def test_sequence_length_two_returns_distinct_observed_ngrams(self):
sequences = create_sequence_list(self.short_corpus, 2)
self.assertEqual(set(sequences), set([
('alpha', 'beta'),
('beta', 'gamma'),
('alef','bet'),
('bet','gimel')]))
def test_sequence_length_one_returns_distinct_words(self):
sequences = create_sequence_list(self.corpus, 1)
self.assertEqual(set(sequences), set([(x,) for y in self.corpus for x in y]))
self.assertEqual(len(sequences), len(set([(x,) for y in self.corpus for x in y])))