Copy data from a Time Machine volume mounted on a Linux box.

copy_from_time_machine.py

"""
A script to rebuild an Apple Time Machine (TM) drive somewhere sensible

The original and best is from here:
https://gist.github.com/vjt/5183305
"""

import os
import sys
import pdb
import shutil
import argparse
import subprocess
import os.path as osp

HFS_PREFIX = '.HFS+ Private Directory Data'
BACKUP_LIST_DIRECTORY = 'Backups.backupdb'

def find_hfs_directory(dirpath):

    root_dirs = os.listdir(dirpath)

    if BACKUP_LIST_DIRECTORY not in root_dirs:
        return

    for d in root_dirs:
        if d.startswith(HFS_PREFIX):
            return osp.join(dirpath, d)


def main(src, target, root_dir=None):

    if not osp.exists(src):
        raise ValueError('Source directory does not exist')

    if root_dir is None:
        root_dir = src
        hfs_dir = None

        while True:
            hfs_dir = find_hfs_directory(root_dir)

            if hfs_dir is not None:
                print('Root directory found at ' + root_dir)
                break

            root_dir = osp.dirname(root_dir)

            if root_dir == '/':
                break

        if hfs_dir is None:
            print('Could not find Time Machine (TM) root directory. Please specify it manually.')

    else:

        if BACKUP_LIST_DIRECTORY not in root_dir:
            raise RuntimeError(BACKUP_LIST_DIRECTORY+' not found in the root directory.')

        if HFS_PREFIX not in root_dir:
            raise RuntimeError(HFS_PREFIX+' not found in the root directory.')

    if not osp.exists(target):
        os.makedirs(target)

    inner(src, target, hfs_dir)


def inner(src, target, hfsd):

    cmd = ['find', src, '-mindepth', '1', '-maxdepth', '1', '-and', '-not', '-name', '.', '-and', '-not', '-name', '..']
    try:
        result = subprocess.check_output(cmd).decode()
    except Exception as e:
        print('failed to run find for', src)
        print(e)
        return

    for entry in result.split('\n')[:-1]:

        if '.DS_Store' in entry:
            continue

        dest = osp.join(target, osp.basename(entry))

        cmd = ['stat', '-c', "'%h %F'", entry]
        try:
            stat_result = subprocess.check_output(cmd).decode().strip("\n'")
        except Exception as e:
            print('failed to stat', entry)
            print(e)
            continue
        hlnum, _, typ = stat_result.partition(' ')

        # if typ in ('regular file', 'symbolic link'):
        if typ in ('regular file',):
            if not osp.exists(dest):
                try:
                    shutil.copy2(entry, dest)
                except Exception as e:
                    print('failed to copy', entry)
                    print(e)
                    continue

        elif typ == 'directory':
            print(dest)
            if not osp.exists(dest):
                os.makedirs(dest)
            inner(entry, dest, hfsd)

        elif typ == 'regular empty file':
            hldir = osp.join(hfsd, 'dir_'+hlnum)
            
            if osp.exists(hldir):
                if not osp.exists(dest):
                    os.makedirs(dest)
                inner(hldir, dest, hfsd)
            

if __name__ == '__main__':

    parser = argparse.ArgumentParser(
        description='Rebuild an Apple Time Machine (TM) drive somewhere sensible',
        formatter_class=argparse.ArgumentDefaultsHelpFormatter)

    parser.add_argument('source', type=str, 
        help='Location of the Time Machine (TM) backup you wish to restore. Should end with something like "Backups.backupdb/<machine name>/<backup version>/Macintosh HD"')
    parser.add_argument('target', type=str, 
        help='Where you want the data to end up')
    parser.add_argument('-r', '--root_dir', type=str, default=None, 
        help='Location of Time Machine (TM) root directory. Only needed if this script fails to infer it from the source argument.')
    params = parser.parse_args()

    main(params.source, params.target, params.root_dir)