crypto.py

#!/usr/bin/env python3

import binascii
import string
from collections import Counter

total_chars = 0

lines = []
result = []
with open("ciphers") as fin:
    for line in fin:
        s = line.strip()
        lines.append(binascii.unhexlify(s))
        result.append([None for c in s])
        total_chars += len(s)

freq_table = [(' ', 0.183),
        ('e', 0.103),
        ('t', 0.075),
        ('a', 0.065),
        ('o', 0.062),
        ('i', 0.057),
        ('s', 0.053)]

char_count = Counter()

for char, freq in freq_table:
    counter = Counter()
    for i in range(len(lines)):
        for j, b in enumerate(lines[i]): 
            if result[i][j] is not None:
                del counter[(i, j)]
                continue

            for k in range(len(lines)):
                if (len(lines[k]) > j):
                    xor_ed = chr(b ^ ord(char) ^ lines[k][j])
                    if xor_ed in string.ascii_lowercase or xor_ed == ' ':
                        counter += Counter({(i, j): 100})
                    elif xor_ed in string.ascii_uppercase:
                        counter += Counter({(i, j): 20 if j != 0 else 100})
                    elif ord(xor_ed) >= 128:
                        del counter[(i, j)]
                        break
                    elif xor_ed in {',', '.'}:
                        counter += Counter({(i, j): 20})
                    elif str.isprintable(xor_ed):
                        counter += Counter({(i, j): 5})

    for (i, j), count in counter.most_common():
        if result[i][j] == None and count > 0:
            for k in range(len(lines)):
                if (len(lines[k]) > j):
                    xor_ed = chr(lines[i][j] ^ ord(char) ^ lines[k][j])
                    char_count.update([xor_ed])
                    result[k][j] = xor_ed

        if char_count[char] > total_chars*freq:
            break

for i in range(len(lines)):
    for j in range(len(lines[i])):
        print(result[i][j] or '?', end='')
    print()