def load_or_convert_to_dataframe()

in src/alpaca_eval/utils.py [0:0]

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

def load_or_convert_to_dataframe(df=Union[AnyPath, AnyData, Callable, list, tuple], **kwargs):
    """Load a dataframe from a path or convert the input to a dataframe if it's not a path."""
    if isinstance(df, Callable):
        df = df(**kwargs)

    if isinstance(df, (tuple, list)) and isinstance(df[0], (str, os.PathLike, pathlib.Path)):
        df = pd.concat(
            [load_or_convert_to_dataframe(f, **kwargs) for f in df],
        )

    if isinstance(df, (str, os.PathLike, pathlib.Path)):
        df = Path(df)

        # check if it's a globbing pattern
        if "*" in str(df):
            df = pd.concat(
                [load_or_convert_to_dataframe(f, **kwargs) for f in glob.glob(str(df))],
            )
        else:
            suffix = df.suffix
            if suffix == ".json":
                df = pd.read_json(df, **kwargs)
            elif suffix == ".csv":
                df = pd.read_csv(df, **kwargs)
                if df.columns[0] == "Unnamed: 0":
                    df.set_index(df.columns[0], inplace=True)
                    df.index.name = None
            elif suffix == ".tsv":
                df = pd.read_table(df, sep="\t", **kwargs)
            else:
                raise ValueError(f"File format {suffix} not supported.")
    else:
        df = convert_to_dataframe(df, **kwargs)

    return df