@vertesia/workflow 0.51.0 → 0.54.0

package/src/dsl/setup/ActivityContext.ts

@@ -1,6 +1,11 @@
-import { log } from "@temporalio/activity";
+import { log, activityInfo } from "@temporalio/activity";
 import { VertesiaClient } from "@vertesia/client";
-import { DSLActivityExecutionPayload, DSLWorkflowExecutionPayload, Project, WorkflowExecutionPayload } from "@vertesia/common";
+import {
+    DSLActivityExecutionPayload,
+    DSLWorkflowExecutionPayload,
+    Project,
+    WorkflowExecutionPayload,
+} from "@vertesia/common";
 import { NoDocumentFound, WorkflowParamNotFound } from "../../errors.js";
 import { getProjectFromToken } from "../../utils/auth.js";
 import { getVertesiaClient } from "../../utils/client.js";
@@ -8,7 +13,6 @@ import { Vars } from "../vars.js";
 import { getFetchProvider, registerFetchProviderFactory } from "./fetch/index.js";
 import { DocumentProvider, DocumentTypeProvider, InteractionRunProvider } from "./fetch/providers.js";
 
-
 registerFetchProviderFactory(DocumentProvider.ID, DocumentProvider.factory);
 registerFetchProviderFactory(DocumentTypeProvider.ID, DocumentTypeProvider.factory);
 registerFetchProviderFactory(InteractionRunProvider.ID, InteractionRunProvider.factory);
@@ -17,7 +21,11 @@ export class ActivityContext<ParamsT extends Record<string, any>> {
     client: VertesiaClient;
     _project?: Promise<Project | undefined>;
 
-    constructor(public payload: DSLActivityExecutionPayload<ParamsT>, client: VertesiaClient, public params: ParamsT) {
+    constructor(
+        public payload: DSLActivityExecutionPayload<ParamsT>,
+        client: VertesiaClient,
+        public params: ParamsT,
+    ) {
         this.client = client;
         this.fetchProject = this.fetchProject.bind(this);
     }
@@ -29,24 +37,54 @@ export class ActivityContext<ParamsT extends Record<string, any>> {
     get objectId() {
         const objectId = this.payload.objectIds && this.payload.objectIds[0];
         if (!objectId) {
-            log.error('No objectId found in payload');
-            throw new WorkflowParamNotFound('objectIds[0]', (this.payload as WorkflowExecutionPayload as DSLWorkflowExecutionPayload).workflow);
+            log.error("No objectId found in payload");
+            throw new WorkflowParamNotFound(
+                "objectIds[0]",
+                (this.payload as WorkflowExecutionPayload as DSLWorkflowExecutionPayload).workflow,
+            );
         }
         return objectId;
     }
 
+    get activityInfo() {
+        return activityInfo();
+    }
+
+    get runId() {
+        const runId = activityInfo().workflowExecution.runId;
+        if (!runId) {
+            log.error("No runId found in activityInfo");
+            throw new WorkflowParamNotFound(
+                "runId",
+                (this.payload as WorkflowExecutionPayload as DSLWorkflowExecutionPayload).workflow,
+            );
+        }
+        return runId;
+    }
+
+    get workflowId() {
+        const workflowId = activityInfo().workflowExecution.workflowId;
+        if (!workflowId) {
+            log.error("No workflowId found in activityInfo");
+            throw new WorkflowParamNotFound(
+                "workflowId",
+                (this.payload as WorkflowExecutionPayload as DSLWorkflowExecutionPayload).workflow,
+            );
+        }
+        return workflowId;
+    }
+
     fetchProject() {
         if (!this._project) {
             this._project = _fetchProject(this.client, this.payload);
         }
         return this._project;
     }
-
 }
 
-
-export async function setupActivity<ParamsT extends Record<string, any>>(payload: DSLActivityExecutionPayload<ParamsT>) {
-
+export async function setupActivity<ParamsT extends Record<string, any>>(
+    payload: DSLActivityExecutionPayload<ParamsT>,
+) {
     const isDebugMode = !!payload.debug_mode;
 
     const vars = new Vars({
@@ -56,13 +94,17 @@ export async function setupActivity<ParamsT extends Record<string, any>>(payload
 
     //}
     if (isDebugMode) {
-        log.info(`Setting up activity ${payload.activity.name}`, { config: payload.config, activity: payload.activity, params: payload.params, vars });
+        log.info(`Setting up activity ${payload.activity.name}`, {
+            config: payload.config,
+            activity: payload.activity,
+            params: payload.params,
+            vars,
+        });
     }
 
     const client = getVertesiaClient(payload);
     const fetchSpecs = payload.activity.fetch;
     if (fetchSpecs) {
-
         const keys = Object.keys(fetchSpecs);
         if (keys.length > 0) {
             // create a new Vars instance to store the fetched data
@@ -74,7 +116,7 @@ export async function setupActivity<ParamsT extends Record<string, any>>(payload
                     fetchSpec = { ...fetchSpec, query };
                 }
 
-                const provider = await getFetchProvider(client, fetchSpec);
+                const provider = getFetchProvider(client, fetchSpec);
 
                 log.info(`Fetching data for ${key} with provider ${provider.name}`, { fetchSpec });
                 const result = await provider.fetch(fetchSpec);
@@ -84,7 +126,7 @@ export async function setupActivity<ParamsT extends Record<string, any>>(payload
                     } else {
                         vars.setValue(key, result);
                     }
-                } else if (fetchSpec.on_not_found === 'throw') {
+                } else if (fetchSpec.on_not_found === "throw") {
                     throw new NoDocumentFound("No documents found for: " + JSON.stringify(fetchSpec));
                 } else {
                     vars.setValue(key, null);
@@ -94,12 +136,11 @@ export async function setupActivity<ParamsT extends Record<string, any>>(payload
     }
 
     const params = vars.resolve() as ParamsT;
-    log.info(`Activity ${payload.activity.name} setup complete`, { params });
+    log.info(`Activity ${payload.activity.name} setup complete`);
 
     return new ActivityContext<ParamsT>(payload, client, params);
 }
 
-
 async function _fetchProject(client: VertesiaClient, payload: WorkflowExecutionPayload) {
     const project = await getProjectFromToken(payload.auth_token);
     return project ? await client.projects.retrieve(project.id) : undefined;

package/src/dsl/validation.test.ts

@@ -18,7 +18,7 @@ describe('workflow validation', () => {
         expect(errors.length).toBe(1);
     })
 
-    test('activities shpuld be an array', () => {
+    test('activities should be an array', () => {
         const workflow: any = {
             name: "test",
             activities: {}
@@ -81,7 +81,7 @@ describe('workflow validation', () => {
         expect(errors.length).toBe(0);
     })
 
-    test('import unknown imorted var through expressions', () => {
+    test('import unknown imported var through expressions', () => {
         const workflow: any = {
             name: "test",
             vars: { "foo": true },

package/src/dsl/vars.test.ts

@@ -125,7 +125,7 @@ describe('Workflow vars', () => {
         expect(resolved.prompt_data.message).toBe("hello");
         expect(resolved.data2).toBeTypeOf("object");
         expect(resolved.data2.message).toBe("hello");
-        expect(resolved.unnown).toBeUndefined();
+        expect(resolved.unknown).toBeUndefined();
     });
 
 

package/src/dsl/vars.ts

@@ -20,9 +20,9 @@ export function splitPath(path: string) {
 
 /**
  * Get the property named by "name" of the given object
- * If an array is idnexed using a string key then a map is done and an array with the content of the properties with that name are returned
+ * If an array is indexed using a string key then a map is done and an array with the content of the properties with that name are returned
  * Ex: docs.text => will return an array of text properties of the docs array
- * @param object the obejct
+ * @param object the object
  * @param name the name of the property.
  * @returns the property value
  */
@@ -128,7 +128,7 @@ export class Vars {
     map: Record<string, any>;
     /**
      * This property is used when resolving params. It contains the list of references that should not be resolved to their value
-     * but instead they need to return the string representation of the expression to eb able to regenrate another Vars instance with the same expression
+     * but instead they need to return the string representation of the expression to be able to regenerate another Vars instance with the same expression
      */
     //TODO this feature is no more used - it was replaced by importVars so we can now delete `preserveRefs`
     preserveRefs: Set<string> | undefined;
@@ -154,7 +154,7 @@ export class Vars {
     }
 
     /**
-     * Set a literal value (canboot set a ref)
+     * Set a literal value (cannot set a ref)
      * To add refs use `append()`
      * @param name
      * @param value
@@ -291,7 +291,7 @@ export class Vars {
     }
 
     getUnknownReferences(obj: any) {
-        const visitor = new UnknownRefrencesVisitor(this);
+        const visitor = new UnknownReferencesVisitor(this);
         new ObjectWalker().walk(obj, visitor);
         return visitor.result;
     }
@@ -311,7 +311,7 @@ function addImportVar(varPath: string, asName: string | undefined, vars: Vars, r
 }
 
 
-class UnknownRefrencesVisitor implements ObjectVisitor {
+class UnknownReferencesVisitor implements ObjectVisitor {
 
     result: { name: string, expression: string }[] = [];
 

package/src/dsl/workflow-exec-child.test.ts

@@ -1,6 +1,7 @@
-import { ContentEventName, DSLActivityExecutionPayload, DSLWorkflowExecutionPayload, DSLWorkflowStep } from '@vertesia/common';
+import * as protos from '@temporalio/proto';
 import { TestWorkflowEnvironment } from '@temporalio/testing';
 import { Worker } from '@temporalio/worker';
+import { ContentEventName, DSLActivityExecutionPayload, DSLWorkflowExecutionPayload, DSLWorkflowStep } from '@vertesia/common';
 import { afterAll, beforeAll, describe, expect, test } from 'vitest';
 import { dslWorkflow } from './dsl-workflow.js';
 import { setupActivity } from './setup/ActivityContext.js';
@@ -86,12 +87,21 @@ const steps2: DSLWorkflowStep[] = [
 // ========== test env setup ==========
 
 
-describe('DSL Workflow with chld workflows', () => {
+describe('DSL Workflow with child workflows', () => {
 
     let testEnv: TestWorkflowEnvironment;
 
     beforeAll(async () => {
         testEnv = await TestWorkflowEnvironment.createLocal();
+        const { connection } = testEnv;
+        await connection.operatorService.addSearchAttributes({
+            namespace: 'default',
+            searchAttributes: {
+                AccountId: protos.temporal.api.enums.v1.IndexedValueType.INDEXED_VALUE_TYPE_KEYWORD,
+                DocumentId: protos.temporal.api.enums.v1.IndexedValueType.INDEXED_VALUE_TYPE_KEYWORD,
+                ProjectId: protos.temporal.api.enums.v1.IndexedValueType.INDEXED_VALUE_TYPE_KEYWORD,
+            },
+        });
     });
 
     afterAll(async () => {
@@ -122,7 +132,7 @@ describe('DSL Workflow with chld workflows', () => {
             auth_token: 'test',
             config: {
                 studio_url: process.env.CP_STUDIO_URL || "http://localhost:8081",
-                store_url: process.env.CP_STODRE_URL || "http://localhost:8082",
+                store_url: process.env.CP_STORE_URL || "http://localhost:8082",
             },
             workflow: {
                 steps: steps1,
@@ -167,7 +177,7 @@ describe('DSL Workflow with chld workflows', () => {
             auth_token: 'test',
             config: {
                 studio_url: process.env.CP_STUDIO_URL || "http://localhost:8081",
-                store_url: process.env.CP_STODRE_URL || "http://localhost:8082",
+                store_url: process.env.CP_STORE_URL || "http://localhost:8082",
             },
             workflow: {
                 steps: steps2,

package/src/dsl/workflow-fetch.test.ts

@@ -110,7 +110,7 @@ describe('DSL Workflow', () => {
             auth_token: 'test',
             config: {
                 studio_url: process.env.CP_STUDIO_URL || "http://localhost:8081",
-                store_url: process.env.CP_STODRE_URL || "http://localhost:8082",
+                store_url: process.env.CP_STORE_URL || "http://localhost:8082",
             },
             workflow: {
                 activities,

package/src/dsl/workflow-import.test.ts

@@ -64,7 +64,7 @@ describe('DSL Workflow import vars', () => {
             auth_token: 'test',
             config: {
                 studio_url: process.env.CP_STUDIO_URL || "http://localhost:8081",
-                store_url: process.env.CP_STODRE_URL || "http://localhost:8082",
+                store_url: process.env.CP_STORE_URL || "http://localhost:8082",
             },
             workflow: {
                 activities,

package/src/dsl/workflow.test.ts

@@ -1,6 +1,7 @@
-import { ContentEventName, DSLActivityExecutionPayload, DSLActivitySpec, DSLWorkflowExecutionPayload } from '@vertesia/common';
+import * as protos from '@temporalio/proto';
 import { TestWorkflowEnvironment } from '@temporalio/testing';
 import { Worker } from '@temporalio/worker';
+import { ContentEventName, DSLActivityExecutionPayload, DSLActivitySpec, DSLWorkflowExecutionPayload } from '@vertesia/common';
 import { afterAll, beforeAll, describe, expect, it } from 'vitest';
 import { dslWorkflow } from './dsl-workflow.js';
 import { setupActivity } from "./setup/ActivityContext.js";
@@ -52,6 +53,15 @@ describe('DSL Workflow', () => {
 
     beforeAll(async () => {
         testEnv = await TestWorkflowEnvironment.createLocal();
+        const { connection } = testEnv;
+        await connection.operatorService.addSearchAttributes({
+            namespace: 'default',
+            searchAttributes: {
+                AccountId: protos.temporal.api.enums.v1.IndexedValueType.INDEXED_VALUE_TYPE_KEYWORD,
+                DocumentId: protos.temporal.api.enums.v1.IndexedValueType.INDEXED_VALUE_TYPE_KEYWORD,
+                ProjectId: protos.temporal.api.enums.v1.IndexedValueType.INDEXED_VALUE_TYPE_KEYWORD,
+            },
+        });
     });
 
     afterAll(async () => {
@@ -82,7 +92,7 @@ describe('DSL Workflow', () => {
             auth_token: 'test',
             config: {
                 studio_url: process.env.CP_STUDIO_URL || "http://localhost:8081",
-                store_url: process.env.CP_STODRE_URL || "http://localhost:8082",
+                store_url: process.env.CP_STORE_URL || "http://localhost:8082",
             },
             workflow: {
                 activities,

package/src/dsl.ts

@@ -1,3 +1,3 @@
 export * from "./dsl/dslProxyActivities.js";
+export * from "./errors.js";
 export * from "./result-types.js";
-export * from "./errors.js";

package/src/errors.ts

@@ -1,24 +1,45 @@
 import { DSLActivitySpec, DSLWorkflowSpec } from "@vertesia/common";
 
-
 export class NoDocumentFound extends Error {
-    constructor(message: string, public ids?: string[]) {
+    constructor(
+        message: string,
+        public ids?: string[],
+    ) {
         super(message);
         this.name = "NoDocumentFound";
-        this.ids = ids
+        this.ids = ids;
     }
 }
 
 export class ActivityParamNotFound extends Error {
-    constructor(public paramName: string, public activity: DSLActivitySpec) {
+    constructor(
+        public paramName: string,
+        public activity: DSLActivitySpec,
+    ) {
         super(`Required parameter ${paramName} not found in activity ${activity.name}`);
         this.name = "ActivityParamNotFound";
     }
 }
 
+export class ActivityParamInvalid extends Error {
+    constructor(
+        public paramName: string,
+        public activity: DSLActivitySpec,
+        reason?: string,
+    ) {
+        super(`${paramName} in activity ${activity.name} is invalid${reason ? ` ${reason}` : ""}`);
+        this.name = "ActivityParamInvalid";
+    }
+}
+
 export class WorkflowParamNotFound extends Error {
-    constructor(public paramName: string, public workflow?: DSLWorkflowSpec) {
+    constructor(
+        public paramName: string,
+        public workflow?: DSLWorkflowSpec,
+    ) {
         super(`Required parameter ${paramName} not found in workflow ${workflow?.name}`);
         this.name = "WorkflowParamNotFound";
     }
-}
+}
+
+export const WF_NON_RETRYABLE_ERRORS = ["NoDocumentFound", "ActivityParamNotFound", "WorkflowParamNotFound"];

package/src/index.ts

@@ -3,7 +3,7 @@
  * Modify the `./activities/index.ts` if you want to modify activity exports
  * 2. Workflows are exported through the '/workflows' named export in package.json.
  * Modify the `./workflows.ts` file if you want to modify workflow exports
- * 3. Here we export the API to be used to validate workflows and the types reuired to create workflow TS definitions.
+ * 3. Here we export the API to be used to validate workflows and the types required to create workflow TS definitions.
  */
 
 //TODO remove this - it is only for backward compat - iot is used from old workflows

package/src/iterative-generation/activities/extractToc.ts

@@ -40,7 +40,7 @@ export async function it_gen_extractToc(payload: WorkflowExecutionPayload): Prom
     await buildAndPublishMemoryPack(client, `${vars.memory}/output`, async () => {
         return {
             toc,
-            lastProcessdPart: undefined, // the part index (a number array)
+            lastProcessedPart: undefined, // the part index (a number array)
             previouslyGenerated: ""
         } as OutputMemoryMeta
     });

package/src/iterative-generation/activities/generatePart.ts

@@ -32,7 +32,7 @@ export async function it_gen_generatePart(payload: WorkflowExecutionPayload, pat
 
     const content = await loadGeneratedContent(outMemory);
 
-    let previously_generated = getPreviouslyGeneratedContent(content, !part, vars.rememberance_strategy);
+    let previously_generated = getPreviouslyGeneratedContent(content, !part, vars.remembrance_strategy);
 
     if (!part) { // a new section
         content.push({
@@ -58,7 +58,7 @@ export async function it_gen_generatePart(payload: WorkflowExecutionPayload, pat
 
     const result = r.result as string;
     content[content.length - 1].content += result;
-    meta.lastProcessdPart = path;
+    meta.lastProcessedPart = path;
     await buildAndPublishMemoryPack(client, `${memory}/output`, async ({ copyText }) => {
         copyText(JSON.stringify(content, null, 2), "content.json");
         return meta;

package/src/iterative-generation/activities/generateToc.ts

@@ -89,7 +89,7 @@ export async function it_gen_generateToc(payload: WorkflowExecutionPayload): Pro
     await buildAndPublishMemoryPack(client, `${vars.memory}/output`, async () => {
         return {
             toc,
-            lastProcessdPart: undefined, // the part index (a number array)
+            lastProcessedPart: undefined, // the part index (a number array)
             previouslyGenerated: ""
         } as OutputMemoryMeta
     });

package/src/iterative-generation/iterativeGenerationWorkflow.ts

@@ -1,6 +1,7 @@
 import { WorkflowExecutionPayload } from "@vertesia/common";
 
 import { log, proxyActivities } from "@temporalio/workflow";
+import { WF_NON_RETRYABLE_ERRORS } from "../errors.js";
 import * as activities from "./activities/index.js";
 import { IterativeGenerationPayload, PartIndex, SECTION_ID_PLACEHOLDER } from "./types.js";
 
@@ -16,7 +17,7 @@ const {
         backoffCoefficient: 2,
         maximumAttempts: 20,
         maximumInterval: 100 * 30 * 1000, //ms
-        nonRetryableErrorTypes: [],
+        nonRetryableErrorTypes: WF_NON_RETRYABLE_ERRORS,
     },
 });
 
@@ -29,7 +30,7 @@ export async function iterativeGenerationWorkflow(payload: WorkflowExecutionPayl
     }
 
     // extractToc tries to extract the toc from the input memory pack (toc.json or toc.yaml)
-    // the generateToc activity is retiurning the toc hierarchy.
+    // the generateToc activity is returning the toc hierarchy.
     // It doesn't include extra TOC details like description etc.
     // To minimize the payload size only the hierarchy and the section/part names are returned
     let toc = await it_gen_extractToc(payload);

package/src/iterative-generation/types.ts

@@ -13,7 +13,7 @@ export interface IterativeGenerationPayload {
     // the main interaction will only be used to prepare the iteration (to generate the TOC)
     // otherwise it will be used for the iterative generation too.
     interaction: string;
-    // if defined this will be used for the iterative interaction which will genrate parts.
+    // if defined this will be used for the iterative interaction which will generate parts.
     // otherwise the main interaction will be used for iterative generation.
     iterative_interaction?: string;
     // the environment to use
@@ -34,10 +34,10 @@ export interface IterativeGenerationPayload {
      * If not set to "none" the previously generated content will be passed to the iteration.
      * If not set at all defaults to "section".
      * If "section" is used only the section content previously generated will be passed to the next iteration.
-     * If "document" is used the whole previopusly generated document content will be passed to the next iteration.
+     * If "document" is used the whole previously generated document content will be passed to the next iteration.
      * Defaults to section.
      */
-    rememberance_strategy?: "document" | "section" | "none";
+    remembrance_strategy?: "document" | "section" | "none";
     /**
      * If present will save sections in files using the pattern
      * The pattern must include a placeholder for the section id: %id.
@@ -88,7 +88,7 @@ export interface TocIndex {
 export interface OutputMemoryMeta {
     toc: Toc;
     previouslyGenerated: string;
-    lastProcessdPart?: number[] | undefined;
+    lastProcessedPart?: number[] | undefined;
 }
 
 export interface Section {

package/src/iterative-generation/utils.ts

@@ -79,19 +79,19 @@ export function getPreviousPathIndex(toc: Toc, pathIndex: number[]): number[] |
 
 
 export function expectMemoryIsConsistent(meta: OutputMemoryMeta, pathIndex: number[]) {
-    const metaLastProcessedPart = meta.lastProcessdPart;
+    const metaLastProcessedPart = meta.lastProcessedPart;
     if (!metaLastProcessedPart) {
         if (pathIndex.length > 1 && pathIndex[0] !== 0) {
-            throw ApplicationFailure.nonRetryable('Memory last processed part is not consitent with the workflow.', 'MemoryPackNotConsistent', { currentIndex: pathIndex, expectedPreviousIndex: [pathIndex[0], pathIndex[1] - 1], previousIndex: metaLastProcessedPart });
+            throw ApplicationFailure.nonRetryable('Memory last processed part is not consistent with the workflow.', 'MemoryPackNotConsistent', { currentIndex: pathIndex, expectedPreviousIndex: [pathIndex[0], pathIndex[1] - 1], previousIndex: metaLastProcessedPart });
         } else {
             return;
         }
     }
     const prevPathIndex = getPreviousPathIndex(meta.toc, pathIndex);
     if (!prevPathIndex) {
-        throw ApplicationFailure.nonRetryable('Memory last processed part is not consitent with the workflow', 'MemoryPackNotConsistent', { currentIndex: pathIndex, expectedPreviousIndex: prevPathIndex, previousIndex: metaLastProcessedPart });
+        throw ApplicationFailure.nonRetryable('Memory last processed part is not consistent with the workflow', 'MemoryPackNotConsistent', { currentIndex: pathIndex, expectedPreviousIndex: prevPathIndex, previousIndex: metaLastProcessedPart });
     } else if (!isSamePartIndex(prevPathIndex, metaLastProcessedPart)) {
-        throw ApplicationFailure.nonRetryable('Memory last processed part is not consitent with the workflow', 'MemoryPackNotConsistent', { currentIndex: pathIndex, expectedPreviousIndex: prevPathIndex, previousIndex: meta.lastProcessdPart });
+        throw ApplicationFailure.nonRetryable('Memory last processed part is not consistent with the workflow', 'MemoryPackNotConsistent', { currentIndex: pathIndex, expectedPreviousIndex: prevPathIndex, previousIndex: meta.lastProcessedPart });
     }
 }
 

package/src/system/notifyWebhookWorkflow.ts

@@ -3,6 +3,7 @@ import { log } from "@temporalio/workflow";
 import { ContentEventName, WorkflowExecutionPayload } from "@vertesia/common";
 import * as activities from "../activities/notifyWebhook.js";
 import { dslProxyActivities } from "../dsl/dslProxyActivities.js";
+import { WF_NON_RETRYABLE_ERRORS } from "../errors.js";
 
 const {
     notifyWebhook
@@ -13,7 +14,7 @@ const {
         backoffCoefficient: 2,
         maximumAttempts: 5,
         maximumInterval: 100 * 30 * 1000, //ms
-        nonRetryableErrorTypes: [],
+        nonRetryableErrorTypes: WF_NON_RETRYABLE_ERRORS,
     },
 });
 

package/src/system/recalculateEmbeddingsWorkflow.ts

@@ -2,7 +2,7 @@
 import { SupportedEmbeddingTypes, WorkflowExecutionPayload } from "@vertesia/common";
 import * as activities from "../activities/index-dsl.js";
 import { dslProxyActivities } from "../dsl/dslProxyActivities.js";
-import { NoDocumentFound } from "../errors.js";
+import { WF_NON_RETRYABLE_ERRORS } from "../errors.js";
 
 const {
     generateEmbeddings,
@@ -13,7 +13,7 @@ const {
         backoffCoefficient: 2,
         maximumAttempts: 10,
         maximumInterval: 100 * 30 * 1000, //ms
-        nonRetryableErrorTypes: [NoDocumentFound.name],
+        nonRetryableErrorTypes: WF_NON_RETRYABLE_ERRORS,
     },
 });
 

package/src/utils/blobs.ts

@@ -10,7 +10,12 @@ export async function fetchBlobAsStream(client: VertesiaClient, blobUri: string)
     try {
         return await client.files.downloadFile(blobUri);
     } catch (err: any) {
-        throw new NoDocumentFound(`Failed to download blob ${blobUri}: ${err.message}`, []);
+        if (err.message.includes("not found")) {
+            //TODO improve error handling with a fetch fail error class in the client
+            throw new NoDocumentFound(`Failed to download blob ${blobUri}: ${err.message}`, []);
+        } else {
+            throw new Error(`Failed to download blob ${blobUri}: ${err.message}`);
+        }
     }
 }
 export async function fetchBlobAsBuffer(client: VertesiaClient, blobUri: string): Promise<Buffer> {
@@ -24,7 +29,7 @@ export async function fetchBlobAsBuffer(client: VertesiaClient, blobUri: string)
 
 export async function fetchBlobAsBase64(client: VertesiaClient, blobUri: string): Promise<string> {
     const buffer = await fetchBlobAsBuffer(client, blobUri);
-    return buffer.toString('base64');
+    return buffer.toString("base64");
 }
 
 export async function saveBlobToFile(client: VertesiaClient, blobUri: string, toFile: string): Promise<void> {
@@ -36,8 +41,8 @@ export async function saveBlobToFile(client: VertesiaClient, blobUri: string, to
 export async function saveBlobToTempFile(client: VertesiaClient, blobUri: string, fileExt?: string): Promise<string> {
     const tmpFile = tmp.fileSync({
         prefix: "vertesia-activity-",
-        postfix: "." + fileExt,
-        discardDescriptor: true
+        postfix: fileExt ? "." + fileExt : "",
+        discardDescriptor: true,
     });
     await saveBlobToFile(client, blobUri, tmpFile.name);
     return tmpFile.name;
@@ -47,12 +52,12 @@ async function writeChunksToStream(chunks: AsyncIterable<Uint8Array>, out: NodeJ
     for await (const chunk of chunks) {
         if (!out.write(chunk)) {
             // If the internal buffer is full, wait until it's drained
-            await new Promise(resolve => out.once('drain', resolve));
+            await new Promise((resolve) => out.once("drain", resolve));
         }
     }
     out.end(); // Close the stream when done
 }
 
 export function md5(contents: string) {
-    return crypto.createHash('md5').update(contents).digest("hex");
+    return crypto.createHash("md5").update(contents).digest("hex");
 }

package/src/utils/chunks.ts

@@ -0,0 +1,17 @@
+
+export interface DocPart {
+    line_number_start: number
+    line_number_end: number
+    name: string
+    type: string
+}
+
+export const getContentPart = (content: string, part: DocPart): string => {
+    const lines = content.split('\n');
+    const text = lines.filter((_l, i) => i >= part.line_number_start && i <= part.line_number_end).join('\n');
+    return text;
+}
+
+export const getContentParts = (content: string, parts: DocPart[]): string[] => {
+    return parts.map(part => getContentPart(content, part));
+}

package/src/utils/client.ts

@@ -4,20 +4,21 @@
 
 import { VertesiaClient } from "@vertesia/client";
 import { WorkflowExecutionBaseParams } from "@vertesia/common";
+import { WorkflowParamNotFound } from "../errors.js";
 
 
 export function getVertesiaClient(payload: WorkflowExecutionBaseParams) {
 
     if (!payload.auth_token) {
-        throw new Error("Authentication Token is missing from WorkflowExecutionPayload.authToken");
+        throw new WorkflowParamNotFound("Authentication Token is missing from WorkflowExecutionPayload.authToken");
     }
 
     if (!payload.config?.studio_url) {
-        throw new Error("Content Store URL is missing from WorkflowExecutionPayload.servers.storeUrl");
+        throw new WorkflowParamNotFound("Content Store URL is missing from WorkflowExecutionPayload.servers.storeUrl");
     }
 
     if (!payload.config?.store_url) {
-        throw new Error("Content Store URL is missing from WorkflowExecutionPayload.servers.storeUrl");
+        throw new WorkflowParamNotFound("Content Store URL is missing from WorkflowExecutionPayload.servers.storeUrl");
     }
 
     const client = new VertesiaClient({