cp with regular expression

nightly_backup.py

"""Behaves pretty much like  cp -Pruv  , except supports skipping directories.

Copy one directory to another recursively.  If any subdirectories match a
specified regular expression pattern, they are not copied.

Also, directories containing a file called .nobackup are not backup up.
"""
import os
import os.path as P
import re
import shutil
import stat

def nightly_backup(source, dest, skip_regex):
    stack = [ P.abspath(source) ]
    while stack:
        from_dir = stack.pop()
        contents = os.listdir(from_dir)

        if ".nobackup" in contents:
            continue

        rel_dirpath = P.relpath(from_dir, source)
        to_dir = P.join(dest, rel_dirpath)
        if not P.isdir(to_dir):
            os.mkdir(to_dir)

        for name in contents:
            from_path = P.join(from_dir, name)
            to_path = P.join(to_dir, name)
            if skip_regex.match(from_path) or P.islink(from_path):
                continue
            
            if P.isdir(from_path):
                stack.append(from_path)
            else:
                if P.exists(to_path):
                    if P.getmtime(from_path) < P.getmtime(to_path):
                        continue
                    #
                    # if a file is read-only, then shutil.copy will fail to
                    # overwrite it.  delete it here to be safe.
                    #
                    os.remove(to_path)

                print "`%s' -> `%s'" % (from_path, to_path)
                shutil.copy(from_path, to_path)

if __name__ == "__main__":
    import sys
    regex = re.compile("""
        ^/home/misha/\.anki |
        ^/home/misha/samba |
        ^/home/misha/\.cache |
        ^/home/misha/Desktop |
        ^/home/misha/Downloads |
        ^/home/misha/.dropbox |
        ^/home/misha/Dropbox |
        ^/home/misha/nobackup |
        ^/home/misha/.local/share/trash |
        ^/home/misha/Music |
        ^/home/misha/Pictures |
        ^/home/misha/Pictures2 |
        ^/home/misha/src |
        ^/home/misha/tmp |
        ^/home/misha/\.thumbnails |
        ^/home/misha/\.Trash |
        ^/home/misha/Videos |
        ^/home/misha/\.gvfs 
        """, re.VERBOSE)
    nightly_backup("/home/misha", "/home/misha/samba/nightly-backup", regex)