Story

Hi HN!

I work on an analytical database engine, and I spend a lot of time trying to understand the structure of Apache Parquet files (row groups, column chunks, data pages, encodings, etc.).

I feel like existing tools for this job such as parquet-tools or pyarrow are cumbersome to use and for anything beyond the basics you end up writing custom code. Datanomy contains some more info, but still requires installing a CLI. Some databases like DuckDB also have a feature to query the metadata, but that means writing SQL for every little thing you want to check, and the output is still text.

I wanted something I could just drop a file into, and immediately see the structure visualized — so I built Parquetastic, a browser-based visual Parquet metadata inspector tool.

What it does:

- Visualizes the full Parquet file structure: row groups, column chunks, pages (if your file has a page index), encodings, compression, statistics

- Shows key specs at a glance without writing any code or queries (e.g., footer size, page index size)

- Runs entirely in the browser — the data never leaves your device

- No download, no installation, no signup

Live demo: https://parquetastic.dev

GitHub: https://github.com/FlorianPfisterer/parquetastic

I'd love your feedback, especially from anyone who regularly works with Parquet files!

Disclaimer: This project was fully built in my free time and has nothing to do with my employer. And before you ask: Yes, this was mostly vibe-coded (or rather "agentically engineered"). Without these tools, I wouldn't have had the time to build it!