def main(unused_argv):
    if not FLAGS.wals_dir:
        raise ValueError("Specify --wals_dir!")
    if not FLAGS.output_dir:
        raise ValueError("Specify --output_dir!")

    logging.info("Preparing dataset ...")
    df, data_info = _make_df(_read(FLAGS.wals_dir),
                             categorical_as_ints=FLAGS.categorical_as_ints)

    output_file = os.path.join(FLAGS.output_dir, _DATASET_NAME + ".csv")
    logging.info("Saving dataset to \"%s\" ...", output_file)
    df.to_csv(output_file, sep="|", index=False, float_format="%g")
    logging.info("Saved %d languages.", len(df))

    output_file = os.path.join(
        FLAGS.output_dir, const.DATA_INFO_FILENAME + "_" + _DATASET_NAME +
        data_lib.FILE_EXTENSION)
    data_lib.write_data_info(output_file, data_info)
Exemplo n.º 2
0
def _write_dict(data, file_type, output_filename):
    """Writes dictionary of a specified type to a file in output directory."""
    output_filename = os.path.join(
        FLAGS.output_dir,
        output_filename + "_" + file_type + data_lib.FILE_EXTENSION)
    data_lib.write_data_info(output_filename, data)