@vertesia/workflow 1.4.0-dev.20260615.042549Z → 1.4.0-dev.20260629.130134Z

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@vertesia/workflow",
-  "version": "1.4.0-dev.20260615.042549Z",
+  "version": "1.4.0-dev.20260629.130134Z",
   "type": "module",
   "description": "Vertesia workflow DSL",
   "main": "./lib/index.js",
@@ -15,23 +15,23 @@
   "license": "Apache-2.0",
   "devDependencies": {
     "@smithy/types": "^4.14.3",
-    "@temporalio/proto": "^1.17.2",
-    "@temporalio/testing": "^1.17.2",
-    "@temporalio/worker": "^1.17.2",
+    "@temporalio/proto": "^1.18.1",
+    "@temporalio/testing": "^1.18.1",
+    "@temporalio/worker": "^1.18.1",
     "@types/jsonwebtoken": "^9.0.10",
-    "@types/node": "^24.12.4",
+    "@types/node": "^24.13.2",
     "@types/papaparse": "^5.5.2",
     "@types/tmp": "^0.2.6",
     "typescript": "^6.0.3",
-    "vitest": "^4.1.8",
+    "vitest": "^4.1.9",
     "@vertesia/tsconfig": "0.1.0"
   },
   "dependencies": {
-    "@aws-sdk/client-s3": "^3.1061.0",
-    "@aws-sdk/client-textract": "^3.1061.0",
-    "@aws-sdk/credential-providers": "^3.1061.0",
-    "@temporalio/activity": "^1.17.2",
-    "@temporalio/workflow": "^1.17.2",
+    "@aws-sdk/client-s3": "^3.1068.0",
+    "@aws-sdk/client-textract": "^3.1068.0",
+    "@aws-sdk/credential-providers": "^3.1068.0",
+    "@temporalio/activity": "^1.18.1",
+    "@temporalio/workflow": "^1.18.1",
     "fast-deep-equal": "^3.1.3",
     "jsonwebtoken": "^9.0.3",
     "mime": "^4.1.0",
@@ -44,12 +44,12 @@
     "tiktoken": "^1.0.22",
     "tmp": "^0.2.7",
     "tmp-promise": "^3.0.3",
-    "undici": "^7.27.0",
+    "undici": "^7.27.2",
     "yaml": "^2.9.0",
-    "@llumiverse/common": "1.3.0",
-    "@vertesia/client": "1.4.0-dev.20260615.042549Z",
-    "@vertesia/common": "1.4.0-dev.20260615.042549Z",
-    "@vertesia/api-fetch-client": "1.4.0-dev.20260615.042549Z"
+    "@vertesia/client": "1.4.0-dev.20260629.130134Z",
+    "@llumiverse/common": "1.4.0-dev.20260629.090753Z",
+    "@vertesia/common": "1.4.0-dev.20260629.130134Z",
+    "@vertesia/api-fetch-client": "1.4.0-dev.20260629.130134Z"
   },
   "exports": {
     ".": {
@@ -107,7 +107,8 @@
     "lint": "biome lint src",
     "test": "vitest run",
     "typecheck:test": "tsc -p tsconfig.test.json --noEmit",
-    "build": "rm -rf ./lib ./tsconfig.tsbuildinfo && tsc && node ./bin/bundle-workflows.mjs lib/workflows.js lib/workflows-bundle.js",
-    "clean": "rm -rf ./lib tsconfig.tsbuildinfo"
+    "clean:lib": "rimraf ./lib ./tsconfig.tsbuildinfo",
+    "build": "pnpm run clean:lib && tsc -p tsconfig.json && node ./bin/bundle-workflows.mjs lib/workflows.js lib/workflows-bundle.js",
+    "clean": "rimraf ./node_modules ./lib ./tsconfig.tsbuildinfo"
   }
 }

package/src/activities/generateEmbeddings.ts

@@ -102,6 +102,10 @@ export async function generateEmbeddings(payload: DSLActivityExecutionPayload<Ge
         throw new DocumentNotFoundError('Document not found', [objectId]);
     }
 
+    if (!document.content) {
+        throw new DocumentNotFoundError('Document content not found', [objectId]);
+    }
+
     let res:
         | Awaited<ReturnType<typeof generateTextEmbeddings>>
         | Awaited<ReturnType<typeof generateImageEmbeddings>>
@@ -168,19 +172,14 @@ async function generateTextEmbeddings({ document, client, type, config, force }:
         };
     }
 
-    const sourceText =
-        type === SupportedEmbeddingTypes.text
-            ? document.text
-            : document.properties
-              ? JSON.stringify(document.properties)
-              : undefined;
-
-    if (!sourceText) {
+    if (type === SupportedEmbeddingTypes.text && !document.text) {
+        return { id: document.id, status: 'failed', message: 'no text found' };
+    }
+    if (type === SupportedEmbeddingTypes.properties && !document?.properties) {
         return {
             id: document.id,
-            type,
-            status: 'skipped',
-            message: type === SupportedEmbeddingTypes.text ? 'no text found' : 'no properties found',
+            status: 'failed',
+            message: 'no properties found',
         };
     }
 
@@ -191,12 +190,12 @@ async function generateTextEmbeddings({ document, client, type, config, force }:
         );
     }
 
-    const sourceEtag =
-        type === SupportedEmbeddingTypes.text ? (document.text_etag ?? md5(sourceText)) : md5(sourceText);
+    // 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 && existingEmbedding.etag === sourceEtag) {
+    if (!force && existingEmbedding?.etag && textEtag && existingEmbedding.etag === textEtag) {
         log.debug(`Skipping ${type} embeddings for document ${document.id} - etag unchanged`);
         return {
             id: document.id,
@@ -207,7 +206,15 @@ async function generateTextEmbeddings({ document, client, type, config, force }:
     }
 
     // Count tokens if needed, do not rely on existing token count
-    const tokenCount = countTokens(sourceText).count;
+    let tokenCount: number | undefined;
+    if (type === SupportedEmbeddingTypes.text && document.text) {
+        tokenCount = countTokens(document.text).count;
+    }
+
+    if (type === SupportedEmbeddingTypes.properties && document.properties) {
+        const propertiesText = JSON.stringify(document.properties);
+        tokenCount = countTokens(propertiesText).count;
+    }
 
     const maxTokens = config.max_tokens ?? 8000;
 
@@ -224,7 +231,7 @@ async function generateTextEmbeddings({ document, client, type, config, force }:
     } else {
         log.debug(`Generating ${type} embeddings for document`);
 
-        const res = await generateEmbeddingsFromStudio(sourceText, environment, client);
+        const res = await generateEmbeddingsFromStudio(JSON.stringify(document[type]), environment, client);
         const values = res?.results?.[0]?.outputs?.[0]?.values;
         if (!values) {
             return {
@@ -240,7 +247,7 @@ async function generateTextEmbeddings({ document, client, type, config, force }:
         await client.objects.setEmbedding(document.id, type, {
             values,
             model: res.model,
-            etag: sourceEtag,
+            etag: textEtag,
         });
 
         return {

package/src/dsl/dslProxyActivities.ts

@@ -14,16 +14,6 @@ export type DslSimplifiedActivityFunction<ParamsT extends object = Record<string
     params: ParamsT,
 ) => Promise<ReturnT>;
 
-export function stripWorkflowContinuationFromVars<T>(vars: T): T {
-    if (!vars || typeof vars !== 'object' || Array.isArray(vars) || !('_continuation' in vars)) {
-        return vars;
-    }
-
-    const rest = { ...(vars as Record<string, unknown>) };
-    delete rest._continuation;
-    return rest as T;
-}
-
 export function dslProxyActivities<ActivitiesT extends object>(workflowName: string, options: ActivityOptions = {}) {
     type DslActivities = {
         [K in keyof ActivitiesT]: ActivitiesT[K] extends DslActivityFunction<infer ParamsT, infer ReturnT>
@@ -44,7 +34,6 @@ export function dslProxyActivities<ActivitiesT extends object>(workflowName: str
                 return (payload: WorkflowExecutionPayload, params: Record<string, unknown>) => {
                     return activityFn({
                         ...payload,
-                        vars: stripWorkflowContinuationFromVars(payload.vars),
                         activity: {
                             name: prop as string,
                         },

package/src/dsl/setup/ActivityContext.ts

@@ -93,7 +93,7 @@ export class ActivityContext<ParamsT extends object> {
      */
     get file(): WorkflowInputFile {
         const input = this.payload.input;
-        if (!input || input.inputType !== 'files') {
+        if (input?.inputType !== 'files') {
             throw new WorkflowExecutionError('Activity expects files but received objectIds');
         }
         // TypeScript now knows input is { inputType: 'files', files: WorkflowInputFile[] }
@@ -115,7 +115,7 @@ export class ActivityContext<ParamsT extends object> {
      */
     get files(): WorkflowInputFile[] {
         const input = this.payload.input;
-        if (!input || input.inputType !== 'files') {
+        if (input?.inputType !== 'files') {
             throw new WorkflowExecutionError('Activity expects files but received objectIds');
         }
         // TypeScript now knows input is { inputType: 'files', files: WorkflowInputFile[] }

package/src/errors.ts

@@ -116,9 +116,4 @@ export const WF_NON_RETRYABLE_ERRORS = [
     'TokenExpiredError',
     'ZenoClientNotFoundError',
     'WorkflowExecutionError',
-    // Temporal raises NotFoundError when an activity isn't registered on the worker
-    // (e.g. a new tool whose activity barrel export was missed). Retrying can never
-    // make it appear within a worker version, and the retried failure carries the
-    // full activity list which blows the payload size limit — so fail fast.
-    'NotFoundError',
 ];

package/src/utils/client.ts

@@ -2,17 +2,19 @@
  * get a zeno client for a given token
  */
 
-import type { FETCH_FN } from '@vertesia/api-fetch-client';
 import { decodeJWT, VertesiaClient, type VertesiaClientProps } from '@vertesia/client';
 import type { WorkflowExecutionBaseParams } from '@vertesia/common';
-import { Agent } from 'undici';
 import { WorkflowParamNotFoundError } from '../errors.js';
 
-const DEFAULT_WORKFLOW_FETCH_TIMEOUT_MS = 30 * 60 * 1000;
+// Short default timeout for ordinary workflow -> server/store calls (object GETs, status updates,
+// etc.). A stale/dead pooled connection (a server pod scaled down/rolled mid-request) used to hang
+// for the whole 30-minute undici headersTimeout; this bounds it to seconds so it fails fast and the
+// activity is retried. The long path — synchronous interaction execution, which blocks on the model —
+// sets its own long per-request timeout in @vertesia/client (executeInteraction*), overriding this.
+// Override the default via VERTESIA_WORKFLOW_FETCH_TIMEOUT_MS (0/false disables it).
+const DEFAULT_WORKFLOW_FETCH_TIMEOUT_MS = 60 * 1000;
 const WORKFLOW_FETCH_TIMEOUT_ENV = 'VERTESIA_WORKFLOW_FETCH_TIMEOUT_MS';
 
-let workflowFetch: Promise<FETCH_FN> | undefined;
-
 export function getVertesiaClient(payload: WorkflowExecutionBaseParams<unknown>) {
     return new VertesiaClient(getVertesiaClientOptions(payload));
 }
@@ -41,46 +43,32 @@ export function getVertesiaClientOptions(payload: WorkflowExecutionBaseParams<un
         storeUrl: payload.config.store_url,
         tokenServerUrl: token.iss,
         apikey: payload.auth_token,
-        fetch: getWorkflowFetch(),
+        timeout: parseWorkflowFetchTimeoutMs(),
     };
 }
 
-function getWorkflowFetch(): Promise<FETCH_FN> {
-    workflowFetch ??= createWorkflowFetch();
-    return workflowFetch;
-}
-
-async function createWorkflowFetch(): Promise<FETCH_FN> {
-    if (typeof globalThis.fetch !== 'function') {
-        throw new Error('No Fetch implementation found');
-    }
-
-    const timeoutMs = parseWorkflowFetchTimeoutMs();
-    if (timeoutMs === 0) {
-        return globalThis.fetch.bind(globalThis);
-    }
-
-    const dispatcher = new Agent({
-        headersTimeout: timeoutMs,
-        bodyTimeout: timeoutMs,
-    });
-
-    return (input, init) =>
-        globalThis.fetch(input, {
-            ...init,
-            dispatcher,
-        } as unknown as RequestInit);
-}
-
-function parseWorkflowFetchTimeoutMs(): number {
-    const raw = process.env[WORKFLOW_FETCH_TIMEOUT_ENV];
+function parseWorkflowFetchTimeoutMs(): number | false {
+    const raw = typeof process !== 'undefined' ? process.env?.[WORKFLOW_FETCH_TIMEOUT_ENV] : undefined;
     if (!raw) {
         return DEFAULT_WORKFLOW_FETCH_TIMEOUT_MS;
     }
 
-    const parsed = Number.parseInt(raw, 10);
-    if (!Number.isFinite(parsed) || parsed < 0) {
+    // Only an explicit 0/false disables the fail-fast timeout.
+    const normalized = raw.trim().toLowerCase();
+    if (normalized === '0' || normalized === 'false') {
+        return false;
+    }
+
+    // Require a plain positive integer count of milliseconds. parseInt() is too lenient — it
+    // reads "30s" as 30 (i.e. 30ms, off by 1000x) and "60000ms" as 60000 — so validate the raw
+    // string first. Any other garbage (typo, negative, unit suffix) keeps the fail-fast default
+    // rather than silently disabling the timeout, which would reintroduce the multi-minute hang.
+    if (!/^\d+$/.test(normalized) || Number.parseInt(normalized, 10) <= 0) {
+        console.warn(
+            `[workflow] Invalid ${WORKFLOW_FETCH_TIMEOUT_ENV}="${raw}"; ` +
+                `expected a positive integer in milliseconds. Falling back to ${DEFAULT_WORKFLOW_FETCH_TIMEOUT_MS}ms`,
+        );
         return DEFAULT_WORKFLOW_FETCH_TIMEOUT_MS;
     }
-    return parsed;
+    return Number.parseInt(normalized, 10);
 }

package/src/activities/generateEmbeddings.test.ts

@@ -1,114 +0,0 @@
-import { MockActivityEnvironment } from '@temporalio/testing';
-import type { VertesiaClient } from '@vertesia/client';
-import {
-    ContentEventName,
-    type ContentObject,
-    type DSLActivityExecutionPayload,
-    SupportedEmbeddingTypes,
-} from '@vertesia/common';
-import { beforeAll, beforeEach, describe, expect, it, vi } from 'vitest';
-import type { ActivityContext } from '../dsl/setup/ActivityContext.js';
-import { type GenerateEmbeddingsParams, generateEmbeddings } from './generateEmbeddings.js';
-
-vi.mock('../dsl/setup/ActivityContext.js', async (importOriginal) => {
-    const actual = await importOriginal<typeof import('../dsl/setup/ActivityContext.js')>();
-    return { ...actual, setupActivity: vi.fn() };
-});
-
-let testEnv: MockActivityEnvironment;
-
-beforeAll(() => {
-    testEnv = new MockActivityEnvironment();
-});
-
-beforeEach(() => {
-    vi.clearAllMocks();
-});
-
-const createPayload = (params: GenerateEmbeddingsParams): DSLActivityExecutionPayload<GenerateEmbeddingsParams> => {
-    const activityParams: GenerateEmbeddingsParams & Record<string, unknown> = { ...params };
-    return {
-        auth_token: 'mock-token',
-        account_id: 'test-account',
-        project_id: 'test-project',
-        params,
-        config: { studio_url: 'http://mock-studio', store_url: 'http://mock-store' },
-        workflow_name: 'StandardDocumentIntake',
-        event: ContentEventName.create,
-        objectIds: ['properties-only-object'],
-        input: { inputType: 'objectIds', objectIds: ['properties-only-object'] },
-        vars: {},
-        activity: { name: 'generateEmbeddings', params: activityParams },
-    };
-};
-
-describe('generateEmbeddings', () => {
-    it('should generate property embeddings for an object without content', async () => {
-        const { setupActivity } = await import('../dsl/setup/ActivityContext.js');
-        const document = {
-            id: 'properties-only-object',
-            properties: {
-                title: 'Properties-only object',
-                category: 'metadata',
-            },
-        } satisfies Partial<ContentObject>;
-        const embeddingResponse = {
-            model: 'embedding-model',
-            results: [{ outputs: [{ values: [0.1, 0.2, 0.3] }] }],
-        };
-        const client = {
-            objects: {
-                retrieve: vi.fn().mockResolvedValue(document),
-                setEmbedding: vi.fn().mockResolvedValue(undefined),
-            },
-            environments: {
-                embeddings: vi.fn().mockResolvedValue(embeddingResponse),
-            },
-        } as unknown as VertesiaClient;
-        const params = {
-            type: SupportedEmbeddingTypes.properties,
-            force: false,
-        } satisfies GenerateEmbeddingsParams;
-
-        vi.mocked(setupActivity).mockResolvedValue({
-            client,
-            objectId: document.id,
-            params,
-            fetchProject: vi.fn().mockResolvedValue({
-                name: 'Test Project',
-                namespace: 'test-project',
-                configuration: {
-                    embeddings: {
-                        [SupportedEmbeddingTypes.properties]: {
-                            enabled: true,
-                            environment: 'test-environment',
-                            max_tokens: 8000,
-                        },
-                    },
-                },
-            }),
-        } as unknown as ActivityContext<GenerateEmbeddingsParams>);
-
-        const result = await testEnv.run(generateEmbeddings, createPayload(params));
-
-        expect(result).toEqual({
-            id: document.id,
-            type: SupportedEmbeddingTypes.properties,
-            status: 'completed',
-            len: 3,
-        });
-        expect(client.objects.retrieve).toHaveBeenCalledWith(
-            document.id,
-            '+text +parts +embeddings +tokens +properties',
-        );
-        expect(client.environments.embeddings).toHaveBeenCalledWith('test-environment', {
-            inputs: [{ type: 'text', text: JSON.stringify(document.properties) }],
-            model: undefined,
-        });
-        expect(client.objects.setEmbedding).toHaveBeenCalledWith(document.id, SupportedEmbeddingTypes.properties, {
-            values: embeddingResponse.results[0].outputs[0].values,
-            model: embeddingResponse.model,
-            etag: expect.any(String),
-        });
-    });
-});

package/src/dsl/dslProxyActivities.test.ts

@@ -1,23 +0,0 @@
-import { describe, expect, it } from 'vitest';
-import { stripWorkflowContinuationFromVars } from './dslProxyActivities.js';
-
-describe('stripWorkflowContinuationFromVars', () => {
-    it('removes continue-as-new state from activity vars', () => {
-        const vars = {
-            interaction: 'sys:AppDeveloper',
-            _continuation: {
-                conversationState: { large: true },
-            },
-        };
-
-        expect(stripWorkflowContinuationFromVars(vars)).toEqual({
-            interaction: 'sys:AppDeveloper',
-        });
-        expect(vars).toHaveProperty('_continuation');
-    });
-
-    it('returns primitive and array vars unchanged', () => {
-        expect(stripWorkflowContinuationFromVars(undefined)).toBeUndefined();
-        expect(stripWorkflowContinuationFromVars(['a'])).toEqual(['a']);
-    });
-});