def resolve_dataset()

in manage.py [0:0]

• 26 lines of code
• 7 McCabe index (conditional complexity)

def resolve_dataset(args, task: str):
    import datasets

    builder = datasets.load_dataset_builder(
        args.dataset_name, use_auth_token=args.token
    )

    if args.dataset_config is None:
        args.dataset_config = builder.config_id
        print(f"Inferred dataset_config {args.dataset_config}")

    splits = builder.info.splits
    if splits is not None:
        if args.dataset_split not in splits:
            raise ValueError(
                f"The split `{args.dataset_split}` is not a valid split, please choose from {','.join(splits.keys())}"
            )

    task_templates = builder.info.task_templates
    if task_templates is not None:
        for task_template in task_templates:
            if task_template.task == task:
                args.dataset_column = task_template.audio_file_path_column
                print(f"Inferred dataset_column {args.dataset_column}")
    return (
        args.dataset_name,
        args.dataset_config,
        args.dataset_split,
        args.dataset_column,
    )