Overview
Introduction
Parquet is a columnar binary format built for analytics, which also means you can't just open a .parquet file in a text editor to see what's inside it the way you can a CSV. This tool reads one in the browser and shows you a normal, readable table instead.
Drop in a file and its rows appear immediately as an HTML table, with no upload, no command-line tooling, and no need to spin up a notebook or a database just to check what a Parquet file contains.
What Is Parquet Reader and Viewer?
A read-only Parquet file viewer that parses the file's footer metadata and column data directly in JavaScript, then renders the decoded rows as a table.
Column order is read from the file's own schema rather than guessed from the first row's object keys, so the table stays correctly ordered even when some rows have null or missing values in a given column.
How Parquet Reader and Viewer Works
The uploaded file is wrapped as an in-memory random-access buffer and handed to a Parquet reader, which first parses the file's footer to recover its schema and row group layout, then decodes the column data pages back into row objects.
Row reading is capped at 500 rows using the reader's own row-range option, so even a Parquet file with millions of rows previews quickly instead of decoding (and rendering) the entire file just to show a sample.
When To Use Parquet Reader and Viewer
Use it whenever you've received or exported a .parquet file and want to quickly confirm what's actually in it before writing code against it.
It's especially handy for spot-checking data pipeline output, verifying column names and a few sample values, without installing a Python environment or a Parquet-aware database.
Often used alongside Parquet Schema Reader, Parquet Metadata Reader and Parquet to CSV Converter.
Features
Advantages
- No install, upload, or command-line tool required, works entirely client-side in the browser.
- Schema-driven column ordering stays correct even for rows with sparse or null values.
- The 500-row cap keeps large files responsive instead of freezing the tab trying to render everything.
Limitations
- This is a preview, not a full data export; use the Parquet to CSV or Parquet to JSON converters to get every row out.
- Deeply nested or repeated (list/map) columns are shown as their raw JSON-like structure rather than flattened into separate sub-columns.
Examples
Best Practices & Notes
Best Practices
- If you see a truncation notice, use the Parquet to CSV or Parquet to JSON converter to work with the full dataset instead of just the preview.
- Check the Parquet Schema Reader first if you only care about column names and types rather than actual data values.
- Use Parquet Metadata Reader alongside this viewer to see the true total row count when the preview itself is truncated.
Developer Notes
The viewer reads a file's schema via `parquetSchema` before reading rows via `parquetReadObjects`, rather than inferring columns from `Object.keys()` of the first returned row, since a row with a null in its first column would otherwise silently omit that column's header.
Parquet Reader and Viewer Use Cases
- Quickly checking a Parquet export's contents without a Python or Spark environment
- Verifying column names and a handful of sample values before writing a data pipeline against a new file
- Spot-checking that a data pipeline's Parquet output looks correct
Common Mistakes
- Assuming the visible table is the complete dataset when a truncation notice is shown, it's the first 500 rows only.
- Expecting nested list or map columns to render as flattened sub-tables rather than as their raw structured value.
Tips
- Use the Parquet Metadata Reader alongside this tool if you want the exact total row count before deciding whether to convert the full file.
- Pair with the Parquet to CSV Converter once you've confirmed a file's shape looks right, to get the full dataset as portable text.