feat: Add message history option (#27)

* messages_history + tests * fix: Changes from lint * fix openai tests * review * fix: Changes from lint --------- Co-authored-by: cloud-sdk-js <[email protected]>
SAP · Jul 19, 2024 · 3d1926a · 3d1926a
1 parent 8814496
commit 3d1926a
Show file tree

Hide file tree

Showing 8 changed files with 183 additions and 129 deletions.
diff --git a/packages/gen-ai-hub/src/client/openai/openai-client.test.ts b/packages/gen-ai-hub/src/client/openai/openai-client.test.ts
@@ -1,14 +1,14 @@
-import fs from 'fs';
-import path from 'path';
 import nock from 'nock';
 import { HttpDestination } from '@sap-cloud-sdk/connectivity';
 import { mockGetAiCoreDestination } from '../../../test-util/mock-context.js';
 import {
-  BaseLlmParameters,
   BaseLlmParametersWithDeploymentId,
   EndpointOptions
 } from '../../core/http-client.js';
-import { mockInference } from '../../../test-util/mock-http.js';
+import {
+  mockInference,
+  parseMockResponse
+} from '../../../test-util/mock-http.js';
 import { OpenAiClient } from './openai-client.js';
 import {
   OpenAiChatCompletionOutput,
@@ -20,17 +20,15 @@ import {
 
 describe('openai client', () => {
   let destination: HttpDestination;
-  let deploymentConfig: BaseLlmParameters;
+  const deploymentConfiguration: BaseLlmParametersWithDeploymentId = {
+    deploymentId: 'deployment-id'
+  };
   let chatCompletionEndpoint: EndpointOptions;
   let embeddingsEndpoint: EndpointOptions;
 
   beforeAll(() => {
     destination = mockGetAiCoreDestination();
-    deploymentConfig = {
-      deploymentConfiguration: {
-        deploymentId: 'deployment-id'
-      } as BaseLlmParametersWithDeploymentId
-    };
+
     chatCompletionEndpoint = {
       url: 'chat/completions',
       apiVersion: '2024-02-01'
@@ -58,60 +56,47 @@ describe('openai client', () => {
       };
       const request: OpenAiChatCompletionParameters = {
         ...prompt,
-        ...deploymentConfig
+        deploymentConfiguration
       };
-      const mockResponse = fs.readFileSync(
-        path.join(
-          'test-util',
-          'mock-data',
-          'openai',
-          'openai-chat-completion-success-response.json'
-        ),
-        'utf8'
+      const mockResponse = parseMockResponse<OpenAiChatCompletionOutput>(
+        'openai',
+        'openai-chat-completion-success-response.json'
       );
 
       mockInference(
         {
           data: request
         },
         {
-          data: JSON.parse(mockResponse),
+          data: mockResponse,
           status: 200
         },
         destination,
         chatCompletionEndpoint
       );
 
-      const result: OpenAiChatCompletionOutput =
-        await new OpenAiClient().chatCompletion(request);
-      const expectedResponse: OpenAiChatCompletionOutput =
-        JSON.parse(mockResponse);
-
-      expect(result).toEqual(expectedResponse);
+      expect(new OpenAiClient().chatCompletion(request)).resolves.toEqual(
+        mockResponse
+      );
     });
 
     it('throws on bad request', async () => {
       const prompt = { messages: [] };
       const request: OpenAiChatCompletionParameters = {
         ...prompt,
-        ...deploymentConfig
+        deploymentConfiguration
       };
-      const mockResponse = fs.readFileSync(
-        path.join(
-          'test-util',
-          'mock-data',
-          'openai',
-          'openai-error-response.json'
-        ),
-        'utf8'
+      const mockResponse = parseMockResponse(
+        'openai',
+        'openai-error-response.json'
       );
 
       mockInference(
         {
           data: request
         },
         {
-          data: JSON.parse(mockResponse),
+          data: mockResponse,
           status: 400
         },
         destination,
@@ -129,66 +114,54 @@ describe('openai client', () => {
       const prompt = { input: ['AI is fascinating'] };
       const request: OpenAiEmbeddingParameters = {
         ...prompt,
-        ...deploymentConfig
+        deploymentConfiguration
       };
-      const mockResponse = fs.readFileSync(
-        path.join(
-          'test-util',
-          'mock-data',
-          'openai',
-          'openai-embeddings-success-response.json'
-        ),
-        'utf8'
+      const mockResponse = parseMockResponse<OpenAiEmbeddingOutput>(
+        'openai',
+        'openai-embeddings-success-response.json'
       );
 
       mockInference(
         {
           data: request
         },
         {
-          data: JSON.parse(mockResponse),
+          data: mockResponse,
           status: 200
         },
         destination,
         embeddingsEndpoint
       );
 
-      const result: OpenAiEmbeddingOutput = await new OpenAiClient().embeddings(
-        request
+      expect(new OpenAiClient().embeddings(request)).resolves.toEqual(
+        mockResponse
       );
-      const expectedResponse: OpenAiEmbeddingOutput = JSON.parse(mockResponse);
-      expect(result).toEqual(expectedResponse);
     });
 
     it('throws on bad request', async () => {
       const prompt = { input: [] };
       const request: OpenAiEmbeddingParameters = {
         ...prompt,
-        ...deploymentConfig
+        deploymentConfiguration
       };
-      const mockResponse = fs.readFileSync(
-        path.join(
-          'test-util',
-          'mock-data',
-          'openai',
-          'openai-error-response.json'
-        ),
-        'utf8'
+      const mockResponse = parseMockResponse(
+        'openai',
+        'openai-error-response.json'
       );
 
       mockInference(
         {
           data: request
         },
         {
-          data: JSON.parse(mockResponse),
+          data: mockResponse,
           status: 400
         },
         destination,
         embeddingsEndpoint
       );
 
-      await expect(new OpenAiClient().embeddings(request)).rejects.toThrow();
+      expect(new OpenAiClient().embeddings(request)).rejects.toThrow();
     });
   });
 });
diff --git a/packages/gen-ai-hub/src/orchestration/api/schema/chat-message.ts b/packages/gen-ai-hub/src/orchestration/api/schema/chat-message.ts
@@ -8,6 +8,6 @@
  * Representation of the 'ChatMessage' schema.
  */
 export type ChatMessage = {
-  role: string;
+  role: 'user' | 'assistant' | 'system';
   content: string;
 } & Record<string, any>;
diff --git a/packages/gen-ai-hub/src/orchestration/orchestration-client.test.ts b/packages/gen-ai-hub/src/orchestration/orchestration-client.test.ts
@@ -1,25 +1,33 @@
-import fs from 'fs';
-import path from 'path';
 import nock from 'nock';
 import { HttpDestination } from '@sap-cloud-sdk/connectivity';
-import { BaseLlmParametersWithDeploymentId } from '../core/index.js';
 import { mockGetAiCoreDestination } from '../../test-util/mock-context.js';
-import { mockInference } from '../../test-util/mock-http.js';
+import { mockInference, parseMockResponse } from '../../test-util/mock-http.js';
+import { BaseLlmParametersWithDeploymentId } from '../core/index.js';
 import {
   GenAiHubClient,
   GenAiHubCompletionParameters
 } from './orchestration-client.js';
-import { CompletionPostResponse, ModuleConfigs } from './api/index.js';
+import {
+  CompletionPostResponse,
+  LLMModuleConfig,
+  ModuleConfigs
+} from './api/index.js';
 
 describe('GenAiHubClient', () => {
   let destination: HttpDestination;
-  let deploymentConfiguration: BaseLlmParametersWithDeploymentId;
   let client: GenAiHubClient;
+  const deploymentConfiguration: BaseLlmParametersWithDeploymentId = {
+    deploymentId: 'deployment-id'
+  };
+  const llm_module_config: LLMModuleConfig = {
+    model_name: 'gpt-35-turbo-16k',
+    model_params: {
+      max_tokens: 50,
+      temperature: 0.1
+    }
+  };
 
   beforeAll(() => {
-    deploymentConfiguration = {
-      deploymentId: 'deployment-id'
-    };
     destination = mockGetAiCoreDestination();
     client = new GenAiHubClient();
   });
@@ -28,92 +36,85 @@ describe('GenAiHubClient', () => {
     nock.cleanAll();
   });
 
-  it(' calls chatCompletion and parses response', async () => {
+  it('calls chatCompletion with minimum configuration and parses response', async () => {
     const module_configurations: ModuleConfigs = {
       templating_module_config: {
         template: [{ role: 'user', content: 'Hello!' }]
       },
-      llm_module_config: {
-        model_name: 'gpt-35-turbo-16k',
-        model_params: {
-          max_tokens: 50,
-          temperature: 0.1
-        }
-      }
+      llm_module_config
     };
     const request: GenAiHubCompletionParameters = {
       deploymentConfiguration,
       orchestration_config: { module_configurations }
     };
 
-    const mockResponse = fs.readFileSync(
-      path.join(
-        'test-util',
-        'mock-data',
-        'orchestration',
-        'genaihub-chat-completion-success-response.json'
-      ),
-      'utf-8'
+    const mockResponse = parseMockResponse<CompletionPostResponse>(
+      'orchestration',
+      'genaihub-chat-completion-success-response.json'
     );
 
     mockInference(
       {
         data: { ...request, input_params: {} }
       },
       {
-        data: JSON.parse(mockResponse),
+        data: mockResponse,
         status: 200
       },
       destination,
       {
         url: 'completion'
       }
     );
-    const result = await client.chatCompletion(request);
-    const expectedResponse: CompletionPostResponse = JSON.parse(mockResponse);
-    expect(result).toEqual(expectedResponse);
+    expect(client.chatCompletion(request)).resolves.toEqual(mockResponse);
   });
 
-  it('throws error for incorrect input parameters', async () => {
+  it('sends message history together with templating config', async () => {
     const module_configurations: ModuleConfigs = {
       templating_module_config: {
-        template: [{ role: 'actor', content: 'Hello' }]
+        template: [{ role: 'user', content: "What's my name?" }]
       },
-      llm_module_config: {
-        model_name: 'gpt-35-turbo-16k',
-        model_params: {
-          max_tokens: 50,
-          temperature: 0.1
-        }
-      }
+      llm_module_config
     };
     const request: GenAiHubCompletionParameters = {
       deploymentConfiguration,
-      orchestration_config: { module_configurations }
+      orchestration_config: { module_configurations },
+      messages_history: [
+        {
+          role: 'system',
+          content:
+            'You are a helpful assistant who remembers all details the user shares with you.'
+        },
+        {
+          role: 'user',
+          content: 'Hi! Im Bob'
+        },
+        {
+          role: 'assistant',
+          content:
+            "Hi Bob, nice to meet you! I'm an AI assistant. I'll remember that your name is Bob as we continue our conversation."
+        }
+      ]
     };
-    const mockResponse = fs.readFileSync(
-      path.join(
-        'test-util',
-        'mock-data',
-        'orchestration',
-        'genaihub-error-response.json'
-      ),
-      'utf-8'
-    );
 
+    const mockResponse = parseMockResponse<CompletionPostResponse>(
+      'orchestration',
+      'genaihub-chat-completion-message-history.json'
+    );
     mockInference(
       {
         data: { ...request, input_params: {} }
       },
       {
-        data: JSON.parse(mockResponse),
-        status: 400
+        data: mockResponse,
+        status: 200
       },
       destination,
       {
         url: 'completion'
       }
     );
-    await expect(client.chatCompletion(request)).rejects.toThrow();
+
+    expect(client.chatCompletion(request)).resolves.toEqual(mockResponse);
   });
 });
diff --git a/packages/gen-ai-hub/src/orchestration/orchestration-client.ts b/packages/gen-ai-hub/src/orchestration/orchestration-client.ts
@@ -12,7 +12,7 @@ import {
  * Input Parameters for GenAI hub chat completion.
  */
 export type GenAiHubCompletionParameters = BaseLlmParameters &
-  Pick<CompletionPostRequest, 'orchestration_config'>;
+  Pick<CompletionPostRequest, 'orchestration_config' | 'messages_history'>;
 
 /**
  * Get the orchestration client.