import { setEmbeddingProvider, embedAndStore, semanticSearch } from './build/rag/vector-store.js';
async function test() {
// Configurer Ollama avec nomic-embed-text (768 dimensions)
setEmbeddingProvider('ollama', 'nomic-embed-text');
// Indexer quelques documents
const docs = [
{ path: '/tmp/test1.txt', content: 'Le chat est un animal domestique' },
{ path: '/tmp/test2.txt', content: 'PostgreSQL est une base de données relationnelle' },
{ path: '/tmp/test3.txt', content: 'React est une librairie JavaScript pour le frontend' }
];
console.log('Indexing documents...');
for (const doc of docs) {
await embedAndStore('/tmp/test_project', doc.path, doc.content);
console.log(`Indexed: ${doc.path}`);
}
// Rechercher
console.log('\nSearching for "animal"...');
const results = await semanticSearch('animal', {
projectFilter: '/tmp/test_project',
limit: 5,
threshold: 0
});
console.log('Results:');
results.forEach((r, i) => {
console.log(`${i+1}. ${r.filePath}: ${r.score.toFixed(4)} - "${r.content.trim()}"`);
});
// Calculer les distances entre les vecteurs
console.log('\nCalculating distances...');
const { Pool } = await import('pg');
const pool = new Pool({
host: "localhost",
port: 16432,
database: "rag_mcp_dedicated",
user: "rag_user",
password: "secure_rag_password",
});
const res = await pool.query(`
SELECT
a.file_path as file1,
b.file_path as file2,
a.vector <=> b.vector as distance
FROM rag_store a, rag_store b
WHERE a.id < b.id AND a.project_path = '/tmp/test_project' AND b.project_path = '/tmp/test_project'
ORDER BY distance;
`);
console.log('Distances between documents:');
res.rows.forEach(row => {
console.log(`${row.file1} <-> ${row.file2}: ${row.distance.toFixed(4)}`);
});
await pool.end();
}
test().catch(console.error);
