Parquet is a column-oriented format widely used in analytics and data pipelines, but it cannot be meaningfully inspected in a normal text editor. Setting up Python, DuckDB, or Spark can feel excessive when you only want a quick look at a file.

If you mainly need the schema, row count, row groups, compression, metadata, and a sample of the actual rows, a local browser viewer can often be enough.

What to inspect first in a Parquet file

Start with the schema to confirm column names and data types. Then check row count and row groups to understand the file's scale and physical organization.

Creator information and custom metadata can also reveal useful clues about the system or pipeline that produced the file.

  • Column names and types
  • Total rows and row groups
  • Compression codec
  • Creator and custom metadata

Do not try to display every row of a large Parquet file at once

Parquet is often used for datasets with hundreds of thousands or millions of rows. For inspection, it is more practical to read metadata first and preview only the pages you need.

Browser Kitty's Parquet Viewer lets you choose the number of rows per page and open multiple files in tabs so you can compare their structure.

  • Preview 50, 100, 250, 500, or 1,000 rows per page
  • Compare multiple files in tabs
  • Copy or export only the current page when needed

The difference between a viewer and a SQL analytics tool

Parquet Viewer is designed for inspection, not as a SQL query engine. If you need aggregation, joins, or a full-file ORDER BY, DuckDB or another analytics environment is a better fit.

Sorting and CSV export apply to the currently displayed preview page. Treat the viewer as a quick investigation tool rather than a full conversion pipeline.

Step by step

  1. Select or drop one or more .parquet files.
  2. Review the schema, row count, row groups, compression codec, and metadata.
  3. Choose a preview page size and move to the page you need.
  4. Sort the current preview page by a column when useful.
  5. Export the displayed page as CSV or copy it to the clipboard.
Try it in Browser Kitty

Parquet Viewer

Inspect Parquet files directly in your browser without uploading them.

Open toolView tool details

Tips and limitations

  • Start with 50 to 100 rows to understand the columns and values, then increase the page size only when useful.
  • When you receive several Parquet files, compare their schemas and row counts in separate tabs to spot differences quickly.
  • If you later need full aggregation or conversion, use the browser inspection as a quick first step before moving to DuckDB, Python, or another analytics tool.

Frequently asked questions

Which Parquet compression codecs are supported?

The viewer supports Uncompressed, Snappy, Gzip, Brotli, LZ4, LZ4_RAW, and Zstandard. Some files may still use encodings or structures the embedded parser cannot read.

Can I convert the entire Parquet file to CSV?

CSV export and copy currently apply to the visible preview page. The tool is intended for inspection rather than full-file conversion.

Is the Parquet file uploaded to a Browser Kitty server?

Files selected in Parquet Viewer are parsed in the browser and are not uploaded to a Browser Kitty backend.