def test_get_similarity_tuples(self):
keyword = k.Keyword('dog')
raw_score = np.float32(0.1234)
similarity = s.Similarity(
k.Keyword('cat'), s.SimilarityScore(raw_score))
similarities = s.Similarities([similarity])
actual = s.SimilarKeywords(
keyword, similarities).get_similarity_tuples()
self.assertEqual(actual, [('cat', raw_score)])
def test_write_similar_kewyords(self):
keyword = k.Keyword('test')
similarity = si.Similarity(k.Keyword('flower'),
si.SimilarityScore(np.float32(0.1)))
similarities = si.Similarities([similarity])
similar_keywords = si.SimilarKeywords(keyword, similarities)
similarities_iterator = iter([similar_keywords])
self.target.write_similar_keywords(similarities_iterator)
with open(self.filename) as f:
actual = [line.strip() for line in f.readlines()]
expected = ['keyword,neighbor 1,similarity 1', 'test,flower,0.1']
self.assertEqual(actual, expected)
def test_create_from_tuples(self):
score = np.float32(0.2)
similarities = iter([('a', score)])
actual = s.Similarities.create_from_tuples(similarities)
self.assertEqual(actual, s.Similarities(
[s.Similarity(k.Keyword('a'),
s.SimilarityScore(score))]))
def test_is_acknowledged_keyword(self):
"""Accept Keyword value."""
self.query_keyed_vectors.__contains__.return_value = True
raw_keyword = 'キーワード'
keyword = k.Keyword(raw_keyword)
actual = self.target.is_acknowledged(keyword)
self.assertTrue(actual)
self.assertEqual(self.query_keyed_vectors.__contains__.call_args_list,
[call(raw_keyword)])
def test_to_query_vectors_multi(self):
query = q.Query([k.Keyword('a'), k.Keyword('quick')])
def contains(k):
return True
def getitem(k):
if k == 'a':
return np.array([0.1, 0.2, 0.3])
if k == 'quick':
return np.array([0.3, 0.2, 0.1])
self.query_keyed_vectors.__contains__.side_effect = contains
self.query_keyed_vectors.__getitem__.side_effect = getitem
actual = self.target._to_query_vectors(query)
self.assertEqual(actual.shape, (2, 3))
self.assertTrue(
(np.linalg.norm(actual, ord=2, axis=1) - np.array([1.0, 1.0]) <
0.001).all())
def test_to_query_vectors_one(self):
"""_to_query_vectors can handle a query with one keyword."""
query = q.Query([k.Keyword('a')])
def contains(k):
return True
def getitem(k):
if k == 'a':
return np.array([0.1, 0.2, 0.3])
self.query_keyed_vectors.__contains__.side_effect = contains
self.query_keyed_vectors.__getitem__.side_effect = getitem
actual = self.target._to_query_vectors(query)
self.assertEqual(actual.shape, (1, 3))
np.testing.assert_approx_equal(np.linalg.norm(actual, ord=2, axis=1),
np.array([1.0]))
def test_rank(self, to_embedding_document, to_query_vectors):
query = q.Query([k.Keyword('a')])
document = d.Document(['a', 'quick'])
self.assertEqual(self.target.rank(query, document),
si.SimilarityScore(np.float32(14.0)))