Worked on Enigma, mainly changing how the notch positions are handled
[cipher-training.git] / cipherbreak.py
index 6f4d2aed7d037e509df96970891a7b9e1808c088..f1ab58e50a4f293368ef9d5382aef9dabf590420 100644 (file)
@@ -13,11 +13,22 @@ from multiprocessing import Pool
 
 import matplotlib.pyplot as plt
 
-logger = logging.getLogger(__name__)
-logger.addHandler(logging.FileHandler('cipher.log'))
+# logging.basicConfig(filename="cipher.log", level=logging.INFO)
+# logger = logging.getLogger(__name__)
+
+logger = logging.getLogger('cipherbreak')
 logger.setLevel(logging.WARNING)
-#logger.setLevel(logging.INFO)
-#logger.setLevel(logging.DEBUG)
+# logger.setLevel(logging.INFO)
+# logger.setLevel(logging.DEBUG)
+
+# create the logging file handler
+fh = logging.FileHandler("cipher.log")
+formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
+fh.setFormatter(formatter)
+
+# add handler to logger object
+logger.addHandler(fh)
+
 
 from cipher import *
 from language_models import *
@@ -199,23 +210,27 @@ def keyword_break_worker(message, keyword, wrap_alphabet, fitness):
     return (keyword, wrap_alphabet), fit
 
 def monoalphabetic_break_hillclimbing(message, max_iterations=10000000, 
-        fitness=Pletters):
+        alphabet=None, fitness=Pletters):
     ciphertext = unaccent(message).lower()
-    alphabet = list(string.ascii_lowercase)
-    random.shuffle(alphabet)
-    alphabet = ''.join(alphabet)
+    if not alphabet:
+        alphabet = list(string.ascii_lowercase)
+        random.shuffle(alphabet)
+        alphabet = cat(alphabet)
     return monoalphabetic_break_hillclimbing_worker(ciphertext, alphabet,
                                                     max_iterations, fitness)
 
 def monoalphabetic_break_hillclimbing_mp(message, workers=10, 
-        max_iterations = 10000000, fitness=Pletters, chunksize=1):
+        max_iterations = 10000000, alphabet=None, fitness=Pletters, chunksize=1):
     worker_args = []
     ciphertext = unaccent(message).lower()
     for i in range(workers):
-        alphabet = list(string.ascii_lowercase)
-        random.shuffle(alphabet)
-        alphabet = ''.join(alphabet)
-        worker_args.append((ciphertext, alphabet, max_iterations, fitness))
+        if alphabet:
+            this_alphabet = alphabet
+        else:
+            this_alphabet = list(string.ascii_lowercase)
+            random.shuffle(this_alphabet)
+            this_alphabet = cat(this_alphabet)
+        worker_args.append((ciphertext, this_alphabet, max_iterations, fitness))
     with Pool() as pool:
         breaks = pool.starmap(monoalphabetic_break_hillclimbing_worker,
                               worker_args, chunksize)
@@ -286,7 +301,7 @@ def vigenere_frequency_break(message, max_key_length=20, fitness=Pletters):
     """
     def worker(message, key_length, fitness):
         splits = every_nth(sanitised_message, key_length)
-        key = ''.join([chr(caesar_break(s)[0] + ord('a')) for s in splits])
+        key = cat([chr(caesar_break(s)[0] + ord('a')) for s in splits])
         plaintext = vigenere_decipher(message, key)
         fit = fitness(plaintext)
         return key, fit
@@ -310,7 +325,7 @@ def beaufort_frequency_break(message, max_key_length=20, fitness=Pletters):
     """
     def worker(message, key_length, fitness):
         splits = every_nth(sanitised_message, key_length)
-        key = ''.join([chr(-caesar_break(s)[0] % 26 + ord('a'))
+        key = cat([chr(-caesar_break(s)[0] % 26 + ord('a'))
                        for s in splits])
         plaintext = beaufort_decipher(message, key)
         fit = fitness(plaintext)
@@ -355,7 +370,7 @@ def column_transposition_break_mp(message, translist=transpositions,
     with Pool() as pool:
         helper_args = [(message, trans, fillcolumnwise, emptycolumnwise,
                         fitness)
-                       for trans in translist.keys()
+                       for trans in translist
                        for fillcolumnwise in [True, False]
                        for emptycolumnwise in [True, False]]
         # Gotcha: the helper function here needs to be defined at the top level
@@ -416,6 +431,19 @@ def scytale_break_mp(message, max_key_length=20,
 scytale_break = scytale_break_mp
 
 
+def railfence_break(message, max_key_length=20,
+                     fitness=Pletters, chunksize=500):
+    """Breaks a hill cipher using a matrix of given rank and letter frequencies
+
+    
+    """
+    
+    sanitised_message = sanitise(message)
+    results = starmap(worker, [(sanitised_message, i, fitness)
+                               for i in range(2, max_key_length+1)])
+    return max(results, key=lambda k: k[1])
+
+
 def railfence_break(message, max_key_length=20,
                      fitness=Pbigrams, chunksize=500):
     """Breaks a railfence cipher using a range of lengths and
@@ -445,24 +473,138 @@ def railfence_break(message, max_key_length=20,
         plaintext = railfence_decipher(message, height)
         fit = fitness(plaintext)
         return height, fit
+
     sanitised_message = sanitise(message)
     results = starmap(worker, [(sanitised_message, i, fitness)
                                for i in range(2, max_key_length+1)])
     return max(results, key=lambda k: k[1])
 
+def amsco_break(message, translist=transpositions, patterns = [(1, 2), (2, 1)],
+                                  fillstyles = [AmscoFillStyle.continuous, 
+                                                AmscoFillStyle.same_each_row, 
+                                                AmscoFillStyle.reverse_each_row],
+                                  fitness=Pbigrams, 
+                                  chunksize=500):
+    """Breaks an AMSCO transposition cipher using a dictionary and
+    n-gram frequency analysis
 
+    >>> amsco_break(amsco_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']}, \
+        patterns=[(1, 2)]) # doctest: +ELLIPSIS
+    (((2, 0, 5, 3, 1, 4, 6), (1, 2), <AmscoFillStyle.continuous: 1>), -709.4646722...)
+    >>> amsco_break(amsco_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', fillpattern=(2, 1)), \
+        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']}, \
+        patterns=[(1, 2), (2, 1)], fitness=Ptrigrams) # doctest: +ELLIPSIS
+    (((2, 0, 5, 3, 1, 4, 6), (2, 1), <AmscoFillStyle.continuous: 1>), -997.0129085...)
+    """
     with Pool() as pool:
-        helper_args = [(message, trans, False, True, fitness)
-            for trans in
-                [[col for col in range(math.ceil(len(message)/rows))]
-                    for rows in range(1,max_key_length+1)]]
+        helper_args = [(message, trans, pattern, fillstyle, fitness)
+                       for trans in translist
+                       for pattern in patterns
+                       for fillstyle in fillstyles]
         # 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)
-        best = max(breaks, key=lambda k: k[1])
-        return math.trunc(len(message) / len(best[0][0])), best[1]
-scytale_break = scytale_break_mp
+        breaks = pool.starmap(amsco_break_worker, helper_args, chunksize) 
+        return max(breaks, key=lambda k: k[1])
+
+def amsco_break_worker(message, transposition,
+        pattern, fillstyle, fitness):
+    plaintext = amsco_transposition_decipher(message, transposition,
+        fillpattern=pattern, fillstyle=fillstyle)
+    fit = fitness(sanitise(plaintext))
+    logger.debug('AMSCO transposition break attempt using key {0} and pattern'
+                         '{1} ({2}) gives fit of {3} and decrypt starting: '
+                         '{4}'.format(
+                             transposition, pattern, fillstyle, fit, 
+                             sanitise(plaintext)[:50]))
+    return (transposition, pattern, fillstyle), fit
+
+
+def hill_break(message, matrix_size=2, fitness=Pletters, 
+    number_of_solutions=1, chunksize=500):
+
+    all_matrices = [np.matrix(list(m)) 
+        for m in itertools.product([list(r) 
+            for r in itertools.product(range(26), repeat=matrix_size)], 
+        repeat=matrix_size)]
+    valid_matrices = [m for m, d in 
+        zip(all_matrices, (int(round(linalg.det(m))) for m in all_matrices))
+                  if d != 0
+                  if d % 2 != 0
+                  if d % 13 != 0 ]
+    with Pool() as pool:
+        helper_args = [(message, matrix, fitness)
+                       for matrix in valid_matrices]
+        # Gotcha: the helper function here needs to be defined at the top level
+        #   (limitation of Pool.starmap)
+        breaks = pool.starmap(hill_break_worker, helper_args, chunksize)
+        if number_of_solutions == 1:
+            return max(breaks, key=lambda k: k[1])
+        else:
+            return sorted(breaks, key=lambda k: k[1], reverse=True)[:number_of_solutions]
+
+def hill_break_worker(message, matrix, fitness):
+    plaintext = hill_decipher(matrix, message)
+    fit = fitness(plaintext)
+    logger.debug('Hill cipher break attempt using key {0} gives fit of '
+                 '{1} and decrypt starting: {2}'.format(matrix, 
+                     fit, sanitise(plaintext)[:50]))
+    return matrix, fit
+
+def bifid_break_mp(message, wordlist=keywords, fitness=Pletters, max_period=10,
+                     number_of_solutions=1, chunksize=500):
+    """Breaks a keyword substitution cipher using a dictionary and
+    frequency analysis
+
+    >>> bifid_break_mp(bifid_encipher('this is a test message for the ' \
+          'keyword decipherment', 'elephant', wrap_alphabet=KeywordWrapAlphabet.from_last), \
+          wordlist=['cat', 'elephant', 'kangaroo']) # doctest: +ELLIPSIS
+    (('elephant', <KeywordWrapAlphabet.from_last: 2>, 0), -52.834575011...)
+    >>> bifid_break_mp(bifid_encipher('this is a test message for the ' \
+          'keyword decipherment', 'elephant', wrap_alphabet=KeywordWrapAlphabet.from_last), \
+          wordlist=['cat', 'elephant', 'kangaroo'], \
+          number_of_solutions=2) # doctest: +ELLIPSIS, +NORMALIZE_WHITESPACE
+    [(('elephant', <KeywordWrapAlphabet.from_last: 2>, 0), -52.834575011...), 
+    (('elephant', <KeywordWrapAlphabet.from_largest: 3>, 0), -52.834575011...)]
+    """
+    with Pool() as pool:
+        helper_args = [(message, word, wrap, period, fitness)
+                       for word in wordlist
+                       for wrap in KeywordWrapAlphabet
+                       for period in range(max_period+1)]
+        # Gotcha: the helper function here needs to be defined at the top level
+        #   (limitation of Pool.starmap)
+        breaks = pool.starmap(bifid_break_worker, helper_args, chunksize)
+        if number_of_solutions == 1:
+            return max(breaks, key=lambda k: k[1])
+        else:
+            return sorted(breaks, key=lambda k: k[1], reverse=True)[:number_of_solutions]
+
+def bifid_break_worker(message, keyword, wrap_alphabet, period, fitness):
+    plaintext = bifid_decipher(message, keyword, wrap_alphabet, period=period)
+    fit = fitness(plaintext)
+    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, period), fit
 
 
 def pocket_enigma_break_by_crib(message, wheel_spec, crib, crib_position):
@@ -494,13 +636,13 @@ def pocket_enigma_break_by_crib(message, wheel_spec, crib, crib_position):
 
 
 def plot_frequency_histogram(freqs, sort_key=None):
-    x = range(len(freqs.keys()))
-    y = [freqs[l] for l in sorted(freqs.keys(), key=sort_key)]
+    x = range(len(freqs))
+    y = [freqs[l] for l in sorted(freqs, key=sort_key)]
     f = plt.figure()
     ax = f.add_axes([0.1, 0.1, 0.9, 0.9])
     ax.bar(x, y, align='center')
     ax.set_xticks(x)
-    ax.set_xticklabels(sorted(freqs.keys(), key=sort_key))
+    ax.set_xticklabels(sorted(freqs, key=sort_key))
     f.show()