What is a Parquet file?

Apache Parquet is a columnar on-disk format optimized for analytical queries. Data is partitioned into row groups, each row group into column chunks, each chunk into pages — so engines can read only the columns they need.

Parquet files routinely run into the hundreds of millions of rows. We materialize a sample so the viewer stays responsive; the row count in the toolbar shows the true total from the footer metadata.

Does it handle nested types?

Yes. Lists, structs, and maps round-trip through hyparquet and we render them as compact JSON. Decimals, dates, and timestamps surface with their logical type annotation in the Schema tab.

No. The bytes are parsed locally in your browser; the file never leaves your device.

Does it work on mobile?

Yes. The results table scrolls horizontally with a sticky row index, and the schema view stacks cleanly on narrow viewports.

Parquet viewer — open .parquet in your browser

About Parquet

Apache columnar data format.

Parquet is the modern columnar analytics format — the backing storage for Spark, Athena, DuckDB, BigQuery external tables, and the Hugging Face datasets hub. hyparquet decodes it directly in the browser without WebAssembly, reading the footer metadata first and then materializing only the rows you ask for. We sample the first 1,000 rows so a multi-gigabyte file opens instantly, surface the leaf-column schema with logical types and repetition modes, and let you export the sample as JSON. Nested values (lists, structs, maps) are rendered as compact JSON; binary columns are summarized by byte length.

FAQ

What is a Parquet file?: Apache Parquet is a columnar on-disk format optimized for analytical queries. Data is partitioned into row groups, each row group into column chunks, each chunk into pages — so engines can read only the columns they need.
Why only 1,000 rows?: Parquet files routinely run into the hundreds of millions of rows. We materialize a sample so the viewer stays responsive; the row count in the toolbar shows the true total from the footer metadata.
Does it handle nested types?: Yes. Lists, structs, and maps round-trip through hyparquet and we render them as compact JSON. Decimals, dates, and timestamps surface with their logical type annotation in the Schema tab.
Is my file uploaded?: No. The bytes are parsed locally in your browser; the file never leaves your device.
Does it work on mobile?: Yes. The results table scrolls horizontally with a sticky row index, and the schema view stacks cleanly on narrow viewports.

More Data & Config

Parquet viewer

Open Parquet files

Open Parquet files