def test_calculate_log_probabilities_empty_frequencies(self):
ngram = NGramTrie(2)
ngram.n_gram_frequencies = {}
actual = ngram.calculate_log_probabilities()
self.assertEqual(ngram.n_gram_log_probabilities, {})
self.assertEqual(1, actual)
def test_calculate_log_probabilities_one_bi_gram(self):
ngram = NGramTrie(2)
ngram.n_gram_frequencies = {(1, 2): 10}
actual = ngram.calculate_log_probabilities()
self.assertEqual(ngram.n_gram_log_probabilities[(1, 2)], 0.0)
self.assertEqual(0, actual)
def test_top_n_grams_more(self):
ngram = NGramTrie(2)
top_n = 2000000
ngram.n_gram_frequencies = {(1, 2): 100, (2, 3): 123, (3, 4): 12345}
expected = ((3, 4), (2, 3), (1, 2))
actual = ngram.top_n_grams(top_n)
self.assertEqual(expected, actual)
def test_calculate_log_probabilities_ideal(self):
ngram = NGramTrie(2)
ngram.n_gram_frequencies = {(1, 2): 10, (1, 3): 2, (2, 5): 5}
first_prob = math.log(10 / 12)
second_prob = math.log(2 / 12)
actual = ngram.calculate_log_probabilities()
self.assertEqual(ngram.n_gram_log_probabilities[(1, 2)], first_prob)
self.assertEqual(ngram.n_gram_log_probabilities[(1, 3)], second_prob)
self.assertEqual(0, actual)