Consider refactoring tables_to_parquet, currently it handles downloading of tables the conversion to parquet - ideally these processes could be separated - the majority of the issues are already handled separately, via get_odata() and convert_table_to_parquet().
Additionally - it is likely better named dataset_urls_to_parquet, or urls_to_parquet if kept together.
Consider refactoring
tables_to_parquet, currently it handles downloading of tables the conversion to parquet - ideally these processes could be separated - the majority of the issues are already handled separately, viaget_odata()andconvert_table_to_parquet().Additionally - it is likely better named
dataset_urls_to_parquet, orurls_to_parquetif kept together.