column_names ignored by load_dataset() when loading CSV file

[luismsgomes]

Describe the bug

load_dataset() ignores the column_names kwarg when loading a CSV file. Instead, it uses whatever values are on the first line of the file.

Steps to reproduce the bug

Call load_dataset to load data from a CSV file and specify column_names kwarg.

Expected behavior

The resulting dataset should have the specified column names and the first line of the file should be considered as data values.

Environment info

• datasets version: 2.20.0
• Platform: Linux-5.10.0-30-cloud-amd64-x86_64-with-glibc2.31
• Python version: 3.9.2
• huggingface_hub version: 0.24.2
• PyArrow version: 17.0.0
• Pandas version: 2.2.2
• fsspec version: 2024.5.0

[albertvillanova]

I confirm that column_names values are not copied to names variable because in this case CsvConfig.__post_init__ is not called: CsvConfig is instantiated with default values and afterwards the config_kwargs are used to overwrite its attributes.

@luismsgomes in the meantime, you can avoid the bug if you pass names instead of column_names.