support/make-cracking-dictionary.py

   1 # import language_models
   2 import utilities
   3
   4 american = set(open('/usr/share/dict/american-english', 'r').readlines())
   5 british = set(open('/usr/share/dict/british-english', 'r').readlines())
   6 cracklib = set(open('/usr/share/dict/cracklib-small', 'r').readlines())
   7
   8 words = american | british | cracklib
   9
  10 # sanitised_words = set()
  11
  12 # for w in words:
  13     # sanitised_words.add(language_models.sanitise(w))
  14
  15 sanitised_words = set(utilities.sanitise(w) for w in words)
  16
  17 sanitised_words.discard('')
  18
  19 with open('words.txt', 'w') as f:
  20     f.write('\n'.join(sorted(sanitised_words, key=lambda w: (len(w), w))))