compression and decompression

compression_decompression.py

def bz2compress(raw_fname, wipe=False, overwrite=False):
    """
    :param raw_fname: string, full path of file to be compressed
    :param wipe: bool, optional. If set to True, raw file will be deleted after compression. Defaults to False.
    :param overwrite: bool, optional. If set to True, overwrites file sharing name of output file. Defaults to False
    :return:
    """

    import bz2
    import os

    compressed_fname = raw_fname + '.bz2'

    if os.path.exists(compressed_fname) and overwrite is False:
        raise ValueError('File {0} already exists. Call bz2compress with '.format(compressed_fname) +
                         'overwrite=True to overwrite.')

    with open(raw_fname, 'rb') as f:
        data = f.read()

    compressed_file = bz2.BZ2File(compressed_fname, "wb")
    compressed_file.write(data)
    compressed_file.close()

    if wipe:
        os.remove(raw_fname)

    return


def bz2decompress(compressed_fname, wipe=False, overwrite=False):
    """
    :param compressed_fname: string, full path of file to be decompressed
    :param wipe: bool, optional. If set to True, raw file will be deleted after decompression
    :param overwrite: bool, optional. If set to True, overwrites file sharing name of output file. Defaults to False.
    :return:
    """

    import bz2
    import os

    raw_fname = compressed_fname[:-4]  # chopping the '.bz2' extension for raw file name

    if os.path.exists(raw_fname) and overwrite is False:
        raise ValueError('File {0} already exists. Call bz2decompress with '.format(raw_fname) +
                         'overwrite=True to overwrite.')

    infile = bz2.BZ2File(compressed_fname, 'rb')
    data = infile.read()
    infile.close()

    with open(raw_fname, 'wb') as f:
        f.write(data)

    if wipe:
        os.remove(compressed_fname)

    return