@vertesia/workflow 0.24.0-dev.202601221707

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.
Files changed (490) hide show
  1. package/LICENSE +13 -0
  2. package/README.md +65 -0
  3. package/bin/bundle-workflows.mjs +39 -0
  4. package/lib/cjs/activities/advanced/createDocumentTypeFromInteractionRun.js +33 -0
  5. package/lib/cjs/activities/advanced/createDocumentTypeFromInteractionRun.js.map +1 -0
  6. package/lib/cjs/activities/advanced/createOrUpdateDocumentFromInteractionRun.js +73 -0
  7. package/lib/cjs/activities/advanced/createOrUpdateDocumentFromInteractionRun.js.map +1 -0
  8. package/lib/cjs/activities/advanced/updateDocumentFromInteractionRun.js +19 -0
  9. package/lib/cjs/activities/advanced/updateDocumentFromInteractionRun.js.map +1 -0
  10. package/lib/cjs/activities/chunkDocument.js +85 -0
  11. package/lib/cjs/activities/chunkDocument.js.map +1 -0
  12. package/lib/cjs/activities/copyParentArtifacts.js +127 -0
  13. package/lib/cjs/activities/copyParentArtifacts.js.map +1 -0
  14. package/lib/cjs/activities/createDocumentFromOther.js +64 -0
  15. package/lib/cjs/activities/createDocumentFromOther.js.map +1 -0
  16. package/lib/cjs/activities/executeInteraction.js +194 -0
  17. package/lib/cjs/activities/executeInteraction.js.map +1 -0
  18. package/lib/cjs/activities/extractDocumentText.js +156 -0
  19. package/lib/cjs/activities/extractDocumentText.js.map +1 -0
  20. package/lib/cjs/activities/generateDocumentProperties.js +83 -0
  21. package/lib/cjs/activities/generateDocumentProperties.js.map +1 -0
  22. package/lib/cjs/activities/generateEmbeddings.js +257 -0
  23. package/lib/cjs/activities/generateEmbeddings.js.map +1 -0
  24. package/lib/cjs/activities/generateOrAssignContentType.js +125 -0
  25. package/lib/cjs/activities/generateOrAssignContentType.js.map +1 -0
  26. package/lib/cjs/activities/getObjectFromStore.js +20 -0
  27. package/lib/cjs/activities/getObjectFromStore.js.map +1 -0
  28. package/lib/cjs/activities/handleError.js +22 -0
  29. package/lib/cjs/activities/handleError.js.map +1 -0
  30. package/lib/cjs/activities/index-dsl.js +51 -0
  31. package/lib/cjs/activities/index-dsl.js.map +1 -0
  32. package/lib/cjs/activities/index.js +21 -0
  33. package/lib/cjs/activities/index.js.map +1 -0
  34. package/lib/cjs/activities/media/prepareAudio.js +239 -0
  35. package/lib/cjs/activities/media/prepareAudio.js.map +1 -0
  36. package/lib/cjs/activities/media/prepareVideo.js +429 -0
  37. package/lib/cjs/activities/media/prepareVideo.js.map +1 -0
  38. package/lib/cjs/activities/media/processPdfWithTextract.js +103 -0
  39. package/lib/cjs/activities/media/processPdfWithTextract.js.map +1 -0
  40. package/lib/cjs/activities/media/saveGladiaTranscription.js +81 -0
  41. package/lib/cjs/activities/media/saveGladiaTranscription.js.map +1 -0
  42. package/lib/cjs/activities/media/transcribeMediaWithGladia.js +82 -0
  43. package/lib/cjs/activities/media/transcribeMediaWithGladia.js.map +1 -0
  44. package/lib/cjs/activities/notifyWebhook.js +158 -0
  45. package/lib/cjs/activities/notifyWebhook.js.map +1 -0
  46. package/lib/cjs/activities/rateLimiter.js +30 -0
  47. package/lib/cjs/activities/rateLimiter.js.map +1 -0
  48. package/lib/cjs/activities/renditions/generateImageRendition.js +66 -0
  49. package/lib/cjs/activities/renditions/generateImageRendition.js.map +1 -0
  50. package/lib/cjs/activities/renditions/generateVideoRendition.js +200 -0
  51. package/lib/cjs/activities/renditions/generateVideoRendition.js.map +1 -0
  52. package/lib/cjs/activities/setDocumentStatus.js +15 -0
  53. package/lib/cjs/activities/setDocumentStatus.js.map +1 -0
  54. package/lib/cjs/conversion/TextractProcessor.js +417 -0
  55. package/lib/cjs/conversion/TextractProcessor.js.map +1 -0
  56. package/lib/cjs/conversion/image.js +149 -0
  57. package/lib/cjs/conversion/image.js.map +1 -0
  58. package/lib/cjs/conversion/markitdown.js +42 -0
  59. package/lib/cjs/conversion/markitdown.js.map +1 -0
  60. package/lib/cjs/conversion/mutool.js +147 -0
  61. package/lib/cjs/conversion/mutool.js.map +1 -0
  62. package/lib/cjs/conversion/pandoc.js +39 -0
  63. package/lib/cjs/conversion/pandoc.js.map +1 -0
  64. package/lib/cjs/dsl/conditions.js +81 -0
  65. package/lib/cjs/dsl/conditions.js.map +1 -0
  66. package/lib/cjs/dsl/dsl-workflow.js +343 -0
  67. package/lib/cjs/dsl/dsl-workflow.js.map +1 -0
  68. package/lib/cjs/dsl/dslProxyActivities.js +23 -0
  69. package/lib/cjs/dsl/dslProxyActivities.js.map +1 -0
  70. package/lib/cjs/dsl/projections.js +59 -0
  71. package/lib/cjs/dsl/projections.js.map +1 -0
  72. package/lib/cjs/dsl/setup/ActivityContext.js +122 -0
  73. package/lib/cjs/dsl/setup/ActivityContext.js.map +1 -0
  74. package/lib/cjs/dsl/setup/fetch/DataProvider.js +51 -0
  75. package/lib/cjs/dsl/setup/fetch/DataProvider.js.map +1 -0
  76. package/lib/cjs/dsl/setup/fetch/index.js +16 -0
  77. package/lib/cjs/dsl/setup/fetch/index.js.map +1 -0
  78. package/lib/cjs/dsl/setup/fetch/providers.js +67 -0
  79. package/lib/cjs/dsl/setup/fetch/providers.js.map +1 -0
  80. package/lib/cjs/dsl/test/test-child-workflow.js +10 -0
  81. package/lib/cjs/dsl/test/test-child-workflow.js.map +1 -0
  82. package/lib/cjs/dsl/validation.js +122 -0
  83. package/lib/cjs/dsl/validation.js.map +1 -0
  84. package/lib/cjs/dsl/vars.js +341 -0
  85. package/lib/cjs/dsl/vars.js.map +1 -0
  86. package/lib/cjs/dsl/walk.js +100 -0
  87. package/lib/cjs/dsl/walk.js.map +1 -0
  88. package/lib/cjs/dsl.js +20 -0
  89. package/lib/cjs/dsl.js.map +1 -0
  90. package/lib/cjs/errors.js +79 -0
  91. package/lib/cjs/errors.js.map +1 -0
  92. package/lib/cjs/index.js +56 -0
  93. package/lib/cjs/index.js.map +1 -0
  94. package/lib/cjs/iterative-generation/activities/extractToc.js +47 -0
  95. package/lib/cjs/iterative-generation/activities/extractToc.js.map +1 -0
  96. package/lib/cjs/iterative-generation/activities/finalizeOutput.js +72 -0
  97. package/lib/cjs/iterative-generation/activities/finalizeOutput.js.map +1 -0
  98. package/lib/cjs/iterative-generation/activities/generatePart.js +78 -0
  99. package/lib/cjs/iterative-generation/activities/generatePart.js.map +1 -0
  100. package/lib/cjs/iterative-generation/activities/generateToc.js +86 -0
  101. package/lib/cjs/iterative-generation/activities/generateToc.js.map +1 -0
  102. package/lib/cjs/iterative-generation/activities/index.js +12 -0
  103. package/lib/cjs/iterative-generation/activities/index.js.map +1 -0
  104. package/lib/cjs/iterative-generation/iterativeGenerationWorkflow.js +56 -0
  105. package/lib/cjs/iterative-generation/iterativeGenerationWorkflow.js.map +1 -0
  106. package/lib/cjs/iterative-generation/types.js +5 -0
  107. package/lib/cjs/iterative-generation/types.js.map +1 -0
  108. package/lib/cjs/iterative-generation/utils.js +121 -0
  109. package/lib/cjs/iterative-generation/utils.js.map +1 -0
  110. package/lib/cjs/package.json +3 -0
  111. package/lib/cjs/result-types.js +10 -0
  112. package/lib/cjs/result-types.js.map +1 -0
  113. package/lib/cjs/system/notifyWebhookWorkflow.js +53 -0
  114. package/lib/cjs/system/notifyWebhookWorkflow.js.map +1 -0
  115. package/lib/cjs/system/recalculateEmbeddingsWorkflow.js +33 -0
  116. package/lib/cjs/system/recalculateEmbeddingsWorkflow.js.map +1 -0
  117. package/lib/cjs/utils/auth.js +15 -0
  118. package/lib/cjs/utils/auth.js.map +1 -0
  119. package/lib/cjs/utils/blobs.js +64 -0
  120. package/lib/cjs/utils/blobs.js.map +1 -0
  121. package/lib/cjs/utils/chunks.js +14 -0
  122. package/lib/cjs/utils/chunks.js.map +1 -0
  123. package/lib/cjs/utils/client.js +31 -0
  124. package/lib/cjs/utils/client.js.map +1 -0
  125. package/lib/cjs/utils/expand-vars.js +33 -0
  126. package/lib/cjs/utils/expand-vars.js.map +1 -0
  127. package/lib/cjs/utils/memory.js +65 -0
  128. package/lib/cjs/utils/memory.js.map +1 -0
  129. package/lib/cjs/utils/renditions.js +88 -0
  130. package/lib/cjs/utils/renditions.js.map +1 -0
  131. package/lib/cjs/utils/storage.js +54 -0
  132. package/lib/cjs/utils/storage.js.map +1 -0
  133. package/lib/cjs/utils/tokens.js +38 -0
  134. package/lib/cjs/utils/tokens.js.map +1 -0
  135. package/lib/cjs/vars.js +20 -0
  136. package/lib/cjs/vars.js.map +1 -0
  137. package/lib/cjs/workflows.js +15 -0
  138. package/lib/cjs/workflows.js.map +1 -0
  139. package/lib/esm/activities/advanced/createDocumentTypeFromInteractionRun.js +30 -0
  140. package/lib/esm/activities/advanced/createDocumentTypeFromInteractionRun.js.map +1 -0
  141. package/lib/esm/activities/advanced/createOrUpdateDocumentFromInteractionRun.js +70 -0
  142. package/lib/esm/activities/advanced/createOrUpdateDocumentFromInteractionRun.js.map +1 -0
  143. package/lib/esm/activities/advanced/updateDocumentFromInteractionRun.js +16 -0
  144. package/lib/esm/activities/advanced/updateDocumentFromInteractionRun.js.map +1 -0
  145. package/lib/esm/activities/chunkDocument.js +82 -0
  146. package/lib/esm/activities/chunkDocument.js.map +1 -0
  147. package/lib/esm/activities/copyParentArtifacts.js +124 -0
  148. package/lib/esm/activities/copyParentArtifacts.js.map +1 -0
  149. package/lib/esm/activities/createDocumentFromOther.js +58 -0
  150. package/lib/esm/activities/createDocumentFromOther.js.map +1 -0
  151. package/lib/esm/activities/executeInteraction.js +190 -0
  152. package/lib/esm/activities/executeInteraction.js.map +1 -0
  153. package/lib/esm/activities/extractDocumentText.js +153 -0
  154. package/lib/esm/activities/extractDocumentText.js.map +1 -0
  155. package/lib/esm/activities/generateDocumentProperties.js +80 -0
  156. package/lib/esm/activities/generateDocumentProperties.js.map +1 -0
  157. package/lib/esm/activities/generateEmbeddings.js +254 -0
  158. package/lib/esm/activities/generateEmbeddings.js.map +1 -0
  159. package/lib/esm/activities/generateOrAssignContentType.js +122 -0
  160. package/lib/esm/activities/generateOrAssignContentType.js.map +1 -0
  161. package/lib/esm/activities/getObjectFromStore.js +17 -0
  162. package/lib/esm/activities/getObjectFromStore.js.map +1 -0
  163. package/lib/esm/activities/handleError.js +19 -0
  164. package/lib/esm/activities/handleError.js.map +1 -0
  165. package/lib/esm/activities/index-dsl.js +25 -0
  166. package/lib/esm/activities/index-dsl.js.map +1 -0
  167. package/lib/esm/activities/index.js +5 -0
  168. package/lib/esm/activities/index.js.map +1 -0
  169. package/lib/esm/activities/media/prepareAudio.js +200 -0
  170. package/lib/esm/activities/media/prepareAudio.js.map +1 -0
  171. package/lib/esm/activities/media/prepareVideo.js +390 -0
  172. package/lib/esm/activities/media/prepareVideo.js.map +1 -0
  173. package/lib/esm/activities/media/processPdfWithTextract.js +99 -0
  174. package/lib/esm/activities/media/processPdfWithTextract.js.map +1 -0
  175. package/lib/esm/activities/media/saveGladiaTranscription.js +78 -0
  176. package/lib/esm/activities/media/saveGladiaTranscription.js.map +1 -0
  177. package/lib/esm/activities/media/transcribeMediaWithGladia.js +79 -0
  178. package/lib/esm/activities/media/transcribeMediaWithGladia.js.map +1 -0
  179. package/lib/esm/activities/notifyWebhook.js +155 -0
  180. package/lib/esm/activities/notifyWebhook.js.map +1 -0
  181. package/lib/esm/activities/rateLimiter.js +27 -0
  182. package/lib/esm/activities/rateLimiter.js.map +1 -0
  183. package/lib/esm/activities/renditions/generateImageRendition.js +63 -0
  184. package/lib/esm/activities/renditions/generateImageRendition.js.map +1 -0
  185. package/lib/esm/activities/renditions/generateVideoRendition.js +194 -0
  186. package/lib/esm/activities/renditions/generateVideoRendition.js.map +1 -0
  187. package/lib/esm/activities/setDocumentStatus.js +12 -0
  188. package/lib/esm/activities/setDocumentStatus.js.map +1 -0
  189. package/lib/esm/conversion/TextractProcessor.js +410 -0
  190. package/lib/esm/conversion/TextractProcessor.js.map +1 -0
  191. package/lib/esm/conversion/image.js +143 -0
  192. package/lib/esm/conversion/image.js.map +1 -0
  193. package/lib/esm/conversion/markitdown.js +36 -0
  194. package/lib/esm/conversion/markitdown.js.map +1 -0
  195. package/lib/esm/conversion/mutool.js +139 -0
  196. package/lib/esm/conversion/mutool.js.map +1 -0
  197. package/lib/esm/conversion/pandoc.js +36 -0
  198. package/lib/esm/conversion/pandoc.js.map +1 -0
  199. package/lib/esm/dsl/conditions.js +75 -0
  200. package/lib/esm/dsl/conditions.js.map +1 -0
  201. package/lib/esm/dsl/dsl-workflow.js +336 -0
  202. package/lib/esm/dsl/dsl-workflow.js.map +1 -0
  203. package/lib/esm/dsl/dslProxyActivities.js +20 -0
  204. package/lib/esm/dsl/dslProxyActivities.js.map +1 -0
  205. package/lib/esm/dsl/projections.js +55 -0
  206. package/lib/esm/dsl/projections.js.map +1 -0
  207. package/lib/esm/dsl/setup/ActivityContext.js +117 -0
  208. package/lib/esm/dsl/setup/ActivityContext.js.map +1 -0
  209. package/lib/esm/dsl/setup/fetch/DataProvider.js +47 -0
  210. package/lib/esm/dsl/setup/fetch/DataProvider.js.map +1 -0
  211. package/lib/esm/dsl/setup/fetch/index.js +12 -0
  212. package/lib/esm/dsl/setup/fetch/index.js.map +1 -0
  213. package/lib/esm/dsl/setup/fetch/providers.js +61 -0
  214. package/lib/esm/dsl/setup/fetch/providers.js.map +1 -0
  215. package/lib/esm/dsl/test/test-child-workflow.js +5 -0
  216. package/lib/esm/dsl/test/test-child-workflow.js.map +1 -0
  217. package/lib/esm/dsl/validation.js +118 -0
  218. package/lib/esm/dsl/validation.js.map +1 -0
  219. package/lib/esm/dsl/vars.js +335 -0
  220. package/lib/esm/dsl/vars.js.map +1 -0
  221. package/lib/esm/dsl/walk.js +96 -0
  222. package/lib/esm/dsl/walk.js.map +1 -0
  223. package/lib/esm/dsl.js +4 -0
  224. package/lib/esm/dsl.js.map +1 -0
  225. package/lib/esm/errors.js +69 -0
  226. package/lib/esm/errors.js.map +1 -0
  227. package/lib/esm/index.js +38 -0
  228. package/lib/esm/index.js.map +1 -0
  229. package/lib/esm/iterative-generation/activities/extractToc.js +44 -0
  230. package/lib/esm/iterative-generation/activities/extractToc.js.map +1 -0
  231. package/lib/esm/iterative-generation/activities/finalizeOutput.js +69 -0
  232. package/lib/esm/iterative-generation/activities/finalizeOutput.js.map +1 -0
  233. package/lib/esm/iterative-generation/activities/generatePart.js +75 -0
  234. package/lib/esm/iterative-generation/activities/generatePart.js.map +1 -0
  235. package/lib/esm/iterative-generation/activities/generateToc.js +83 -0
  236. package/lib/esm/iterative-generation/activities/generateToc.js.map +1 -0
  237. package/lib/esm/iterative-generation/activities/index.js +5 -0
  238. package/lib/esm/iterative-generation/activities/index.js.map +1 -0
  239. package/lib/esm/iterative-generation/iterativeGenerationWorkflow.js +53 -0
  240. package/lib/esm/iterative-generation/iterativeGenerationWorkflow.js.map +1 -0
  241. package/lib/esm/iterative-generation/types.js +2 -0
  242. package/lib/esm/iterative-generation/types.js.map +1 -0
  243. package/lib/esm/iterative-generation/utils.js +112 -0
  244. package/lib/esm/iterative-generation/utils.js.map +1 -0
  245. package/lib/esm/result-types.js +7 -0
  246. package/lib/esm/result-types.js.map +1 -0
  247. package/lib/esm/system/notifyWebhookWorkflow.js +50 -0
  248. package/lib/esm/system/notifyWebhookWorkflow.js.map +1 -0
  249. package/lib/esm/system/recalculateEmbeddingsWorkflow.js +30 -0
  250. package/lib/esm/system/recalculateEmbeddingsWorkflow.js.map +1 -0
  251. package/lib/esm/utils/auth.js +8 -0
  252. package/lib/esm/utils/auth.js.map +1 -0
  253. package/lib/esm/utils/blobs.js +54 -0
  254. package/lib/esm/utils/blobs.js.map +1 -0
  255. package/lib/esm/utils/chunks.js +9 -0
  256. package/lib/esm/utils/chunks.js.map +1 -0
  257. package/lib/esm/utils/client.js +27 -0
  258. package/lib/esm/utils/client.js.map +1 -0
  259. package/lib/esm/utils/expand-vars.js +30 -0
  260. package/lib/esm/utils/expand-vars.js.map +1 -0
  261. package/lib/esm/utils/memory.js +55 -0
  262. package/lib/esm/utils/memory.js.map +1 -0
  263. package/lib/esm/utils/renditions.js +80 -0
  264. package/lib/esm/utils/renditions.js.map +1 -0
  265. package/lib/esm/utils/storage.js +45 -0
  266. package/lib/esm/utils/storage.js.map +1 -0
  267. package/lib/esm/utils/tokens.js +34 -0
  268. package/lib/esm/utils/tokens.js.map +1 -0
  269. package/lib/esm/vars.js +4 -0
  270. package/lib/esm/vars.js.map +1 -0
  271. package/lib/esm/workflows.js +8 -0
  272. package/lib/esm/workflows.js.map +1 -0
  273. package/lib/tsconfig.tsbuildinfo +1 -0
  274. package/lib/types/activities/advanced/createDocumentTypeFromInteractionRun.d.ts +17 -0
  275. package/lib/types/activities/advanced/createDocumentTypeFromInteractionRun.d.ts.map +1 -0
  276. package/lib/types/activities/advanced/createOrUpdateDocumentFromInteractionRun.d.ts +39 -0
  277. package/lib/types/activities/advanced/createOrUpdateDocumentFromInteractionRun.d.ts.map +1 -0
  278. package/lib/types/activities/advanced/updateDocumentFromInteractionRun.d.ts +19 -0
  279. package/lib/types/activities/advanced/updateDocumentFromInteractionRun.d.ts.map +1 -0
  280. package/lib/types/activities/chunkDocument.d.ts +33 -0
  281. package/lib/types/activities/chunkDocument.d.ts.map +1 -0
  282. package/lib/types/activities/copyParentArtifacts.d.ts +19 -0
  283. package/lib/types/activities/copyParentArtifacts.d.ts.map +1 -0
  284. package/lib/types/activities/createDocumentFromOther.d.ts +21 -0
  285. package/lib/types/activities/createDocumentFromOther.d.ts.map +1 -0
  286. package/lib/types/activities/executeInteraction.d.ts +61 -0
  287. package/lib/types/activities/executeInteraction.d.ts.map +1 -0
  288. package/lib/types/activities/extractDocumentText.d.ts +10 -0
  289. package/lib/types/activities/extractDocumentText.d.ts.map +1 -0
  290. package/lib/types/activities/generateDocumentProperties.d.ts +32 -0
  291. package/lib/types/activities/generateDocumentProperties.d.ts.map +1 -0
  292. package/lib/types/activities/generateEmbeddings.d.ts +53 -0
  293. package/lib/types/activities/generateEmbeddings.d.ts.map +1 -0
  294. package/lib/types/activities/generateOrAssignContentType.d.ts +44 -0
  295. package/lib/types/activities/generateOrAssignContentType.d.ts.map +1 -0
  296. package/lib/types/activities/getObjectFromStore.d.ts +14 -0
  297. package/lib/types/activities/getObjectFromStore.d.ts.map +1 -0
  298. package/lib/types/activities/handleError.d.ts +6 -0
  299. package/lib/types/activities/handleError.d.ts.map +1 -0
  300. package/lib/types/activities/index-dsl.d.ts +25 -0
  301. package/lib/types/activities/index-dsl.d.ts.map +1 -0
  302. package/lib/types/activities/index.d.ts +5 -0
  303. package/lib/types/activities/index.d.ts.map +1 -0
  304. package/lib/types/activities/media/prepareAudio.d.ts +25 -0
  305. package/lib/types/activities/media/prepareAudio.d.ts.map +1 -0
  306. package/lib/types/activities/media/prepareVideo.d.ts +30 -0
  307. package/lib/types/activities/media/prepareVideo.d.ts.map +1 -0
  308. package/lib/types/activities/media/processPdfWithTextract.d.ts +26 -0
  309. package/lib/types/activities/media/processPdfWithTextract.d.ts.map +1 -0
  310. package/lib/types/activities/media/saveGladiaTranscription.d.ts +14 -0
  311. package/lib/types/activities/media/saveGladiaTranscription.d.ts.map +1 -0
  312. package/lib/types/activities/media/transcribeMediaWithGladia.d.ts +19 -0
  313. package/lib/types/activities/media/transcribeMediaWithGladia.d.ts.map +1 -0
  314. package/lib/types/activities/notifyWebhook.d.ts +27 -0
  315. package/lib/types/activities/notifyWebhook.d.ts.map +1 -0
  316. package/lib/types/activities/rateLimiter.d.ts +11 -0
  317. package/lib/types/activities/rateLimiter.d.ts.map +1 -0
  318. package/lib/types/activities/renditions/generateImageRendition.d.ts +14 -0
  319. package/lib/types/activities/renditions/generateImageRendition.d.ts.map +1 -0
  320. package/lib/types/activities/renditions/generateVideoRendition.d.ts +15 -0
  321. package/lib/types/activities/renditions/generateVideoRendition.d.ts.map +1 -0
  322. package/lib/types/activities/setDocumentStatus.d.ts +15 -0
  323. package/lib/types/activities/setDocumentStatus.d.ts.map +1 -0
  324. package/lib/types/conversion/TextractProcessor.d.ts +45 -0
  325. package/lib/types/conversion/TextractProcessor.d.ts.map +1 -0
  326. package/lib/types/conversion/image.d.ts +13 -0
  327. package/lib/types/conversion/image.d.ts.map +1 -0
  328. package/lib/types/conversion/markitdown.d.ts +2 -0
  329. package/lib/types/conversion/markitdown.d.ts.map +1 -0
  330. package/lib/types/conversion/mutool.d.ts +19 -0
  331. package/lib/types/conversion/mutool.d.ts.map +1 -0
  332. package/lib/types/conversion/pandoc.d.ts +2 -0
  333. package/lib/types/conversion/pandoc.d.ts.map +1 -0
  334. package/lib/types/dsl/conditions.d.ts +2 -0
  335. package/lib/types/dsl/conditions.d.ts.map +1 -0
  336. package/lib/types/dsl/dsl-workflow.d.ts +5 -0
  337. package/lib/types/dsl/dsl-workflow.d.ts.map +1 -0
  338. package/lib/types/dsl/dslProxyActivities.d.ts +10 -0
  339. package/lib/types/dsl/dslProxyActivities.d.ts.map +1 -0
  340. package/lib/types/dsl/projections.d.ts +4 -0
  341. package/lib/types/dsl/projections.d.ts.map +1 -0
  342. package/lib/types/dsl/setup/ActivityContext.d.ts +17 -0
  343. package/lib/types/dsl/setup/ActivityContext.d.ts.map +1 -0
  344. package/lib/types/dsl/setup/fetch/DataProvider.d.ts +9 -0
  345. package/lib/types/dsl/setup/fetch/DataProvider.d.ts.map +1 -0
  346. package/lib/types/dsl/setup/fetch/index.d.ts +6 -0
  347. package/lib/types/dsl/setup/fetch/index.d.ts.map +1 -0
  348. package/lib/types/dsl/setup/fetch/providers.d.ts +25 -0
  349. package/lib/types/dsl/setup/fetch/providers.d.ts.map +1 -0
  350. package/lib/types/dsl/test/test-child-workflow.d.ts +4 -0
  351. package/lib/types/dsl/test/test-child-workflow.d.ts.map +1 -0
  352. package/lib/types/dsl/validation.d.ts +4 -0
  353. package/lib/types/dsl/validation.d.ts.map +1 -0
  354. package/lib/types/dsl/vars.d.ts +48 -0
  355. package/lib/types/dsl/vars.d.ts.map +1 -0
  356. package/lib/types/dsl/walk.d.ts +18 -0
  357. package/lib/types/dsl/walk.d.ts.map +1 -0
  358. package/lib/types/dsl.d.ts +4 -0
  359. package/lib/types/dsl.d.ts.map +1 -0
  360. package/lib/types/errors.d.ts +37 -0
  361. package/lib/types/errors.d.ts.map +1 -0
  362. package/lib/types/index.d.ts +37 -0
  363. package/lib/types/index.d.ts.map +1 -0
  364. package/lib/types/iterative-generation/activities/extractToc.d.ts +10 -0
  365. package/lib/types/iterative-generation/activities/extractToc.d.ts.map +1 -0
  366. package/lib/types/iterative-generation/activities/finalizeOutput.d.ts +3 -0
  367. package/lib/types/iterative-generation/activities/finalizeOutput.d.ts.map +1 -0
  368. package/lib/types/iterative-generation/activities/generatePart.d.ts +3 -0
  369. package/lib/types/iterative-generation/activities/generatePart.d.ts.map +1 -0
  370. package/lib/types/iterative-generation/activities/generateToc.d.ts +4 -0
  371. package/lib/types/iterative-generation/activities/generateToc.d.ts.map +1 -0
  372. package/lib/types/iterative-generation/activities/index.d.ts +5 -0
  373. package/lib/types/iterative-generation/activities/index.d.ts.map +1 -0
  374. package/lib/types/iterative-generation/iterativeGenerationWorkflow.d.ts +3 -0
  375. package/lib/types/iterative-generation/iterativeGenerationWorkflow.d.ts.map +1 -0
  376. package/lib/types/iterative-generation/types.d.ts +79 -0
  377. package/lib/types/iterative-generation/types.d.ts.map +1 -0
  378. package/lib/types/iterative-generation/utils.d.ts +26 -0
  379. package/lib/types/iterative-generation/utils.d.ts.map +1 -0
  380. package/lib/types/result-types.d.ts +22 -0
  381. package/lib/types/result-types.d.ts.map +1 -0
  382. package/lib/types/system/notifyWebhookWorkflow.d.ts +8 -0
  383. package/lib/types/system/notifyWebhookWorkflow.d.ts.map +1 -0
  384. package/lib/types/system/recalculateEmbeddingsWorkflow.d.ts +25 -0
  385. package/lib/types/system/recalculateEmbeddingsWorkflow.d.ts.map +1 -0
  386. package/lib/types/utils/auth.d.ts +4 -0
  387. package/lib/types/utils/auth.d.ts.map +1 -0
  388. package/lib/types/utils/blobs.d.ts +7 -0
  389. package/lib/types/utils/blobs.d.ts.map +1 -0
  390. package/lib/types/utils/chunks.d.ts +9 -0
  391. package/lib/types/utils/chunks.d.ts.map +1 -0
  392. package/lib/types/utils/client.d.ts +8 -0
  393. package/lib/types/utils/client.d.ts.map +1 -0
  394. package/lib/types/utils/expand-vars.d.ts +8 -0
  395. package/lib/types/utils/expand-vars.d.ts.map +1 -0
  396. package/lib/types/utils/memory.d.ts +8 -0
  397. package/lib/types/utils/memory.d.ts.map +1 -0
  398. package/lib/types/utils/renditions.d.ts +23 -0
  399. package/lib/types/utils/renditions.d.ts.map +1 -0
  400. package/lib/types/utils/storage.d.ts +16 -0
  401. package/lib/types/utils/storage.d.ts.map +1 -0
  402. package/lib/types/utils/tokens.d.ts +11 -0
  403. package/lib/types/utils/tokens.d.ts.map +1 -0
  404. package/lib/types/vars.d.ts +3 -0
  405. package/lib/types/vars.d.ts.map +1 -0
  406. package/lib/types/workflows.d.ts +8 -0
  407. package/lib/types/workflows.d.ts.map +1 -0
  408. package/lib/workflows-bundle.js +17213 -0
  409. package/package.json +146 -0
  410. package/src/activities/advanced/createDocumentTypeFromInteractionRun.ts +55 -0
  411. package/src/activities/advanced/createOrUpdateDocumentFromInteractionRun.ts +119 -0
  412. package/src/activities/advanced/updateDocumentFromInteractionRun.ts +35 -0
  413. package/src/activities/chunkDocument.ts +146 -0
  414. package/src/activities/copyParentArtifacts.ts +162 -0
  415. package/src/activities/createDocumentFromOther.ts +92 -0
  416. package/src/activities/executeInteraction.ts +300 -0
  417. package/src/activities/extractDocumentText.ts +205 -0
  418. package/src/activities/generateDocumentProperties.ts +120 -0
  419. package/src/activities/generateEmbeddings.ts +387 -0
  420. package/src/activities/generateOrAssignContentType.ts +218 -0
  421. package/src/activities/getObjectFromStore.ts +31 -0
  422. package/src/activities/handleError.ts +25 -0
  423. package/src/activities/index-dsl.ts +25 -0
  424. package/src/activities/index.ts +4 -0
  425. package/src/activities/media/prepareAudio.ts +334 -0
  426. package/src/activities/media/prepareVideo.ts +622 -0
  427. package/src/activities/media/processPdfWithTextract.ts +141 -0
  428. package/src/activities/media/saveGladiaTranscription.ts +128 -0
  429. package/src/activities/media/transcribeMediaWithGladia.ts +117 -0
  430. package/src/activities/notifyWebhook.test.ts +134 -0
  431. package/src/activities/notifyWebhook.ts +199 -0
  432. package/src/activities/rateLimiter.ts +41 -0
  433. package/src/activities/renditions/generateImageRendition.ts +111 -0
  434. package/src/activities/renditions/generateVideoRendition.ts +293 -0
  435. package/src/activities/setDocumentStatus.ts +25 -0
  436. package/src/conversion/TextractProcessor.ts +506 -0
  437. package/src/conversion/image.test.ts +118 -0
  438. package/src/conversion/image.ts +168 -0
  439. package/src/conversion/markitdown.ts +41 -0
  440. package/src/conversion/mutool.test.ts +74 -0
  441. package/src/conversion/mutool.ts +180 -0
  442. package/src/conversion/pandoc.test.ts +24 -0
  443. package/src/conversion/pandoc.ts +40 -0
  444. package/src/dsl/conditions.ts +76 -0
  445. package/src/dsl/dsl-workflow.test.ts +58 -0
  446. package/src/dsl/dsl-workflow.ts +397 -0
  447. package/src/dsl/dslProxyActivities.ts +38 -0
  448. package/src/dsl/ms.d.ts +11 -0
  449. package/src/dsl/projections.test.ts +159 -0
  450. package/src/dsl/projections.ts +72 -0
  451. package/src/dsl/setup/ActivityContext.ts +178 -0
  452. package/src/dsl/setup/fetch/DataProvider.ts +45 -0
  453. package/src/dsl/setup/fetch/index.ts +19 -0
  454. package/src/dsl/setup/fetch/providers.ts +67 -0
  455. package/src/dsl/test/test-child-workflow.ts +6 -0
  456. package/src/dsl/validation.test.ts +257 -0
  457. package/src/dsl/validation.ts +125 -0
  458. package/src/dsl/vars.test.ts +245 -0
  459. package/src/dsl/vars.ts +340 -0
  460. package/src/dsl/walk.test.ts +81 -0
  461. package/src/dsl/walk.ts +103 -0
  462. package/src/dsl/workflow-exec-child.test.ts +273 -0
  463. package/src/dsl/workflow-fetch.test.ts +138 -0
  464. package/src/dsl/workflow-import.test.ts +89 -0
  465. package/src/dsl/workflow.test.ts +122 -0
  466. package/src/dsl.ts +3 -0
  467. package/src/errors.ts +101 -0
  468. package/src/index.ts +41 -0
  469. package/src/iterative-generation/activities/extractToc.ts +63 -0
  470. package/src/iterative-generation/activities/finalizeOutput.ts +100 -0
  471. package/src/iterative-generation/activities/generatePart.ts +123 -0
  472. package/src/iterative-generation/activities/generateToc.ts +116 -0
  473. package/src/iterative-generation/activities/index.ts +4 -0
  474. package/src/iterative-generation/iterativeGenerationWorkflow.ts +68 -0
  475. package/src/iterative-generation/types.ts +99 -0
  476. package/src/iterative-generation/utils.ts +126 -0
  477. package/src/result-types.ts +25 -0
  478. package/src/system/notifyWebhookWorkflow.ts +70 -0
  479. package/src/system/recalculateEmbeddingsWorkflow.ts +41 -0
  480. package/src/utils/auth.ts +10 -0
  481. package/src/utils/blobs.ts +59 -0
  482. package/src/utils/chunks.ts +17 -0
  483. package/src/utils/client.ts +46 -0
  484. package/src/utils/expand-vars.ts +31 -0
  485. package/src/utils/memory.ts +61 -0
  486. package/src/utils/renditions.ts +127 -0
  487. package/src/utils/storage.ts +60 -0
  488. package/src/utils/tokens.ts +44 -0
  489. package/src/vars.ts +3 -0
  490. package/src/workflows.ts +7 -0
@@ -0,0 +1 @@
1
+ {"version":3,"file":"updateDocumentFromInteractionRun.js","sourceRoot":"","sources":["../../../../src/activities/advanced/updateDocumentFromInteractionRun.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,iBAAiB,EAAE,MAAM,kBAAkB,CAAC;AAErD,OAAO,EAAE,aAAa,EAAE,MAAM,oCAAoC,CAAC;AACnE,OAAO,EAAE,0BAA0B,EAAE,MAAM,iBAAiB,CAAC;AAe7D,MAAM,CAAC,KAAK,UAAU,gCAAgC,CAAC,OAA4E;IAC/H,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,GAAG,MAAM,aAAa,CAAyC,OAAO,CAAC,CAAC;IAE1G,IAAI,CAAC,MAAM,CAAC,GAAG,EAAE,CAAC;QACd,MAAM,IAAI,0BAA0B,CAAC,KAAK,EAAE,OAAO,CAAC,QAAQ,CAAC,CAAC;IAClE,CAAC;IAED,MAAM,QAAQ,GAAG,iBAAiB,CAAC,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,MAAM,EAAE,CAAC;IAEpE,IAAI,CAAC,QAAQ,EAAE,CAAC;QACZ,OAAO,EAAE,MAAM,EAAE,QAAQ,EAAE,KAAK,EAAE,UAAU,EAAE,CAAC;IACnD,CAAC;IAED,MAAM,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,QAAQ,EAAE,QAAQ,EAAE,EAAE,iBAAiB,EAAE,IAAI,EAAE,CAAC,CAAC;IAE7E,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,CAAC;AACjC,CAAC"}
@@ -0,0 +1,82 @@
1
+ import { log } from "@temporalio/activity";
2
+ import { setupActivity } from "../dsl/setup/ActivityContext.js";
3
+ import { executeInteractionFromActivity } from "./executeInteraction.js";
4
+ const INT_CHUNK_DOCUMENT = "sys:ChunkDocument";
5
+ export async function chunkDocument(payload) {
6
+ const { params, client, objectId } = await setupActivity(payload);
7
+ const { force } = params;
8
+ const interactionName = params.interactionName ?? INT_CHUNK_DOCUMENT;
9
+ log.info(`Object ${objectId} chunking started`);
10
+ const document = await client.objects.retrieve(objectId, "+text");
11
+ const type = document.type ? await client.types.retrieve(document.type.id) : undefined;
12
+ if (!type?.is_chunkable) {
13
+ log.warn('Type is not chunkable for object ID: ' + objectId);
14
+ return { id: objectId, status: "skipped", message: "type not chunkable" };
15
+ }
16
+ //check if text is present
17
+ if (!document.text) {
18
+ log.warn('No text found for object ID: ' + objectId);
19
+ return { id: objectId, status: "failed", message: "no text found" };
20
+ }
21
+ if (!force && document.parts && document.parts.length > 0 && document.parts_etag === document.text_etag) {
22
+ log.info('Document already chunked for object ID: ' + objectId);
23
+ return { id: objectId, status: "skipped", message: "document already chunked with correct etag" };
24
+ }
25
+ //instrument the text with line numbers
26
+ const lines = document.text.split('\n');
27
+ const instrumented = lines.map((l, i) => `{%${i}%}${l}`).join('\n');
28
+ const res = await executeInteractionFromActivity(client, interactionName, params, {
29
+ objectId: objectId,
30
+ content: instrumented
31
+ });
32
+ const jsonResult = res.result.object();
33
+ const parts = jsonResult.parts;
34
+ if (!parts || parts.length === 0) {
35
+ log.warn('No parts found for object ID: ' + objectId, res);
36
+ return { id: objectId, status: "failed", parts: [], message: "no parts found" };
37
+ }
38
+ /**
39
+ * Only create parts as document if the flag is set
40
+ */
41
+ if (params.createParts) {
42
+ const partDocs = await Promise.all(parts.map(async (part, i) => {
43
+ const text = lines.filter((_l, i) => i >= part.line_number_start && i <= part.line_number_end).join('\n');
44
+ const location = () => {
45
+ let location = document.location;
46
+ if (location.endsWith('/')) {
47
+ location += document.name + "/" + part.type;
48
+ }
49
+ location += '/' + document.name + "/" + part.type;
50
+ return location;
51
+ };
52
+ const docPart = await client.objects.create({
53
+ name: part.name,
54
+ parent: objectId,
55
+ text: text,
56
+ location: location(),
57
+ properties: {
58
+ part_number: i + 1,
59
+ etag: document.text_etag,
60
+ source_line_start: part.line_number_start,
61
+ source_line_end: part.line_number_end,
62
+ title: part.name
63
+ }
64
+ });
65
+ return docPart;
66
+ }));
67
+ //delete previous parts
68
+ if (document.parts && document.parts.length > 0) {
69
+ log.info('Deleting previous parts for object ID: ' + objectId, { parts: document.parts });
70
+ await Promise.all(document.parts.map(async (partId) => {
71
+ await client.objects.delete(partId);
72
+ }));
73
+ }
74
+ await client.objects.update(objectId, {
75
+ parts: partDocs.map(p => p.id),
76
+ parts_etag: document.text_etag
77
+ });
78
+ }
79
+ log.info(`Object ${objectId} chunking completed`, { parts: document.parts });
80
+ return { id: objectId, status: "completed", parts: document.parts };
81
+ }
82
+ //# sourceMappingURL=chunkDocument.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"chunkDocument.js","sourceRoot":"","sources":["../../../src/activities/chunkDocument.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAC;AAE3C,OAAO,EAAE,aAAa,EAAE,MAAM,iCAAiC,CAAC;AAEhE,OAAO,EAA8B,8BAA8B,EAAE,MAAM,yBAAyB,CAAC;AAErG,MAAM,kBAAkB,GAAG,mBAAmB,CAAA;AAyC9C,MAAM,CAAC,KAAK,UAAU,aAAa,CAAC,OAAyD;IACzF,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,GAAG,MAAM,aAAa,CAAsB,OAAO,CAAC,CAAC;IAEvF,MAAM,EAAE,KAAK,EAAE,GAAG,MAAM,CAAC;IACzB,MAAM,eAAe,GAAG,MAAM,CAAC,eAAe,IAAI,kBAAkB,CAAC;IAErE,GAAG,CAAC,IAAI,CAAC,UAAU,QAAQ,mBAAmB,CAAC,CAAC;IAEhD,MAAM,QAAQ,GAAG,MAAM,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC;IAElE,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,MAAM,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;IAEvF,IAAI,CAAC,IAAI,EAAE,YAAY,EAAE,CAAC;QACtB,GAAG,CAAC,IAAI,CAAC,uCAAuC,GAAG,QAAQ,CAAC,CAAC;QAC7D,OAAO,EAAE,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,oBAAoB,EAAE,CAAA;IAC7E,CAAC;IAED,0BAA0B;IAC1B,IAAI,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAC;QACjB,GAAG,CAAC,IAAI,CAAC,+BAA+B,GAAG,QAAQ,CAAC,CAAC;QACrD,OAAO,EAAE,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,QAAQ,EAAE,OAAO,EAAE,eAAe,EAAE,CAAA;IACvE,CAAC;IAED,IAAI,CAAC,KAAK,IAAI,QAAQ,CAAC,KAAK,IAAI,QAAQ,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,QAAQ,CAAC,UAAU,KAAK,QAAQ,CAAC,SAAS,EAAE,CAAC;QACtG,GAAG,CAAC,IAAI,CAAC,0CAA0C,GAAG,QAAQ,CAAC,CAAC;QAChE,OAAO,EAAE,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,4CAA4C,EAAE,CAAA;IACrG,CAAC;IAED,uCAAuC;IACvC,MAAM,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAA;IACvC,MAAM,YAAY,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;IAEnE,MAAM,GAAG,GAAG,MAAM,8BAA8B,CAAC,MAAM,EAAE,eAAe,EAAE,MAAM,EAAE;QAC9E,QAAQ,EAAE,QAAQ;QAClB,OAAO,EAAE,YAAY;KACxB,CAAC,CAAC;IAEH,MAAM,UAAU,GAAG,GAAG,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC;IAEvC,MAAM,KAAK,GAAG,UAAU,CAAC,KAAkB,CAAC;IAC5C,IAAI,CAAC,KAAK,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC/B,GAAG,CAAC,IAAI,CAAC,gCAAgC,GAAG,QAAQ,EAAE,GAAG,CAAC,CAAC;QAC3D,OAAO,EAAE,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,QAAQ,EAAE,KAAK,EAAE,EAAE,EAAE,OAAO,EAAE,gBAAgB,EAAE,CAAA;IACnF,CAAC;IAGD;;OAEG;IACH,IAAI,MAAM,CAAC,WAAW,EAAE,CAAC;QAErB,MAAM,QAAQ,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,KAAK,EAAE,IAAI,EAAE,CAAC,EAAE,EAAE;YAE3D,MAAM,IAAI,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC,EAAE,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,IAAI,IAAI,CAAC,iBAAiB,IAAI,CAAC,IAAI,IAAI,CAAC,eAAe,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YAE1G,MAAM,QAAQ,GAAG,GAAG,EAAE;gBAClB,IAAI,QAAQ,GAAG,QAAQ,CAAC,QAAQ,CAAC;gBACjC,IAAI,QAAQ,CAAC,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC;oBACzB,QAAQ,IAAI,QAAQ,CAAC,IAAI,GAAG,GAAG,GAAG,IAAI,CAAC,IAAI,CAAA;gBAC/C,CAAC;gBACD,QAAQ,IAAI,GAAG,GAAG,QAAQ,CAAC,IAAI,GAAG,GAAG,GAAG,IAAI,CAAC,IAAI,CAAC;gBAClD,OAAO,QAAQ,CAAC;YACpB,CAAC,CAAA;YAED,MAAM,OAAO,GAAG,MAAM,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC;gBACxC,IAAI,EAAE,IAAI,CAAC,IAAI;gBACf,MAAM,EAAE,QAAQ;gBAChB,IAAI,EAAE,IAAI;gBACV,QAAQ,EAAE,QAAQ,EAAE;gBACpB,UAAU,EAAE;oBACR,WAAW,EAAE,CAAC,GAAG,CAAC;oBAClB,IAAI,EAAE,QAAQ,CAAC,SAAS;oBACxB,iBAAiB,EAAE,IAAI,CAAC,iBAAiB;oBACzC,eAAe,EAAE,IAAI,CAAC,eAAe;oBACrC,KAAK,EAAE,IAAI,CAAC,IAAI;iBACnB;aACJ,CAAC,CAAC;YACH,OAAO,OAAO,CAAC;QACnB,CAAC,CAAC,CAAC,CAAC;QAEJ,uBAAuB;QACvB,IAAI,QAAQ,CAAC,KAAK,IAAI,QAAQ,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAC9C,GAAG,CAAC,IAAI,CAAC,yCAAyC,GAAG,QAAQ,EAAE,EAAE,KAAK,EAAE,QAAQ,CAAC,KAAK,EAAE,CAAC,CAAC;YAC1F,MAAM,OAAO,CAAC,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,GAAG,CAAC,KAAK,EAAE,MAAM,EAAE,EAAE;gBAClD,MAAM,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;YACxC,CAAC,CAAC,CAAC,CAAC;QACR,CAAC;QAED,MAAM,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,QAAQ,EAAE;YAClC,KAAK,EAAE,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;YAC9B,UAAU,EAAE,QAAQ,CAAC,SAAS;SACjC,CAAC,CAAC;IACP,CAAC;IAED,GAAG,CAAC,IAAI,CAAC,UAAU,QAAQ,qBAAqB,EAAE,EAAE,KAAK,EAAE,QAAQ,CAAC,KAAK,EAAE,CAAC,CAAC;IAE7E,OAAO,EAAE,EAAE,EAAE,QAAQ,EAAE,MAAM,EAAE,WAAW,EAAE,KAAK,EAAE,QAAQ,CAAC,KAAK,EAAE,CAAA;AAEvE,CAAC"}
@@ -0,0 +1,124 @@
1
+ import { log, activityInfo } from "@temporalio/activity";
2
+ import { NodeStreamSource } from "@vertesia/client/node";
3
+ import { Readable } from "stream";
4
+ import { setupActivity } from "../dsl/setup/ActivityContext.js";
5
+ /**
6
+ * Directories in the agent workspace to copy from parent to child.
7
+ */
8
+ const WORKSPACE_DIRECTORIES = ['scripts', 'files', 'skills', 'docs', 'out'];
9
+ /**
10
+ * Files that should never be copied (child has its own).
11
+ */
12
+ const EXCLUDED_FILES = ['conversation.json', 'tools.json'];
13
+ /**
14
+ * Copy workspace artifacts from parent workflow's agent space to current workflow's agent space.
15
+ *
16
+ * Copies: scripts/, files/, skills/, docs/, out/
17
+ * Excludes: conversation.json, tools.json
18
+ */
19
+ export async function copyParentArtifacts(payload) {
20
+ const { client, params } = await setupActivity(payload);
21
+ const childRunId = activityInfo().workflowExecution.runId;
22
+ const parentRunId = params.parent_run_id;
23
+ log.info(`Copying artifacts from parent ${parentRunId} to child ${childRunId}`);
24
+ // List all files in parent's agent space
25
+ const parentFiles = await client.files.listArtifacts(parentRunId);
26
+ if (parentFiles.length === 0) {
27
+ log.info("No artifacts in parent agent space");
28
+ return { copied: 0, files: [] };
29
+ }
30
+ const copiedFiles = [];
31
+ for (const fullPath of parentFiles) {
32
+ const relativePath = extractRelativePath(fullPath, parentRunId);
33
+ if (!relativePath)
34
+ continue;
35
+ // Only copy workspace directories
36
+ const dir = relativePath.split('/')[0];
37
+ if (!WORKSPACE_DIRECTORIES.includes(dir))
38
+ continue;
39
+ // Skip excluded files
40
+ const fileName = relativePath.split('/').pop() || '';
41
+ if (EXCLUDED_FILES.includes(fileName))
42
+ continue;
43
+ try {
44
+ await copyArtifact(client, parentRunId, childRunId, relativePath);
45
+ copiedFiles.push(relativePath);
46
+ }
47
+ catch (err) {
48
+ log.warn(`Failed to copy ${relativePath}: ${err.message}`);
49
+ }
50
+ }
51
+ log.info(`Copied ${copiedFiles.length} artifacts from parent`);
52
+ return { copied: copiedFiles.length, files: copiedFiles };
53
+ }
54
+ /**
55
+ * Extract relative path from full artifact storage path.
56
+ * Storage paths can be: "store_xxx/agents/{runId}/..." or "agents/{runId}/..."
57
+ */
58
+ function extractRelativePath(fullPath, runId) {
59
+ // Look for the pattern: agents/{runId}/
60
+ const marker = `/agents/${runId}/`;
61
+ const idx = fullPath.lastIndexOf(marker);
62
+ if (idx >= 0) {
63
+ return fullPath.slice(idx + marker.length);
64
+ }
65
+ // Also try without leading slash
66
+ const legacyPrefix = `agents/${runId}/`;
67
+ if (fullPath.startsWith(legacyPrefix)) {
68
+ return fullPath.slice(legacyPrefix.length);
69
+ }
70
+ // If path contains the runId, try to extract after it
71
+ const runIdIdx = fullPath.indexOf(runId);
72
+ if (runIdIdx >= 0) {
73
+ const afterRunId = fullPath.slice(runIdIdx + runId.length);
74
+ if (afterRunId.startsWith('/')) {
75
+ return afterRunId.slice(1);
76
+ }
77
+ }
78
+ return null;
79
+ }
80
+ /**
81
+ * Copy a single artifact from parent to child agent space
82
+ */
83
+ async function copyArtifact(client, parentRunId, childRunId, relativePath) {
84
+ // Download from parent's agent space
85
+ const stream = await client.files.downloadArtifact(parentRunId, relativePath);
86
+ // Convert web stream to node stream for upload
87
+ const nodeStream = Readable.fromWeb(stream);
88
+ // Determine mime type from extension
89
+ const mimeType = getMimeType(relativePath);
90
+ const fileName = relativePath.split('/').pop() || relativePath;
91
+ // Upload to child's agent space at the same relative path
92
+ const source = new NodeStreamSource(nodeStream, fileName, mimeType);
93
+ await client.files.uploadArtifact(childRunId, relativePath, source);
94
+ }
95
+ /**
96
+ * Get mime type from file extension
97
+ */
98
+ function getMimeType(filePath) {
99
+ const ext = filePath.split('.').pop()?.toLowerCase();
100
+ const mimeTypes = {
101
+ 'json': 'application/json',
102
+ 'txt': 'text/plain',
103
+ 'md': 'text/markdown',
104
+ 'html': 'text/html',
105
+ 'csv': 'text/csv',
106
+ 'png': 'image/png',
107
+ 'jpg': 'image/jpeg',
108
+ 'jpeg': 'image/jpeg',
109
+ 'gif': 'image/gif',
110
+ 'svg': 'image/svg+xml',
111
+ 'pdf': 'application/pdf',
112
+ 'xml': 'application/xml',
113
+ 'zip': 'application/zip',
114
+ 'js': 'application/javascript',
115
+ 'ts': 'application/typescript',
116
+ 'py': 'text/x-python',
117
+ 'sh': 'text/x-shellscript',
118
+ 'bash': 'text/x-shellscript',
119
+ 'yaml': 'text/yaml',
120
+ 'yml': 'text/yaml',
121
+ };
122
+ return mimeTypes[ext || ''] || 'application/octet-stream';
123
+ }
124
+ //# sourceMappingURL=copyParentArtifacts.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"copyParentArtifacts.js","sourceRoot":"","sources":["../../../src/activities/copyParentArtifacts.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,GAAG,EAAE,YAAY,EAAE,MAAM,sBAAsB,CAAC;AACzD,OAAO,EAAE,gBAAgB,EAAE,MAAM,uBAAuB,CAAC;AAEzD,OAAO,EAAE,QAAQ,EAAE,MAAM,QAAQ,CAAC;AAClC,OAAO,EAAE,aAAa,EAAE,MAAM,iCAAiC,CAAC;AAEhE;;GAEG;AACH,MAAM,qBAAqB,GAAG,CAAC,SAAS,EAAE,OAAO,EAAE,QAAQ,EAAE,MAAM,EAAE,KAAK,CAAC,CAAC;AAE5E;;GAEG;AACH,MAAM,cAAc,GAAG,CAAC,mBAAmB,EAAE,YAAY,CAAC,CAAC;AAW3D;;;;;GAKG;AACH,MAAM,CAAC,KAAK,UAAU,mBAAmB,CACrC,OAA+D;IAE/D,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,GAAG,MAAM,aAAa,CAA4B,OAAO,CAAC,CAAC;IACnF,MAAM,UAAU,GAAG,YAAY,EAAE,CAAC,iBAAiB,CAAC,KAAK,CAAC;IAC1D,MAAM,WAAW,GAAG,MAAM,CAAC,aAAa,CAAC;IAEzC,GAAG,CAAC,IAAI,CAAC,iCAAiC,WAAW,aAAa,UAAU,EAAE,CAAC,CAAC;IAEhF,yCAAyC;IACzC,MAAM,WAAW,GAAG,MAAM,MAAM,CAAC,KAAK,CAAC,aAAa,CAAC,WAAW,CAAC,CAAC;IAElE,IAAI,WAAW,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC3B,GAAG,CAAC,IAAI,CAAC,oCAAoC,CAAC,CAAC;QAC/C,OAAO,EAAE,MAAM,EAAE,CAAC,EAAE,KAAK,EAAE,EAAE,EAAE,CAAC;IACpC,CAAC;IAED,MAAM,WAAW,GAAa,EAAE,CAAC;IAEjC,KAAK,MAAM,QAAQ,IAAI,WAAW,EAAE,CAAC;QACjC,MAAM,YAAY,GAAG,mBAAmB,CAAC,QAAQ,EAAE,WAAW,CAAC,CAAC;QAChE,IAAI,CAAC,YAAY;YAAE,SAAS;QAE5B,kCAAkC;QAClC,MAAM,GAAG,GAAG,YAAY,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;QACvC,IAAI,CAAC,qBAAqB,CAAC,QAAQ,CAAC,GAAG,CAAC;YAAE,SAAS;QAEnD,sBAAsB;QACtB,MAAM,QAAQ,GAAG,YAAY,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,IAAI,EAAE,CAAC;QACrD,IAAI,cAAc,CAAC,QAAQ,CAAC,QAAQ,CAAC;YAAE,SAAS;QAEhD,IAAI,CAAC;YACD,MAAM,YAAY,CAAC,MAAM,EAAE,WAAW,EAAE,UAAU,EAAE,YAAY,CAAC,CAAC;YAClE,WAAW,CAAC,IAAI,CAAC,YAAY,CAAC,CAAC;QACnC,CAAC;QAAC,OAAO,GAAQ,EAAE,CAAC;YAChB,GAAG,CAAC,IAAI,CAAC,kBAAkB,YAAY,KAAK,GAAG,CAAC,OAAO,EAAE,CAAC,CAAC;QAC/D,CAAC;IACL,CAAC;IAED,GAAG,CAAC,IAAI,CAAC,UAAU,WAAW,CAAC,MAAM,wBAAwB,CAAC,CAAC;IAC/D,OAAO,EAAE,MAAM,EAAE,WAAW,CAAC,MAAM,EAAE,KAAK,EAAE,WAAW,EAAE,CAAC;AAC9D,CAAC;AAED;;;GAGG;AACH,SAAS,mBAAmB,CAAC,QAAgB,EAAE,KAAa;IACxD,wCAAwC;IACxC,MAAM,MAAM,GAAG,WAAW,KAAK,GAAG,CAAC;IACnC,MAAM,GAAG,GAAG,QAAQ,CAAC,WAAW,CAAC,MAAM,CAAC,CAAC;IACzC,IAAI,GAAG,IAAI,CAAC,EAAE,CAAC;QACX,OAAO,QAAQ,CAAC,KAAK,CAAC,GAAG,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC;IAC/C,CAAC;IAED,iCAAiC;IACjC,MAAM,YAAY,GAAG,UAAU,KAAK,GAAG,CAAC;IACxC,IAAI,QAAQ,CAAC,UAAU,CAAC,YAAY,CAAC,EAAE,CAAC;QACpC,OAAO,QAAQ,CAAC,KAAK,CAAC,YAAY,CAAC,MAAM,CAAC,CAAC;IAC/C,CAAC;IAED,sDAAsD;IACtD,MAAM,QAAQ,GAAG,QAAQ,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC;IACzC,IAAI,QAAQ,IAAI,CAAC,EAAE,CAAC;QAChB,MAAM,UAAU,GAAG,QAAQ,CAAC,KAAK,CAAC,QAAQ,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC;QAC3D,IAAI,UAAU,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;YAC7B,OAAO,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;QAC/B,CAAC;IACL,CAAC;IAED,OAAO,IAAI,CAAC;AAChB,CAAC;AAED;;GAEG;AACH,KAAK,UAAU,YAAY,CACvB,MAAW,EACX,WAAmB,EACnB,UAAkB,EAClB,YAAoB;IAEpB,qCAAqC;IACrC,MAAM,MAAM,GAAG,MAAM,MAAM,CAAC,KAAK,CAAC,gBAAgB,CAAC,WAAW,EAAE,YAAY,CAAC,CAAC;IAE9E,+CAA+C;IAC/C,MAAM,UAAU,GAAG,QAAQ,CAAC,OAAO,CAAC,MAAa,CAAC,CAAC;IAEnD,qCAAqC;IACrC,MAAM,QAAQ,GAAG,WAAW,CAAC,YAAY,CAAC,CAAC;IAC3C,MAAM,QAAQ,GAAG,YAAY,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,IAAI,YAAY,CAAC;IAE/D,0DAA0D;IAC1D,MAAM,MAAM,GAAG,IAAI,gBAAgB,CAC/B,UAAU,EACV,QAAQ,EACR,QAAQ,CACX,CAAC;IAEF,MAAM,MAAM,CAAC,KAAK,CAAC,cAAc,CAAC,UAAU,EAAE,YAAY,EAAE,MAAM,CAAC,CAAC;AACxE,CAAC;AAED;;GAEG;AACH,SAAS,WAAW,CAAC,QAAgB;IACjC,MAAM,GAAG,GAAG,QAAQ,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,WAAW,EAAE,CAAC;IACrD,MAAM,SAAS,GAA2B;QACtC,MAAM,EAAE,kBAAkB;QAC1B,KAAK,EAAE,YAAY;QACnB,IAAI,EAAE,eAAe;QACrB,MAAM,EAAE,WAAW;QACnB,KAAK,EAAE,UAAU;QACjB,KAAK,EAAE,WAAW;QAClB,KAAK,EAAE,YAAY;QACnB,MAAM,EAAE,YAAY;QACpB,KAAK,EAAE,WAAW;QAClB,KAAK,EAAE,eAAe;QACtB,KAAK,EAAE,iBAAiB;QACxB,KAAK,EAAE,iBAAiB;QACxB,KAAK,EAAE,iBAAiB;QACxB,IAAI,EAAE,wBAAwB;QAC9B,IAAI,EAAE,wBAAwB;QAC9B,IAAI,EAAE,eAAe;QACrB,IAAI,EAAE,oBAAoB;QAC1B,MAAM,EAAE,oBAAoB;QAC5B,MAAM,EAAE,WAAW;QACnB,KAAK,EAAE,WAAW;KACrB,CAAC;IACF,OAAO,SAAS,CAAC,GAAG,IAAI,EAAE,CAAC,IAAI,0BAA0B,CAAC;AAC9D,CAAC"}
@@ -0,0 +1,58 @@
1
+ import { log } from "@temporalio/activity";
2
+ import { NodeStreamSource } from "@vertesia/client/node";
3
+ import fs from 'fs';
4
+ import { pdfExtractPages } from "../conversion/mutool.js";
5
+ import { setupActivity } from "../dsl/setup/ActivityContext.js";
6
+ import { DocumentNotFoundError } from "../errors.js";
7
+ import { saveBlobToTempFile } from "../utils/blobs.js";
8
+ /**
9
+ * Create a new PDF by extracting pages from a source PDF
10
+ * @returns
11
+ */
12
+ export async function createPdfDocumentFromSource(payload) {
13
+ const { client, objectId, params } = await setupActivity(payload);
14
+ const inputObject = await client.objects.retrieve(objectId);
15
+ const { pages, filename, title } = params;
16
+ log.info(`Creating PDF from source`, { objectId, pages, filename, title });
17
+ if (!pages || pages.length === 0) {
18
+ log.error(`No pages provided`);
19
+ throw new Error(`No pages provided`);
20
+ }
21
+ if (!inputObject) {
22
+ log.error(`Document ${objectId} not found`);
23
+ throw new DocumentNotFoundError(`Document ${objectId} not found`, [objectId]);
24
+ }
25
+ if (!inputObject.content?.source) {
26
+ log.error(`Document ${objectId} has no source`);
27
+ throw new DocumentNotFoundError(`Document ${objectId} has no source`, [objectId]);
28
+ }
29
+ if (!inputObject.content.type || (!inputObject.content.type?.startsWith('application/pdf'))) {
30
+ log.error(`Document ${objectId} is not an image`);
31
+ throw new DocumentNotFoundError(`Document ${objectId} is not an image or pdf: ${inputObject.content.type}`, [objectId]);
32
+ }
33
+ const targetType = await client.types.getTypeByName(params.target_object_type);
34
+ if (!targetType) {
35
+ log.error(`Type ${params.target_object_type} not found`);
36
+ throw new DocumentNotFoundError(`Type ${params.target_object_type} not found`);
37
+ }
38
+ const tmpFile = await saveBlobToTempFile(client, inputObject.content.source, ".pdf");
39
+ const newPdf = await pdfExtractPages(tmpFile, pages);
40
+ log.info(`PDF created from pages ${pages.join(', ')} `, { newPdf });
41
+ const name = `pages-${pages.join('-')}.pdf`;
42
+ const sourceToUpload = new NodeStreamSource(fs.createReadStream(newPdf), name, "application/pdf");
43
+ log.info(`Uploading file ${newPdf} `);
44
+ const upload = await client.objects.upload(sourceToUpload);
45
+ log.info(`File uploaded ${upload.source} `);
46
+ const newObject = await client.objects.create({
47
+ type: targetType.id,
48
+ name: title || targetType.name,
49
+ parent: objectId,
50
+ content: {
51
+ source: upload.source,
52
+ name: upload.name,
53
+ type: 'application/pdf'
54
+ }
55
+ });
56
+ return { newObjectId: newObject.id, uploadedFile: upload.name };
57
+ }
58
+ //# sourceMappingURL=createDocumentFromOther.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"createDocumentFromOther.js","sourceRoot":"","sources":["../../../src/activities/createDocumentFromOther.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAC;AAC3C,OAAO,EAAE,gBAAgB,EAAE,MAAM,uBAAuB,CAAC;AAEzD,OAAO,EAAE,MAAM,IAAI,CAAC;AACpB,OAAO,EAAE,eAAe,EAAE,MAAM,yBAAyB,CAAC;AAC1D,OAAO,EAAE,aAAa,EAAE,MAAM,iCAAiC,CAAC;AAChE,OAAO,EAAE,qBAAqB,EAAE,MAAM,cAAc,CAAC;AACrD,OAAO,EAAE,kBAAkB,EAAE,MAAM,mBAAmB,CAAC;AAkBvD;;;GAGG;AACH,MAAM,CAAC,KAAK,UAAU,2BAA2B,CAAC,OAAuE;IACrH,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,EAAE,GAAG,MAAM,aAAa,CAAoC,OAAO,CAAC,CAAC;IACrG,MAAM,WAAW,GAAG,MAAM,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC;IAE5D,MAAM,EAAE,KAAK,EAAE,QAAQ,EAAE,KAAK,EAAE,GAAG,MAAM,CAAC;IAC1C,GAAG,CAAC,IAAI,CAAC,0BAA0B,EAAE,EAAE,QAAQ,EAAE,KAAK,EAAE,QAAQ,EAAE,KAAK,EAAE,CAAC,CAAC;IAE3E,IAAI,CAAC,KAAK,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC/B,GAAG,CAAC,KAAK,CAAC,mBAAmB,CAAC,CAAC;QAC/B,MAAM,IAAI,KAAK,CAAC,mBAAmB,CAAC,CAAC;IACzC,CAAC;IAED,IAAI,CAAC,WAAW,EAAE,CAAC;QACf,GAAG,CAAC,KAAK,CAAC,YAAY,QAAQ,YAAY,CAAC,CAAC;QAC5C,MAAM,IAAI,qBAAqB,CAAC,YAAY,QAAQ,YAAY,EAAE,CAAC,QAAQ,CAAC,CAAC,CAAC;IAClF,CAAC;IAED,IAAI,CAAC,WAAW,CAAC,OAAO,EAAE,MAAM,EAAE,CAAC;QAC/B,GAAG,CAAC,KAAK,CAAC,YAAY,QAAQ,gBAAgB,CAAC,CAAC;QAChD,MAAM,IAAI,qBAAqB,CAAC,YAAY,QAAQ,gBAAgB,EAAE,CAAC,QAAQ,CAAC,CAAC,CAAC;IACtF,CAAC;IAED,IAAI,CAAC,WAAW,CAAC,OAAO,CAAC,IAAI,IAAI,CAAC,CAAC,WAAW,CAAC,OAAO,CAAC,IAAI,EAAE,UAAU,CAAC,iBAAiB,CAAC,CAAC,EAAE,CAAC;QAC1F,GAAG,CAAC,KAAK,CAAC,YAAY,QAAQ,kBAAkB,CAAC,CAAC;QAClD,MAAM,IAAI,qBAAqB,CAAC,YAAY,QAAQ,4BAA4B,WAAW,CAAC,OAAO,CAAC,IAAI,EAAE,EAAE,CAAC,QAAQ,CAAC,CAAC,CAAC;IAC5H,CAAC;IAED,MAAM,UAAU,GAAG,MAAM,MAAM,CAAC,KAAK,CAAC,aAAa,CAAC,MAAM,CAAC,kBAAkB,CAAC,CAAC;IAC/E,IAAI,CAAC,UAAU,EAAE,CAAC;QACd,GAAG,CAAC,KAAK,CAAC,QAAQ,MAAM,CAAC,kBAAkB,YAAY,CAAC,CAAC;QACzD,MAAM,IAAI,qBAAqB,CAAC,QAAQ,MAAM,CAAC,kBAAkB,YAAY,CAAC,CAAC;IACnF,CAAC;IAED,MAAM,OAAO,GAAG,MAAM,kBAAkB,CAAC,MAAM,EAAE,WAAW,CAAC,OAAO,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;IACrF,MAAM,MAAM,GAAG,MAAM,eAAe,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC;IACrD,GAAG,CAAC,IAAI,CAAC,0BAA0B,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,EAAE,EAAE,MAAM,EAAE,CAAC,CAAC;IACpE,MAAM,IAAI,GAAG,SAAS,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC;IAE5C,MAAM,cAAc,GAAG,IAAI,gBAAgB,CACvC,EAAE,CAAC,gBAAgB,CAAC,MAAM,CAAC,EAC3B,IAAI,EACJ,iBAAiB,CACpB,CAAA;IAED,GAAG,CAAC,IAAI,CAAC,kBAAkB,MAAM,GAAG,CAAC,CAAC;IACtC,MAAM,MAAM,GAAG,MAAM,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,cAAc,CAAC,CAAC;IAC3D,GAAG,CAAC,IAAI,CAAC,iBAAiB,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC;IAE5C,MAAM,SAAS,GAAG,MAAM,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC;QAC1C,IAAI,EAAE,UAAU,CAAC,EAAE;QACnB,IAAI,EAAE,KAAK,IAAI,UAAU,CAAC,IAAI;QAC9B,MAAM,EAAE,QAAQ;QAChB,OAAO,EAAE;YACL,MAAM,EAAE,MAAM,CAAC,MAAM;YACrB,IAAI,EAAE,MAAM,CAAC,IAAI;YACjB,IAAI,EAAE,iBAAiB;SAC1B;KACJ,CAAC,CAAC;IAEH,OAAO,EAAE,WAAW,EAAE,SAAS,CAAC,EAAE,EAAE,YAAY,EAAE,MAAM,CAAC,IAAI,EAAE,CAAC;AAGpE,CAAC"}
@@ -0,0 +1,190 @@
1
+ import { activityInfo, log } from "@temporalio/activity";
2
+ import { NodeStreamSource } from "@vertesia/client/node";
3
+ import { ExecutionRunStatus, } from "@vertesia/common";
4
+ import { projectResult } from "../dsl/projections.js";
5
+ import { setupActivity } from "../dsl/setup/ActivityContext.js";
6
+ import { ActivityParamInvalidError, ActivityParamNotFoundError, ResourceExhaustedError } from "../errors.js";
7
+ import { truncByMaxTokens } from "../utils/tokens.js";
8
+ import { Readable } from "stream";
9
+ //Example:
10
+ //@ts-ignore
11
+ const JSON = {
12
+ name: "executeInteraction",
13
+ import: ["defaultModel", "guidlineId", "docTypeId"],
14
+ params: {
15
+ defaultModel: "${model}",
16
+ interactionName: "GenerateSummary",
17
+ model: "${defaultModel ?? 'gpt4'}",
18
+ environment: "13456",
19
+ max_tokens: 100,
20
+ temperature: 0.5,
21
+ tags: ["test"],
22
+ result_schema: "${docType.object_schema}",
23
+ prompt_data: {
24
+ documents: "${documents}",
25
+ guidline: "${guidline.text}",
26
+ },
27
+ },
28
+ fetch: {
29
+ documents: {
30
+ type: "document",
31
+ query: {
32
+ id: { $in: "${objectIds}" },
33
+ },
34
+ select: "+text",
35
+ },
36
+ guidline: {
37
+ type: "document",
38
+ limit: 1,
39
+ query: {
40
+ id: "${guidlineId}",
41
+ },
42
+ select: "+text",
43
+ on_not_found: "throw",
44
+ },
45
+ docType: {
46
+ type: "document_type",
47
+ limit: 1,
48
+ query: {
49
+ id: "${docTypeId}",
50
+ },
51
+ select: "+object_schema",
52
+ },
53
+ },
54
+ };
55
+ export async function executeInteraction(payload) {
56
+ const { client, params } = await setupActivity(payload);
57
+ const { interactionName, prompt_data, static_prompt_data: wf_prompt_data } = params;
58
+ if (wf_prompt_data) {
59
+ Object.assign(prompt_data, wf_prompt_data);
60
+ }
61
+ if (!interactionName) {
62
+ log.error("Missing interactionName", { params });
63
+ throw new ActivityParamNotFoundError("interactionName", payload.activity);
64
+ }
65
+ if (params.truncate) {
66
+ const truncate = params.truncate;
67
+ for (const [key, value] of Object.entries(truncate)) {
68
+ prompt_data[key] = truncByMaxTokens(prompt_data[key], value);
69
+ }
70
+ }
71
+ try {
72
+ const res = await executeInteractionFromActivity(client, interactionName, params, prompt_data, payload.debug_mode);
73
+ let completionResult = res.result;
74
+ // Handle image uploads if the result contains base64 images
75
+ const imageResults = completionResult.filter(r => r.type === "image");
76
+ if (imageResults.length > 0) {
77
+ const uploadedImages = await Promise.all(completionResult.map(async (item, index) => {
78
+ if (item.type === "image") {
79
+ const image = item.value;
80
+ // Extract base64 data and create buffer
81
+ const base64Data = image.replace(/^data:image\/[a-z]+;base64,/, "");
82
+ const buffer = Buffer.from(base64Data, 'base64');
83
+ // Generate filename
84
+ const { runId } = activityInfo().workflowExecution;
85
+ const { activityId } = activityInfo();
86
+ const filename = `generated-image-${runId}-${activityId}-${index}.png`;
87
+ // Create a readable stream from the buffer
88
+ const stream = Readable.from(buffer);
89
+ const source = new NodeStreamSource(stream, filename, "image/png");
90
+ const file = await client.files.uploadFile(source);
91
+ return { type: "image", value: file };
92
+ }
93
+ return item;
94
+ }));
95
+ completionResult = uploadedImages;
96
+ }
97
+ return projectResult(payload, params, res, {
98
+ runId: res.id,
99
+ status: res.status,
100
+ result: completionResult,
101
+ });
102
+ }
103
+ catch (error) {
104
+ log.error(`Failed to execute interaction ${interactionName}`, { error });
105
+ if (error.statusCode === 429 && params.exit_on_resource_exhaustion) {
106
+ throw new ResourceExhaustedError(error.statusCode, "Resource exhausted - rate limit exceeded");
107
+ }
108
+ else if (error.message.includes("Failed to validate merged prompt schema")) {
109
+ //issue with the input data, don't retry
110
+ throw new ActivityParamInvalidError("prompt_data", payload.activity, error.message);
111
+ }
112
+ else if (error.message.includes("modelId: Path `modelId` is required")) {
113
+ //issue with the input data, don't retry
114
+ throw new ActivityParamInvalidError("model", payload.activity, error.message);
115
+ }
116
+ else {
117
+ throw new Error(`Interaction Execution failed ${interactionName}: ${error.message}`);
118
+ }
119
+ }
120
+ }
121
+ export async function executeInteractionFromActivity(client, interactionName, params, prompt_data, debug) {
122
+ const userTags = params.tags;
123
+ const info = activityInfo();
124
+ const runId = info.workflowExecution.runId;
125
+ let tags = ["workflow"];
126
+ if (userTags) {
127
+ tags = tags.concat(userTags);
128
+ }
129
+ const workflow = {
130
+ run_id: info.workflowExecution.runId,
131
+ workflow_id: info.workflowExecution.workflowId,
132
+ activity_type: info.activityType,
133
+ };
134
+ let previousStudioExecutionRun = undefined;
135
+ if (params.include_previous_error) {
136
+ //retrieve last failed run if any
137
+ if (info.attempt > 1) {
138
+ log.info("Retrying, searching for previous run", { prev_run_id: runId });
139
+ const payload = {
140
+ query: { workflow_run_ids: [runId] },
141
+ limit: 1,
142
+ };
143
+ const previousRun = await client.runs.search(payload).then((res) => {
144
+ log.info("Search results", { results: res });
145
+ return res ? (res[0] ?? undefined) : undefined;
146
+ });
147
+ if (previousRun) {
148
+ log.info("Found previous run", { previousRun });
149
+ previousStudioExecutionRun = await client.runs.retrieve(previousRun.id);
150
+ }
151
+ }
152
+ }
153
+ if (debug && previousStudioExecutionRun?.error) {
154
+ log.info(`Found previous run error`, { error: previousStudioExecutionRun?.error });
155
+ }
156
+ const config = {
157
+ environment: params.environment,
158
+ model: params.model,
159
+ model_options: params.model_options,
160
+ do_validate: params.validate_result,
161
+ };
162
+ const data = {
163
+ ...prompt_data,
164
+ previous_error: previousStudioExecutionRun?.error,
165
+ };
166
+ const result_schema = params.result_schema;
167
+ log.debug(`About to execute interaction ${interactionName}`, { config, data, result_schema, tags, workflow });
168
+ const res = await client.interactions
169
+ .executeByName(interactionName, {
170
+ config,
171
+ data,
172
+ result_schema,
173
+ tags,
174
+ stream: false,
175
+ workflow,
176
+ })
177
+ .catch((err) => {
178
+ log.error(`Error executing interaction ${interactionName}`, { err });
179
+ throw err;
180
+ });
181
+ if (debug) {
182
+ log.info(`Interaction executed ${interactionName}`, res);
183
+ }
184
+ if (res.error || res.status === ExecutionRunStatus.failed) {
185
+ log.error(`Error executing interaction ${interactionName}`, { error: res.error });
186
+ throw new Error(`Interaction Execution failed ${interactionName}: ${res.error}`);
187
+ }
188
+ return res;
189
+ }
190
+ //# sourceMappingURL=executeInteraction.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"executeInteraction.js","sourceRoot":"","sources":["../../../src/activities/executeInteraction.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,YAAY,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAC;AAEzD,OAAO,EAAE,gBAAgB,EAAE,MAAM,uBAAuB,CAAC;AACzD,OAAO,EAIH,kBAAkB,GAIrB,MAAM,kBAAkB,CAAC;AAC1B,OAAO,EAAE,aAAa,EAAE,MAAM,uBAAuB,CAAC;AACtD,OAAO,EAAE,aAAa,EAAE,MAAM,iCAAiC,CAAC;AAChE,OAAO,EAAE,yBAAyB,EAAE,0BAA0B,EAAE,sBAAsB,EAAE,MAAM,cAAc,CAAC;AAC7G,OAAO,EAAgB,gBAAgB,EAAE,MAAM,oBAAoB,CAAC;AACpE,OAAO,EAAE,QAAQ,EAAE,MAAM,QAAQ,CAAC;AAElC,UAAU;AACV,YAAY;AACZ,MAAM,IAAI,GAAoB;IAC1B,IAAI,EAAE,oBAAoB;IAC1B,MAAM,EAAE,CAAC,cAAc,EAAE,YAAY,EAAE,WAAW,CAAC;IACnD,MAAM,EAAE;QACJ,YAAY,EAAE,UAAU;QACxB,eAAe,EAAE,iBAAiB;QAClC,KAAK,EAAE,2BAA2B;QAClC,WAAW,EAAE,OAAO;QACpB,UAAU,EAAE,GAAG;QACf,WAAW,EAAE,GAAG;QAChB,IAAI,EAAE,CAAC,MAAM,CAAC;QACd,aAAa,EAAE,0BAA0B;QACzC,WAAW,EAAE;YACT,SAAS,EAAE,cAAc;YACzB,QAAQ,EAAE,kBAAkB;SAC/B;KACJ;IACD,KAAK,EAAE;QACH,SAAS,EAAE;YACP,IAAI,EAAE,UAAU;YAChB,KAAK,EAAE;gBACH,EAAE,EAAE,EAAE,GAAG,EAAE,cAAc,EAAE;aAC9B;YACD,MAAM,EAAE,OAAO;SAClB;QACD,QAAQ,EAAE;YACN,IAAI,EAAE,UAAU;YAChB,KAAK,EAAE,CAAC;YACR,KAAK,EAAE;gBACH,EAAE,EAAE,eAAe;aACtB;YACD,MAAM,EAAE,OAAO;YACf,YAAY,EAAE,OAAO;SACxB;QACD,OAAO,EAAE;YACL,IAAI,EAAE,eAAe;YACrB,KAAK,EAAE,CAAC;YACR,KAAK,EAAE;gBACH,EAAE,EAAE,cAAc;aACrB;YACD,MAAM,EAAE,gBAAgB;SAC3B;KACJ;CACJ,CAAC;AAiEF,MAAM,CAAC,KAAK,UAAU,kBAAkB,CAAC,OAA8D;IACnG,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,GAAG,MAAM,aAAa,CAA2B,OAAO,CAAC,CAAC;IAElF,MAAM,EAAE,eAAe,EAAE,WAAW,EAAE,kBAAkB,EAAE,cAAc,EAAE,GAAG,MAAM,CAAC;IACpF,IAAI,cAAc,EAAE,CAAC;QACjB,MAAM,CAAC,MAAM,CAAC,WAAW,EAAE,cAAc,CAAC,CAAC;IAC/C,CAAC;IAED,IAAI,CAAC,eAAe,EAAE,CAAC;QACnB,GAAG,CAAC,KAAK,CAAC,yBAAyB,EAAE,EAAE,MAAM,EAAE,CAAC,CAAC;QACjD,MAAM,IAAI,0BAA0B,CAAC,iBAAiB,EAAE,OAAO,CAAC,QAAQ,CAAC,CAAC;IAC9E,CAAC;IAED,IAAI,MAAM,CAAC,QAAQ,EAAE,CAAC;QAClB,MAAM,QAAQ,GAAG,MAAM,CAAC,QAAQ,CAAC;QACjC,KAAK,MAAM,CAAC,GAAG,EAAE,KAAK,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,EAAE,CAAC;YAClD,WAAW,CAAC,GAAG,CAAC,GAAG,gBAAgB,CAAC,WAAW,CAAC,GAAG,CAAC,EAAE,KAAK,CAAC,CAAC;QACjE,CAAC;IACL,CAAC;IAED,IAAI,CAAC;QACD,MAAM,GAAG,GAAG,MAAM,8BAA8B,CAC5C,MAAM,EACN,eAAe,EACf,MAAM,EACN,WAAW,EACX,OAAO,CAAC,UAAU,CACrB,CAAC;QAEF,IAAI,gBAAgB,GAAuB,GAAG,CAAC,MAAM,CAAC;QAEtD,4DAA4D;QAC5D,MAAM,YAAY,GAAG,gBAAgB,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,OAAO,CAAC,CAAC;QACtE,IAAI,YAAY,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YAC1B,MAAM,cAAc,GAAG,MAAM,OAAO,CAAC,GAAG,CACpC,gBAAgB,CAAC,GAAG,CAAC,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,EAAE;gBACvC,IAAI,IAAI,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;oBACxB,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC;oBACzB,wCAAwC;oBACxC,MAAM,UAAU,GAAG,KAAK,CAAC,OAAO,CAAC,6BAA6B,EAAE,EAAE,CAAC,CAAC;oBACpE,MAAM,MAAM,GAAG,MAAM,CAAC,IAAI,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAC;oBAEjD,oBAAoB;oBACpB,MAAM,EAAE,KAAK,EAAE,GAAG,YAAY,EAAE,CAAC,iBAAiB,CAAC;oBACnD,MAAM,EAAE,UAAU,EAAE,GAAG,YAAY,EAAE,CAAC;oBACtC,MAAM,QAAQ,GAAG,mBAAmB,KAAK,IAAI,UAAU,IAAI,KAAK,MAAM,CAAC;oBAEvE,2CAA2C;oBAC3C,MAAM,MAAM,GAAG,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;oBAErC,MAAM,MAAM,GAAG,IAAI,gBAAgB,CAC/B,MAAM,EACN,QAAQ,EACR,WAAW,CACd,CAAC;oBAEF,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,KAAK,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC;oBACnD,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,IAAI,EAAsB,CAAC;gBAC9D,CAAC;gBACD,OAAO,IAAI,CAAC;YAChB,CAAC,CAAC,CACL,CAAC;YACF,gBAAgB,GAAG,cAAc,CAAC;QACtC,CAAC;QAED,OAAO,aAAa,CAAC,OAAO,EAAE,MAAM,EAAE,GAAG,EAAE;YACvC,KAAK,EAAE,GAAG,CAAC,EAAE;YACb,MAAM,EAAE,GAAG,CAAC,MAAM;YAClB,MAAM,EAAE,gBAAgB;SAC3B,CAAC,CAAC;IAEP,CAAC;IAAC,OAAO,KAAU,EAAE,CAAC;QAClB,GAAG,CAAC,KAAK,CAAC,iCAAiC,eAAe,EAAE,EAAE,EAAE,KAAK,EAAE,CAAC,CAAC;QACzE,IAAI,KAAK,CAAC,UAAU,KAAK,GAAG,IAAI,MAAM,CAAC,2BAA2B,EAAE,CAAC;YACjE,MAAM,IAAI,sBAAsB,CAAC,KAAK,CAAC,UAAU,EAAE,0CAA0C,CAAC,CAAC;QACnG,CAAC;aAAM,IAAI,KAAK,CAAC,OAAO,CAAC,QAAQ,CAAC,yCAAyC,CAAC,EAAE,CAAC;YAC3E,wCAAwC;YACxC,MAAM,IAAI,yBAAyB,CAAC,aAAa,EAAE,OAAO,CAAC,QAAQ,EAAE,KAAK,CAAC,OAAO,CAAC,CAAC;QACxF,CAAC;aAAM,IAAI,KAAK,CAAC,OAAO,CAAC,QAAQ,CAAC,qCAAqC,CAAC,EAAE,CAAC;YACvE,wCAAwC;YACxC,MAAM,IAAI,yBAAyB,CAAC,OAAO,EAAE,OAAO,CAAC,QAAQ,EAAE,KAAK,CAAC,OAAO,CAAC,CAAC;QAClF,CAAC;aAAM,CAAC;YACJ,MAAM,IAAI,KAAK,CAAC,gCAAgC,eAAe,KAAK,KAAK,CAAC,OAAO,EAAE,CAAC,CAAC;QACzF,CAAC;IACL,CAAC;AACL,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,8BAA8B,CAChD,MAAsB,EACtB,eAAuB,EACvB,MAAkC,EAClC,WAAgB,EAChB,KAAe;IAEf,MAAM,QAAQ,GAAG,MAAM,CAAC,IAAI,CAAC;IAC7B,MAAM,IAAI,GAAG,YAAY,EAAE,CAAC;IAC5B,MAAM,KAAK,GAAG,IAAI,CAAC,iBAAiB,CAAC,KAAK,CAAC;IAC3C,IAAI,IAAI,GAAG,CAAC,UAAU,CAAC,CAAC;IACxB,IAAI,QAAQ,EAAE,CAAC;QACX,IAAI,GAAG,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC;IACjC,CAAC;IACD,MAAM,QAAQ,GAAyB;QACnC,MAAM,EAAE,IAAI,CAAC,iBAAiB,CAAC,KAAK;QACpC,WAAW,EAAE,IAAI,CAAC,iBAAiB,CAAC,UAAU;QAC9C,aAAa,EAAE,IAAI,CAAC,YAAY;KACnC,CAAC;IAEF,IAAI,0BAA0B,GAA6B,SAAS,CAAC;IACrE,IAAI,MAAM,CAAC,sBAAsB,EAAE,CAAC;QAChC,iCAAiC;QACjC,IAAI,IAAI,CAAC,OAAO,GAAG,CAAC,EAAE,CAAC;YACnB,GAAG,CAAC,IAAI,CAAC,sCAAsC,EAAE,EAAE,WAAW,EAAE,KAAK,EAAE,CAAC,CAAC;YACzE,MAAM,OAAO,GAAqB;gBAC9B,KAAK,EAAE,EAAE,gBAAgB,EAAE,CAAC,KAAK,CAAC,EAAE;gBACpC,KAAK,EAAE,CAAC;aACX,CAAC;YACF,MAAM,WAAW,GAAG,MAAM,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,EAAE,EAAE;gBAC/D,GAAG,CAAC,IAAI,CAAC,gBAAgB,EAAE,EAAE,OAAO,EAAE,GAAG,EAAE,CAAC,CAAC;gBAC7C,OAAO,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,SAAS,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;YACnD,CAAC,CAAC,CAAC;YAEH,IAAI,WAAW,EAAE,CAAC;gBACd,GAAG,CAAC,IAAI,CAAC,oBAAoB,EAAE,EAAE,WAAW,EAAE,CAAC,CAAC;gBAChD,0BAA0B,GAAG,MAAM,MAAM,CAAC,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,EAAE,CAAC,CAAC;YAC5E,CAAC;QACL,CAAC;IACL,CAAC;IACD,IAAI,KAAK,IAAI,0BAA0B,EAAE,KAAK,EAAE,CAAC;QAC7C,GAAG,CAAC,IAAI,CAAC,2BAA2B,EAAE,EAAE,KAAK,EAAE,0BAA0B,EAAE,KAAK,EAAE,CAAC,CAAC;IACxF,CAAC;IAED,MAAM,MAAM,GAAsC;QAC9C,WAAW,EAAE,MAAM,CAAC,WAAW;QAC/B,KAAK,EAAE,MAAM,CAAC,KAAK;QACnB,aAAa,EAAE,MAAM,CAAC,aAAa;QACnC,WAAW,EAAE,MAAM,CAAC,eAAe;KACtC,CAAC;IACF,MAAM,IAAI,GAAG;QACT,GAAG,WAAW;QACd,cAAc,EAAE,0BAA0B,EAAE,KAAK;KACpD,CAAC;IAEF,MAAM,aAAa,GAAG,MAAM,CAAC,aAAa,CAAC;IAE3C,GAAG,CAAC,KAAK,CAAC,gCAAgC,eAAe,EAAE,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,aAAa,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAC,CAAC;IAE9G,MAAM,GAAG,GAAG,MAAM,MAAM,CAAC,YAAY;SAChC,aAAa,CAAC,eAAe,EAAE;QAC5B,MAAM;QACN,IAAI;QACJ,aAAa;QACb,IAAI;QACJ,MAAM,EAAE,KAAK;QACb,QAAQ;KACX,CAAC;SACD,KAAK,CAAC,CAAC,GAAG,EAAE,EAAE;QACX,GAAG,CAAC,KAAK,CAAC,+BAA+B,eAAe,EAAE,EAAE,EAAE,GAAG,EAAE,CAAC,CAAC;QACrE,MAAM,GAAG,CAAC;IACd,CAAC,CAAC,CAAC;IAEP,IAAI,KAAK,EAAE,CAAC;QACR,GAAG,CAAC,IAAI,CAAC,wBAAwB,eAAe,EAAE,EAAE,GAAG,CAAC,CAAC;IAC7D,CAAC;IAED,IAAI,GAAG,CAAC,KAAK,IAAI,GAAG,CAAC,MAAM,KAAK,kBAAkB,CAAC,MAAM,EAAE,CAAC;QACxD,GAAG,CAAC,KAAK,CAAC,+BAA+B,eAAe,EAAE,EAAE,EAAE,KAAK,EAAE,GAAG,CAAC,KAAK,EAAE,CAAC,CAAC;QAClF,MAAM,IAAI,KAAK,CAAC,gCAAgC,eAAe,KAAK,GAAG,CAAC,KAAK,EAAE,CAAC,CAAC;IACrF,CAAC;IAED,OAAO,GAAG,CAAC;AACf,CAAC"}
@@ -0,0 +1,153 @@
1
+ import { log } from "@temporalio/activity";
2
+ import { markdownWithMarkitdown } from "../conversion/markitdown.js";
3
+ import { mutoolPdfToText } from "../conversion/mutool.js";
4
+ import { markdownWithPandoc } from "../conversion/pandoc.js";
5
+ import { setupActivity } from "../dsl/setup/ActivityContext.js";
6
+ import { DocumentNotFoundError } from "../errors.js";
7
+ import { TextExtractionStatus } from "../result-types.js";
8
+ import { fetchBlobAsBuffer, md5 } from "../utils/blobs.js";
9
+ import { countTokens } from "../utils/tokens.js";
10
+ //@ts-ignore
11
+ const JSON = {
12
+ name: "extractDocumentText",
13
+ };
14
+ export async function extractDocumentText(payload) {
15
+ const { client, objectId } = await setupActivity(payload);
16
+ const r = await client.objects.find({
17
+ query: { _id: objectId },
18
+ limit: 1,
19
+ select: "+text",
20
+ });
21
+ const doc = r[0];
22
+ if (!doc) {
23
+ log.error(`Document ${objectId} not found`);
24
+ throw new DocumentNotFoundError(`Document ${objectId} not found`, payload.objectIds);
25
+ }
26
+ log.info(`Extracting text for object ${doc.id}`);
27
+ if (!doc.content?.type || !doc.content?.source) {
28
+ if (doc.text) {
29
+ return createResponse(doc, doc.text, TextExtractionStatus.skipped, "Text present and no source or type");
30
+ }
31
+ else {
32
+ return createResponse(doc, "", TextExtractionStatus.error, "No source or type found");
33
+ }
34
+ }
35
+ //skip if text already extracted and proper etag
36
+ if (doc.text && doc.text.length > 0 && doc.text_etag === doc.content.etag) {
37
+ return createResponse(doc, doc.text, TextExtractionStatus.skipped, "Text already extracted");
38
+ }
39
+ let fileBuffer;
40
+ try {
41
+ fileBuffer = await fetchBlobAsBuffer(client, doc.content.source);
42
+ }
43
+ catch (e) {
44
+ log.error(`Error reading file: ${e}`);
45
+ return createResponse(doc, "", TextExtractionStatus.error, e.message);
46
+ }
47
+ let txt;
48
+ switch (doc.content.type) {
49
+ case "application/pdf":
50
+ txt = await mutoolPdfToText(fileBuffer);
51
+ break;
52
+ case "text/plain":
53
+ txt = fileBuffer.toString("utf8");
54
+ break;
55
+ //docx
56
+ case "application/vnd.openxmlformats-officedocument.wordprocessingml.document":
57
+ txt = await markdownWithMarkitdown(fileBuffer, "docx");
58
+ break;
59
+ //pptx
60
+ case "application/vnd.openxmlformats-officedocument.presentationml.presentation":
61
+ txt = await markdownWithMarkitdown(fileBuffer, "pptx");
62
+ break;
63
+ //html
64
+ case "text/html":
65
+ txt = await markdownWithPandoc(fileBuffer, "html");
66
+ break;
67
+ //opendocument
68
+ case "application/vnd.oasis.opendocument.text":
69
+ txt = await markdownWithPandoc(fileBuffer, "odt");
70
+ break;
71
+ //rtf
72
+ case "application/rtf":
73
+ txt = await markdownWithPandoc(fileBuffer, "rtf");
74
+ break;
75
+ //markdown
76
+ case "text/markdown":
77
+ txt = fileBuffer.toString("utf8");
78
+ break;
79
+ //csv
80
+ case "text/csv":
81
+ txt = fileBuffer.toString("utf8");
82
+ break;
83
+ //typescript
84
+ case "application/typescript":
85
+ txt = fileBuffer.toString("utf8");
86
+ break;
87
+ //javascript
88
+ case "application/javascript":
89
+ txt = fileBuffer.toString("utf8");
90
+ break;
91
+ //json
92
+ case "application/json":
93
+ txt = fileBuffer.toString("utf8");
94
+ break;
95
+ default:
96
+ if (sniffIfText(fileBuffer)) {
97
+ txt = fileBuffer.toString("utf8"); //TODO: add charset detection
98
+ break;
99
+ }
100
+ return createResponse(doc, doc.text ?? "", TextExtractionStatus.skipped, `Unsupported mime type: ${doc.content.type}`);
101
+ }
102
+ const tokensData = countTokens(txt);
103
+ const etag = doc.content.etag ?? md5(txt);
104
+ const updateData = {
105
+ text: txt,
106
+ text_etag: etag,
107
+ tokens: {
108
+ ...tokensData,
109
+ etag: etag,
110
+ },
111
+ };
112
+ await client.objects.update(doc.id, updateData);
113
+ return createResponse(doc, txt, TextExtractionStatus.success);
114
+ }
115
+ function createResponse(doc, text, status, message) {
116
+ return {
117
+ status,
118
+ message,
119
+ tokens: doc.tokens,
120
+ len: text.length,
121
+ objectId: doc.id,
122
+ hasText: !!text,
123
+ };
124
+ }
125
+ function sniffIfText(buf) {
126
+ // If file is too large, don't even try
127
+ if (buf.length > 500 * 1024) {
128
+ return false;
129
+ }
130
+ // Count binary/control characters
131
+ let binaryCount = 0;
132
+ const sampleSize = Math.min(buf.length, 1000); // Check first 1000 bytes
133
+ for (let i = 0; i < sampleSize; i++) {
134
+ // Count control characters (except common whitespace)
135
+ const byte = buf[i];
136
+ if ((byte < 32 && ![9, 10, 13].includes(byte)) || byte === 0) {
137
+ binaryCount++;
138
+ }
139
+ }
140
+ // If more than 10% binary/control chars, probably not text
141
+ if (binaryCount / sampleSize > 0.1) {
142
+ return false;
143
+ }
144
+ // Additional check for valid UTF-8 encoding
145
+ try {
146
+ const s = buf.toString("utf8");
147
+ return s.length > 0 && !s.includes("\uFFFD"); // Replacement character
148
+ }
149
+ catch (e) {
150
+ return false;
151
+ }
152
+ }
153
+ //# sourceMappingURL=extractDocumentText.js.map