Python Differential Backup Script for Immutable Files

backup.py

import os
import shutil

__author__ = 'faisal'


backup_folders = [
    # folders to backup, backup location, ignore top level directories
    ('D:\\', 'H:\\Drive_D\\', ['_temp', 'Downloads', '$RECYCLE.BIN']),
    ('E:\\', 'H:\\Drive_E\\', ['$RECYCLE.BIN'])
]


cached_list = {}


def get_files(path, ignores=[]):
    if path in cached_list:
        return cached_list[path]

    f = []
    for (dirpath, dirnames, filenames) in os.walk(path):
        prefix = dirpath.replace(path, '')
        f.extend([os.path.join(prefix, filename) for filename in filenames if prefix.split(os.sep).pop(0) not in ignores])
    cached_list[path] = f
    return f


for folder in backup_folders:
    print 'Getting source files...'
    src = get_files(folder[0], folder[2])
    print 'Getting destination files...'
    dest = get_files(folder[1])
    print 'Finding diff files...'
    diff = list(set(src) - set(dest))
    diff_rem = list(set(dest) - set(src))
    print 'Found {} files to copy and {} files to remove'.format(len(diff), len(diff_rem))

    # change current dir to source for copying
    os.chdir(folder[0])
    for f in diff:
        print 'Copying {}...'.format(f)
        df = os.path.join(folder[1], f)
        dd = os.path.dirname(df)
        if not os.path.exists(dd):
            os.makedirs(dd)
        shutil.copy2(f, df)

    # change current dir to destination for deletions
    os.chdir(folder[1])
    for f in diff_rem:
        print 'Deleting {}...'.format(f)
        os.unlink(f)


print 'Completed'