modelfusion 0.128.0 → 0.129.1

package/CHANGELOG.md

@@ -1,5 +1,32 @@
 # Changelog
 
+## v0.129.0 - 2024-01-20
+
+### Changed
+
+- **breaking change**: Usage of Node `async_hooks` has been renamed from `node:async_hooks` to `async_hooks` for easier Webpack configuration. To exclude the `async_hooks` from client-side bundling, you can use the following config for Next.js (`next.config.mjs` or `next.config.js`):
+
+  ```js
+  /**
+   * @type {import('next').NextConfig}
+   */
+  const nextConfig = {
+    webpack: (config, { isServer }) => {
+      if (isServer) {
+        return config;
+      }
+
+      config.resolve = config.resolve ?? {};
+      config.resolve.fallback = config.resolve.fallback ?? {};
+
+      // async hooks is not available in the browser:
+      config.resolve.fallback.async_hooks = false;
+
+      return config;
+    },
+  };
+  ```
+
 ## v0.128.0 - 2024-01-20
 
 ### Changed
@@ -8,9 +35,7 @@
 - **breaking change**: Image content in multi-modal instruction and chat inputs (e.g. for GPT Vision) is passed in the `image` property (instead of `base64Image`) and supports both base64 strings and `Uint8Array` inputs:
 
   ```ts
-  const image = fs.readFileSync(path.join("data", "example-image.png"), {
-    encoding: "base64",
-  });
+  const image = fs.readFileSync(path.join("data", "example-image.png"));
 
   const textStream = await streamText({
     model: openai.ChatTextGenerator({

package/README.md

@@ -86,7 +86,7 @@ Multi-modal vision models such as GPT 4 Vision can process images as part of the
 import { streamText, openai } from "modelfusion";
 import { readFileSync } from "fs";
 
-const image = readFileSync("./image.png").toString("base64");
+const image = readFileSync("./image.png");
 
 const textStream = await streamText({
   model: openai.ChatTextGenerator({ model: "gpt-4-vision-preview" }),
@@ -219,7 +219,7 @@ Synthesize speech (audio) from text. Also called TTS (text-to-speech).
 ```ts
 import { generateSpeech, lmnt } from "modelfusion";
 
-// `speech` is a Buffer with MP3 audio data
+// `speech` is a Uint8Array with MP3 audio data
 const speech = await generateSpeech({
   model: lmnt.SpeechGenerator({
     voice: "034b632b-df71-46c8-b440-86a42ffc3cf3", // Henry
@@ -257,7 +257,7 @@ const speechStream = await streamSpeech({
 });
 
 for await (const part of speechStream) {
-  // each part is a Buffer with MP3 audio data
+  // each part is a Uint8Array with MP3 audio data
 }
 ```
 
@@ -635,17 +635,14 @@ A web chat with an AI assistant, implemented as a Next.js app.
 
 Ask questions about a PDF document and get answers from the document.
 
-### [Image generator (Next.js)](https://github.com/lgrammel/modelfusion/tree/main/examples/image-generator-next-js)
+### [Next.js / ModelFusion Demos](https://github.com/lgrammel/modelfusion/tree/main/examples/nextjs)
 
-> _Next.js app_, _Stability AI image generation_
+> _Next.js app_, _image generation_, _transcription_, _OpenAI_, _Stability AI_
 
-Create an 19th century painting image for your input.
+Examples of using ModelFusion with Next.js 14 (App Router):
 
-### [Voice recording and transcription (Next.js)](https://github.com/lgrammel/modelfusion/tree/main/examples/voice-recording-next-js)
-
-> _Next.js app_, _OpenAI Whisper_
-
-Record audio with push-to-talk and transcribe it using Whisper, implemented as a Next.js app. The app shows a list of the transcriptions.
+- image generation
+- voice recording & transcription
 
 ### [Duplex Speech Streaming (using Vite/React & ModelFusion Server/Fastify)](https://github.com/lgrammel/modelfusion/tree/main/examples/speech-streaming-vite-react-fastify)
 

package/core/getRun.cjs

@@ -34,7 +34,8 @@ async function ensureLoaded() {
     if (!isNode)
         return Promise.resolve();
     if (!runStorage) {
-        const { AsyncLocalStorage } = await Promise.resolve().then(() => __importStar(require("node:async_hooks")));
+        // Note: using "async_hooks" instead of "node:async_hooks" to avoid webpack fallback problems.
+        const { AsyncLocalStorage } = await Promise.resolve().then(() => __importStar(require("async_hooks")));
         runStorage = new AsyncLocalStorage();
     }
     return Promise.resolve();

package/core/getRun.js

@@ -8,7 +8,8 @@ async function ensureLoaded() {
     if (!isNode)
         return Promise.resolve();
     if (!runStorage) {
-        const { AsyncLocalStorage } = await import("node:async_hooks");
+        // Note: using "async_hooks" instead of "node:async_hooks" to avoid webpack fallback problems.
+        const { AsyncLocalStorage } = await import("async_hooks");
         runStorage = new AsyncLocalStorage();
     }
     return Promise.resolve();

package/core/schema/ZodSchema.cjs

@@ -14,12 +14,24 @@ class ZodSchema {
             writable: true,
             value: void 0
         });
+        /**
+         * Use only for typing purposes. The value is always `undefined`.
+         */
         Object.defineProperty(this, "_type", {
             enumerable: true,
             configurable: true,
             writable: true,
             value: void 0
         });
+        /**
+         * Use only for typing purposes. The value is always `undefined`.
+         */
+        Object.defineProperty(this, "_partialType", {
+            enumerable: true,
+            configurable: true,
+            writable: true,
+            value: void 0
+        });
         this.zodSchema = zodSchema;
     }
     validate(data) {

package/core/schema/ZodSchema.d.ts

@@ -1,6 +1,7 @@
 import { z } from "zod";
 import { JsonSchemaProducer } from "./JsonSchemaProducer.js";
 import { Schema } from "./Schema.js";
+import { PartialDeep } from "type-fest";
 export declare function zodSchema<STRUCTURE>(zodSchema: z.Schema<STRUCTURE>): ZodSchema<STRUCTURE>;
 export declare class ZodSchema<STRUCTURE> implements Schema<STRUCTURE>, JsonSchemaProducer {
     readonly zodSchema: z.Schema<STRUCTURE>;
@@ -13,5 +14,14 @@ export declare class ZodSchema<STRUCTURE> implements Schema<STRUCTURE>, JsonSche
         error: unknown;
     };
     getJsonSchema(): unknown;
+    /**
+     * Use only for typing purposes. The value is always `undefined`.
+     */
     readonly _type: STRUCTURE;
+    /**
+     * Use only for typing purposes. The value is always `undefined`.
+     */
+    readonly _partialType: PartialDeep<STRUCTURE, {
+        recurseIntoArrays: true;
+    }>;
 }

package/core/schema/ZodSchema.js

@@ -10,12 +10,24 @@ export class ZodSchema {
             writable: true,
             value: void 0
         });
+        /**
+         * Use only for typing purposes. The value is always `undefined`.
+         */
         Object.defineProperty(this, "_type", {
             enumerable: true,
             configurable: true,
             writable: true,
             value: void 0
         });
+        /**
+         * Use only for typing purposes. The value is always `undefined`.
+         */
+        Object.defineProperty(this, "_partialType", {
+            enumerable: true,
+            configurable: true,
+            writable: true,
+            value: void 0
+        });
         this.zodSchema = zodSchema;
     }
     validate(data) {

package/model-function/generate-text/prompt-template/ContentPart.d.ts

@@ -1,3 +1,4 @@
+/// <reference types="node" />
 export interface TextPart {
     type: "text";
     /**
@@ -8,9 +9,9 @@ export interface TextPart {
 export interface ImagePart {
     type: "image";
     /**
-     * Image data. Can either be a base64-encoded string or a Uint8Array.
+     * Image data. Can either be a base64-encoded string, a Uint8Array, or a Buffer.
      */
-    image: string | Uint8Array;
+    image: string | Uint8Array | Buffer;
     /**
      * Optional mime type of the image.
      */
@@ -27,5 +28,5 @@ export interface ToolResponsePart {
     id: string;
     response: unknown;
 }
-export declare function getImageAsBase64(image: string | Uint8Array): string;
+export declare function getImageAsBase64(image: string | Uint8Array | Buffer): string;
 export declare function validateContentIsString(content: string | unknown, prompt: unknown): string;

package/model-provider/stability/StabilityImageGenerationModel.d.ts

@@ -69,9 +69,9 @@ export declare class StabilityImageGenerationModel extends AbstractModel<Stabili
     doGenerateImages(prompt: StabilityImageGenerationPrompt, callOptions: FunctionCallOptions): Promise<{
         rawResponse: {
             artifacts: {
+                base64: string;
                 finishReason: "ERROR" | "SUCCESS" | "CONTENT_FILTERED";
                 seed: number;
-                base64: string;
             }[];
         };
         base64Images: string[];
@@ -86,25 +86,25 @@ declare const stabilityImageGenerationResponseSchema: z.ZodObject<{
         seed: z.ZodNumber;
         finishReason: z.ZodEnum<["SUCCESS", "ERROR", "CONTENT_FILTERED"]>;
     }, "strip", z.ZodTypeAny, {
+        base64: string;
         finishReason: "ERROR" | "SUCCESS" | "CONTENT_FILTERED";
         seed: number;
-        base64: string;
     }, {
+        base64: string;
         finishReason: "ERROR" | "SUCCESS" | "CONTENT_FILTERED";
         seed: number;
-        base64: string;
     }>, "many">;
 }, "strip", z.ZodTypeAny, {
     artifacts: {
+        base64: string;
         finishReason: "ERROR" | "SUCCESS" | "CONTENT_FILTERED";
         seed: number;
-        base64: string;
     }[];
 }, {
     artifacts: {
+        base64: string;
         finishReason: "ERROR" | "SUCCESS" | "CONTENT_FILTERED";
         seed: number;
-        base64: string;
     }[];
 }>;
 export type StabilityImageGenerationResponse = z.infer<typeof stabilityImageGenerationResponseSchema>;

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "modelfusion",
   "description": "The TypeScript library for building AI applications.",
-  "version": "0.128.0",
+  "version": "0.129.1",
   "author": "Lars Grammel",
   "license": "MIT",
   "keywords": [
@@ -80,7 +80,7 @@
     "@vitest/ui": "1.1.0",
     "eslint": "^8.45.0",
     "eslint-config-prettier": "9.1.0",
-    "msw": "2.0.11",
+    "msw": "2.1.2",
     "type-fest": "4.9.0"
   }
 }