Re:Modern CSV version 2 is now available
by TristanBall on Tuesday August 15, 2023 @09:15PM
(#37141588)
Duckdb does this straight from csv. You can treat them pretty much just like un-indexed tables, including queries, joins and output back to csv or parquet.
its
Really nice.
I do that. Many, maybe linked, CSVs. Load all into SQLIte, query, filter, sort, etc.
In the repos to process we have only scripts to load and filter. The SQLite things are short lived.
Also, memory only SQLite FTW!