@vertesia/workflow 0.81.1 → 1.0.0-dev.20260203.130115Z

package/bin/bundle-workflows.mjs

@@ -7,6 +7,19 @@ import path from 'path';
 async function bundle(wsPath, bundlePath) {
     const { code } = await bundleWorkflowCode({
         workflowsPath: path.resolve(wsPath),
+        webpackConfigHook: (config) => {
+            // Fix for Temporal's VM sandbox environment:
+            // 1. publicPath: '' - disable auto-detection (no document.currentScript)
+            // 2. chunkLoading: 'import' - use import() instead of JSONP (no 'self' global)
+            // 3. globalObject: 'globalThis' - use globalThis instead of self/window
+            config.output = {
+                ...config.output,
+                publicPath: '',
+                chunkLoading: 'import',
+                globalObject: 'globalThis',
+            };
+            return config;
+        },
     });
     const codePath = path.resolve(bundlePath);
     await writeFile(codePath, code);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@vertesia/workflow",
-  "version": "0.81.1",
+  "version": "1.0.0-dev.20260203.130115Z",
   "type": "module",
   "description": "Vertesia workflow DSL",
   "main": "./lib/esm/index.js",
@@ -17,38 +17,38 @@
     "@temporalio/proto": "^1.11.5",
     "@temporalio/testing": "^1.11.5",
     "@temporalio/worker": "^1.11.5",
-    "@types/jsonwebtoken": "^9.0.7",
-    "@types/node": "^22.5.1",
-    "@types/papaparse": "^5.3.15",
+    "@types/jsonwebtoken": "^9.0.10",
+    "@types/node": "^22.13.5",
+    "@types/papaparse": "^5.5.1",
     "@types/tmp": "^0.2.6",
     "ts-dual-module": "^0.6.3",
-    "vitest": "^3.0.9"
+    "vitest": "^4.0.16"
   },
   "dependencies": {
-    "@aws-sdk/client-s3": "^3.693.0",
-    "@aws-sdk/client-textract": "^3.693.0",
-    "@aws-sdk/credential-providers": "^3.693.0",
+    "@aws-sdk/client-s3": "^3.948.0",
+    "@aws-sdk/client-textract": "^3.948.0",
+    "@aws-sdk/credential-providers": "^3.948.0",
     "@temporalio/activity": "^1.11.5",
     "@temporalio/workflow": "^1.11.5",
     "@types/json-schema": "^7.0.15",
     "fast-deep-equal": "^3.1.3",
-    "jsonwebtoken": "^9.0.2",
+    "jsonwebtoken": "^9.0.3",
     "mime": "^4.0.0",
     "ms": "3.0.0-canary.1",
     "node-web-stream-adapters": "^0.2.1",
     "p-limit": "^6.2.0",
-    "papaparse": "^5.4.1",
+    "papaparse": "^5.5.3",
     "seedrandom": "^3.0.5",
     "sharp": "^0.33.4",
-    "tiktoken": "^1.0.15",
+    "tiktoken": "^1.0.22",
     "tmp": "^0.2.4",
     "tmp-promise": "^3.0.3",
     "yaml": "^2.6.0",
-    "@vertesia/memory": "0.81.1",
-    "@vertesia/common": "0.81.1",
-    "@vertesia/client": "0.81.1",
-    "@llumiverse/common": "0.24.0",
-    "@vertesia/api-fetch-client": "0.81.1"
+    "@llumiverse/common": "1.0.0-dev.20260202.145450Z",
+    "@vertesia/api-fetch-client": "1.0.0-dev.20260203.130115Z",
+    "@vertesia/client": "1.0.0-dev.20260203.130115Z",
+    "@vertesia/common": "1.0.0-dev.20260203.130115Z",
+    "@vertesia/memory": "1.0.0-dev.20260203.130115Z"
   },
   "ts_dual_module": {
     "outDir": "lib",

package/src/activities/advanced/createOrUpdateDocumentFromInteractionRun.ts

@@ -107,7 +107,7 @@ export async function createOrUpdateDocumentFromInteractionRun(payload: DSLActiv
     let doc = undefined;
     if (params.update_existing_id) {
         log.info(`Updating existing document ${params.update_existing_id}`);
-        doc = await client.objects.update(params.update_existing_id, docPayload);
+        doc = await client.objects.update(params.update_existing_id, docPayload, { suppressWorkflows: true });
     } else {
         log.info(`Creating new document of type ${objectTypeName}`);
         doc = await client.objects.create(docPayload);

package/src/activities/advanced/updateDocumentFromInteractionRun.ts

@@ -29,7 +29,7 @@ export async function updateDocumentFromInteractionRun(payload: DSLActivityExecu
         return { status: "failed", error: "no-props" };
     }
 
-    await client.objects.update(objectId, docProps);
+    await client.objects.update(objectId, docProps, { suppressWorkflows: true });
 
     return { status: "success" };
 }

package/src/activities/copyParentArtifacts.ts

@@ -0,0 +1,162 @@
+import { log, activityInfo } from "@temporalio/activity";
+import { NodeStreamSource } from "@vertesia/client/node";
+import { DSLActivityExecutionPayload, DSLActivitySpec } from "@vertesia/common";
+import { Readable } from "stream";
+import { setupActivity } from "../dsl/setup/ActivityContext.js";
+
+/**
+ * Directories in the agent workspace to copy from parent to child.
+ */
+const WORKSPACE_DIRECTORIES = ['scripts', 'files', 'skills', 'docs', 'out'];
+
+/**
+ * Files that should never be copied (child has its own).
+ */
+const EXCLUDED_FILES = ['conversation.json', 'tools.json'];
+
+export interface CopyParentArtifactsParams {
+    parent_run_id: string;
+}
+
+export interface CopyParentArtifacts extends DSLActivitySpec<CopyParentArtifactsParams> {
+    name: 'copyParentArtifacts';
+    projection?: never;
+}
+
+/**
+ * Copy workspace artifacts from parent workflow's agent space to current workflow's agent space.
+ *
+ * Copies: scripts/, files/, skills/, docs/, out/
+ * Excludes: conversation.json, tools.json
+ */
+export async function copyParentArtifacts(
+    payload: DSLActivityExecutionPayload<CopyParentArtifactsParams>
+): Promise<{ copied: number; files: string[] }> {
+    const { client, params } = await setupActivity<CopyParentArtifactsParams>(payload);
+    const childRunId = activityInfo().workflowExecution.runId;
+    const parentRunId = params.parent_run_id;
+
+    log.info(`Copying artifacts from parent ${parentRunId} to child ${childRunId}`);
+
+    // List all files in parent's agent space
+    const parentFiles = await client.files.listArtifacts(parentRunId);
+
+    if (parentFiles.length === 0) {
+        log.info("No artifacts in parent agent space");
+        return { copied: 0, files: [] };
+    }
+
+    const copiedFiles: string[] = [];
+
+    for (const fullPath of parentFiles) {
+        const relativePath = extractRelativePath(fullPath, parentRunId);
+        if (!relativePath) continue;
+
+        // Only copy workspace directories
+        const dir = relativePath.split('/')[0];
+        if (!WORKSPACE_DIRECTORIES.includes(dir)) continue;
+
+        // Skip excluded files
+        const fileName = relativePath.split('/').pop() || '';
+        if (EXCLUDED_FILES.includes(fileName)) continue;
+
+        try {
+            await copyArtifact(client, parentRunId, childRunId, relativePath);
+            copiedFiles.push(relativePath);
+        } catch (err: any) {
+            log.warn(`Failed to copy ${relativePath}: ${err.message}`);
+        }
+    }
+
+    log.info(`Copied ${copiedFiles.length} artifacts from parent`);
+    return { copied: copiedFiles.length, files: copiedFiles };
+}
+
+/**
+ * Extract relative path from full artifact storage path.
+ * Storage paths can be: "store_xxx/agents/{runId}/..." or "agents/{runId}/..."
+ */
+function extractRelativePath(fullPath: string, runId: string): string | null {
+    // Look for the pattern: agents/{runId}/
+    const marker = `/agents/${runId}/`;
+    const idx = fullPath.lastIndexOf(marker);
+    if (idx >= 0) {
+        return fullPath.slice(idx + marker.length);
+    }
+
+    // Also try without leading slash
+    const legacyPrefix = `agents/${runId}/`;
+    if (fullPath.startsWith(legacyPrefix)) {
+        return fullPath.slice(legacyPrefix.length);
+    }
+
+    // If path contains the runId, try to extract after it
+    const runIdIdx = fullPath.indexOf(runId);
+    if (runIdIdx >= 0) {
+        const afterRunId = fullPath.slice(runIdIdx + runId.length);
+        if (afterRunId.startsWith('/')) {
+            return afterRunId.slice(1);
+        }
+    }
+
+    return null;
+}
+
+/**
+ * Copy a single artifact from parent to child agent space
+ */
+async function copyArtifact(
+    client: any,
+    parentRunId: string,
+    childRunId: string,
+    relativePath: string
+): Promise<void> {
+    // Download from parent's agent space
+    const stream = await client.files.downloadArtifact(parentRunId, relativePath);
+
+    // Convert web stream to node stream for upload
+    const nodeStream = Readable.fromWeb(stream as any);
+
+    // Determine mime type from extension
+    const mimeType = getMimeType(relativePath);
+    const fileName = relativePath.split('/').pop() || relativePath;
+
+    // Upload to child's agent space at the same relative path
+    const source = new NodeStreamSource(
+        nodeStream,
+        fileName,
+        mimeType
+    );
+
+    await client.files.uploadArtifact(childRunId, relativePath, source);
+}
+
+/**
+ * Get mime type from file extension
+ */
+function getMimeType(filePath: string): string {
+    const ext = filePath.split('.').pop()?.toLowerCase();
+    const mimeTypes: Record<string, string> = {
+        'json': 'application/json',
+        'txt': 'text/plain',
+        'md': 'text/markdown',
+        'html': 'text/html',
+        'csv': 'text/csv',
+        'png': 'image/png',
+        'jpg': 'image/jpeg',
+        'jpeg': 'image/jpeg',
+        'gif': 'image/gif',
+        'svg': 'image/svg+xml',
+        'pdf': 'application/pdf',
+        'xml': 'application/xml',
+        'zip': 'application/zip',
+        'js': 'application/javascript',
+        'ts': 'application/typescript',
+        'py': 'text/x-python',
+        'sh': 'text/x-shellscript',
+        'bash': 'text/x-shellscript',
+        'yaml': 'text/yaml',
+        'yml': 'text/yaml',
+    };
+    return mimeTypes[ext || ''] || 'application/octet-stream';
+}

package/src/activities/extractDocumentText.ts

@@ -67,7 +67,10 @@ export async function extractDocumentText(
 
     let txt: string;
 
-    switch (doc.content.type) {
+    // Normalize mime type by stripping charset and other parameters (e.g., "text/html; charset=UTF-8" -> "text/html")
+    const mimeType = doc.content.type.split(';')[0].trim();
+
+    switch (mimeType) {
         case "application/pdf":
             txt = await mutoolPdfToText(fileBuffer);
             break;

package/src/activities/generateDocumentProperties.ts

@@ -114,7 +114,7 @@ export async function generateDocumentProperties(
             date: new Date().toISOString(),
             model: infoRes.modelId,
         },
-    });
+    }, { suppressWorkflows: true });
 
     return { status: "completed" };
 }

package/src/activities/generateEmbeddings.ts

@@ -1,6 +1,6 @@
 import { EmbeddingsResult } from "@llumiverse/common";
 import { log } from "@temporalio/activity";
-import { VertesiaClient } from "@vertesia/client";
+import { VertesiaClient, ZenoClientNotFoundError } from "@vertesia/client";
 import {
     ContentObject,
     DSLActivityExecutionPayload,
@@ -11,7 +11,7 @@ import {
 } from "@vertesia/common";
 import { setupActivity } from "../dsl/setup/ActivityContext.js";
 import { DocumentNotFoundError } from "../errors.js";
-import { fetchBlobAsBase64 } from "../utils/blobs.js";
+import { fetchBlobAsBase64, md5 } from "../utils/blobs.js";
 import { DocPart } from "../utils/chunks.js";
 import { countTokens } from "../utils/tokens.js";
 
@@ -94,10 +94,18 @@ export async function generateEmbeddings(
         );
     }
 
-    const document = await client.objects.retrieve(
-        objectId,
-        "+text +parts +embeddings +tokens +properties",
-    );
+    let document;
+    try {
+        document = await client.objects.retrieve(
+            objectId,
+            "+text +parts +embeddings +tokens +properties",
+        );
+    } catch (error) {
+        if (error instanceof ZenoClientNotFoundError) {
+            throw new DocumentNotFoundError(`Document not found: ${objectId}`, [objectId]);
+        }
+        throw error;
+    }
 
     if (!document) {
         throw new DocumentNotFoundError("Document not found", [objectId]);
@@ -116,6 +124,7 @@ export async function generateEmbeddings(
                 config,
                 document,
                 type,
+                force,
             });
             break;
         case SupportedEmbeddingTypes.properties:
@@ -124,6 +133,7 @@ export async function generateEmbeddings(
                 config,
                 document,
                 type,
+                force,
             });
             break;
         case SupportedEmbeddingTypes.image:
@@ -132,6 +142,7 @@ export async function generateEmbeddings(
                 config,
                 document,
                 type,
+                force,
             });
             break;
         default:
@@ -155,7 +166,7 @@ interface ExecuteGenerateEmbeddingsParams {
 }
 
 async function generateTextEmbeddings(
-    { document, client, type, config }: ExecuteGenerateEmbeddingsParams
+    { document, client, type, config, force }: ExecuteGenerateEmbeddingsParams
 ) {
 
     if (!document) {
@@ -186,6 +197,21 @@ async function generateTextEmbeddings(
 
     const { environment } = config;
 
+    // Compute text etag for comparison
+    const textEtag = document.text_etag ?? (document.text ? md5(document.text) : undefined);
+
+    // Skip if embeddings already exist with matching etag (unless force=true)
+    const existingEmbedding = document.embeddings?.[type];
+    if (!force && existingEmbedding?.etag && textEtag && existingEmbedding.etag === textEtag) {
+        log.info(`Skipping ${type} embeddings for document ${document.id} - etag unchanged`);
+        return {
+            id: document.id,
+            type,
+            status: "skipped",
+            message: "embeddings already exist with matching etag",
+        };
+    }
+
     // Count tokens if needed, do not rely on existing token count
     let tokenCount : number | undefined = undefined;
     if (type === SupportedEmbeddingTypes.text && document.text) {
@@ -232,7 +258,7 @@ async function generateTextEmbeddings(
         await client.objects.setEmbedding(document.id, type, {
             values: res.values,
             model: res.model,
-            etag: document.text_etag,
+            etag: textEtag,
         });
 
         return {
@@ -249,6 +275,7 @@ async function generateImageEmbeddings({
     client,
     type,
     config,
+    force,
 }: ExecuteGenerateEmbeddingsParams) {
     log.info("Generating image embeddings for document " + document.id, {
         content: document.content,
@@ -264,6 +291,22 @@ async function generateImageEmbeddings({
             message: "content is not an image",
         };
     }
+
+    // Use content etag for image change detection
+    const contentEtag = document.content?.etag;
+
+    // Skip if embeddings already exist with matching etag (unless force=true)
+    const existingEmbedding = document.embeddings?.[type];
+    if (!force && existingEmbedding?.etag && contentEtag && existingEmbedding.etag === contentEtag) {
+        log.info(`Skipping ${type} embeddings for document ${document.id} - content etag unchanged`);
+        return {
+            id: document.id,
+            type,
+            status: "skipped",
+            message: "embeddings already exist with matching etag",
+        };
+    }
+
     const { environment, model } = config;
 
     const resRnd = await client.store.objects.getRendition(document.id, {
@@ -317,7 +360,7 @@ async function generateImageEmbeddings({
         {
             values: res.values,
             model: res.model,
-            etag: document.text_etag,
+            etag: contentEtag,
         },
     );
 

package/src/activities/index-dsl.ts

@@ -12,13 +12,15 @@ export { generateOrAssignContentType } from "./generateOrAssignContentType.js";
 export { getObjectFromStore } from "./getObjectFromStore.js";
 export { handleDslError } from "./handleError.js";
 export { prepareVideo } from "./media/prepareVideo.js";
+export { prepareAudio } from "./media/prepareAudio.js";
 export { convertPdfToStructuredText } from "./media/processPdfWithTextract.js";
 export { saveGladiaTranscription } from "./media/saveGladiaTranscription.js";
 export { transcribeMedia } from "./media/transcribeMediaWithGladia.js";
 export type { TranscriptMediaResult } from "./media/transcribeMediaWithGladia.js";
 export { notifyWebhook } from "./notifyWebhook.js";
 export { checkRateLimit } from "./rateLimiter.js";
+export { copyParentArtifacts } from "./copyParentArtifacts.js";
+export { mergeChildArtifacts } from "./mergeChildArtifacts.js";
 export { generateImageRendition } from "./renditions/generateImageRendition.js";
 export { generateVideoRendition } from "./renditions/generateVideoRendition.js";
 export { setDocumentStatus } from "./setDocumentStatus.js";
-