feat (providers): support custom fetch implementations (#1955)

Author: lgrammel

.changeset/nasty-goats-hammer.md

@@ -0,0 +1,10 @@
+---
+'@ai-sdk/provider-utils': patch
+'@ai-sdk/anthropic': patch
+'@ai-sdk/mistral': patch
+'@ai-sdk/google': patch
+'@ai-sdk/openai': patch
+'@ai-sdk/azure': patch
+---
+
+feat (providers): support custom fetch implementations

content/providers/01-ai-sdk-providers/01-openai.mdx

@@ -67,6 +67,13 @@ You can use the following optional settings to customize the OpenAI provider ins
 
   Custom headers to include in the requests.
 
+- **fetch** _(input: RequestInfo, init?: RequestInit) => Promise<Response>_
+
+  Custom [fetch](https://developer.mozilla.org/en-US/docs/Web/API/fetch) implementation.
+  Defaults to the global `fetch` function.
+  You can use it as a middleware to intercept requests,
+  or to provide a custom fetch implementation for e.g. testing.
+
 - **compatibility** _"strict" | "compatible"_
 
   OpenAI compatibility mode. Should be set to `strict` when using the OpenAI API,

content/providers/01-ai-sdk-providers/02-azure.mdx

@@ -54,6 +54,13 @@ You can use the following optional settings to customize the OpenAI provider ins
   API key that is being send using the `api-key` header.
   It defaults to the `AZURE_API_KEY` environment variable.
 
+- **fetch** _(input: RequestInfo, init?: RequestInit) => Promise<Response>_
+
+  Custom [fetch](https://developer.mozilla.org/en-US/docs/Web/API/fetch) implementation.
+  Defaults to the global `fetch` function.
+  You can use it as a middleware to intercept requests,
+  or to provide a custom fetch implementation for e.g. testing.
+
 ## Language Models
 
 The Azure OpenAI provider instance is a function that you can invoke to create a language model:

content/providers/01-ai-sdk-providers/05-anthropic.mdx

@@ -57,6 +57,13 @@ You can use the following optional settings to customize the Google Generative A
 
   Custom headers to include in the requests.
 
+- **fetch** _(input: RequestInfo, init?: RequestInit) => Promise<Response>_
+
+  Custom [fetch](https://developer.mozilla.org/en-US/docs/Web/API/fetch) implementation.
+  Defaults to the global `fetch` function.
+  You can use it as a middleware to intercept requests,
+  or to provide a custom fetch implementation for e.g. testing.
+
 ## Language Models
 
 You can create models that call the [Anthropic Messages API](https://docs.anthropic.com/claude/reference/messages_post) using the provider instance.

content/providers/01-ai-sdk-providers/10-google-generative-ai.mdx

@@ -57,6 +57,13 @@ You can use the following optional settings to customize the Google Generative A
 
   Custom headers to include in the requests.
 
+- **fetch** _(input: RequestInfo, init?: RequestInit) => Promise<Response>_
+
+  Custom [fetch](https://developer.mozilla.org/en-US/docs/Web/API/fetch) implementation.
+  Defaults to the global `fetch` function.
+  You can use it as a middleware to intercept requests,
+  or to provide a custom fetch implementation for e.g. testing.
+
 ## Language Models
 
 You can create models that call the [Google Generative AI API](https://ai.google.dev/api/rest) using the provider instance.

content/providers/01-ai-sdk-providers/20-mistral.mdx

@@ -58,6 +58,13 @@ You can use the following optional settings to customize the Mistral provider in
 
   Custom headers to include in the requests.
 
+- **fetch** _(input: RequestInfo, init?: RequestInit) => Promise<Response>_
+
+  Custom [fetch](https://developer.mozilla.org/en-US/docs/Web/API/fetch) implementation.
+  Defaults to the global `fetch` function.
+  You can use it as a middleware to intercept requests,
+  or to provide a custom fetch implementation for e.g. testing.
+
 ## Language Models
 
 You can create models that call the [Mistral chat API](https://docs.mistral.ai/api/#operation/createChatCompletion) using provider instance.

examples/ai-core/src/generate-text/anthropic-custom-fetch.ts

@@ -0,0 +1,27 @@
+import { createAnthropic } from '@ai-sdk/anthropic';
+import { generateText } from 'ai';
+import dotenv from 'dotenv';
+
+dotenv.config();
+
+const anthropic = createAnthropic({
+  // example fetch wrapper that logs the URL:
+  fetch: async (url, options) => {
+    console.log(`Fetching ${url}`);
+    const result = await fetch(url, options);
+    console.log(`Fetched ${url}`);
+    console.log();
+    return result;
+  },
+});
+
+async function main() {
+  const result = await generateText({
+    model: anthropic('claude-3-haiku-20240307'),
+    prompt: 'Invent a new holiday and describe its traditions.',
+  });
+
+  console.log(result.text);
+}
+
+main().catch(console.error);

examples/ai-core/src/generate-text/azure-custom-fetch.ts

@@ -0,0 +1,27 @@
+import { createAzure } from '@ai-sdk/azure';
+import { generateText } from 'ai';
+import dotenv from 'dotenv';
+
+dotenv.config();
+
+const azure = createAzure({
+  // example fetch wrapper that logs the URL:
+  fetch: async (url, options) => {
+    console.log(`Fetching ${url}`);
+    const result = await fetch(url, options);
+    console.log(`Fetched ${url}`);
+    console.log();
+    return result;
+  },
+});
+
+async function main() {
+  const result = await generateText({
+    model: azure('v0-gpt-35-turbo'), // use your own deployment
+    prompt: 'Invent a new holiday and describe its traditions.',
+  });
+
+  console.log(result.text);
+}
+
+main().catch(console.error);

examples/ai-core/src/generate-text/google-custom-fetch.ts

@@ -0,0 +1,27 @@
+import { createGoogleGenerativeAI } from '@ai-sdk/google';
+import { generateText } from 'ai';
+import dotenv from 'dotenv';
+
+dotenv.config();
+
+const google = createGoogleGenerativeAI({
+  // example fetch wrapper that logs the URL:
+  fetch: async (url, options) => {
+    console.log(`Fetching ${url}`);
+    const result = await fetch(url, options);
+    console.log(`Fetched ${url}`);
+    console.log();
+    return result;
+  },
+});
+
+async function main() {
+  const result = await generateText({
+    model: google('models/gemini-pro'),
+    prompt: 'Invent a new holiday and describe its traditions.',
+  });
+
+  console.log(result.text);
+}
+
+main().catch(console.error);

examples/ai-core/src/generate-text/mistral-custom-fetch.ts

@@ -0,0 +1,27 @@
+import { createMistral } from '@ai-sdk/mistral';
+import { generateText } from 'ai';
+import dotenv from 'dotenv';
+
+dotenv.config();
+
+const mistral = createMistral({
+  // example fetch wrapper that logs the URL:
+  fetch: async (url, options) => {
+    console.log(`Fetching ${url}`);
+    const result = await fetch(url, options);
+    console.log(`Fetched ${url}`);
+    console.log();
+    return result;
+  },
+});
+
+async function main() {
+  const result = await generateText({
+    model: mistral('open-mistral-7b'),
+    prompt: 'Invent a new holiday and describe its traditions.',
+  });
+
+  console.log(result.text);
+}
+
+main().catch(console.error);

examples/ai-core/src/generate-text/openai-custom-fetch.ts

@@ -0,0 +1,27 @@
+import { createOpenAI } from '@ai-sdk/openai';
+import { generateText } from 'ai';
+import dotenv from 'dotenv';
+
+dotenv.config();
+
+const openai = createOpenAI({
+  // example fetch wrapper that logs the URL:
+  fetch: async (url, options) => {
+    console.log(`Fetching ${url}`);
+    const result = await fetch(url, options);
+    console.log(`Fetched ${url}`);
+    console.log();
+    return result;
+  },
+});
+
+async function main() {
+  const result = await generateText({
+    model: openai('gpt-3.5-turbo'),
+    prompt: 'Invent a new holiday and describe its traditions.',
+  });
+
+  console.log(result.text);
+}
+
+main().catch(console.error);

packages/anthropic/src/anthropic-messages-language-model.ts

@@ -25,6 +25,7 @@ type AnthropicMessagesConfig = {
   provider: string;
   baseURL: string;
   headers: () => Record<string, string | undefined>;
+  fetch?: typeof fetch;
 };
 
 export class AnthropicMessagesLanguageModel implements LanguageModelV1 {
@@ -163,6 +164,7 @@ export class AnthropicMessagesLanguageModel implements LanguageModelV1 {
         anthropicMessagesResponseSchema,
       ),
       abortSignal: options.abortSignal,
+      fetch: this.config.fetch,
     });
 
     const { messages: rawPrompt, ...rawSettings } = args;
@@ -222,6 +224,7 @@ export class AnthropicMessagesLanguageModel implements LanguageModelV1 {
         anthropicMessagesChunkSchema,
       ),
       abortSignal: options.abortSignal,
+      fetch: this.config.fetch,
     });
 
     const { messages: rawPrompt, ...rawSettings } = args;

packages/anthropic/src/anthropic-provider.ts

@@ -54,6 +54,12 @@ Custom headers to include in the requests.
      */
   headers?: Record<string, string>;
 
+  /**
+Custom fetch implementation. You can use it as a middleware to intercept requests,
+or to provide a custom fetch implementation for e.g. testing.
+    */
+  fetch?: typeof fetch;
+
   generateId?: () => string;
 }
 
@@ -86,6 +92,7 @@ export function createAnthropic(
       provider: 'anthropic.messages',
       baseURL,
       headers: getHeaders,
+      fetch: options.fetch,
     });
 
   const provider = function (

packages/azure/src/azure-openai-provider.ts

@@ -29,6 +29,12 @@ Name of the Azure OpenAI resource.
 API key for authenticating requests.
      */
   apiKey?: string;
+
+  /**
+Custom fetch implementation. You can use it as a middleware to intercept requests,
+or to provide a custom fetch implementation for e.g. testing.
+    */
+  fetch?: typeof fetch;
 }
 
 /**
@@ -63,6 +69,7 @@ export function createAzure(
       url: ({ path, modelId }) =>
         `https://${getResourceName()}.openai.azure.com/openai/deployments/${modelId}${path}?api-version=2024-05-01-preview`,
       compatibility: 'compatible',
+      fetch: options.fetch,
     });
 
   const provider = function (

packages/google/src/google-generative-ai-language-model.ts

@@ -26,6 +26,7 @@ type GoogleGenerativeAIConfig = {
   baseURL: string;
   headers: () => Record<string, string | undefined>;
   generateId: () => string;
+  fetch?: typeof fetch;
 };
 
 export class GoogleGenerativeAILanguageModel implements LanguageModelV1 {
@@ -156,6 +157,7 @@ export class GoogleGenerativeAILanguageModel implements LanguageModelV1 {
       failedResponseHandler: googleFailedResponseHandler,
       successfulResponseHandler: createJsonResponseHandler(responseSchema),
       abortSignal: options.abortSignal,
+      fetch: this.config.fetch,
     });
 
     const { contents: rawPrompt, ...rawSettings } = args;
@@ -197,6 +199,7 @@ export class GoogleGenerativeAILanguageModel implements LanguageModelV1 {
       failedResponseHandler: googleFailedResponseHandler,
       successfulResponseHandler: createEventSourceResponseHandler(chunkSchema),
       abortSignal: options.abortSignal,
+      fetch: this.config.fetch,
     });
 
     const { contents: rawPrompt, ...rawSettings } = args;

packages/google/src/google-provider.ts

@@ -1,4 +1,8 @@
-import { Google } from './google-facade';
+import {
+  generateId,
+  loadApiKey,
+  withoutTrailingSlash,
+} from '@ai-sdk/provider-utils';
 import { GoogleGenerativeAILanguageModel } from './google-generative-ai-language-model';
 import {
   GoogleGenerativeAIModelId,
@@ -48,6 +52,12 @@ Custom headers to include in the requests.
      */
   headers?: Record<string, string>;
 
+  /**
+Custom fetch implementation. You can use it as a middleware to intercept requests,
+or to provide a custom fetch implementation for e.g. testing.
+    */
+  fetch?: typeof fetch;
+
   generateId?: () => string;
 }
 
@@ -57,7 +67,30 @@ Create a Google Generative AI provider instance.
 export function createGoogleGenerativeAI(
   options: GoogleGenerativeAIProviderSettings = {},
 ): GoogleGenerativeAIProvider {
-  const google = new Google(options);
+  const baseURL =
+    withoutTrailingSlash(options.baseURL ?? options.baseUrl) ??
+    'https://generativelanguage.googleapis.com/v1beta';
+
+  const getHeaders = () => ({
+    'x-goog-api-key': loadApiKey({
+      apiKey: options.apiKey,
+      environmentVariableName: 'GOOGLE_GENERATIVE_AI_API_KEY',
+      description: 'Google Generative AI',
+    }),
+    ...options.headers,
+  });
+
+  const createChatModel = (
+    modelId: GoogleGenerativeAIModelId,
+    settings: GoogleGenerativeAISettings = {},
+  ) =>
+    new GoogleGenerativeAILanguageModel(modelId, settings, {
+      provider: 'google.generative-ai',
+      baseURL,
+      headers: getHeaders,
+      generateId: options.generateId ?? generateId,
+      fetch: options.fetch,
+    });
 
   const provider = function (
     modelId: GoogleGenerativeAIModelId,
@@ -69,11 +102,11 @@ export function createGoogleGenerativeAI(
       );
     }
 
-    return google.chat(modelId, settings);
+    return createChatModel(modelId, settings);
   };
 
-  provider.chat = google.chat.bind(google);
-  provider.generativeAI = google.generativeAI.bind(google);
+  provider.chat = createChatModel;
+  provider.generativeAI = createChatModel;
 
   return provider as GoogleGenerativeAIProvider;
 }