scai/dist/db/fileIndex.js

> AI-powered CLI tool for commit messages **and** pull request reviews — using local models.

import fs from 'fs';
import path from 'path';
import { generateEmbedding } from '../lib/generateEmbedding.js';
import { sanitizeQueryForFts } from '../utils/sanitizeQuery.js';
import * as sqlTemplates from './sqlTemplates.js';
import { CANDIDATE_LIMIT } from '../constants.js';
import { getDbForRepo } from './client.js';
import { scoreFiles } from '../fileRules/scoreFiles.js'; // 👈 NEW
export function indexFile(filePath, summary, type) {
    const stats = fs.statSync(filePath);
    const lastModified = stats.mtime.toISOString();
    const indexedAt = new Date().toISOString();
    const normalizedPath = path.normalize(filePath).replace(/\\/g, '/');
    const fileName = path.basename(normalizedPath);
    const db = getDbForRepo();
    db.prepare(sqlTemplates.upsertFileTemplate).run({
        path: normalizedPath,
        filename: fileName,
        summary,
        type,
        lastModified,
        indexedAt,
        embedding: null
    });
    db.prepare(`
    INSERT OR REPLACE INTO files_fts (rowid, filename, summary, path)
    VALUES ((SELECT id FROM files WHERE path = :path), :filename, :summary, :path)
  `).run({
        path: normalizedPath,
        filename: fileName,
        summary,
    });
    console.log(`📄 Indexed: ${normalizedPath}`);
}
export function queryFiles(safeQuery, limit = 10) {
    console.log(`Executing search query: ${safeQuery}`);
    const db = getDbForRepo();
    return db.prepare(`
    SELECT f.id, f.path, f.filename, f.summary, f.type, f.last_modified, f.indexed_at
    FROM files f
    JOIN files_fts fts ON f.id = fts.rowid
    WHERE fts.files_fts MATCH ?
    LIMIT ?
  `).all(safeQuery, limit);
}
export async function searchFiles(query, topK = 5) {
    console.log(`🧠 Searching for query: "${query}"`);
    const embedding = await generateEmbedding(query);
    if (!embedding) {
        console.log('⚠️ Failed to generate embedding for query');
        return [];
    }
    const safeQuery = sanitizeQueryForFts(query);
    console.log(`Executing search query in FTS5: ${safeQuery}`);
    const db = getDbForRepo();
    const ftsResults = db.prepare(`
    SELECT fts.rowid AS id, f.path, f.filename, f.summary, f.type, bm25(files_fts) AS bm25Score, f.embedding
    FROM files f
    JOIN files_fts fts ON f.id = fts.rowid
    WHERE fts.files_fts MATCH ?
    ORDER BY bm25Score ASC
    LIMIT ?
  `).all(safeQuery, CANDIDATE_LIMIT);
    console.log(`FTS search returned ${ftsResults.length} results`);
    if (ftsResults.length === 0)
        return [];
    const scored = scoreFiles(query, embedding, ftsResults);
    return scored.slice(0, topK);
}
export function getFunctionsForFiles(fileIds) {
    if (!fileIds.length)
        return {};
    const placeholders = fileIds.map(() => '?').join(',');
    const db = getDbForRepo();
    const stmt = db.prepare(`
    SELECT f.file_id, f.name, f.start_line, f.end_line, f.content
    FROM functions f
    WHERE f.file_id IN (${placeholders})
  `);
    const rows = stmt.all(...fileIds);
    const grouped = {};
    for (const row of rows) {
        if (!grouped[row.file_id])
            grouped[row.file_id] = [];
        grouped[row.file_id].push({
            name: row.name,
            start_line: row.start_line,
            end_line: row.end_line,
            content: row.content,
        });
    }
    return grouped;
}