|
| 1 | +import { afterEach, beforeEach, describe, expect, it, vi } from 'vitest'; |
| 2 | +import { promises as fs } from 'fs'; |
| 3 | +import os from 'os'; |
| 4 | +import path from 'path'; |
| 5 | +import type { SearchResult } from '../src/types/index.js'; |
| 6 | +import { rmWithRetries } from './test-helpers.js'; |
| 7 | + |
| 8 | +const transformersMocks = vi.hoisted(() => ({ |
| 9 | + tokenizerFromPretrained: vi.fn(), |
| 10 | + modelFromPretrained: vi.fn(), |
| 11 | + env: { cacheDir: null as string | null } |
| 12 | +})); |
| 13 | + |
| 14 | +vi.mock('@huggingface/transformers', () => ({ |
| 15 | + AutoTokenizer: { |
| 16 | + from_pretrained: transformersMocks.tokenizerFromPretrained |
| 17 | + }, |
| 18 | + AutoModelForSequenceClassification: { |
| 19 | + from_pretrained: transformersMocks.modelFromPretrained |
| 20 | + }, |
| 21 | + env: transformersMocks.env |
| 22 | +})); |
| 23 | + |
| 24 | +function makeResult(score: number, filePath: string): SearchResult { |
| 25 | + return { |
| 26 | + summary: `Result from ${filePath}`, |
| 27 | + snippet: 'export class Foo {}', |
| 28 | + filePath, |
| 29 | + startLine: 1, |
| 30 | + endLine: 10, |
| 31 | + score, |
| 32 | + language: 'typescript', |
| 33 | + metadata: {} |
| 34 | + } as SearchResult; |
| 35 | +} |
| 36 | + |
| 37 | +describe('reranker corruption recovery', () => { |
| 38 | + let tempCacheRoot: string | null = null; |
| 39 | + let consoleErrorSpy: ReturnType<typeof vi.spyOn>; |
| 40 | + |
| 41 | + beforeEach(async () => { |
| 42 | + vi.resetModules(); |
| 43 | + transformersMocks.tokenizerFromPretrained.mockReset(); |
| 44 | + transformersMocks.modelFromPretrained.mockReset(); |
| 45 | + tempCacheRoot = await fs.mkdtemp(path.join(os.tmpdir(), 'reranker-cache-')); |
| 46 | + transformersMocks.env.cacheDir = tempCacheRoot; |
| 47 | + consoleErrorSpy = vi.spyOn(console, 'error').mockImplementation(() => {}); |
| 48 | + }); |
| 49 | + |
| 50 | + afterEach(async () => { |
| 51 | + consoleErrorSpy.mockRestore(); |
| 52 | + transformersMocks.env.cacheDir = null; |
| 53 | + if (tempCacheRoot) { |
| 54 | + await rmWithRetries(tempCacheRoot); |
| 55 | + tempCacheRoot = null; |
| 56 | + } |
| 57 | + }); |
| 58 | + |
| 59 | + it('clears corrupt cache and retries on the next same-process call', async () => { |
| 60 | + if (!tempCacheRoot) throw new Error('tempCacheRoot not initialized'); |
| 61 | + |
| 62 | + const modelCacheDir = path.join(tempCacheRoot, 'Xenova', 'ms-marco-MiniLM-L-6-v2'); |
| 63 | + await fs.mkdir(modelCacheDir, { recursive: true }); |
| 64 | + await fs.writeFile(path.join(modelCacheDir, 'model.onnx'), 'corrupt'); |
| 65 | + |
| 66 | + const tokenizer = vi.fn((query: string, passage: string) => ({ query, passage })); |
| 67 | + const model = vi.fn(async (inputs: { passage: string }) => { |
| 68 | + if (inputs.passage.includes('/a.ts')) return { logits: { data: [1] } }; |
| 69 | + if (inputs.passage.includes('/b.ts')) return { logits: { data: [3] } }; |
| 70 | + return { logits: { data: [2] } }; |
| 71 | + }); |
| 72 | + |
| 73 | + transformersMocks.tokenizerFromPretrained.mockResolvedValue(tokenizer); |
| 74 | + transformersMocks.modelFromPretrained |
| 75 | + .mockRejectedValueOnce(new Error('Protobuf parse failure while loading model')) |
| 76 | + .mockResolvedValueOnce(model); |
| 77 | + |
| 78 | + const { rerank, getRerankerStatus } = await import('../src/core/reranker.js'); |
| 79 | + const results = [ |
| 80 | + makeResult(0.85, '/a.ts'), |
| 81 | + makeResult(0.83, '/b.ts'), |
| 82 | + makeResult(0.82, '/c.ts') |
| 83 | + ]; |
| 84 | + |
| 85 | + await expect(rerank('auth token', results)).rejects.toThrow(/Protobuf|parse/i); |
| 86 | + await expect(fs.access(modelCacheDir)).rejects.toThrow(); |
| 87 | + expect(getRerankerStatus()).toBe('unavailable'); |
| 88 | + |
| 89 | + const reranked = await rerank('auth token', results); |
| 90 | + expect(transformersMocks.tokenizerFromPretrained).toHaveBeenCalledTimes(2); |
| 91 | + expect(transformersMocks.modelFromPretrained).toHaveBeenCalledTimes(2); |
| 92 | + expect(reranked.map((result) => result.filePath)).toEqual(['/b.ts', '/c.ts', '/a.ts']); |
| 93 | + expect(getRerankerStatus()).toBe('active'); |
| 94 | + }); |
| 95 | +}); |
0 commit comments