|
+
this.filePath = filePath;
|
|
44
|
+
this.name = 'OCRFileError';
|
|
45
|
+
}
|
|
46
|
+
}
|
|
47
|
+
export class OCRAuthenticationError extends OCRError {
|
|
48
|
+
statusCode;
|
|
49
|
+
constructor(message, statusCode) {
|
|
50
|
+
super(message, 'OCR_AUTHENTICATION_ERROR');
|
|
51
|
+
this.statusCode = statusCode;
|
|
52
|
+
this.name = 'OCRAuthenticationError';
|
|
53
|
+
}
|
|
54
|
+
}
|
|
55
|
+
/**
|
|
56
|
+
* Map Python error JSON to TypeScript error class
|
|
57
|
+
* FAIL-FAST: Unknown categories throw OCRError
|
|
58
|
+
*/
|
|
59
|
+
export function mapPythonError(category, message, details) {
|
|
60
|
+
switch (category) {
|
|
61
|
+
case 'OCR_API_ERROR':
|
|
62
|
+
case 'OCR_SERVER_ERROR':
|
|
63
|
+
case 'FORM_FILL_API_ERROR':
|
|
64
|
+
case 'FORM_FILL_SERVER_ERROR':
|
|
65
|
+
case 'FILE_MANAGER_API_ERROR':
|
|
66
|
+
case 'FILE_MANAGER_SERVER_ERROR':
|
|
67
|
+
return new OCRAPIError(message, details.status_code ?? 500, details.request_id);
|
|
68
|
+
case 'OCR_RATE_LIMIT':
|
|
69
|
+
return new OCRRateLimitError(message, details.retry_after ?? 60);
|
|
70
|
+
case 'OCR_TIMEOUT':
|
|
71
|
+
case 'FORM_FILL_TIMEOUT':
|
|
72
|
+
return new OCRTimeoutError(message, details.request_id);
|
|
73
|
+
case 'OCR_FILE_ERROR':
|
|
74
|
+
case 'FORM_FILL_FILE_ERROR':
|
|
75
|
+
case 'FILE_MANAGER_FILE_ERROR':
|
|
76
|
+
return new OCRFileError(message, details.file_path ?? 'unknown');
|
|
77
|
+
case 'OCR_AUTHENTICATION_ERROR':
|
|
78
|
+
return new OCRAuthenticationError(message, details.status_code ?? 401);
|
|
79
|
+
default:
|
|
80
|
+
throw new OCRError(`Unknown error category: ${category}. Message: ${message}`, 'OCR_API_ERROR');
|
|
81
|
+
}
|
|
82
|
+
}
|
|
83
|
+
//# sourceMappingURL=errors.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"errors.js","sourceRoot":"","sources":["../../../src/services/ocr/errors.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAiBH,MAAM,OAAO,QAAS,SAAQ,KAAK;IAGf;IACA;IAHlB,YACE,OAAe,EACC,QAA0B,EAC1B,SAAkB;QAElC,KAAK,CAAC,OAAO,CAAC,CAAC;QAHC,aAAQ,GAAR,QAAQ,CAAkB;QAC1B,cAAS,GAAT,SAAS,CAAS;QAGlC,IAAI,CAAC,IAAI,GAAG,UAAU,CAAC;IACzB,CAAC;CACF;AAED,MAAM,OAAO,WAAY,SAAQ,QAAQ;IAGrB;IAFlB,YACE,OAAe,EACC,UAAkB,EAClC,SAAkB;QAElB,KAAK,CAAC,OAAO,EAAE,UAAU,IAAI,GAAG,CAAC,CAAC,CAAC,kBAAkB,CAAC,CAAC,CAAC,eAAe,EAAE,SAAS,CAAC,CAAC;QAHpE,eAAU,GAAV,UAAU,CAAQ;QAIlC,IAAI,CAAC,IAAI,GAAG,aAAa,CAAC;IAC5B,CAAC;CACF;AAED,MAAM,OAAO,iBAAkB,SAAQ,QAAQ;IAG3B;IAFlB,YACE,UAAkB,qBAAqB,EACvB,aAAqB,EAAE;QAEvC,KAAK,CAAC,OAAO,EAAE,gBAAgB,CAAC,CAAC;QAFjB,eAAU,GAAV,UAAU,CAAa;QAGvC,IAAI,CAAC,IAAI,GAAG,mBAAmB,CAAC;IAClC,CAAC;CACF;AAED,MAAM,OAAO,eAAgB,SAAQ,QAAQ;IAC3C,YAAY,OAAe,EAAE,SAAkB;QAC7C,KAAK,CAAC,OAAO,EAAE,aAAa,EAAE,SAAS,CAAC,CAAC;QACzC,IAAI,CAAC,IAAI,GAAG,iBAAiB,CAAC;IAChC,CAAC;CACF;AAED,MAAM,OAAO,YAAa,SAAQ,QAAQ;IAGtB;IAFlB,YACE,OAAe,EACC,QAAgB;QAEhC,KAAK,CAAC,OAAO,EAAE,gBAAgB,CAAC,CAAC;QAFjB,aAAQ,GAAR,QAAQ,CAAQ;QAGhC,IAAI,CAAC,IAAI,GAAG,cAAc,CAAC;IAC7B,CAAC;CACF;AAED,MAAM,OAAO,sBAAuB,SAAQ,QAAQ;IAGhC;IAFlB,YACE,OAAe,EACC,UAAkB;QAElC,KAAK,CAAC,OAAO,EAAE,0BAA0B,CAAC,CAAC;QAF3B,eAAU,GAAV,UAAU,CAAQ;QAGlC,IAAI,CAAC,IAAI,GAAG,wBAAwB,CAAC;IACvC,CAAC;CACF;AAED;;;GAGG;AACH,MAAM,UAAU,cAAc,CAC5B,QAAgB,EAChB,OAAe,EACf,OAAgC;IAEhC,QAAQ,QAAQ,EAAE,CAAC;QACjB,KAAK,eAAe,CAAC;QACrB,KAAK,kBAAkB,CAAC;QACxB,KAAK,qBAAqB,CAAC;QAC3B,KAAK,wBAAwB,CAAC;QAC9B,KAAK,wBAAwB,CAAC;QAC9B,KAAK,2BAA2B;YAC9B,OAAO,IAAI,WAAW,CACpB,OAAO,EACN,OAAO,CAAC,WAAsB,IAAI,GAAG,EACtC,OAAO,CAAC,UAAoB,CAC7B,CAAC;QACJ,KAAK,gBAAgB;YACnB,OAAO,IAAI,iBAAiB,CAAC,OAAO,EAAG,OAAO,CAAC,WAAsB,IAAI,EAAE,CAAC,CAAC;QAC/E,KAAK,aAAa,CAAC;QACnB,KAAK,mBAAmB;YACtB,OAAO,IAAI,eAAe,CAAC,OAAO,EAAE,OAAO,CAAC,UAAoB,CAAC,CAAC;QACpE,KAAK,gBAAgB,CAAC;QACtB,KAAK,sBAAsB,CAAC;QAC5B,KAAK,yBAAyB;YAC5B,OAAO,IAAI,YAAY,CAAC,OAAO,EAAG,OAAO,CAAC,SAAoB,IAAI,SAAS,CAAC,CAAC;QAC/E,KAAK,0BAA0B;YAC7B,OAAO,IAAI,sBAAsB,CAAC,OAAO,EAAG,OAAO,CAAC,WAAsB,IAAI,GAAG,CAAC,CAAC;QACrF;YACE,MAAM,IAAI,QAAQ,CAChB,2BAA2B,QAAQ,cAAc,OAAO,EAAE,EAC1D,eAAe,CAChB,CAAC;IACN,CAAC;AACH,CAAC"}
|
|
@@ -0,0 +1,76 @@
|
|
|
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
|
+
interface FileUploadResult {
|
|
8
|
+
fileId: string;
|
|
9
|
+
reference: string | null;
|
|
10
|
+
fileName: string;
|
|
11
|
+
fileHash: string;
|
|
12
|
+
fileSize: number;
|
|
13
|
+
contentType: string;
|
|
14
|
+
status: 'complete' | 'failed';
|
|
15
|
+
error: string | null;
|
|
16
|
+
processingDurationMs: number;
|
|
17
|
+
}
|
|
18
|
+
interface FileInfo {
|
|
19
|
+
fileId: string;
|
|
20
|
+
fileName: string | null;
|
|
21
|
+
fileSize: number | null;
|
|
22
|
+
contentType: string | null;
|
|
23
|
+
createdAt: string | null;
|
|
24
|
+
reference: string | null;
|
|
25
|
+
status: string | null;
|
|
26
|
+
}
|
|
27
|
+
interface FileListResult {
|
|
28
|
+
files: Record<string, unknown>[];
|
|
29
|
+
total: number;
|
|
30
|
+
}
|
|
31
|
+
interface FileManagerClientConfig {
|
|
32
|
+
pythonPath?: string;
|
|
33
|
+
timeout?: number;
|
|
34
|
+
}
|
|
35
|
+
export declare class FileManagerClient {
|
|
36
|
+
private readonly pythonPath;
|
|
37
|
+
private readonly workerPath;
|
|
38
|
+
private readonly timeout;
|
|
39
|
+
constructor(config?: FileManagerClientConfig);
|
|
40
|
+
/**
|
|
41
|
+
* Upload a file to Datalab cloud storage
|
|
42
|
+
*
|
|
43
|
+
* FAIL-FAST: Throws on any error, no fallbacks
|
|
44
|
+
*/
|
|
45
|
+
uploadFile(filePath: string): Promise<FileUploadResult>;
|
|
46
|
+
/**
|
|
47
|
+
* List files in Datalab cloud storage
|
|
48
|
+
*/
|
|
49
|
+
listFiles(limit?: number, offset?: number): Promise<FileListResult>;
|
|
50
|
+
/**
|
|
51
|
+
* Get file metadata by ID
|
|
52
|
+
*/
|
|
53
|
+
getFile(fileId: string): Promise<FileInfo>;
|
|
54
|
+
/**
|
|
55
|
+
* Get download URL for a file
|
|
56
|
+
*
|
|
57
|
+
* @param fileId - Datalab file ID
|
|
58
|
+
* @param expiresIn - URL expiry time in seconds (default: 3600, min: 60, max: 86400)
|
|
59
|
+
*/
|
|
60
|
+
getDownloadUrl(fileId: string, expiresIn?: number): Promise<{
|
|
61
|
+
downloadUrl: string;
|
|
62
|
+
expiresIn: number;
|
|
63
|
+
fileId: string;
|
|
64
|
+
}>;
|
|
65
|
+
/**
|
|
66
|
+
* Delete a file from Datalab cloud storage
|
|
67
|
+
*/
|
|
68
|
+
deleteFile(fileId: string): Promise<boolean>;
|
|
69
|
+
/**
|
|
70
|
+
* Run the Python worker with given args and parse JSON output
|
|
71
|
+
*/
|
|
72
|
+
private runWorker;
|
|
73
|
+
private isErrorResponse;
|
|
74
|
+
}
|
|
75
|
+
export {};
|
|
76
|
+
//# sourceMappingURL=file-manager.d.ts.map
|