Blame · Line-by-line history
semantic-index.test.ts
Each line is annotated with the commit that last touched it. Click any SHA to jump to that commit and see the surrounding change.
| a686079 | 1 | /** |
| 2 | * Tests for src/lib/semantic-index.ts and the /api/v2/.../semantic-search | |
| 3 | * endpoint that wraps it. | |
| 4 | * | |
| 5 | * Layered: | |
| 6 | * - Pure (no DB, no network): fallback embedder shape + determinism, | |
| 7 | * `indexChangedFiles` short-circuits when there are no candidate | |
| 8 | * paths, file-type filtering, and the test-only embedder seam. | |
| 9 | * - DB + pgvector: full upsert → search ranking via the real | |
| 10 | * `code_embeddings` table. Skipped unless DATABASE_URL is present | |
| 11 | * AND the running Postgres has pgvector installed (probed lazily). | |
| 12 | * | |
| 13 | * The HAS_PGVECTOR probe inserts and then deletes a single throwaway | |
| 14 | * row so a missing extension or table is detected at runtime. | |
| 15 | */ | |
| 16 | ||
| 17 | import { | |
| 18 | describe, | |
| 19 | it, | |
| 20 | expect, | |
| 21 | beforeAll, | |
| 22 | afterAll, | |
| 23 | beforeEach, | |
| 24 | } from "bun:test"; | |
| 25 | import { join } from "path"; | |
| 26 | import { rm, mkdir } from "fs/promises"; | |
| 27 | import { randomBytes } from "crypto"; | |
| 28 | import app from "../app"; | |
| 29 | import { clearRateLimitStore } from "../middleware/rate-limit"; | |
| 30 | import { initBareRepo, getRepoPath } from "../git/repository"; | |
| 31 | import { | |
| 32 | indexChangedFiles, | |
| 33 | searchSemantic, | |
| 34 | embedOne, | |
| 35 | __setEmbedderForTests, | |
| 36 | __test, | |
| 37 | EMBEDDING_DIM, | |
| 38 | } from "../lib/semantic-index"; | |
| 39 | ||
| 40 | const HAS_DB = Boolean(process.env.DATABASE_URL); | |
| 41 | ||
| 42 | const TEST_REPOS = join( | |
| 43 | import.meta.dir, | |
| 44 | "../../.test-repos-semantic-index-" + Date.now() | |
| 45 | ); | |
| 46 | ||
| 47 | // Lazily probed — true only if DATABASE_URL is set, the table exists, | |
| 48 | // AND a basic UPSERT/SELECT round-trip through pgvector succeeds. | |
| 49 | let HAS_PGVECTOR = false; | |
| 50 | ||
| 51 | beforeAll(async () => { | |
| 52 | process.env.GIT_REPOS_PATH = TEST_REPOS; | |
| 53 | process.env.GLUECRON_SEMANTIC_CACHE_DIR = join(TEST_REPOS, "_cache"); | |
| 54 | clearRateLimitStore(); | |
| 55 | await rm(TEST_REPOS, { recursive: true, force: true }); | |
| 56 | await mkdir(TEST_REPOS, { recursive: true }); | |
| 57 | ||
| 58 | if (HAS_DB) { | |
| 59 | HAS_PGVECTOR = await probePgvector(); | |
| 60 | } | |
| 61 | }); | |
| 62 | ||
| 63 | afterAll(async () => { | |
| 64 | __setEmbedderForTests(null); | |
| 65 | await rm(TEST_REPOS, { recursive: true, force: true }); | |
| 66 | }); | |
| 67 | ||
| 68 | beforeEach(() => { | |
| 69 | __setEmbedderForTests(null); | |
| 70 | }); | |
| 71 | ||
| 72 | // --------------------------------------------------------------------------- | |
| 73 | // Helpers | |
| 74 | // --------------------------------------------------------------------------- | |
| 75 | ||
| 76 | async function run(cmd: string[], cwd: string) { | |
| 77 | const proc = Bun.spawn(cmd, { cwd, stdout: "pipe", stderr: "pipe" }); | |
| 78 | await new Response(proc.stdout).text(); | |
| 779c681 | 79 | const __code = await proc.exited; |
| 80 | // Fail loudly. These fixtures sit under the project dir, so a silently | |
| 81 | // failed clone leaves cwd inside the REAL repo and the following git | |
| 82 | // config/add/commit then mutate it. See week3.test.ts. | |
| 83 | if (__code !== 0) throw new Error(`git fixture command failed (exit ${__code})`); | |
| a686079 | 84 | } |
| 85 | ||
| 86 | async function seedRepo(owner: string, name: string, files: Record<string, string>) { | |
| 87 | await initBareRepo(owner, name); | |
| 88 | const bare = getRepoPath(owner, name); | |
| 89 | const work = join(TEST_REPOS, "_work_" + randomBytes(4).toString("hex")); | |
| 90 | await mkdir(work, { recursive: true }); | |
| 91 | await run(["git", "clone", bare, work], TEST_REPOS); | |
| 92 | await run(["git", "config", "user.email", "t@gluecron.com"], work); | |
| 93 | await run(["git", "config", "user.name", "T"], work); | |
| 94 | await run(["git", "checkout", "-B", "main"], work); | |
| 95 | for (const [path, content] of Object.entries(files)) { | |
| 96 | const full = join(work, path); | |
| 97 | await mkdir(join(full, ".."), { recursive: true }); | |
| 98 | await Bun.write(full, content); | |
| 99 | } | |
| 100 | await run(["git", "add", "-A"], work); | |
| 101 | await run(["git", "commit", "-m", "seed"], work); | |
| 102 | await run(["git", "push", "-u", "origin", "main"], work); | |
| 103 | // Capture the commit sha | |
| 104 | const { stdout } = await Bun.spawn(["git", "rev-parse", "main"], { | |
| 105 | cwd: work, | |
| 106 | stdout: "pipe", | |
| 107 | }); | |
| 108 | const sha = (await new Response(stdout).text()).trim(); | |
| 109 | await rm(work, { recursive: true, force: true }); | |
| 110 | return sha; | |
| 111 | } | |
| 112 | ||
| 113 | // Deterministic stub embedder used by ranking tests — encodes the | |
| 114 | // first 8 unique characters of the text into the first 8 dimensions | |
| 115 | // so semantically-similar text produces similar vectors. Pads to the | |
| 116 | // full embedding dim with zeros + L2-normalises. | |
| 117 | function makeStubEmbedder(): ( | |
| 118 | text: string, | |
| 119 | inputType: "document" | "query" | |
| 120 | ) => Promise<{ vector: number[]; model: string }> { | |
| 121 | return async (text: string) => { | |
| 122 | const v = new Array<number>(EMBEDDING_DIM).fill(0); | |
| 123 | const lower = text.toLowerCase(); | |
| 124 | // Token-frequency over a small bag of high-signal keywords. The | |
| 125 | // ranking test below relies on "fetch" / "database" being distinct. | |
| 126 | const KEYS = [ | |
| 127 | "fetch", | |
| 128 | "database", | |
| 129 | "config", | |
| 130 | "user", | |
| 131 | "render", | |
| 132 | "test", | |
| 133 | "embed", | |
| 134 | "route", | |
| 135 | ]; | |
| 136 | for (let i = 0; i < KEYS.length; i++) { | |
| 137 | const k = KEYS[i]; | |
| 138 | const matches = lower.split(k).length - 1; | |
| 139 | v[i] = matches; | |
| 140 | } | |
| 141 | // Normalise so cosine == dot. | |
| 142 | let sumsq = 0; | |
| 143 | for (let i = 0; i < EMBEDDING_DIM; i++) sumsq += v[i] * v[i]; | |
| 144 | if (sumsq > 0) { | |
| 145 | const inv = 1 / Math.sqrt(sumsq); | |
| 146 | for (let i = 0; i < EMBEDDING_DIM; i++) v[i] *= inv; | |
| 147 | } | |
| 148 | return { vector: v, model: "stub-1024" }; | |
| 149 | }; | |
| 150 | } | |
| 151 | ||
| 152 | async function probePgvector(): Promise<boolean> { | |
| 153 | if (!HAS_DB) return false; | |
| 154 | try { | |
| 155 | const { db } = await import("../db"); | |
| 156 | const { codeEmbeddings } = await import("../db/schema"); | |
| 157 | const { eq } = await import("drizzle-orm"); | |
| 158 | const fakeRepoId = "00000000-0000-0000-0000-000000000000"; | |
| 159 | // Try a SELECT — fast probe that the table + vector column exist. | |
| 160 | await db | |
| 161 | .select({ id: codeEmbeddings.id }) | |
| 162 | .from(codeEmbeddings) | |
| 163 | .where(eq(codeEmbeddings.repositoryId, fakeRepoId)) | |
| 164 | .limit(1); | |
| 165 | return true; | |
| 166 | } catch { | |
| 167 | return false; | |
| 168 | } | |
| 169 | } | |
| 170 | ||
| 171 | // --------------------------------------------------------------------------- | |
| 172 | // 1. Pure helpers — no DB required | |
| 173 | // --------------------------------------------------------------------------- | |
| 174 | ||
| 175 | describe("semantic-index — pure helpers", () => { | |
| 176 | it("fallbackEmbed returns a 1024-dim vector", () => { | |
| 177 | const v = __test.fallbackEmbed("hello world function getUser"); | |
| 178 | expect(v.length).toBe(EMBEDDING_DIM); | |
| 179 | }); | |
| 180 | ||
| 181 | it("fallbackEmbed is deterministic", () => { | |
| 182 | const a = __test.fallbackEmbed("function indexFiles()"); | |
| 183 | const b = __test.fallbackEmbed("function indexFiles()"); | |
| 184 | expect(a).toEqual(b); | |
| 185 | }); | |
| 186 | ||
| 187 | it("fallbackEmbed produces different vectors for different inputs", () => { | |
| 188 | const a = __test.fallbackEmbed("database connection pool"); | |
| 189 | const b = __test.fallbackEmbed("react component render hook"); | |
| 190 | let differ = false; | |
| 191 | for (let i = 0; i < EMBEDDING_DIM; i++) { | |
| 192 | if (Math.abs(a[i] - b[i]) > 1e-9) { | |
| 193 | differ = true; | |
| 194 | break; | |
| 195 | } | |
| 196 | } | |
| 197 | expect(differ).toBe(true); | |
| 198 | }); | |
| 199 | ||
| 200 | it("deriveBlobSha is deterministic + 64 hex chars", async () => { | |
| 201 | const a = await __test.deriveBlobSha("hello world"); | |
| 202 | const b = await __test.deriveBlobSha("hello world"); | |
| 203 | expect(a).toBe(b); | |
| 204 | expect(/^[0-9a-f]{64}$/.test(a)).toBe(true); | |
| 205 | }); | |
| 206 | ||
| 207 | it("MAX_FILES_PER_PUSH is the documented cap (~50)", () => { | |
| 208 | expect(__test.MAX_FILES_PER_PUSH).toBeGreaterThanOrEqual(20); | |
| 209 | expect(__test.MAX_FILES_PER_PUSH).toBeLessThanOrEqual(200); | |
| 210 | }); | |
| 211 | }); | |
| 212 | ||
| 213 | describe("embedOne — test seam", () => { | |
| 214 | it("respects __setEmbedderForTests override", async () => { | |
| 215 | __setEmbedderForTests(async () => ({ | |
| 216 | vector: new Array(EMBEDDING_DIM).fill(0.001), | |
| 217 | model: "fake-model", | |
| 218 | })); | |
| 219 | const out = await embedOne("anything", "document"); | |
| 220 | expect(out.model).toBe("fake-model"); | |
| 221 | expect(out.vector.length).toBe(EMBEDDING_DIM); | |
| 222 | expect(out.vector[0]).toBe(0.001); | |
| 223 | __setEmbedderForTests(null); | |
| 224 | }); | |
| 225 | ||
| 226 | it("falls back to the deterministic embedder when no key + no override", async () => { | |
| 227 | delete process.env.VOYAGE_API_KEY; | |
| 228 | const out = await embedOne("hello world", "document"); | |
| 229 | expect(out.vector.length).toBe(EMBEDDING_DIM); | |
| 230 | // Fallback model name is stable. | |
| 231 | expect(out.model).toBe(__test.FALLBACK_MODEL); | |
| 232 | }); | |
| 233 | }); | |
| 234 | ||
| 235 | describe("indexChangedFiles — graceful no-ops", () => { | |
| 236 | it("returns 0/0 for empty path list", async () => { | |
| 237 | const out = await indexChangedFiles({ | |
| 238 | repositoryId: "00000000-0000-0000-0000-000000000000", | |
| 239 | ownerName: "nobody", | |
| 240 | repoName: "nothing", | |
| 241 | commitSha: "0000000000000000000000000000000000000000", | |
| 242 | changedPaths: [], | |
| 243 | }); | |
| 244 | expect(out.indexed).toBe(0); | |
| 245 | }); | |
| 246 | ||
| 247 | it("returns 0/0 for empty repositoryId", async () => { | |
| 248 | const out = await indexChangedFiles({ | |
| 249 | repositoryId: "", | |
| 250 | ownerName: "x", | |
| 251 | repoName: "y", | |
| 252 | commitSha: "deadbeef", | |
| 253 | changedPaths: ["src/foo.ts"], | |
| 254 | }); | |
| 255 | expect(out.indexed).toBe(0); | |
| 256 | }); | |
| 257 | ||
| 258 | it("filters out non-code files before doing any work", async () => { | |
| 259 | // We can't easily assert what was filtered without a stub git, but | |
| 260 | // we can verify that *only* non-code paths produce indexed=0 even | |
| 261 | // when a stub embedder would have succeeded. | |
| 262 | __setEmbedderForTests(makeStubEmbedder()); | |
| 263 | const out = await indexChangedFiles({ | |
| 264 | repositoryId: "00000000-0000-0000-0000-000000000000", | |
| 265 | ownerName: "nobody", | |
| 266 | repoName: "nothing", | |
| 267 | commitSha: "0000000000000000000000000000000000000000", | |
| 268 | // None of these are code files we'd index. | |
| 269 | changedPaths: ["LICENSE", "image.png", "binary.zip"], | |
| 270 | }); | |
| 271 | expect(out.indexed).toBe(0); | |
| 272 | __setEmbedderForTests(null); | |
| 273 | }); | |
| 274 | }); | |
| 275 | ||
| 276 | describe("searchSemantic — graceful no-ops", () => { | |
| 277 | it("returns [] for empty query", async () => { | |
| 278 | const out = await searchSemantic({ | |
| 279 | repositoryId: "00000000-0000-0000-0000-000000000000", | |
| 280 | query: "", | |
| 281 | }); | |
| 282 | expect(out).toEqual([]); | |
| 283 | }); | |
| 284 | ||
| 285 | it("returns [] for empty repositoryId", async () => { | |
| 286 | const out = await searchSemantic({ | |
| 287 | repositoryId: "", | |
| 288 | query: "hello", | |
| 289 | }); | |
| 290 | expect(out).toEqual([]); | |
| 291 | }); | |
| 292 | ||
| 293 | it("clamps limit to [1, 100]", async () => { | |
| 294 | // We can't assert the bound directly without DB; but we can verify | |
| 295 | // that absurd values don't throw — graceful behaviour is the | |
| 296 | // contract. | |
| 297 | const a = await searchSemantic({ | |
| 298 | repositoryId: "", | |
| 299 | query: "hi", | |
| 300 | limit: 0, | |
| 301 | }); | |
| 302 | const b = await searchSemantic({ | |
| 303 | repositoryId: "", | |
| 304 | query: "hi", | |
| 305 | limit: 999999, | |
| 306 | }); | |
| 307 | expect(a).toEqual([]); | |
| 308 | expect(b).toEqual([]); | |
| 309 | }); | |
| 310 | }); | |
| 311 | ||
| 312 | // --------------------------------------------------------------------------- | |
| 313 | // 2. API endpoint — auth/validation surface (no DB) | |
| 314 | // --------------------------------------------------------------------------- | |
| 315 | ||
| 316 | describe("GET /api/v2/repos/:o/:r/semantic-search — validation", () => { | |
| b08d63b | 317 | it("returns 404 for a nonexistent repo (or 500 when no DB)", async () => { |
| a686079 | 318 | const res = await app.request( |
| 319 | "/api/v2/repos/nobody/nothing/semantic-search?q=foo" | |
| 320 | ); | |
| b08d63b | 321 | // 404 with a DB, 500 without (resolveRepo throws on missing DATABASE_URL). |
| 322 | expect([404, 500]).toContain(res.status); | |
| a686079 | 323 | }); |
| 324 | ||
| b08d63b | 325 | it("returns 404 (or 500 without DB) when called without ?q on a missing repo", async () => { |
| a686079 | 326 | const res = await app.request( |
| 327 | "/api/v2/repos/nobody/nothing/semantic-search" | |
| 328 | ); | |
| 329 | // Missing repo dominates over empty query. | |
| b08d63b | 330 | expect([404, 500]).toContain(res.status); |
| a686079 | 331 | }); |
| 332 | }); | |
| 333 | ||
| 334 | // --------------------------------------------------------------------------- | |
| 335 | // 3. End-to-end with DB + pgvector | |
| 336 | // --------------------------------------------------------------------------- | |
| 337 | ||
| 338 | describe.skipIf(!HAS_DB)("semantic-index — DB-backed flows", () => { | |
| 339 | it.skipIf(!HAS_DB)("upserts a row per file and ranks by similarity", async () => { | |
| 340 | if (!HAS_PGVECTOR) { | |
| 341 | // Re-check at runtime — the beforeAll probe may have raced before | |
| 342 | // migrations ran. Skip cleanly so the suite stays green when | |
| 343 | // pgvector is unavailable. | |
| 344 | return; | |
| 345 | } | |
| 346 | const { db } = await import("../db"); | |
| 347 | const { users, repositories, codeEmbeddings } = await import( | |
| 348 | "../db/schema" | |
| 349 | ); | |
| 350 | const { eq, and } = await import("drizzle-orm"); | |
| 351 | ||
| 352 | const stamp = randomBytes(4).toString("hex"); | |
| 353 | const username = `semuser-${stamp}`; | |
| 354 | const reponame = `semrepo-${stamp}`; | |
| 355 | ||
| 356 | const [u] = await db | |
| 357 | .insert(users) | |
| 358 | .values({ | |
| 359 | username, | |
| 360 | email: `${username}@test.local`, | |
| 361 | passwordHash: "x", | |
| 362 | }) | |
| 363 | .returning(); | |
| 364 | if (!u) return; | |
| 365 | ||
| 366 | const sha = await seedRepo(username, reponame, { | |
| 367 | "src/fetch.ts": "export function fetchData() { return fetch('/api'); }\n", | |
| 368 | "src/db.ts": "export function connectDatabase() { /* database pool */ }\n", | |
| 369 | }); | |
| 370 | ||
| 371 | const [r] = await db | |
| 372 | .insert(repositories) | |
| 373 | .values({ | |
| 374 | name: reponame, | |
| 375 | ownerId: u.id, | |
| 376 | diskPath: getRepoPath(username, reponame), | |
| 377 | defaultBranch: "main", | |
| 378 | }) | |
| 379 | .returning(); | |
| 380 | if (!r) return; | |
| 381 | ||
| 382 | // Use the deterministic stub so the test doesn't rely on Voyage. | |
| 383 | __setEmbedderForTests(makeStubEmbedder()); | |
| 384 | ||
| 385 | const out = await indexChangedFiles({ | |
| 386 | repositoryId: r.id, | |
| 387 | ownerName: username, | |
| 388 | repoName: reponame, | |
| 389 | commitSha: sha, | |
| 390 | changedPaths: ["src/fetch.ts", "src/db.ts"], | |
| 391 | }); | |
| 392 | ||
| 393 | expect(out.indexed).toBe(2); | |
| 394 | ||
| 395 | // Verify rows landed. | |
| 396 | const rows = await db | |
| 397 | .select({ | |
| 398 | path: codeEmbeddings.filePath, | |
| 399 | snippet: codeEmbeddings.contentSnippet, | |
| 400 | }) | |
| 401 | .from(codeEmbeddings) | |
| 402 | .where(eq(codeEmbeddings.repositoryId, r.id)); | |
| 403 | expect(rows.length).toBe(2); | |
| 404 | const paths = rows.map((r) => r.path).sort(); | |
| 405 | expect(paths).toEqual(["src/db.ts", "src/fetch.ts"]); | |
| 406 | for (const row of rows) { | |
| 407 | // Snippet is the first 500 chars of file content — should be non-empty. | |
| 408 | expect(row.snippet.length).toBeGreaterThan(0); | |
| 409 | expect(row.snippet.length).toBeLessThanOrEqual(500); | |
| 410 | } | |
| 411 | ||
| 412 | // Ranking: query "fetch" should rank src/fetch.ts higher than src/db.ts. | |
| 413 | const hits = await searchSemantic({ | |
| 414 | repositoryId: r.id, | |
| 415 | query: "fetch data", | |
| 416 | limit: 5, | |
| 417 | }); | |
| 418 | expect(hits.length).toBeGreaterThanOrEqual(1); | |
| 419 | expect(hits[0].filePath).toBe("src/fetch.ts"); | |
| 420 | ||
| 421 | // Query "database" should rank src/db.ts higher. | |
| 422 | const hits2 = await searchSemantic({ | |
| 423 | repositoryId: r.id, | |
| 424 | query: "database connection", | |
| 425 | limit: 5, | |
| 426 | }); | |
| 427 | expect(hits2.length).toBeGreaterThanOrEqual(1); | |
| 428 | expect(hits2[0].filePath).toBe("src/db.ts"); | |
| 429 | ||
| 430 | // Idempotency: re-indexing the same files should not duplicate rows | |
| 431 | // (unique index on (repository_id, file_path) enforces this). | |
| 432 | const out2 = await indexChangedFiles({ | |
| 433 | repositoryId: r.id, | |
| 434 | ownerName: username, | |
| 435 | repoName: reponame, | |
| 436 | commitSha: sha, | |
| 437 | changedPaths: ["src/fetch.ts", "src/db.ts"], | |
| 438 | }); | |
| 439 | expect(out2.indexed).toBe(2); | |
| 440 | ||
| 441 | const rowsAfter = await db | |
| 442 | .select({ id: codeEmbeddings.id }) | |
| 443 | .from(codeEmbeddings) | |
| 444 | .where(eq(codeEmbeddings.repositoryId, r.id)); | |
| 445 | expect(rowsAfter.length).toBe(2); | |
| 446 | ||
| 447 | // Cleanup | |
| 448 | await db.delete(codeEmbeddings).where(eq(codeEmbeddings.repositoryId, r.id)); | |
| 449 | await db | |
| 450 | .delete(repositories) | |
| 451 | .where(and(eq(repositories.ownerId, u.id), eq(repositories.name, reponame))); | |
| 452 | await db.delete(users).where(eq(users.id, u.id)); | |
| 453 | __setEmbedderForTests(null); | |
| 454 | }); | |
| 455 | ||
| 456 | it.skipIf(!HAS_DB)( | |
| 457 | "GET /api/v2/repos/:o/:r/semantic-search returns the indexed hits", | |
| 458 | async () => { | |
| 459 | if (!HAS_PGVECTOR) return; | |
| 460 | const { db } = await import("../db"); | |
| 461 | const { users, repositories, codeEmbeddings } = await import( | |
| 462 | "../db/schema" | |
| 463 | ); | |
| 464 | const { eq, and } = await import("drizzle-orm"); | |
| 465 | ||
| 466 | const stamp = randomBytes(4).toString("hex"); | |
| 467 | const username = `semapi-${stamp}`; | |
| 468 | const reponame = `semapi-${stamp}`; | |
| 469 | ||
| 470 | const [u] = await db | |
| 471 | .insert(users) | |
| 472 | .values({ | |
| 473 | username, | |
| 474 | email: `${username}@test.local`, | |
| 475 | passwordHash: "x", | |
| 476 | }) | |
| 477 | .returning(); | |
| 478 | if (!u) return; | |
| 479 | ||
| 480 | const sha = await seedRepo(username, reponame, { | |
| 481 | "src/main.ts": "export function fetchUserData() { return fetch('/u'); }\n", | |
| 482 | }); | |
| 483 | ||
| 484 | const [r] = await db | |
| 485 | .insert(repositories) | |
| 486 | .values({ | |
| 487 | name: reponame, | |
| 488 | ownerId: u.id, | |
| 489 | diskPath: getRepoPath(username, reponame), | |
| 490 | defaultBranch: "main", | |
| 491 | isPrivate: false, | |
| 492 | }) | |
| 493 | .returning(); | |
| 494 | if (!r) return; | |
| 495 | ||
| 496 | __setEmbedderForTests(makeStubEmbedder()); | |
| 497 | ||
| 498 | await indexChangedFiles({ | |
| 499 | repositoryId: r.id, | |
| 500 | ownerName: username, | |
| 501 | repoName: reponame, | |
| 502 | commitSha: sha, | |
| 503 | changedPaths: ["src/main.ts"], | |
| 504 | }); | |
| 505 | ||
| 506 | const res = await app.request( | |
| 507 | `/api/v2/repos/${username}/${reponame}/semantic-search?q=fetch` | |
| 508 | ); | |
| 509 | expect(res.status).toBe(200); | |
| 510 | const body = (await res.json()) as Array<{ | |
| 511 | file_path: string; | |
| 512 | snippet: string; | |
| 513 | score: number; | |
| 514 | blob_sha: string; | |
| 515 | }>; | |
| 516 | expect(Array.isArray(body)).toBe(true); | |
| 517 | expect(body.length).toBeGreaterThanOrEqual(1); | |
| 518 | expect(body[0].file_path).toBe("src/main.ts"); | |
| 519 | expect(typeof body[0].score).toBe("number"); | |
| 520 | expect(typeof body[0].blob_sha).toBe("string"); | |
| 521 | ||
| 522 | // Cleanup | |
| 523 | await db | |
| 524 | .delete(codeEmbeddings) | |
| 525 | .where(eq(codeEmbeddings.repositoryId, r.id)); | |
| 526 | await db | |
| 527 | .delete(repositories) | |
| 528 | .where( | |
| 529 | and(eq(repositories.ownerId, u.id), eq(repositories.name, reponame)) | |
| 530 | ); | |
| 531 | await db.delete(users).where(eq(users.id, u.id)); | |
| 532 | __setEmbedderForTests(null); | |
| 533 | } | |
| 534 | ); | |
| 535 | }); |