Get files out of git-annex without any dependencies except Python 3

deannex.py

#!/usr/bin/python3
#
# A small utility for pulling files out of git-annex without
# actually having git-annex installed, with minimal
# dependencies.
import sys
import os
import argparse
import shutil
import hashlib


def print_win(f):
    try:
        print(f)
    except UnicodeEncodeError:
        print(f.encode('utf-8', 'surrogateescape').decode('ISO-8859-1'))


def parseargs():
    parser = argparse.ArgumentParser(description='Extract files from a git-annex.')
    parser.add_argument('--to', dest='output', required=True,
                        help='path to the output directory')
    parser.add_argument('--from', dest='annex', required=True,
                        help='path to the git annex')
    parser.add_argument('--action', dest='action', choices=['copy', 'move', 'none'], default='none',
                        help='action to perform for each file')

    return parser.parse_args()


def do_copy(src, dst):
    print_win("cp {} {}".format(src, dst))
    try:
        shutil.copy2(src, dst)
        return True
    except shutil.Error as err:
        print(err)
        return False
    except PermissionError as err:
        print(err)
        return False


def do_remove(src):
    print_win("rm {}".format(src))
    try:
        os.remove(src)
        return True
    except OSError as err:
        print(err)
        return False


def main():
    args = parseargs()
    output = args.output
    annex = args.annex
    action = args.action

    tgttail = len(annex)
    for dirpath, dirnames, filenames in os.walk(annex):
        if dirpath.startswith(os.path.join(annex, ".git")):
            continue
        target = output + dirpath[tgttail:]
        if not os.path.isdir(target):
            print_win("mkdir {}".format(target))
            if action in ['copy', 'move']:
                os.mkdir(target)
        for f in filenames:
            src = os.path.join(dirpath, f)
            dst = os.path.join(target, f)
            if os.path.exists(dst):
                if action == 'move' and os.path.exists(src):
                    print_win("md5sum {} ...".format(src))
                    m = hashlib.md5()
                    m.update(open(src, 'rb').read())
                    srcdigest = m.digest()
                    print_win("md5sum {} ...".format(dst))
                    m = hashlib.md5()
                    m.update(open(dst, 'rb').read())
                    dstdigest = m.digest()
                    if srcdigest == dstdigest:
                        if do_remove(dst) and do_copy(src, dst) and action == 'move':
                            do_remove(src)
                    else:
                        print_win("diff! ls -l \"{}\" \"{}\"".format(src, dst))
            else:
                if action in ['copy', 'move']:
                    if do_copy(src, dst) and action == 'move':
                        do_remove(src)


if __name__ == "__main__":
    main()