def validate_preprocessing_args()

in pytorch_translate/options.py [0:0]

• 75 lines of code
• 34 McCabe index (conditional complexity)

def validate_preprocessing_args(args):
    if (
        not (
            (
                getattr(args, "train_source_text_file", None)
                or getattr(args, "train_source_binary_path", None)
                or getattr(args, "multiling_train_source_text_file", None)
                or getattr(args, "multilingual_train_text_file", None)
                or getattr(args, "multilingual_train_binary_path", None)
            )
            and (
                getattr(args, "train_target_text_file", None)
                or getattr(args, "train_target_binary_path", None)
                or getattr(args, "multiling_train_target_text_file", None)
                or getattr(args, "multilingual_train_text_file", None)
                or getattr(args, "multilingual_train_binary_path", None)
            )
            and (
                getattr(args, "eval_source_text_file", None)
                or getattr(args, "eval_source_binary_path", None)
                or getattr(args, "multiling_eval_source_text_file", None)
                or getattr(args, "multilingual_eval_text_file", None)
                or getattr(args, "multilingual_eval_binary_path", None)
            )
            and (
                getattr(args, "eval_target_text_file", None)
                or getattr(args, "eval_target_binary_path", None)
                or getattr(args, "multiling_eval_target_text_file", None)
                or getattr(args, "multilingual_eval_text_file", None)
                or getattr(args, "multilingual_eval_binary_path", None)
            )
        )
        and args.task != "dual_learning_task"
    ):
        raise ValueError(
            "At least one of --*_text_file or --*_binary_path flags must be "
            "specified for each of --{train, eval}_{source, target}_*"
        )

    for file_type in (
        "train_source_text_file",
        "train_target_text_file",
        "eval_source_text_file",
        "eval_target_text_file",
    ):
        text_file = getattr(args, file_type)
        if text_file and not os.path.isfile(text_file):
            raise ValueError(
                f"Please specify an existing text file for "
                f"--{file_type}={text_file}"
            )

    for file_type in ("source_vocab_file", "target_vocab_file"):
        vocab_file = getattr(args, file_type)
        if not vocab_file:
            raise ValueError(
                f"--{file_type} must be specified - even if you don't have "
                f"a vocab file, you must still specify a location "
                f"for it to be written to."
            )

    if args.task == constants.SEMI_SUPERVISED_TASK and not (
        getattr(args, "train_mono_source_binary_path", None)
        or getattr(args, "train_mono_target_binary_path", None)
        or getattr(args, "train_mono_source_text_file", None)
    ):
        raise ValueError(
            "For semisupervised training, at least one of --*_text_file or "
            "--*_binary_path flags must be specified for at least one of "
            "--train_mono_{source, target}_*"
        )

        for file_type in ("train_mono_source_text_file", "train_mono_target_text_file"):
            file_path = getattr(args, file_type)
            if file_path and not os.path.isfile(file_path):
                raise ValueError(
                    f"Please specify an existing text file for --{file_type}="
                    f"{file_path}"
                )