def _cast_pandas_column()

in awswrangler/_data_types.py [0:0]

• 29 lines of code
• 9 McCabe index (conditional complexity)

def _cast_pandas_column(df: pd.DataFrame, col: str, current_type: str, desired_type: str) -> pd.DataFrame:
    if desired_type == "datetime64":
        df[col] = pd.to_datetime(df[col])
    elif desired_type == "date":
        df[col] = df[col].apply(lambda x: _cast2date(value=x)).replace(to_replace={pd.NaT: None})
    elif desired_type == "bytes":
        df[col] = df[col].astype("string").str.encode(encoding="utf-8").replace(to_replace={pd.NA: None})
    elif desired_type == "decimal":
        # First cast to string
        df = _cast_pandas_column(df=df, col=col, current_type=current_type, desired_type="string")
        # Then cast to decimal
        df[col] = df[col].apply(lambda x: Decimal(str(x)) if str(x) not in ("", "none", "None", " ", "<NA>") else None)
    else:
        try:
            df[col] = df[col].astype(desired_type)
        except TypeError as ex:
            _logger.debug("Column: %s", col)
            if "object cannot be converted to an IntegerDtype" not in str(ex):
                raise ex
            warnings.warn(
                "Object cannot be converted to an IntegerDtype. Integer columns in Python cannot contain "
                "missing values. If your input data contains missing values, it will be encoded as floats"
                "which may cause precision loss.",
                UserWarning,
            )
            df[col] = (
                df[col]
                .apply(lambda x: int(x) if str(x) not in ("", "none", "None", " ", "<NA>") else None)
                .astype(desired_type)
            )
    return df