chore: Fix resource group header (#185)

.changeset/two-bottles-juggle.md

@@ -0,0 +1,9 @@
+---
+'@sap-ai-sdk/foundation-models': minor
+'@sap-ai-sdk/orchestration': minor
+'@sap-ai-sdk/langchain': minor
+'@sap-ai-sdk/ai-api': minor
+'@sap-ai-sdk/core': minor
+---
+
+[Fixed Issue] Fix sending the correct resource group headers when custom resource group is set.

packages/ai-api/src/utils/deployment-resolver.ts

@@ -48,6 +48,17 @@ export type ModelDeployment<ModelNameT = string> =
   | ModelNameT
   | ((ModelConfig<ModelNameT> | DeploymentIdConfig) & ResourceGroupConfig);
 
+/**
+ * @internal
+ */
+export function getResourceGroup(
+  modelDeployment: ModelDeployment
+): string | undefined {
+  return typeof modelDeployment === 'object'
+    ? modelDeployment.resourceGroup
+    : undefined;
+}
+
 /**
  * Type guard to check if the given deployment configuration is a deployment ID configuration.
  * @param modelDeployment - Configuration to check.

packages/core/src/http-client.test.ts

@@ -36,4 +36,31 @@ describe('http-client', () => {
     expect(res.status).toBe(200);
     expect(res.data).toEqual(mockPromptResponse);
   }, 10000);
+
+  it('should execute a request to the AI Core service with a custom resource group', async () => {
+    const mockPrompt = { prompt: 'some test prompt' };
+    const mockPromptResponse = { completion: 'some test completion' };
+
+    const scope = nock(aiCoreDestination.url, {
+      reqheaders: {
+        'ai-resource-group': 'custom-resource-group',
+        'ai-client-type': 'AI SDK JavaScript'
+      }
+    })
+      .post('/v2/some/endpoint', mockPrompt)
+      .query({ 'api-version': 'mock-api-version' })
+      .reply(200, mockPromptResponse);
+
+    const res = await executeRequest(
+      {
+        url: '/some/endpoint',
+        apiVersion: 'mock-api-version',
+        resourceGroup: 'custom-resource-group'
+      },
+      mockPrompt
+    );
+    expect(scope.isDone()).toBe(true);
+    expect(res.status).toBe(200);
+    expect(res.data).toEqual(mockPromptResponse);
+  });
 });

packages/core/src/http-client.ts

@@ -30,11 +30,14 @@ export interface EndpointOptions {
    * The specific endpoint to call.
    */
   url: string;
-
   /**
    * The API version to use.
    */
   apiVersion?: string;
+  /**
+   * The resource group to use.
+   */
+  resourceGroup?: string;
 }
 /**
  * Executes a request to the AI Core service.
@@ -49,10 +52,10 @@ export async function executeRequest(
   requestConfig?: CustomRequestConfig
 ): Promise<HttpResponse> {
   const aiCoreDestination = await getAiCoreDestination();
-  const { url, apiVersion } = endpointOptions;
+  const { url, apiVersion, resourceGroup = 'default' } = endpointOptions;
 
   const mergedRequestConfig = {
-    ...mergeWithDefaultRequestConfig(apiVersion, requestConfig),
+    ...mergeWithDefaultRequestConfig(apiVersion, resourceGroup, requestConfig),
     data: JSON.stringify(data)
   };
 
@@ -69,13 +72,14 @@ export async function executeRequest(
 
 function mergeWithDefaultRequestConfig(
   apiVersion?: string,
+  resourceGroup?: string,
   requestConfig?: CustomRequestConfig
 ): HttpRequestConfig {
   const defaultConfig: HttpRequestConfig = {
     method: 'post',
     headers: {
       'content-type': 'application/json',
-      'ai-resource-group': 'default',
+      'ai-resource-group': resourceGroup,
       'ai-client-type': 'AI SDK JavaScript'
     },
     params: apiVersion ? { 'api-version': apiVersion } : {}

packages/foundation-models/src/azure-openai/azure-openai-chat-client.test.ts

@@ -1,6 +1,7 @@
 import nock from 'nock';
 import {
   mockClientCredentialsGrantCall,
+  mockDeploymentsList,
   mockInference,
   parseMockResponse
 } from '../../../../test-util/mock-http.js';
@@ -75,4 +76,55 @@ describe('Azure OpenAI chat client', () => {
 
     await expect(client.run(prompt)).rejects.toThrow('status code 400');
   });
+
+  it('executes a request with the custom resource group', async () => {
+    const customChatCompletionEndpoint = {
+      url: 'inference/deployments/1234/chat/completions',
+      apiVersion,
+      resourceGroup: 'custom-resource-group'
+    };
+
+    const mockResponse =
+      parseMockResponse<AzureOpenAiCreateChatCompletionResponse>(
+        'foundation-models',
+        'azure-openai-chat-completion-success-response.json'
+      );
+
+    const prompt = {
+      messages: [
+        {
+          role: 'user' as const,
+          content: 'Where is the deepest place on earth located'
+        }
+      ]
+    };
+
+    mockDeploymentsList(
+      {
+        scenarioId: 'foundation-models',
+        resourceGroup: 'custom-resource-group',
+        executableId: 'azure-openai'
+      },
+      { id: '1234', model: { name: 'gpt-4o', version: 'latest' } }
+    );
+
+    mockInference(
+      {
+        data: prompt
+      },
+      {
+        data: mockResponse,
+        status: 200
+      },
+      customChatCompletionEndpoint
+    );
+
+    const clientWithResourceGroup = new AzureOpenAiChatClient({
+      modelName: 'gpt-4o',
+      resourceGroup: 'custom-resource-group'
+    });
+
+    const response = await clientWithResourceGroup.run(prompt);
+    expect(response.data).toEqual(mockResponse);
+  });
 });

packages/foundation-models/src/azure-openai/azure-openai-chat-client.ts

@@ -1,6 +1,7 @@
 import { type CustomRequestConfig, executeRequest } from '@sap-ai-sdk/core';
 import {
   getDeploymentId,
+  getResourceGroup,
   type ModelDeployment
 } from '@sap-ai-sdk/ai-api/internal.js';
 import type { AzureOpenAiCreateChatCompletionRequest } from './client/inference/schema/index.js';
@@ -31,10 +32,12 @@ export class AzureOpenAiChatClient {
       this.modelDeployment,
       'azure-openai'
     );
+    const resourceGroup = getResourceGroup(this.modelDeployment);
     const response = await executeRequest(
       {
         url: `/inference/deployments/${deploymentId}/chat/completions`,
-        apiVersion
+        apiVersion,
+        resourceGroup
       },
       data,
       requestConfig

packages/foundation-models/src/azure-openai/azure-openai-embedding-client.ts

@@ -1,6 +1,7 @@
 import { type CustomRequestConfig, executeRequest } from '@sap-ai-sdk/core';
 import {
   getDeploymentId,
+  getResourceGroup,
   type ModelDeployment
 } from '@sap-ai-sdk/ai-api/internal.js';
 import { AzureOpenAiEmbeddingResponse } from './azure-openai-embedding-response.js';
@@ -34,8 +35,13 @@ export class AzureOpenAiEmbeddingClient {
       this.modelDeployment,
       'azure-openai'
     );
+    const resourceGroup = getResourceGroup(this.modelDeployment);
     const response = await executeRequest(
-      { url: `/inference/deployments/${deploymentId}/embeddings`, apiVersion },
+      {
+        url: `/inference/deployments/${deploymentId}/embeddings`,
+        apiVersion,
+        resourceGroup
+      },
       data,
       requestConfig
     );

packages/langchain/src/openai/chat.ts

@@ -3,7 +3,6 @@ import { BaseMessage } from '@langchain/core/messages';
 import type { ChatResult } from '@langchain/core/outputs';
 import { AzureOpenAiChatClient as AzureOpenAiChatClientBase } from '@sap-ai-sdk/foundation-models';
 import { BaseChatModel } from '@langchain/core/language_models/chat_models';
-import { AzureOpenAiChatModel } from '@sap-ai-sdk/core';
 import { mapLangchainToAiClient, mapOutputToChatResult } from './util.js';
 import type {
   AzureOpenAiChatCallOptions,
@@ -13,13 +12,7 @@ import type {
 /**
  * LangChain chat client for Azure OpenAI consumption on SAP BTP.
  */
-export class AzureOpenAiChatClient
-  extends BaseChatModel<AzureOpenAiChatCallOptions>
-  implements AzureOpenAiChatModelParams
-{
-  modelName: AzureOpenAiChatModel;
-  modelVersion?: string;
-  resourceGroup?: string;
+export class AzureOpenAiChatClient extends BaseChatModel<AzureOpenAiChatCallOptions> {
   temperature?: number | null;
   top_p?: number | null;
   logit_bias?: Record<string, any> | null | undefined;
@@ -33,9 +26,6 @@ export class AzureOpenAiChatClient
   constructor(fields: AzureOpenAiChatModelParams) {
     super(fields);
     this.openAiChatClient = new AzureOpenAiChatClientBase(fields);
-    this.modelName = fields.modelName;
-    this.modelVersion = fields.modelVersion;
-    this.resourceGroup = fields.resourceGroup;
     this.temperature = fields.temperature;
     this.top_p = fields.top_p;
     this.logit_bias = fields.logit_bias;

packages/langchain/src/openai/embedding.ts

@@ -9,10 +9,7 @@ import { AzureOpenAiEmbeddingModelParams } from './types.js';
 /**
  * LangChain embedding client for Azure OpenAI consumption on SAP BTP.
  */
-export class AzureOpenAiEmbeddingClient
-  extends Embeddings
-  implements AzureOpenAiEmbeddingModelParams
-{
+export class AzureOpenAiEmbeddingClient extends Embeddings {
   modelName: AzureOpenAiEmbeddingModel;
   modelVersion?: string;
   resourceGroup?: string;

packages/orchestration/src/orchestration-client.test.ts

@@ -12,7 +12,7 @@ import {
 } from './orchestration-client.js';
 import { buildAzureContentFilter } from './orchestration-filter-utility.js';
 import { OrchestrationResponse } from './orchestration-response.js';
-import { OrchestrationModuleConfig } from './orchestration-types.js';
+import { OrchestrationModuleConfig, Prompt } from './orchestration-types.js';
 
 describe('orchestration service client', () => {
   beforeEach(() => {
@@ -220,4 +220,58 @@ describe('orchestration service client', () => {
     );
     expect(response.data).toEqual(mockResponse);
   });
+
+  it('executes a request with the custom resource group', async () => {
+    const prompt: Prompt = {
+      messagesHistory: [
+        {
+          role: 'user',
+          content: 'Where is the deepest place on earth located'
+        }
+      ]
+    };
+
+    const config: OrchestrationModuleConfig = {
+      llm: {
+        model_name: 'gpt-4o',
+        model_params: {}
+      },
+      templating: {
+        template: [{ role: 'user', content: "What's my name?" }]
+      }
+    };
+
+    const customChatCompletionEndpoint = {
+      url: 'inference/deployments/1234/completion',
+      resourceGroup: 'custom-resource-group'
+    };
+
+    const mockResponse = parseMockResponse<CompletionPostResponse>(
+      'orchestration',
+      'orchestration-chat-completion-message-history.json'
+    );
+
+    mockDeploymentsList(
+      { scenarioId: 'orchestration', resourceGroup: 'custom-resource-group' },
+      { id: '1234', model: { name: 'gpt-4o', version: 'latest' } }
+    );
+
+    mockInference(
+      {
+        data: constructCompletionPostRequest(config, prompt)
+      },
+      {
+        data: mockResponse,
+        status: 200
+      },
+      customChatCompletionEndpoint
+    );
+
+    const clientWithResourceGroup = new OrchestrationClient(config, {
+      resourceGroup: 'custom-resource-group'
+    });
+
+    const response = await clientWithResourceGroup.chatCompletion(prompt);
+    expect(response.data).toEqual(mockResponse);
+  });
 });

packages/orchestration/src/orchestration-client.ts

@@ -39,7 +39,8 @@ export class OrchestrationClient {
 
     const response = await executeRequest(
       {
-        url: `/inference/deployments/${deploymentId}/completion`
+        url: `/inference/deployments/${deploymentId}/completion`,
+        resourceGroup: this.deploymentConfig?.resourceGroup
       },
       body,
       requestConfig

test-util/mock-http.ts

@@ -99,11 +99,11 @@ export function mockInference(
   },
   endpoint: EndpointOptions = mockEndpoint
 ): nock.Scope {
-  const { url, apiVersion } = endpoint;
+  const { url, apiVersion, resourceGroup = 'default' } = endpoint;
   const destination = getMockedAiCoreDestination();
   return nock(destination.url, {
     reqheaders: {
-      'ai-resource-group': 'default',
+      'ai-resource-group': resourceGroup,
       authorization: `Bearer ${destination.authTokens?.[0].value}`
     }
   })