Function try_format_code

codegeex/data/process_pretrain_dataset.py:19–29

(code: str)

def try_format_code(code: str):
    # Auto-correct to PEP8 format (Change tab to 4-whitespaces;
    # add whitespace around some special symbols;
    # reformat line length < 100, etc.)
    try:
        res = format_str(code, mode=FileMode(line_length=200))
    except Exception as e:
        res = code
        print(e)
        print("Wrong python format: {}".format(code))
    return res


def load_pretrain_dataset(dataset_path: Union[str, List[str]]) -> Dict: