@vertesia/workflow 1.3.0 → 1.4.0-dev.20260615.051508Z
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/README.md +1 -2
- package/lib/{types/activities → activities}/advanced/createDocumentTypeFromInteractionRun.d.ts +2 -2
- package/lib/activities/advanced/createDocumentTypeFromInteractionRun.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/advanced/createDocumentTypeFromInteractionRun.js +9 -9
- package/lib/activities/advanced/createDocumentTypeFromInteractionRun.js.map +1 -0
- package/lib/{types/activities → activities}/advanced/createOrUpdateDocumentFromInteractionRun.d.ts +2 -2
- package/lib/activities/advanced/createOrUpdateDocumentFromInteractionRun.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/advanced/createOrUpdateDocumentFromInteractionRun.js +29 -22
- package/lib/activities/advanced/createOrUpdateDocumentFromInteractionRun.js.map +1 -0
- package/lib/{types/activities → activities}/advanced/updateDocumentFromInteractionRun.d.ts +1 -1
- package/lib/activities/advanced/updateDocumentFromInteractionRun.d.ts.map +1 -0
- package/lib/activities/advanced/updateDocumentFromInteractionRun.js +16 -0
- package/lib/activities/advanced/updateDocumentFromInteractionRun.js.map +1 -0
- package/lib/{types/activities → activities}/chunkDocument.d.ts +3 -3
- package/lib/activities/chunkDocument.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/chunkDocument.js +38 -33
- package/lib/activities/chunkDocument.js.map +1 -0
- package/lib/{types/activities → activities}/createDocumentFromOther.d.ts +1 -1
- package/lib/activities/createDocumentFromOther.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/createDocumentFromOther.js +15 -13
- package/lib/activities/createDocumentFromOther.js.map +1 -0
- package/lib/{types/activities → activities}/executeInteraction.d.ts +19 -10
- package/lib/activities/executeInteraction.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/executeInteraction.js +97 -78
- package/lib/activities/executeInteraction.js.map +1 -0
- package/lib/{types/activities → activities}/executeRemoteActivity.d.ts +4 -4
- package/lib/activities/executeRemoteActivity.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/executeRemoteActivity.js +48 -23
- package/lib/activities/executeRemoteActivity.js.map +1 -0
- package/lib/{types/activities → activities}/extractDocumentText.d.ts +3 -3
- package/lib/activities/extractDocumentText.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/extractDocumentText.js +65 -52
- package/lib/activities/extractDocumentText.js.map +1 -0
- package/lib/{types/activities → activities}/generateDocumentProperties.d.ts +4 -4
- package/lib/activities/generateDocumentProperties.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/generateDocumentProperties.js +37 -33
- package/lib/activities/generateDocumentProperties.js.map +1 -0
- package/lib/{types/activities → activities}/generateEmbeddings.d.ts +3 -3
- package/lib/activities/generateEmbeddings.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/generateEmbeddings.js +80 -80
- package/lib/activities/generateEmbeddings.js.map +1 -0
- package/lib/{types/activities → activities}/generateOrAssignContentType.d.ts +4 -4
- package/lib/activities/generateOrAssignContentType.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/generateOrAssignContentType.js +51 -49
- package/lib/activities/generateOrAssignContentType.js.map +1 -0
- package/lib/{types/activities → activities}/getObjectFromStore.d.ts +1 -1
- package/lib/activities/getObjectFromStore.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/getObjectFromStore.js +4 -3
- package/lib/activities/getObjectFromStore.js.map +1 -0
- package/lib/{types/activities → activities}/handleError.d.ts +1 -1
- package/lib/activities/handleError.d.ts.map +1 -0
- package/lib/activities/handleError.js +37 -0
- package/lib/activities/handleError.js.map +1 -0
- package/lib/activities/index-dsl.d.ts +31 -0
- package/lib/activities/index-dsl.d.ts.map +1 -0
- package/lib/activities/index-dsl.js +29 -0
- package/lib/activities/index-dsl.js.map +1 -0
- package/lib/{types/activities → activities}/index.d.ts +1 -1
- package/lib/activities/index.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/index.js +1 -1
- package/lib/activities/index.js.map +1 -0
- package/lib/{types/activities → activities}/loadChildWorkflowSpec.d.ts +1 -1
- package/lib/activities/loadChildWorkflowSpec.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/loadChildWorkflowSpec.js +4 -2
- package/lib/activities/loadChildWorkflowSpec.js.map +1 -0
- package/lib/{types/activities → activities}/media/prepareAudio.d.ts +1 -1
- package/lib/activities/media/prepareAudio.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/media/prepareAudio.js +27 -23
- package/lib/activities/media/prepareAudio.js.map +1 -0
- package/lib/{types/activities → activities}/media/prepareVideo.d.ts +2 -2
- package/lib/activities/media/prepareVideo.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/media/prepareVideo.js +72 -54
- package/lib/activities/media/prepareVideo.js.map +1 -0
- package/lib/{types/activities → activities}/media/probeMediaStreams.d.ts +2 -3
- package/lib/activities/media/probeMediaStreams.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/media/probeMediaStreams.js +7 -5
- package/lib/activities/media/probeMediaStreams.js.map +1 -0
- package/lib/{types/activities → activities}/media/processPdfWithTextract.d.ts +3 -3
- package/lib/activities/media/processPdfWithTextract.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/media/processPdfWithTextract.js +43 -26
- package/lib/activities/media/processPdfWithTextract.js.map +1 -0
- package/lib/{types/activities → activities}/media/saveGladiaTranscription.d.ts +2 -2
- package/lib/activities/media/saveGladiaTranscription.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/media/saveGladiaTranscription.js +35 -25
- package/lib/activities/media/saveGladiaTranscription.js.map +1 -0
- package/lib/{types/activities → activities}/media/transcribeMediaWithGladia.d.ts +2 -2
- package/lib/activities/media/transcribeMediaWithGladia.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/media/transcribeMediaWithGladia.js +26 -21
- package/lib/activities/media/transcribeMediaWithGladia.js.map +1 -0
- package/lib/{types/activities → activities}/mergeChildArtifacts.d.ts +1 -1
- package/lib/activities/mergeChildArtifacts.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/mergeChildArtifacts.js +27 -26
- package/lib/activities/mergeChildArtifacts.js.map +1 -0
- package/lib/{types/activities → activities}/notifyWebhook.d.ts +6 -5
- package/lib/activities/notifyWebhook.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/notifyWebhook.js +20 -17
- package/lib/activities/notifyWebhook.js.map +1 -0
- package/lib/{types/activities → activities}/rateLimiter.d.ts +1 -1
- package/lib/activities/rateLimiter.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/rateLimiter.js +4 -3
- package/lib/activities/rateLimiter.js.map +1 -0
- package/lib/{types/activities → activities}/renditions/generateImageRendition.d.ts +3 -3
- package/lib/activities/renditions/generateImageRendition.d.ts.map +1 -0
- package/lib/activities/renditions/generateImageRendition.js +75 -0
- package/lib/activities/renditions/generateImageRendition.js.map +1 -0
- package/lib/{types/activities → activities}/renditions/generateVideoRendition.d.ts +3 -3
- package/lib/activities/renditions/generateVideoRendition.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/renditions/generateVideoRendition.js +51 -54
- package/lib/activities/renditions/generateVideoRendition.js.map +1 -0
- package/lib/{types/activities → activities}/resolveRemoteActivities.d.ts +3 -4
- package/lib/activities/resolveRemoteActivities.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/resolveRemoteActivities.js +16 -15
- package/lib/activities/resolveRemoteActivities.js.map +1 -0
- package/lib/{types/activities → activities}/setDocumentStatus.d.ts +1 -1
- package/lib/activities/setDocumentStatus.d.ts.map +1 -0
- package/lib/{esm/activities → activities}/setDocumentStatus.js +5 -3
- package/lib/activities/setDocumentStatus.js.map +1 -0
- package/lib/{types/bulk-import.d.ts → bulk-import.d.ts} +10 -1
- package/lib/bulk-import.d.ts.map +1 -0
- package/lib/bulk-import.js.map +1 -0
- package/lib/{types/conversion → conversion}/TextractProcessor.d.ts +7 -4
- package/lib/conversion/TextractProcessor.d.ts.map +1 -0
- package/lib/{esm/conversion → conversion}/TextractProcessor.js +38 -38
- package/lib/conversion/TextractProcessor.js.map +1 -0
- package/lib/conversion/image.d.ts.map +1 -0
- package/lib/{esm/conversion → conversion}/image.js +35 -35
- package/lib/conversion/image.js.map +1 -0
- package/lib/conversion/markitdown.d.ts.map +1 -0
- package/lib/{esm/conversion → conversion}/markitdown.js +11 -11
- package/lib/conversion/markitdown.js.map +1 -0
- package/lib/conversion/mutool.d.ts.map +1 -0
- package/lib/{esm/conversion → conversion}/mutool.js +19 -24
- package/lib/conversion/mutool.js.map +1 -0
- package/lib/conversion/pandoc.d.ts.map +1 -0
- package/lib/{esm/conversion → conversion}/pandoc.js +11 -11
- package/lib/conversion/pandoc.js.map +1 -0
- package/lib/dsl/conditions.d.ts +2 -0
- package/lib/dsl/conditions.d.ts.map +1 -0
- package/lib/dsl/conditions.js +90 -0
- package/lib/dsl/conditions.js.map +1 -0
- package/lib/{types/dsl → dsl}/dsl-workflow.d.ts +2 -2
- package/lib/dsl/dsl-workflow.d.ts.map +1 -0
- package/lib/{esm/dsl → dsl}/dsl-workflow.js +93 -65
- package/lib/dsl/dsl-workflow.js.map +1 -0
- package/lib/dsl/dslProxyActivities.d.ts +7 -0
- package/lib/dsl/dslProxyActivities.d.ts.map +1 -0
- package/lib/{esm/dsl → dsl}/dslProxyActivities.js +10 -1
- package/lib/dsl/dslProxyActivities.js.map +1 -0
- package/lib/dsl/projections.d.ts +4 -0
- package/lib/dsl/projections.d.ts.map +1 -0
- package/lib/{esm/dsl → dsl}/projections.js +22 -8
- package/lib/dsl/projections.js.map +1 -0
- package/lib/{types/dsl → dsl}/setup/ActivityContext.d.ts +4 -4
- package/lib/dsl/setup/ActivityContext.d.ts.map +1 -0
- package/lib/{esm/dsl → dsl}/setup/ActivityContext.js +22 -25
- package/lib/dsl/setup/ActivityContext.js.map +1 -0
- package/lib/{types/dsl → dsl}/setup/fetch/DataProvider.d.ts +3 -3
- package/lib/dsl/setup/fetch/DataProvider.d.ts.map +1 -0
- package/lib/{esm/dsl → dsl}/setup/fetch/DataProvider.js +2 -1
- package/lib/dsl/setup/fetch/DataProvider.js.map +1 -0
- package/lib/dsl/setup/fetch/index.d.ts +6 -0
- package/lib/dsl/setup/fetch/index.d.ts.map +1 -0
- package/lib/{esm/dsl → dsl}/setup/fetch/index.js +1 -1
- package/lib/dsl/setup/fetch/index.js.map +1 -0
- package/lib/{types/dsl → dsl}/setup/fetch/providers.d.ts +6 -6
- package/lib/dsl/setup/fetch/providers.d.ts.map +1 -0
- package/lib/{esm/dsl → dsl}/setup/fetch/providers.js +30 -16
- package/lib/dsl/setup/fetch/providers.js.map +1 -0
- package/lib/dsl/test/test-child-workflow.d.ts +4 -0
- package/lib/dsl/test/test-child-workflow.d.ts.map +1 -0
- package/lib/{esm/dsl → dsl}/test/test-child-workflow.js +1 -1
- package/lib/dsl/test/test-child-workflow.js.map +1 -0
- package/lib/{types/dsl → dsl}/validation.d.ts +1 -1
- package/lib/dsl/validation.d.ts.map +1 -0
- package/lib/{esm/dsl → dsl}/validation.js +8 -8
- package/lib/dsl/validation.js.map +1 -0
- package/lib/{types/dsl → dsl}/vars.d.ts +16 -16
- package/lib/dsl/vars.d.ts.map +1 -0
- package/lib/{esm/dsl → dsl}/vars.js +20 -15
- package/lib/dsl/vars.js.map +1 -0
- package/lib/dsl/walk.d.ts +18 -0
- package/lib/dsl/walk.d.ts.map +1 -0
- package/lib/{esm/dsl → dsl}/walk.js +37 -16
- package/lib/dsl/walk.js.map +1 -0
- package/lib/dsl.d.ts +4 -0
- package/lib/dsl.d.ts.map +1 -0
- package/lib/dsl.js +4 -0
- package/lib/dsl.js.map +1 -0
- package/lib/{types/errors.d.ts → errors.d.ts} +2 -2
- package/lib/errors.d.ts.map +1 -0
- package/lib/{esm/errors.js → errors.js} +22 -17
- package/lib/errors.js.map +1 -0
- package/lib/index.d.ts +36 -0
- package/lib/index.d.ts.map +1 -0
- package/lib/index.js +37 -0
- package/lib/index.js.map +1 -0
- package/lib/result-types.d.ts.map +1 -0
- package/lib/result-types.js.map +1 -0
- package/lib/security/ssrf.d.ts.map +1 -0
- package/lib/security/ssrf.js.map +1 -0
- package/lib/system/notifyWebhookWorkflow.d.ts +8 -0
- package/lib/system/notifyWebhookWorkflow.d.ts.map +1 -0
- package/lib/{esm/system → system}/notifyWebhookWorkflow.js +13 -12
- package/lib/system/notifyWebhookWorkflow.js.map +1 -0
- package/lib/{types/system → system}/recalculateEmbeddingsWorkflow.d.ts +1 -1
- package/lib/system/recalculateEmbeddingsWorkflow.d.ts.map +1 -0
- package/lib/{esm/system → system}/recalculateEmbeddingsWorkflow.js +7 -7
- package/lib/system/recalculateEmbeddingsWorkflow.js.map +1 -0
- package/lib/utils/activity-info.d.ts +12 -0
- package/lib/utils/activity-info.d.ts.map +1 -0
- package/lib/utils/activity-info.js +15 -0
- package/lib/utils/activity-info.js.map +1 -0
- package/lib/{types/utils → utils}/auth.d.ts +1 -1
- package/lib/utils/auth.d.ts.map +1 -0
- package/lib/utils/auth.js.map +1 -0
- package/lib/{types/utils → utils}/blobs.d.ts +1 -1
- package/lib/utils/blobs.d.ts.map +1 -0
- package/lib/{esm/utils → utils}/blobs.js +17 -16
- package/lib/utils/blobs.js.map +1 -0
- package/lib/utils/chunks.d.ts.map +1 -0
- package/lib/{esm/utils → utils}/chunks.js +1 -1
- package/lib/utils/chunks.js.map +1 -0
- package/lib/utils/client.d.ts +8 -0
- package/lib/utils/client.d.ts.map +1 -0
- package/lib/utils/client.js +64 -0
- package/lib/utils/client.js.map +1 -0
- package/lib/{types/utils → utils}/expand-vars.d.ts +1 -1
- package/lib/utils/expand-vars.d.ts.map +1 -0
- package/lib/{esm/utils → utils}/expand-vars.js +4 -1
- package/lib/utils/expand-vars.js.map +1 -0
- package/lib/{types/utils → utils}/renditions.d.ts +2 -2
- package/lib/utils/renditions.d.ts.map +1 -0
- package/lib/{esm/utils → utils}/renditions.js +29 -18
- package/lib/utils/renditions.js.map +1 -0
- package/lib/{types/utils → utils}/storage.d.ts +2 -2
- package/lib/utils/storage.d.ts.map +1 -0
- package/lib/{esm/utils → utils}/storage.js +13 -12
- package/lib/utils/storage.js.map +1 -0
- package/lib/{types/utils → utils}/text-preview-utils.d.ts +2 -2
- package/lib/utils/text-preview-utils.d.ts.map +1 -0
- package/lib/{esm/utils → utils}/text-preview-utils.js +4 -4
- package/lib/utils/text-preview-utils.js.map +1 -0
- package/lib/{types/utils → utils}/tokens.d.ts +1 -1
- package/lib/utils/tokens.d.ts.map +1 -0
- package/lib/{esm/utils → utils}/tokens.js +3 -3
- package/lib/utils/tokens.js.map +1 -0
- package/lib/vars.d.ts +3 -0
- package/lib/vars.d.ts.map +1 -0
- package/lib/vars.js +4 -0
- package/lib/vars.js.map +1 -0
- package/lib/workflows-bundle.js +5561 -2815
- package/lib/workflows.d.ts +7 -0
- package/lib/workflows.d.ts.map +1 -0
- package/lib/workflows.js +7 -0
- package/lib/workflows.js.map +1 -0
- package/package.json +47 -88
- package/src/activities/advanced/createDocumentTypeFromInteractionRun.ts +30 -17
- package/src/activities/advanced/createOrUpdateDocumentFromInteractionRun.ts +51 -37
- package/src/activities/advanced/updateDocumentFromInteractionRun.ts +12 -11
- package/src/activities/chunkDocument.ts +87 -77
- package/src/activities/createDocumentFromOther.ts +19 -25
- package/src/activities/executeInteraction.test.ts +81 -8
- package/src/activities/executeInteraction.ts +147 -102
- package/src/activities/executeRemoteActivity.test.ts +61 -67
- package/src/activities/executeRemoteActivity.ts +56 -29
- package/src/activities/extractDocumentText.ts +75 -64
- package/src/activities/generateDocumentProperties.ts +69 -50
- package/src/activities/generateEmbeddings.test.ts +114 -0
- package/src/activities/generateEmbeddings.ts +113 -141
- package/src/activities/generateOrAssignContentType.ts +219 -231
- package/src/activities/getObjectFromStore.ts +11 -10
- package/src/activities/handleError.ts +23 -11
- package/src/activities/index-dsl.ts +30 -30
- package/src/activities/index.ts +1 -1
- package/src/activities/loadChildWorkflowSpec.ts +8 -3
- package/src/activities/media/prepareAudio.ts +47 -56
- package/src/activities/media/prepareVideo.ts +107 -122
- package/src/activities/media/probeMediaStreams.test.ts +11 -13
- package/src/activities/media/probeMediaStreams.ts +13 -15
- package/src/activities/media/processPdfWithTextract.ts +65 -46
- package/src/activities/media/saveGladiaTranscription.test.ts +122 -121
- package/src/activities/media/saveGladiaTranscription.ts +52 -26
- package/src/activities/media/transcribeMediaWithGladia.test.ts +173 -165
- package/src/activities/media/transcribeMediaWithGladia.ts +44 -28
- package/src/activities/mergeChildArtifacts.ts +36 -38
- package/src/activities/notifyWebhook.test.ts +223 -217
- package/src/activities/notifyWebhook.ts +67 -40
- package/src/activities/rateLimiter.ts +31 -30
- package/src/activities/renditions/generateImageRendition.ts +54 -64
- package/src/activities/renditions/generateVideoRendition.ts +87 -141
- package/src/activities/resolveRemoteActivities.test.ts +96 -88
- package/src/activities/resolveRemoteActivities.ts +32 -23
- package/src/activities/setDocumentStatus.ts +11 -7
- package/src/bulk-import.ts +11 -5
- package/src/conversion/TextractProcessor.ts +69 -71
- package/src/conversion/image.test.ts +31 -31
- package/src/conversion/image.ts +41 -38
- package/src/conversion/markitdown.ts +11 -11
- package/src/conversion/mutool.test.ts +77 -62
- package/src/conversion/mutool.ts +27 -49
- package/src/conversion/pandoc.test.ts +5 -6
- package/src/conversion/pandoc.ts +11 -11
- package/src/dsl/conditions.ts +35 -17
- package/src/dsl/dsl-workflow.test.ts +49 -39
- package/src/dsl/dsl-workflow.ts +200 -112
- package/src/dsl/dslProxyActivities.test.ts +23 -0
- package/src/dsl/dslProxyActivities.ts +46 -26
- package/src/dsl/ms.d.ts +34 -3
- package/src/dsl/projections.test.ts +108 -100
- package/src/dsl/projections.ts +46 -26
- package/src/dsl/setup/ActivityContext.test.ts +22 -23
- package/src/dsl/setup/ActivityContext.ts +39 -82
- package/src/dsl/setup/fetch/DataProvider.ts +11 -8
- package/src/dsl/setup/fetch/index.ts +6 -8
- package/src/dsl/setup/fetch/providers.ts +38 -24
- package/src/dsl/test/test-child-workflow.ts +3 -2
- package/src/dsl/validation.test.ts +232 -213
- package/src/dsl/validation.ts +17 -12
- package/src/dsl/vars.test.ts +143 -140
- package/src/dsl/vars.ts +61 -54
- package/src/dsl/walk.test.ts +61 -56
- package/src/dsl/walk.ts +61 -35
- package/src/dsl/workflow-exec-child.test.ts +77 -61
- package/src/dsl/workflow-fetch.test.ts +40 -38
- package/src/dsl/workflow-import.test.ts +27 -27
- package/src/dsl/workflow.test.ts +34 -33
- package/src/dsl.ts +3 -4
- package/src/errors.ts +31 -23
- package/src/index.ts +28 -33
- package/src/result-types.ts +6 -9
- package/src/system/notifyWebhookWorkflow.ts +23 -23
- package/src/system/recalculateEmbeddingsWorkflow.ts +14 -17
- package/src/utils/activity-info.ts +15 -0
- package/src/utils/auth.ts +1 -1
- package/src/utils/blobs.ts +19 -18
- package/src/utils/chunks.ts +7 -8
- package/src/utils/client.ts +57 -17
- package/src/utils/expand-vars.ts +9 -6
- package/src/utils/renditions.ts +68 -72
- package/src/utils/storage.ts +18 -22
- package/src/utils/text-preview-utils.ts +9 -19
- package/src/utils/tokens.ts +9 -11
- package/src/vars.ts +2 -2
- package/src/workflows.ts +3 -4
- package/lib/cjs/activities/advanced/createDocumentTypeFromInteractionRun.js +0 -33
- package/lib/cjs/activities/advanced/createDocumentTypeFromInteractionRun.js.map +0 -1
- package/lib/cjs/activities/advanced/createOrUpdateDocumentFromInteractionRun.js +0 -73
- package/lib/cjs/activities/advanced/createOrUpdateDocumentFromInteractionRun.js.map +0 -1
- package/lib/cjs/activities/advanced/updateDocumentFromInteractionRun.js +0 -19
- package/lib/cjs/activities/advanced/updateDocumentFromInteractionRun.js.map +0 -1
- package/lib/cjs/activities/chunkDocument.js +0 -111
- package/lib/cjs/activities/chunkDocument.js.map +0 -1
- package/lib/cjs/activities/createDocumentFromOther.js +0 -64
- package/lib/cjs/activities/createDocumentFromOther.js.map +0 -1
- package/lib/cjs/activities/executeInteraction.js +0 -241
- package/lib/cjs/activities/executeInteraction.js.map +0 -1
- package/lib/cjs/activities/executeRemoteActivity.js +0 -126
- package/lib/cjs/activities/executeRemoteActivity.js.map +0 -1
- package/lib/cjs/activities/extractDocumentText.js +0 -196
- package/lib/cjs/activities/extractDocumentText.js.map +0 -1
- package/lib/cjs/activities/generateDocumentProperties.js +0 -109
- package/lib/cjs/activities/generateDocumentProperties.js.map +0 -1
- package/lib/cjs/activities/generateEmbeddings.js +0 -269
- package/lib/cjs/activities/generateEmbeddings.js.map +0 -1
- package/lib/cjs/activities/generateOrAssignContentType.js +0 -173
- package/lib/cjs/activities/generateOrAssignContentType.js.map +0 -1
- package/lib/cjs/activities/getObjectFromStore.js +0 -31
- package/lib/cjs/activities/getObjectFromStore.js.map +0 -1
- package/lib/cjs/activities/handleError.js +0 -29
- package/lib/cjs/activities/handleError.js.map +0 -1
- package/lib/cjs/activities/index-dsl.js +0 -59
- package/lib/cjs/activities/index-dsl.js.map +0 -1
- package/lib/cjs/activities/index.js +0 -21
- package/lib/cjs/activities/index.js.map +0 -1
- package/lib/cjs/activities/loadChildWorkflowSpec.js +0 -15
- package/lib/cjs/activities/loadChildWorkflowSpec.js.map +0 -1
- package/lib/cjs/activities/media/prepareAudio.js +0 -240
- package/lib/cjs/activities/media/prepareAudio.js.map +0 -1
- package/lib/cjs/activities/media/prepareVideo.js +0 -432
- package/lib/cjs/activities/media/prepareVideo.js.map +0 -1
- package/lib/cjs/activities/media/probeMediaStreams.js +0 -49
- package/lib/cjs/activities/media/probeMediaStreams.js.map +0 -1
- package/lib/cjs/activities/media/processPdfWithTextract.js +0 -103
- package/lib/cjs/activities/media/processPdfWithTextract.js.map +0 -1
- package/lib/cjs/activities/media/saveGladiaTranscription.js +0 -103
- package/lib/cjs/activities/media/saveGladiaTranscription.js.map +0 -1
- package/lib/cjs/activities/media/transcribeMediaWithGladia.js +0 -107
- package/lib/cjs/activities/media/transcribeMediaWithGladia.js.map +0 -1
- package/lib/cjs/activities/mergeChildArtifacts.js +0 -150
- package/lib/cjs/activities/mergeChildArtifacts.js.map +0 -1
- package/lib/cjs/activities/notifyWebhook.js +0 -167
- package/lib/cjs/activities/notifyWebhook.js.map +0 -1
- package/lib/cjs/activities/rateLimiter.js +0 -30
- package/lib/cjs/activities/rateLimiter.js.map +0 -1
- package/lib/cjs/activities/renditions/generateImageRendition.js +0 -66
- package/lib/cjs/activities/renditions/generateImageRendition.js.map +0 -1
- package/lib/cjs/activities/renditions/generateVideoRendition.js +0 -200
- package/lib/cjs/activities/renditions/generateVideoRendition.js.map +0 -1
- package/lib/cjs/activities/resolveRemoteActivities.js +0 -133
- package/lib/cjs/activities/resolveRemoteActivities.js.map +0 -1
- package/lib/cjs/activities/setDocumentStatus.js +0 -26
- package/lib/cjs/activities/setDocumentStatus.js.map +0 -1
- package/lib/cjs/bulk-import.js +0 -12
- package/lib/cjs/bulk-import.js.map +0 -1
- package/lib/cjs/conversion/TextractProcessor.js +0 -417
- package/lib/cjs/conversion/TextractProcessor.js.map +0 -1
- package/lib/cjs/conversion/image.js +0 -149
- package/lib/cjs/conversion/image.js.map +0 -1
- package/lib/cjs/conversion/markitdown.js +0 -42
- package/lib/cjs/conversion/markitdown.js.map +0 -1
- package/lib/cjs/conversion/mutool.js +0 -147
- package/lib/cjs/conversion/mutool.js.map +0 -1
- package/lib/cjs/conversion/pandoc.js +0 -39
- package/lib/cjs/conversion/pandoc.js.map +0 -1
- package/lib/cjs/dsl/conditions.js +0 -81
- package/lib/cjs/dsl/conditions.js.map +0 -1
- package/lib/cjs/dsl/dsl-workflow.js +0 -458
- package/lib/cjs/dsl/dsl-workflow.js.map +0 -1
- package/lib/cjs/dsl/dslProxyActivities.js +0 -23
- package/lib/cjs/dsl/dslProxyActivities.js.map +0 -1
- package/lib/cjs/dsl/projections.js +0 -59
- package/lib/cjs/dsl/projections.js.map +0 -1
- package/lib/cjs/dsl/setup/ActivityContext.js +0 -188
- package/lib/cjs/dsl/setup/ActivityContext.js.map +0 -1
- package/lib/cjs/dsl/setup/fetch/DataProvider.js +0 -51
- package/lib/cjs/dsl/setup/fetch/DataProvider.js.map +0 -1
- package/lib/cjs/dsl/setup/fetch/index.js +0 -16
- package/lib/cjs/dsl/setup/fetch/index.js.map +0 -1
- package/lib/cjs/dsl/setup/fetch/providers.js +0 -67
- package/lib/cjs/dsl/setup/fetch/providers.js.map +0 -1
- package/lib/cjs/dsl/test/test-child-workflow.js +0 -10
- package/lib/cjs/dsl/test/test-child-workflow.js.map +0 -1
- package/lib/cjs/dsl/validation.js +0 -122
- package/lib/cjs/dsl/validation.js.map +0 -1
- package/lib/cjs/dsl/vars.js +0 -341
- package/lib/cjs/dsl/vars.js.map +0 -1
- package/lib/cjs/dsl/walk.js +0 -100
- package/lib/cjs/dsl/walk.js.map +0 -1
- package/lib/cjs/dsl.js +0 -20
- package/lib/cjs/dsl.js.map +0 -1
- package/lib/cjs/errors.js +0 -89
- package/lib/cjs/errors.js.map +0 -1
- package/lib/cjs/index.js +0 -57
- package/lib/cjs/index.js.map +0 -1
- package/lib/cjs/iterative-generation/activities/extractToc.js +0 -47
- package/lib/cjs/iterative-generation/activities/extractToc.js.map +0 -1
- package/lib/cjs/iterative-generation/activities/finalizeOutput.js +0 -72
- package/lib/cjs/iterative-generation/activities/finalizeOutput.js.map +0 -1
- package/lib/cjs/iterative-generation/activities/generatePart.js +0 -78
- package/lib/cjs/iterative-generation/activities/generatePart.js.map +0 -1
- package/lib/cjs/iterative-generation/activities/generateToc.js +0 -86
- package/lib/cjs/iterative-generation/activities/generateToc.js.map +0 -1
- package/lib/cjs/iterative-generation/activities/index.js +0 -12
- package/lib/cjs/iterative-generation/activities/index.js.map +0 -1
- package/lib/cjs/iterative-generation/iterativeGenerationWorkflow.js +0 -56
- package/lib/cjs/iterative-generation/iterativeGenerationWorkflow.js.map +0 -1
- package/lib/cjs/iterative-generation/types.js +0 -5
- package/lib/cjs/iterative-generation/types.js.map +0 -1
- package/lib/cjs/iterative-generation/utils.js +0 -121
- package/lib/cjs/iterative-generation/utils.js.map +0 -1
- package/lib/cjs/package.json +0 -3
- package/lib/cjs/result-types.js +0 -10
- package/lib/cjs/result-types.js.map +0 -1
- package/lib/cjs/security/ssrf.js +0 -34
- package/lib/cjs/security/ssrf.js.map +0 -1
- package/lib/cjs/system/notifyWebhookWorkflow.js +0 -53
- package/lib/cjs/system/notifyWebhookWorkflow.js.map +0 -1
- package/lib/cjs/system/recalculateEmbeddingsWorkflow.js +0 -33
- package/lib/cjs/system/recalculateEmbeddingsWorkflow.js.map +0 -1
- package/lib/cjs/utils/auth.js +0 -15
- package/lib/cjs/utils/auth.js.map +0 -1
- package/lib/cjs/utils/blobs.js +0 -64
- package/lib/cjs/utils/blobs.js.map +0 -1
- package/lib/cjs/utils/chunks.js +0 -14
- package/lib/cjs/utils/chunks.js.map +0 -1
- package/lib/cjs/utils/client.js +0 -31
- package/lib/cjs/utils/client.js.map +0 -1
- package/lib/cjs/utils/expand-vars.js +0 -33
- package/lib/cjs/utils/expand-vars.js.map +0 -1
- package/lib/cjs/utils/memory.js +0 -65
- package/lib/cjs/utils/memory.js.map +0 -1
- package/lib/cjs/utils/renditions.js +0 -92
- package/lib/cjs/utils/renditions.js.map +0 -1
- package/lib/cjs/utils/storage.js +0 -54
- package/lib/cjs/utils/storage.js.map +0 -1
- package/lib/cjs/utils/text-preview-utils.js +0 -43
- package/lib/cjs/utils/text-preview-utils.js.map +0 -1
- package/lib/cjs/utils/tokens.js +0 -38
- package/lib/cjs/utils/tokens.js.map +0 -1
- package/lib/cjs/vars.js +0 -20
- package/lib/cjs/vars.js.map +0 -1
- package/lib/cjs/workflows.js +0 -15
- package/lib/cjs/workflows.js.map +0 -1
- package/lib/esm/activities/advanced/createDocumentTypeFromInteractionRun.js.map +0 -1
- package/lib/esm/activities/advanced/createOrUpdateDocumentFromInteractionRun.js.map +0 -1
- package/lib/esm/activities/advanced/updateDocumentFromInteractionRun.js +0 -16
- package/lib/esm/activities/advanced/updateDocumentFromInteractionRun.js.map +0 -1
- package/lib/esm/activities/chunkDocument.js.map +0 -1
- package/lib/esm/activities/createDocumentFromOther.js.map +0 -1
- package/lib/esm/activities/executeInteraction.js.map +0 -1
- package/lib/esm/activities/executeRemoteActivity.js.map +0 -1
- package/lib/esm/activities/extractDocumentText.js.map +0 -1
- package/lib/esm/activities/generateDocumentProperties.js.map +0 -1
- package/lib/esm/activities/generateEmbeddings.js.map +0 -1
- package/lib/esm/activities/generateOrAssignContentType.js.map +0 -1
- package/lib/esm/activities/getObjectFromStore.js.map +0 -1
- package/lib/esm/activities/handleError.js +0 -26
- package/lib/esm/activities/handleError.js.map +0 -1
- package/lib/esm/activities/index-dsl.js +0 -29
- package/lib/esm/activities/index-dsl.js.map +0 -1
- package/lib/esm/activities/index.js.map +0 -1
- package/lib/esm/activities/loadChildWorkflowSpec.js.map +0 -1
- package/lib/esm/activities/media/prepareAudio.js.map +0 -1
- package/lib/esm/activities/media/prepareVideo.js.map +0 -1
- package/lib/esm/activities/media/probeMediaStreams.js.map +0 -1
- package/lib/esm/activities/media/processPdfWithTextract.js.map +0 -1
- package/lib/esm/activities/media/saveGladiaTranscription.js.map +0 -1
- package/lib/esm/activities/media/transcribeMediaWithGladia.js.map +0 -1
- package/lib/esm/activities/mergeChildArtifacts.js.map +0 -1
- package/lib/esm/activities/notifyWebhook.js.map +0 -1
- package/lib/esm/activities/rateLimiter.js.map +0 -1
- package/lib/esm/activities/renditions/generateImageRendition.js +0 -63
- package/lib/esm/activities/renditions/generateImageRendition.js.map +0 -1
- package/lib/esm/activities/renditions/generateVideoRendition.js.map +0 -1
- package/lib/esm/activities/resolveRemoteActivities.js.map +0 -1
- package/lib/esm/activities/setDocumentStatus.js.map +0 -1
- package/lib/esm/bulk-import.js.map +0 -1
- package/lib/esm/conversion/TextractProcessor.js.map +0 -1
- package/lib/esm/conversion/image.js.map +0 -1
- package/lib/esm/conversion/markitdown.js.map +0 -1
- package/lib/esm/conversion/mutool.js.map +0 -1
- package/lib/esm/conversion/pandoc.js.map +0 -1
- package/lib/esm/dsl/conditions.js +0 -75
- package/lib/esm/dsl/conditions.js.map +0 -1
- package/lib/esm/dsl/dsl-workflow.js.map +0 -1
- package/lib/esm/dsl/dslProxyActivities.js.map +0 -1
- package/lib/esm/dsl/projections.js.map +0 -1
- package/lib/esm/dsl/setup/ActivityContext.js.map +0 -1
- package/lib/esm/dsl/setup/fetch/DataProvider.js.map +0 -1
- package/lib/esm/dsl/setup/fetch/index.js.map +0 -1
- package/lib/esm/dsl/setup/fetch/providers.js.map +0 -1
- package/lib/esm/dsl/test/test-child-workflow.js.map +0 -1
- package/lib/esm/dsl/validation.js.map +0 -1
- package/lib/esm/dsl/vars.js.map +0 -1
- package/lib/esm/dsl/walk.js.map +0 -1
- package/lib/esm/dsl.js +0 -4
- package/lib/esm/dsl.js.map +0 -1
- package/lib/esm/errors.js.map +0 -1
- package/lib/esm/index.js +0 -39
- package/lib/esm/index.js.map +0 -1
- package/lib/esm/iterative-generation/activities/extractToc.js +0 -44
- package/lib/esm/iterative-generation/activities/extractToc.js.map +0 -1
- package/lib/esm/iterative-generation/activities/finalizeOutput.js +0 -69
- package/lib/esm/iterative-generation/activities/finalizeOutput.js.map +0 -1
- package/lib/esm/iterative-generation/activities/generatePart.js +0 -75
- package/lib/esm/iterative-generation/activities/generatePart.js.map +0 -1
- package/lib/esm/iterative-generation/activities/generateToc.js +0 -83
- package/lib/esm/iterative-generation/activities/generateToc.js.map +0 -1
- package/lib/esm/iterative-generation/activities/index.js +0 -5
- package/lib/esm/iterative-generation/activities/index.js.map +0 -1
- package/lib/esm/iterative-generation/iterativeGenerationWorkflow.js +0 -53
- package/lib/esm/iterative-generation/iterativeGenerationWorkflow.js.map +0 -1
- package/lib/esm/iterative-generation/types.js +0 -2
- package/lib/esm/iterative-generation/types.js.map +0 -1
- package/lib/esm/iterative-generation/utils.js +0 -112
- package/lib/esm/iterative-generation/utils.js.map +0 -1
- package/lib/esm/result-types.js.map +0 -1
- package/lib/esm/security/ssrf.js.map +0 -1
- package/lib/esm/system/notifyWebhookWorkflow.js.map +0 -1
- package/lib/esm/system/recalculateEmbeddingsWorkflow.js.map +0 -1
- package/lib/esm/utils/auth.js.map +0 -1
- package/lib/esm/utils/blobs.js.map +0 -1
- package/lib/esm/utils/chunks.js.map +0 -1
- package/lib/esm/utils/client.js +0 -27
- package/lib/esm/utils/client.js.map +0 -1
- package/lib/esm/utils/expand-vars.js.map +0 -1
- package/lib/esm/utils/memory.js +0 -55
- package/lib/esm/utils/memory.js.map +0 -1
- package/lib/esm/utils/renditions.js.map +0 -1
- package/lib/esm/utils/storage.js.map +0 -1
- package/lib/esm/utils/text-preview-utils.js.map +0 -1
- package/lib/esm/utils/tokens.js.map +0 -1
- package/lib/esm/vars.js +0 -4
- package/lib/esm/vars.js.map +0 -1
- package/lib/esm/workflows.js +0 -8
- package/lib/esm/workflows.js.map +0 -1
- package/lib/tsconfig.tsbuildinfo +0 -1
- package/lib/types/activities/advanced/createDocumentTypeFromInteractionRun.d.ts.map +0 -1
- package/lib/types/activities/advanced/createOrUpdateDocumentFromInteractionRun.d.ts.map +0 -1
- package/lib/types/activities/advanced/updateDocumentFromInteractionRun.d.ts.map +0 -1
- package/lib/types/activities/chunkDocument.d.ts.map +0 -1
- package/lib/types/activities/createDocumentFromOther.d.ts.map +0 -1
- package/lib/types/activities/executeInteraction.d.ts.map +0 -1
- package/lib/types/activities/executeRemoteActivity.d.ts.map +0 -1
- package/lib/types/activities/extractDocumentText.d.ts.map +0 -1
- package/lib/types/activities/generateDocumentProperties.d.ts.map +0 -1
- package/lib/types/activities/generateEmbeddings.d.ts.map +0 -1
- package/lib/types/activities/generateOrAssignContentType.d.ts.map +0 -1
- package/lib/types/activities/getObjectFromStore.d.ts.map +0 -1
- package/lib/types/activities/handleError.d.ts.map +0 -1
- package/lib/types/activities/index-dsl.d.ts +0 -31
- package/lib/types/activities/index-dsl.d.ts.map +0 -1
- package/lib/types/activities/index.d.ts.map +0 -1
- package/lib/types/activities/loadChildWorkflowSpec.d.ts.map +0 -1
- package/lib/types/activities/media/prepareAudio.d.ts.map +0 -1
- package/lib/types/activities/media/prepareVideo.d.ts.map +0 -1
- package/lib/types/activities/media/probeMediaStreams.d.ts.map +0 -1
- package/lib/types/activities/media/processPdfWithTextract.d.ts.map +0 -1
- package/lib/types/activities/media/saveGladiaTranscription.d.ts.map +0 -1
- package/lib/types/activities/media/transcribeMediaWithGladia.d.ts.map +0 -1
- package/lib/types/activities/mergeChildArtifacts.d.ts.map +0 -1
- package/lib/types/activities/notifyWebhook.d.ts.map +0 -1
- package/lib/types/activities/rateLimiter.d.ts.map +0 -1
- package/lib/types/activities/renditions/generateImageRendition.d.ts.map +0 -1
- package/lib/types/activities/renditions/generateVideoRendition.d.ts.map +0 -1
- package/lib/types/activities/resolveRemoteActivities.d.ts.map +0 -1
- package/lib/types/activities/setDocumentStatus.d.ts.map +0 -1
- package/lib/types/bulk-import.d.ts.map +0 -1
- package/lib/types/conversion/TextractProcessor.d.ts.map +0 -1
- package/lib/types/conversion/image.d.ts.map +0 -1
- package/lib/types/conversion/markitdown.d.ts.map +0 -1
- package/lib/types/conversion/mutool.d.ts.map +0 -1
- package/lib/types/conversion/pandoc.d.ts.map +0 -1
- package/lib/types/dsl/conditions.d.ts +0 -2
- package/lib/types/dsl/conditions.d.ts.map +0 -1
- package/lib/types/dsl/dsl-workflow.d.ts.map +0 -1
- package/lib/types/dsl/dslProxyActivities.d.ts +0 -10
- package/lib/types/dsl/dslProxyActivities.d.ts.map +0 -1
- package/lib/types/dsl/projections.d.ts +0 -4
- package/lib/types/dsl/projections.d.ts.map +0 -1
- package/lib/types/dsl/setup/ActivityContext.d.ts.map +0 -1
- package/lib/types/dsl/setup/fetch/DataProvider.d.ts.map +0 -1
- package/lib/types/dsl/setup/fetch/index.d.ts +0 -6
- package/lib/types/dsl/setup/fetch/index.d.ts.map +0 -1
- package/lib/types/dsl/setup/fetch/providers.d.ts.map +0 -1
- package/lib/types/dsl/test/test-child-workflow.d.ts +0 -4
- package/lib/types/dsl/test/test-child-workflow.d.ts.map +0 -1
- package/lib/types/dsl/validation.d.ts.map +0 -1
- package/lib/types/dsl/vars.d.ts.map +0 -1
- package/lib/types/dsl/walk.d.ts +0 -18
- package/lib/types/dsl/walk.d.ts.map +0 -1
- package/lib/types/dsl.d.ts +0 -4
- package/lib/types/dsl.d.ts.map +0 -1
- package/lib/types/errors.d.ts.map +0 -1
- package/lib/types/index.d.ts +0 -38
- package/lib/types/index.d.ts.map +0 -1
- package/lib/types/iterative-generation/activities/extractToc.d.ts +0 -10
- package/lib/types/iterative-generation/activities/extractToc.d.ts.map +0 -1
- package/lib/types/iterative-generation/activities/finalizeOutput.d.ts +0 -3
- package/lib/types/iterative-generation/activities/finalizeOutput.d.ts.map +0 -1
- package/lib/types/iterative-generation/activities/generatePart.d.ts +0 -3
- package/lib/types/iterative-generation/activities/generatePart.d.ts.map +0 -1
- package/lib/types/iterative-generation/activities/generateToc.d.ts +0 -4
- package/lib/types/iterative-generation/activities/generateToc.d.ts.map +0 -1
- package/lib/types/iterative-generation/activities/index.d.ts +0 -5
- package/lib/types/iterative-generation/activities/index.d.ts.map +0 -1
- package/lib/types/iterative-generation/iterativeGenerationWorkflow.d.ts +0 -3
- package/lib/types/iterative-generation/iterativeGenerationWorkflow.d.ts.map +0 -1
- package/lib/types/iterative-generation/types.d.ts +0 -79
- package/lib/types/iterative-generation/types.d.ts.map +0 -1
- package/lib/types/iterative-generation/utils.d.ts +0 -26
- package/lib/types/iterative-generation/utils.d.ts.map +0 -1
- package/lib/types/result-types.d.ts.map +0 -1
- package/lib/types/security/ssrf.d.ts.map +0 -1
- package/lib/types/system/notifyWebhookWorkflow.d.ts +0 -8
- package/lib/types/system/notifyWebhookWorkflow.d.ts.map +0 -1
- package/lib/types/system/recalculateEmbeddingsWorkflow.d.ts.map +0 -1
- package/lib/types/utils/auth.d.ts.map +0 -1
- package/lib/types/utils/blobs.d.ts.map +0 -1
- package/lib/types/utils/chunks.d.ts.map +0 -1
- package/lib/types/utils/client.d.ts +0 -8
- package/lib/types/utils/client.d.ts.map +0 -1
- package/lib/types/utils/expand-vars.d.ts.map +0 -1
- package/lib/types/utils/memory.d.ts +0 -8
- package/lib/types/utils/memory.d.ts.map +0 -1
- package/lib/types/utils/renditions.d.ts.map +0 -1
- package/lib/types/utils/storage.d.ts.map +0 -1
- package/lib/types/utils/text-preview-utils.d.ts.map +0 -1
- package/lib/types/utils/tokens.d.ts.map +0 -1
- package/lib/types/vars.d.ts +0 -3
- package/lib/types/vars.d.ts.map +0 -1
- package/lib/types/workflows.d.ts +0 -8
- package/lib/types/workflows.d.ts.map +0 -1
- package/src/iterative-generation/activities/extractToc.ts +0 -63
- package/src/iterative-generation/activities/finalizeOutput.ts +0 -100
- package/src/iterative-generation/activities/generatePart.ts +0 -123
- package/src/iterative-generation/activities/generateToc.ts +0 -116
- package/src/iterative-generation/activities/index.ts +0 -4
- package/src/iterative-generation/iterativeGenerationWorkflow.ts +0 -68
- package/src/iterative-generation/types.ts +0 -99
- package/src/iterative-generation/utils.ts +0 -126
- package/src/utils/memory.ts +0 -61
- /package/lib/{esm/bulk-import.js → bulk-import.js} +0 -0
- /package/lib/{types/conversion → conversion}/image.d.ts +0 -0
- /package/lib/{types/conversion → conversion}/markitdown.d.ts +0 -0
- /package/lib/{types/conversion → conversion}/mutool.d.ts +0 -0
- /package/lib/{types/conversion → conversion}/pandoc.d.ts +0 -0
- /package/lib/{types/result-types.d.ts → result-types.d.ts} +0 -0
- /package/lib/{esm/result-types.js → result-types.js} +0 -0
- /package/lib/{types/security → security}/ssrf.d.ts +0 -0
- /package/lib/{esm/security → security}/ssrf.js +0 -0
- /package/lib/{esm/utils → utils}/auth.js +0 -0
- /package/lib/{types/utils → utils}/chunks.d.ts +0 -0
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"setDocumentStatus.js","sourceRoot":"","sources":["../../src/activities/setDocumentStatus.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAC;AAE3C,OAAO,EAAE,aAAa,EAAE,MAAM,iCAAiC,CAAC;AAWhE;;;;GAIG;AACH,MAAM,CAAC,KAAK,UAAU,iBAAiB,CAAC,OAA6D;IACjG,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,GAAG,MAAM,aAAa,CAA0B,OAAO,CAAC,CAAC;IAE3F,IAAI,CAAC;QACD,MAAM,GAAG,GAAG,MAAM,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,QAAQ,EAAE,EAAE,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,CAAC,CAAC;QAC7E,OAAO,GAAG,CAAC,MAAM,CAAC;IACtB,CAAC;IAAC,OAAO,GAAY,EAAE,CAAC;QACpB,wEAAwE;QACxE,MAAM,MAAM,GAAG,GAAG,IAAI,OAAO,GAAG,KAAK,QAAQ,IAAI,QAAQ,IAAI,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,SAAS,CAAC;QAC1F,MAAM,IAAI,GAAG,GAAG,YAAY,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,SAAS,CAAC;QACzD,IAAI,MAAM,KAAK,GAAG,IAAI,IAAI,KAAK,yBAAyB,EAAE,CAAC;YACvD,GAAG,CAAC,IAAI,CACJ,YAAY,QAAQ,kEAAkE,MAAM,CAAC,MAAM,GAAG,CACzG,CAAC;YACF,OAAO,SAAS,CAAC,CAAC,oCAAoC;QAC1D,CAAC;QACD,MAAM,GAAG,CAAC;IACd,CAAC;AACL,CAAC"}
|
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
import type { CreateCollectionPayload, CreateContentObjectPayload } from '@vertesia/common';
|
|
1
|
+
import type { ContentObjectProcessingPriority, CreateCollectionPayload, CreateContentObjectPayload } from '@vertesia/common';
|
|
2
2
|
export declare enum ItemTypes {
|
|
3
3
|
CONTENT_OBJECT = "ContentObject",
|
|
4
4
|
COLLECTION = "Collection",
|
|
@@ -74,10 +74,19 @@ export interface BulkImportParams {
|
|
|
74
74
|
partitionSize?: number;
|
|
75
75
|
dryRun?: boolean;
|
|
76
76
|
updateByContentSource?: boolean;
|
|
77
|
+
skipWorkflows?: boolean;
|
|
78
|
+
/**
|
|
79
|
+
* Processing priority for the document-processing workflows triggered by created objects.
|
|
80
|
+
* Defaults to `low` so bulk imports run on the low-priority ("bulk") task queue and don't
|
|
81
|
+
* compete with interactive traffic.
|
|
82
|
+
*/
|
|
83
|
+
processingPriority?: ContentObjectProcessingPriority;
|
|
77
84
|
}
|
|
78
85
|
export interface PartitionError {
|
|
79
86
|
partitionIndex: number;
|
|
80
87
|
errorCount: number;
|
|
88
|
+
/** Workflow-level error message if the partition itself failed (vs. per-batch failures). */
|
|
89
|
+
message?: string;
|
|
81
90
|
}
|
|
82
91
|
export interface BulkImportResult {
|
|
83
92
|
totalItems: number;
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"bulk-import.d.ts","sourceRoot":"","sources":["../src/bulk-import.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EACR,+BAA+B,EAC/B,uBAAuB,EACvB,0BAA0B,EAC7B,MAAM,kBAAkB,CAAC;AAE1B,oBAAY,SAAS;IACjB,cAAc,kBAAkB;IAChC,UAAU,eAAe;IACzB,QAAQ,aAAa;IACrB,cAAc,kBAAkB;IAChC,IAAI,SAAS;CAChB;AAED,MAAM,WAAW,QAAQ;IACrB,IAAI,EAAE,SAAS,CAAC;CACnB;AAED,MAAM,WAAW,uBAAwB,SAAQ,QAAQ;IACrD,IAAI,EAAE,SAAS,CAAC,cAAc,GAAG,SAAS,CAAC,IAAI,CAAC;IAChD,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,IAAI,EAAE,0BAA0B,CAAC;IACjC,OAAO,EAAE;QACL,aAAa,CAAC,EAAE,MAAM,CAAC;KAC1B,CAAC;CACL;AAED,MAAM,WAAW,oBAAqB,SAAQ,QAAQ;IAClD,IAAI,EAAE,SAAS,CAAC,UAAU,CAAC;IAC3B,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,IAAI,EAAE,uBAAuB,CAAC;CACjC;AAED,MAAM,WAAW,kBAAmB,SAAQ,QAAQ;IAChD,IAAI,EAAE,SAAS,CAAC,QAAQ,CAAC;IACzB,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,IAAI,EAAE,IAAI,CAAC,0BAA0B,EAAE,SAAS,CAAC,CAAC;IAClD,OAAO,EAAE;QACL,aAAa,CAAC,EAAE,MAAM,CAAC;KAC1B,CAAC;CACL;AAED,MAAM,WAAW,uBAAwB,SAAQ,QAAQ;IACrD,IAAI,EAAE,SAAS,CAAC,cAAc,CAAC;IAC/B,QAAQ,EAAE,MAAM,CAAC;IACjB,UAAU,EAAE,MAAM,CAAC;IACnB,SAAS,EAAE,MAAM,CAAC;IAClB,QAAQ,EAAE,MAAM,CAAC;CACpB;AAED,MAAM,MAAM,UAAU,GAAG,uBAAuB,GAAG,oBAAoB,GAAG,kBAAkB,GAAG,uBAAuB,CAAC;AAEvH,MAAM,WAAW,eAAe;IAC5B,KAAK,EAAE,MAAM,CAAC;IACd,KAAK,EAAE,UAAU,EAAE,CAAC;CACvB;AAED,MAAM,WAAW,iBAAiB;IAC9B,OAAO,EAAE,eAAe,EAAE,CAAC;CAC9B;AAED,MAAM,WAAW,qBAAqB;IAClC,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,aAAa,EAAE,MAAM,CAAC;IACtB,WAAW,EAAE,MAAM,CAAC;CACvB;AAED,MAAM,WAAW,kBAAkB,CAAC,CAAC,SAAS,QAAQ,GAAG,UAAU;IAC/D,EAAE,EAAE,MAAM,CAAC;IACX,IAAI,EAAE,CAAC,CAAC;IACR,QAAQ,EAAE,OAAO,CAAC;CACrB;AAED,MAAM,WAAW,yBAAyB;IACtC,aAAa,EAAE,MAAM,CAAC;CACzB;AAED,MAAM,WAAW,SAAS;IACtB,IAAI,EAAE,MAAM,CAAC;IACb,aAAa,EAAE,MAAM,CAAC;IACtB,WAAW,EAAE,MAAM,CAAC;CACvB;AAED,MAAM,WAAW,UAAU;IACvB,SAAS,EAAE,MAAM,CAAC;IAClB,UAAU,EAAE,SAAS,EAAE,CAAC;CAC3B;AAED,MAAM,WAAW,gBAAgB;IAC7B,oBAAoB,CAAC,EAAE,MAAM,CAAC;IAC9B,uBAAuB,CAAC,EAAE,MAAM,CAAC;IACjC,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB,qBAAqB,CAAC,EAAE,OAAO,CAAC;IAChC,aAAa,CAAC,EAAE,OAAO,CAAC;IACxB;;;;OAIG;IACH,kBAAkB,CAAC,EAAE,+BAA+B,CAAC;CACxD;AAED,MAAM,WAAW,cAAc;IAC3B,cAAc,EAAE,MAAM,CAAC;IACvB,UAAU,EAAE,MAAM,CAAC;IACnB,4FAA4F;IAC5F,OAAO,CAAC,EAAE,MAAM,CAAC;CACpB;AAED,MAAM,WAAW,gBAAgB;IAC7B,UAAU,EAAE,MAAM,CAAC;IACnB,cAAc,EAAE,MAAM,CAAC;IACvB,YAAY,EAAE,MAAM,CAAC;IACrB,YAAY,EAAE,MAAM,CAAC;IACrB,YAAY,EAAE,MAAM,CAAC;IACrB,gBAAgB,EAAE,MAAM,CAAC;IACzB,MAAM,EAAE,cAAc,EAAE,CAAC;IACzB,SAAS,EAAE,MAAM,CAAC;CACrB;AAED,MAAM,WAAW,UAAU;IACvB,UAAU,EAAE,MAAM,CAAC;IACnB,KAAK,EAAE,MAAM,CAAC;IACd,IAAI,EAAE,UAAU,CAAC;IACjB,GAAG,EAAE,MAAM,CAAC;CACf;AAED,MAAM,WAAW,gBAAgB;IAC7B,aAAa,EAAE,kBAAkB,EAAE,CAAC;IACpC,MAAM,EAAE,UAAU,EAAE,CAAC;CACxB"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"bulk-import.js","sourceRoot":"","sources":["../src/bulk-import.ts"],"names":[],"mappings":"AAMA,MAAM,CAAN,IAAY,SAMX;AAND,WAAY,SAAS;IACjB,6CAAgC,CAAA;IAChC,sCAAyB,CAAA;IACzB,kCAAqB,CAAA;IACrB,6CAAgC,CAAA;IAChC,0BAAa,CAAA;AACjB,CAAC,EANW,SAAS,KAAT,SAAS,QAMpB"}
|
|
@@ -1,28 +1,31 @@
|
|
|
1
|
-
import type { AwsCredentialIdentityProvider } from
|
|
1
|
+
import type { AwsCredentialIdentityProvider } from '@smithy/types';
|
|
2
2
|
interface TextractProcessorOptions {
|
|
3
3
|
fileKey: string;
|
|
4
4
|
region: string;
|
|
5
5
|
bucket: string;
|
|
6
6
|
credentials?: AwsCredentialIdentityProvider;
|
|
7
|
-
log?:
|
|
7
|
+
log?: TextractLogger;
|
|
8
8
|
detectImages?: boolean;
|
|
9
9
|
/**
|
|
10
10
|
* NEW: If true, includes cell-confidence information in the table CSV
|
|
11
11
|
*/
|
|
12
12
|
includeConfidenceInTables?: boolean;
|
|
13
13
|
}
|
|
14
|
+
interface TextractLogger {
|
|
15
|
+
info(message: string, metadata?: Record<string, unknown>): void;
|
|
16
|
+
}
|
|
14
17
|
export declare class TextractProcessor {
|
|
15
18
|
private textractClient;
|
|
16
19
|
private s3Client;
|
|
17
20
|
private fileKey;
|
|
18
21
|
private bucket;
|
|
19
|
-
private log
|
|
22
|
+
private log?;
|
|
20
23
|
private detectImages;
|
|
21
24
|
/**
|
|
22
25
|
* Whether or not to include confidence values in CSV output for tables.
|
|
23
26
|
*/
|
|
24
27
|
private includeConfidenceInTables;
|
|
25
|
-
constructor({ fileKey, region, bucket, credentials, log, detectImages, includeConfidenceInTables }: TextractProcessorOptions);
|
|
28
|
+
constructor({ fileKey, region, bucket, credentials, log, detectImages, includeConfidenceInTables, }: TextractProcessorOptions);
|
|
26
29
|
private getText;
|
|
27
30
|
private isBlockInTable;
|
|
28
31
|
private isWordInTableCell;
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"TextractProcessor.d.ts","sourceRoot":"","sources":["../../src/conversion/TextractProcessor.ts"],"names":[],"mappings":"AAGA,OAAO,KAAK,EAAE,6BAA6B,EAAE,MAAM,eAAe,CAAC;AAuBnE,UAAU,wBAAwB;IAC9B,OAAO,EAAE,MAAM,CAAC;IAChB,MAAM,EAAE,MAAM,CAAC;IACf,MAAM,EAAE,MAAM,CAAC;IACf,WAAW,CAAC,EAAE,6BAA6B,CAAC;IAC5C,GAAG,CAAC,EAAE,cAAc,CAAC;IACrB,YAAY,CAAC,EAAE,OAAO,CAAC;IACvB;;OAEG;IACH,yBAAyB,CAAC,EAAE,OAAO,CAAC;CACvC;AAED,UAAU,cAAc;IACpB,IAAI,CAAC,OAAO,EAAE,MAAM,EAAE,QAAQ,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,IAAI,CAAC;CACnE;AAED,qBAAa,iBAAiB;IAC1B,OAAO,CAAC,cAAc,CAAiB;IACvC,OAAO,CAAC,QAAQ,CAAW;IAC3B,OAAO,CAAC,OAAO,CAAS;IACxB,OAAO,CAAC,MAAM,CAAS;IACvB,OAAO,CAAC,GAAG,CAAC,CAAiB;IAC7B,OAAO,CAAC,YAAY,CAAU;IAC9B;;OAEG;IACH,OAAO,CAAC,yBAAyB,CAAU;gBAE/B,EACR,OAAO,EACP,MAAM,EACN,MAAM,EACN,WAAW,EACX,GAAG,EACH,YAAoB,EACpB,yBAAiC,GACpC,EAAE,wBAAwB;IAiB3B,OAAO,CAAC,OAAO;IA0Bf,OAAO,CAAC,cAAc;IAmBtB,OAAO,CAAC,iBAAiB;IA4BzB;;OAEG;IACH,OAAO,CAAC,iBAAiB;IAuCzB,OAAO,CAAC,gBAAgB;IA0ClB,MAAM,CAAC,OAAO,EAAE,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC;IAUtC,aAAa,CAAC,KAAK,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC;IAe7C,cAAc,CAAC,KAAK,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC;IAOpD,OAAO,CAAC,mBAAmB;IAoB3B,OAAO,CAAC,mBAAmB;IAO3B,OAAO,CAAC,cAAc;IAQtB,OAAO,CAAC,eAAe;IAWvB,OAAO,CAAC,gBAAgB;IAYlB,cAAc,CAAC,KAAK,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC;CAgLvD"}
|
|
@@ -1,5 +1,5 @@
|
|
|
1
|
-
import { PutObjectCommand, S3Client } from
|
|
2
|
-
import { GetDocumentAnalysisCommand, StartDocumentAnalysisCommand, TextractClient } from
|
|
1
|
+
import { PutObjectCommand, S3Client } from '@aws-sdk/client-s3';
|
|
2
|
+
import { GetDocumentAnalysisCommand, StartDocumentAnalysisCommand, TextractClient } from '@aws-sdk/client-textract';
|
|
3
3
|
import Papa from 'papaparse';
|
|
4
4
|
export class TextractProcessor {
|
|
5
5
|
textractClient;
|
|
@@ -12,7 +12,7 @@ export class TextractProcessor {
|
|
|
12
12
|
* Whether or not to include confidence values in CSV output for tables.
|
|
13
13
|
*/
|
|
14
14
|
includeConfidenceInTables;
|
|
15
|
-
constructor({ fileKey, region, bucket, credentials, log, detectImages = false, includeConfidenceInTables = false // NEW default = false
|
|
15
|
+
constructor({ fileKey, region, bucket, credentials, log, detectImages = false, includeConfidenceInTables = false, // NEW default = false
|
|
16
16
|
}) {
|
|
17
17
|
this.fileKey = fileKey;
|
|
18
18
|
this.bucket = bucket;
|
|
@@ -21,11 +21,11 @@ export class TextractProcessor {
|
|
|
21
21
|
this.includeConfidenceInTables = includeConfidenceInTables;
|
|
22
22
|
this.textractClient = new TextractClient({
|
|
23
23
|
region,
|
|
24
|
-
credentials
|
|
24
|
+
credentials,
|
|
25
25
|
});
|
|
26
26
|
this.s3Client = new S3Client({
|
|
27
27
|
region,
|
|
28
|
-
credentials
|
|
28
|
+
credentials,
|
|
29
29
|
});
|
|
30
30
|
}
|
|
31
31
|
getText(result, blocksMap) {
|
|
@@ -38,16 +38,14 @@ export class TextractProcessor {
|
|
|
38
38
|
if (word.BlockType === 'WORD') {
|
|
39
39
|
const wordText = word.Text || '';
|
|
40
40
|
// Example logic to quote numeric text with commas
|
|
41
|
-
if (wordText.includes(',') &&
|
|
42
|
-
wordText.replace(',', '').match(/^\d+$/)) {
|
|
41
|
+
if (wordText.includes(',') && wordText.replace(',', '').match(/^\d+$/)) {
|
|
43
42
|
text += `"${wordText}" `;
|
|
44
43
|
}
|
|
45
44
|
else {
|
|
46
45
|
text += `${wordText} `;
|
|
47
46
|
}
|
|
48
47
|
}
|
|
49
|
-
if (word.BlockType === 'SELECTION_ELEMENT' &&
|
|
50
|
-
word.SelectionStatus === 'SELECTED') {
|
|
48
|
+
if (word.BlockType === 'SELECTION_ELEMENT' && word.SelectionStatus === 'SELECTED') {
|
|
51
49
|
text += 'X ';
|
|
52
50
|
}
|
|
53
51
|
}
|
|
@@ -86,6 +84,7 @@ export class TextractProcessor {
|
|
|
86
84
|
if (cell.BlockType === 'CELL' && cell.Relationships) {
|
|
87
85
|
for (const cellRel of cell.Relationships) {
|
|
88
86
|
if (cellRel.Type === 'CHILD' &&
|
|
87
|
+
// biome-ignore lint/style/noNonNullAssertion: intentional non-null assertion; TS can't prove narrowing here
|
|
89
88
|
cellRel.Ids?.includes(wordBlock.Id)) {
|
|
90
89
|
return true;
|
|
91
90
|
}
|
|
@@ -103,9 +102,9 @@ export class TextractProcessor {
|
|
|
103
102
|
*/
|
|
104
103
|
getRowsColumnsMap(tableResult, blocksMap) {
|
|
105
104
|
const rows = [];
|
|
106
|
-
tableResult.Relationships?.forEach(relationship => {
|
|
105
|
+
tableResult.Relationships?.forEach((relationship) => {
|
|
107
106
|
if (relationship.Type === 'CHILD') {
|
|
108
|
-
relationship.Ids?.forEach(childId => {
|
|
107
|
+
relationship.Ids?.forEach((childId) => {
|
|
109
108
|
const cell = blocksMap[childId];
|
|
110
109
|
if (cell.BlockType === 'CELL') {
|
|
111
110
|
const rowIndex = cell.RowIndex || 1;
|
|
@@ -147,7 +146,7 @@ export class TextractProcessor {
|
|
|
147
146
|
csvData.push(rowData);
|
|
148
147
|
}
|
|
149
148
|
// Compute average confidence (or any other method you prefer)
|
|
150
|
-
const tableConfidence = cellCount > 0 ?
|
|
149
|
+
const tableConfidence = cellCount > 0 ? totalConfidence / cellCount : 0;
|
|
151
150
|
// Convert to CSV
|
|
152
151
|
const csv = Papa.unparse(csvData, {
|
|
153
152
|
delimiter: ',',
|
|
@@ -156,12 +155,12 @@ export class TextractProcessor {
|
|
|
156
155
|
escapeChar: '"',
|
|
157
156
|
header: false,
|
|
158
157
|
newline: '\n',
|
|
159
|
-
skipEmptyLines: false
|
|
158
|
+
skipEmptyLines: false,
|
|
160
159
|
});
|
|
161
160
|
return { csv, tableConfidence };
|
|
162
161
|
}
|
|
163
162
|
async upload(fileBuf) {
|
|
164
|
-
this.log
|
|
163
|
+
this.log?.info('Uploading file to S3', { fileKey: this.fileKey });
|
|
165
164
|
const command = new PutObjectCommand({
|
|
166
165
|
Bucket: this.bucket,
|
|
167
166
|
Key: this.fileKey,
|
|
@@ -174,17 +173,19 @@ export class TextractProcessor {
|
|
|
174
173
|
DocumentLocation: {
|
|
175
174
|
S3Object: {
|
|
176
175
|
Bucket: this.bucket,
|
|
177
|
-
Name: s3Key
|
|
178
|
-
}
|
|
176
|
+
Name: s3Key,
|
|
177
|
+
},
|
|
179
178
|
},
|
|
180
|
-
FeatureTypes: [
|
|
179
|
+
FeatureTypes: ['TABLES'],
|
|
181
180
|
});
|
|
182
181
|
const response = await this.textractClient.send(command);
|
|
182
|
+
// biome-ignore lint/style/noNonNullAssertion: intentional non-null assertion; TS can't prove narrowing here
|
|
183
183
|
return response.JobId;
|
|
184
184
|
}
|
|
185
185
|
async checkJobStatus(jobId) {
|
|
186
186
|
const command = new GetDocumentAnalysisCommand({ JobId: jobId });
|
|
187
187
|
const response = await this.textractClient.send(command);
|
|
188
|
+
// biome-ignore lint/style/noNonNullAssertion: intentional non-null assertion; TS can't prove narrowing here
|
|
188
189
|
return response.JobStatus;
|
|
189
190
|
}
|
|
190
191
|
getImagePlaceholder(block) {
|
|
@@ -221,8 +222,7 @@ export class TextractProcessor {
|
|
|
221
222
|
if (!prevBlock)
|
|
222
223
|
return true;
|
|
223
224
|
const gap = (block.Geometry?.BoundingBox?.Top || 0) -
|
|
224
|
-
((prevBlock.Geometry?.BoundingBox?.Top || 0) +
|
|
225
|
-
(prevBlock.Geometry?.BoundingBox?.Height || 0));
|
|
225
|
+
((prevBlock.Geometry?.BoundingBox?.Top || 0) + (prevBlock.Geometry?.BoundingBox?.Height || 0));
|
|
226
226
|
return gap > 0.03;
|
|
227
227
|
}
|
|
228
228
|
formatTextBlock(block, prevBlock) {
|
|
@@ -235,8 +235,7 @@ export class TextractProcessor {
|
|
|
235
235
|
return `${indent}${text}\n`;
|
|
236
236
|
}
|
|
237
237
|
shouldMergeLines(prev, current) {
|
|
238
|
-
const prevBottom = (prev.Geometry?.BoundingBox?.Top || 0)
|
|
239
|
-
+ (prev.Geometry?.BoundingBox?.Height || 0);
|
|
238
|
+
const prevBottom = (prev.Geometry?.BoundingBox?.Top || 0) + (prev.Geometry?.BoundingBox?.Height || 0);
|
|
240
239
|
const currentTop = current.Geometry?.BoundingBox?.Top || 0;
|
|
241
240
|
const gap = currentTop - prevBottom;
|
|
242
241
|
// For example, if gap < 0.02, treat them as contiguous
|
|
@@ -251,7 +250,7 @@ export class TextractProcessor {
|
|
|
251
250
|
do {
|
|
252
251
|
const command = new GetDocumentAnalysisCommand({
|
|
253
252
|
JobId: jobId,
|
|
254
|
-
NextToken: nextToken
|
|
253
|
+
NextToken: nextToken,
|
|
255
254
|
});
|
|
256
255
|
const response = await this.textractClient.send(command);
|
|
257
256
|
allBlocks = allBlocks.concat(response.Blocks || []);
|
|
@@ -260,13 +259,14 @@ export class TextractProcessor {
|
|
|
260
259
|
// Create blocks map
|
|
261
260
|
const blocksMap = {};
|
|
262
261
|
for (const block of allBlocks) {
|
|
262
|
+
// biome-ignore lint/style/noNonNullAssertion: intentional non-null assertion; TS can't prove narrowing here
|
|
263
263
|
blocksMap[block.Id] = block;
|
|
264
264
|
}
|
|
265
265
|
// We'll store each page's content in sequence
|
|
266
266
|
const pageContents = [];
|
|
267
267
|
let currentPage = null;
|
|
268
268
|
// We'll keep track of a "current text block" that we're building
|
|
269
|
-
let currentTextContent =
|
|
269
|
+
let currentTextContent = '';
|
|
270
270
|
let prevLineBlock = null;
|
|
271
271
|
// Sort by page and vertical position
|
|
272
272
|
allBlocks.sort((a, b) => {
|
|
@@ -280,7 +280,7 @@ export class TextractProcessor {
|
|
|
280
280
|
if (currentTextContent.trim().length > 0 && currentPage) {
|
|
281
281
|
currentPage.blocks.push({
|
|
282
282
|
type: 'text',
|
|
283
|
-
content: currentTextContent
|
|
283
|
+
content: currentTextContent,
|
|
284
284
|
});
|
|
285
285
|
}
|
|
286
286
|
if (currentPage) {
|
|
@@ -288,9 +288,9 @@ export class TextractProcessor {
|
|
|
288
288
|
}
|
|
289
289
|
currentPage = {
|
|
290
290
|
pageNumber: block.Page || 0,
|
|
291
|
-
blocks: []
|
|
291
|
+
blocks: [],
|
|
292
292
|
};
|
|
293
|
-
currentTextContent =
|
|
293
|
+
currentTextContent = '';
|
|
294
294
|
prevLineBlock = null;
|
|
295
295
|
}
|
|
296
296
|
else if (currentPage && block.Page === currentPage.pageNumber) {
|
|
@@ -300,15 +300,15 @@ export class TextractProcessor {
|
|
|
300
300
|
if (currentTextContent.trim().length > 0) {
|
|
301
301
|
currentPage.blocks.push({
|
|
302
302
|
type: 'text',
|
|
303
|
-
content: currentTextContent
|
|
303
|
+
content: currentTextContent,
|
|
304
304
|
});
|
|
305
|
-
currentTextContent =
|
|
305
|
+
currentTextContent = '';
|
|
306
306
|
}
|
|
307
|
-
const { csv, tableConfidence } = this.generateTableCSV(block, blocksMap, currentPage.blocks.filter(b => b.type === 'table').length + 1, currentPage.pageNumber);
|
|
307
|
+
const { csv, tableConfidence } = this.generateTableCSV(block, blocksMap, currentPage.blocks.filter((b) => b.type === 'table').length + 1, currentPage.pageNumber);
|
|
308
308
|
currentPage.blocks.push({
|
|
309
309
|
type: 'table',
|
|
310
310
|
content: csv,
|
|
311
|
-
confidence: tableConfidence
|
|
311
|
+
confidence: tableConfidence,
|
|
312
312
|
});
|
|
313
313
|
prevLineBlock = null;
|
|
314
314
|
}
|
|
@@ -321,15 +321,15 @@ export class TextractProcessor {
|
|
|
321
321
|
const formatted = this.formatTextBlock(block, prevLineBlock);
|
|
322
322
|
// formatTextBlock might include a leading newline if isLikelyHeader = true
|
|
323
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 +=
|
|
324
|
+
const mergedText = formatted.replace(/^\s*\n/, ' ');
|
|
325
|
+
currentTextContent += ` ${mergedText.trim()}`;
|
|
326
326
|
}
|
|
327
327
|
else {
|
|
328
328
|
// If there's an existing text block, push it
|
|
329
329
|
if (currentTextContent.trim().length > 0) {
|
|
330
330
|
currentPage.blocks.push({
|
|
331
331
|
type: 'text',
|
|
332
|
-
content: currentTextContent
|
|
332
|
+
content: currentTextContent,
|
|
333
333
|
});
|
|
334
334
|
}
|
|
335
335
|
// Start a new text block
|
|
@@ -340,16 +340,16 @@ export class TextractProcessor {
|
|
|
340
340
|
// IMAGES (if detectImages)
|
|
341
341
|
else if (this.detectImages) {
|
|
342
342
|
const geometry = block.Geometry?.BoundingBox;
|
|
343
|
-
if (geometry
|
|
343
|
+
if (geometry?.Width && geometry.Height) {
|
|
344
344
|
const imagePlaceholder = this.getImagePlaceholder(block);
|
|
345
345
|
if (imagePlaceholder) {
|
|
346
346
|
// If there's a pending text block, push it first
|
|
347
347
|
if (currentTextContent.trim().length > 0) {
|
|
348
348
|
currentPage.blocks.push({
|
|
349
349
|
type: 'text',
|
|
350
|
-
content: currentTextContent
|
|
350
|
+
content: currentTextContent,
|
|
351
351
|
});
|
|
352
|
-
currentTextContent =
|
|
352
|
+
currentTextContent = '';
|
|
353
353
|
}
|
|
354
354
|
currentPage.blocks.push({
|
|
355
355
|
type: 'image',
|
|
@@ -357,7 +357,7 @@ export class TextractProcessor {
|
|
|
357
357
|
left: geometry.Left,
|
|
358
358
|
top: geometry.Top,
|
|
359
359
|
width: geometry.Width,
|
|
360
|
-
height: geometry.Height
|
|
360
|
+
height: geometry.Height,
|
|
361
361
|
});
|
|
362
362
|
}
|
|
363
363
|
}
|
|
@@ -370,7 +370,7 @@ export class TextractProcessor {
|
|
|
370
370
|
if (currentTextContent.trim().length > 0) {
|
|
371
371
|
currentPage.blocks.push({
|
|
372
372
|
type: 'text',
|
|
373
|
-
content: currentTextContent
|
|
373
|
+
content: currentTextContent,
|
|
374
374
|
});
|
|
375
375
|
}
|
|
376
376
|
pageContents.push(currentPage);
|
|
@@ -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,EAAE,0BAA0B,EAAE,4BAA4B,EAAE,cAAc,EAAE,MAAM,0BAA0B,CAAC;AAEpH,OAAO,IAAI,MAAM,WAAW,CAAC;AAuC7B,MAAM,OAAO,iBAAiB;IAClB,cAAc,CAAiB;IAC/B,QAAQ,CAAW;IACnB,OAAO,CAAS;IAChB,MAAM,CAAS;IACf,GAAG,CAAkB;IACrB,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,EAAE,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,IAAI,QAAQ,CAAC,OAAO,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,EAAE,CAAC;gCACrE,IAAI,IAAI,IAAI,QAAQ,IAAI,CAAC;4BAC7B,CAAC;iCAAM,CAAC;gCACJ,IAAI,IAAI,GAAG,QAAQ,GAAG,CAAC;4BAC3B,CAAC;wBACL,CAAC;wBACD,IAAI,IAAI,CAAC,SAAS,KAAK,mBAAmB,IAAI,IAAI,CAAC,eAAe,KAAK,UAAU,EAAE,CAAC;4BAChF,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,4GAA4G;wCAC5G,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,CAAC,YAAY,EAAE,EAAE;YAChD,IAAI,YAAY,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;gBAChC,YAAY,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC,OAAO,EAAE,EAAE;oBAClC,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,eAAe,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC;QAExE,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,EAAE,IAAI,CAAC,sBAAsB,EAAE,EAAE,OAAO,EAAE,IAAI,CAAC,OAAO,EAAE,CAAC,CAAC;QAClE,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,4GAA4G;QAC5G,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,4GAA4G;QAC5G,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,GACL,CAAC,KAAK,CAAC,QAAQ,EAAE,WAAW,EAAE,GAAG,IAAI,CAAC,CAAC;YACvC,CAAC,CAAC,SAAS,CAAC,QAAQ,EAAE,WAAW,EAAE,GAAG,IAAI,CAAC,CAAC,GAAG,CAAC,SAAS,CAAC,QAAQ,EAAE,WAAW,EAAE,MAAM,IAAI,CAAC,CAAC,CAAC,CAAC;QACnG,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,GAAG,CAAC,IAAI,CAAC,QAAQ,EAAE,WAAW,EAAE,MAAM,IAAI,CAAC,CAAC,CAAC;QACtG,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,4GAA4G;YAC5G,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;iBAAM,IAAI,WAAW,IAAI,KAAK,CAAC,IAAI,KAAK,WAAW,CAAC,UAAU,EAAE,CAAC;gBAC9D,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,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,OAAO,CAAC,CAAC,MAAM,GAAG,CAAC,EAC/D,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,IAAI,UAAU,CAAC,IAAI,EAAE,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,EAAE,KAAK,IAAI,QAAQ,CAAC,MAAM,EAAE,CAAC;wBACrC,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,IAAI,WAAW,GAAG,CAAC,CAAC;QACpB,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,GAChB,KAAK,CAAC,UAAU,KAAK,SAAS,IAAI,IAAI,CAAC,yBAAyB;wBAC5D,CAAC,CAAC,gBAAgB,KAAK,CAAC,UAAU,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG;wBAChD,CAAC,CAAC,EAAE,CAAC;oBACb,QAAQ,IAAI,iBAAiB,WAAW,EAAE,eAAe,cAAc,KAAK,CAAC;oBAC7E,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;CACJ"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"image.d.ts","sourceRoot":"","sources":["../../src/conversion/image.ts"],"names":[],"mappings":"AAQA;;;;;;;;;;GAUG;AACH,wBAAsB,YAAY,CAC9B,SAAS,EAAE,MAAM,EACjB,MAAM,EAAE,MAAM,EACd,MAAM,EAAE,MAAM,EACd,WAAW,GAAE,OAAc,EAC3B,oBAAoB,GAAE,OAAc,EACpC,UAAU,GAAE,KAAK,GAAG,KAAK,GAAG,KAAK,GAAG,WAAmB,GACxD,OAAO,CAAC,MAAM,CAAC,CAgJjB"}
|
|
@@ -1,8 +1,8 @@
|
|
|
1
|
-
import {
|
|
2
|
-
import
|
|
3
|
-
import
|
|
4
|
-
import {
|
|
5
|
-
import {
|
|
1
|
+
import { execFile as execFileCallback } from 'node:child_process';
|
|
2
|
+
import fs from 'node:fs';
|
|
3
|
+
import { promisify } from 'node:util';
|
|
4
|
+
import { log } from '@temporalio/activity';
|
|
5
|
+
import { file } from 'tmp-promise';
|
|
6
6
|
const execFile = promisify(execFileCallback);
|
|
7
7
|
/**
|
|
8
8
|
* Resizes an image to a maximum height or width using ImageMagick
|
|
@@ -17,9 +17,9 @@ const execFile = promisify(execFileCallback);
|
|
|
17
17
|
*/
|
|
18
18
|
export async function imageResizer(inputPath, max_hw, format, progressive = true, colorspaceCorrection = true, colorspace = 'RGB') {
|
|
19
19
|
log.debug(`[image-resizer] Resizing image: ${inputPath} to max_hw: ${max_hw}, format: ${format}, progressive: ${progressive}, colorspaceCorrection: ${colorspaceCorrection ? colorspace : 'disabled'}`);
|
|
20
|
-
const allowedFormats = [
|
|
21
|
-
if (!format || format.trim() ===
|
|
22
|
-
throw new Error(`Invalid format: ${format}.Supported : ${allowedFormats.join(
|
|
20
|
+
const allowedFormats = ['jpg', 'jpeg', 'png', 'webp'];
|
|
21
|
+
if (!format || format.trim() === '') {
|
|
22
|
+
throw new Error(`Invalid format: ${format}.Supported : ${allowedFormats.join(', ')}`);
|
|
23
23
|
}
|
|
24
24
|
//check that max_hw is valid
|
|
25
25
|
if (!Number.isInteger(max_hw) || max_hw <= 0) {
|
|
@@ -41,31 +41,31 @@ export async function imageResizer(inputPath, max_hw, format, progressive = true
|
|
|
41
41
|
throw new Error(`Invalid max_hw value: ${max_hw}`);
|
|
42
42
|
}
|
|
43
43
|
// Progressive loading options
|
|
44
|
-
let conversionOption =
|
|
44
|
+
let conversionOption = '';
|
|
45
45
|
// Only add progressive option for formats that support it
|
|
46
46
|
if (progressive) {
|
|
47
47
|
// JPEG and some other formats support progressive loading
|
|
48
48
|
const lowerFormat = format.toLowerCase();
|
|
49
|
-
if (lowerFormat ===
|
|
50
|
-
conversionOption =
|
|
49
|
+
if (lowerFormat === 'jpg' || lowerFormat === 'jpeg') {
|
|
50
|
+
conversionOption = '-interlace JPEG';
|
|
51
51
|
log.debug(`Enabling interlaced ${lowerFormat.toUpperCase()} format`);
|
|
52
52
|
}
|
|
53
|
-
else if (lowerFormat ===
|
|
54
|
-
conversionOption =
|
|
53
|
+
else if (lowerFormat === 'png') {
|
|
54
|
+
conversionOption = '-interlace PNG';
|
|
55
55
|
log.debug(`Enabling interlaced ${lowerFormat.toUpperCase()} format`);
|
|
56
56
|
}
|
|
57
|
-
else if (lowerFormat ===
|
|
58
|
-
conversionOption =
|
|
57
|
+
else if (lowerFormat === 'gif') {
|
|
58
|
+
conversionOption = '-interlace GIF';
|
|
59
59
|
log.debug(`Enabling interlaced ${lowerFormat.toUpperCase()} format`);
|
|
60
60
|
}
|
|
61
61
|
}
|
|
62
62
|
log.debug(`Resizing image using ImageMagick: ${inputPath} -> ${outputPath}`);
|
|
63
|
-
const command =
|
|
64
|
-
|
|
65
|
-
// Add JPEG shrink-on-load optimization
|
|
66
|
-
args.push(
|
|
63
|
+
const command = 'convert';
|
|
64
|
+
const args = ['-define', `jpeg:size=${max_hw * 3}x${max_hw * 3}`];
|
|
65
|
+
// Add input after JPEG shrink-on-load optimization so ImageMagick can apply it while decoding.
|
|
66
|
+
args.push(inputPath);
|
|
67
67
|
// Remove metadata
|
|
68
|
-
args.push(
|
|
68
|
+
args.push('-strip');
|
|
69
69
|
// https://usage.imagemagick.org/filter/nicolas/#downsample
|
|
70
70
|
// Add colorspace correction if enabled
|
|
71
71
|
if (colorspaceCorrection) {
|
|
@@ -73,32 +73,32 @@ export async function imageResizer(inputPath, max_hw, format, progressive = true
|
|
|
73
73
|
case 'RGB':
|
|
74
74
|
// Linear light, recommended default
|
|
75
75
|
// Convert from sRGB to linear RGB for processing
|
|
76
|
-
args.push(
|
|
77
|
-
log.debug(
|
|
76
|
+
args.push('-colorspace', 'RGB');
|
|
77
|
+
log.debug('Using linear RGB colorspace for resize processing');
|
|
78
78
|
break;
|
|
79
79
|
case 'LAB':
|
|
80
80
|
// Perceptual linear light
|
|
81
81
|
// Use LAB colorspace which separates intensity from color
|
|
82
82
|
// Better for avoiding color clipping and distortion
|
|
83
|
-
args.push(
|
|
84
|
-
log.debug(
|
|
83
|
+
args.push('-colorspace', 'LAB');
|
|
84
|
+
log.debug('Using LAB colorspace for resize processing');
|
|
85
85
|
break;
|
|
86
86
|
case 'LUV':
|
|
87
87
|
// Perceptual linear light
|
|
88
88
|
// Alternative to LAB with perceptually uniform color deltas
|
|
89
|
-
args.push(
|
|
90
|
-
log.debug(
|
|
89
|
+
args.push('-colorspace', 'LUV');
|
|
90
|
+
log.debug('Using LUV colorspace for resize processing');
|
|
91
91
|
break;
|
|
92
92
|
case 'sigmoidal':
|
|
93
93
|
// Sigmoidal colorspace modification to reduce ringing artifacts
|
|
94
|
-
args.push(
|
|
95
|
-
args.push(
|
|
96
|
-
log.debug(
|
|
94
|
+
args.push('-colorspace', 'RGB');
|
|
95
|
+
args.push('+sigmoidal-contrast', '6.5,50%');
|
|
96
|
+
log.debug('Using sigmoidal contrast modification for resize processing');
|
|
97
97
|
break;
|
|
98
98
|
}
|
|
99
99
|
}
|
|
100
100
|
// Resize operation
|
|
101
|
-
args.push(
|
|
101
|
+
args.push('-resize', `${max_hw}x${max_hw}>`);
|
|
102
102
|
// Restore colorspace after processing
|
|
103
103
|
if (colorspaceCorrection) {
|
|
104
104
|
switch (colorspace) {
|
|
@@ -106,22 +106,22 @@ export async function imageResizer(inputPath, max_hw, format, progressive = true
|
|
|
106
106
|
case 'LAB':
|
|
107
107
|
case 'LUV':
|
|
108
108
|
// Convert back to sRGB for output
|
|
109
|
-
args.push(
|
|
109
|
+
args.push('-colorspace', 'sRGB');
|
|
110
110
|
break;
|
|
111
111
|
case 'sigmoidal':
|
|
112
112
|
// Restore from sigmoidal modification and convert to sRGB
|
|
113
|
-
args.push(
|
|
114
|
-
args.push(
|
|
113
|
+
args.push('-sigmoidal-contrast', '6.5,50%');
|
|
114
|
+
args.push('-colorspace', 'sRGB');
|
|
115
115
|
break;
|
|
116
116
|
}
|
|
117
117
|
}
|
|
118
118
|
// Add progressive/interlace options
|
|
119
119
|
if (conversionOption) {
|
|
120
|
-
args.push(...conversionOption.split(
|
|
120
|
+
args.push(...conversionOption.split(' '));
|
|
121
121
|
}
|
|
122
122
|
// Output path
|
|
123
123
|
args.push(outputPath);
|
|
124
|
-
log.debug(`ImageMagick command: ${command} ${args.join(
|
|
124
|
+
log.debug(`ImageMagick command: ${command} ${args.join(' ')}`);
|
|
125
125
|
const { stderr } = await execFile(command, args);
|
|
126
126
|
if (stderr) {
|
|
127
127
|
log.warn(`ImageMagick warning: ${stderr}`);
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"image.js","sourceRoot":"","sources":["../../src/conversion/image.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,IAAI,gBAAgB,EAAE,MAAM,oBAAoB,CAAC;AAClE,OAAO,EAAE,MAAM,SAAS,CAAC;AACzB,OAAO,EAAE,SAAS,EAAE,MAAM,WAAW,CAAC;AACtC,OAAO,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAC;AAC3C,OAAO,EAAE,IAAI,EAAE,MAAM,aAAa,CAAC;AAEnC,MAAM,QAAQ,GAAG,SAAS,CAAC,gBAAgB,CAAC,CAAC;AAE7C;;;;;;;;;;GAUG;AACH,MAAM,CAAC,KAAK,UAAU,YAAY,CAC9B,SAAiB,EACjB,MAAc,EACd,MAAc,EACd,cAAuB,IAAI,EAC3B,uBAAgC,IAAI,EACpC,aAAkD,KAAK;IAEvD,GAAG,CAAC,KAAK,CACL,mCAAmC,SAAS,eAAe,MAAM,aAAa,MAAM,kBAAkB,WAAW,2BAA2B,oBAAoB,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,UAAU,EAAE,CAC/L,CAAC;IAEF,MAAM,cAAc,GAAG,CAAC,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,CAAC,CAAC;IAEtD,IAAI,CAAC,MAAM,IAAI,MAAM,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC;QAClC,MAAM,IAAI,KAAK,CAAC,mBAAmB,MAAM,gBAAgB,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IAC1F,CAAC;IAED,4BAA4B;IAC5B,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,MAAM,CAAC,IAAI,MAAM,IAAI,CAAC,EAAE,CAAC;QAC3C,MAAM,IAAI,KAAK,CAAC,yBAAyB,MAAM,EAAE,CAAC,CAAC;IACvD,CAAC;IAED,6BAA6B;IAC7B,IAAI,CAAC,EAAE,CAAC,UAAU,CAAC,SAAS,CAAC,EAAE,CAAC;QAC5B,MAAM,IAAI,KAAK,CAAC,8BAA8B,SAAS,EAAE,CAAC,CAAC;IAC/D,CAAC;IAED,0BAA0B;IAC1B,MAAM,EAAE,IAAI,EAAE,UAAU,EAAE,OAAO,EAAE,GAAG,MAAM,IAAI,CAAC,EAAE,OAAO,EAAE,IAAI,MAAM,EAAE,EAAE,CAAC,CAAC;IAC5E,IAAI,CAAC;QACD,6BAA6B;QAC7B,IAAI,CAAC,EAAE,CAAC,UAAU,CAAC,SAAS,CAAC,EAAE,CAAC;YAC5B,MAAM,IAAI,KAAK,CAAC,8BAA8B,SAAS,EAAE,CAAC,CAAC;QAC/D,CAAC;QACD,kBAAkB;QAClB,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,MAAM,CAAC,IAAI,MAAM,IAAI,CAAC,EAAE,CAAC;YAC3C,MAAM,IAAI,KAAK,CAAC,yBAAyB,MAAM,EAAE,CAAC,CAAC;QACvD,CAAC;QAED,8BAA8B;QAC9B,IAAI,gBAAgB,GAAG,EAAE,CAAC;QAE1B,0DAA0D;QAC1D,IAAI,WAAW,EAAE,CAAC;YACd,0DAA0D;YAC1D,MAAM,WAAW,GAAG,MAAM,CAAC,WAAW,EAAE,CAAC;YACzC,IAAI,WAAW,KAAK,KAAK,IAAI,WAAW,KAAK,MAAM,EAAE,CAAC;gBAClD,gBAAgB,GAAG,iBAAiB,CAAC;gBACrC,GAAG,CAAC,KAAK,CAAC,uBAAuB,WAAW,CAAC,WAAW,EAAE,SAAS,CAAC,CAAC;YACzE,CAAC;iBAAM,IAAI,WAAW,KAAK,KAAK,EAAE,CAAC;gBAC/B,gBAAgB,GAAG,gBAAgB,CAAC;gBACpC,GAAG,CAAC,KAAK,CAAC,uBAAuB,WAAW,CAAC,WAAW,EAAE,SAAS,CAAC,CAAC;YACzE,CAAC;iBAAM,IAAI,WAAW,KAAK,KAAK,EAAE,CAAC;gBAC/B,gBAAgB,GAAG,gBAAgB,CAAC;gBACpC,GAAG,CAAC,KAAK,CAAC,uBAAuB,WAAW,CAAC,WAAW,EAAE,SAAS,CAAC,CAAC;YACzE,CAAC;QACL,CAAC;QAED,GAAG,CAAC,KAAK,CAAC,qCAAqC,SAAS,OAAO,UAAU,EAAE,CAAC,CAAC;QAE7E,MAAM,OAAO,GAAG,SAAS,CAAC;QAC1B,MAAM,IAAI,GAAG,CAAC,SAAS,EAAE,aAAa,MAAM,GAAG,CAAC,IAAI,MAAM,GAAG,CAAC,EAAE,CAAC,CAAC;QAElE,+FAA+F;QAC/F,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;QAErB,kBAAkB;QAClB,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;QAEpB,2DAA2D;QAC3D,uCAAuC;QACvC,IAAI,oBAAoB,EAAE,CAAC;YACvB,QAAQ,UAAU,EAAE,CAAC;gBACjB,KAAK,KAAK;oBACN,oCAAoC;oBACpC,iDAAiD;oBACjD,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,KAAK,CAAC,CAAC;oBAChC,GAAG,CAAC,KAAK,CAAC,mDAAmD,CAAC,CAAC;oBAC/D,MAAM;gBACV,KAAK,KAAK;oBACN,0BAA0B;oBAC1B,0DAA0D;oBAC1D,oDAAoD;oBACpD,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,KAAK,CAAC,CAAC;oBAChC,GAAG,CAAC,KAAK,CAAC,4CAA4C,CAAC,CAAC;oBACxD,MAAM;gBACV,KAAK,KAAK;oBACN,0BAA0B;oBAC1B,4DAA4D;oBAC5D,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,KAAK,CAAC,CAAC;oBAChC,GAAG,CAAC,KAAK,CAAC,4CAA4C,CAAC,CAAC;oBACxD,MAAM;gBACV,KAAK,WAAW;oBACZ,gEAAgE;oBAChE,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,KAAK,CAAC,CAAC;oBAChC,IAAI,CAAC,IAAI,CAAC,qBAAqB,EAAE,SAAS,CAAC,CAAC;oBAC5C,GAAG,CAAC,KAAK,CAAC,6DAA6D,CAAC,CAAC;oBACzE,MAAM;YACd,CAAC;QACL,CAAC;QAED,mBAAmB;QACnB,IAAI,CAAC,IAAI,CAAC,SAAS,EAAE,GAAG,MAAM,IAAI,MAAM,GAAG,CAAC,CAAC;QAE7C,sCAAsC;QACtC,IAAI,oBAAoB,EAAE,CAAC;YACvB,QAAQ,UAAU,EAAE,CAAC;gBACjB,KAAK,KAAK,CAAC;gBACX,KAAK,KAAK,CAAC;gBACX,KAAK,KAAK;oBACN,kCAAkC;oBAClC,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,MAAM,CAAC,CAAC;oBACjC,MAAM;gBACV,KAAK,WAAW;oBACZ,0DAA0D;oBAC1D,IAAI,CAAC,IAAI,CAAC,qBAAqB,EAAE,SAAS,CAAC,CAAC;oBAC5C,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,MAAM,CAAC,CAAC;oBACjC,MAAM;YACd,CAAC;QACL,CAAC;QAED,oCAAoC;QACpC,IAAI,gBAAgB,EAAE,CAAC;YACnB,IAAI,CAAC,IAAI,CAAC,GAAG,gBAAgB,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC;QAC9C,CAAC;QAED,cAAc;QACd,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;QAEtB,GAAG,CAAC,KAAK,CAAC,wBAAwB,OAAO,IAAI,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QAE/D,MAAM,EAAE,MAAM,EAAE,GAAG,MAAM,QAAQ,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC;QAEjD,IAAI,MAAM,EAAE,CAAC;YACT,GAAG,CAAC,IAAI,CAAC,wBAAwB,MAAM,EAAE,CAAC,CAAC;QAC/C,CAAC;QAED,uCAAuC;QACvC,IAAI,CAAC,EAAE,CAAC,UAAU,CAAC,UAAU,CAAC,IAAI,EAAE,CAAC,QAAQ,CAAC,UAAU,CAAC,CAAC,IAAI,KAAK,CAAC,EAAE,CAAC;YACnE,MAAM,IAAI,KAAK,CAAC,iEAAiE,CAAC,CAAC;QACvF,CAAC;QAED,OAAO,UAAU,CAAC;IACtB,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACb,8BAA8B;QAC9B,MAAM,OAAO,EAAE,CAAC;QAChB,MAAM,YAAY,GAAG,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;QAC5E,GAAG,CAAC,KAAK,CAAC,4BAA4B,YAAY,EAAE,CAAC,CAAC;QACtD,MAAM,IAAI,KAAK,CAAC,4BAA4B,YAAY,EAAE,CAAC,CAAC;IAChE,CAAC;AACL,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"markitdown.d.ts","sourceRoot":"","sources":["../../src/conversion/markitdown.ts"],"names":[],"mappings":"AAKA,wBAAgB,sBAAsB,CAAC,MAAM,EAAE,MAAM,EAAE,GAAG,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC,CAmCpF"}
|
|
@@ -1,26 +1,26 @@
|
|
|
1
|
-
import {
|
|
2
|
-
import
|
|
3
|
-
import
|
|
4
|
-
import tmp from
|
|
1
|
+
import { spawn } from 'node:child_process';
|
|
2
|
+
import fs from 'node:fs';
|
|
3
|
+
import { log } from '@temporalio/activity';
|
|
4
|
+
import tmp from 'tmp';
|
|
5
5
|
export function markdownWithMarkitdown(buffer, ext) {
|
|
6
6
|
const inputFile = tmp.fileSync({ postfix: ext });
|
|
7
|
-
const targetFileName = tmp.tmpNameSync({ postfix:
|
|
7
|
+
const targetFileName = tmp.tmpNameSync({ postfix: '.md' });
|
|
8
8
|
fs.writeSync(inputFile.fd, buffer);
|
|
9
9
|
return new Promise((resolve, reject) => {
|
|
10
|
-
const tool =
|
|
10
|
+
const tool = 'markitdown';
|
|
11
11
|
log.info(`Converting document to markdown with ${tool}`, { inputFile: inputFile.name, targetFileName });
|
|
12
|
-
const command = spawn(tool, [inputFile.name,
|
|
13
|
-
command.on(
|
|
12
|
+
const command = spawn(tool, [inputFile.name, '-o', targetFileName]);
|
|
13
|
+
command.on('exit', (code) => {
|
|
14
14
|
if (code) {
|
|
15
15
|
reject(new Error(`${tool} exited with code ${code}`));
|
|
16
16
|
}
|
|
17
17
|
});
|
|
18
|
-
command.on(
|
|
18
|
+
command.on('close', (code) => {
|
|
19
19
|
if (code) {
|
|
20
20
|
reject(new Error(`${tool} exited with code ${code}`));
|
|
21
21
|
}
|
|
22
22
|
else {
|
|
23
|
-
return fs.readFile(targetFileName,
|
|
23
|
+
return fs.readFile(targetFileName, 'utf8', (err, data) => {
|
|
24
24
|
if (err) {
|
|
25
25
|
reject(err);
|
|
26
26
|
}
|
|
@@ -28,7 +28,7 @@ export function markdownWithMarkitdown(buffer, ext) {
|
|
|
28
28
|
});
|
|
29
29
|
}
|
|
30
30
|
});
|
|
31
|
-
command.on(
|
|
31
|
+
command.on('error', (err) => {
|
|
32
32
|
reject(err);
|
|
33
33
|
});
|
|
34
34
|
});
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"markitdown.js","sourceRoot":"","sources":["../../src/conversion/markitdown.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,KAAK,EAAE,MAAM,oBAAoB,CAAC;AAC3C,OAAO,EAAE,MAAM,SAAS,CAAC;AACzB,OAAO,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAC;AAC3C,OAAO,GAAG,MAAM,KAAK,CAAC;AAEtB,MAAM,UAAU,sBAAsB,CAAC,MAAc,EAAE,GAAY;IAC/D,MAAM,SAAS,GAAG,GAAG,CAAC,QAAQ,CAAC,EAAE,OAAO,EAAE,GAAG,EAAE,CAAC,CAAC;IACjD,MAAM,cAAc,GAAG,GAAG,CAAC,WAAW,CAAC,EAAE,OAAO,EAAE,KAAK,EAAE,CAAC,CAAC;IAE3D,EAAE,CAAC,SAAS,CAAC,SAAS,CAAC,EAAE,EAAE,MAAM,CAAC,CAAC;IAEnC,OAAO,IAAI,OAAO,CAAC,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE;QACnC,MAAM,IAAI,GAAG,YAAY,CAAC;QAC1B,GAAG,CAAC,IAAI,CAAC,wCAAwC,IAAI,EAAE,EAAE,EAAE,SAAS,EAAE,SAAS,CAAC,IAAI,EAAE,cAAc,EAAE,CAAC,CAAC;QAExG,MAAM,OAAO,GAAG,KAAK,CAAC,IAAI,EAAE,CAAC,SAAS,CAAC,IAAI,EAAE,IAAI,EAAE,cAAc,CAAC,CAAC,CAAC;QAEpE,OAAO,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,IAAI,EAAE,EAAE;YACxB,IAAI,IAAI,EAAE,CAAC;gBACP,MAAM,CAAC,IAAI,KAAK,CAAC,GAAG,IAAI,qBAAqB,IAAI,EAAE,CAAC,CAAC,CAAC;YAC1D,CAAC;QACL,CAAC,CAAC,CAAC;QAEH,OAAO,CAAC,EAAE,CAAC,OAAO,EAAE,CAAC,IAAI,EAAE,EAAE;YACzB,IAAI,IAAI,EAAE,CAAC;gBACP,MAAM,CAAC,IAAI,KAAK,CAAC,GAAG,IAAI,qBAAqB,IAAI,EAAE,CAAC,CAAC,CAAC;YAC1D,CAAC;iBAAM,CAAC;gBACJ,OAAO,EAAE,CAAC,QAAQ,CAAC,cAAc,EAAE,MAAM,EAAE,CAAC,GAAG,EAAE,IAAI,EAAE,EAAE;oBACrD,IAAI,GAAG,EAAE,CAAC;wBACN,MAAM,CAAC,GAAG,CAAC,CAAC;oBAChB,CAAC;oBACD,OAAO,OAAO,CAAC,IAAI,CAAC,CAAC;gBACzB,CAAC,CAAC,CAAC;YACP,CAAC;QACL,CAAC,CAAC,CAAC;QAEH,OAAO,CAAC,EAAE,CAAC,OAAO,EAAE,CAAC,GAAG,EAAE,EAAE;YACxB,MAAM,CAAC,GAAG,CAAC,CAAC;QAChB,CAAC,CAAC,CAAC;IACP,CAAC,CAAC,CAAC;AACP,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"mutool.d.ts","sourceRoot":"","sources":["../../src/conversion/mutool.ts"],"names":[],"mappings":"AAKA;;;GAGG;AAEH,wBAAgB,eAAe,CAAC,MAAM,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC,CAkC/D;AAED;;;;;;;GAOG;AACH,wBAAsB,WAAW,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,EAAE,KAAK,CAAC,EAAE,MAAM,EAAE,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,CA+C5F;AAED;;GAEG;AAEH,wBAAsB,eAAe,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,GAAG,OAAO,CAAC,MAAM,CAAC,CAkD7F"}
|