SmartQueryTools

Transpose Parquet Files Online

Transpose Parquet files — flip rows and columns — directly in your browser. No upload, no server.

How to transpose Parquet files

  1. Drop your file onto the upload area. The tool shows the row and column count and previews the first 200 rows.
  2. Check the notice above the preview. Files with more than 500 rows are transposed using only the first 500 rows.
  3. Click Transpose. Each original column becomes a row, and the counts update to show the new shape.
  4. Review the transposed table, then click Download CSV. The result is always saved as a CSV file named after your input with _transposed added.

Your file is processed locally in your browser and is never uploaded. The free limit is 50 MB per file; larger files work if your device has the memory for them.

Worked example

A web team keeps quarterly visit totals with one row per region. A slide template needs the quarters running down the page and the regions across the top.

Input (Parquet)

Parquet file (binary, columnar) — shown as a table with its schema

regionq1_visitsq2_visitsq3_visits
North184021051990
South9601130NULL
West241022902675

Schema: region VARCHAR, q1_visits BIGINT, q2_visits BIGINT, q3_visits BIGINT

Settings

  • No options. The first 500 rows are transposed (this file has 3).

Result

FieldRow_1Row_2Row_3
regionNorthSouthWest
q1_visits18409602410
q2_visits210511302290
q3_visits19902675

Three rows and four columns become four rows and four columns. The old header names move into a new first column called Field. The new headers are Row_1, Row_2 and Row_3, not the region names. Every value is written as text, and the missing Q3 figure for South becomes an empty cell.

Working with Parquet files

Parquet columns are typed, but a transposed table cannot keep those types. One output column would have to hold a date, a number and a string at once. The tool converts every value to text instead. Dates are written as YYYY-MM-DD, timestamps as YYYY-MM-DD HH:MM:SS, and decimals keep their exact digits.

The download is a CSV file even though the input was Parquet. Struct and list cells are written as JSON text in a single quoted cell, such as {"city":"Oslo"} or ["a","b"]. Flatten nested columns first if you want each field on its own line. Only the first 500 rows in file order are used, so on a large Parquet file this gives you a sideways view of the first rows, not the whole file. Parquet files often have dozens of columns, which makes wide records hard to read in a normal preview. Transposing the first few rows gives one line per column, so you can scan names and sample values together.

Frequently Asked Questions

Why is my transposed Parquet file a CSV?

After transposing, each column mixes values from different source columns and types, which Parquet cannot store in one typed column. The tool writes the result as CSV text. Convert it back with CSV to Parquet if you need to.

What happens to struct columns when I transpose a Parquet file?

They are not expanded. Each struct cell is written as JSON text, such as {"city":"Oslo","zip":"0150"}. Flatten the file first if you want each field as its own row.

What do the column headers look like after transposing?

The first column is called Field and holds your original column names. The other columns are named Row_1, Row_2 and so on, one per original row. The values in your first column do not become headers.

Is there a row limit?

Yes. Only the first 500 rows are transposed, which gives at most 501 output columns. The tool warns you before you click Transpose if your file has more rows than that.

What format is the transposed file saved in?

Always CSV, whatever format you uploaded. All values are written as text and empty values become empty cells.

Related Tools