read_csv with with store_rejects=true does not work as intented with constraints about duckdb HOT 4 CLOSED

Thanks for the report, the error is unrelated to read_csv however.
read_csv store_rejects handles errors that happen during reading of the csv file
This error originates from writing the data to the table, which is not related to csv read.

The constraint information could perhaps be pushed down into the csv reader so it's aware and can act accordingly, but that's definitely a feature request

from duckdb.

As mentioned this is not an issue with the CSV file or the rejects table - as such this is out-of-scope for the CSV reader. You could perhaps use e.g. INSERT OR IGNORE or other SQL constructs to get the desired behavior instead.

from duckdb.

@Mytherin @Tishj Thank you for your answers. I understand the rationale.

How could I add such cases to reject_errors table? My goal is to use duckdb + sqlalchemy as a data quality assessment framework.

from duckdb.

The reject_errors table is specifically meant for finding errors within CSV files. Duplicate keys is not an error within a CSV file but a manually inserted constraint. As such you would need to do this outside of the rejects table.

from duckdb.