find_best_caesar_break_parameters-2.py

   1 import random
   2 import collections
   3 from cipher import *
   4 from cipherbreak import *
   5 import itertools
   6
   7 print('Loading...')
   8
   9 corpus = sanitise(''.join([open('shakespeare.txt', 'r').read(),
  10     open('sherlock-holmes.txt', 'r').read(),
  11     open('war-and-peace.txt', 'r').read()]))
  12 corpus_length = len(corpus)
  13
  14 euclidean_scaled_english_counts = norms.euclidean_scale(english_counts)
  15
  16 metrics = [{'func': norms.l1, 'name': 'l1'},
  17     {'func': norms.l2, 'name': 'l2'},
  18     {'func': norms.l3, 'name': 'l2'},
  19     {'func': norms.cosine_distance, 'name': 'cosine_distance'},
  20     {'func': norms.harmonic_mean, 'name': 'harminic_mean'},
  21     {'func': norms.geometric_mean, 'name': 'geometric_mean'},
  22     {'func': norms.inverse_log_pl, 'name': 'inverse_log_pl'}]
  23 scalings = [{'corpus_frequency': normalised_english_counts,
  24          'scaling': norms.normalise,
  25          'name': 'normalised'},
  26         {'corpus_frequency': euclidean_scaled_english_counts,
  27          'scaling': norms.euclidean_scale,
  28          'name': 'euclidean_scaled'},
  29          {'corpus_frequency': normalised_english_counts,
  30          'scaling': norms.identity_scale,
  31          'name': 'normalised_with_identity'}]
  32 message_lengths = [300, 100, 50, 30, 20, 10, 5]
  33
  34 trials = 5000
  35
  36 scores = collections.defaultdict(int)
  37
  38 def eval_all():
  39     list(itertools.starmap(eval_one_parameter_set,
  40         itertools.product(metrics, scalings, message_lengths)))
  41
  42 def eval_one_parameter_set(metric, scaling, message_length):
  43     for i in range(trials):
  44         sample_start = random.randint(0, corpus_length - message_length)
  45         sample = corpus[sample_start:(sample_start + message_length)]
  46         key = random.randint(1, 25)
  47         sample_ciphertext = caesar_encipher(sample, key)
  48         (found_key, score) = caesar_break(sample_ciphertext,
  49                                           metric=metric['func'],
  50                                           target_counts=scaling['corpus_frequency'],
  51                                           message_frequency_scaling=scaling['scaling'])
  52         if found_key == key:
  53             scores[(metric['name'], scaling['name'], message_length)] += 1
  54     return scores[(metric['name'], scaling['name'], message_length)]
  55
  56 def show_results():
  57     with open('caesar_break_parameter_trials.csv', 'w') as f:
  58         for (k, v) in scores.items():
  59             print(str(k)[1:-1], v, sep=",", file=f)
  60
  61 eval_all()
  62 show_results()