anagram phrase comparison

anagram_lookup.py

from hashlib import md5
from collections import Counter
import itertools
import sys
import time

MATCH_MSG = '>>> We have a match!!! This phrase matches the mt5 hash'


def get_relevant_words(target):
    # get a set of unique dictionary words
    dictionary = set([w.strip() for w in open('wordlist').readlines()])
    target_counter = Counter(target)

    relevant_words = []

    # a single iteration to reduce the number of words to ~ 1.5k
    for word in dictionary:
        c = Counter(word)
        if set(c) - set(target_counter):
            continue  # contains letters that are not in our target
        if any(1 for k in c if c[k] > target_counter[k]):
            continue  # contains letters that repeat more than in our target
        relevant_words.append(word)


def main(phrase):
    target = sorted(''.join(phrase.split()))
    md5_match = md5(target).hexdigest()
    relevant_words = get_relevant_words(target)

    anagrams = 0
    start = time.time()

    # in increasing difficulty, attempt to
    for n in range(1, 10):
        print('Looking for anagram of "%s" with %s word/s' % (phrase, n))
        for i, words in enumerate(itertools.combinations(relevant_words, n)):
            # check if this word combination is Anagram of target
            if sorted(''.join(words)) == target:
                # look for all permutations of these words
                for anagram_words in itertools.permutations(words):
                    anagrams += 1
                    sentence = ' '.join(anagram_words)
                    if md5(sentence).hexdigest() == md5_match:
                        print('[+] Target anagram found:', sentence,
                              md5(sentence).hexdigest(), MATCH_MSG)
                        return sentence
                    # else:
                    #     print('[-] Anagram found:', sentence,
                    #           md5(sentence).hexdigest())
                    if anagrams % 200 == 0:
                        if anagrams == 0:
                            continue
                        print('processed {} combinations, {} anagrams found, '
                              'in {} seconds ({}/ps, {}/as)'.format(
                                  i, anagrams, int(time.time() - start),
                                  round(i / (time.time() - start)),
                                  round(anagrams / (time.time() - start), 1)))


if __name__ == '__main__':
    main(sys.argv[1])