Code Duplication - milvus-io/milvus - Measure and Improve Code Quality continuously with Scrutinizer

Code Duplication Length = 21-24 lines in 6 locations

tests/milvus_python_test/collection/test_collection_count.py 4 locations


        status, res = connect.count_entities(ham_collection)
        assert res == 0

    def test_collection_rows_count_multi_collections(self, connect):
        '''
        target: test collection rows_count is correct or not with multiple collections of IP
        method: create collection and add vectors in it,
            assert the value returned by count_entities method is equal to length of vectors
        expected: the count is equal to the length of vectors
        '''
        nq = 100
        tmp, vectors = gen_binary_vectors(nq, dim)
        collection_list = []
        for i in range(20):
            collection_name = gen_unique_str('test_collection_rows_count_multi_collections')
            collection_list.append(collection_name)
            param = {'collection_name': collection_name,
                     'dimension': dim,
                     'index_file_size': index_file_size,
                     'metric_type': MetricType.HAMMING}
            connect.create_collection(param)
            res = connect.insert(collection_name=collection_name, records=vectors)
        connect.flush(collection_list)
        for i in range(20):
            status, res = connect.count_entities(collection_list[i])
            assert status.OK()
            assert res == nq


class TestCollectionCountTANIMOTO:

        status, res = connect.count_entities(jac_collection)
        assert res == 0

    def test_collection_rows_count_multi_collections(self, connect):
        '''
        target: test collection rows_count is correct or not with multiple collections of IP
        method: create collection and add vectors in it,
            assert the value returned by count_entities method is equal to length of vectors
        expected: the count is equal to the length of vectors
        '''
        nq = 100
        tmp, vectors = gen_binary_vectors(nq, dim)
        collection_list = []
        for i in range(20):
            collection_name = gen_unique_str('test_collection_rows_count_multi_collections')
            collection_list.append(collection_name)
            param = {'collection_name': collection_name,
                     'dimension': dim,
                     'index_file_size': index_file_size,
                     'metric_type': MetricType.JACCARD}
            connect.create_collection(param)
            res = connect.insert(collection_name=collection_name, records=vectors)
        connect.flush(collection_list)
        for i in range(20):
            status, res = connect.count_entities(collection_list[i])
            assert status.OK()
            assert res == nq

class TestCollectionCountBinary:
    """

        for p in processes:
            p.join()

    def test_collection_rows_count_multi_collections(self, connect):
        '''
        target: test collection rows_count is correct or not with multiple collections of IP
        method: create collection and add vectors in it,
            assert the value returned by count_entities method is equal to length of vectors
        expected: the count is equal to the length of vectors
        '''
        nq = 100
        vectors = gen_vectors(nq, dim)
        collection_list = []
        for i in range(20):
            collection_name = gen_unique_str('test_collection_rows_count_multi_collections')
            collection_list.append(collection_name)
            param = {'collection_name': collection_name,
                     'dimension': dim,
                     'index_file_size': index_file_size,
                     'metric_type': MetricType.IP}
            connect.create_collection(param)
            res = connect.insert(collection_name=collection_name, records=vectors)
        connect.flush(collection_list)
        for i in range(20):
            status, res = connect.count_entities(collection_list[i])
            assert status.OK()
            assert res == nq


class TestCollectionCountJAC:

        for p in processes:
            p.join()

    def test_collection_rows_count_multi_collections(self, connect):
        '''
        target: test collection rows_count is correct or not with multiple collections of L2
        method: create collection and add vectors in it,
            assert the value returned by count_entities method is equal to length of vectors
        expected: the count is equal to the length of vectors
        '''
        nq = 100
        vectors = gen_vectors(nq, dim)
        collection_list = []
        for i in range(20):
            collection_name = gen_unique_str()
            collection_list.append(collection_name)
            param = {'collection_name': collection_name,
                     'dimension': dim,
                     'index_file_size': index_file_size,
                     'metric_type': MetricType.L2}
            connect.create_collection(param)
            res = connect.insert(collection_name=collection_name, records=vectors)
        connect.flush(collection_list)
        for i in range(20):
            status, res = connect.count_entities(collection_list[i])
            assert status.OK()
            assert res == nq


class TestCollectionCountIP:

tests/milvus_python_test/entity/test_insert.py 2 locations


        assert status.OK()
        assert len(result) == 1

    def test_add_vector_multi_collections(self, connect):
        '''
        target: test add vectors is correct or not with multiple collections of IP
        method: create 50 collections and add vectors into them in turn
        expected: status ok
        '''
        nq = 100
        vectors = gen_vectors(nq, dim)
        collection_list = []
        for i in range(20):
            collection_name = gen_unique_str(
                'test_add_vector_multi_collections')
            collection_list.append(collection_name)
            param = {'collection_name': collection_name,
                     'dimension': dim,
                     'index_file_size': index_file_size,
                     'metric_type': MetricType.IP}
            connect.create_collection(param)
        for j in range(10):
            for i in range(20):
                status, ids = connect.insert(
                    collection_name=collection_list[i], records=vectors)
                assert status.OK()


        status, res = milvus.count_entities(collection)
        assert res == thread_num * nb

    def test_add_vector_multi_collections(self, connect):
        '''
        target: test add vectors is correct or not with multiple collections of L2
        method: create 50 collections and add vectors into them in turn
        expected: status ok
        '''
        nq = 100
        vectors = gen_vectors(nq, dim)
        collection_list = []
        for i in range(20):
            collection_name = gen_unique_str(
                'test_add_vector_multi_collections')
            collection_list.append(collection_name)
            param = {'collection_name': collection_name,
                     'dimension': dim,
                     'index_file_size': index_file_size,
                     'metric_type': MetricType.L2}
            connect.create_collection(param)
        for j in range(5):
            for i in range(20):
                status, ids = connect.insert(
                    collection_name=collection_list[i], records=vectors)
                assert status.OK()


		@@ 652-675 (lines=24) @@
649		status, res = connect.count_entities(ham_collection)
650		assert res == 0
651
652		def test_collection_rows_count_multi_collections(self, connect):
653		'''
654		target: test collection rows_count is correct or not with multiple collections of IP
655		method: create collection and add vectors in it,
656		assert the value returned by count_entities method is equal to length of vectors
657		expected: the count is equal to the length of vectors
658		'''
659		nq = 100
660		tmp, vectors = gen_binary_vectors(nq, dim)
661		collection_list = []
662		for i in range(20):
663		collection_name = gen_unique_str('test_collection_rows_count_multi_collections')
664		collection_list.append(collection_name)
665		param = {'collection_name': collection_name,
666		'dimension': dim,
667		'index_file_size': index_file_size,
668		'metric_type': MetricType.HAMMING}
669		connect.create_collection(param)
670		res = connect.insert(collection_name=collection_name, records=vectors)
671		connect.flush(collection_list)
672		for i in range(20):
673		status, res = connect.count_entities(collection_list[i])
674		assert status.OK()
675		assert res == nq
676
677
678		class TestCollectionCountTANIMOTO:
		@@ 459-482 (lines=24) @@
456		status, res = connect.count_entities(jac_collection)
457		assert res == 0
458
459		def test_collection_rows_count_multi_collections(self, connect):
460		'''
461		target: test collection rows_count is correct or not with multiple collections of IP
462		method: create collection and add vectors in it,
463		assert the value returned by count_entities method is equal to length of vectors
464		expected: the count is equal to the length of vectors
465		'''
466		nq = 100
467		tmp, vectors = gen_binary_vectors(nq, dim)
468		collection_list = []
469		for i in range(20):
470		collection_name = gen_unique_str('test_collection_rows_count_multi_collections')
471		collection_list.append(collection_name)
472		param = {'collection_name': collection_name,
473		'dimension': dim,
474		'index_file_size': index_file_size,
475		'metric_type': MetricType.JACCARD}
476		connect.create_collection(param)
477		res = connect.insert(collection_name=collection_name, records=vectors)
478		connect.flush(collection_list)
479		for i in range(20):
480		status, res = connect.count_entities(collection_list[i])
481		assert status.OK()
482		assert res == nq
483
484		class TestCollectionCountBinary:
485		"""
		@@ 347-370 (lines=24) @@
344		for p in processes:
345		p.join()
346
347		def test_collection_rows_count_multi_collections(self, connect):
348		'''
349		target: test collection rows_count is correct or not with multiple collections of IP
350		method: create collection and add vectors in it,
351		assert the value returned by count_entities method is equal to length of vectors
352		expected: the count is equal to the length of vectors
353		'''
354		nq = 100
355		vectors = gen_vectors(nq, dim)
356		collection_list = []
357		for i in range(20):
358		collection_name = gen_unique_str('test_collection_rows_count_multi_collections')
359		collection_list.append(collection_name)
360		param = {'collection_name': collection_name,
361		'dimension': dim,
362		'index_file_size': index_file_size,
363		'metric_type': MetricType.IP}
364		connect.create_collection(param)
365		res = connect.insert(collection_name=collection_name, records=vectors)
366		connect.flush(collection_list)
367		for i in range(20):
368		status, res = connect.count_entities(collection_list[i])
369		assert status.OK()
370		assert res == nq
371
372
373		class TestCollectionCountJAC:
		@@ 204-227 (lines=24) @@
201		for p in processes:
202		p.join()
203
204		def test_collection_rows_count_multi_collections(self, connect):
205		'''
206		target: test collection rows_count is correct or not with multiple collections of L2
207		method: create collection and add vectors in it,
208		assert the value returned by count_entities method is equal to length of vectors
209		expected: the count is equal to the length of vectors
210		'''
211		nq = 100
212		vectors = gen_vectors(nq, dim)
213		collection_list = []
214		for i in range(20):
215		collection_name = gen_unique_str()
216		collection_list.append(collection_name)
217		param = {'collection_name': collection_name,
218		'dimension': dim,
219		'index_file_size': index_file_size,
220		'metric_type': MetricType.L2}
221		connect.create_collection(param)
222		res = connect.insert(collection_name=collection_name, records=vectors)
223		connect.flush(collection_list)
224		for i in range(20):
225		status, res = connect.count_entities(collection_list[i])
226		assert status.OK()
227		assert res == nq
228
229
230		class TestCollectionCountIP:

		@@ 1289-1309 (lines=21) @@
1286		assert status.OK()
1287		assert len(result) == 1
1288
1289		def test_add_vector_multi_collections(self, connect):
1290		'''
1291		target: test add vectors is correct or not with multiple collections of IP
1292		method: create 50 collections and add vectors into them in turn
1293		expected: status ok
1294		'''
1295		nq = 100
1296		vectors = gen_vectors(nq, dim)
1297		collection_list = []
1298		for i in range(20):
1299		collection_name = gen_unique_str(
1300		'test_add_vector_multi_collections')
1301		collection_list.append(collection_name)
1302		param = {'collection_name': collection_name,
1303		'dimension': dim,
1304		'index_file_size': index_file_size,
1305		'metric_type': MetricType.IP}
1306		connect.create_collection(param)
1307		for j in range(10):
1308		for i in range(20):
1309		status, ids = connect.insert(
1310		collection_name=collection_list[i], records=vectors)
1311		assert status.OK()
1312
		@@ 646-666 (lines=21) @@
643		status, res = milvus.count_entities(collection)
644		assert res == thread_num * nb
645
646		def test_add_vector_multi_collections(self, connect):
647		'''
648		target: test add vectors is correct or not with multiple collections of L2
649		method: create 50 collections and add vectors into them in turn
650		expected: status ok
651		'''
652		nq = 100
653		vectors = gen_vectors(nq, dim)
654		collection_list = []
655		for i in range(20):
656		collection_name = gen_unique_str(
657		'test_add_vector_multi_collections')
658		collection_list.append(collection_name)
659		param = {'collection_name': collection_name,
660		'dimension': dim,
661		'index_file_size': index_file_size,
662		'metric_type': MetricType.L2}
663		connect.create_collection(param)
664		for j in range(5):
665		for i in range(20):
666		status, ids = connect.insert(
667		collection_name=collection_list[i], records=vectors)
668		assert status.OK()
669

milvus-io / milvus

Code Duplication Length = 21-24 lines in 6 locations

tests/milvus_python_test/collection/test_collection_count.py 4 locations

tests/milvus_python_test/entity/test_insert.py 2 locations