Parquet Viewer
Browse the rows, schema and statistics of an Apache Parquet file. It opens in this tab and is never uploaded.
- Opens
.parquet.parq.pq- Compression
- All 8 Parquet codecs
- Runs in
- This browser tab
- Cost
- Free, no account
Drop a Parquet file here
It opens in this tab. Nothing is uploaded.
See it work
Half a minute with the sample file: open it, check a column, search, export, and read the result.
How it reads your file
Parquet is built for reading only what you need. The viewer leans on that, so big files open fast.
Footer first
A Parquet file keeps its schema and column statistics in a small footer at the end. The viewer reads that first, so columns, types and the row count appear almost at once, whatever the file size.
Only what’s on screen
Rows are decoded in a background worker, only for the part of the file you scroll to. The page stays responsive with millions of rows.
Search and export stream through
They read every row group in order, with progress and a cancel button. Search keeps the first 5,000 matching rows; export builds the CSV or JSON on your device.
What you see
Three views of the same file. These fragments come from the sample file.
Data
Every column header shows its type and how many values are null. Tap a cell for its full value, a header for the column’s statistics.
| tags | list<string> |
|---|---|
| shipping | struct |
| method | string |
| weight_kg | double |
| delivery_date | date |
Schema
Nested lists, maps and structs, with min, max and null counts for every column, overall and per row group: handy for checking partitioning.
- Written by
- Prefect Tools sample generator
- Codec
- SNAPPY
- On disk
- 106 KB
- Compression
- 1.4×
Metadata
The program that wrote the file, format version, codecs and compression ratio, row groups and key-value metadata such as the pandas schema.
Limits
What it doesn’t do, so you know before you open a 5 GB file.
| Search | Reads the whole file and keeps the first 5,000 matching rows. Very large files take a while. |
|---|---|
| Export | Built in memory. For files above a few hundred megabytes, a desktop browser works best. |
| Sorting, SQL | Not supported. |
| Statistics | Shown only when the file contains them. |
| Encryption | Encrypted Parquet files can’t be opened. |
| Big integers | 64-bit integers are shown exactly. JSON export writes values above 253 as strings to keep precision. |
Questions
What is a Parquet file?
Apache Parquet is a columnar file format for analytics and data engineering. It stores data by column with compression and per-column statistics, which keeps files small and fast to query. pandas, Spark, DuckDB, Polars and BigQuery all read and write it.
Is my file uploaded?
No. The viewer reads the file with your browser’s File API inside a Web Worker. You can confirm it in your browser’s developer tools: the Network tab shows no upload while a file is open.
Which compression codecs are supported?
Uncompressed, Snappy, GZip, Brotli, ZSTD, LZ4, LZ4_RAW and LZO: every codec in the Parquet specification.
Is there a file size limit?
There is no fixed limit. Browsing reads the file lazily, so size mostly affects search and export, which read every row. The practical limit is your device’s memory.
More tools in Data viewers.