Parquet to CSV — Export Rows Locally

What this Parquet to CSV export includes

Choose the current page or all matching rows. Tables show at most 100 rows per page. Generated exports have a 50 MiB cap. Search, filter and sort the table to find the records you want, then check the displayed page before exporting. All matching rows can include the complete dataset when no filters are active and the CSV fits within the cap. Processing and downloads stay local; there is no cloud conversion service.

How to export the current page or all matching rows

Choose your local .parquet file in the Parquet viewer. Review its schema and browse the table. Use the table CSV export control to download the current page or all matching rows. Formula escaping is enabled by default to reduce spreadsheet formula interpretation; an explicit raw opt-out is available. Review exported values before opening them in another application.

Input limits and data fidelity

Parquet input is limited to 200 MiB. DuckDB-Wasm runs locally in your browser, and compressed data can reach memory limits sooner. Encryption and some codec variants are unsupported. CSV represents the exported rows as text, whether you choose the current page or all matching rows. It does not preserve Parquet schema or provide a round-trip archive. Keep the original when types and complete data matter.

When you need every row

Use All matching rows for a filtered CSV export, or clear search and filters to include every row within the 50 MiB cap. Larger conversions need a dedicated data processing tool. This viewer has no SQL editor. Download original preserves the selected Parquet bytes rather than generating CSV. Closing or reloading clears the session, while files you downloaded remain on your device.

Questions and answers