Comment tidying

[cipher-tools.git] / cipher.py

diff --git a/cipher.py b/cipher.py
index 446a2711bab1aeaa4b6eef951e3f08c85e0f35c8..7b0b13f9ece0c87f995fc2d80e4690ba60588456 100644 (file)
--- a/cipher.py
+++ b/cipher.py
@@ -5,13 +5,14 @@ import logging
 import math
 from itertools import zip_longest
 from segment import segment
+from multiprocessing import Pool
 
 # To time a run:
 #
 # import timeit
 # c5a = open('2012/5a.ciphertext', 'r').read()
 # timeit.timeit('keyword_break(c5a)', setup='gc.enable() ; from __main__ import c5a ; from cipher import keyword_break', number=1)
-
+# timeit.repeat('keyword_break_mp(c5a, chunksize=500)', setup='gc.enable() ; from __main__ import c5a ; from cipher import keyword_break_mp', repeat=5, number=1
 
 logger = logging.getLogger(__name__)
 logger.addHandler(logging.FileHandler('cipher.log'))
@@ -405,7 +406,6 @@ def affine_break(message, metric=norms.euclidean_distance, target_counts=normali
     logger.info('Affine break best fit with key {0}x+{1} ({2}) gives fit of {3} and decrypt starting: {4}'.format(best_multiplier, best_adder, best_one_based, best_fit, affine_decipher(sanitised_message, best_multiplier, best_adder, best_one_based)[:50]))
     return (best_multiplier, best_adder, best_one_based), best_fit
 
-
 def keyword_break(message, wordlist=keywords, metric=norms.euclidean_distance, target_counts=normalised_english_counts, message_frequency_scaling=norms.normalise):
     """Breaks a keyword substitution cipher using a dictionary and frequency analysis
 
@@ -428,6 +428,24 @@ def keyword_break(message, wordlist=keywords, metric=norms.euclidean_distance, t
     logger.info('Keyword break best fit with key {0} (wrap={1}) gives fit of {2} and decrypt starting: {3}'.format(best_keyword, best_wrap_alphabet, best_fit, sanitise(keyword_decipher(message, best_keyword))[:50]))
     return (best_keyword, best_wrap_alphabet), best_fit
 
+def keyword_break_mp(message, wordlist=keywords, metric=norms.euclidean_distance, target_counts=normalised_english_counts, message_frequency_scaling=norms.normalise, chunksize=500):
+    """Breaks a keyword substitution cipher using a dictionary and frequency analysis
+
+    >>> keyword_break_mp(keyword_encipher('this is a test message for the keyword decipherment', 'elephant', 1), wordlist=['cat', 'elephant', 'kangaroo']) # doctest: +ELLIPSIS
+    (('elephant', 1), 0.41643991598441...)
+    """
+    with Pool() as pool:
+        helper_args = [(message, word, wrap, metric, target_counts, message_frequency_scaling) for word in wordlist for wrap in range(3)]
+        breaks = pool.starmap(keyword_break_one, helper_args, chunksize) # Gotcha: the helper function here needs to be defined at the top level (limitation of Pool.starmap)
+        return min(breaks, key=lambda k: k[1])
+
+def keyword_break_one(message, keyword, wrap_alphabet, metric, target_counts, message_frequency_scaling):
+    plaintext = keyword_decipher(message, keyword, wrap_alphabet)
+    counts = message_frequency_scaling(letter_frequencies(plaintext))
+    fit = metric(target_counts, counts)
+    logger.debug('Keyword break attempt using key {0} (wrap={1}) gives fit of {2} and decrypt starting: {3}'.format(keyword, wrap_alphabet, fit, sanitise(plaintext)[:50]))
+    return (keyword, wrap_alphabet), fit
+
 def scytale_break(message, metric=norms.euclidean_distance, target_counts=normalised_english_bigram_counts, message_frequency_scaling=norms.normalise):
     """Breaks a Scytale cipher