Approximate file renaming script

approx-rename.py

#!/usr/bin/env python2

import sys, os, shutil, argparse, string, readline
from fuzzywuzzy import process

parser = argparse.ArgumentParser()
parser.add_argument('directory', type=str)
parser.add_argument('namelist', type=argparse.FileType(mode='r'))
# TODO: option to include extension in renaming?
args = parser.parse_args()

normal_characters = string.ascii_letters + string.digits
def normalize(s):
  return s
  ret = ''
  for c in s:
    if c in normal_characters:
      ret += c
  return ret

namelist = [x.strip() for x in args.namelist.readlines()]
normalized = {x: normalize(x) for x in namelist}
oldnames = list(os.listdir(args.directory))
for old in oldnames:
  if not os.path.isfile(os.path.join(args.directory, old)):
    print old, "not a file?"
    sys.exit(1)

mapping = {}
gave_up = []
for old in oldnames:
  normal = normalize(old.partition('.')[0])
  matches = process.extract(normal, normalized, limit=5)
  print old
  for m in matches:
    answer = '?'
    while answer not in ('y', 'Y', 'n', 'N'):
      answer = raw_input('{} {} y/n? '.format(m[1], m[2]))
    if answer in ('y', 'Y'):
      mapping[old] = m[2]
      break
  else:
    print 'Giving up'
    gave_up.append(old)

inverse = {}
for g in gave_up:
  print 'gave up:', g
for n in namelist:
  if n not in mapping.itervalues():
    print 'unused: '+n
for k, v in mapping.iteritems():
  inverse.setdefault(v, []).append(k)
for k, v in inverse.iteritems():
  if len(v) > 1:
    print 'clash: ', k, v
    for q in v:
      del mapping[q]

for k, v in mapping.iteritems():
  name, dot, ext = k.partition('.')
  newname = v + dot + ext
  shutil.move(os.path.join(args.directory, k), os.path.join(args.directory, newname))