How to open a Parquet file (no Python needed)

Parquet is a binary columnar format, not plain text like CSV or JSON. A Parquet file begins and ends with the 4 bytes PAR1, so a text editor shows PAR1 followed by unreadable compressed bytes.

Desktop Excel has no Parquet connector either. Power Query's Parquet connector exists only in Power BI and Power Query Online, so double-clicking a .parquet file won't load it into a spreadsheet.

rows.page needs no install: DuckDB-WASM reads the file inside your browser, so your data stays on your machine unless you click Share.

1. Open https://rows.page and drop your .parquet or .pq file onto the page, or choose it from your computer.
2. Scroll the grid: there is no row limit, and rows load in pages of 100 as you scroll.
3. Open the Columns panel for each column's type, distinct count, share of empty values, min and max for numbers and dates, and a histogram or the top 10 values.
4. Filter with the search box (country:NL, price:>20, name:*), or type a DuckDB SQL statement that starts with SELECT, WITH or FROM against the data table to get up to 1,000,000 rows.
5. Use the Export menu to save the current view or SQL result as CSV, JSON, JSONL, zstd-compressed Parquet, or a Markdown table of the first 1,000 rows.

CSV, TSV, JSON and JSONL files are capped at 1.5 GB per browser tab, but Parquet skips that check because DuckDB reads it in place. To view a public remote file without downloading it, put its URL after rows.page/, like https://rows.page/https://example.com/data.parquet; if the host blocks cross-origin requests, the rows.page proxy fetches up to 100 MB.

rows.page can't edit cells, join two files, compare two local files side by side or export to .xlsx. The Columns panel doesn't show mean, median or standard deviation, so run SELECT avg(price), median(price), stddev(price) FROM data or FROM (SUMMARIZE data) in the SQL box.

With DuckDB installed, you can query a Parquet file in place without loading it into a database first. In the DuckDB CLI, this shows the first 10 rows:

SELECT * FROM 'file.parquet' LIMIT 10;

Since DuckDB 1.2.1 (March 2025), DuckDB also has a browser interface built by MotherDuck. This command opens it at localhost:4213:

duckdb -ui

pandas reads Parquet with pd.read_parquet(), but it doesn't include a Parquet engine. Install pyarrow or fastparquet first.

import pandas as pd

df = pd.read_parquet('file.parquet')
print(df.head())

If you want a desktop window instead of a browser tab or terminal, these 3 tools are free:

• Tad: a desktop app (MIT) with DuckDB inside for viewing, pivoting and filtering tabular files.
• ParquetViewer: a standalone viewer (GPL-3.0) that runs on Windows only.
• Parquet Viewer by dvirtz: a VS Code extension (MIT) that previews a .parquet file from your editor sidebar but truncates files over 50 MB.

To see column names and types before you write a pipeline, open the Columns panel on rows.page or run DESCRIBE in DuckDB:

DESCRIBE SELECT * FROM 'file.parquet';

To get a Parquet file into desktop Excel, open it on rows.page, filter or hide the columns you don't need, and choose Export > CSV. You get a comma-delimited UTF-8 CSV without a BOM, and any cell that starts with =, +, -, @, a tab or a carriage return gets a leading apostrophe so Excel won't run it as a formula. Filtered views export every matching row, while SQL results stop at 1,000,000 rows.