Parquet Viewer
Read every layer of a Parquet file the way a query engine does — then fix what would hurt you in production.
Stays on your device — 100% in-browser, nothing uploaded
Drag a .parquet file here
or pick one — it opens instantly, nothing leaves your browser.
🔒 Read locally with DuckDB-WASM — your file is never uploaded.
Parquet files — FAQ
- How do I open or view a Parquet file?
- Drag a .parquet file onto this page (or load the built-in sample). It opens instantly in your browser — no upload, no account — showing the schema, metadata, row groups, encodings, compression, statistics, and a data preview.
- Is my file uploaded to a server?
- No. The file is read entirely on your device with an in-browser engine (DuckDB-WASM); nothing is sent to a server. You can open DevTools → Network to confirm, and the tool keeps working offline once loaded.
- What can I inspect in a Parquet file?
- The full schema (including nested/repeated fields), file and row-group metadata, per-column chunks, encodings, compression codecs and ratios, min/max/null statistics, and a preview of the actual rows.
- What is a good Parquet row group size?
- A common target is 128 MB–1 GB per row group so engines can prune and parallelize efficiently. Many tiny row groups add footer overhead and hurt scan performance — the advisor flags row groups that are too small for the file.
- Why does Parquet have min/max statistics?
- Per-column min/max (and null counts) let a query engine skip row groups and pages that can't match a filter (predicate pushdown). Missing or unsorted statistics mean the engine scans data it could have skipped; the advisor flags when stats are absent.
- Does it work online and offline?
- Both. It runs entirely in your browser with no server round-trips, so it works online instantly — and once the page and its in-browser engine (DuckDB-WASM) have loaded, it keeps working offline too.
- Is the Parquet Viewer free?
- Yes — it is completely free with no sign-up. The entire viewer and advisor run as JavaScript/WASM in your browser.
Understand why these numbers matter
The Parquet Format Deep Dive teaches the footer, encodings, statistics, and page index from first principles — so you can read any file like the engine does.