Time Machine diff

diff_tm.py

#!/usr/bin/env python
import sys, os


def safe_index(l, val):
    try:
        return l.index(val)
    except ValueError:
        return -1


def onerror(err):
    print >>sys.stderr, err


def lsamefile(f1, f2):
    s1 = os.lstat(f1)
    s2 = os.lstat(f2)
    return os.path.samestat(s1, s2)


def diff_tm(old_dir, new_dir):
    """
    Yields tuples like:
        is_dir, event, path

    Where:
        is_dir is boolean showing wherther item is dir or file;
        event is:
        - '+' for new item,
        - '-' for deleted item,
        - '!' for changed item,
        - ' ' for not changed item
        path is relative path of item
    """

    old_walk = os.walk(old_dir, onerror=onerror)
    new_walk = os.walk(new_dir, onerror=onerror)
    while True:
        try:
            old_cur, old_subdirs, old_files = old_walk.next()
            new_cur, new_subdirs, new_files = new_walk.next()
        except StopIteration:
            break
        subdirs = []
        for d in old_subdirs:
            path1 = os.path.join(old_cur,d)
            rel = os.path.relpath(path1,old_dir)
            ind = safe_index(new_subdirs,d)
            if ind == -1:
                yield (True, '-', rel)
                continue
            path2 = os.path.join(new_cur,d)
            if lsamefile(path1,path2):
                yield (True, ' ', rel)
            else:
                yield (True, '!', rel)
                subdirs.append(d)
            del new_subdirs[ind]
        for d in new_subdirs:
            yield (True, '+', os.path.relpath(os.path.join(new_cur, d), new_dir))
        old_subdirs[:] = new_subdirs[:] = subdirs
        for f in old_files:
            path1 = os.path.join(old_cur, f)
            rel = os.path.relpath(path1, old_dir)
            ind = safe_index(new_files, f)
            if ind == -1:
                yield (False, '-', rel)
                continue
            path2 = os.path.join(new_cur,f)
            if lsamefile(path1,path2):
                yield (False, ' ', rel)
            else:
                yield (False, '!', rel)
            del new_files[ind]
        for f in new_files:
            yield (False, '+', os.path.relpath(os.path.join(new_cur, f), new_dir))

if __name__ == '__main__':
    import optparse
    
    parser = optparse.OptionParser(
        usage="Usage: %prog old_dir new_dir",
        description="Gathers difference between two directory trees by inode "
                    "numbers. By default prints only new and changed files. "
                    "All output paths is relative to old_dir/new_dir.")
    parser.add_option("-f", "--files", action="store_true",
            dest="show_files", default=True, help="Show files in output")
    parser.add_option("-F", "--no-files", action="store_false",
            dest="show_files", default=True, help="Don't show files in output")
    parser.add_option("-d", "--dirs", action="store_true",
            dest="show_dirs", default=False, help="Show dirs in output")
    parser.add_option("-D", "--no-dirs", action="store_true",
            dest="show_dirs", default=False, help="Don't show dirs in output")

    parser.add_option("-s", action="store_true",
            dest="names_only", default=True, help="Show only file/dir paths")
    parser.add_option("-l", action="store_false",
            dest="names_only", help="Show what happend with file/dir")

    parser.add_option("-u", action="store_true", dest="show_unchanged",
            default=False, help="Show unchanged items")
    parser.add_option("-U", action="store_false", dest="show_unchanged",
            help="Don't show unchanged items")
    parser.add_option("-c", action="store_true", dest="show_changed",
            default=True, help="Show changed items")
    parser.add_option("-C", action="store_false", dest="show_changed",
            help="Don't show changed items")
    parser.add_option("-r", action="store_true", dest="show_removed",
            default=False, help="Show removed items")
    parser.add_option("-R", action="store_false", dest="show_removed",
            help="Don't show removed items")
    parser.add_option("-n", action="store_true", dest="show_new",
            default=True, help="Show new items")
    parser.add_option("-N", action="store_false", dest="show_new",
            help="Don't show new items")

    (options, args) = parser.parse_args()

    if len(args) != 2:
        parser.print_help()
        sys.exit(1)
    
    event_dict = {' ':'Unchanged', '!':'Changed', '-':'Removed', '+':'New'}
    event_opts = {' ':options.show_unchanged, '!':options.show_changed,
                  '-':options.show_removed, '+':options.show_new}

    for t in diff_tm(args[0], args[1]):
        if (t[0] and options.show_dirs or not t[0] and options.show_files) \
                and event_opts[t[1]]:
            if options.names_only:
                print t[2]
            else:
                print "%s %s: %s" % (event_dict[t[1]],
                                    "dir" if t[0] else "file", t[2])

From blog: http://procrastinatrix.yorik.name/2010/10/time-machine-backup-diffs.html

Run it like this:
cd /Volumes//Backups.backupdb//
diff_tm yyyy-MM-dd-HHmmss1 yyyy-MM-dd-HHmmss2
By default it shows only new or changed files, just everything that was sent to server during last backup. It runs very fast (around a second for not-so-far backups).
Hope, it will help someone. Testing, changes, comments appreciated.
PS: I've found that the fattest files in every backup is Chrome's history index. I think, I'll keep them in backup.