DuckDB
PackageEmbedded analytical SQL over local CSV, Parquet, JSON and DataFrames
- Price
- Free, open source
- Access
- No account needed
About
DuckDB is an MIT-licensed analytical database embedded in your process. Query local files and DataFrames using SQL without running a database service or signing up. Its Python client requires Python 3.9 or newer; a CLI and other language clients are available.
What you can do with it
- Query local CSV and Parquet files with SQL without a database server
- Join and aggregate Pandas, Polars or Arrow data in Python
- Embed an analytical database in an application or data-processing script
Get started
- Install DuckDB with pip install duckdb or use the CLI
- Point a query at your local CSV or Parquet file
- Fetch query results directly in your application
Example
import duckdb
duckdb.sql("SELECT * FROM 'example.parquet' LIMIT 5").show()Details
- Hosting
- Runs locally
- Available in
- Worldwide
- Official SDKs
- Python
- MCP server
- None