def test_banana(self):
string = "blah blah blah blah blah blah baabab blah blah"
markov = MarkovModel(list(string))
results = {'a': 0, 'b': 0}
for i in range(1000):
token = markov.get_n_tokens(list("ba"), 1)[0]
results[token] += 1
bwt_log_ratio = self.log_ratio(results, 'a', 'b')
print "BWT: log ratio is", bwt_log_ratio
results = {'a': 0, 'b': 0}
for i in range(1000):
start = randrange(len(string))
substring = None
while (substring != "ba"):
substring = string[start:start+2] if start < (len(string) - 2) else (string[-1:]+string[:1] if start == len(string) - 1 else string[-2:]+string[:0])
start = (start + 1) % len(string)
token = string[start+1]
results[token] += 1
plain_log_ratio = self.log_ratio(results, 'a', 'b')
print "PLAIN: log ratio is", plain_log_ratio
self.assertTrue(abs(bwt_log_ratio)<abs(plain_log_ratio))
def test_markov(self):
markov = MarkovModel(list("Tom Tucker"))
exp_tok = [["o"], ["u"]]
token = markov.get_n_tokens(list("T"), 1)
self.assertIn(token, exp_tok)
tokens = [token for token in markov.get_all_possible_n_grams(list("T"), 1)]
self.assertEqual(tokens, exp_tok)