source: misc/music-files-sanitize.py@ 386

#!/usr/bin/env python
# -*- coding: utf-8 -*-
#
"""
Simple trick to rename all files to be compliant with FAT32 filesystems
Usefull if you like to copy your files from a UFS/EXT filesystem to a FAT
filesystem.

By default it will dryrun to avoid (major) disasters.

Version: $Id$
License: BSDLike http://rickvanderzwet.nl/LICENSE
Rick van der Zwet <info@rickvanderzwet.nl>
"""

import re
import os
import sys
import string
import argparse


parser = argparse.ArgumentParser(description=__doc__,formatter_class=argparse.RawTextHelpFormatter)
parser.add_argument('-r', '--rename', dest="dryrun", action='store_false',help="Rename the files",default=True)
parser.add_argument('directories', nargs='+', metavar='DIR', type=str)
args = parser.parse_args()

print "# INFO: Going to parse directories %s" % args.directories

trans_table = {
  u'é' : 'e',
  u'&' : 'and',
  u'Ø' : 'O',
  u'Ì' : 'u',
  u'’' : "'",
  u'ö' : 'o',
  u'á' : 'a',
}

word_table = {
  "'m"  : ' am',
  "'d"  : ' would',
  "n't" : ' not',
  "n's" : 'n his',
  "'n'" : 'and',
  "n'"  : 'ng',
  "'s"  : ' is',
  "'ll" :  ' will',
  "A'k" : "Als ik",
  "'T"  : "Het",
  "'t"  : "het",
}

replace_re = re.compile('[^a-zA-Z0-9\-\.\ ]+')

def make_ascii(source):
  source = source.decode('utf-8')
  dest = ""
  for i in source:
    if trans_table.has_key(i):
      dest += trans_table[i]
    else:
      dest += i
  for key,value in word_table.iteritems():
    dest = dest.replace(key,value)

  # Debug decoding
  # print "---"
  # print "Original : ", source, 
  # for c in source: print ord(c),
  # print "\nConverted: ", dest, 
  # for c in dest: print ord(c),
  # print "\n"

  dest = replace_re.sub('', dest)
  return str(dest)

for dir_root in args.directories:
  for root, dirs, files in os.walk(dir_root,topdown=False):
    for file in files + dirs:
      new_file = make_ascii(file)
      if new_file != file:
        source = os.path.join(root,file)
        target = os.path.join(root,new_file)
        prefix = '# (dryrun) ' if args.dryrun else ''
        print prefix, "Source:", source
        print prefix, "Target:", target
        if not args.dryrun:
          os.rename(source,target)