@vertesia/workflow 1.3.0 → 1.4.0-dev.20260615.051508Z

package/lib/{esm/activities → activities}/extractDocumentText.js

@@ -1,16 +1,16 @@
-import { log } from "@temporalio/activity";
-import { markdownWithMarkitdown } from "../conversion/markitdown.js";
-import { mutoolPdfToText } from "../conversion/mutool.js";
-import { markdownWithPandoc } from "../conversion/pandoc.js";
-import { setupActivity } from "../dsl/setup/ActivityContext.js";
-import { DocumentNotFoundError } from "../errors.js";
-import { TextExtractionStatus } from "../result-types.js";
-import { fetchBlobAsBuffer, md5 } from "../utils/blobs.js";
-import { createFileSourceResult, uploadTextPreviewToStorage } from "../utils/text-preview-utils.js";
-import { countTokens } from "../utils/tokens.js";
-//@ts-ignore
-const JSON = {
-    name: "extractDocumentText",
+import { log } from '@temporalio/activity';
+import { markdownWithMarkitdown } from '../conversion/markitdown.js';
+import { mutoolPdfToText } from '../conversion/mutool.js';
+import { markdownWithPandoc } from '../conversion/pandoc.js';
+import { setupActivity } from '../dsl/setup/ActivityContext.js';
+import { DocumentNotFoundError } from '../errors.js';
+import { TextExtractionStatus } from '../result-types.js';
+import { fetchBlobAsBuffer, md5 } from '../utils/blobs.js';
+import { createFileSourceResult, uploadTextPreviewToStorage } from '../utils/text-preview-utils.js';
+import { countTokens } from '../utils/tokens.js';
+//@ts-expect-error
+const _JSON = {
+    name: 'extractDocumentText',
 };
 export async function extractDocumentText(payload) {
     const context = await setupActivity(payload);
@@ -29,12 +29,24 @@ export async function extractDocumentText(payload) {
     }
 }
 async function extractFromObject(client, objectId, objectIds) {
-    const r = await client.objects.find({
-        query: { _id: objectId },
-        limit: 1,
-        select: "+text",
-    });
-    const doc = r[0];
+    // Fetch the exact revision by id rather than `find({_id, revision.head:true})`.
+    // `find` filters to HEAD revisions, so if a newer revision is created while this
+    // intake is in flight (e.g. an update with `createRevision: true`), the original
+    // revision this workflow was queued for becomes non-head and `find` returns empty —
+    // failing intake with a spurious "not found" even though the revision still exists.
+    // A direct GET by id resolves the specific revision regardless of head.
+    let doc;
+    try {
+        doc = (await client.objects.retrieve(objectId, '+text'));
+    }
+    catch (err) {
+        const status = err && typeof err === 'object' && 'status' in err ? err.status : undefined;
+        if (status === 404) {
+            log.error(`Document ${objectId} not found`);
+            throw new DocumentNotFoundError(`Document ${objectId} not found`, objectIds);
+        }
+        throw err;
+    }
     if (!doc) {
         log.error(`Document ${objectId} not found`);
         throw new DocumentNotFoundError(`Document ${objectId} not found`, objectIds);
@@ -42,27 +54,28 @@ async function extractFromObject(client, objectId, objectIds) {
     log.info(`Extracting text for object ${doc.id}`);
     if (!doc.content?.type || !doc.content?.source) {
         if (doc.text) {
-            return createResponse(doc, doc.text, TextExtractionStatus.skipped, "Text present and no source or type");
+            return createResponse(doc, doc.text, TextExtractionStatus.skipped, 'Text present and no source or type');
         }
         else {
-            return createResponse(doc, "", TextExtractionStatus.error, "No source or type found");
+            return createResponse(doc, '', TextExtractionStatus.error, 'No source or type found');
         }
     }
     //skip if text already extracted and proper etag
     if (doc.text && doc.text.length > 0 && doc.text_etag === doc.content.etag) {
-        return createResponse(doc, doc.text, TextExtractionStatus.skipped, "Text already extracted");
+        return createResponse(doc, doc.text, TextExtractionStatus.skipped, 'Text already extracted');
     }
     let fileBuffer;
     try {
         fileBuffer = await fetchBlobAsBuffer(client, doc.content.source);
     }
     catch (e) {
+        const message = e instanceof Error ? e.message : String(e);
         log.error(`Error reading file: ${e}`);
-        return createResponse(doc, "", TextExtractionStatus.error, e.message);
+        return createResponse(doc, '', TextExtractionStatus.error, message);
     }
     const txt = await extractTextFromBuffer(fileBuffer, doc.content.type);
     if (!txt) {
-        return createResponse(doc, doc.text ?? "", TextExtractionStatus.skipped, `Unsupported mime type: ${doc.content.type}`);
+        return createResponse(doc, doc.text ?? '', TextExtractionStatus.skipped, `Unsupported mime type: ${doc.content.type}`);
     }
     const tokensData = countTokens(txt);
     const etag = doc.content.etag ?? md5(txt);
@@ -90,62 +103,62 @@ async function extractFromFileSource(client, input_file, output_storage_path) {
     const txt = await extractTextFromBuffer(fileBuffer, input_file.mimetype);
     // Upload extracted text to storage
     if (txt && output_storage_path) {
-        await uploadTextPreviewToStorage(client, txt, output_storage_path, "Document");
+        await uploadTextPreviewToStorage(client, txt, output_storage_path, 'Document');
     }
     return createFileSourceResult(input_file.url, output_storage_path, txt);
 }
 async function extractTextFromBuffer(fileBuffer, mimeType) {
     let txt;
     switch (mimeType) {
-        case "application/pdf":
+        case 'application/pdf':
             txt = await mutoolPdfToText(fileBuffer);
             break;
-        case "text/plain":
-            txt = fileBuffer.toString("utf8");
+        case 'text/plain':
+            txt = fileBuffer.toString('utf8');
             break;
         //docx
-        case "application/vnd.openxmlformats-officedocument.wordprocessingml.document":
-            txt = await markdownWithMarkitdown(fileBuffer, "docx");
+        case 'application/vnd.openxmlformats-officedocument.wordprocessingml.document':
+            txt = await markdownWithMarkitdown(fileBuffer, 'docx');
             break;
         //pptx
-        case "application/vnd.openxmlformats-officedocument.presentationml.presentation":
-            txt = await markdownWithMarkitdown(fileBuffer, "pptx");
+        case 'application/vnd.openxmlformats-officedocument.presentationml.presentation':
+            txt = await markdownWithMarkitdown(fileBuffer, 'pptx');
             break;
         //html
-        case "text/html":
-            txt = await markdownWithPandoc(fileBuffer, "html");
+        case 'text/html':
+            txt = await markdownWithPandoc(fileBuffer, 'html');
             break;
         //opendocument
-        case "application/vnd.oasis.opendocument.text":
-            txt = await markdownWithPandoc(fileBuffer, "odt");
+        case 'application/vnd.oasis.opendocument.text':
+            txt = await markdownWithPandoc(fileBuffer, 'odt');
             break;
         //rtf
-        case "application/rtf":
-            txt = await markdownWithPandoc(fileBuffer, "rtf");
+        case 'application/rtf':
+            txt = await markdownWithPandoc(fileBuffer, 'rtf');
             break;
         //markdown
-        case "text/markdown":
-            txt = fileBuffer.toString("utf8");
+        case 'text/markdown':
+            txt = fileBuffer.toString('utf8');
             break;
         //csv
-        case "text/csv":
-            txt = fileBuffer.toString("utf8");
+        case 'text/csv':
+            txt = fileBuffer.toString('utf8');
             break;
         //typescript
-        case "application/typescript":
-            txt = fileBuffer.toString("utf8");
+        case 'application/typescript':
+            txt = fileBuffer.toString('utf8');
             break;
         //javascript
-        case "application/javascript":
-            txt = fileBuffer.toString("utf8");
+        case 'application/javascript':
+            txt = fileBuffer.toString('utf8');
             break;
         //json
-        case "application/json":
-            txt = fileBuffer.toString("utf8");
+        case 'application/json':
+            txt = fileBuffer.toString('utf8');
             break;
         default:
             if (sniffIfText(fileBuffer)) {
-                txt = fileBuffer.toString("utf8"); //TODO: add charset detection
+                txt = fileBuffer.toString('utf8'); //TODO: add charset detection
                 break;
             }
             return null;
@@ -183,10 +196,10 @@ function sniffIfText(buf) {
     }
     // Additional check for valid UTF-8 encoding
     try {
-        const s = buf.toString("utf8");
-        return s.length > 0 && !s.includes("\uFFFD"); // Replacement character
+        const s = buf.toString('utf8');
+        return s.length > 0 && !s.includes('\uFFFD'); // Replacement character
     }
-    catch (e) {
+    catch {
         return false;
     }
 }

package/lib/activities/extractDocumentText.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"extractDocumentText.js","sourceRoot":"","sources":["../../src/activities/extractDocumentText.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAC;AAS3C,OAAO,EAAE,sBAAsB,EAAE,MAAM,6BAA6B,CAAC;AACrE,OAAO,EAAE,eAAe,EAAE,MAAM,yBAAyB,CAAC;AAC1D,OAAO,EAAE,kBAAkB,EAAE,MAAM,yBAAyB,CAAC;AAC7D,OAAO,EAAE,aAAa,EAAE,MAAM,iCAAiC,CAAC;AAChE,OAAO,EAAE,qBAAqB,EAAE,MAAM,cAAc,CAAC;AACrD,OAAO,EAA6B,oBAAoB,EAAE,MAAM,oBAAoB,CAAC;AACrF,OAAO,EAAE,iBAAiB,EAAE,GAAG,EAAE,MAAM,mBAAmB,CAAC;AAC3D,OAAO,EAAE,sBAAsB,EAAE,0BAA0B,EAAE,MAAM,gCAAgC,CAAC;AACpG,OAAO,EAAE,WAAW,EAAE,MAAM,oBAAoB,CAAC;AAEjD,kBAAkB;AAClB,MAAM,KAAK,GAAoB;IAC3B,IAAI,EAAE,qBAAqB;CAC9B,CAAC;AAUF,MAAM,CAAC,KAAK,UAAU,mBAAmB,CACrC,OAA+D;IAE/D,MAAM,OAAO,GAAG,MAAM,aAAa,CAAC,OAAO,CAAC,CAAC;IAC7C,MAAM,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,EAAE,GAAG,OAAO,CAAC;IAC9C,MAAM,EAAE,mBAAmB,EAAE,GAAG,MAAM,CAAC;IAEvC,IAAI,SAAS,KAAK,OAAO,EAAE,CAAC;QACxB,sCAAsC;QACtC,IAAI,CAAC,mBAAmB,EAAE,CAAC;YACvB,MAAM,IAAI,KAAK,CAAC,wEAAwE,CAAC,CAAC;QAC9F,CAAC;QACD,OAAO,qBAAqB,CAAC,MAAM,EAAE,OAAO,CAAC,IAAI,EAAE,mBAAmB,CAAC,CAAC;IAC5E,CAAC;SAAM,CAAC;QACJ,uCAAuC;QACvC,OAAO,iBAAiB,CAAC,MAAM,EAAE,OAAO,CAAC,QAAQ,EAAE,OAAO,CAAC,SAAS,IAAI,EAAE,CAAC,CAAC;IAChF,CAAC;AACL,CAAC;AAED,KAAK,UAAU,iBAAiB,CAC5B,MAAsB,EACtB,QAAgB,EAChB,SAAmB;IAEnB,gFAAgF;IAChF,iFAAiF;IACjF,iFAAiF;IACjF,oFAAoF;IACpF,oFAAoF;IACpF,wEAAwE;IACxE,IAAI,GAAkB,CAAC;IACvB,IAAI,CAAC;QACD,GAAG,GAAG,CAAC,MAAM,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAkB,CAAC;IAC9E,CAAC;IAAC,OAAO,GAAY,EAAE,CAAC;QACpB,MAAM,MAAM,GACR,GAAG,IAAI,OAAO,GAAG,KAAK,QAAQ,IAAI,QAAQ,IAAI,GAAG,CAAC,CAAC,CAAE,GAA2B,CAAC,MAAM,CAAC,CAAC,CAAC,SAAS,CAAC;QACxG,IAAI,MAAM,KAAK,GAAG,EAAE,CAAC;YACjB,GAAG,CAAC,KAAK,CAAC,YAAY,QAAQ,YAAY,CAAC,CAAC;YAC5C,MAAM,IAAI,qBAAqB,CAAC,YAAY,QAAQ,YAAY,EAAE,SAAS,CAAC,CAAC;QACjF,CAAC;QACD,MAAM,GAAG,CAAC;IACd,CAAC;IACD,IAAI,CAAC,GAAG,EAAE,CAAC;QACP,GAAG,CAAC,KAAK,CAAC,YAAY,QAAQ,YAAY,CAAC,CAAC;QAC5C,MAAM,IAAI,qBAAqB,CAAC,YAAY,QAAQ,YAAY,EAAE,SAAS,CAAC,CAAC;IACjF,CAAC;IAED,GAAG,CAAC,IAAI,CAAC,8BAA8B,GAAG,CAAC,EAAE,EAAE,CAAC,CAAC;IAEjD,IAAI,CAAC,GAAG,CAAC,OAAO,EAAE,IAAI,IAAI,CAAC,GAAG,CAAC,OAAO,EAAE,MAAM,EAAE,CAAC;QAC7C,IAAI,GAAG,CAAC,IAAI,EAAE,CAAC;YACX,OAAO,cAAc,CAAC,GAAG,EAAE,GAAG,CAAC,IAAI,EAAE,oBAAoB,CAAC,OAAO,EAAE,oCAAoC,CAAC,CAAC;QAC7G,CAAC;aAAM,CAAC;YACJ,OAAO,cAAc,CAAC,GAAG,EAAE,EAAE,EAAE,oBAAoB,CAAC,KAAK,EAAE,yBAAyB,CAAC,CAAC;QAC1F,CAAC;IACL,CAAC;IAED,gDAAgD;IAChD,IAAI,GAAG,CAAC,IAAI,IAAI,GAAG,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,IAAI,GAAG,CAAC,SAAS,KAAK,GAAG,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC;QACxE,OAAO,cAAc,CAAC,GAAG,EAAE,GAAG,CAAC,IAAI,EAAE,oBAAoB,CAAC,OAAO,EAAE,wBAAwB,CAAC,CAAC;IACjG,CAAC;IAED,IAAI,UAAkB,CAAC;IACvB,IAAI,CAAC;QACD,UAAU,GAAG,MAAM,iBAAiB,CAAC,MAAM,EAAE,GAAG,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;IACrE,CAAC;IAAC,OAAO,CAAU,EAAE,CAAC;QAClB,MAAM,OAAO,GAAG,CAAC,YAAY,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC;QAC3D,GAAG,CAAC,KAAK,CAAC,uBAAuB,CAAC,EAAE,CAAC,CAAC;QACtC,OAAO,cAAc,CAAC,GAAG,EAAE,EAAE,EAAE,oBAAoB,CAAC,KAAK,EAAE,OAAO,CAAC,CAAC;IACxE,CAAC;IAED,MAAM,GAAG,GAAG,MAAM,qBAAqB,CAAC,UAAU,EAAE,GAAG,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC;IACtE,IAAI,CAAC,GAAG,EAAE,CAAC;QACP,OAAO,cAAc,CACjB,GAAG,EACH,GAAG,CAAC,IAAI,IAAI,EAAE,EACd,oBAAoB,CAAC,OAAO,EAC5B,0BAA0B,GAAG,CAAC,OAAO,CAAC,IAAI,EAAE,CAC/C,CAAC;IACN,CAAC;IAED,MAAM,UAAU,GAAG,WAAW,CAAC,GAAG,CAAC,CAAC;IACpC,MAAM,IAAI,GAAG,GAAG,CAAC,OAAO,CAAC,IAAI,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC;IAE1C,MAAM,UAAU,GAA+B;QAC3C,IAAI,EAAE,GAAG;QACT,SAAS,EAAE,IAAI;QACf,MAAM,EAAE;YACJ,GAAG,UAAU;YACb,IAAI,EAAE,IAAI;SACb;KACJ,CAAC;IAEF,MAAM,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,GAAG,CAAC,EAAE,EAAE,UAAU,CAAC,CAAC;IAEhD,OAAO,cAAc,CAAC,GAAG,EAAE,GAAG,EAAE,oBAAoB,CAAC,OAAO,CAAC,CAAC;AAClE,CAAC;AAED,KAAK,UAAU,qBAAqB,CAChC,MAAsB,EACtB,UAA6B,EAC7B,mBAA2B;IAE3B,GAAG,CAAC,IAAI,CAAC,wBAAwB,UAAU,EAAE,CAAC,CAAC;IAE/C,IAAI,UAAkB,CAAC;IACvB,IAAI,CAAC;QACD,UAAU,GAAG,MAAM,iBAAiB,CAAC,MAAM,EAAE,UAAU,CAAC,GAAG,CAAC,CAAC;IACjE,CAAC;IAAC,OAAO,CAAU,EAAE,CAAC;QAClB,GAAG,CAAC,KAAK,CAAC,uBAAuB,CAAC,EAAE,CAAC,CAAC;QACtC,OAAO,sBAAsB,CAAC,UAAU,CAAC,GAAG,EAAE,mBAAmB,EAAE,IAAI,CAAC,CAAC;IAC7E,CAAC;IAED,MAAM,GAAG,GAAG,MAAM,qBAAqB,CAAC,UAAU,EAAE,UAAU,CAAC,QAAQ,CAAC,CAAC;IAEzE,mCAAmC;IACnC,IAAI,GAAG,IAAI,mBAAmB,EAAE,CAAC;QAC7B,MAAM,0BAA0B,CAAC,MAAM,EAAE,GAAG,EAAE,mBAAmB,EAAE,UAAU,CAAC,CAAC;IACnF,CAAC;IAED,OAAO,sBAAsB,CAAC,UAAU,CAAC,GAAG,EAAE,mBAAmB,EAAE,GAAG,CAAC,CAAC;AAC5E,CAAC;AAED,KAAK,UAAU,qBAAqB,CAAC,UAAkB,EAAE,QAAgB;IACrE,IAAI,GAAW,CAAC;IAEhB,QAAQ,QAAQ,EAAE,CAAC;QACf,KAAK,iBAAiB;YAClB,GAAG,GAAG,MAAM,eAAe,CAAC,UAAU,CAAC,CAAC;YACxC,MAAM;QAEV,KAAK,YAAY;YACb,GAAG,GAAG,UAAU,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YAClC,MAAM;QAEV,MAAM;QACN,KAAK,yEAAyE;YAC1E,GAAG,GAAG,MAAM,sBAAsB,CAAC,UAAU,EAAE,MAAM,CAAC,CAAC;YACvD,MAAM;QAEV,MAAM;QACN,KAAK,2EAA2E;YAC5E,GAAG,GAAG,MAAM,sBAAsB,CAAC,UAAU,EAAE,MAAM,CAAC,CAAC;YACvD,MAAM;QAEV,MAAM;QACN,KAAK,WAAW;YACZ,GAAG,GAAG,MAAM,kBAAkB,CAAC,UAAU,EAAE,MAAM,CAAC,CAAC;YACnD,MAAM;QAEV,cAAc;QACd,KAAK,yCAAyC;YAC1C,GAAG,GAAG,MAAM,kBAAkB,CAAC,UAAU,EAAE,KAAK,CAAC,CAAC;YAClD,MAAM;QAEV,KAAK;QACL,KAAK,iBAAiB;YAClB,GAAG,GAAG,MAAM,kBAAkB,CAAC,UAAU,EAAE,KAAK,CAAC,CAAC;YAClD,MAAM;QAEV,UAAU;QACV,KAAK,eAAe;YAChB,GAAG,GAAG,UAAU,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YAClC,MAAM;QAEV,KAAK;QACL,KAAK,UAAU;YACX,GAAG,GAAG,UAAU,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YAClC,MAAM;QAEV,YAAY;QACZ,KAAK,wBAAwB;YACzB,GAAG,GAAG,UAAU,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YAClC,MAAM;QAEV,YAAY;QACZ,KAAK,wBAAwB;YACzB,GAAG,GAAG,UAAU,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YAClC,MAAM;QAEV,MAAM;QACN,KAAK,kBAAkB;YACnB,GAAG,GAAG,UAAU,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YAClC,MAAM;QAEV;YACI,IAAI,WAAW,CAAC,UAAU,CAAC,EAAE,CAAC;gBAC1B,GAAG,GAAG,UAAU,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,6BAA6B;gBAChE,MAAM;YACV,CAAC;YACD,OAAO,IAAI,CAAC;IACpB,CAAC;IAED,OAAO,GAAG,CAAC;AACf,CAAC;AAED,SAAS,cAAc,CACnB,GAAkB,EAClB,IAAY,EACZ,MAA4B,EAC5B,OAAgB;IAEhB,OAAO;QACH,MAAM;QACN,OAAO;QACP,MAAM,EAAE,GAAG,CAAC,MAAM;QAClB,GAAG,EAAE,IAAI,CAAC,MAAM;QAChB,QAAQ,EAAE,GAAG,CAAC,EAAE;QAChB,OAAO,EAAE,CAAC,CAAC,IAAI;KAClB,CAAC;AACN,CAAC;AAED,SAAS,WAAW,CAAC,GAAW;IAC5B,uCAAuC;IACvC,IAAI,GAAG,CAAC,MAAM,GAAG,GAAG,GAAG,IAAI,EAAE,CAAC;QAC1B,OAAO,KAAK,CAAC;IACjB,CAAC;IAED,kCAAkC;IAClC,IAAI,WAAW,GAAG,CAAC,CAAC;IACpB,MAAM,UAAU,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC,CAAC,yBAAyB;IAExE,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,UAAU,EAAE,CAAC,EAAE,EAAE,CAAC;QAClC,sDAAsD;QACtD,MAAM,IAAI,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC;QACpB,IAAI,CAAC,IAAI,GAAG,EAAE,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,IAAI,IAAI,KAAK,CAAC,EAAE,CAAC;YAC3D,WAAW,EAAE,CAAC;QAClB,CAAC;IACL,CAAC;IAED,2DAA2D;IAC3D,IAAI,WAAW,GAAG,UAAU,GAAG,GAAG,EAAE,CAAC;QACjC,OAAO,KAAK,CAAC;IACjB,CAAC;IAED,4CAA4C;IAC5C,IAAI,CAAC;QACD,MAAM,CAAC,GAAG,GAAG,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;QAC/B,OAAO,CAAC,CAAC,MAAM,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC,CAAC,wBAAwB;IAC1E,CAAC;IAAC,MAAM,CAAC;QACL,OAAO,KAAK,CAAC;IACjB,CAAC;AACL,CAAC"}

package/lib/{types/activities → activities}/generateDocumentProperties.d.ts

@@ -1,6 +1,6 @@
-import { DSLActivityExecutionPayload, DSLActivitySpec } from "@vertesia/common";
-import { TruncateSpec } from "../utils/tokens.js";
-import { InteractionExecutionParams } from "./executeInteraction.js";
+import type { DSLActivityExecutionPayload, DSLActivitySpec } from '@vertesia/common';
+import { type TruncateSpec } from '../utils/tokens.js';
+import { type InteractionExecutionParams } from './executeInteraction.js';
 export interface GenerateDocumentPropertiesParams extends InteractionExecutionParams {
     typesHint?: string[];
     /**
@@ -11,7 +11,7 @@ export interface GenerateDocumentPropertiesParams extends InteractionExecutionPa
     use_vision?: boolean;
 }
 export interface GenerateDocumentProperties extends DSLActivitySpec<GenerateDocumentPropertiesParams> {
-    name: "generateDocumentProperties";
+    name: 'generateDocumentProperties';
 }
 export declare function generateDocumentProperties(payload: DSLActivityExecutionPayload<GenerateDocumentPropertiesParams>): Promise<{
     status: string;

package/lib/activities/generateDocumentProperties.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"generateDocumentProperties.d.ts","sourceRoot":"","sources":["../../src/activities/generateDocumentProperties.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,2BAA2B,EAAE,eAAe,EAAc,MAAM,kBAAkB,CAAC;AAEjG,OAAO,EAAE,KAAK,YAAY,EAAoB,MAAM,oBAAoB,CAAC;AACzE,OAAO,EAAkC,KAAK,0BAA0B,EAAE,MAAM,yBAAyB,CAAC;AAoB1G,MAAM,WAAW,gCAAiC,SAAQ,0BAA0B;IAChF,SAAS,CAAC,EAAE,MAAM,EAAE,CAAC;IACrB;;OAEG;IACH,QAAQ,CAAC,EAAE,YAAY,CAAC;IAExB,eAAe,CAAC,EAAE,MAAM,CAAC;IAEzB,UAAU,CAAC,EAAE,OAAO,CAAC;CACxB;AACD,MAAM,WAAW,0BAA2B,SAAQ,eAAe,CAAC,gCAAgC,CAAC;IACjG,IAAI,EAAE,4BAA4B,CAAC;CACtC;AAED,wBAAsB,0BAA0B,CAC5C,OAAO,EAAE,2BAA2B,CAAC,gCAAgC,CAAC;;;;;;;;;;;;;;;GA6HzE"}

package/lib/{esm/activities → activities}/generateDocumentProperties.js

@@ -1,38 +1,40 @@
-import { ApplicationFailure, log } from "@temporalio/activity";
-import { setupActivity } from "../dsl/setup/ActivityContext.js";
-import { truncByMaxTokens } from "../utils/tokens.js";
-import { executeInteractionFromActivity } from "./executeInteraction.js";
-const INT_EXTRACT_INFORMATION = "sys:ExtractInformation";
+import { ApplicationFailure, log } from '@temporalio/activity';
+import { setupActivity } from '../dsl/setup/ActivityContext.js';
+import { truncByMaxTokens } from '../utils/tokens.js';
+import { executeInteractionFromActivity } from './executeInteraction.js';
+const INT_EXTRACT_INFORMATION = 'sys:ExtractInformation';
+function toRetryableError(error) {
+    if (error instanceof Error) {
+        return error;
+    }
+    return new Error(String(error));
+}
 export async function generateDocumentProperties(payload) {
     const context = await setupActivity(payload);
     const { params, client, objectId } = context;
     const interactionName = params.interactionName ?? INT_EXTRACT_INFORMATION;
     const project = await context.fetchProject();
-    const doc = await client.objects.retrieve(objectId, "+text");
-    const type = doc.type
-        ? await client.types.catalog.resolve(doc.type)
-        : undefined;
-    if (!doc?.text && !params.use_vision && !doc?.content?.type?.startsWith("image/")) {
+    const doc = await client.objects.retrieve(objectId, '+text');
+    const type = doc.type ? await client.types.catalog.resolve(doc.type) : undefined;
+    if (!doc?.text && !params.use_vision && !doc?.content?.type?.startsWith('image/')) {
         log.warn(`Object ${objectId} not found or text is empty`);
-        return { status: "failed", error: "no-text" };
+        return { status: 'failed', error: 'no-text' };
     }
-    if (!type || !type.object_schema) {
+    if (!type?.object_schema) {
         log.info(`Object ${objectId} has no schema`);
-        return { document: objectId, status: "skipped", message: "no schema defined on type" };
+        return { document: objectId, status: 'skipped', message: 'no schema defined on type' };
     }
     const getImageRef = () => {
-        if (doc.content?.type?.startsWith("image/")) {
-            return "store:" + doc.id;
+        if (doc.content?.type?.startsWith('image/')) {
+            return `store:${doc.id}`;
         }
-        if (params.use_vision && doc.content?.type?.startsWith("application/pdf")) {
-            return "store:" + doc.id;
+        if (params.use_vision && doc.content?.type?.startsWith('application/pdf')) {
+            return `store:${doc.id}`;
         }
         log.info(`Object ${objectId} is not an image or pdf`);
         return undefined;
     };
-    const content = doc.text
-        ? truncByMaxTokens(doc.text, params.truncate || 30000)
-        : undefined;
+    const content = doc.text ? truncByMaxTokens(doc.text, params.truncate || 30000) : undefined;
     const promptData = {
         content: content,
         image: getImageRef() ?? undefined,
@@ -49,20 +51,22 @@ export async function generateDocumentProperties(payload) {
         }, promptData, payload.debug_mode ?? false);
     }
     catch (error) {
-        log.error(`Failed to extract document properties for ${objectId}`, { error, retryable: error.retryable });
-        const isRetryable = error.retryable !== undefined
-            ? error.retryable !== false
-            : undefined;
+        const extractionError = toRetryableError(error);
+        log.error(`Failed to extract document properties for ${objectId}`, {
+            error: extractionError,
+            retryable: extractionError.retryable,
+        });
+        const isRetryable = extractionError.retryable !== undefined ? extractionError.retryable !== false : undefined;
         if (isRetryable !== undefined) {
             if (isRetryable) {
                 throw ApplicationFailure.create({
-                    message: `Document property extraction failed for ${objectId}: ${error.message}`,
+                    message: `Document property extraction failed for ${objectId}: ${extractionError.message}`,
                     nonRetryable: false,
                 });
             }
             else {
                 throw ApplicationFailure.create({
-                    message: `Non-retryable document property extraction failed for ${objectId}: ${error.message}`,
+                    message: `Non-retryable document property extraction failed for ${objectId}: ${extractionError.message}`,
                     nonRetryable: true,
                 });
             }
@@ -73,12 +77,12 @@ export async function generateDocumentProperties(payload) {
         if (doc.text) {
             return undefined;
         }
-        let text = "";
+        let text = '';
         const jsonResult = infoRes.result.object();
-        if (jsonResult.title) {
-            text += jsonResult.title + "\n";
+        if (typeof jsonResult.title === 'string') {
+            text += `${jsonResult.title}\n`;
         }
-        if (jsonResult.description) {
+        if (typeof jsonResult.description === 'string') {
             text += jsonResult.description;
         }
         if (text) {
@@ -92,15 +96,15 @@ export async function generateDocumentProperties(payload) {
     await client.objects.update(doc.id, {
         properties: {
             ...infoRes.result.object(),
-            etag: doc.text_etag,
+            ...(doc.text_etag !== undefined ? { etag: doc.text_etag } : {}),
         },
         text: getText(),
         generation_run_info: {
             id: infoRes.id,
             date: new Date().toISOString(),
-            model: infoRes.modelId ?? "",
+            model: infoRes.modelId ?? '',
         },
     }, { suppressWorkflows: true });
-    return { status: "completed" };
+    return { status: 'completed' };
 }
 //# sourceMappingURL=generateDocumentProperties.js.map

package/lib/activities/generateDocumentProperties.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"generateDocumentProperties.js","sourceRoot":"","sources":["../../src/activities/generateDocumentProperties.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,kBAAkB,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAC;AAE/D,OAAO,EAAE,aAAa,EAAE,MAAM,iCAAiC,CAAC;AAChE,OAAO,EAAqB,gBAAgB,EAAE,MAAM,oBAAoB,CAAC;AACzE,OAAO,EAAE,8BAA8B,EAAmC,MAAM,yBAAyB,CAAC;AAE1G,MAAM,uBAAuB,GAAG,wBAAwB,CAAC;AAWzD,SAAS,gBAAgB,CAAC,KAAc;IACpC,IAAI,KAAK,YAAY,KAAK,EAAE,CAAC;QACzB,OAAO,KAAuB,CAAC;IACnC,CAAC;IACD,OAAO,IAAI,KAAK,CAAC,MAAM,CAAC,KAAK,CAAC,CAAmB,CAAC;AACtD,CAAC;AAiBD,MAAM,CAAC,KAAK,UAAU,0BAA0B,CAC5C,OAAsE;IAEtE,MAAM,OAAO,GAAG,MAAM,aAAa,CAAmC,OAAO,CAAC,CAAC;IAC/E,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,GAAG,OAAO,CAAC;IAC7C,MAAM,eAAe,GAAG,MAAM,CAAC,eAAe,IAAI,uBAAuB,CAAC;IAE1E,MAAM,OAAO,GAAG,MAAM,OAAO,CAAC,YAAY,EAAE,CAAC;IAE7C,MAAM,GAAG,GAAG,MAAM,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC;IAC7D,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,MAAM,MAAM,CAAC,KAAK,CAAC,OAAO,CAAC,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;IAEjF,IAAI,CAAC,GAAG,EAAE,IAAI,IAAI,CAAC,MAAM,CAAC,UAAU,IAAI,CAAC,GAAG,EAAE,OAAO,EAAE,IAAI,EAAE,UAAU,CAAC,QAAQ,CAAC,EAAE,CAAC;QAChF,GAAG,CAAC,IAAI,CAAC,UAAU,QAAQ,6BAA6B,CAAC,CAAC;QAC1D,OAAO,EAAE,MAAM,EAAE,QAAQ,EAAE,KAAK,EAAE,SAAS,EAAE,CAAC;IAClD,CAAC;IAED,IAAI,CAAC,IAAI,EAAE,aAAa,EAAE,CAAC;QACvB,GAAG,CAAC,IAAI,CAAC,UAAU,QAAQ,gBAAgB,CAAC,CAAC;QAC7C,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,2BAA2B,EAAE,CAAC;IAC3F,CAAC;IAED,MAAM,WAAW,GAAG,GAAG,EAAE;QACrB,IAAI,GAAG,CAAC,OAAO,EAAE,IAAI,EAAE,UAAU,CAAC,QAAQ,CAAC,EAAE,CAAC;YAC1C,OAAO,SAAS,GAAG,CAAC,EAAE,EAAE,CAAC;QAC7B,CAAC;QAED,IAAI,MAAM,CAAC,UAAU,IAAI,GAAG,CAAC,OAAO,EAAE,IAAI,EAAE,UAAU,CAAC,iBAAiB,CAAC,EAAE,CAAC;YACxE,OAAO,SAAS,GAAG,CAAC,EAAE,EAAE,CAAC;QAC7B,CAAC;QAED,GAAG,CAAC,IAAI,CAAC,UAAU,QAAQ,yBAAyB,CAAC,CAAC;QACtD,OAAO,SAAS,CAAC;IACrB,CAAC,CAAC;IAEF,MAAM,OAAO,GAAG,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,gBAAgB,CAAC,GAAG,CAAC,IAAI,EAAE,MAAM,CAAC,QAAQ,IAAI,KAAK,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;IAE5F,MAAM,UAAU,GAAG;QACf,OAAO,EAAE,OAAO;QAChB,KAAK,EAAE,WAAW,EAAE,IAAI,SAAS;QACjC,aAAa,EAAE,OAAO,EAAE,aAAa,EAAE,aAAa,IAAI,SAAS;KACpE,CAAC;IAEF,GAAG,CAAC,IAAI,CACJ,uCAAuC,QAAQ,cAAc,IAAI,CAAC,IAAI,EAAE,EACxE,OAAO,CAAC,UAAU,CAAC,CAAC,CAAC,EAAE,MAAM,EAAE,CAAC,CAAC,CAAC,SAAS,CAC9C,CAAC;IAEF,IAAI,OAAmE,CAAC;IACxE,IAAI,CAAC;QACD,OAAO,GAAG,MAAM,8BAA8B,CAC1C,MAAM,EACN,eAAe,EACf;YACI,GAAG,MAAM;YACT,sBAAsB,EAAE,IAAI;YAC5B,aAAa,EAAE,IAAI,CAAC,aAAa;YACjC,eAAe,EAAE,IAAI,CAAC,WAAW;SACpC,EACD,UAAU,EACV,OAAO,CAAC,UAAU,IAAI,KAAK,CAC9B,CAAC;IACN,CAAC;IAAC,OAAO,KAAc,EAAE,CAAC;QACtB,MAAM,eAAe,GAAG,gBAAgB,CAAC,KAAK,CAAC,CAAC;QAChD,GAAG,CAAC,KAAK,CAAC,6CAA6C,QAAQ,EAAE,EAAE;YAC/D,KAAK,EAAE,eAAe;YACtB,SAAS,EAAE,eAAe,CAAC,SAAS;SACvC,CAAC,CAAC;QAEH,MAAM,WAAW,GAAG,eAAe,CAAC,SAAS,KAAK,SAAS,CAAC,CAAC,CAAC,eAAe,CAAC,SAAS,KAAK,KAAK,CAAC,CAAC,CAAC,SAAS,CAAC;QAE9G,IAAI,WAAW,KAAK,SAAS,EAAE,CAAC;YAC5B,IAAI,WAAW,EAAE,CAAC;gBACd,MAAM,kBAAkB,CAAC,MAAM,CAAC;oBAC5B,OAAO,EAAE,2CAA2C,QAAQ,KAAK,eAAe,CAAC,OAAO,EAAE;oBAC1F,YAAY,EAAE,KAAK;iBACtB,CAAC,CAAC;YACP,CAAC;iBAAM,CAAC;gBACJ,MAAM,kBAAkB,CAAC,MAAM,CAAC;oBAC5B,OAAO,EAAE,yDAAyD,QAAQ,KAAK,eAAe,CAAC,OAAO,EAAE;oBACxG,YAAY,EAAE,IAAI;iBACrB,CAAC,CAAC;YACP,CAAC;QACL,CAAC;QAED,MAAM,KAAK,CAAC;IAChB,CAAC;IAED,MAAM,OAAO,GAAG,GAAG,EAAE;QACjB,IAAI,GAAG,CAAC,IAAI,EAAE,CAAC;YACX,OAAO,SAAS,CAAC;QACrB,CAAC;QACD,IAAI,IAAI,GAAG,EAAE,CAAC;QACd,MAAM,UAAU,GAAG,OAAO,CAAC,MAAM,CAAC,MAAM,EAA8B,CAAC;QACvE,IAAI,OAAO,UAAU,CAAC,KAAK,KAAK,QAAQ,EAAE,CAAC;YACvC,IAAI,IAAI,GAAG,UAAU,CAAC,KAAK,IAAI,CAAC;QACpC,CAAC;QACD,IAAI,OAAO,UAAU,CAAC,WAAW,KAAK,QAAQ,EAAE,CAAC;YAC7C,IAAI,IAAI,UAAU,CAAC,WAAW,CAAC;QACnC,CAAC;QACD,IAAI,IAAI,EAAE,CAAC;YACP,OAAO,IAAI,CAAC;QAChB,CAAC;aAAM,CAAC;YACJ,OAAO,SAAS,CAAC;QACrB,CAAC;IACL,CAAC,CAAC;IAEF,GAAG,CAAC,IAAI,CAAC,qCAAqC,QAAQ,cAAc,IAAI,CAAC,IAAI,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,CAAC,EAAE,EAAE,CAAC,CAAC;IACxG,MAAM,MAAM,CAAC,OAAO,CAAC,MAAM,CACvB,GAAG,CAAC,EAAE,EACN;QACI,UAAU,EAAE;YACR,GAAG,OAAO,CAAC,MAAM,CAAC,MAAM,EAAc;YACtC,GAAG,CAAC,GAAG,CAAC,SAAS,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,GAAG,CAAC,SAAS,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;SAClE;QACD,IAAI,EAAE,OAAO,EAAE;QACf,mBAAmB,EAAE;YACjB,EAAE,EAAE,OAAO,CAAC,EAAE;YACd,IAAI,EAAE,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE;YAC9B,KAAK,EAAE,OAAO,CAAC,OAAO,IAAI,EAAE;SAC/B;KACJ,EACD,EAAE,iBAAiB,EAAE,IAAI,EAAE,CAC9B,CAAC;IAEF,OAAO,EAAE,MAAM,EAAE,WAAW,EAAE,CAAC;AACnC,CAAC"}

package/lib/{types/activities → activities}/generateEmbeddings.d.ts

@@ -1,5 +1,5 @@
-import { DSLActivityExecutionPayload, DSLActivitySpec, SupportedEmbeddingTypes } from "@vertesia/common";
-import { DocPart } from "../utils/chunks.js";
+import { type DSLActivityExecutionPayload, type DSLActivitySpec, SupportedEmbeddingTypes } from '@vertesia/common';
+import type { DocPart } from '../utils/chunks.js';
 export interface GenerateEmbeddingsParams {
     /**
      * The model to use for embedding generation
@@ -25,7 +25,7 @@ export interface GenerateEmbeddingsParams {
     parts?: DocPart[];
 }
 export interface GenerateEmbeddings extends DSLActivitySpec<GenerateEmbeddingsParams> {
-    name: "generateEmbeddings";
+    name: 'generateEmbeddings';
 }
 export declare function generateEmbeddings(payload: DSLActivityExecutionPayload<GenerateEmbeddingsParams>): Promise<{
     status: string;

package/lib/activities/generateEmbeddings.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"generateEmbeddings.d.ts","sourceRoot":"","sources":["../../src/activities/generateEmbeddings.ts"],"names":[],"mappings":"AAEA,OAAO,EAEH,KAAK,2BAA2B,EAChC,KAAK,eAAe,EAIpB,uBAAuB,EAC1B,MAAM,kBAAkB,CAAC;AAI1B,OAAO,KAAK,EAAE,OAAO,EAAE,MAAM,oBAAoB,CAAC;AAGlD,MAAM,WAAW,wBAAwB;IACrC;;;OAGG;IACH,KAAK,CAAC,EAAE,MAAM,CAAC;IAEf;;;OAGG;IACH,WAAW,CAAC,EAAE,MAAM,CAAC;IAErB;;OAEG;IACH,KAAK,CAAC,EAAE,OAAO,CAAC;IAEhB;;OAEG;IACH,IAAI,EAAE,uBAAuB,CAAC;IAE9B;;OAEG;IACH,KAAK,CAAC,EAAE,OAAO,EAAE,CAAC;CACrB;AAED,MAAM,WAAW,kBAAmB,SAAQ,eAAe,CAAC,wBAAwB,CAAC;IACjF,IAAI,EAAE,oBAAoB,CAAC;CAC9B;AAED,wBAAsB,kBAAkB,CAAC,OAAO,EAAE,2BAA2B,CAAC,wBAAwB,CAAC;;;;;;;;;;;;;;;;;;;;;;;;GAgGtG"}