In the civis.io.civis_file_to_table function, the Python API client appears to tell platform to change SQL types like VARCHAR(1) in the table_columns argument to more generic values like VARCHAR(1024) when there are multiple files. The behavior seems reasonable when data types are automatically inferred (i.e., when need_table_columns is true), as noted in the code comment, but if the user has specified a dtype with specific length like VARCHAR(1), the import should just use that.
In the
civis.io.civis_file_to_table
function, the Python API client appears to tell platform to change SQL types likeVARCHAR(1)
in thetable_columns
argument to more generic values likeVARCHAR(1024)
when there are multiple files. The behavior seems reasonable when data types are automatically inferred (i.e., whenneed_table_columns
is true), as noted in the code comment, but if the user has specified a dtype with specific length likeVARCHAR(1)
, the import should just use that.