Recursively Unpack Archives

unpr.py

#!/usr/bin/env python3
import glob
import os
import logging
import multiprocessing
import re
import shutil
import sys

from pyunpack import Archive

logging.basicConfig(level=logging.WARN)

ARCHIVE_RE = re.compile(".*(\.7z|\.ace|\.alz|\.a|\.arc|\.arj|\.tar\.bz2|\.tar\.gz|\.tbz2|\.tgz|\.bz2|\.cab|\.Z|\.cpio|\.deb|\.dms|\.gz|\.lrz|\.lha|\.lzh\.lz|\.lzma|\.lzo|\.rpm|\.rar|\.rz|\.tar|\.xz|\.zip|\.jar|\.zoo)")

ARCHIVE_SINK_UNPACKED = "./unpack_success"
ARCHIVE_SINK_ERROR = "./unpack_error"

os.makedirs(ARCHIVE_SINK_UNPACKED, exist_ok=True)
os.makedirs(ARCHIVE_SINK_ERROR, exist_ok=True)

def get_archives(path):
     for f in glob.glob(path + '/**/*.*', recursive=True):
        if ARCHIVE_RE.match(f) and os.path.isfile(f):   
            yield f

def move_archive(archive, path):
    new_archive_path = "{}/{}".format(path, os.path.basename(archive))
    
    if os.path.exists(new_archive_path):
        os.remove(new_archive_path)

    os.makedirs(path, exist_ok=True)
    shutil.move(archive, path)

def unpack(archive):
    archive_dir = os.path.dirname(archive)

    archive_unpack_dir = "{}/{}".format(archive_dir, '_unpack')
    os.makedirs(archive_unpack_dir, exist_ok=True)

    try:
        logging.info("UNARCHIVING: {}".format(archive))    
        Archive(archive).extractall(archive_unpack_dir)
    except:
        move_archive(archive, "{}/{}".format(ARCHIVE_SINK_ERROR, archive_dir))
        logging.info("UNARCHIVE_FAILURE: {}".format(archive))
        return

    logging.info("UNARCHIVE_SUCCESS: {}".format(archive))
 
    move_archive(archive, "{}/{}".format(ARCHIVE_SINK_UNPACKED, archive_dir))

    for sub_archive in get_archives(archive_unpack_dir):
        unpack(sub_archive)

def main():
    with multiprocessing.Pool(16) as pool:
        pool.map(unpack, get_archives(sys.argv[1]))

if __name__ == '__main__':
    main()