[chore]: Make TSConfig stricter and fix as many errors as possible (#…

…1596)

* Tighten TSConfig; fix new errors

* Fix tests; required re-adding ts-node

* Checkpoint. Stuck on @types/json-schema

* Fix ollama and core

* cleanup

* Move ts-node back to root

* pnpm lock updated

* Remove skipLibCheck from firebase; google-cloud still uses outdated long

* Add back gablorkenTool

* Revert "Remove skipLibCheck from firebase; google-cloud still uses outdated long"

This reverts commit 0ba65df.

js/ai/src/formats/types.d.ts → js/ai/src/formats/types.ts

@@ -19,14 +19,14 @@ import { GenerateResponseChunk } from '../generate.js';
 import { Message } from '../message.js';
 import { ModelRequest } from '../model.js';
 
-type OutputContentTypes = 'application/json' | 'text/plain';
+export type OutputContentTypes = 'application/json' | 'text/plain';
 
 export interface Formatter<O = unknown, CO = unknown> {
   name: string;
   config: ModelRequest['output'];
   handler: (schema?: JSONSchema) => {
     parseMessage(message: Message): O;
-    parseChunk?: (chunk: GenerateResponseChunk, cursor?: CC) => CO;
+    parseChunk?: (chunk: GenerateResponseChunk) => CO;
     instructions?: string;
   };
 }

js/ai/src/generate/action.ts

@@ -152,17 +152,15 @@ async function generate(
     streamingCallback
       ? (chunk: GenerateResponseChunkData) => {
           // Store accumulated chunk data
-          if (streamingCallback) {
-            streamingCallback!(
-              new GenerateResponseChunk(chunk, {
-                index: 0,
-                role: 'model',
-                previousChunks: accumulatedChunks,
-                parser: resolvedFormat?.handler(request.output?.schema)
-                  .parseChunk,
-              })
-            );
-          }
+          streamingCallback(
+            new GenerateResponseChunk(chunk, {
+              index: 0,
+              role: 'model',
+              previousChunks: accumulatedChunks,
+              parser: resolvedFormat?.handler(request.output?.schema)
+                .parseChunk,
+            })
+          );
           accumulatedChunks.push(chunk);
         }
       : undefined,

js/ai/src/generate/response.ts

@@ -73,14 +73,6 @@ export class GenerateResponse<O = unknown> implements ModelResponseData {
     this.request = options?.request;
   }
 
-  private get assertMessage(): Message<O> {
-    if (!this.message)
-      throw new Error(
-        'Operation could not be completed because the response does not contain a generated message.'
-      );
-    return this.message;
-  }
-
   /**
    * Throws an error if the response does not contain valid output.
    */

js/ai/src/model/middleware.ts

@@ -15,7 +15,7 @@
  */
 
 import { Document } from '../document.js';
-import {
+import type {
   MediaPart,
   MessageData,
   ModelInfo,
@@ -129,12 +129,15 @@ export function validateSupport(options: {
   };
 }
 
+// N.B. Figure out why array.findLast isn't available despite setting target
+// to ES2022 (Node 16.14.0)
 function lastUserMessage(messages: MessageData[]) {
   for (let i = messages.length - 1; i >= 0; i--) {
     if (messages[i].role === 'user') {
       return messages[i];
     }
   }
+  return undefined;
 }
 
 /**

js/ai/src/testing/model-tester.ts

@@ -53,9 +53,9 @@ const tests: Record<string, TestCase> = {
       ],
     });
 
-    const want = '';
+    const want = /plus/i;
     const got = response.text.trim();
-    assert.match(got, /plus/i);
+    assert.match(got, want);
   },
   history: async (registry: Registry, model: string) => {
     const resolvedModel = (await registry.lookupAction(
@@ -155,7 +155,7 @@ export async function testModels(
   registry: Registry,
   models: string[]
 ): Promise<TestReport> {
-  const gablorkenTool = defineTool(
+  defineTool(
     registry,
     {
       name: 'gablorkenTool',

js/ai/tests/formats/array_test.ts

@@ -68,7 +68,6 @@ describe('arrayFormat', () => {
     it(st.desc, () => {
       const parser = arrayFormatter.handler();
       const chunks: GenerateResponseChunkData[] = [];
-      let lastCursor = 0;
 
       for (const chunk of st.chunks) {
         const newChunk: GenerateResponseChunkData = {

js/ai/tests/formats/json_test.ts

@@ -64,16 +64,14 @@ describe('jsonFormat', () => {
     it(st.desc, () => {
       const parser = jsonFormatter.handler();
       const chunks: GenerateResponseChunkData[] = [];
-      let lastCursor = '';
 
       for (const chunk of st.chunks) {
         const newChunk: GenerateResponseChunkData = {
           content: [{ text: chunk.text }],
         };
 
         const result = parser.parseChunk!(
-          new GenerateResponseChunk(newChunk, { previousChunks: [...chunks] }),
-          lastCursor
+          new GenerateResponseChunk(newChunk, { previousChunks: [...chunks] })
         );
         chunks.push(newChunk);
 

js/core/package.json

@@ -32,11 +32,12 @@
     "@opentelemetry/sdk-metrics": "^1.25.0",
     "@opentelemetry/sdk-node": "^0.52.0",
     "@opentelemetry/sdk-trace-base": "^1.25.0",
+    "@types/json-schema": "^7.0.15",
     "ajv": "^8.12.0",
-    "body-parser": "^1.20.3",
-    "cors": "^2.8.5",
     "ajv-formats": "^3.0.1",
     "async-mutex": "^0.5.0",
+    "body-parser": "^1.20.3",
+    "cors": "^2.8.5",
     "express": "^4.21.0",
     "get-port": "^5.1.0",
     "json-schema": "^0.4.0",

js/core/src/tracing.ts

@@ -79,24 +79,17 @@ export async function enableTelemetry(
 }
 
 export async function cleanUpTracing(): Promise<void> {
-  return new Promise((resolve) => {
-    if (telemetrySDK) {
-      // Metrics are not flushed as part of the shutdown operation. If metrics
-      // are enabled, we need to manually flush them *before* the reader
-      // receives shutdown order.
-      const metricFlush = maybeFlushMetrics();
+  if (!telemetrySDK) {
+    return;
+  }
 
-      return metricFlush.then(() => {
-        return telemetrySDK!.shutdown().then(() => {
-          logger.debug('OpenTelemetry SDK shut down.');
-          telemetrySDK = null;
-          resolve();
-        });
-      });
-    } else {
-      resolve();
-    }
-  });
+  // Metrics are not flushed as part of the shutdown operation. If metrics
+  // are enabled, we need to manually flush them *before* the reader
+  // receives shutdown order.
+  await maybeFlushMetrics();
+  await telemetrySDK.shutdown();
+  logger.debug('OpenTelemetry SDK shut down.');
+  telemetrySDK = null;
 }
 
 /**

js/core/tsconfig.json

@@ -1,4 +1,8 @@
 {
   "extends": "../tsconfig.json",
-  "include": ["src"]
+  "include": ["src"],
+  "compilerOptions": {
+    // DOM is needed for streaming APIs.
+    "lib": ["es2022", "dom"]
+  }
 }

js/genkit/src/client/client.ts

@@ -194,6 +194,19 @@ export async function runFlow<O = any>({
       `Server returned: ${response.status}: ${await response.text()}`
     );
   }
-  const wrappedDesult = await response.json();
-  return wrappedDesult.result;
+  const wrappedResult = (await response.json()) as
+    | { result: O }
+    | { error: unknown };
+  if ('error' in wrappedResult) {
+    if (typeof wrappedResult.error === 'string') {
+      throw new Error(wrappedResult.error);
+    }
+    // TODO: The callable protocol defines an HttpError that has a JSON format of
+    // details?: string
+    // httpErrorCode: { canonicalName: string }
+    // message: string
+    // Should we create a new error class that parses this and exposes it as fields?
+    throw new Error(JSON.stringify(wrappedResult.error));
+  }
+  return wrappedResult.result;
 }

js/plugins/chroma/tsconfig.json

@@ -1,4 +1,9 @@
 {
   "extends": "../../tsconfig.json",
-  "include": ["src"]
+  "include": ["src"],
+  "compilerOptions": {
+    // chromadb does not compile cleany. It tries to import @xenova/transformers and
+    // chromadb-default-embed, which do not seem to be packaged with the library
+    "skipLibCheck": true
+  }
 }

js/plugins/firebase/tsconfig.json

@@ -1,4 +1,13 @@
 {
   "extends": "../../tsconfig.json",
+  "compilerOptions": {
+    // Google protobuf libraries have a transitive dependency on the long
+    // library, which incorrectly implements the ESM/CSM dual mode. This
+    // started in 5.0.0 and protobufjs requires ^5.0.0.
+    // This can be removed if https://github.com/dcodeIO/long.js/pull/130 gets
+    // approved, though it's been sitting around for over a year without
+    // acceptance.
+    "skipLibCheck": true
+  },
   "include": ["src"]
 }

js/plugins/google-cloud/src/gcpOpenTelemetry.ts

@@ -219,7 +219,7 @@ export class GcpOpenTelemetry {
  */
 class MetricExporterWrapper extends MetricExporter {
   constructor(
-    private options?: ExporterOptions,
+    options?: ExporterOptions,
     private errorHandler?: (error: Error) => void
   ) {
     super(options);

js/plugins/google-cloud/tests/metrics_test.ts

@@ -243,7 +243,7 @@ describe('GoogleCloudMetrics', () => {
     await ai.generate({ model: testModel, prompt: 'Hi' });
     await ai.generate({ model: testModel, prompt: 'Yo' });
 
-    const spans = await getExportedSpans();
+    await getExportedSpans();
 
     const requestCounter = await getCounterMetric('genkit/feature/requests');
     const latencyHistogram = await getHistogramMetric('genkit/feature/latency');
@@ -544,7 +544,7 @@ describe('GoogleCloudMetrics', () => {
 
   it('writes path metrics for a failing flow with exception in root', async () => {
     const flow = createFlow(ai, 'testFlow', async () => {
-      const subPath = await ai.run('sub-action', async () => {
+      await ai.run('sub-action', async () => {
         return 'done';
       });
       return Promise.reject(new Error('failed'));
@@ -582,8 +582,8 @@ describe('GoogleCloudMetrics', () => {
 
   it('writes path metrics for a failing flow with exception in subaction', async () => {
     const flow = createFlow(ai, 'testFlow', async () => {
-      const subPath1 = await ai.run('sub-action-1', async () => {
-        const subPath2 = await ai.run('sub-action-2', async () => {
+      await ai.run('sub-action-1', async () => {
+        await ai.run('sub-action-2', async () => {
           return Promise.reject(new Error('failed'));
         });
         return 'done';
@@ -627,8 +627,8 @@ describe('GoogleCloudMetrics', () => {
 
   it('writes path metrics for a flow with exception in action', async () => {
     const flow = createFlow(ai, 'testFlow', async () => {
-      const subPath1 = await ai.run('sub-action-1', async () => {
-        const subPath2 = await ai.run('sub-action-2', async () => {
+      await ai.run('sub-action-1', async () => {
+        await ai.run('sub-action-2', async () => {
           return 'done';
         });
         return Promise.reject(new Error('failed'));
@@ -674,10 +674,10 @@ describe('GoogleCloudMetrics', () => {
 
   it('writes path metrics for a flow with an exception in a serial action', async () => {
     const flow = createFlow(ai, 'testFlow', async () => {
-      const subPath1 = await ai.run('sub-action-1', async () => {
+      await ai.run('sub-action-1', async () => {
         return 'done';
       });
-      const subPath2 = await ai.run('sub-action-2', async () => {
+      await ai.run('sub-action-2', async () => {
         return Promise.reject(new Error('failed'));
       });
       return 'done';

js/plugins/google-cloud/tsconfig.json

@@ -1,4 +1,13 @@
 {
   "extends": "../../tsconfig.json",
-  "include": ["src"]
+  "include": ["src"],
+  "compilerOptions": {
+    // Google protobuf libraries have a transitive dependency on the long
+    // library, which incorrectly implements the ESM/CSM dual mode. This
+    // started in 5.0.0 and protobufjs requires ^5.0.0.
+    // This can be removed if https://github.com/dcodeIO/long.js/pull/130 gets
+    // approved, though it's been sitting around for over a year without
+    // acceptance.
+    "skipLibCheck": true
+  }
 }

js/plugins/langchain/tsconfig.json

@@ -1,4 +1,15 @@
 {
   "extends": "../../tsconfig.json",
-  "include": ["src"]
+  "include": ["src"],
+  "compilerOptions": {
+    // The langchain+core package has multiple ts errors:
+    // 1. In chat_models.d.ts and llms.d.ts, TypeScript cannot infer the type of `this`.
+    //    this error might be resolved by upgrading typescript versions
+    // 2. outputs.d.cts and base.d.cts mix up ESM and CSM and are using the wrong loader.
+    //
+    // The langchain package also has errors:
+    // 1. callbacks.d.ts, evaluation.d.cts, and base.d.cts mix up ESM and CSM and are
+    //    using the wrong loader
+    "skipLibCheck": true
+  }
 }

js/plugins/mcp/src/client/tools.ts

@@ -56,7 +56,7 @@ function registerTool(
     {
       name: `${params.name}/${tool.name}`,
       description: tool.description || '',
-      inputJsonSchema: tool.inputSchema,
+      inputJsonSchema: tool.inputSchema.properties,
       outputSchema: z.any(),
     },
     async (args) => {

js/plugins/mcp/src/index.ts

@@ -44,7 +44,7 @@ async function transportFrom(params: McpClientOptions): Promise<Transport> {
     const { SSEClientTransport } = await import(
       '@modelcontextprotocol/sdk/client/sse.js'
     );
-    return new SSEClientTransport(URL.parse(params.serverUrl)!);
+    return new SSEClientTransport(new URL(params.serverUrl));
   }
   if (params.serverProcess) {
     const { StdioClientTransport } = await import(

js/plugins/mcp/src/server.ts

@@ -184,7 +184,9 @@ export class GenkitMcpServer {
   }
 }
 
-function toMcpPromptArguments(p: PromptAction): Prompt['arguments'] {
+function toMcpPromptArguments(
+  p: PromptAction
+): Prompt['arguments'] | undefined {
   const jsonSchema = toJsonSchema({
     schema: p.__action.inputSchema,
     jsonSchema: p.__action.inputJsonSchema,
@@ -216,6 +218,7 @@ function toMcpPromptArguments(p: PromptAction): Prompt['arguments'] {
       required: jsonSchema.required?.includes(k),
     });
   }
+  return args;
 }
 
 const ROLE_MAP = { model: 'assistant', user: 'user' } as const;

js/plugins/ollama/package.json

@@ -31,10 +31,12 @@
   "peerDependencies": {
     "genkit": "workspace:*"
   },
+  "dependencies": {
+    "ollama": "^0.5.9"
+  },
   "devDependencies": {
     "@types/node": "^20.11.16",
     "npm-run-all": "^4.1.5",
-    "ollama": "^0.5.9",
     "rimraf": "^6.0.1",
     "tsup": "^8.3.5",
     "tsx": "^4.19.2",

js/plugins/ollama/src/embeddings.ts

@@ -98,7 +98,7 @@ export function defineOllamaEmbedder(
         );
       }
 
-      const payload: EmbedResponse = await response.json();
+      const payload = (await response.json()) as EmbedResponse;
 
       const embeddings: { embedding: number[] }[] = [];