Parquet Viewer
View a Parquet file's schema and data — free, unlimited, entirely private.
Drop files here or click to choose
🔒 Your files never leave your browserHow it works
This tool reads your Parquet file's footer metadata to learn its schema, then decodes the column data using hyparquet, a pure-JavaScript Parquet reader — no server, no conversion service. Because Parquet readers only need the footer and the specific row groups being requested, this stays fast even for large files without reading the whole thing into memory at once.
Limitations
The preview reads up to 10,000 rows; both CSV and JSON export cover everything that was read. Nested (struct/array) column values are shown and exported as their JSON representation rather than being flattened into separate columns.
FAQ
- What is a Parquet file?
- A columnar storage format widely used in data engineering (Spark, Pandas, data lakes) — efficient for large analytical datasets, but not human-readable without a tool like this one or a code library.
- Is there a row limit?
- Rows are read up to 10,000 for the preview, rendered with virtualization so browsing them stays responsive regardless of count. Both exports (CSV and JSON) include everything that was read.
- Is my file uploaded anywhere?
- No. This tool reads only the specific byte ranges it needs — the file's footer metadata and the requested columns — directly from your device, using hyparquet (a pure-JavaScript Parquet reader). Nothing is uploaded.