verify-backup.py

import os
import hashlib
import sys

from pathlib import Path  # python < 3.4 : pip install pathlib


def md5(path):
    content = path.open('br').read()
    return hashlib.md5(content).hexdigest()


def does_md5_match(path1, path2):
    return md5(path1) == md5(path2)


def compare_dirs(dir1, dir2):
    delta = {'missing': {'files': [], 'dirs': []}, 'differed': []}

    for curdir, subdir_list, file_list in os.walk(dir1):

        # compting relative path eg. src/foo -> dst/foo
        dst_dir_path = Path(dir2) / Path(curdir).relative_to(dir1)

        if not dst_dir_path.exists():
            delta['missing']['files'].extend([(Path(curdir) / p) for p in subdir_list])
            delta['missing']['dirs'].extend([(Path(curdir) / p) for p in file_list])
        else:
            for path in file_list:
                src_path = Path(curdir) / path
                dst_path = dst_dir_path / path
                if not dst_path.exists():
                    delta['missing']['files'].append(src_path)
                elif not does_md5_match(src_path, dst_path):
                    delta['differed'].append(src_path)
            for path in subdir_list:
                src_path = Path(curdir) / path
                dst_path = dst_dir_path / path
                if not dst_path.exists():
                    delta['missing']['dirs'].append(src_path)
    return delta


def test():
    src, dst = sys.argv[1:3]
    print(compare_dirs(src, dst))

test()