def get_args()

in obelics/callers/download_warc.py [0:0]


def get_args():
    parser = argparse.ArgumentParser(description="Download warc files from Common Crawl pointers.")
    parser.add_argument(
        "--path_metadata_dataset",
        type=str,
        default="./large_files/metadata_dataset_10000",
        help="Path of the dataset containing the metadata to retrieve the warc files.",
    )
    parser.add_argument(
        "--path_save_dir_warc_dataset",
        type=str,
        default="./large_files/warc_dataset_10000",
        help="The directory to save the warc dataset.",
    )
    parser.add_argument(
        "--num_proc",
        type=int,
        default=cpu_count(),
        help="Number of processes to use for the multiprocessing.",
    )
    args = parser.parse_args()
    return args