|
| 1 | +/** |
| 2 | + * Unit tests for EmbeddingCache |
| 3 | + * |
| 4 | + * Tests in-memory embedding cache with TTL expiration, |
| 5 | + * content hash validation, and LRU-style eviction. |
| 6 | + */ |
| 7 | + |
| 8 | +import { EmbeddingCache } from '../../src/cache/EmbeddingCache'; |
| 9 | + |
| 10 | +describe('EmbeddingCache', () => { |
| 11 | + let cache: EmbeddingCache; |
| 12 | + |
| 13 | + beforeEach(() => { |
| 14 | + jest.useFakeTimers(); |
| 15 | + cache = new EmbeddingCache(); |
| 16 | + }); |
| 17 | + |
| 18 | + afterEach(() => { |
| 19 | + jest.useRealTimers(); |
| 20 | + }); |
| 21 | + |
| 22 | + describe('Basic Operations', () => { |
| 23 | + it('should set and get embedding', () => { |
| 24 | + const embedding = [0.1, 0.2, 0.3, 0.4, 0.5]; |
| 25 | + const contentHash = EmbeddingCache.computeContentHash('title', 'body'); |
| 26 | + |
| 27 | + cache.set('issue-1', contentHash, embedding); |
| 28 | + const result = cache.get('issue-1', contentHash); |
| 29 | + |
| 30 | + expect(result).toEqual(embedding); |
| 31 | + }); |
| 32 | + |
| 33 | + it('should return null for non-existent entry', () => { |
| 34 | + const contentHash = EmbeddingCache.computeContentHash('title', 'body'); |
| 35 | + const result = cache.get('non-existent', contentHash); |
| 36 | + |
| 37 | + expect(result).toBeNull(); |
| 38 | + }); |
| 39 | + |
| 40 | + it('should check if issue exists with has()', () => { |
| 41 | + const contentHash = EmbeddingCache.computeContentHash('title', 'body'); |
| 42 | + cache.set('issue-1', contentHash, [0.1]); |
| 43 | + |
| 44 | + expect(cache.has('issue-1')).toBe(true); |
| 45 | + expect(cache.has('issue-2')).toBe(false); |
| 46 | + }); |
| 47 | + |
| 48 | + it('should clear all entries', () => { |
| 49 | + const contentHash = EmbeddingCache.computeContentHash('title', 'body'); |
| 50 | + cache.set('issue-1', contentHash, [0.1]); |
| 51 | + cache.set('issue-2', contentHash, [0.2]); |
| 52 | + |
| 53 | + cache.clear(); |
| 54 | + |
| 55 | + expect(cache.size()).toBe(0); |
| 56 | + expect(cache.has('issue-1')).toBe(false); |
| 57 | + }); |
| 58 | + |
| 59 | + it('should return correct size', () => { |
| 60 | + const contentHash = EmbeddingCache.computeContentHash('title', 'body'); |
| 61 | + |
| 62 | + expect(cache.size()).toBe(0); |
| 63 | + |
| 64 | + cache.set('issue-1', contentHash, [0.1]); |
| 65 | + expect(cache.size()).toBe(1); |
| 66 | + |
| 67 | + cache.set('issue-2', contentHash, [0.2]); |
| 68 | + expect(cache.size()).toBe(2); |
| 69 | + }); |
| 70 | + |
| 71 | + it('should return all keys', () => { |
| 72 | + const contentHash = EmbeddingCache.computeContentHash('title', 'body'); |
| 73 | + cache.set('issue-1', contentHash, [0.1]); |
| 74 | + cache.set('issue-2', contentHash, [0.2]); |
| 75 | + |
| 76 | + const keys = cache.keys(); |
| 77 | + |
| 78 | + expect(keys).toContain('issue-1'); |
| 79 | + expect(keys).toContain('issue-2'); |
| 80 | + expect(keys).toHaveLength(2); |
| 81 | + }); |
| 82 | + |
| 83 | + it('should overwrite existing entry', () => { |
| 84 | + const contentHash = EmbeddingCache.computeContentHash('title', 'body'); |
| 85 | + cache.set('issue-1', contentHash, [0.1]); |
| 86 | + cache.set('issue-1', contentHash, [0.9]); |
| 87 | + |
| 88 | + const result = cache.get('issue-1', contentHash); |
| 89 | + expect(result).toEqual([0.9]); |
| 90 | + }); |
| 91 | + }); |
| 92 | + |
| 93 | + describe('TTL Expiration', () => { |
| 94 | + it('should return null for expired entry', () => { |
| 95 | + // Create cache with 1 hour TTL |
| 96 | + const cacheWithTTL = new EmbeddingCache(60 * 60 * 1000); // 1 hour |
| 97 | + const contentHash = EmbeddingCache.computeContentHash('title', 'body'); |
| 98 | + |
| 99 | + cacheWithTTL.set('issue-1', contentHash, [0.1]); |
| 100 | + |
| 101 | + // Advance time past TTL |
| 102 | + jest.advanceTimersByTime(2 * 60 * 60 * 1000); // 2 hours |
| 103 | + |
| 104 | + const result = cacheWithTTL.get('issue-1', contentHash); |
| 105 | + expect(result).toBeNull(); |
| 106 | + }); |
| 107 | + |
| 108 | + it('should return entry before TTL expires', () => { |
| 109 | + const cacheWithTTL = new EmbeddingCache(60 * 60 * 1000); // 1 hour |
| 110 | + const contentHash = EmbeddingCache.computeContentHash('title', 'body'); |
| 111 | + |
| 112 | + cacheWithTTL.set('issue-1', contentHash, [0.1]); |
| 113 | + |
| 114 | + // Advance time but not past TTL |
| 115 | + jest.advanceTimersByTime(30 * 60 * 1000); // 30 minutes |
| 116 | + |
| 117 | + const result = cacheWithTTL.get('issue-1', contentHash); |
| 118 | + expect(result).toEqual([0.1]); |
| 119 | + }); |
| 120 | + |
| 121 | + it('should delete expired entry on get()', () => { |
| 122 | + const cacheWithTTL = new EmbeddingCache(60 * 60 * 1000); |
| 123 | + const contentHash = EmbeddingCache.computeContentHash('title', 'body'); |
| 124 | + |
| 125 | + cacheWithTTL.set('issue-1', contentHash, [0.1]); |
| 126 | + jest.advanceTimersByTime(2 * 60 * 60 * 1000); |
| 127 | + |
| 128 | + cacheWithTTL.get('issue-1', contentHash); |
| 129 | + |
| 130 | + // Entry should be removed |
| 131 | + expect(cacheWithTTL.has('issue-1')).toBe(false); |
| 132 | + }); |
| 133 | + |
| 134 | + it('should use default TTL of 24 hours', () => { |
| 135 | + const defaultCache = new EmbeddingCache(); |
| 136 | + const contentHash = EmbeddingCache.computeContentHash('title', 'body'); |
| 137 | + |
| 138 | + defaultCache.set('issue-1', contentHash, [0.1]); |
| 139 | + |
| 140 | + // 12 hours later should still work |
| 141 | + jest.advanceTimersByTime(12 * 60 * 60 * 1000); |
| 142 | + expect(defaultCache.get('issue-1', contentHash)).toEqual([0.1]); |
| 143 | + |
| 144 | + // 25 hours total should expire |
| 145 | + jest.advanceTimersByTime(13 * 60 * 60 * 1000); |
| 146 | + expect(defaultCache.get('issue-1', contentHash)).toBeNull(); |
| 147 | + }); |
| 148 | + |
| 149 | + it('should clean expired entries with cleanExpired()', () => { |
| 150 | + const cacheWithTTL = new EmbeddingCache(60 * 60 * 1000); |
| 151 | + const contentHash = EmbeddingCache.computeContentHash('title', 'body'); |
| 152 | + |
| 153 | + cacheWithTTL.set('issue-1', contentHash, [0.1]); |
| 154 | + cacheWithTTL.set('issue-2', contentHash, [0.2]); |
| 155 | + |
| 156 | + jest.advanceTimersByTime(30 * 60 * 1000); // 30 min |
| 157 | + cacheWithTTL.set('issue-3', contentHash, [0.3]); // Added later |
| 158 | + |
| 159 | + jest.advanceTimersByTime(45 * 60 * 1000); // 75 min total |
| 160 | + |
| 161 | + const removed = cacheWithTTL.cleanExpired(); |
| 162 | + |
| 163 | + expect(removed).toBe(2); // issue-1 and issue-2 expired |
| 164 | + expect(cacheWithTTL.has('issue-3')).toBe(true); |
| 165 | + }); |
| 166 | + }); |
| 167 | + |
| 168 | + describe('Content Hash Validation', () => { |
| 169 | + it('should return null if content hash changes', () => { |
| 170 | + const hash1 = EmbeddingCache.computeContentHash('title', 'body'); |
| 171 | + const hash2 = EmbeddingCache.computeContentHash('title', 'modified body'); |
| 172 | + |
| 173 | + cache.set('issue-1', hash1, [0.1]); |
| 174 | + |
| 175 | + // Request with different hash |
| 176 | + const result = cache.get('issue-1', hash2); |
| 177 | + expect(result).toBeNull(); |
| 178 | + }); |
| 179 | + |
| 180 | + it('should delete entry when content hash changes', () => { |
| 181 | + const hash1 = EmbeddingCache.computeContentHash('title', 'body'); |
| 182 | + const hash2 = EmbeddingCache.computeContentHash('title', 'modified body'); |
| 183 | + |
| 184 | + cache.set('issue-1', hash1, [0.1]); |
| 185 | + cache.get('issue-1', hash2); |
| 186 | + |
| 187 | + // Entry should be removed |
| 188 | + expect(cache.has('issue-1')).toBe(false); |
| 189 | + }); |
| 190 | + |
| 191 | + it('should compute consistent hash for same content', () => { |
| 192 | + const hash1 = EmbeddingCache.computeContentHash('title', 'body'); |
| 193 | + const hash2 = EmbeddingCache.computeContentHash('title', 'body'); |
| 194 | + |
| 195 | + expect(hash1).toBe(hash2); |
| 196 | + }); |
| 197 | + |
| 198 | + it('should compute different hash for different content', () => { |
| 199 | + const hash1 = EmbeddingCache.computeContentHash('title1', 'body'); |
| 200 | + const hash2 = EmbeddingCache.computeContentHash('title2', 'body'); |
| 201 | + |
| 202 | + expect(hash1).not.toBe(hash2); |
| 203 | + }); |
| 204 | + |
| 205 | + it('should normalize content before hashing (trim + lowercase)', () => { |
| 206 | + const hash1 = EmbeddingCache.computeContentHash(' Title ', ' Body '); |
| 207 | + const hash2 = EmbeddingCache.computeContentHash('title', 'body'); |
| 208 | + |
| 209 | + expect(hash1).toBe(hash2); |
| 210 | + }); |
| 211 | + |
| 212 | + it('should handle empty title or body', () => { |
| 213 | + const hashEmptyTitle = EmbeddingCache.computeContentHash('', 'body'); |
| 214 | + const hashEmptyBody = EmbeddingCache.computeContentHash('title', ''); |
| 215 | + const hashBothEmpty = EmbeddingCache.computeContentHash('', ''); |
| 216 | + |
| 217 | + expect(hashEmptyTitle.length).toBe(64); // SHA256 hex length |
| 218 | + expect(hashEmptyBody.length).toBe(64); |
| 219 | + expect(hashBothEmpty.length).toBe(64); |
| 220 | + }); |
| 221 | + |
| 222 | + it('should handle null-ish values', () => { |
| 223 | + const hash = EmbeddingCache.computeContentHash(null as any, undefined as any); |
| 224 | + expect(hash.length).toBe(64); |
| 225 | + }); |
| 226 | + }); |
| 227 | + |
| 228 | + describe('Eviction', () => { |
| 229 | + it('should evict oldest entries when exceeding maxSize', () => { |
| 230 | + const smallCache = new EmbeddingCache(24 * 60 * 60 * 1000, 5); // Max 5 entries |
| 231 | + |
| 232 | + // Add 5 entries |
| 233 | + for (let i = 1; i <= 5; i++) { |
| 234 | + const hash = EmbeddingCache.computeContentHash(`title-${i}`, 'body'); |
| 235 | + smallCache.set(`issue-${i}`, hash, [i * 0.1]); |
| 236 | + jest.advanceTimersByTime(1000); // Space them out in time |
| 237 | + } |
| 238 | + |
| 239 | + expect(smallCache.size()).toBe(5); |
| 240 | + |
| 241 | + // Add 6th entry - should trigger eviction |
| 242 | + const hash6 = EmbeddingCache.computeContentHash('title-6', 'body'); |
| 243 | + smallCache.set('issue-6', hash6, [0.6]); |
| 244 | + |
| 245 | + // Should have evicted oldest (issue-1) |
| 246 | + expect(smallCache.size()).toBeLessThanOrEqual(5); |
| 247 | + expect(smallCache.has('issue-6')).toBe(true); |
| 248 | + }); |
| 249 | + |
| 250 | + it('should evict approximately 10% of maxSize entries', () => { |
| 251 | + const cache100 = new EmbeddingCache(24 * 60 * 60 * 1000, 100); |
| 252 | + |
| 253 | + // Fill to capacity |
| 254 | + for (let i = 1; i <= 100; i++) { |
| 255 | + const hash = EmbeddingCache.computeContentHash(`title-${i}`, 'body'); |
| 256 | + cache100.set(`issue-${i}`, hash, [i * 0.01]); |
| 257 | + jest.advanceTimersByTime(10); |
| 258 | + } |
| 259 | + |
| 260 | + // Add one more |
| 261 | + const hashNew = EmbeddingCache.computeContentHash('new', 'body'); |
| 262 | + cache100.set('issue-new', hashNew, [0.99]); |
| 263 | + |
| 264 | + // Should have evicted ~10 entries |
| 265 | + expect(cache100.size()).toBeLessThanOrEqual(91); |
| 266 | + }); |
| 267 | + |
| 268 | + it('should not evict when updating existing entry', () => { |
| 269 | + const smallCache = new EmbeddingCache(24 * 60 * 60 * 1000, 3); |
| 270 | + |
| 271 | + const hash = EmbeddingCache.computeContentHash('title', 'body'); |
| 272 | + smallCache.set('issue-1', hash, [0.1]); |
| 273 | + smallCache.set('issue-2', hash, [0.2]); |
| 274 | + smallCache.set('issue-3', hash, [0.3]); |
| 275 | + |
| 276 | + // Update existing |
| 277 | + smallCache.set('issue-1', hash, [0.9]); |
| 278 | + |
| 279 | + expect(smallCache.size()).toBe(3); |
| 280 | + expect(smallCache.get('issue-1', hash)).toEqual([0.9]); |
| 281 | + }); |
| 282 | + |
| 283 | + it('should use default maxSize of 10000', () => { |
| 284 | + const defaultCache = new EmbeddingCache(); |
| 285 | + const stats = defaultCache.getStats(); |
| 286 | + |
| 287 | + expect(stats.maxSize).toBe(10000); |
| 288 | + }); |
| 289 | + }); |
| 290 | + |
| 291 | + describe('Statistics', () => { |
| 292 | + it('should return cache statistics', () => { |
| 293 | + const contentHash = EmbeddingCache.computeContentHash('title', 'body'); |
| 294 | + cache.set('issue-1', contentHash, [0.1]); |
| 295 | + |
| 296 | + const stats = cache.getStats(); |
| 297 | + |
| 298 | + expect(stats).toHaveProperty('size'); |
| 299 | + expect(stats).toHaveProperty('maxSize'); |
| 300 | + expect(stats).toHaveProperty('ttlMs'); |
| 301 | + expect(stats.size).toBe(1); |
| 302 | + }); |
| 303 | + |
| 304 | + it('should track oldest and newest entry ages', () => { |
| 305 | + const contentHash = EmbeddingCache.computeContentHash('title', 'body'); |
| 306 | + |
| 307 | + cache.set('issue-1', contentHash, [0.1]); |
| 308 | + jest.advanceTimersByTime(60 * 1000); // 1 minute |
| 309 | + cache.set('issue-2', contentHash, [0.2]); |
| 310 | + jest.advanceTimersByTime(30 * 1000); // 30 more seconds |
| 311 | + |
| 312 | + const stats = cache.getStats(); |
| 313 | + |
| 314 | + expect(stats.oldestEntryAge).toBeGreaterThan(stats.newestEntryAge!); |
| 315 | + }); |
| 316 | + |
| 317 | + it('should handle empty cache stats', () => { |
| 318 | + const stats = cache.getStats(); |
| 319 | + |
| 320 | + expect(stats.size).toBe(0); |
| 321 | + expect(stats.oldestEntryAge).toBeUndefined(); |
| 322 | + expect(stats.newestEntryAge).toBeUndefined(); |
| 323 | + }); |
| 324 | + }); |
| 325 | +}); |
0 commit comments