How to Query Parquet Files in the Browser

Parquet is a columnar file format designed for analytical queries. It stores schema information and is often smaller and faster to scan than equivalent CSV exports.

• Column types are preserved better than in plain text files.
• Analytical queries can read only the columns they need.
• Parquet is common in data warehouses, lakehouses, and Python/R workflows.

1. Open the SQL for Files editor.
2. Add a .parquet file from your device.
3. Inspect the table and schema in the Database sidebar.
4. Write SQL against the generated table name.

Because Parquet stores typed columns, the schema view helps you quickly identify dates, numeric measures, dimensions, booleans, and nested fields.

SELECT *
FROM transactions
LIMIT 25;

Parquet is strongest when you want to summarize large sets of rows by a smaller number of dimensions.

SELECT
  date_trunc('month', order_date) AS month,
  product_category,
  SUM(revenue) AS revenue
FROM transactions
GROUP BY month, product_category
ORDER BY month, revenue DESC;

Parquet can be efficient, but SQL for Files still runs inside browser memory. Very large files may hit browser limits earlier than a native DuckDB process or server-side database.