def update_dataset()

in utils/gradio_utils.py [0:0]

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

def update_dataset(dataset_name: str, ds_name_to_dict):
    # choose a config to analyze
    ds_configs = ds_name_to_dict[dataset_name]
    # special handling for the largest-by-far dataset, C4
    if dataset_name == "c4":
        config_names = ['en', 'en.noblocklist', 'realnewslike']
    else:
        config_names = list(ds_configs.keys())

    config_name = config_names[0]
    ds_config = ds_configs[config_name]

    text_features = ds_config[HF_FEATURE_FIELD]["string"]
    text_features = [('text',)] if dataset_name == "c4" else [tp for tp in text_features if tp[0] != "id"]
    feature = str(text_features[0])
    text_features = [str(f) for f in text_features]

    avail_splits = list(ds_config["splits"].keys())
    split = avail_splits[0]

    return [(config_names, config_name), (text_features, feature), (avail_splits, split)]