LSQL

Visual SQL query builder for your CSV files. Drop CSVs into the browser, get a live schema with inferred relationships, and build real SELECT queries by clicking — no SQL typing required. Every query runs against an in-memory SQLite database and returns real rows plus the exact SQL it generated.

Features

• Drag-and-drop CSV ingestion — parse multiple files at once, add more incrementally. Delimiter (,; tab |), quoted fields, and BOM handled automatically.
• Automatic schema inference — column types (INTEGER / REAL / TEXT), primary keys (an INTEGER id column), and foreign keys detected from <name>_id → <name>s naming heuristics.
• Visual query builder — pick FROM, add JOINs (INNER / LEFT / RIGHT), SELECT columns with aggregates (COUNT/SUM/AVG/MIN/MAX), WHERE conditions (AND/OR), GROUP BY, ORDER BY, LIMIT, and DISTINCT.
• Query graph — see tables and their join relationships visualized.
• Generated SQL + live results — the SQL is shown and results returned (capped at 1,000 rows for display) with query timing.
• Safe by design — the client only sends a structured query model; SQL is rebuilt on the server, SELECT-only. No raw SQL from the client is ever executed.
• Light / dark theme.

How it works

1. You drop CSV files. The server parses them, infers column types and foreign keys, and loads the rows (up to 100k per table) into a fresh in-memory SQLite database (Node's built-in node:sqlite), one per session.
2. As you build a query in the UI, a structured Query object is maintained.
3. On run, the server turns that model into a parameter-free SELECT statement, executes it, and returns the columns, rows, and elapsed time.

Nothing is persisted — each upload session lives in memory and disappears when the server restarts.

Getting started

Requires Node.js 22+ (for the built-in node:sqlite module).

cd web
npm install
npm run dev

Open http://localhost:3000 and drop in some CSVs. The samples/ directory ships a small e-commerce dataset (customers, orders, order_items, products, categories, employees, shipments, reviews) with ready-made relationships to try joins.

Scripts

Command	Action
npm run dev	Start the dev server
npm run build	Production build
npm run start	Serve the production build
npm run lint	Lint

Tech stack

• Next.js 15 (App Router) + React 19
• Node.js built-in node:sqlite — no external database
• Tailwind CSS 4 + shadcn/ui (Radix primitives)

Limits

• Max 20 MB per CSV file, 100,000 rows loaded per table.
• 1,000 rows returned per query (results are truncated for display, not the underlying query).
• Foreign keys are inferred from naming conventions, so unusually named columns may not auto-link — you can still join them manually.

Project layout

web/ Next.js app (UI + API + SQLite/CSV engine)
samples/ Example e-commerce CSV dataset