Question 1

How do I convert Parquet to CSV?

Accepted Answer

Open the file and press .csv. Every row is written, not only the page on screen, one row group at a time so a large file does not have to be held as text all at once. Nested columns - lists, maps and structs - are written as JSON inside their cell, since CSV has no way to hold them otherwise.

Question 2

Why do some large numbers come out as text in the JSON?

Accepted Answer

Because JSON numbers are 64-bit floating point and cannot hold every 64-bit integer: 9007199254740993 would silently become 9007199254740992. Integers that fit exactly are written as numbers; the ones that would change are written as strings instead, so nothing is quietly altered.

Question 3

Can it open a folder of Parquet files, or a Delta or Iceberg table?

Accepted Answer

No - one .parquet file at a time. A partitioned dataset is a folder of such files, so any single one of them will open, but the table formats built on top of Parquet need their own reader.

Question 4

Is the data uploaded to be read?

Accepted Answer

No. The file is decoded in this page by hyparquet. Parquet files are usually exports of whole tables - customers, transactions, logs - which is exactly the kind of file that should not go to a stranger's server to be looked at.

Free Parquet file viewer online