AI MCP Gateway

/** * Unit tests for EmbeddingService */ import { describe, it, expect, vi, beforeEach, afterEach } from 'vitest'; import { EmbeddingService } from '../EmbeddingService'; // Mock dependencies vi.mock('../../../logging/logger', () => ({ logger: { debug: vi.fn(), info: vi.fn(), warn: vi.fn(), error: vi.fn(), }, })); vi.mock('../../../cache/redis', () => ({ redisCache: { get: vi.fn(), set: vi.fn(), }, })); vi.mock('../../../config/provider-manager', () => ({ providerManager: { getApiKey: vi.fn().mockResolvedValue('test-api-key'), }, })); vi.mock('../../../config/env', () => ({ env: { EMBEDDING_PROVIDER: 'openai', EMBEDDING_MODEL_ID: 'text-embedding-3-small', OPENAI_API_KEY: 'test-key', }, })); describe('EmbeddingService', () => { let service: EmbeddingService; beforeEach(() => { service = new EmbeddingService({ provider: 'openai', modelId: 'text-embedding-3-small', dimension: 1536, enableCache: false, // Disable cache for tests }); vi.clearAllMocks(); }); afterEach(() => { vi.resetAllMocks(); }); describe('constructor', () => { it('should initialize with default config', () => { const defaultService = new EmbeddingService(); expect(defaultService).toBeDefined(); }); it('should accept custom config', () => { const customService = new EmbeddingService({ provider: 'ollama', modelId: 'custom-model', dimension: 384, }); expect(customService).toBeDefined(); }); }); describe('cosineSimilarity', () => { it('should calculate similarity correctly', () => { const a = [1, 0, 0]; const b = [1, 0, 0]; const similarity = EmbeddingService.cosineSimilarity(a, b); expect(similarity).toBeCloseTo(1.0); }); it('should handle orthogonal vectors', () => { const a = [1, 0, 0]; const b = [0, 1, 0]; const similarity = EmbeddingService.cosineSimilarity(a, b); expect(similarity).toBeCloseTo(0.0); }); it('should handle opposite vectors', () => { const a = [1, 0, 0]; const b = [-1, 0, 0]; const similarity = EmbeddingService.cosineSimilarity(a, b); expect(similarity).toBeCloseTo(-1.0); }); it('should throw on dimension mismatch', () => { const a = [1, 0]; const b = [1, 0, 0]; expect(() => EmbeddingService.cosineSimilarity(a, b)).toThrow(); }); }); describe('formatForPostgres', () => { it('should format embedding for postgres', () => { const embedding = [0.1, 0.2, 0.3]; const formatted = EmbeddingService.formatForPostgres(embedding); expect(formatted).toBe('[0.1,0.2,0.3]'); }); it('should handle empty array', () => { const embedding: number[] = []; const formatted = EmbeddingService.formatForPostgres(embedding); expect(formatted).toBe('[]'); }); it('should handle single element', () => { const embedding = [0.5]; const formatted = EmbeddingService.formatForPostgres(embedding); expect(formatted).toBe('[0.5]'); }); }); describe('getEmbedding', () => { it('should reject empty text', async () => { await expect(service.getEmbedding('')).rejects.toThrow(); }); it('should normalize whitespace', async () => { // This test would need actual API mocking // For now, just check it doesn't throw on valid input const text = ' test text \n\n\n '; // Would test: normalized version sent to API }); }); describe('getBatchEmbeddings', () => { it('should handle empty batch', async () => { const result = await service.getBatchEmbeddings({ texts: [] }); expect(result).toEqual([]); }); it('should process multiple texts', async () => { // This would need API mocking const texts = ['text 1', 'text 2', 'text 3']; // Would verify: batch processing with concurrency limit }); }); });