Why not parquet, json or something else? My main issue with CSV is that it doesn’t retain types and inferencing seems to never work quite right. You can still open parquet & json in Pandas. CSV introduces a ton of edge cases that don’t exist with other formats. It seems simple until you get serious.
This could be definitely the extension of the tool. Mainly it was build for the project needs I was working on, and there we only required to export the data in simple CSV format (or mainly the part of it, using Query). Will create the ticket to consider implement parquet.
Also one of the goal was to escape from internal DynamoDB JSON format, where each field has a type, and have this data stripped. For parquet format although it seems more logical to keep that data.
Thank you for suggestion.