@vertesia/workflow 0.42.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/LICENSE +13 -0
- package/README.md +24 -0
- package/bin/bundle-workflows.mjs +26 -0
- package/lib/cjs/activities/advanced/createDocumentTypeFromInteractionRun.js +32 -0
- package/lib/cjs/activities/advanced/createDocumentTypeFromInteractionRun.js.map +1 -0
- package/lib/cjs/activities/advanced/createOrUpdateDocumentFromInteractionRun.js +66 -0
- package/lib/cjs/activities/advanced/createOrUpdateDocumentFromInteractionRun.js.map +1 -0
- package/lib/cjs/activities/advanced/updateDocumentFromInteractionRun.js +18 -0
- package/lib/cjs/activities/advanced/updateDocumentFromInteractionRun.js.map +1 -0
- package/lib/cjs/activities/chunkDocument.js +79 -0
- package/lib/cjs/activities/chunkDocument.js.map +1 -0
- package/lib/cjs/activities/createDocumentFromOther.js +64 -0
- package/lib/cjs/activities/createDocumentFromOther.js.map +1 -0
- package/lib/cjs/activities/executeInteraction.js +135 -0
- package/lib/cjs/activities/executeInteraction.js.map +1 -0
- package/lib/cjs/activities/extractDocumentText.js +140 -0
- package/lib/cjs/activities/extractDocumentText.js.map +1 -0
- package/lib/cjs/activities/generateDocumentProperties.js +59 -0
- package/lib/cjs/activities/generateDocumentProperties.js.map +1 -0
- package/lib/cjs/activities/generateEmbeddings.js +292 -0
- package/lib/cjs/activities/generateEmbeddings.js.map +1 -0
- package/lib/cjs/activities/generateImageRendition.js +104 -0
- package/lib/cjs/activities/generateImageRendition.js.map +1 -0
- package/lib/cjs/activities/generateOrAssignContentType.js +103 -0
- package/lib/cjs/activities/generateOrAssignContentType.js.map +1 -0
- package/lib/cjs/activities/getObjectFromStore.js +20 -0
- package/lib/cjs/activities/getObjectFromStore.js.map +1 -0
- package/lib/cjs/activities/index.js +54 -0
- package/lib/cjs/activities/index.js.map +1 -0
- package/lib/cjs/activities/media/processPdfWithTextract.js +102 -0
- package/lib/cjs/activities/media/processPdfWithTextract.js.map +1 -0
- package/lib/cjs/activities/media/transcribeMediaWithGladia.js +51 -0
- package/lib/cjs/activities/media/transcribeMediaWithGladia.js.map +1 -0
- package/lib/cjs/activities/notifyWebhook.js +34 -0
- package/lib/cjs/activities/notifyWebhook.js.map +1 -0
- package/lib/cjs/activities/setDocumentStatus.js +15 -0
- package/lib/cjs/activities/setDocumentStatus.js.map +1 -0
- package/lib/cjs/conversion/TextractProcessor.js +416 -0
- package/lib/cjs/conversion/TextractProcessor.js.map +1 -0
- package/lib/cjs/conversion/image.js +22 -0
- package/lib/cjs/conversion/image.js.map +1 -0
- package/lib/cjs/conversion/mutool.js +147 -0
- package/lib/cjs/conversion/mutool.js.map +1 -0
- package/lib/cjs/conversion/pandoc.js +39 -0
- package/lib/cjs/conversion/pandoc.js.map +1 -0
- package/lib/cjs/conversion/pdf.js +13 -0
- package/lib/cjs/conversion/pdf.js.map +1 -0
- package/lib/cjs/dsl/conditions.js +81 -0
- package/lib/cjs/dsl/conditions.js.map +1 -0
- package/lib/cjs/dsl/dsl-workflow.js +223 -0
- package/lib/cjs/dsl/dsl-workflow.js.map +1 -0
- package/lib/cjs/dsl/projections.js +59 -0
- package/lib/cjs/dsl/projections.js.map +1 -0
- package/lib/cjs/dsl/setup/ActivityContext.js +96 -0
- package/lib/cjs/dsl/setup/ActivityContext.js.map +1 -0
- package/lib/cjs/dsl/setup/fetch/DataProvider.js +51 -0
- package/lib/cjs/dsl/setup/fetch/DataProvider.js.map +1 -0
- package/lib/cjs/dsl/setup/fetch/index.js +16 -0
- package/lib/cjs/dsl/setup/fetch/index.js.map +1 -0
- package/lib/cjs/dsl/setup/fetch/providers.js +67 -0
- package/lib/cjs/dsl/setup/fetch/providers.js.map +1 -0
- package/lib/cjs/dsl/test/test-child-workflow.js +10 -0
- package/lib/cjs/dsl/test/test-child-workflow.js.map +1 -0
- package/lib/cjs/dsl/validation.js +122 -0
- package/lib/cjs/dsl/validation.js.map +1 -0
- package/lib/cjs/dsl/vars.js +341 -0
- package/lib/cjs/dsl/vars.js.map +1 -0
- package/lib/cjs/dsl/walk.js +100 -0
- package/lib/cjs/dsl/walk.js.map +1 -0
- package/lib/cjs/errors.js +36 -0
- package/lib/cjs/errors.js.map +1 -0
- package/lib/cjs/index.js +43 -0
- package/lib/cjs/index.js.map +1 -0
- package/lib/cjs/iterative-generation/activities/extractToc.js +47 -0
- package/lib/cjs/iterative-generation/activities/extractToc.js.map +1 -0
- package/lib/cjs/iterative-generation/activities/finalizeOutput.js +69 -0
- package/lib/cjs/iterative-generation/activities/finalizeOutput.js.map +1 -0
- package/lib/cjs/iterative-generation/activities/generatePart.js +73 -0
- package/lib/cjs/iterative-generation/activities/generatePart.js.map +1 -0
- package/lib/cjs/iterative-generation/activities/generateToc.js +91 -0
- package/lib/cjs/iterative-generation/activities/generateToc.js.map +1 -0
- package/lib/cjs/iterative-generation/activities/index.js +12 -0
- package/lib/cjs/iterative-generation/activities/index.js.map +1 -0
- package/lib/cjs/iterative-generation/iterativeGenerationWorkflow.js +55 -0
- package/lib/cjs/iterative-generation/iterativeGenerationWorkflow.js.map +1 -0
- package/lib/cjs/iterative-generation/types.js +5 -0
- package/lib/cjs/iterative-generation/types.js.map +1 -0
- package/lib/cjs/iterative-generation/utils.js +118 -0
- package/lib/cjs/iterative-generation/utils.js.map +1 -0
- package/lib/cjs/package.json +3 -0
- package/lib/cjs/result-types.js +10 -0
- package/lib/cjs/result-types.js.map +1 -0
- package/lib/cjs/system/generateObjectText.js +89 -0
- package/lib/cjs/system/generateObjectText.js.map +1 -0
- package/lib/cjs/system/notifyWebhookWorkflow.js +52 -0
- package/lib/cjs/system/notifyWebhookWorkflow.js.map +1 -0
- package/lib/cjs/system/recalculateEmbeddingsWorkflow.js +37 -0
- package/lib/cjs/system/recalculateEmbeddingsWorkflow.js.map +1 -0
- package/lib/cjs/utils/auth.js +15 -0
- package/lib/cjs/utils/auth.js.map +1 -0
- package/lib/cjs/utils/blobs.js +63 -0
- package/lib/cjs/utils/blobs.js.map +1 -0
- package/lib/cjs/utils/client.js +25 -0
- package/lib/cjs/utils/client.js.map +1 -0
- package/lib/cjs/utils/expand-vars.js +33 -0
- package/lib/cjs/utils/expand-vars.js.map +1 -0
- package/lib/cjs/utils/memory.js +72 -0
- package/lib/cjs/utils/memory.js.map +1 -0
- package/lib/cjs/utils/tokens.js +38 -0
- package/lib/cjs/utils/tokens.js.map +1 -0
- package/lib/cjs/vars.js +20 -0
- package/lib/cjs/vars.js.map +1 -0
- package/lib/cjs/workflows.js +17 -0
- package/lib/cjs/workflows.js.map +1 -0
- package/lib/esm/activities/advanced/createDocumentTypeFromInteractionRun.js +29 -0
- package/lib/esm/activities/advanced/createDocumentTypeFromInteractionRun.js.map +1 -0
- package/lib/esm/activities/advanced/createOrUpdateDocumentFromInteractionRun.js +63 -0
- package/lib/esm/activities/advanced/createOrUpdateDocumentFromInteractionRun.js.map +1 -0
- package/lib/esm/activities/advanced/updateDocumentFromInteractionRun.js +15 -0
- package/lib/esm/activities/advanced/updateDocumentFromInteractionRun.js.map +1 -0
- package/lib/esm/activities/chunkDocument.js +76 -0
- package/lib/esm/activities/chunkDocument.js.map +1 -0
- package/lib/esm/activities/createDocumentFromOther.js +58 -0
- package/lib/esm/activities/createDocumentFromOther.js.map +1 -0
- package/lib/esm/activities/executeInteraction.js +131 -0
- package/lib/esm/activities/executeInteraction.js.map +1 -0
- package/lib/esm/activities/extractDocumentText.js +137 -0
- package/lib/esm/activities/extractDocumentText.js.map +1 -0
- package/lib/esm/activities/generateDocumentProperties.js +56 -0
- package/lib/esm/activities/generateDocumentProperties.js.map +1 -0
- package/lib/esm/activities/generateEmbeddings.js +256 -0
- package/lib/esm/activities/generateEmbeddings.js.map +1 -0
- package/lib/esm/activities/generateImageRendition.js +98 -0
- package/lib/esm/activities/generateImageRendition.js.map +1 -0
- package/lib/esm/activities/generateOrAssignContentType.js +100 -0
- package/lib/esm/activities/generateOrAssignContentType.js.map +1 -0
- package/lib/esm/activities/getObjectFromStore.js +17 -0
- package/lib/esm/activities/getObjectFromStore.js.map +1 -0
- package/lib/esm/activities/index.js +21 -0
- package/lib/esm/activities/index.js.map +1 -0
- package/lib/esm/activities/media/processPdfWithTextract.js +98 -0
- package/lib/esm/activities/media/processPdfWithTextract.js.map +1 -0
- package/lib/esm/activities/media/transcribeMediaWithGladia.js +48 -0
- package/lib/esm/activities/media/transcribeMediaWithGladia.js.map +1 -0
- package/lib/esm/activities/notifyWebhook.js +31 -0
- package/lib/esm/activities/notifyWebhook.js.map +1 -0
- package/lib/esm/activities/setDocumentStatus.js +12 -0
- package/lib/esm/activities/setDocumentStatus.js.map +1 -0
- package/lib/esm/conversion/TextractProcessor.js +409 -0
- package/lib/esm/conversion/TextractProcessor.js.map +1 -0
- package/lib/esm/conversion/image.js +16 -0
- package/lib/esm/conversion/image.js.map +1 -0
- package/lib/esm/conversion/mutool.js +139 -0
- package/lib/esm/conversion/mutool.js.map +1 -0
- package/lib/esm/conversion/pandoc.js +36 -0
- package/lib/esm/conversion/pandoc.js.map +1 -0
- package/lib/esm/conversion/pdf.js +7 -0
- package/lib/esm/conversion/pdf.js.map +1 -0
- package/lib/esm/dsl/conditions.js +75 -0
- package/lib/esm/dsl/conditions.js.map +1 -0
- package/lib/esm/dsl/dsl-workflow.js +216 -0
- package/lib/esm/dsl/dsl-workflow.js.map +1 -0
- package/lib/esm/dsl/projections.js +55 -0
- package/lib/esm/dsl/projections.js.map +1 -0
- package/lib/esm/dsl/setup/ActivityContext.js +91 -0
- package/lib/esm/dsl/setup/ActivityContext.js.map +1 -0
- package/lib/esm/dsl/setup/fetch/DataProvider.js +47 -0
- package/lib/esm/dsl/setup/fetch/DataProvider.js.map +1 -0
- package/lib/esm/dsl/setup/fetch/index.js +12 -0
- package/lib/esm/dsl/setup/fetch/index.js.map +1 -0
- package/lib/esm/dsl/setup/fetch/providers.js +61 -0
- package/lib/esm/dsl/setup/fetch/providers.js.map +1 -0
- package/lib/esm/dsl/test/test-child-workflow.js +5 -0
- package/lib/esm/dsl/test/test-child-workflow.js.map +1 -0
- package/lib/esm/dsl/validation.js +118 -0
- package/lib/esm/dsl/validation.js.map +1 -0
- package/lib/esm/dsl/vars.js +335 -0
- package/lib/esm/dsl/vars.js.map +1 -0
- package/lib/esm/dsl/walk.js +96 -0
- package/lib/esm/dsl/walk.js.map +1 -0
- package/lib/esm/errors.js +30 -0
- package/lib/esm/errors.js.map +1 -0
- package/lib/esm/index.js +25 -0
- package/lib/esm/index.js.map +1 -0
- package/lib/esm/iterative-generation/activities/extractToc.js +44 -0
- package/lib/esm/iterative-generation/activities/extractToc.js.map +1 -0
- package/lib/esm/iterative-generation/activities/finalizeOutput.js +66 -0
- package/lib/esm/iterative-generation/activities/finalizeOutput.js.map +1 -0
- package/lib/esm/iterative-generation/activities/generatePart.js +70 -0
- package/lib/esm/iterative-generation/activities/generatePart.js.map +1 -0
- package/lib/esm/iterative-generation/activities/generateToc.js +88 -0
- package/lib/esm/iterative-generation/activities/generateToc.js.map +1 -0
- package/lib/esm/iterative-generation/activities/index.js +5 -0
- package/lib/esm/iterative-generation/activities/index.js.map +1 -0
- package/lib/esm/iterative-generation/iterativeGenerationWorkflow.js +52 -0
- package/lib/esm/iterative-generation/iterativeGenerationWorkflow.js.map +1 -0
- package/lib/esm/iterative-generation/types.js +2 -0
- package/lib/esm/iterative-generation/types.js.map +1 -0
- package/lib/esm/iterative-generation/utils.js +109 -0
- package/lib/esm/iterative-generation/utils.js.map +1 -0
- package/lib/esm/result-types.js +7 -0
- package/lib/esm/result-types.js.map +1 -0
- package/lib/esm/system/generateObjectText.js +86 -0
- package/lib/esm/system/generateObjectText.js.map +1 -0
- package/lib/esm/system/notifyWebhookWorkflow.js +49 -0
- package/lib/esm/system/notifyWebhookWorkflow.js.map +1 -0
- package/lib/esm/system/recalculateEmbeddingsWorkflow.js +34 -0
- package/lib/esm/system/recalculateEmbeddingsWorkflow.js.map +1 -0
- package/lib/esm/utils/auth.js +8 -0
- package/lib/esm/utils/auth.js.map +1 -0
- package/lib/esm/utils/blobs.js +52 -0
- package/lib/esm/utils/blobs.js.map +1 -0
- package/lib/esm/utils/client.js +22 -0
- package/lib/esm/utils/client.js.map +1 -0
- package/lib/esm/utils/expand-vars.js +30 -0
- package/lib/esm/utils/expand-vars.js.map +1 -0
- package/lib/esm/utils/memory.js +60 -0
- package/lib/esm/utils/memory.js.map +1 -0
- package/lib/esm/utils/tokens.js +34 -0
- package/lib/esm/utils/tokens.js.map +1 -0
- package/lib/esm/vars.js +4 -0
- package/lib/esm/vars.js.map +1 -0
- package/lib/esm/workflows.js +9 -0
- package/lib/esm/workflows.js.map +1 -0
- package/lib/types/activities/advanced/createDocumentTypeFromInteractionRun.d.ts +17 -0
- package/lib/types/activities/advanced/createDocumentTypeFromInteractionRun.d.ts.map +1 -0
- package/lib/types/activities/advanced/createOrUpdateDocumentFromInteractionRun.d.ts +29 -0
- package/lib/types/activities/advanced/createOrUpdateDocumentFromInteractionRun.d.ts.map +1 -0
- package/lib/types/activities/advanced/updateDocumentFromInteractionRun.d.ts +19 -0
- package/lib/types/activities/advanced/updateDocumentFromInteractionRun.d.ts.map +1 -0
- package/lib/types/activities/chunkDocument.d.ts +18 -0
- package/lib/types/activities/chunkDocument.d.ts.map +1 -0
- package/lib/types/activities/createDocumentFromOther.d.ts +21 -0
- package/lib/types/activities/createDocumentFromOther.d.ts.map +1 -0
- package/lib/types/activities/executeInteraction.d.ts +40 -0
- package/lib/types/activities/executeInteraction.d.ts.map +1 -0
- package/lib/types/activities/extractDocumentText.d.ts +9 -0
- package/lib/types/activities/extractDocumentText.d.ts.map +1 -0
- package/lib/types/activities/generateDocumentProperties.d.ts +32 -0
- package/lib/types/activities/generateDocumentProperties.d.ts.map +1 -0
- package/lib/types/activities/generateEmbeddings.d.ts +49 -0
- package/lib/types/activities/generateEmbeddings.d.ts.map +1 -0
- package/lib/types/activities/generateImageRendition.d.ts +17 -0
- package/lib/types/activities/generateImageRendition.d.ts.map +1 -0
- package/lib/types/activities/generateOrAssignContentType.d.ts +44 -0
- package/lib/types/activities/generateOrAssignContentType.d.ts.map +1 -0
- package/lib/types/activities/getObjectFromStore.d.ts +14 -0
- package/lib/types/activities/getObjectFromStore.d.ts.map +1 -0
- package/lib/types/activities/index.d.ts +21 -0
- package/lib/types/activities/index.d.ts.map +1 -0
- package/lib/types/activities/media/processPdfWithTextract.d.ts +26 -0
- package/lib/types/activities/media/processPdfWithTextract.d.ts.map +1 -0
- package/lib/types/activities/media/transcribeMediaWithGladia.d.ts +14 -0
- package/lib/types/activities/media/transcribeMediaWithGladia.d.ts.map +1 -0
- package/lib/types/activities/notifyWebhook.d.ts +17 -0
- package/lib/types/activities/notifyWebhook.d.ts.map +1 -0
- package/lib/types/activities/setDocumentStatus.d.ts +15 -0
- package/lib/types/activities/setDocumentStatus.d.ts.map +1 -0
- package/lib/types/conversion/TextractProcessor.d.ts +45 -0
- package/lib/types/conversion/TextractProcessor.d.ts.map +1 -0
- package/lib/types/conversion/image.d.ts +9 -0
- package/lib/types/conversion/image.d.ts.map +1 -0
- package/lib/types/conversion/mutool.d.ts +19 -0
- package/lib/types/conversion/mutool.d.ts.map +1 -0
- package/lib/types/conversion/pandoc.d.ts +2 -0
- package/lib/types/conversion/pandoc.d.ts.map +1 -0
- package/lib/types/conversion/pdf.d.ts +2 -0
- package/lib/types/conversion/pdf.d.ts.map +1 -0
- package/lib/types/dsl/conditions.d.ts +2 -0
- package/lib/types/dsl/conditions.d.ts.map +1 -0
- package/lib/types/dsl/dsl-workflow.d.ts +5 -0
- package/lib/types/dsl/dsl-workflow.d.ts.map +1 -0
- package/lib/types/dsl/projections.d.ts +4 -0
- package/lib/types/dsl/projections.d.ts.map +1 -0
- package/lib/types/dsl/setup/ActivityContext.d.ts +14 -0
- package/lib/types/dsl/setup/ActivityContext.d.ts.map +1 -0
- package/lib/types/dsl/setup/fetch/DataProvider.d.ts +9 -0
- package/lib/types/dsl/setup/fetch/DataProvider.d.ts.map +1 -0
- package/lib/types/dsl/setup/fetch/index.d.ts +6 -0
- package/lib/types/dsl/setup/fetch/index.d.ts.map +1 -0
- package/lib/types/dsl/setup/fetch/providers.d.ts +25 -0
- package/lib/types/dsl/setup/fetch/providers.d.ts.map +1 -0
- package/lib/types/dsl/test/test-child-workflow.d.ts +4 -0
- package/lib/types/dsl/test/test-child-workflow.d.ts.map +1 -0
- package/lib/types/dsl/validation.d.ts +4 -0
- package/lib/types/dsl/validation.d.ts.map +1 -0
- package/lib/types/dsl/vars.d.ts +48 -0
- package/lib/types/dsl/vars.d.ts.map +1 -0
- package/lib/types/dsl/walk.d.ts +18 -0
- package/lib/types/dsl/walk.d.ts.map +1 -0
- package/lib/types/errors.d.ts +16 -0
- package/lib/types/errors.d.ts.map +1 -0
- package/lib/types/index.d.ts +24 -0
- package/lib/types/index.d.ts.map +1 -0
- package/lib/types/iterative-generation/activities/extractToc.d.ts +10 -0
- package/lib/types/iterative-generation/activities/extractToc.d.ts.map +1 -0
- package/lib/types/iterative-generation/activities/finalizeOutput.d.ts +3 -0
- package/lib/types/iterative-generation/activities/finalizeOutput.d.ts.map +1 -0
- package/lib/types/iterative-generation/activities/generatePart.d.ts +3 -0
- package/lib/types/iterative-generation/activities/generatePart.d.ts.map +1 -0
- package/lib/types/iterative-generation/activities/generateToc.d.ts +4 -0
- package/lib/types/iterative-generation/activities/generateToc.d.ts.map +1 -0
- package/lib/types/iterative-generation/activities/index.d.ts +5 -0
- package/lib/types/iterative-generation/activities/index.d.ts.map +1 -0
- package/lib/types/iterative-generation/iterativeGenerationWorkflow.d.ts +3 -0
- package/lib/types/iterative-generation/iterativeGenerationWorkflow.d.ts.map +1 -0
- package/lib/types/iterative-generation/types.d.ts +79 -0
- package/lib/types/iterative-generation/types.d.ts.map +1 -0
- package/lib/types/iterative-generation/utils.d.ts +27 -0
- package/lib/types/iterative-generation/utils.d.ts.map +1 -0
- package/lib/types/result-types.d.ts +22 -0
- package/lib/types/result-types.d.ts.map +1 -0
- package/lib/types/system/generateObjectText.d.ts +4 -0
- package/lib/types/system/generateObjectText.d.ts.map +1 -0
- package/lib/types/system/notifyWebhookWorkflow.d.ts +6 -0
- package/lib/types/system/notifyWebhookWorkflow.d.ts.map +1 -0
- package/lib/types/system/recalculateEmbeddingsWorkflow.d.ts +40 -0
- package/lib/types/system/recalculateEmbeddingsWorkflow.d.ts.map +1 -0
- package/lib/types/utils/auth.d.ts +4 -0
- package/lib/types/utils/auth.d.ts.map +1 -0
- package/lib/types/utils/blobs.d.ts +8 -0
- package/lib/types/utils/blobs.d.ts.map +1 -0
- package/lib/types/utils/client.d.ts +7 -0
- package/lib/types/utils/client.d.ts.map +1 -0
- package/lib/types/utils/expand-vars.d.ts +8 -0
- package/lib/types/utils/expand-vars.d.ts.map +1 -0
- package/lib/types/utils/memory.d.ts +12 -0
- package/lib/types/utils/memory.d.ts.map +1 -0
- package/lib/types/utils/tokens.d.ts +11 -0
- package/lib/types/utils/tokens.d.ts.map +1 -0
- package/lib/types/vars.d.ts +3 -0
- package/lib/types/vars.d.ts.map +1 -0
- package/lib/types/workflows.d.ts +9 -0
- package/lib/types/workflows.d.ts.map +1 -0
- package/lib/workflows-bundle.js +18394 -0
- package/package.json +109 -0
- package/src/activities/advanced/createDocumentTypeFromInteractionRun.ts +54 -0
- package/src/activities/advanced/createOrUpdateDocumentFromInteractionRun.ts +97 -0
- package/src/activities/advanced/updateDocumentFromInteractionRun.ts +34 -0
- package/src/activities/chunkDocument.ts +124 -0
- package/src/activities/createDocumentFromOther.ts +92 -0
- package/src/activities/executeInteraction.ts +191 -0
- package/src/activities/extractDocumentText.ts +174 -0
- package/src/activities/generateDocumentProperties.ts +93 -0
- package/src/activities/generateEmbeddings.ts +345 -0
- package/src/activities/generateImageRendition.ts +134 -0
- package/src/activities/generateOrAssignContentType.ts +152 -0
- package/src/activities/getObjectFromStore.ts +31 -0
- package/src/activities/index.ts +21 -0
- package/src/activities/media/processPdfWithTextract.ts +141 -0
- package/src/activities/media/transcribeMediaWithGladia.ts +83 -0
- package/src/activities/notifyWebhook.test.ts +32 -0
- package/src/activities/notifyWebhook.ts +51 -0
- package/src/activities/setDocumentStatus.ts +25 -0
- package/src/conversion/TextractProcessor.ts +505 -0
- package/src/conversion/image.test.ts +26 -0
- package/src/conversion/image.ts +22 -0
- package/src/conversion/mutool.test.ts +74 -0
- package/src/conversion/mutool.ts +180 -0
- package/src/conversion/pandoc.test.ts +22 -0
- package/src/conversion/pandoc.ts +44 -0
- package/src/conversion/pdf.test.ts +35 -0
- package/src/conversion/pdf.ts +8 -0
- package/src/dsl/conditions.ts +76 -0
- package/src/dsl/dsl-workflow.test.ts +58 -0
- package/src/dsl/dsl-workflow.ts +235 -0
- package/src/dsl/ms.d.ts +11 -0
- package/src/dsl/projections.test.ts +159 -0
- package/src/dsl/projections.ts +72 -0
- package/src/dsl/setup/ActivityContext.ts +106 -0
- package/src/dsl/setup/fetch/DataProvider.ts +45 -0
- package/src/dsl/setup/fetch/index.ts +19 -0
- package/src/dsl/setup/fetch/providers.ts +67 -0
- package/src/dsl/test/test-child-workflow.ts +6 -0
- package/src/dsl/validation.test.ts +257 -0
- package/src/dsl/validation.ts +125 -0
- package/src/dsl/vars.test.ts +245 -0
- package/src/dsl/vars.ts +340 -0
- package/src/dsl/walk.test.ts +81 -0
- package/src/dsl/walk.ts +103 -0
- package/src/dsl/workflow-exec-child.test.ts +182 -0
- package/src/dsl/workflow-fetch.test.ts +135 -0
- package/src/dsl/workflow-import.test.ts +89 -0
- package/src/dsl/workflow.test.ts +110 -0
- package/src/errors.ts +24 -0
- package/src/index.ts +27 -0
- package/src/iterative-generation/activities/extractToc.ts +49 -0
- package/src/iterative-generation/activities/finalizeOutput.ts +77 -0
- package/src/iterative-generation/activities/generatePart.ts +82 -0
- package/src/iterative-generation/activities/generateToc.ts +98 -0
- package/src/iterative-generation/activities/index.ts +4 -0
- package/src/iterative-generation/iterativeGenerationWorkflow.ts +67 -0
- package/src/iterative-generation/types.ts +99 -0
- package/src/iterative-generation/utils.ts +123 -0
- package/src/result-types.ts +25 -0
- package/src/system/generateObjectText.ts +109 -0
- package/src/system/notifyWebhookWorkflow.ts +64 -0
- package/src/system/recalculateEmbeddingsWorkflow.ts +46 -0
- package/src/utils/auth.ts +10 -0
- package/src/utils/blobs.ts +58 -0
- package/src/utils/client.ts +31 -0
- package/src/utils/expand-vars.ts +31 -0
- package/src/utils/memory.ts +66 -0
- package/src/utils/tokens.ts +44 -0
- package/src/vars.ts +3 -0
- package/src/workflows.ts +9 -0
@@ -0,0 +1,98 @@
|
|
1
|
+
/**
|
2
|
+
* Use textract to convert a pdf into a data structure of the following format:
|
3
|
+
* <document>
|
4
|
+
* <page number="n">
|
5
|
+
* <text/>
|
6
|
+
* <table/>
|
7
|
+
* <text/>
|
8
|
+
* <figure/>
|
9
|
+
* ...
|
10
|
+
* </page>
|
11
|
+
*/
|
12
|
+
import { fromWebToken } from "@aws-sdk/credential-providers";
|
13
|
+
import { SupportedIntegrations } from "@vertesia/common";
|
14
|
+
import { log } from "@temporalio/activity";
|
15
|
+
import { TextractProcessor } from "../../conversion/TextractProcessor.js";
|
16
|
+
import { setupActivity } from "../../dsl/setup/ActivityContext.js";
|
17
|
+
import { NoDocumentFound } from "../../errors.js";
|
18
|
+
import { TextExtractionStatus } from "../../result-types.js";
|
19
|
+
import { fetchBlobAsBuffer, md5 } from "../../utils/blobs.js";
|
20
|
+
import { countTokens } from "../../utils/tokens.js";
|
21
|
+
export async function convertPdfToStructuredText(payload) {
|
22
|
+
const { params, client, objectId } = await setupActivity(payload);
|
23
|
+
const object = await client.objects.retrieve(objectId, "+text");
|
24
|
+
if (object.text && !params.force) {
|
25
|
+
return { hasText: true, objectId, status: TextExtractionStatus.skipped, message: "text already present and force not enabled" };
|
26
|
+
}
|
27
|
+
if (!object.content?.source) {
|
28
|
+
throw new NoDocumentFound(`No source found for object ${objectId}`);
|
29
|
+
}
|
30
|
+
const pdfUrl = await client.store.objects.getContentSource(objectId).then(res => res.source);
|
31
|
+
if (!pdfUrl) {
|
32
|
+
throw new NoDocumentFound(`Error fetching source ${object.content.source}`);
|
33
|
+
}
|
34
|
+
const awsConfig = (await client.projects.integrations.retrieve(client.project, SupportedIntegrations.aws));
|
35
|
+
const credentials = await getS3AWSCredentials(awsConfig, payload.auth_token, client.project);
|
36
|
+
const processor = new TextractProcessor({
|
37
|
+
fileKey: objectId,
|
38
|
+
region: "us-west-2",
|
39
|
+
bucket: "cp-textract-tests",
|
40
|
+
credentials,
|
41
|
+
log: log,
|
42
|
+
detectImages: true,
|
43
|
+
includeConfidenceInTables: true,
|
44
|
+
});
|
45
|
+
try {
|
46
|
+
if (!object.content.source.startsWith("s3://")) {
|
47
|
+
const buf = await fetchBlobAsBuffer(client, object.content.source);
|
48
|
+
await processor.upload(buf);
|
49
|
+
}
|
50
|
+
const jobId = await processor.startAnalysis(objectId);
|
51
|
+
let jobStatus = await processor.checkJobStatus(jobId);
|
52
|
+
while (jobStatus === "IN_PROGRESS") {
|
53
|
+
await new Promise(resolve => setTimeout(resolve, 5000));
|
54
|
+
jobStatus = await processor.checkJobStatus(jobId);
|
55
|
+
}
|
56
|
+
if (jobStatus === "SUCCEEDED") {
|
57
|
+
log.info(`Job ${jobId} succeeded, saving results`, { jobId });
|
58
|
+
const ftext = await processor.processResults(jobId);
|
59
|
+
const tokensData = countTokens(ftext);
|
60
|
+
const etag = object.content.etag ?? md5(ftext);
|
61
|
+
const updateData = {
|
62
|
+
text: ftext,
|
63
|
+
text_etag: etag,
|
64
|
+
tokens: {
|
65
|
+
...tokensData,
|
66
|
+
etag: etag,
|
67
|
+
}
|
68
|
+
};
|
69
|
+
await client.objects.update(objectId, updateData);
|
70
|
+
console.log("Full text updated");
|
71
|
+
return { hasText: true, objectId, status: TextExtractionStatus.success, message: "Text extracted successfully" };
|
72
|
+
}
|
73
|
+
else {
|
74
|
+
throw new Error(`Job failed with status: ${jobStatus}`);
|
75
|
+
}
|
76
|
+
}
|
77
|
+
catch (error) {
|
78
|
+
console.error("Error processing document:", error);
|
79
|
+
throw error;
|
80
|
+
}
|
81
|
+
}
|
82
|
+
export async function getS3AWSCredentials(awsConfig, composableAuthToken, projectId) {
|
83
|
+
// fetch s3 role ARN
|
84
|
+
if (!awsConfig || !awsConfig.enabled) {
|
85
|
+
throw new NoDocumentFound("AWS integration is not enabled for this project");
|
86
|
+
}
|
87
|
+
if (!awsConfig.s3_role_arn) {
|
88
|
+
throw new NoDocumentFound("S3 Role ARN is not defined in AWS project integration");
|
89
|
+
}
|
90
|
+
log.info("Getting AWS credentials for Textract", { projectId, composableAuthToken, roleArn: awsConfig.s3_role_arn });
|
91
|
+
const credentials = fromWebToken({
|
92
|
+
webIdentityToken: composableAuthToken,
|
93
|
+
roleArn: awsConfig.s3_role_arn,
|
94
|
+
roleSessionName: `cp-project-textract-${projectId}`,
|
95
|
+
});
|
96
|
+
return credentials;
|
97
|
+
}
|
98
|
+
//# sourceMappingURL=processPdfWithTextract.js.map
|
@@ -0,0 +1 @@
|
|
1
|
+
{"version":3,"file":"processPdfWithTextract.js","sourceRoot":"","sources":["../../../../src/activities/media/processPdfWithTextract.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;GAUG;AAEH,OAAO,EAAE,YAAY,EAAE,MAAM,+BAA+B,CAAC;AAC7D,OAAO,EAA8F,qBAAqB,EAAE,MAAM,kBAAkB,CAAC;AAErJ,OAAO,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAC;AAC3C,OAAO,EAAE,iBAAiB,EAAE,MAAM,uCAAuC,CAAC;AAC1E,OAAO,EAAE,aAAa,EAAE,MAAM,oCAAoC,CAAC;AACnE,OAAO,EAAE,eAAe,EAAE,MAAM,iBAAiB,CAAC;AAClD,OAAO,EAAwB,oBAAoB,EAAE,MAAM,uBAAuB,CAAC;AACnF,OAAO,EAAE,iBAAiB,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAC;AAC9D,OAAO,EAAE,WAAW,EAAE,MAAM,uBAAuB,CAAC;AAkBpD,MAAM,CAAC,KAAK,UAAU,0BAA0B,CAAC,OAAoC;IAGjF,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,GAAG,MAAM,aAAa,CAAmC,OAAO,CAAC,CAAC;IAEpG,MAAM,MAAM,GAAG,MAAM,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC;IAEhE,IAAI,MAAM,CAAC,IAAI,IAAI,CAAC,MAAM,CAAC,KAAK,EAAE,CAAC;QAC/B,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,QAAQ,EAAE,MAAM,EAAE,oBAAoB,CAAC,OAAO,EAAE,OAAO,EAAE,4CAA4C,EAAE,CAAA;IACnI,CAAC;IAED,IAAI,CAAC,MAAM,CAAC,OAAO,EAAE,MAAM,EAAE,CAAC;QAC1B,MAAM,IAAI,eAAe,CAAC,8BAA8B,QAAQ,EAAE,CAAC,CAAC;IACxE,CAAC;IAED,MAAM,MAAM,GAAG,MAAM,MAAM,CAAC,KAAK,CAAC,OAAO,CAAC,gBAAgB,CAAC,QAAQ,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;IAE7F,IAAI,CAAC,MAAM,EAAE,CAAC;QACV,MAAM,IAAI,eAAe,CAAC,yBAAyB,MAAM,CAAC,OAAO,CAAC,MAAM,EAAE,CAAC,CAAC;IAChF,CAAC;IAGD,MAAM,SAAS,GAAG,CAAC,MAAM,MAAM,CAAC,QAAQ,CAAC,YAAY,CAAC,QAAQ,CAAC,MAAM,CAAC,OAAQ,EAAE,qBAAqB,CAAC,GAAG,CAAC,CAAqB,CAAC;IAChI,MAAM,WAAW,GAAG,MAAM,mBAAmB,CAAC,SAAS,EAAE,OAAO,CAAC,UAAU,EAAE,MAAM,CAAC,OAAQ,CAAC,CAAC;IAE9F,MAAM,SAAS,GAAG,IAAI,iBAAiB,CAAC;QACpC,OAAO,EAAE,QAAQ;QACjB,MAAM,EAAE,WAAW;QACnB,MAAM,EAAE,mBAAmB;QAC3B,WAAW;QACX,GAAG,EAAE,GAAG;QACR,YAAY,EAAE,IAAI;QAClB,yBAAyB,EAAE,IAAI;KAClC,CAAC,CAAC;IAIH,IAAI,CAAC;QAED,IAAI,CAAC,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,UAAU,CAAC,OAAO,CAAC,EAAE,CAAC;YAC7C,MAAM,GAAG,GAAG,MAAM,iBAAiB,CAAC,MAAM,EAAE,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;YACnE,MAAM,SAAS,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;QAChC,CAAC;QAED,MAAM,KAAK,GAAG,MAAM,SAAS,CAAC,aAAa,CAAC,QAAQ,CAAC,CAAC;QAEtD,IAAI,SAAS,GAAG,MAAM,SAAS,CAAC,cAAc,CAAC,KAAK,CAAC,CAAC;QACtD,OAAO,SAAS,KAAK,aAAa,EAAE,CAAC;YACjC,MAAM,IAAI,OAAO,CAAC,OAAO,CAAC,EAAE,CAAC,UAAU,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC,CAAC;YACxD,SAAS,GAAG,MAAM,SAAS,CAAC,cAAc,CAAC,KAAK,CAAC,CAAC;QACtD,CAAC;QAED,IAAI,SAAS,KAAK,WAAW,EAAE,CAAC;YAC5B,GAAG,CAAC,IAAI,CAAC,OAAO,KAAK,4BAA4B,EAAE,EAAE,KAAK,EAAE,CAAC,CAAC;YAC9D,MAAM,KAAK,GAAG,MAAM,SAAS,CAAC,cAAc,CAAC,KAAK,CAAC,CAAC;YACpD,MAAM,UAAU,GAAG,WAAW,CAAC,KAAK,CAAC,CAAC;YACtC,MAAM,IAAI,GAAG,MAAM,CAAC,OAAO,CAAC,IAAI,IAAI,GAAG,CAAC,KAAK,CAAC,CAAC;YAC/C,MAAM,UAAU,GAA+B;gBAC3C,IAAI,EAAE,KAAK;gBACX,SAAS,EAAE,IAAI;gBACf,MAAM,EAAE;oBACJ,GAAG,UAAU;oBACb,IAAI,EAAE,IAAI;iBACb;aACJ,CAAA;YAED,MAAM,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,QAAQ,EAAE,UAAU,CAAC,CAAC;YAClD,OAAO,CAAC,GAAG,CAAC,mBAAmB,CAAC,CAAC;YAEjC,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,QAAQ,EAAE,MAAM,EAAE,oBAAoB,CAAC,OAAO,EAAE,OAAO,EAAE,6BAA6B,EAAE,CAAA;QAEpH,CAAC;aAAM,CAAC;YACJ,MAAM,IAAI,KAAK,CAAC,2BAA2B,SAAS,EAAE,CAAC,CAAC;QAC5D,CAAC;IACL,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACb,OAAO,CAAC,KAAK,CAAC,4BAA4B,EAAE,KAAK,CAAC,CAAC;QACnD,MAAM,KAAK,CAAC;IAChB,CAAC;AAGL,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,mBAAmB,CAAC,SAA2B,EAAE,mBAA2B,EAAE,SAAiB;IAEjH,oBAAoB;IACpB,IAAI,CAAC,SAAS,IAAI,CAAC,SAAS,CAAC,OAAO,EAAE,CAAC;QACnC,MAAM,IAAI,eAAe,CAAC,iDAAiD,CAAC,CAAC;IACjF,CAAC;IACD,IAAI,CAAC,SAAS,CAAC,WAAW,EAAE,CAAC;QACzB,MAAM,IAAI,eAAe,CAAC,uDAAuD,CAAC,CAAC;IACvF,CAAC;IAED,GAAG,CAAC,IAAI,CAAC,sCAAsC,EAAE,EAAE,SAAS,EAAE,mBAAmB,EAAE,OAAO,EAAE,SAAS,CAAC,WAAW,EAAE,CAAC,CAAC;IAErH,MAAM,WAAW,GAAG,YAAY,CAAC;QAC7B,gBAAgB,EAAE,mBAAmB;QACrC,OAAO,EAAE,SAAS,CAAC,WAAW;QAC9B,eAAe,EAAE,uBAAuB,SAAS,EAAE;KACtD,CAAC,CAAC;IAEH,OAAO,WAAW,CAAC;AACvB,CAAC"}
|
@@ -0,0 +1,48 @@
|
|
1
|
+
import { SupportedIntegrations } from "@vertesia/common";
|
2
|
+
import { activityInfo, CompleteAsyncError, log } from "@temporalio/activity";
|
3
|
+
import { FetchClient } from "api-fetch-client";
|
4
|
+
import { setupActivity } from "../../dsl/setup/ActivityContext.js";
|
5
|
+
import { NoDocumentFound } from "../../errors.js";
|
6
|
+
import { TextExtractionStatus } from "../../index.js";
|
7
|
+
const GLADIA_URL = "https://api.gladia.io/v2";
|
8
|
+
export async function transcribeMedia(payload) {
|
9
|
+
const { params, client, objectId } = await setupActivity(payload);
|
10
|
+
const gladiaConfig = await client.projects.integrations.retrieve(payload.project_id, SupportedIntegrations.gladia);
|
11
|
+
if (!gladiaConfig || !gladiaConfig.enabled) {
|
12
|
+
throw new NoDocumentFound("Gladia integration not enabled");
|
13
|
+
}
|
14
|
+
const object = await client.objects.retrieve(objectId, "+text");
|
15
|
+
const gladiaClient = new FetchClient(gladiaConfig.url ?? GLADIA_URL);
|
16
|
+
gladiaClient.withHeaders({ "x-gladia-key": gladiaConfig.api_key });
|
17
|
+
if (object.text && !params.force) {
|
18
|
+
return { hasText: true, objectId, status: TextExtractionStatus.skipped, message: "text already present and force not enabled" };
|
19
|
+
}
|
20
|
+
if (!object.content?.source) {
|
21
|
+
throw new NoDocumentFound(`No source found for object ${objectId}`);
|
22
|
+
}
|
23
|
+
const mediaUrl = await client.store.objects.getContentSource(objectId).then(res => res.source);
|
24
|
+
if (!mediaUrl) {
|
25
|
+
throw new NoDocumentFound(`Error fetching source ${object.content.source}`);
|
26
|
+
}
|
27
|
+
const taskToken = Buffer.from(activityInfo().taskToken).toString('base64url');
|
28
|
+
const callbackUrl = generateCallbackUrlForGladia(client.store.baseUrl, payload.auth_token, taskToken, objectId);
|
29
|
+
log.info(`Transcribing media ${mediaUrl} with Gladia`, { objectId, callbackUrl });
|
30
|
+
const res = await gladiaClient.post("/transcription", {
|
31
|
+
payload: {
|
32
|
+
audio_url: mediaUrl,
|
33
|
+
callback_url: callbackUrl,
|
34
|
+
diarization_enhanced: true,
|
35
|
+
enable_code_switching: true,
|
36
|
+
subtitles: true,
|
37
|
+
subtitles_config: {
|
38
|
+
formats: ["vtt"],
|
39
|
+
}
|
40
|
+
}
|
41
|
+
});
|
42
|
+
log.info(`Transcription request sent to Gladia`, { objectId, res });
|
43
|
+
throw new CompleteAsyncError();
|
44
|
+
}
|
45
|
+
function generateCallbackUrlForGladia(baseUrl, authToken, taskToken, objectId) {
|
46
|
+
return `${baseUrl}/api/v1/webhooks/gladia/${objectId}?auth_token=${authToken}&task_token=${taskToken}`;
|
47
|
+
}
|
48
|
+
//# sourceMappingURL=transcribeMediaWithGladia.js.map
|
@@ -0,0 +1 @@
|
|
1
|
+
{"version":3,"file":"transcribeMediaWithGladia.js","sourceRoot":"","sources":["../../../../src/activities/media/transcribeMediaWithGladia.ts"],"names":[],"mappings":"AAAA,OAAO,EAAqE,qBAAqB,EAAE,MAAM,kBAAkB,CAAC;AAC5H,OAAO,EAAE,YAAY,EAAE,kBAAkB,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAC;AAC7E,OAAO,EAAE,WAAW,EAAE,MAAM,kBAAkB,CAAC;AAC/C,OAAO,EAAE,aAAa,EAAE,MAAM,oCAAoC,CAAC;AACnE,OAAO,EAAE,eAAe,EAAE,MAAM,iBAAiB,CAAC;AAClD,OAAO,EAAwB,oBAAoB,EAAE,MAAM,gBAAgB,CAAC;AAgB5E,MAAM,UAAU,GAAG,0BAA0B,CAAC;AAE9C,MAAM,CAAC,KAAK,UAAU,eAAe,CAAC,OAAoC;IAEtE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,GAAG,MAAM,aAAa,CAAwB,OAAO,CAAC,CAAC;IAEzF,MAAM,YAAY,GAAG,MAAM,MAAM,CAAC,QAAQ,CAAC,YAAY,CAAC,QAAQ,CAAC,OAAO,CAAC,UAAU,EAAE,qBAAqB,CAAC,MAAM,CAAoC,CAAC;IACtJ,IAAI,CAAC,YAAY,IAAI,CAAC,YAAY,CAAC,OAAO,EAAE,CAAC;QACzC,MAAM,IAAI,eAAe,CAAC,gCAAgC,CAAC,CAAC;IAChE,CAAC;IAED,MAAM,MAAM,GAAG,MAAM,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC;IAChE,MAAM,YAAY,GAAG,IAAI,WAAW,CAAC,YAAY,CAAC,GAAG,IAAI,UAAU,CAAC,CAAC;IACrE,YAAY,CAAC,WAAW,CAAC,EAAE,cAAc,EAAE,YAAY,CAAC,OAAO,EAAE,CAAC,CAAC;IAEnE,IAAI,MAAM,CAAC,IAAI,IAAI,CAAC,MAAM,CAAC,KAAK,EAAE,CAAC;QAC/B,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,QAAQ,EAAE,MAAM,EAAE,oBAAoB,CAAC,OAAO,EAAE,OAAO,EAAE,4CAA4C,EAAE,CAAA;IACnI,CAAC;IAED,IAAI,CAAC,MAAM,CAAC,OAAO,EAAE,MAAM,EAAE,CAAC;QAC1B,MAAM,IAAI,eAAe,CAAC,8BAA8B,QAAQ,EAAE,CAAC,CAAC;IACxE,CAAC;IAED,MAAM,QAAQ,GAAG,MAAM,MAAM,CAAC,KAAK,CAAC,OAAO,CAAC,gBAAgB,CAAC,QAAQ,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;IAE/F,IAAI,CAAC,QAAQ,EAAE,CAAC;QACZ,MAAM,IAAI,eAAe,CAAC,yBAAyB,MAAM,CAAC,OAAO,CAAC,MAAM,EAAE,CAAC,CAAC;IAChF,CAAC;IAED,MAAM,SAAS,GAAG,MAAM,CAAC,IAAI,CAAC,YAAY,EAAE,CAAC,SAAS,CAAC,CAAC,QAAQ,CAAC,WAAW,CAAC,CAAC;IAC9E,MAAM,WAAW,GAAG,4BAA4B,CAAC,MAAM,CAAC,KAAK,CAAC,OAAO,EAAE,OAAO,CAAC,UAAU,EAAE,SAAS,EAAE,QAAQ,CAAC,CAAC;IAEhH,GAAG,CAAC,IAAI,CAAC,sBAAsB,QAAQ,cAAc,EAAE,EAAE,QAAQ,EAAE,WAAW,EAAE,CAAC,CAAC;IAElF,MAAM,GAAG,GAAG,MAAM,YAAY,CAAC,IAAI,CAAC,gBAAgB,EAAE;QAClD,OAAO,EAAE;YACL,SAAS,EAAE,QAAQ;YACnB,YAAY,EAAE,WAAW;YACzB,oBAAoB,EAAE,IAAI;YAC1B,qBAAqB,EAAE,IAAI;YAC3B,SAAS,EAAE,IAAI;YACf,gBAAgB,EAAE;gBACd,OAAO,EAAE,CAAC,KAAK,CAAC;aACnB;SACJ;KACJ,CAAoC,CAAC;IAEtC,GAAG,CAAC,IAAI,CAAC,sCAAsC,EAAE,EAAE,QAAQ,EAAE,GAAG,EAAE,CAAC,CAAC;IAEpE,MAAM,IAAI,kBAAkB,EAAE,CAAC;AAEnC,CAAC;AAGD,SAAS,4BAA4B,CAAC,OAAe,EAAE,SAAiB,EAAE,SAAiB,EAAE,QAAgB;IACzG,OAAO,GAAG,OAAO,2BAA2B,QAAQ,eAAe,SAAS,eAAe,SAAS,EAAE,CAAC;AAC3G,CAAC"}
|
@@ -0,0 +1,31 @@
|
|
1
|
+
import { log } from "@temporalio/activity";
|
2
|
+
import { setupActivity } from "../dsl/setup/ActivityContext.js";
|
3
|
+
import { WorkflowParamNotFound } from "../errors.js";
|
4
|
+
export async function notifyWebhook(payload) {
|
5
|
+
const { params } = await setupActivity(payload);
|
6
|
+
const { target_url, method, payload: requestPayload, headers } = params;
|
7
|
+
if (!target_url)
|
8
|
+
throw new WorkflowParamNotFound('target_url');
|
9
|
+
const body = method === 'POST' ? JSON.stringify({
|
10
|
+
...requestPayload,
|
11
|
+
...params
|
12
|
+
}) : undefined;
|
13
|
+
log.info(`Notifying webhook at ${target_url}`);
|
14
|
+
const res = await fetch(target_url, {
|
15
|
+
method,
|
16
|
+
body,
|
17
|
+
headers: {
|
18
|
+
'Content-Type': 'application/json',
|
19
|
+
...headers
|
20
|
+
},
|
21
|
+
}).catch(err => {
|
22
|
+
log.warn(`Failed to notify webhook ${target_url}: ${err}`);
|
23
|
+
throw new Error(`Failed to notify webhook ${target_url}: ${err}`);
|
24
|
+
});
|
25
|
+
if (!res.ok) {
|
26
|
+
log.warn(`Failed to notify webhook ${target_url} - ${res.status}: ${res.statusText}`, { res });
|
27
|
+
throw new Error(`Failed to notify webhook ${target_url}: ${res.statusText}`);
|
28
|
+
}
|
29
|
+
return { status: res.status, message: res.statusText, url: res.url };
|
30
|
+
}
|
31
|
+
//# sourceMappingURL=notifyWebhook.js.map
|
@@ -0,0 +1 @@
|
|
1
|
+
{"version":3,"file":"notifyWebhook.js","sourceRoot":"","sources":["../../../src/activities/notifyWebhook.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAC;AAC3C,OAAO,EAAE,aAAa,EAAE,MAAM,iCAAiC,CAAC;AAChE,OAAO,EAAE,qBAAqB,EAAE,MAAM,cAAc,CAAC;AAerD,MAAM,CAAC,KAAK,UAAU,aAAa,CAAC,OAAoC;IAEpE,MAAM,EAAE,MAAM,EAAE,GAAG,MAAM,aAAa,CAAsB,OAAO,CAAC,CAAC;IACrE,MAAM,EAAE,UAAU,EAAE,MAAM,EAAE,OAAO,EAAE,cAAc,EAAE,OAAO,EAAE,GAAG,MAAM,CAAA;IAEvE,IAAI,CAAC,UAAU;QAAE,MAAM,IAAI,qBAAqB,CAAC,YAAY,CAAC,CAAC;IAE/D,MAAM,IAAI,GAAG,MAAM,KAAK,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC;QAC5C,GAAG,cAAc;QACjB,GAAG,MAAM;KACZ,CAAC,CAAC,CAAC,CAAC,SAAS,CAAA;IAEd,GAAG,CAAC,IAAI,CAAC,wBAAwB,UAAU,EAAE,CAAC,CAAC;IAC/C,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,UAAU,EAAE;QAChC,MAAM;QACN,IAAI;QACJ,OAAO,EAAE;YACL,cAAc,EAAE,kBAAkB;YAClC,GAAG,OAAO;SACb;KACJ,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE;QACX,GAAG,CAAC,IAAI,CAAC,4BAA4B,UAAU,KAAK,GAAG,EAAE,CAAC,CAAC;QAC3D,MAAM,IAAI,KAAK,CAAC,4BAA4B,UAAU,KAAK,GAAG,EAAE,CAAC,CAAC;IACtE,CAAC,CAAC,CAAC;IAEH,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC;QACV,GAAG,CAAC,IAAI,CAAC,4BAA4B,UAAU,MAAM,GAAG,CAAC,MAAM,KAAK,GAAG,CAAC,UAAU,EAAE,EAAE,EAAE,GAAG,EAAE,CAAC,CAAC;QAC/F,MAAM,IAAI,KAAK,CAAC,4BAA4B,UAAU,KAAK,GAAG,CAAC,UAAU,EAAE,CAAC,CAAC;IACjF,CAAC;IAED,OAAO,EAAC,MAAM,EAAE,GAAG,CAAC,MAAM,EAAE,OAAO,EAAE,GAAG,CAAC,UAAU,EAAE,GAAG,EAAE,GAAG,CAAC,GAAG,EAAE,CAAA;AAEvE,CAAC"}
|
@@ -0,0 +1,12 @@
|
|
1
|
+
import { setupActivity } from "../dsl/setup/ActivityContext.js";
|
2
|
+
/**
|
3
|
+
* We are using a union type for the status parameter since typescript enumbs breaks the workflow code generation
|
4
|
+
* @param objectId
|
5
|
+
* @param status
|
6
|
+
*/
|
7
|
+
export async function setDocumentStatus(payload) {
|
8
|
+
const { client, params, objectId } = await setupActivity(payload);
|
9
|
+
const res = await client.objects.update(objectId, { status: params.status });
|
10
|
+
return res.status;
|
11
|
+
}
|
12
|
+
//# sourceMappingURL=setDocumentStatus.js.map
|
@@ -0,0 +1 @@
|
|
1
|
+
{"version":3,"file":"setDocumentStatus.js","sourceRoot":"","sources":["../../../src/activities/setDocumentStatus.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,aAAa,EAAE,MAAM,iCAAiC,CAAC;AAWhE;;;;GAIG;AACH,MAAM,CAAC,KAAK,UAAU,iBAAiB,CAAC,OAAoC;IACxE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,GAAG,MAAM,aAAa,CAA0B,OAAO,CAAC,CAAC;IAE3F,MAAM,GAAG,GAAG,MAAM,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,QAAQ,EAAE,EAAE,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,CAAC,CAAC;IAE7E,OAAO,GAAG,CAAC,MAAM,CAAC;AAEtB,CAAC"}
|
@@ -0,0 +1,409 @@
|
|
1
|
+
import { PutObjectCommand, S3Client } from "@aws-sdk/client-s3";
|
2
|
+
import { GetDocumentAnalysisCommand, StartDocumentAnalysisCommand, TextractClient } from "@aws-sdk/client-textract";
|
3
|
+
import Papa from 'papaparse';
|
4
|
+
export class TextractProcessor {
|
5
|
+
textractClient;
|
6
|
+
s3Client;
|
7
|
+
fileKey;
|
8
|
+
bucket;
|
9
|
+
log;
|
10
|
+
detectImages;
|
11
|
+
/**
|
12
|
+
* Whether or not to include confidence values in CSV output for tables.
|
13
|
+
*/
|
14
|
+
includeConfidenceInTables;
|
15
|
+
constructor({ fileKey, region, bucket, credentials, log, detectImages = false, includeConfidenceInTables = false // NEW default = false
|
16
|
+
}) {
|
17
|
+
this.fileKey = fileKey;
|
18
|
+
this.bucket = bucket;
|
19
|
+
this.log = log;
|
20
|
+
this.detectImages = detectImages;
|
21
|
+
this.includeConfidenceInTables = includeConfidenceInTables;
|
22
|
+
this.textractClient = new TextractClient({
|
23
|
+
region,
|
24
|
+
credentials
|
25
|
+
});
|
26
|
+
this.s3Client = new S3Client({
|
27
|
+
region,
|
28
|
+
credentials
|
29
|
+
});
|
30
|
+
}
|
31
|
+
getText(result, blocksMap) {
|
32
|
+
let text = '';
|
33
|
+
if (result.Relationships) {
|
34
|
+
for (const relationship of result.Relationships) {
|
35
|
+
if (relationship.Type === 'CHILD') {
|
36
|
+
for (const childId of relationship.Ids || []) {
|
37
|
+
const word = blocksMap[childId];
|
38
|
+
if (word.BlockType === 'WORD') {
|
39
|
+
const wordText = word.Text || '';
|
40
|
+
// Example logic to quote numeric text with commas
|
41
|
+
if (wordText.includes(',') &&
|
42
|
+
wordText.replace(',', '').match(/^\d+$/)) {
|
43
|
+
text += `"${wordText}" `;
|
44
|
+
}
|
45
|
+
else {
|
46
|
+
text += `${wordText} `;
|
47
|
+
}
|
48
|
+
}
|
49
|
+
if (word.BlockType === 'SELECTION_ELEMENT' &&
|
50
|
+
word.SelectionStatus === 'SELECTED') {
|
51
|
+
text += 'X ';
|
52
|
+
}
|
53
|
+
}
|
54
|
+
}
|
55
|
+
}
|
56
|
+
}
|
57
|
+
return text.trim();
|
58
|
+
}
|
59
|
+
isBlockInTable(block, blocksMap) {
|
60
|
+
if (block.BlockType !== 'LINE') {
|
61
|
+
return false;
|
62
|
+
}
|
63
|
+
if (block.Relationships) {
|
64
|
+
for (const relationship of block.Relationships) {
|
65
|
+
if (relationship.Type === 'CHILD') {
|
66
|
+
for (const childId of relationship.Ids || []) {
|
67
|
+
const wordBlock = blocksMap[childId];
|
68
|
+
if (this.isWordInTableCell(wordBlock, blocksMap)) {
|
69
|
+
return true;
|
70
|
+
}
|
71
|
+
}
|
72
|
+
}
|
73
|
+
}
|
74
|
+
}
|
75
|
+
return false;
|
76
|
+
}
|
77
|
+
isWordInTableCell(wordBlock, blocksMap) {
|
78
|
+
// Check if the wordBlock is a descendant of any TABLE->CELL block
|
79
|
+
for (const blockId in blocksMap) {
|
80
|
+
const potentialTable = blocksMap[blockId];
|
81
|
+
if (potentialTable.BlockType === 'TABLE' && potentialTable.Relationships) {
|
82
|
+
for (const relationship of potentialTable.Relationships) {
|
83
|
+
if (relationship.Type === 'CHILD') {
|
84
|
+
for (const cellId of relationship.Ids || []) {
|
85
|
+
const cell = blocksMap[cellId];
|
86
|
+
if (cell.BlockType === 'CELL' && cell.Relationships) {
|
87
|
+
for (const cellRel of cell.Relationships) {
|
88
|
+
if (cellRel.Type === 'CHILD' &&
|
89
|
+
cellRel.Ids?.includes(wordBlock.Id)) {
|
90
|
+
return true;
|
91
|
+
}
|
92
|
+
}
|
93
|
+
}
|
94
|
+
}
|
95
|
+
}
|
96
|
+
}
|
97
|
+
}
|
98
|
+
}
|
99
|
+
return false;
|
100
|
+
}
|
101
|
+
/**
|
102
|
+
* NEW: Helper type to store row and column text along with confidence.
|
103
|
+
*/
|
104
|
+
getRowsColumnsMap(tableResult, blocksMap) {
|
105
|
+
const rows = [];
|
106
|
+
tableResult.Relationships?.forEach(relationship => {
|
107
|
+
if (relationship.Type === 'CHILD') {
|
108
|
+
relationship.Ids?.forEach(childId => {
|
109
|
+
const cell = blocksMap[childId];
|
110
|
+
if (cell.BlockType === 'CELL') {
|
111
|
+
const rowIndex = cell.RowIndex || 1;
|
112
|
+
const colIndex = cell.ColumnIndex || 1;
|
113
|
+
// Expand the array if needed
|
114
|
+
if (!rows[rowIndex - 1]) {
|
115
|
+
rows[rowIndex - 1] = [];
|
116
|
+
}
|
117
|
+
// Prepare cell text and confidence
|
118
|
+
const text = this.getText(cell, blocksMap);
|
119
|
+
const confidence = cell.Confidence || 0;
|
120
|
+
// If there's a gap, fill it with placeholders
|
121
|
+
// so that we can safely place text at colIndex - 1
|
122
|
+
for (let i = rows[rowIndex - 1].length; i < colIndex - 1; i++) {
|
123
|
+
rows[rowIndex - 1].push({ text: '', confidence: 0 });
|
124
|
+
}
|
125
|
+
rows[rowIndex - 1][colIndex - 1] = { text, confidence };
|
126
|
+
}
|
127
|
+
});
|
128
|
+
}
|
129
|
+
});
|
130
|
+
return { rows };
|
131
|
+
}
|
132
|
+
generateTableCSV(tableResult, blocksMap, _tableIndex, _pageNumber) {
|
133
|
+
const { rows } = this.getRowsColumnsMap(tableResult, blocksMap);
|
134
|
+
let totalConfidence = 0;
|
135
|
+
let cellCount = 0;
|
136
|
+
// Prepare CSV data
|
137
|
+
const csvData = [];
|
138
|
+
for (const row of rows) {
|
139
|
+
const rowData = [];
|
140
|
+
for (const cell of row) {
|
141
|
+
// Add to CSV
|
142
|
+
rowData.push(cell.text.trim());
|
143
|
+
// Accumulate confidence
|
144
|
+
totalConfidence += cell.confidence;
|
145
|
+
cellCount++;
|
146
|
+
}
|
147
|
+
csvData.push(rowData);
|
148
|
+
}
|
149
|
+
// Compute average confidence (or any other method you prefer)
|
150
|
+
const tableConfidence = cellCount > 0 ? (totalConfidence / cellCount) : 0;
|
151
|
+
// Convert to CSV
|
152
|
+
const csv = Papa.unparse(csvData, {
|
153
|
+
delimiter: ',',
|
154
|
+
quotes: true,
|
155
|
+
quoteChar: '"',
|
156
|
+
escapeChar: '"',
|
157
|
+
header: false,
|
158
|
+
newline: '\n',
|
159
|
+
skipEmptyLines: false
|
160
|
+
});
|
161
|
+
return { csv, tableConfidence };
|
162
|
+
}
|
163
|
+
async upload(fileBuf) {
|
164
|
+
this.log.info('Uploading file to S3', { fileKey: this.fileKey });
|
165
|
+
const command = new PutObjectCommand({
|
166
|
+
Bucket: this.bucket,
|
167
|
+
Key: this.fileKey,
|
168
|
+
Body: fileBuf,
|
169
|
+
});
|
170
|
+
await this.s3Client.send(command);
|
171
|
+
}
|
172
|
+
async startAnalysis(s3Key) {
|
173
|
+
const command = new StartDocumentAnalysisCommand({
|
174
|
+
DocumentLocation: {
|
175
|
+
S3Object: {
|
176
|
+
Bucket: this.bucket,
|
177
|
+
Name: s3Key
|
178
|
+
}
|
179
|
+
},
|
180
|
+
FeatureTypes: ["TABLES"]
|
181
|
+
});
|
182
|
+
const response = await this.textractClient.send(command);
|
183
|
+
return response.JobId;
|
184
|
+
}
|
185
|
+
async checkJobStatus(jobId) {
|
186
|
+
const command = new GetDocumentAnalysisCommand({ JobId: jobId });
|
187
|
+
const response = await this.textractClient.send(command);
|
188
|
+
return response.JobStatus;
|
189
|
+
}
|
190
|
+
getImagePlaceholder(block) {
|
191
|
+
const geometry = block.Geometry?.BoundingBox;
|
192
|
+
if (!geometry)
|
193
|
+
return '';
|
194
|
+
const area = (geometry.Width || 0) * (geometry.Height || 0);
|
195
|
+
if (area < 0.05)
|
196
|
+
return ''; // skip small images
|
197
|
+
const top = geometry.Top || 0;
|
198
|
+
const left = geometry.Left || 0;
|
199
|
+
let position = '';
|
200
|
+
if (top < 0.3)
|
201
|
+
position += 'TOP_';
|
202
|
+
else if (top > 0.7)
|
203
|
+
position += 'BOTTOM_';
|
204
|
+
if (left < 0.3)
|
205
|
+
position += 'LEFT';
|
206
|
+
else if (left > 0.7)
|
207
|
+
position += 'RIGHT';
|
208
|
+
else
|
209
|
+
position += 'CENTER';
|
210
|
+
return `[IMAGE_${position}]\n`;
|
211
|
+
}
|
212
|
+
getIndentationLevel(block) {
|
213
|
+
const left = block.Geometry?.BoundingBox?.Left || 0;
|
214
|
+
if (left < 0.15)
|
215
|
+
return 0;
|
216
|
+
if (left < 0.25)
|
217
|
+
return 1;
|
218
|
+
return 2;
|
219
|
+
}
|
220
|
+
isLikelyHeader(block, prevBlock) {
|
221
|
+
if (!prevBlock)
|
222
|
+
return true;
|
223
|
+
const gap = (block.Geometry?.BoundingBox?.Top || 0) -
|
224
|
+
((prevBlock.Geometry?.BoundingBox?.Top || 0) +
|
225
|
+
(prevBlock.Geometry?.BoundingBox?.Height || 0));
|
226
|
+
return gap > 0.03;
|
227
|
+
}
|
228
|
+
formatTextBlock(block, prevBlock) {
|
229
|
+
const text = block.Text || '';
|
230
|
+
const indentLevel = this.getIndentationLevel(block);
|
231
|
+
const indent = ' '.repeat(indentLevel);
|
232
|
+
if (this.isLikelyHeader(block, prevBlock)) {
|
233
|
+
return `\n${indent}${text}\n`;
|
234
|
+
}
|
235
|
+
return `${indent}${text}\n`;
|
236
|
+
}
|
237
|
+
shouldMergeLines(prev, current) {
|
238
|
+
const prevBottom = (prev.Geometry?.BoundingBox?.Top || 0)
|
239
|
+
+ (prev.Geometry?.BoundingBox?.Height || 0);
|
240
|
+
const currentTop = current.Geometry?.BoundingBox?.Top || 0;
|
241
|
+
const gap = currentTop - prevBottom;
|
242
|
+
// For example, if gap < 0.02, treat them as contiguous
|
243
|
+
if (gap < 0.02) {
|
244
|
+
return true;
|
245
|
+
}
|
246
|
+
return false;
|
247
|
+
}
|
248
|
+
async processResults(jobId) {
|
249
|
+
let nextToken;
|
250
|
+
let allBlocks = [];
|
251
|
+
do {
|
252
|
+
const command = new GetDocumentAnalysisCommand({
|
253
|
+
JobId: jobId,
|
254
|
+
NextToken: nextToken
|
255
|
+
});
|
256
|
+
const response = await this.textractClient.send(command);
|
257
|
+
allBlocks = allBlocks.concat(response.Blocks || []);
|
258
|
+
nextToken = response.NextToken;
|
259
|
+
} while (nextToken);
|
260
|
+
// Create blocks map
|
261
|
+
const blocksMap = {};
|
262
|
+
for (const block of allBlocks) {
|
263
|
+
blocksMap[block.Id] = block;
|
264
|
+
}
|
265
|
+
// We'll store each page's content in sequence
|
266
|
+
const pageContents = [];
|
267
|
+
let currentPage = null;
|
268
|
+
// We'll keep track of a "current text block" that we're building
|
269
|
+
let currentTextContent = "";
|
270
|
+
let prevLineBlock = null;
|
271
|
+
// Sort by page and vertical position
|
272
|
+
allBlocks.sort((a, b) => {
|
273
|
+
if (a.Page !== b.Page)
|
274
|
+
return (a.Page || 0) - (b.Page || 0);
|
275
|
+
return (a.Geometry?.BoundingBox?.Top || 0) - (b.Geometry?.BoundingBox?.Top || 0);
|
276
|
+
});
|
277
|
+
for (const block of allBlocks) {
|
278
|
+
if (block.BlockType === 'PAGE') {
|
279
|
+
// If we were building a text block, push it before starting a new page
|
280
|
+
if (currentTextContent.trim().length > 0 && currentPage) {
|
281
|
+
currentPage.blocks.push({
|
282
|
+
type: 'text',
|
283
|
+
content: currentTextContent
|
284
|
+
});
|
285
|
+
}
|
286
|
+
if (currentPage) {
|
287
|
+
pageContents.push(currentPage);
|
288
|
+
}
|
289
|
+
currentPage = {
|
290
|
+
pageNumber: block.Page || 0,
|
291
|
+
blocks: []
|
292
|
+
};
|
293
|
+
currentTextContent = "";
|
294
|
+
prevLineBlock = null;
|
295
|
+
}
|
296
|
+
else if (currentPage && block.Page === currentPage.pageNumber) {
|
297
|
+
// TABLE handling
|
298
|
+
if (block.BlockType === 'TABLE') {
|
299
|
+
// If there's a pending text block, push it first
|
300
|
+
if (currentTextContent.trim().length > 0) {
|
301
|
+
currentPage.blocks.push({
|
302
|
+
type: 'text',
|
303
|
+
content: currentTextContent
|
304
|
+
});
|
305
|
+
currentTextContent = "";
|
306
|
+
}
|
307
|
+
const { csv, tableConfidence } = this.generateTableCSV(block, blocksMap, currentPage.blocks.filter(b => b.type === 'table').length + 1, currentPage.pageNumber);
|
308
|
+
currentPage.blocks.push({
|
309
|
+
type: 'table',
|
310
|
+
content: csv,
|
311
|
+
confidence: tableConfidence
|
312
|
+
});
|
313
|
+
prevLineBlock = null;
|
314
|
+
}
|
315
|
+
// LINE handling (merge or start new)
|
316
|
+
else if (block.BlockType === 'LINE' && !this.isBlockInTable(block, blocksMap)) {
|
317
|
+
if (prevLineBlock && this.shouldMergeLines(prevLineBlock, block)) {
|
318
|
+
// If we consider this line to be part of the same paragraph,
|
319
|
+
// just append the text. We'll call formatTextBlock to get
|
320
|
+
// indentation/header logic, but we won't add a leading newline.
|
321
|
+
const formatted = this.formatTextBlock(block, prevLineBlock);
|
322
|
+
// formatTextBlock might include a leading newline if isLikelyHeader = true
|
323
|
+
// so you can strip it out if you want them truly "merged" into one paragraph:
|
324
|
+
const mergedText = formatted.replace(/^\s*\n/, " ");
|
325
|
+
currentTextContent += " " + mergedText.trim();
|
326
|
+
}
|
327
|
+
else {
|
328
|
+
// If there's an existing text block, push it
|
329
|
+
if (currentTextContent.trim().length > 0) {
|
330
|
+
currentPage.blocks.push({
|
331
|
+
type: 'text',
|
332
|
+
content: currentTextContent
|
333
|
+
});
|
334
|
+
}
|
335
|
+
// Start a new text block
|
336
|
+
currentTextContent = this.formatTextBlock(block, prevLineBlock).trim();
|
337
|
+
}
|
338
|
+
prevLineBlock = block;
|
339
|
+
}
|
340
|
+
// IMAGES (if detectImages)
|
341
|
+
else if (this.detectImages) {
|
342
|
+
const geometry = block.Geometry?.BoundingBox;
|
343
|
+
if (geometry && geometry.Width && geometry.Height) {
|
344
|
+
const imagePlaceholder = this.getImagePlaceholder(block);
|
345
|
+
if (imagePlaceholder) {
|
346
|
+
// If there's a pending text block, push it first
|
347
|
+
if (currentTextContent.trim().length > 0) {
|
348
|
+
currentPage.blocks.push({
|
349
|
+
type: 'text',
|
350
|
+
content: currentTextContent
|
351
|
+
});
|
352
|
+
currentTextContent = "";
|
353
|
+
}
|
354
|
+
currentPage.blocks.push({
|
355
|
+
type: 'image',
|
356
|
+
content: imagePlaceholder,
|
357
|
+
left: geometry.Left,
|
358
|
+
top: geometry.Top,
|
359
|
+
width: geometry.Width,
|
360
|
+
height: geometry.Height
|
361
|
+
});
|
362
|
+
}
|
363
|
+
}
|
364
|
+
// No line update to prevLineBlock here
|
365
|
+
}
|
366
|
+
}
|
367
|
+
}
|
368
|
+
// Handle last page
|
369
|
+
if (currentPage) {
|
370
|
+
if (currentTextContent.trim().length > 0) {
|
371
|
+
currentPage.blocks.push({
|
372
|
+
type: 'text',
|
373
|
+
content: currentTextContent
|
374
|
+
});
|
375
|
+
}
|
376
|
+
pageContents.push(currentPage);
|
377
|
+
}
|
378
|
+
// Build final output
|
379
|
+
let fulltext = '';
|
380
|
+
let imgNumber = 1;
|
381
|
+
for (const page of pageContents) {
|
382
|
+
fulltext += `<page number="${page.pageNumber}">\n`;
|
383
|
+
for (const block of page.blocks) {
|
384
|
+
if (block.type === 'text') {
|
385
|
+
fulltext += `<text>\n${block.content}\n</text>\n\n`;
|
386
|
+
}
|
387
|
+
else if (block.type === 'table') {
|
388
|
+
const confidenceAttr = block.confidence !== undefined && this.includeConfidenceInTables
|
389
|
+
? ` confidence="${block.confidence.toFixed(2)}"`
|
390
|
+
: '';
|
391
|
+
fulltext += `<table type="csv"${confidenceAttr}>\n`;
|
392
|
+
fulltext += `${block.content}\n`;
|
393
|
+
fulltext += `</table>\n\n`;
|
394
|
+
}
|
395
|
+
else if (block.type === 'image') {
|
396
|
+
// Include geometry if you like
|
397
|
+
const leftAttr = block.left ? ` left="${block.left.toFixed(4)}"` : '';
|
398
|
+
const topAttr = block.top ? ` top="${block.top.toFixed(4)}"` : '';
|
399
|
+
const widthAttr = block.width ? ` width="${block.width.toFixed(4)}"` : '';
|
400
|
+
const heightAttr = block.height ? ` height="${block.height.toFixed(4)}"` : '';
|
401
|
+
fulltext += `<image id="${imgNumber++}" ${leftAttr}${topAttr}${widthAttr}${heightAttr}>\n${block.content.trim()}\n</image>\n\n`;
|
402
|
+
}
|
403
|
+
}
|
404
|
+
fulltext += `</page>\n\n`;
|
405
|
+
}
|
406
|
+
return fulltext;
|
407
|
+
}
|
408
|
+
}
|
409
|
+
//# sourceMappingURL=TextractProcessor.js.map
|
@@ -0,0 +1 @@
|
|
1
|
+
{"version":3,"file":"TextractProcessor.js","sourceRoot":"","sources":["../../../src/conversion/TextractProcessor.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,gBAAgB,EAAE,QAAQ,EAAE,MAAM,oBAAoB,CAAC;AAEhE,OAAO,EACH,0BAA0B,EAC1B,4BAA4B,EAC5B,cAAc,EACjB,MAAM,0BAA0B,CAAC;AAElC,OAAO,IAAI,MAAM,WAAW,CAAC;AAmC7B,MAAM,OAAO,iBAAiB;IAClB,cAAc,CAAiB;IAC/B,QAAQ,CAAW;IACnB,OAAO,CAAS;IAChB,MAAM,CAAS;IACf,GAAG,CAAM;IACT,YAAY,CAAU;IAC9B;;OAEG;IACK,yBAAyB,CAAU;IAE3C,YAAY,EACR,OAAO,EACP,MAAM,EACN,MAAM,EACN,WAAW,EACX,GAAG,EACH,YAAY,GAAG,KAAK,EACpB,yBAAyB,GAAG,KAAK,CAAE,sBAAsB;MAClC;QACvB,IAAI,CAAC,OAAO,GAAG,OAAO,CAAC;QACvB,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC;QACrB,IAAI,CAAC,GAAG,GAAG,GAAG,CAAC;QACf,IAAI,CAAC,YAAY,GAAG,YAAY,CAAC;QACjC,IAAI,CAAC,yBAAyB,GAAG,yBAAyB,CAAC;QAE3D,IAAI,CAAC,cAAc,GAAG,IAAI,cAAc,CAAC;YACrC,MAAM;YACN,WAAW;SACd,CAAC,CAAC;QACH,IAAI,CAAC,QAAQ,GAAG,IAAI,QAAQ,CAAC;YACzB,MAAM;YACN,WAAW;SACd,CAAC,CAAC;IACP,CAAC;IAEO,OAAO,CAAC,MAAa,EAAE,SAAoB;QAC/C,IAAI,IAAI,GAAG,EAAE,CAAC;QACd,IAAI,MAAM,CAAC,aAAa,EAAE,CAAC;YACvB,KAAK,MAAM,YAAY,IAAI,MAAM,CAAC,aAAa,EAAE,CAAC;gBAC9C,IAAI,YAAY,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;oBAChC,KAAK,MAAM,OAAO,IAAI,YAAY,CAAC,GAAG,IAAI,EAAE,EAAE,CAAC;wBAC3C,MAAM,IAAI,GAAG,SAAS,CAAC,OAAO,CAAC,CAAC;wBAChC,IAAI,IAAI,CAAC,SAAS,KAAK,MAAM,EAAE,CAAC;4BAC5B,MAAM,QAAQ,GAAG,IAAI,CAAC,IAAI,IAAI,EAAE,CAAC;4BACjC,kDAAkD;4BAClD,IAAI,QAAQ,CAAC,QAAQ,CAAC,GAAG,CAAC;gCACtB,QAAQ,CAAC,OAAO,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,EAAE,CAAC;gCAC3C,IAAI,IAAI,IAAI,QAAQ,IAAI,CAAC;4BAC7B,CAAC;iCAAM,CAAC;gCACJ,IAAI,IAAI,GAAG,QAAQ,GAAG,CAAC;4BAC3B,CAAC;wBACL,CAAC;wBACD,IACI,IAAI,CAAC,SAAS,KAAK,mBAAmB;4BACtC,IAAI,CAAC,eAAe,KAAK,UAAU,EACrC,CAAC;4BACC,IAAI,IAAI,IAAI,CAAC;wBACjB,CAAC;oBACL,CAAC;gBACL,CAAC;YACL,CAAC;QACL,CAAC;QACD,OAAO,IAAI,CAAC,IAAI,EAAE,CAAC;IACvB,CAAC;IAEO,cAAc,CAAC,KAAY,EAAE,SAAoB;QACrD,IAAI,KAAK,CAAC,SAAS,KAAK,MAAM,EAAE,CAAC;YAC7B,OAAO,KAAK,CAAC;QACjB,CAAC;QACD,IAAI,KAAK,CAAC,aAAa,EAAE,CAAC;YACtB,KAAK,MAAM,YAAY,IAAI,KAAK,CAAC,aAAa,EAAE,CAAC;gBAC7C,IAAI,YAAY,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;oBAChC,KAAK,MAAM,OAAO,IAAI,YAAY,CAAC,GAAG,IAAI,EAAE,EAAE,CAAC;wBAC3C,MAAM,SAAS,GAAG,SAAS,CAAC,OAAO,CAAC,CAAC;wBACrC,IAAI,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,SAAS,CAAC,EAAE,CAAC;4BAC/C,OAAO,IAAI,CAAC;wBAChB,CAAC;oBACL,CAAC;gBACL,CAAC;YACL,CAAC;QACL,CAAC;QACD,OAAO,KAAK,CAAC;IACjB,CAAC;IAEO,iBAAiB,CAAC,SAAgB,EAAE,SAAoB;QAC5D,kEAAkE;QAClE,KAAK,MAAM,OAAO,IAAI,SAAS,EAAE,CAAC;YAC9B,MAAM,cAAc,GAAG,SAAS,CAAC,OAAO,CAAC,CAAC;YAC1C,IAAI,cAAc,CAAC,SAAS,KAAK,OAAO,IAAI,cAAc,CAAC,aAAa,EAAE,CAAC;gBACvE,KAAK,MAAM,YAAY,IAAI,cAAc,CAAC,aAAa,EAAE,CAAC;oBACtD,IAAI,YAAY,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;wBAChC,KAAK,MAAM,MAAM,IAAI,YAAY,CAAC,GAAG,IAAI,EAAE,EAAE,CAAC;4BAC1C,MAAM,IAAI,GAAG,SAAS,CAAC,MAAM,CAAC,CAAC;4BAC/B,IAAI,IAAI,CAAC,SAAS,KAAK,MAAM,IAAI,IAAI,CAAC,aAAa,EAAE,CAAC;gCAClD,KAAK,MAAM,OAAO,IAAI,IAAI,CAAC,aAAa,EAAE,CAAC;oCACvC,IACI,OAAO,CAAC,IAAI,KAAK,OAAO;wCACxB,OAAO,CAAC,GAAG,EAAE,QAAQ,CAAC,SAAS,CAAC,EAAG,CAAC,EACtC,CAAC;wCACC,OAAO,IAAI,CAAC;oCAChB,CAAC;gCACL,CAAC;4BACL,CAAC;wBACL,CAAC;oBACL,CAAC;gBACL,CAAC;YACL,CAAC;QACL,CAAC;QACD,OAAO,KAAK,CAAC;IACjB,CAAC;IAED;;OAEG;IACK,iBAAiB,CACrB,WAAkB,EAClB,SAAoB;QAIpB,MAAM,IAAI,GAAuD,EAAE,CAAC;QAEpE,WAAW,CAAC,aAAa,EAAE,OAAO,CAAC,YAAY,CAAC,EAAE;YAC9C,IAAI,YAAY,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;gBAChC,YAAY,CAAC,GAAG,EAAE,OAAO,CAAC,OAAO,CAAC,EAAE;oBAChC,MAAM,IAAI,GAAG,SAAS,CAAC,OAAO,CAAC,CAAC;oBAChC,IAAI,IAAI,CAAC,SAAS,KAAK,MAAM,EAAE,CAAC;wBAC5B,MAAM,QAAQ,GAAG,IAAI,CAAC,QAAQ,IAAI,CAAC,CAAC;wBACpC,MAAM,QAAQ,GAAG,IAAI,CAAC,WAAW,IAAI,CAAC,CAAC;wBAEvC,6BAA6B;wBAC7B,IAAI,CAAC,IAAI,CAAC,QAAQ,GAAG,CAAC,CAAC,EAAE,CAAC;4BACtB,IAAI,CAAC,QAAQ,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC;wBAC5B,CAAC;wBAED,mCAAmC;wBACnC,MAAM,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,SAAS,CAAC,CAAC;wBAC3C,MAAM,UAAU,GAAG,IAAI,CAAC,UAAU,IAAI,CAAC,CAAC;wBAExC,8CAA8C;wBAC9C,mDAAmD;wBACnD,KAAK,IAAI,CAAC,GAAG,IAAI,CAAC,QAAQ,GAAG,CAAC,CAAC,CAAC,MAAM,EAAE,CAAC,GAAG,QAAQ,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;4BAC5D,IAAI,CAAC,QAAQ,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE,UAAU,EAAE,CAAC,EAAE,CAAC,CAAC;wBACzD,CAAC;wBACD,IAAI,CAAC,QAAQ,GAAG,CAAC,CAAC,CAAC,QAAQ,GAAG,CAAC,CAAC,GAAG,EAAE,IAAI,EAAE,UAAU,EAAE,CAAC;oBAC5D,CAAC;gBACL,CAAC,CAAC,CAAC;YACP,CAAC;QACL,CAAC,CAAC,CAAC;QAEH,OAAO,EAAE,IAAI,EAAE,CAAC;IACpB,CAAC;IAEO,gBAAgB,CACpB,WAAkB,EAClB,SAAoB,EACpB,WAAmB,EACnB,WAAmB;QAEnB,MAAM,EAAE,IAAI,EAAE,GAAG,IAAI,CAAC,iBAAiB,CAAC,WAAW,EAAE,SAAS,CAAC,CAAC;QAEhE,IAAI,eAAe,GAAG,CAAC,CAAC;QACxB,IAAI,SAAS,GAAG,CAAC,CAAC;QAElB,mBAAmB;QACnB,MAAM,OAAO,GAAe,EAAE,CAAC;QAC/B,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;YACrB,MAAM,OAAO,GAAa,EAAE,CAAC;YAC7B,KAAK,MAAM,IAAI,IAAI,GAAG,EAAE,CAAC;gBACrB,aAAa;gBACb,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC;gBAC/B,wBAAwB;gBACxB,eAAe,IAAI,IAAI,CAAC,UAAU,CAAC;gBACnC,SAAS,EAAE,CAAC;YAChB,CAAC;YACD,OAAO,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;QAC1B,CAAC;QAED,8DAA8D;QAC9D,MAAM,eAAe,GAAG,SAAS,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,eAAe,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAE1E,iBAAiB;QACjB,MAAM,GAAG,GAAG,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE;YAC9B,SAAS,EAAE,GAAG;YACd,MAAM,EAAE,IAAI;YACZ,SAAS,EAAE,GAAG;YACd,UAAU,EAAE,GAAG;YACf,MAAM,EAAE,KAAK;YACb,OAAO,EAAE,IAAI;YACb,cAAc,EAAE,KAAK;SACxB,CAAC,CAAC;QAEH,OAAO,EAAE,GAAG,EAAE,eAAe,EAAE,CAAC;IACpC,CAAC;IAED,KAAK,CAAC,MAAM,CAAC,OAAe;QACxB,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,sBAAsB,EAAE,EAAE,OAAO,EAAE,IAAI,CAAC,OAAO,EAAE,CAAC,CAAC;QACjE,MAAM,OAAO,GAAG,IAAI,gBAAgB,CAAC;YACjC,MAAM,EAAE,IAAI,CAAC,MAAM;YACnB,GAAG,EAAE,IAAI,CAAC,OAAO;YACjB,IAAI,EAAE,OAAO;SAChB,CAAC,CAAC;QACH,MAAM,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IACtC,CAAC;IAED,KAAK,CAAC,aAAa,CAAC,KAAa;QAC7B,MAAM,OAAO,GAAG,IAAI,4BAA4B,CAAC;YAC7C,gBAAgB,EAAE;gBACd,QAAQ,EAAE;oBACN,MAAM,EAAE,IAAI,CAAC,MAAM;oBACnB,IAAI,EAAE,KAAK;iBACd;aACJ;YACD,YAAY,EAAE,CAAC,QAAQ,CAAC;SAC3B,CAAC,CAAC;QACH,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;QACzD,OAAO,QAAQ,CAAC,KAAM,CAAC;IAC3B,CAAC;IAED,KAAK,CAAC,cAAc,CAAC,KAAa;QAC9B,MAAM,OAAO,GAAG,IAAI,0BAA0B,CAAC,EAAE,KAAK,EAAE,KAAK,EAAE,CAAC,CAAC;QACjE,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;QACzD,OAAO,QAAQ,CAAC,SAAU,CAAC;IAC/B,CAAC;IAEO,mBAAmB,CAAC,KAAY;QACpC,MAAM,QAAQ,GAAG,KAAK,CAAC,QAAQ,EAAE,WAAW,CAAC;QAC7C,IAAI,CAAC,QAAQ;YAAE,OAAO,EAAE,CAAC;QACzB,MAAM,IAAI,GAAG,CAAC,QAAQ,CAAC,KAAK,IAAI,CAAC,CAAC,GAAG,CAAC,QAAQ,CAAC,MAAM,IAAI,CAAC,CAAC,CAAC;QAC5D,IAAI,IAAI,GAAG,IAAI;YAAE,OAAO,EAAE,CAAC,CAAC,oBAAoB;QAEhD,MAAM,GAAG,GAAG,QAAQ,CAAC,GAAG,IAAI,CAAC,CAAC;QAC9B,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI,IAAI,CAAC,CAAC;QAEhC,IAAI,QAAQ,GAAG,EAAE,CAAC;QAClB,IAAI,GAAG,GAAG,GAAG;YAAE,QAAQ,IAAI,MAAM,CAAC;aAC7B,IAAI,GAAG,GAAG,GAAG;YAAE,QAAQ,IAAI,SAAS,CAAC;QAE1C,IAAI,IAAI,GAAG,GAAG;YAAE,QAAQ,IAAI,MAAM,CAAC;aAC9B,IAAI,IAAI,GAAG,GAAG;YAAE,QAAQ,IAAI,OAAO,CAAC;;YACpC,QAAQ,IAAI,QAAQ,CAAC;QAE1B,OAAO,UAAU,QAAQ,KAAK,CAAC;IACnC,CAAC;IAEO,mBAAmB,CAAC,KAAY;QACpC,MAAM,IAAI,GAAG,KAAK,CAAC,QAAQ,EAAE,WAAW,EAAE,IAAI,IAAI,CAAC,CAAC;QACpD,IAAI,IAAI,GAAG,IAAI;YAAE,OAAO,CAAC,CAAC;QAC1B,IAAI,IAAI,GAAG,IAAI;YAAE,OAAO,CAAC,CAAC;QAC1B,OAAO,CAAC,CAAC;IACb,CAAC;IAEO,cAAc,CAAC,KAAY,EAAE,SAAuB;QACxD,IAAI,CAAC,SAAS;YAAE,OAAO,IAAI,CAAC;QAC5B,MAAM,GAAG,GAAG,CAAC,KAAK,CAAC,QAAQ,EAAE,WAAW,EAAE,GAAG,IAAI,CAAC,CAAC;YAC/C,CAAC,CAAC,SAAS,CAAC,QAAQ,EAAE,WAAW,EAAE,GAAG,IAAI,CAAC,CAAC;gBACxC,CAAC,SAAS,CAAC,QAAQ,EAAE,WAAW,EAAE,MAAM,IAAI,CAAC,CAAC,CAAC,CAAC;QACxD,OAAO,GAAG,GAAG,IAAI,CAAC;IACtB,CAAC;IAEO,eAAe,CAAC,KAAY,EAAE,SAAuB;QACzD,MAAM,IAAI,GAAG,KAAK,CAAC,IAAI,IAAI,EAAE,CAAC;QAC9B,MAAM,WAAW,GAAG,IAAI,CAAC,mBAAmB,CAAC,KAAK,CAAC,CAAC;QACpD,MAAM,MAAM,GAAG,MAAM,CAAC,MAAM,CAAC,WAAW,CAAC,CAAC;QAE1C,IAAI,IAAI,CAAC,cAAc,CAAC,KAAK,EAAE,SAAS,CAAC,EAAE,CAAC;YACxC,OAAO,KAAK,MAAM,GAAG,IAAI,IAAI,CAAC;QAClC,CAAC;QACD,OAAO,GAAG,MAAM,GAAG,IAAI,IAAI,CAAC;IAChC,CAAC;IAEO,gBAAgB,CAAC,IAAW,EAAE,OAAc;QAChD,MAAM,UAAU,GAAG,CAAC,IAAI,CAAC,QAAQ,EAAE,WAAW,EAAE,GAAG,IAAI,CAAC,CAAC;cACnD,CAAC,IAAI,CAAC,QAAQ,EAAE,WAAW,EAAE,MAAM,IAAI,CAAC,CAAC,CAAC;QAChD,MAAM,UAAU,GAAG,OAAO,CAAC,QAAQ,EAAE,WAAW,EAAE,GAAG,IAAI,CAAC,CAAC;QAC3D,MAAM,GAAG,GAAG,UAAU,GAAG,UAAU,CAAC;QAEpC,uDAAuD;QACvD,IAAI,GAAG,GAAG,IAAI,EAAE,CAAC;YACb,OAAO,IAAI,CAAC;QAChB,CAAC;QACD,OAAO,KAAK,CAAC;IACjB,CAAC;IAED,KAAK,CAAC,cAAc,CAAC,KAAa;QAC9B,IAAI,SAA6B,CAAC;QAClC,IAAI,SAAS,GAAY,EAAE,CAAC;QAE5B,GAAG,CAAC;YACA,MAAM,OAAO,GAAG,IAAI,0BAA0B,CAAC;gBAC3C,KAAK,EAAE,KAAK;gBACZ,SAAS,EAAE,SAAS;aACvB,CAAC,CAAC;YACH,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;YACzD,SAAS,GAAG,SAAS,CAAC,MAAM,CAAC,QAAQ,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC;YACpD,SAAS,GAAG,QAAQ,CAAC,SAAS,CAAC;QACnC,CAAC,QAAQ,SAAS,EAAE;QAEpB,oBAAoB;QACpB,MAAM,SAAS,GAAc,EAAE,CAAC;QAChC,KAAK,MAAM,KAAK,IAAI,SAAS,EAAE,CAAC;YAC5B,SAAS,CAAC,KAAK,CAAC,EAAG,CAAC,GAAG,KAAK,CAAC;QACjC,CAAC;QAED,8CAA8C;QAC9C,MAAM,YAAY,GAAkB,EAAE,CAAC;QACvC,IAAI,WAAW,GAAuB,IAAI,CAAC;QAE3C,iEAAiE;QACjE,IAAI,kBAAkB,GAAG,EAAE,CAAC;QAC5B,IAAI,aAAa,GAAiB,IAAI,CAAC;QAEvC,qCAAqC;QACrC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE;YACpB,IAAI,CAAC,CAAC,IAAI,KAAK,CAAC,CAAC,IAAI;gBAAE,OAAO,CAAC,CAAC,CAAC,IAAI,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC;YAC5D,OAAO,CAAC,CAAC,CAAC,QAAQ,EAAE,WAAW,EAAE,GAAG,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,QAAQ,EAAE,WAAW,EAAE,GAAG,IAAI,CAAC,CAAC,CAAC;QACrF,CAAC,CAAC,CAAC;QAEH,KAAK,MAAM,KAAK,IAAI,SAAS,EAAE,CAAC;YAC5B,IAAI,KAAK,CAAC,SAAS,KAAK,MAAM,EAAE,CAAC;gBAC7B,uEAAuE;gBACvE,IAAI,kBAAkB,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,IAAI,WAAW,EAAE,CAAC;oBACtD,WAAW,CAAC,MAAM,CAAC,IAAI,CAAC;wBACpB,IAAI,EAAE,MAAM;wBACZ,OAAO,EAAE,kBAAkB;qBAC9B,CAAC,CAAC;gBACP,CAAC;gBACD,IAAI,WAAW,EAAE,CAAC;oBACd,YAAY,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC;gBACnC,CAAC;gBACD,WAAW,GAAG;oBACV,UAAU,EAAE,KAAK,CAAC,IAAI,IAAI,CAAC;oBAC3B,MAAM,EAAE,EAAE;iBACb,CAAC;gBACF,kBAAkB,GAAG,EAAE,CAAC;gBACxB,aAAa,GAAG,IAAI,CAAC;YACzB,CAAC;iBACI,IAAI,WAAW,IAAI,KAAK,CAAC,IAAI,KAAK,WAAW,CAAC,UAAU,EAAE,CAAC;gBAC5D,iBAAiB;gBACjB,IAAI,KAAK,CAAC,SAAS,KAAK,OAAO,EAAE,CAAC;oBAC9B,iDAAiD;oBACjD,IAAI,kBAAkB,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;wBACvC,WAAW,CAAC,MAAM,CAAC,IAAI,CAAC;4BACpB,IAAI,EAAE,MAAM;4BACZ,OAAO,EAAE,kBAAkB;yBAC9B,CAAC,CAAC;wBACH,kBAAkB,GAAG,EAAE,CAAC;oBAC5B,CAAC;oBACD,MAAM,EAAE,GAAG,EAAE,eAAe,EAAE,GAAG,IAAI,CAAC,gBAAgB,CAClD,KAAK,EACL,SAAS,EACT,WAAW,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,OAAO,CAAC,CAAC,MAAM,GAAG,CAAC,EAC7D,WAAW,CAAC,UAAU,CACzB,CAAC;oBACF,WAAW,CAAC,MAAM,CAAC,IAAI,CAAC;wBACpB,IAAI,EAAE,OAAO;wBACb,OAAO,EAAE,GAAG;wBACZ,UAAU,EAAE,eAAe;qBAC9B,CAAC,CAAC;oBACH,aAAa,GAAG,IAAI,CAAC;gBACzB,CAAC;gBACD,qCAAqC;qBAChC,IAAI,KAAK,CAAC,SAAS,KAAK,MAAM,IAAI,CAAC,IAAI,CAAC,cAAc,CAAC,KAAK,EAAE,SAAS,CAAC,EAAE,CAAC;oBAC5E,IAAI,aAAa,IAAI,IAAI,CAAC,gBAAgB,CAAC,aAAa,EAAE,KAAK,CAAC,EAAE,CAAC;wBAC/D,6DAA6D;wBAC7D,0DAA0D;wBAC1D,gEAAgE;wBAChE,MAAM,SAAS,GAAG,IAAI,CAAC,eAAe,CAAC,KAAK,EAAE,aAAa,CAAC,CAAC;wBAE7D,2EAA2E;wBAC3E,8EAA8E;wBAC9E,MAAM,UAAU,GAAG,SAAS,CAAC,OAAO,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC;wBAEpD,kBAAkB,IAAI,GAAG,GAAG,UAAU,CAAC,IAAI,EAAE,CAAC;oBAClD,CAAC;yBAAM,CAAC;wBACJ,6CAA6C;wBAC7C,IAAI,kBAAkB,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;4BACvC,WAAW,CAAC,MAAM,CAAC,IAAI,CAAC;gCACpB,IAAI,EAAE,MAAM;gCACZ,OAAO,EAAE,kBAAkB;6BAC9B,CAAC,CAAC;wBACP,CAAC;wBACD,yBAAyB;wBACzB,kBAAkB,GAAG,IAAI,CAAC,eAAe,CAAC,KAAK,EAAE,aAAa,CAAC,CAAC,IAAI,EAAE,CAAC;oBAC3E,CAAC;oBACD,aAAa,GAAG,KAAK,CAAC;gBAC1B,CAAC;gBACD,2BAA2B;qBACtB,IAAI,IAAI,CAAC,YAAY,EAAE,CAAC;oBACzB,MAAM,QAAQ,GAAG,KAAK,CAAC,QAAQ,EAAE,WAAW,CAAC;oBAC7C,IAAI,QAAQ,IAAI,QAAQ,CAAC,KAAK,IAAI,QAAQ,CAAC,MAAM,EAAE,CAAC;wBAChD,MAAM,gBAAgB,GAAG,IAAI,CAAC,mBAAmB,CAAC,KAAK,CAAC,CAAC;wBACzD,IAAI,gBAAgB,EAAE,CAAC;4BACnB,iDAAiD;4BACjD,IAAI,kBAAkB,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gCACvC,WAAW,CAAC,MAAM,CAAC,IAAI,CAAC;oCACpB,IAAI,EAAE,MAAM;oCACZ,OAAO,EAAE,kBAAkB;iCAC9B,CAAC,CAAC;gCACH,kBAAkB,GAAG,EAAE,CAAC;4BAC5B,CAAC;4BAED,WAAW,CAAC,MAAM,CAAC,IAAI,CAAC;gCACpB,IAAI,EAAE,OAAO;gCACb,OAAO,EAAE,gBAAgB;gCACzB,IAAI,EAAE,QAAQ,CAAC,IAAI;gCACnB,GAAG,EAAE,QAAQ,CAAC,GAAG;gCACjB,KAAK,EAAE,QAAQ,CAAC,KAAK;gCACrB,MAAM,EAAE,QAAQ,CAAC,MAAM;6BAC1B,CAAC,CAAC;wBACP,CAAC;oBACL,CAAC;oBACD,uCAAuC;gBAC3C,CAAC;YACL,CAAC;QACL,CAAC;QAED,mBAAmB;QACnB,IAAI,WAAW,EAAE,CAAC;YACd,IAAI,kBAAkB,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBACvC,WAAW,CAAC,MAAM,CAAC,IAAI,CAAC;oBACpB,IAAI,EAAE,MAAM;oBACZ,OAAO,EAAE,kBAAkB;iBAC9B,CAAC,CAAC;YACP,CAAC;YACD,YAAY,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC;QACnC,CAAC;QAED,qBAAqB;QACrB,IAAI,QAAQ,GAAG,EAAE,CAAC;QAClB,IAAI,SAAS,GAAG,CAAC,CAAC;QAClB,KAAK,MAAM,IAAI,IAAI,YAAY,EAAE,CAAC;YAC9B,QAAQ,IAAI,iBAAiB,IAAI,CAAC,UAAU,MAAM,CAAC;YACnD,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,MAAM,EAAE,CAAC;gBAC9B,IAAI,KAAK,CAAC,IAAI,KAAK,MAAM,EAAE,CAAC;oBACxB,QAAQ,IAAI,WAAW,KAAK,CAAC,OAAO,eAAe,CAAC;gBACxD,CAAC;qBAAM,IAAI,KAAK,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;oBAChC,MAAM,cAAc,GAAG,KAAK,CAAC,UAAU,KAAK,SAAS,IAAI,IAAI,CAAC,yBAAyB;wBACnF,CAAC,CAAC,gBAAgB,KAAK,CAAC,UAAU,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG;wBAChD,CAAC,CAAC,EAAE,CAAC;oBACT,QAAQ,IAAI,oBAAoB,cAAc,KAAK,CAAC;oBACpD,QAAQ,IAAI,GAAG,KAAK,CAAC,OAAO,IAAI,CAAC;oBACjC,QAAQ,IAAI,cAAc,CAAC;gBAC/B,CAAC;qBAAM,IAAI,KAAK,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;oBAChC,+BAA+B;oBAC/B,MAAM,QAAQ,GAAG,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,UAAU,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC;oBACtE,MAAM,OAAO,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,SAAS,KAAK,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC;oBAClE,MAAM,SAAS,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,WAAW,KAAK,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC;oBAC1E,MAAM,UAAU,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,YAAY,KAAK,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC;oBAE9E,QAAQ,IAAI,cAAc,SAAS,EAAE,KAAK,QAAQ,GAAG,OAAO,GAAG,SAAS,GAAG,UAAU,MAAM,KAAK,CAAC,OAAO,CAAC,IAAI,EAAE,gBAAgB,CAAC;gBACpI,CAAC;YACL,CAAC;YACD,QAAQ,IAAI,aAAa,CAAC;QAC9B,CAAC;QAED,OAAO,QAAQ,CAAC;IACpB,CAAC;CAEJ"}
|