179 lines
6.1 KiB
TypeScript
179 lines
6.1 KiB
TypeScript
/**
|
|
* Test du ChunkIndexingService — dedup, stale deletion, upsert.
|
|
* Mocke l'embedding (pas d'appel API), utilise la vraie DB.
|
|
*/
|
|
import { prisma } from '../../lib/prisma'
|
|
import { ChunkIndexingService } from '../../lib/ai/services/chunk-indexing.service'
|
|
|
|
const testNoteId = 'test-chunk-000001'
|
|
|
|
function test(name: string, fn: () => Promise<void>) {
|
|
return fn()
|
|
.then(() => console.log(` ✓ ${name}`))
|
|
.catch((err: any) => {
|
|
console.error(` ✗ ${name}: ${err.message}`)
|
|
process.exitCode = 1
|
|
})
|
|
}
|
|
|
|
// Mock l'embedding service : vecteur déterministe basé sur le hash du contenu
|
|
const originalEmbedText =
|
|
require('../../lib/ai/services/embedding.service').embeddingService.embedText
|
|
|
|
function mockEmbedding() {
|
|
const svc = require('../../lib/ai/services/embedding.service').embeddingService
|
|
svc.embedText = async (text: string) => {
|
|
const crypto = require('crypto')
|
|
const hash = crypto.createHash('md5').update(text).digest()
|
|
return Array.from({ length: 1536 }, (_, i) => hash[i % 16] / 255)
|
|
}
|
|
}
|
|
|
|
function restoreEmbedding() {
|
|
const svc = require('../../lib/ai/services/embedding.service').embeddingService
|
|
svc.embedText = originalEmbedText
|
|
}
|
|
|
|
async function ensureTestNote() {
|
|
await prisma.note.upsert({
|
|
where: { id: testNoteId },
|
|
create: {
|
|
id: testNoteId,
|
|
title: 'Test Note for Chunk Indexing',
|
|
content: '<p>Test</p>',
|
|
},
|
|
update: {},
|
|
})
|
|
}
|
|
|
|
async function cleanup() {
|
|
await prisma.noteEmbeddingChunk.deleteMany({ where: { noteId: testNoteId } })
|
|
}
|
|
|
|
async function removeTestNote() {
|
|
await prisma.note.delete({ where: { id: testNoteId } }).catch(() => {})
|
|
}
|
|
|
|
async function main() {
|
|
mockEmbedding()
|
|
await ensureTestNote()
|
|
await cleanup()
|
|
|
|
console.log('\n=== US-CHUNK-2 : Indexation incrémentale avec dedup ===\n')
|
|
|
|
const service = new ChunkIndexingService()
|
|
|
|
const longContent = Array.from({ length: 8 }, (_, i) =>
|
|
`Section ${i} de la note de test. `.repeat(60).trim(),
|
|
).join('\n\n')
|
|
|
|
await test('première indexation → tous les fragments sont nouveaux', async () => {
|
|
const result = await service.indexNote(testNoteId, 'Note de test', longContent)
|
|
|
|
if (result.newFragments === 0)
|
|
throw new Error(`attendu newFragments > 0, reçu ${result.newFragments}`)
|
|
if (result.deleted !== 0)
|
|
throw new Error(`attendu deleted=0, reçu ${result.deleted}`)
|
|
if (result.skipped !== 0)
|
|
throw new Error(`attendu skipped=0, reçu ${result.skipped}`)
|
|
|
|
console.log(` → ${result.newFragments} nouveaux, ${result.totalFragments} total`)
|
|
})
|
|
|
|
await test('deuxième indexation (même contenu) → tout skipped, 0 nouveau', async () => {
|
|
const result = await service.indexNote(testNoteId, 'Note de test', longContent)
|
|
|
|
if (result.newFragments !== 0)
|
|
throw new Error(`attendu 0 nouveau, reçu ${result.newFragments}`)
|
|
if (result.skipped === 0)
|
|
throw new Error(`attendu skipped > 0, reçu ${result.skipped}`)
|
|
if (result.deleted !== 0)
|
|
throw new Error(`attendu deleted=0, reçu ${result.deleted}`)
|
|
|
|
console.log(` → ${result.skipped} skip, 0 nouveau ✓`)
|
|
})
|
|
|
|
await test('modification d\'une section → 1 nouveau, reste skip', async () => {
|
|
const sections = Array.from({ length: 8 }, (_, i) =>
|
|
`Section ${i === 3 ? 'MODIFIÉE' : i} de la note de test. `.repeat(60).trim(),
|
|
)
|
|
const modified = sections.join('\n\n')
|
|
|
|
const result = await service.indexNote(testNoteId, 'Note de test', modified)
|
|
|
|
if (result.newFragments === 0)
|
|
throw new Error(`attendu au moins 1 nouveau fragment, reçu ${result.newFragments}`)
|
|
if (result.deleted === 0)
|
|
throw new Error(`attendu au moins 1 stale supprimé, reçu ${result.deleted}`)
|
|
|
|
console.log(` → ${result.newFragments} nouveau(x), ${result.skipped} skip, ${result.deleted} supprimé(s)`)
|
|
})
|
|
|
|
await test('suppression d\'une section → fragments stale nettoyés', async () => {
|
|
const sections = Array.from({ length: 8 }, (_, i) =>
|
|
`Section ${i} de la note de test. `.repeat(60).trim(),
|
|
)
|
|
await service.indexNote(testNoteId, 'Note de test', sections.join('\n\n'))
|
|
|
|
const beforeCount = await prisma.noteEmbeddingChunk.count({
|
|
where: { noteId: testNoteId },
|
|
})
|
|
|
|
const shorter = sections.slice(0, 4).join('\n\n')
|
|
const result = await service.indexNote(testNoteId, 'Note de test', shorter)
|
|
|
|
const afterCount = await prisma.noteEmbeddingChunk.count({
|
|
where: { noteId: testNoteId },
|
|
})
|
|
|
|
if (afterCount >= beforeCount)
|
|
throw new Error(`attendu count < ${beforeCount}, reçu ${afterCount}`)
|
|
if (result.deleted === 0)
|
|
throw new Error(`attendu deleted > 0, reçu ${result.deleted}`)
|
|
|
|
console.log(` → ${beforeCount} → ${afterCount} fragments (${result.deleted} supprimés)`)
|
|
})
|
|
|
|
await test('note vide → tous les fragments supprimés', async () => {
|
|
await service.indexNote(testNoteId, 'Note de test', longContent)
|
|
const result = await service.indexNote(testNoteId, '', '')
|
|
|
|
const count = await prisma.noteEmbeddingChunk.count({
|
|
where: { noteId: testNoteId },
|
|
})
|
|
|
|
if (count !== 0) throw new Error(`attendu 0 fragments, reçu ${count}`)
|
|
console.log(` → tous les fragments supprimés ✓`)
|
|
})
|
|
|
|
await test('deleteNoteChunks → supprime tout', async () => {
|
|
await service.indexNote(testNoteId, 'Note de test', longContent)
|
|
await service.deleteNoteChunks(testNoteId)
|
|
|
|
const count = await prisma.noteEmbeddingChunk.count({
|
|
where: { noteId: testNoteId },
|
|
})
|
|
if (count !== 0) throw new Error(`attendu 0, reçu ${count}`)
|
|
console.log(` → 0 fragment restant ✓`)
|
|
})
|
|
|
|
await test('hasChunks → détection correcte', async () => {
|
|
const before = await service.hasChunks(testNoteId)
|
|
if (before) throw new Error('attendu false avant indexation')
|
|
|
|
await service.indexNote(testNoteId, 'Note de test', longContent)
|
|
const after = await service.hasChunks(testNoteId)
|
|
if (!after) throw new Error('attendu true après indexation')
|
|
|
|
console.log(` → false avant, true après ✓`)
|
|
})
|
|
|
|
await cleanup()
|
|
restoreEmbedding()
|
|
await prisma.$disconnect()
|
|
|
|
console.log('\n=== Tests terminés ===')
|
|
}
|
|
|
|
main().catch(console.error)
|