def test_load_data_power_set(self):
morpheme_loader = MorphemesLoader()
morpheme_loader.use_power_set = True
data = morpheme_loader.load_data()
self.assertGreater(len(data['train']), 0)
self.assertEqual(data['train'][0][5:,].sum(),0) # 5 words, after it only 0s
self.assertEqual(data['train'][0].shape,(self.max_word_length,)) #49 morphemes, max 80 words
self.assertEqual(data['train'][0][0],0) #first word is quotes, this time it's a unique key
def test__set_to_vec_power_set(self):
morpheme_loader = MorphemesLoader()
morpheme_loader.use_power_set = True
morpheme_loader.max_morpheme_count=2
self.assertEqual(morpheme_loader._set_to_vec({0,1}),[0])
self.assertEqual(morpheme_loader._set_to_vec({}),[1])
self.assertEqual(morpheme_loader._set_to_vec({0}),[2])
self.assertEqual(morpheme_loader.max_power_set_key,3)
self.assertEqual(morpheme_loader._set_to_vec({0, 1}), [0])
self.assertEqual(morpheme_loader.max_power_set_key, 3)
def test__get_sentence_vector_power_set(self):
morpheme_loader = MorphemesLoader()
morpheme_loader.use_power_set = True
test_string = """0 1 " _ yyQUOT yyQUOT _ 1
1 2 תהיה היה COP COP gen=F|num=S|per=3 2
2 3 נקמה נקמה NN NN gen=F|num=S 3
3 4 ו ו CONJ CONJ _ 4
4 5 בגדול בגדול RB RB _ 4
5 6 . _ yyDOT yyDOT _ 5"""
test_tensor = morpheme_loader._get_sentence_vector(test_string)
self.assertEqual(test_tensor.shape,(self.max_word_length,))