Parquet to CSV converter
Open a Parquet file, see what's in it, and save it as CSV or Excel. Your file never leaves your browser.
Excel
Google SheetsDrop a Parquet file here
A .parquet file, up to 100 MB. It stays on your computer.
- No sign-up
- Nothing uploaded
- Snappy, gzip and zstd files
Look inside a Parquet file without installing anything.
Your file stays on your computer
The Parquet file is read in your browser. Nothing is sent to Mammoth or anyone else.
Every column type written sensibly
Timestamps become readable date-times, 64-bit IDs keep every digit, and lists become JSON text in one cell.
See the rows before you save
A preview shows the first 50 rows and the full row count, so you know what's in the file before you download it.
How to convert Parquet to CSV
The converter reads the file's schema, decompresses each column and writes one CSV line per row.
- 1
Add your Parquet file
Drop a .parquet file on the box or choose one.
- 2
Check the rows
A preview shows the columns and the first rows of the table.
- 3
Download
Pick a separator and save the CSV, or take an Excel file instead.
Other ways to do it
If you have Python or DuckDB installed, one line will do it.
Python with pandas
A job you script once
- Reads compressed filesYes
- Shows the rows firstNo
- No code or installNo
- Runs on a scheduleYes
DuckDB
Large files, SQL users
- Reads compressed filesYes
- Shows the rows firstYes
- No code or installNo
- Runs on a scheduleYes
- This page
This converter
Looking inside one file
- Reads compressed filesYes
- Shows the rows firstYes
- No code or installYes
- Runs on a scheduleNo
How to convert Parquet to CSV in Python or DuckDB
pandas does it in one line once pyarrow is installed. DuckDB does it from the command line, and copes with files bigger than your memory.
Need the rows as JSON instead? Save the CSV here, then use the CSV to JSON converter.
import pandas as pd
pd.read_parquet("orders.parquet").to_csv("orders.csv", index=False)read_parquet needs pyarrow installed. pandas writes booleans as True and False.
duckdb -c "COPY (SELECT * FROM 'orders.parquet') TO 'orders.csv' (HEADER, DELIMITER ',')"
DuckDB streams the file, so it handles Parquet far larger than memory. Add a WHERE clause to export only some rows.
Your file stays private
The converter runs inside this page. Your Parquet file is read and written out as CSV on your computer, and nothing is sent to Mammoth. That makes it safe for warehouse extracts with customer data in them.
Once the page has loaded, it works with your internet switched off.
Sending data on as Parquet?
Mammoth works the other way too. Build the table once from your files and download it as Parquet, CSV, Excel or JSON whenever it changes.
Source files
Drive
DropboxOneDrive
SFTP| Shop | Nov 02 | Order 1041 | 129.50 | |
| Wholesale | Nov 02 | Order 7710 | 1,200.00 | |
| Returns | Nov 03 | Return 88 | -48.00 | |
| Shop | Nov 09 | Order 1077 | 310.25 | |
| Wholesale | Nov 09 | Order 7744 | 875.90 |
Illustration. Every new file in the folder is added to the same table.