--- /dev/null
+{
+ "metadata": {
+ "name": "Untitled0"
+ },
+ "nbformat": 3,
+ "nbformat_minor": 0,
+ "worksheets": []
+}
\ No newline at end of file
+++ /dev/null
-import string
-import collections
-import norms
-import logging
-from itertools import zip_longest, cycle
-from segment import segment, Pwords
-from multiprocessing import Pool
-
-from cipher import *
-
-# 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)
-
-
-english_counts = collections.defaultdict(int)
-with open('count_1l.txt', 'r') as f:
- for line in f:
- (letter, count) = line.split("\t")
- english_counts[letter] = int(count)
-normalised_english_counts = norms.normalise(english_counts)
-
-english_bigram_counts = collections.defaultdict(int)
-with open('count_2l.txt', 'r') as f:
- for line in f:
- (bigram, count) = line.split("\t")
- english_bigram_counts[bigram] = int(count)
-normalised_english_bigram_counts = norms.normalise(english_bigram_counts)
-
-english_trigram_counts = collections.defaultdict(int)
-with open('count_3l.txt', 'r') as f:
- for line in f:
- (trigram, count) = line.split("\t")
- english_trigram_counts[trigram] = int(count)
-normalised_english_trigram_counts = norms.normalise(english_trigram_counts)
-
-
-with open('words.txt', 'r') as f:
- keywords = [line.rstrip() for line in f]
-
-transpositions = collections.defaultdict(list)
-for word in keywords:
- transpositions[transpositions_of(word)] += [word]
-
-def frequencies(text):
- """Count the number of occurrences of each character in text
-
- >>> sorted(frequencies('abcdefabc').items())
- [('a', 2), ('b', 2), ('c', 2), ('d', 1), ('e', 1), ('f', 1)]
- >>> sorted(frequencies('the quick brown fox jumped over the lazy ' \
- 'dog').items()) # doctest: +NORMALIZE_WHITESPACE
- [(' ', 8), ('a', 1), ('b', 1), ('c', 1), ('d', 2), ('e', 4), ('f', 1),
- ('g', 1), ('h', 2), ('i', 1), ('j', 1), ('k', 1), ('l', 1), ('m', 1),
- ('n', 1), ('o', 4), ('p', 1), ('q', 1), ('r', 2), ('t', 2), ('u', 2),
- ('v', 1), ('w', 1), ('x', 1), ('y', 1), ('z', 1)]
- >>> sorted(frequencies('The Quick BROWN fox jumped! over... the ' \
- '(9lazy) DOG').items()) # doctest: +NORMALIZE_WHITESPACE
- [(' ', 8), ('!', 1), ('(', 1), (')', 1), ('.', 3), ('9', 1), ('B', 1),
- ('D', 1), ('G', 1), ('N', 1), ('O', 2), ('Q', 1), ('R', 1), ('T', 1),
- ('W', 1), ('a', 1), ('c', 1), ('d', 1), ('e', 4), ('f', 1), ('h', 2),
- ('i', 1), ('j', 1), ('k', 1), ('l', 1), ('m', 1), ('o', 2), ('p', 1),
- ('r', 1), ('t', 1), ('u', 2), ('v', 1), ('x', 1), ('y', 1), ('z', 1)]
- >>> sorted(frequencies(sanitise('The Quick BROWN fox jumped! over... ' \
- 'the (9lazy) DOG')).items()) # doctest: +NORMALIZE_WHITESPACE
- [('a', 1), ('b', 1), ('c', 1), ('d', 2), ('e', 4), ('f', 1), ('g', 1),
- ('h', 2), ('i', 1), ('j', 1), ('k', 1), ('l', 1), ('m', 1), ('n', 1),
- ('o', 4), ('p', 1), ('q', 1), ('r', 2), ('t', 2), ('u', 2), ('v', 1),
- ('w', 1), ('x', 1), ('y', 1), ('z', 1)]
- >>> frequencies('abcdefabcdef')['x']
- 0
- """
- #counts = collections.defaultdict(int)
- #for c in text:
- # counts[c] += 1
- #return counts
- return collections.Counter(c for c in text)
-letter_frequencies = frequencies
-
-
-
-def caesar_break(message,
- metric=norms.euclidean_distance,
- target_counts=normalised_english_counts,
- message_frequency_scaling=norms.normalise):
- """Breaks a Caesar cipher using frequency analysis
-
- >>> caesar_break('ibxcsyorsaqcheyklxivoexlevmrimwxsfiqevvmihrsasrxliwyrh' \
- 'ecjsppsamrkwleppfmergefifvmhixscsymjcsyqeoixlm') # doctest: +ELLIPSIS
- (4, 0.080345432737...)
- >>> caesar_break('wxwmaxdgheetgwuxztgptedbgznitgwwhpguxyhkxbmhvvtlbhgtee' \
- 'raxlmhiixweblmxgxwmhmaxybkbgztgwztsxwbgmxgmert') # doctest: +ELLIPSIS
- (19, 0.11189290326...)
- >>> caesar_break('yltbbqnqnzvguvaxurorgenafsbezqvagbnornfgsbevpnaabjurer' \
- 'svaquvzyvxrnznazlybequrvfohgriraabjtbaruraprur') # doctest: +ELLIPSIS
- (13, 0.08293968842...)
- """
- sanitised_message = sanitise(message)
- best_shift = 0
- best_fit = float("inf")
- for shift in range(26):
- plaintext = caesar_decipher(sanitised_message, shift)
- counts = message_frequency_scaling(letter_frequencies(plaintext))
- fit = metric(target_counts, counts)
- logger.debug('Caesar break attempt using key {0} gives fit of {1} '
- 'and decrypt starting: {2}'.format(shift, fit, plaintext[:50]))
- if fit < best_fit:
- best_fit = fit
- best_shift = shift
- logger.info('Caesar break best fit: key {0} gives fit of {1} and '
- 'decrypt starting: {2}'.format(best_shift, best_fit,
- caesar_decipher(sanitised_message, best_shift)[:50]))
- return best_shift, best_fit
-
-def affine_break(message,
- metric=norms.euclidean_distance,
- target_counts=normalised_english_counts,
- message_frequency_scaling=norms.normalise):
- """Breaks an affine cipher using frequency analysis
-
- >>> affine_break('lmyfu bkuusd dyfaxw claol psfaom jfasd snsfg jfaoe ls ' \
- 'omytd jlaxe mh jm bfmibj umis hfsul axubafkjamx. ls kffkxwsd jls ' \
- 'ofgbjmwfkiu olfmxmtmwaokttg jlsx ls kffkxwsd jlsi zg tsxwjl. jlsx ' \
- 'ls umfjsd jlsi zg hfsqysxog. ls dmmdtsd mx jls bats mh bkbsf. ls ' \
- 'bfmctsd kfmyxd jls lyj, mztanamyu xmc jm clm cku tmmeaxw kj lai kxd ' \
- 'clm ckuxj.') # doctest: +ELLIPSIS
- ((15, 22, True), 0.0598745365924...)
- """
- sanitised_message = sanitise(message)
- best_multiplier = 0
- best_adder = 0
- best_one_based = True
- best_fit = float("inf")
- for one_based in [True, False]:
- for multiplier in range(1, 26, 2):
- for adder in range(26):
- plaintext = affine_decipher(sanitised_message,
- multiplier, adder, one_based)
- counts = message_frequency_scaling(letter_frequencies(plaintext))
- fit = metric(target_counts, counts)
- logger.debug('Affine break attempt using key {0}x+{1} ({2}) '
- 'gives fit of {3} and decrypt starting: {4}'.
- format(multiplier, adder, one_based, fit,
- plaintext[:50]))
- if fit < best_fit:
- best_fit = fit
- best_multiplier = multiplier
- best_adder = adder
- best_one_based = one_based
- 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
-
- >>> keyword_break(keyword_encipher('this is a test message for the ' \
- 'keyword decipherment', 'elephant', 1), \
- wordlist=['cat', 'elephant', 'kangaroo']) # doctest: +ELLIPSIS
- (('elephant', 1), 0.1066453448861...)
- """
- best_keyword = ''
- best_wrap_alphabet = True
- best_fit = float("inf")
- for wrap_alphabet in range(3):
- for keyword in wordlist:
- 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]))
- if fit < best_fit:
- best_fit = fit
- best_keyword = keyword
- best_wrap_alphabet = wrap_alphabet
- 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,
- best_wrap_alphabet))[: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), 0.106645344886...)
- """
- with Pool() as pool:
- helper_args = [(message, word, wrap, metric, target_counts,
- message_frequency_scaling)
- for word in wordlist for wrap in range(3)]
- # Gotcha: the helper function here needs to be defined at the top level
- # (limitation of Pool.starmap)
- breaks = pool.starmap(keyword_break_worker, helper_args, chunksize)
- return min(breaks, key=lambda k: k[1])
-
-def keyword_break_worker(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
-
- >>> scytale_break('tfeulchtrtteehwahsdehneoifeayfsondmwpltmaoalhikotoere' \
- 'dcweatehiplwxsnhooacgorrcrcraotohsgullasenylrendaianeplscdriioto' \
- 'aek') # doctest: +ELLIPSIS
- (6, 0.092599933059...)
- """
- best_key = 0
- best_fit = float("inf")
- ngram_length = len(next(iter(target_counts.keys())))
- for key in range(1, 20):
- if len(message) % key == 0:
- plaintext = scytale_decipher(message, key)
- counts = message_frequency_scaling(frequencies(
- ngrams(sanitise(plaintext), ngram_length)))
- fit = metric(target_counts, counts)
- logger.debug('Scytale break attempt using key {0} gives fit of '
- '{1} and decrypt starting: {2}'.format(key,
- fit, sanitise(plaintext)[:50]))
- if fit < best_fit:
- best_fit = fit
- best_key = key
- logger.info('Scytale break best fit with key {0} gives fit of {1} and '
- 'decrypt starting: {2}'.format(best_key, best_fit,
- sanitise(scytale_decipher(message, best_key))[:50]))
- return best_key, best_fit
-
-def column_transposition_break(message,
- translist=transpositions,
- metric=norms.euclidean_distance,
- target_counts=normalised_english_bigram_counts,
- message_frequency_scaling=norms.normalise):
- """Breaks a column transposition cipher using a dictionary and
- n-gram frequency analysis
-
- >>> column_transposition_break(column_transposition_encipher(sanitise( \
- "It is a truth universally acknowledged, that a single man in \
- possession of a good fortune, must be in want of a wife. However \
- little known the feelings or views of such a man may be on his \
- first entering a neighbourhood, this truth is so well fixed in the \
- minds of the surrounding families, that he is considered the \
- rightful property of some one or other of their daughters."), \
- 'encipher'), \
- translist={(2, 0, 5, 3, 1, 4, 6): ['encipher'], \
- (5, 0, 6, 1, 3, 4, 2): ['fourteen'], \
- (6, 1, 0, 4, 5, 3, 2): ['keyword']}) # doctest: +ELLIPSIS
- ((2, 0, 5, 3, 1, 4, 6), 0.0628106372...)
- >>> column_transposition_break(column_transposition_encipher(sanitise( \
- "It is a truth universally acknowledged, that a single man in \
- possession of a good fortune, must be in want of a wife. However \
- little known the feelings or views of such a man may be on his \
- first entering a neighbourhood, this truth is so well fixed in the \
- minds of the surrounding families, that he is considered the \
- rightful property of some one or other of their daughters."), \
- 'encipher'), \
- translist={(2, 0, 5, 3, 1, 4, 6): ['encipher'], \
- (5, 0, 6, 1, 3, 4, 2): ['fourteen'], \
- (6, 1, 0, 4, 5, 3, 2): ['keyword']}, \
- target_counts=normalised_english_trigram_counts) # doctest: +ELLIPSIS
- ((2, 0, 5, 3, 1, 4, 6), 0.0592259560...)
- """
- best_transposition = ''
- best_fit = float("inf")
- ngram_length = len(next(iter(target_counts.keys())))
- for transposition in translist.keys():
- if len(message) % len(transposition) == 0:
- plaintext = column_transposition_decipher(message, transposition)
- counts = message_frequency_scaling(frequencies(
- ngrams(sanitise(plaintext), ngram_length)))
- fit = metric(target_counts, counts)
- logger.debug('Column transposition break attempt using key {0} '
- 'gives fit of {1} and decrypt starting: {2}'.format(
- translist[transposition][0], fit,
- sanitise(plaintext)[:50]))
- if fit < best_fit:
- best_fit = fit
- best_transposition = transposition
- logger.info('Column transposition break best fit with key {0} gives fit '
- 'of {1} and decrypt starting: {2}'.format(
- translist[best_transposition][0],
- best_fit, sanitise(
- column_transposition_decipher(message,
- best_transposition))[:50]))
- return best_transposition, best_fit
-
-
-def column_transposition_break_mp(message,
- translist=transpositions,
- metric=norms.euclidean_distance,
- target_counts=normalised_english_bigram_counts,
- message_frequency_scaling=norms.normalise,
- chunksize=500):
- """Breaks a column transposition cipher using a dictionary and
- n-gram frequency analysis
-
- >>> column_transposition_break_mp(column_transposition_encipher(sanitise( \
- "It is a truth universally acknowledged, that a single man in \
- possession of a good fortune, must be in want of a wife. However \
- little known the feelings or views of such a man may be on his \
- first entering a neighbourhood, this truth is so well fixed in the \
- minds of the surrounding families, that he is considered the \
- rightful property of some one or other of their daughters."), \
- 'encipher'), \
- translist={(2, 0, 5, 3, 1, 4, 6): ['encipher'], \
- (5, 0, 6, 1, 3, 4, 2): ['fourteen'], \
- (6, 1, 0, 4, 5, 3, 2): ['keyword']}) # doctest: +ELLIPSIS
- ((2, 0, 5, 3, 1, 4, 6), 0.0628106372...)
- >>> column_transposition_break_mp(column_transposition_encipher(sanitise( \
- "It is a truth universally acknowledged, that a single man in \
- possession of a good fortune, must be in want of a wife. However \
- little known the feelings or views of such a man may be on his \
- first entering a neighbourhood, this truth is so well fixed in the \
- minds of the surrounding families, that he is considered the \
- rightful property of some one or other of their daughters."), \
- 'encipher'), \
- translist={(2, 0, 5, 3, 1, 4, 6): ['encipher'], \
- (5, 0, 6, 1, 3, 4, 2): ['fourteen'], \
- (6, 1, 0, 4, 5, 3, 2): ['keyword']}, \
- target_counts=normalised_english_trigram_counts) # doctest: +ELLIPSIS
- ((2, 0, 5, 3, 1, 4, 6), 0.0592259560...)
- """
- ngram_length = len(next(iter(target_counts.keys())))
- with Pool() as pool:
- helper_args = [(message, trans, metric, target_counts, ngram_length,
- message_frequency_scaling)
- for trans in translist.keys()]
- # Gotcha: the helper function here needs to be defined at the top level
- # (limitation of Pool.starmap)
- breaks = pool.starmap(column_transposition_break_worker, helper_args, chunksize)
- return min(breaks, key=lambda k: k[1])
-
-def column_transposition_break_worker(message, transposition, metric, target_counts,
- ngram_length, message_frequency_scaling):
- plaintext = column_transposition_decipher(message, transposition)
- counts = message_frequency_scaling(frequencies(
- ngrams(sanitise(plaintext), ngram_length)))
- fit = metric(target_counts, counts)
- logger.debug('Column transposition break attempt using key {0} '
- 'gives fit of {1} and decrypt starting: {2}'.format(
- transposition, fit,
- sanitise(plaintext)[:50]))
- return transposition, fit
-
-def vigenere_keyword_break(message,
- wordlist=keywords,
- metric=norms.euclidean_distance,
- target_counts=normalised_english_counts,
- message_frequency_scaling=norms.normalise):
- """Breaks a vigenere cipher using a dictionary and
- frequency analysis
-
- >>> vigenere_keyword_break(vigenere_encipher(sanitise('this is a test ' \
- 'message for the vigenere decipherment'), 'cat'), \
- wordlist=['cat', 'elephant', 'kangaroo']) # doctest: +ELLIPSIS
- ('cat', 0.15965224935...)
- """
- best_keyword = ''
- best_fit = float("inf")
- for keyword in wordlist:
- plaintext = vigenere_decipher(message, keyword)
- counts = message_frequency_scaling(letter_frequencies(plaintext))
- fit = metric(target_counts, counts)
- logger.debug('Vigenere break attempt using key {0} '
- 'gives fit of {1} and decrypt starting: {2}'.format(
- keyword, fit,
- sanitise(plaintext)[:50]))
- if fit < best_fit:
- best_fit = fit
- best_keyword = keyword
- logger.info('Vigenere break best fit with key {0} gives fit '
- 'of {1} and decrypt starting: {2}'.format(best_keyword,
- best_fit, sanitise(
- vigenere_decipher(message, best_keyword))[:50]))
- return best_keyword, best_fit
-
-def vigenere_keyword_break_mp(message,
- wordlist=keywords,
- metric=norms.euclidean_distance,
- target_counts=normalised_english_counts,
- message_frequency_scaling=norms.normalise,
- chunksize=500):
- """Breaks a vigenere cipher using a dictionary and
- frequency analysis
-
- >>> vigenere_keyword_break_mp(vigenere_encipher(sanitise('this is a test ' \
- 'message for the vigenere decipherment'), 'cat'), \
- wordlist=['cat', 'elephant', 'kangaroo']) # doctest: +ELLIPSIS
- ('cat', 0.159652249358...)
- """
- with Pool() as pool:
- helper_args = [(message, word, metric, target_counts,
- message_frequency_scaling)
- for word in wordlist]
- # Gotcha: the helper function here needs to be defined at the top level
- # (limitation of Pool.starmap)
- breaks = pool.starmap(vigenere_keyword_break_worker, helper_args, chunksize)
- return min(breaks, key=lambda k: k[1])
-
-def vigenere_keyword_break_worker(message, keyword, metric, target_counts,
- message_frequency_scaling):
- plaintext = vigenere_decipher(message, keyword)
- counts = message_frequency_scaling(letter_frequencies(plaintext))
- fit = metric(target_counts, counts)
- logger.debug('Vigenere keyword break attempt using key {0} gives fit of '
- '{1} and decrypt starting: {2}'.format(keyword,
- fit, sanitise(plaintext)[:50]))
- return keyword, fit
-
-
-
-def vigenere_frequency_break(message,
- metric=norms.euclidean_distance,
- target_counts=normalised_english_counts,
- message_frequency_scaling=norms.normalise):
- """Breaks a Vigenere cipher with frequency analysis
-
- >>> vigenere_frequency_break(vigenere_encipher(sanitise("It is time to " \
- "run. She is ready and so am I. I stole Daniel's pocketbook this " \
- "afternoon when he left his jacket hanging on the easel in the " \
- "attic."), 'florence')) # doctest: +ELLIPSIS
- ('florence', 0.077657073...)
- """
- best_fit = float("inf")
- best_key = ''
- sanitised_message = sanitise(message)
- for trial_length in range(1, 20):
- splits = every_nth(sanitised_message, trial_length)
- key = ''.join([chr(caesar_break(s, target_counts=target_counts)[0] + ord('a')) for s in splits])
- plaintext = vigenere_decipher(sanitised_message, key)
- counts = message_frequency_scaling(frequencies(plaintext))
- fit = metric(target_counts, counts)
- logger.debug('Vigenere key length of {0} ({1}) gives fit of {2}'.
- format(trial_length, key, fit))
- if fit < best_fit:
- best_fit = fit
- best_key = key
- logger.info('Vigenere break best fit with key {0} gives fit '
- 'of {1} and decrypt starting: {2}'.format(best_key,
- best_fit, sanitise(
- vigenere_decipher(message, best_key))[:50]))
- return best_key, best_fit
-
-
-
-if __name__ == "__main__":
- import doctest
- doctest.testmod()
-
--- /dev/null
+import string
+import collections
+import norms
+import logging
+from itertools import zip_longest, cycle
+from segment import segment, Pwords
+from multiprocessing import Pool
+
+from cipher import *
+
+# 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)
+
+
+english_counts = collections.defaultdict(int)
+with open('count_1l.txt', 'r') as f:
+ for line in f:
+ (letter, count) = line.split("\t")
+ english_counts[letter] = int(count)
+normalised_english_counts = norms.normalise(english_counts)
+
+english_bigram_counts = collections.defaultdict(int)
+with open('count_2l.txt', 'r') as f:
+ for line in f:
+ (bigram, count) = line.split("\t")
+ english_bigram_counts[bigram] = int(count)
+normalised_english_bigram_counts = norms.normalise(english_bigram_counts)
+
+english_trigram_counts = collections.defaultdict(int)
+with open('count_3l.txt', 'r') as f:
+ for line in f:
+ (trigram, count) = line.split("\t")
+ english_trigram_counts[trigram] = int(count)
+normalised_english_trigram_counts = norms.normalise(english_trigram_counts)
+
+
+with open('words.txt', 'r') as f:
+ keywords = [line.rstrip() for line in f]
+
+transpositions = collections.defaultdict(list)
+for word in keywords:
+ transpositions[transpositions_of(word)] += [word]
+
+def frequencies(text):
+ """Count the number of occurrences of each character in text
+
+ >>> sorted(frequencies('abcdefabc').items())
+ [('a', 2), ('b', 2), ('c', 2), ('d', 1), ('e', 1), ('f', 1)]
+ >>> sorted(frequencies('the quick brown fox jumped over the lazy ' \
+ 'dog').items()) # doctest: +NORMALIZE_WHITESPACE
+ [(' ', 8), ('a', 1), ('b', 1), ('c', 1), ('d', 2), ('e', 4), ('f', 1),
+ ('g', 1), ('h', 2), ('i', 1), ('j', 1), ('k', 1), ('l', 1), ('m', 1),
+ ('n', 1), ('o', 4), ('p', 1), ('q', 1), ('r', 2), ('t', 2), ('u', 2),
+ ('v', 1), ('w', 1), ('x', 1), ('y', 1), ('z', 1)]
+ >>> sorted(frequencies('The Quick BROWN fox jumped! over... the ' \
+ '(9lazy) DOG').items()) # doctest: +NORMALIZE_WHITESPACE
+ [(' ', 8), ('!', 1), ('(', 1), (')', 1), ('.', 3), ('9', 1), ('B', 1),
+ ('D', 1), ('G', 1), ('N', 1), ('O', 2), ('Q', 1), ('R', 1), ('T', 1),
+ ('W', 1), ('a', 1), ('c', 1), ('d', 1), ('e', 4), ('f', 1), ('h', 2),
+ ('i', 1), ('j', 1), ('k', 1), ('l', 1), ('m', 1), ('o', 2), ('p', 1),
+ ('r', 1), ('t', 1), ('u', 2), ('v', 1), ('x', 1), ('y', 1), ('z', 1)]
+ >>> sorted(frequencies(sanitise('The Quick BROWN fox jumped! over... ' \
+ 'the (9lazy) DOG')).items()) # doctest: +NORMALIZE_WHITESPACE
+ [('a', 1), ('b', 1), ('c', 1), ('d', 2), ('e', 4), ('f', 1), ('g', 1),
+ ('h', 2), ('i', 1), ('j', 1), ('k', 1), ('l', 1), ('m', 1), ('n', 1),
+ ('o', 4), ('p', 1), ('q', 1), ('r', 2), ('t', 2), ('u', 2), ('v', 1),
+ ('w', 1), ('x', 1), ('y', 1), ('z', 1)]
+ >>> frequencies('abcdefabcdef')['x']
+ 0
+ """
+ #counts = collections.defaultdict(int)
+ #for c in text:
+ # counts[c] += 1
+ #return counts
+ return collections.Counter(c for c in text)
+letter_frequencies = frequencies
+
+
+
+def caesar_break(message,
+ metric=norms.euclidean_distance,
+ target_counts=normalised_english_counts,
+ message_frequency_scaling=norms.normalise):
+ """Breaks a Caesar cipher using frequency analysis
+
+ >>> caesar_break('ibxcsyorsaqcheyklxivoexlevmrimwxsfiqevvmihrsasrxliwyrh' \
+ 'ecjsppsamrkwleppfmergefifvmhixscsymjcsyqeoixlm') # doctest: +ELLIPSIS
+ (4, 0.080345432737...)
+ >>> caesar_break('wxwmaxdgheetgwuxztgptedbgznitgwwhpguxyhkxbmhvvtlbhgtee' \
+ 'raxlmhiixweblmxgxwmhmaxybkbgztgwztsxwbgmxgmert') # doctest: +ELLIPSIS
+ (19, 0.11189290326...)
+ >>> caesar_break('yltbbqnqnzvguvaxurorgenafsbezqvagbnornfgsbevpnaabjurer' \
+ 'svaquvzyvxrnznazlybequrvfohgriraabjtbaruraprur') # doctest: +ELLIPSIS
+ (13, 0.08293968842...)
+ """
+ sanitised_message = sanitise(message)
+ best_shift = 0
+ best_fit = float("inf")
+ for shift in range(26):
+ plaintext = caesar_decipher(sanitised_message, shift)
+ counts = message_frequency_scaling(letter_frequencies(plaintext))
+ fit = metric(target_counts, counts)
+ logger.debug('Caesar break attempt using key {0} gives fit of {1} '
+ 'and decrypt starting: {2}'.format(shift, fit, plaintext[:50]))
+ if fit < best_fit:
+ best_fit = fit
+ best_shift = shift
+ logger.info('Caesar break best fit: key {0} gives fit of {1} and '
+ 'decrypt starting: {2}'.format(best_shift, best_fit,
+ caesar_decipher(sanitised_message, best_shift)[:50]))
+ return best_shift, best_fit
+
+def affine_break(message,
+ metric=norms.euclidean_distance,
+ target_counts=normalised_english_counts,
+ message_frequency_scaling=norms.normalise):
+ """Breaks an affine cipher using frequency analysis
+
+ >>> affine_break('lmyfu bkuusd dyfaxw claol psfaom jfasd snsfg jfaoe ls ' \
+ 'omytd jlaxe mh jm bfmibj umis hfsul axubafkjamx. ls kffkxwsd jls ' \
+ 'ofgbjmwfkiu olfmxmtmwaokttg jlsx ls kffkxwsd jlsi zg tsxwjl. jlsx ' \
+ 'ls umfjsd jlsi zg hfsqysxog. ls dmmdtsd mx jls bats mh bkbsf. ls ' \
+ 'bfmctsd kfmyxd jls lyj, mztanamyu xmc jm clm cku tmmeaxw kj lai kxd ' \
+ 'clm ckuxj.') # doctest: +ELLIPSIS
+ ((15, 22, True), 0.0598745365924...)
+ """
+ sanitised_message = sanitise(message)
+ best_multiplier = 0
+ best_adder = 0
+ best_one_based = True
+ best_fit = float("inf")
+ for one_based in [True, False]:
+ for multiplier in range(1, 26, 2):
+ for adder in range(26):
+ plaintext = affine_decipher(sanitised_message,
+ multiplier, adder, one_based)
+ counts = message_frequency_scaling(letter_frequencies(plaintext))
+ fit = metric(target_counts, counts)
+ logger.debug('Affine break attempt using key {0}x+{1} ({2}) '
+ 'gives fit of {3} and decrypt starting: {4}'.
+ format(multiplier, adder, one_based, fit,
+ plaintext[:50]))
+ if fit < best_fit:
+ best_fit = fit
+ best_multiplier = multiplier
+ best_adder = adder
+ best_one_based = one_based
+ 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
+
+ >>> keyword_break(keyword_encipher('this is a test message for the ' \
+ 'keyword decipherment', 'elephant', 1), \
+ wordlist=['cat', 'elephant', 'kangaroo']) # doctest: +ELLIPSIS
+ (('elephant', 1), 0.1066453448861...)
+ """
+ best_keyword = ''
+ best_wrap_alphabet = True
+ best_fit = float("inf")
+ for wrap_alphabet in range(3):
+ for keyword in wordlist:
+ 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]))
+ if fit < best_fit:
+ best_fit = fit
+ best_keyword = keyword
+ best_wrap_alphabet = wrap_alphabet
+ 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,
+ best_wrap_alphabet))[: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), 0.106645344886...)
+ """
+ with Pool() as pool:
+ helper_args = [(message, word, wrap, metric, target_counts,
+ message_frequency_scaling)
+ for word in wordlist for wrap in range(3)]
+ # Gotcha: the helper function here needs to be defined at the top level
+ # (limitation of Pool.starmap)
+ breaks = pool.starmap(keyword_break_worker, helper_args, chunksize)
+ return min(breaks, key=lambda k: k[1])
+
+def keyword_break_worker(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
+
+ >>> scytale_break('tfeulchtrtteehwahsdehneoifeayfsondmwpltmaoalhikotoere' \
+ 'dcweatehiplwxsnhooacgorrcrcraotohsgullasenylrendaianeplscdriioto' \
+ 'aek') # doctest: +ELLIPSIS
+ (6, 0.092599933059...)
+ """
+ best_key = 0
+ best_fit = float("inf")
+ ngram_length = len(next(iter(target_counts.keys())))
+ for key in range(1, 20):
+ if len(message) % key == 0:
+ plaintext = scytale_decipher(message, key)
+ counts = message_frequency_scaling(frequencies(
+ ngrams(sanitise(plaintext), ngram_length)))
+ fit = metric(target_counts, counts)
+ logger.debug('Scytale break attempt using key {0} gives fit of '
+ '{1} and decrypt starting: {2}'.format(key,
+ fit, sanitise(plaintext)[:50]))
+ if fit < best_fit:
+ best_fit = fit
+ best_key = key
+ logger.info('Scytale break best fit with key {0} gives fit of {1} and '
+ 'decrypt starting: {2}'.format(best_key, best_fit,
+ sanitise(scytale_decipher(message, best_key))[:50]))
+ return best_key, best_fit
+
+def column_transposition_break(message,
+ translist=transpositions,
+ metric=norms.euclidean_distance,
+ target_counts=normalised_english_bigram_counts,
+ message_frequency_scaling=norms.normalise):
+ """Breaks a column transposition cipher using a dictionary and
+ n-gram frequency analysis
+
+ >>> column_transposition_break(column_transposition_encipher(sanitise( \
+ "It is a truth universally acknowledged, that a single man in \
+ possession of a good fortune, must be in want of a wife. However \
+ little known the feelings or views of such a man may be on his \
+ first entering a neighbourhood, this truth is so well fixed in the \
+ minds of the surrounding families, that he is considered the \
+ rightful property of some one or other of their daughters."), \
+ 'encipher'), \
+ translist={(2, 0, 5, 3, 1, 4, 6): ['encipher'], \
+ (5, 0, 6, 1, 3, 4, 2): ['fourteen'], \
+ (6, 1, 0, 4, 5, 3, 2): ['keyword']}) # doctest: +ELLIPSIS
+ ((2, 0, 5, 3, 1, 4, 6), 0.0628106372...)
+ >>> column_transposition_break(column_transposition_encipher(sanitise( \
+ "It is a truth universally acknowledged, that a single man in \
+ possession of a good fortune, must be in want of a wife. However \
+ little known the feelings or views of such a man may be on his \
+ first entering a neighbourhood, this truth is so well fixed in the \
+ minds of the surrounding families, that he is considered the \
+ rightful property of some one or other of their daughters."), \
+ 'encipher'), \
+ translist={(2, 0, 5, 3, 1, 4, 6): ['encipher'], \
+ (5, 0, 6, 1, 3, 4, 2): ['fourteen'], \
+ (6, 1, 0, 4, 5, 3, 2): ['keyword']}, \
+ target_counts=normalised_english_trigram_counts) # doctest: +ELLIPSIS
+ ((2, 0, 5, 3, 1, 4, 6), 0.0592259560...)
+ """
+ best_transposition = ''
+ best_fit = float("inf")
+ ngram_length = len(next(iter(target_counts.keys())))
+ for transposition in translist.keys():
+ if len(message) % len(transposition) == 0:
+ plaintext = column_transposition_decipher(message, transposition)
+ counts = message_frequency_scaling(frequencies(
+ ngrams(sanitise(plaintext), ngram_length)))
+ fit = metric(target_counts, counts)
+ logger.debug('Column transposition break attempt using key {0} '
+ 'gives fit of {1} and decrypt starting: {2}'.format(
+ translist[transposition][0], fit,
+ sanitise(plaintext)[:50]))
+ if fit < best_fit:
+ best_fit = fit
+ best_transposition = transposition
+ logger.info('Column transposition break best fit with key {0} gives fit '
+ 'of {1} and decrypt starting: {2}'.format(
+ translist[best_transposition][0],
+ best_fit, sanitise(
+ column_transposition_decipher(message,
+ best_transposition))[:50]))
+ return best_transposition, best_fit
+
+
+def column_transposition_break_mp(message,
+ translist=transpositions,
+ metric=norms.euclidean_distance,
+ target_counts=normalised_english_bigram_counts,
+ message_frequency_scaling=norms.normalise,
+ chunksize=500):
+ """Breaks a column transposition cipher using a dictionary and
+ n-gram frequency analysis
+
+ >>> column_transposition_break_mp(column_transposition_encipher(sanitise( \
+ "It is a truth universally acknowledged, that a single man in \
+ possession of a good fortune, must be in want of a wife. However \
+ little known the feelings or views of such a man may be on his \
+ first entering a neighbourhood, this truth is so well fixed in the \
+ minds of the surrounding families, that he is considered the \
+ rightful property of some one or other of their daughters."), \
+ 'encipher'), \
+ translist={(2, 0, 5, 3, 1, 4, 6): ['encipher'], \
+ (5, 0, 6, 1, 3, 4, 2): ['fourteen'], \
+ (6, 1, 0, 4, 5, 3, 2): ['keyword']}) # doctest: +ELLIPSIS
+ ((2, 0, 5, 3, 1, 4, 6), 0.0628106372...)
+ >>> column_transposition_break_mp(column_transposition_encipher(sanitise( \
+ "It is a truth universally acknowledged, that a single man in \
+ possession of a good fortune, must be in want of a wife. However \
+ little known the feelings or views of such a man may be on his \
+ first entering a neighbourhood, this truth is so well fixed in the \
+ minds of the surrounding families, that he is considered the \
+ rightful property of some one or other of their daughters."), \
+ 'encipher'), \
+ translist={(2, 0, 5, 3, 1, 4, 6): ['encipher'], \
+ (5, 0, 6, 1, 3, 4, 2): ['fourteen'], \
+ (6, 1, 0, 4, 5, 3, 2): ['keyword']}, \
+ target_counts=normalised_english_trigram_counts) # doctest: +ELLIPSIS
+ ((2, 0, 5, 3, 1, 4, 6), 0.0592259560...)
+ """
+ ngram_length = len(next(iter(target_counts.keys())))
+ with Pool() as pool:
+ helper_args = [(message, trans, metric, target_counts, ngram_length,
+ message_frequency_scaling)
+ for trans in translist.keys()]
+ # Gotcha: the helper function here needs to be defined at the top level
+ # (limitation of Pool.starmap)
+ breaks = pool.starmap(column_transposition_break_worker, helper_args, chunksize)
+ return min(breaks, key=lambda k: k[1])
+
+def column_transposition_break_worker(message, transposition, metric, target_counts,
+ ngram_length, message_frequency_scaling):
+ plaintext = column_transposition_decipher(message, transposition)
+ counts = message_frequency_scaling(frequencies(
+ ngrams(sanitise(plaintext), ngram_length)))
+ fit = metric(target_counts, counts)
+ logger.debug('Column transposition break attempt using key {0} '
+ 'gives fit of {1} and decrypt starting: {2}'.format(
+ transposition, fit,
+ sanitise(plaintext)[:50]))
+ return transposition, fit
+
+def vigenere_keyword_break(message,
+ wordlist=keywords,
+ metric=norms.euclidean_distance,
+ target_counts=normalised_english_counts,
+ message_frequency_scaling=norms.normalise):
+ """Breaks a vigenere cipher using a dictionary and
+ frequency analysis
+
+ >>> vigenere_keyword_break(vigenere_encipher(sanitise('this is a test ' \
+ 'message for the vigenere decipherment'), 'cat'), \
+ wordlist=['cat', 'elephant', 'kangaroo']) # doctest: +ELLIPSIS
+ ('cat', 0.15965224935...)
+ """
+ best_keyword = ''
+ best_fit = float("inf")
+ for keyword in wordlist:
+ plaintext = vigenere_decipher(message, keyword)
+ counts = message_frequency_scaling(letter_frequencies(plaintext))
+ fit = metric(target_counts, counts)
+ logger.debug('Vigenere break attempt using key {0} '
+ 'gives fit of {1} and decrypt starting: {2}'.format(
+ keyword, fit,
+ sanitise(plaintext)[:50]))
+ if fit < best_fit:
+ best_fit = fit
+ best_keyword = keyword
+ logger.info('Vigenere break best fit with key {0} gives fit '
+ 'of {1} and decrypt starting: {2}'.format(best_keyword,
+ best_fit, sanitise(
+ vigenere_decipher(message, best_keyword))[:50]))
+ return best_keyword, best_fit
+
+def vigenere_keyword_break_mp(message,
+ wordlist=keywords,
+ metric=norms.euclidean_distance,
+ target_counts=normalised_english_counts,
+ message_frequency_scaling=norms.normalise,
+ chunksize=500):
+ """Breaks a vigenere cipher using a dictionary and
+ frequency analysis
+
+ >>> vigenere_keyword_break_mp(vigenere_encipher(sanitise('this is a test ' \
+ 'message for the vigenere decipherment'), 'cat'), \
+ wordlist=['cat', 'elephant', 'kangaroo']) # doctest: +ELLIPSIS
+ ('cat', 0.159652249358...)
+ """
+ with Pool() as pool:
+ helper_args = [(message, word, metric, target_counts,
+ message_frequency_scaling)
+ for word in wordlist]
+ # Gotcha: the helper function here needs to be defined at the top level
+ # (limitation of Pool.starmap)
+ breaks = pool.starmap(vigenere_keyword_break_worker, helper_args, chunksize)
+ return min(breaks, key=lambda k: k[1])
+
+def vigenere_keyword_break_worker(message, keyword, metric, target_counts,
+ message_frequency_scaling):
+ plaintext = vigenere_decipher(message, keyword)
+ counts = message_frequency_scaling(letter_frequencies(plaintext))
+ fit = metric(target_counts, counts)
+ logger.debug('Vigenere keyword break attempt using key {0} gives fit of '
+ '{1} and decrypt starting: {2}'.format(keyword,
+ fit, sanitise(plaintext)[:50]))
+ return keyword, fit
+
+
+
+def vigenere_frequency_break(message,
+ metric=norms.euclidean_distance,
+ target_counts=normalised_english_counts,
+ message_frequency_scaling=norms.normalise):
+ """Breaks a Vigenere cipher with frequency analysis
+
+ >>> vigenere_frequency_break(vigenere_encipher(sanitise("It is time to " \
+ "run. She is ready and so am I. I stole Daniel's pocketbook this " \
+ "afternoon when he left his jacket hanging on the easel in the " \
+ "attic."), 'florence')) # doctest: +ELLIPSIS
+ ('florence', 0.077657073...)
+ """
+ best_fit = float("inf")
+ best_key = ''
+ sanitised_message = sanitise(message)
+ for trial_length in range(1, 20):
+ splits = every_nth(sanitised_message, trial_length)
+ key = ''.join([chr(caesar_break(s, target_counts=target_counts)[0] + ord('a')) for s in splits])
+ plaintext = vigenere_decipher(sanitised_message, key)
+ counts = message_frequency_scaling(frequencies(plaintext))
+ fit = metric(target_counts, counts)
+ logger.debug('Vigenere key length of {0} ({1}) gives fit of {2}'.
+ format(trial_length, key, fit))
+ if fit < best_fit:
+ best_fit = fit
+ best_key = key
+ logger.info('Vigenere break best fit with key {0} gives fit '
+ 'of {1} and decrypt starting: {2}'.format(best_key,
+ best_fit, sanitise(
+ vigenere_decipher(message, best_key))[:50]))
+ return best_key, best_fit
+
+
+
+if __name__ == "__main__":
+ import doctest
+ doctest.testmod()
+