Add func check of output fields and modify cases (#25510)

Signed-off-by: nico <cheng.yuan@zilliz.com>

tests/python_client/base/client_base.py

@@ -196,7 +196,7 @@ def insert_data_general(self, prefix="test", insert_data=False, nb=ct.default_nb
 
     def init_collection_general(self, prefix="test", insert_data=False, nb=ct.default_nb,
                                 partition_num=0, is_binary=False, is_all_data_type=False,
-                                auto_id=False, dim=ct.default_dim, is_index=False,
+                                auto_id=False, dim=ct.default_dim, is_index=True,
                                 primary_field=ct.default_int64_field_name, is_flush=True, name=None,
                                 enable_dynamic_field=False, with_json=True, **kwargs):
         """
@@ -243,13 +243,13 @@ def init_collection_general(self, prefix="test", insert_data=False, nb=ct.defaul
                 assert collection_w.is_empty is False
                 assert collection_w.num_entities == nb
             # This condition will be removed after auto index feature
-            if not is_index:
+            if is_index:
                 if is_binary:
                     collection_w.create_index(ct.default_binary_vec_field_name, ct.default_bin_flat_index)
                 else:
                     collection_w.create_index(ct.default_float_vec_field_name, ct.default_flat_index)
                 collection_w.load()
-        elif not is_index:
+        elif is_index:
             if is_binary:
                 collection_w.create_index(ct.default_binary_vec_field_name, ct.default_bin_flat_index)
             else:

tests/python_client/check/func_check.py

@@ -287,6 +287,12 @@ def check_search_results(search_res, func_name, check_items):
             if check_items["_async"]:
                 search_res.done()
                 search_res = search_res.result()
+        if check_items.get("output_fields", None):
+            assert set(search_res[0][0].entity.fields) == set(check_items["output_fields"])
+            log.info('search_results_check: Output fields of query searched is correct')
+            if check_items.get("original_entities", None):
+                original_entities = check_items["original_entities"][0]
+                pc.output_field_value_check(search_res, original_entities)
         if len(search_res) != check_items["nq"]:
             log.error("search_results_check: Numbers of query searched (%d) "
                       "is not equal with expected (%d)"

tests/python_client/check/param_check.py

@@ -224,3 +224,23 @@ def equal_entities_list(exp, actual, primary_field, with_vec=False):
                 except Exception as ex:
                     log.error(ex)
     return True if len(exp) == 0 else False
+
+
+def output_field_value_check(search_res, original):
+    """
+    check if the value of output fields is correct
+    :param search_res: the search result of specific output fields
+    :param original: the data in the collection
+    :return: True or False
+    """
+    limit = len(search_res[0])
+    for i in range(limit):
+        entity = eval(str(search_res[0][i]).split('entity: ', 1)[1])
+        _id = search_res[0][i].id
+        for field in entity.keys():
+            if isinstance(entity[field], list):
+                for order in range(0, len(entity[field]), 4):
+                    assert abs(original[field][_id][order] - entity[field][order]) < ct.epsilon
+            else:
+                assert original[field][_id] == entity[field]
+    return True

tests/python_client/common/common_func.py

@@ -1016,3 +1016,11 @@ def install_milvus_operator_specific_config(namespace, milvus_mode, release_name
         raise MilvusException(message=f'Milvus healthy timeout 1800s')
 
     return host
+
+
+def get_wildcard_output_field_names(collection_w, output_fields):
+    all_fields = [field.name for field in collection_w.schema.fields]
+    if "*" in output_fields:
+        output_fields.remove("*")
+        output_fields.extend(all_fields)
+    return output_fields

tests/python_client/testcases/test_collection.py

@@ -2543,7 +2543,7 @@ def test_load_collection_without_creating_index(self):
         method: create a collection without index, then load
         expected: raise exception
         """
-        collection_w = self.init_collection_general(prefix, True, is_index=True)[0]
+        collection_w = self.init_collection_general(prefix, True, is_index=False)[0]
         collection_w.load(check_task=CheckTasks.err_res,
                           check_items={"err_code": 1,
                                        "err_msg": "index not exist"})
@@ -2581,7 +2581,7 @@ def test_release_partition_during_searching(self):
         """
         self._connect()
         partition_num = 1
-        collection_w = self.init_collection_general(prefix, True, 10, partition_num, is_index=True)[0]
+        collection_w = self.init_collection_general(prefix, True, 10, partition_num, is_index=False)[0]
         collection_w.create_index(ct.default_float_vec_field_name, index_params=ct.default_flat_index)
         par = collection_w.partitions
         par_name = par[partition_num].name
@@ -2607,7 +2607,7 @@ def test_release_indexed_collection_during_searching(self):
         """
         self._connect()
         partition_num = 1
-        collection_w = self.init_collection_general(prefix, True, 10, partition_num, is_index=True)[0]
+        collection_w = self.init_collection_general(prefix, True, 10, partition_num, is_index=False)[0]
         collection_w.create_index(ct.default_float_vec_field_name, index_params=ct.default_flat_index)
         par = collection_w.partitions
         par_name = par[partition_num].name
@@ -2662,7 +2662,7 @@ def test_load_partition_after_index_binary(self, binary_index, metric_type):
         self._connect()
         partition_num = 1
         collection_w = self.init_collection_general(prefix, True, ct.default_nb, partition_num,
-                                                    is_binary=True, is_index=True)[0]
+                                                    is_binary=True, is_index=False)[0]
 
         # for metric_type in ct.binary_metrics:
         binary_index["metric_type"] = metric_type

tests/python_client/testcases/test_delete.py

@@ -321,7 +321,7 @@ def test_delete_after_index(self):
         expected: assert index and deleted id not in search result
         """
         # create collection, insert tmp_nb, flush and load
-        collection_w, vectors = self.init_collection_general(prefix, insert_data=True, is_index=True)[0:2]
+        collection_w, vectors = self.init_collection_general(prefix, insert_data=True, is_index=False)[0:2]
 
         # create index
         index_params = {"index_type": "IVF_SQ8",
@@ -1174,7 +1174,7 @@ def test_delete_all_index_with_string(self):
         expected: assert index and deleted id not in search result
         """
         # create collection, insert tmp_nb, flush and load
-        collection_w, vectors = self.init_collection_general(prefix, insert_data=True, is_index=True,
+        collection_w, vectors = self.init_collection_general(prefix, insert_data=True, is_index=False,
                                                              primary_field=ct.default_string_field_name)[0:2]
 
         # create index

tests/python_client/testcases/test_index.py

@@ -225,7 +225,7 @@ def test_index_create_indexes_for_different_fields(self):
         method: create two different indexes with default index name
         expected: create successfully
         """
-        collection_w = self.init_collection_general(prefix, True, is_index=True)[0]
+        collection_w = self.init_collection_general(prefix, True, is_index=False)[0]
         default_index = {"index_type": "IVF_FLAT", "params": {"nlist": 128}, "metric_type": "L2"}
         collection_w.create_index(default_field_name, default_index)
         collection_w.create_index(ct.default_int64_field_name, {})
@@ -237,7 +237,7 @@ def test_index_create_on_scalar_field(self):
         method: create index on scalar field and load
         expected: raise exception
         """
-        collection_w = self.init_collection_general(prefix, True, is_index=True)[0]
+        collection_w = self.init_collection_general(prefix, True, is_index=False)[0]
         collection_w.create_index(ct.default_int64_field_name, {})
         collection_w.load(check_task=CheckTasks.err_res,
                           check_items={ct.err_code: 1, ct.err_msg: "there is no vector index on collection, "
@@ -1259,10 +1259,7 @@ def test_drop_index_without_release(self):
                 2. drop the index
         expected: raise exception
         """
-        collection_w = self.init_collection_general(prefix, True, is_index=True)[0]
-        default_index = {"index_type": "IVF_FLAT", "params": {"nlist": 128}, "metric_type": "L2"}
-        collection_w.create_index("float_vector", default_index)
-        collection_w.load()
+        collection_w = self.init_collection_general(prefix, True)[0]
         collection_w.drop_index(check_task=CheckTasks.err_res,
                                 check_items={"err_code": 1,
                                              "err_msg": "index cannot be dropped, collection is "
@@ -1853,7 +1850,7 @@ def test_create_autoindex_with_no_params(self):
         method: create index with only one field name
         expected: create successfully
         """
-        collection_w = self.init_collection_general(prefix, is_index=True)[0]
+        collection_w = self.init_collection_general(prefix, is_index=False)[0]
         collection_w.create_index(field_name)
         actual_index_params = collection_w.index()[0].params
         assert default_autoindex_params == actual_index_params
@@ -1866,7 +1863,7 @@ def test_create_autoindex_with_params(self, index_params):
         method: create index with different params
         expected: create successfully
         """
-        collection_w = self.init_collection_general(prefix, is_index=True)[0]
+        collection_w = self.init_collection_general(prefix, is_index=False)[0]
         collection_w.create_index(field_name, index_params)
         actual_index_params = collection_w.index()[0].params
         log.info(collection_w.index()[0].params)
@@ -1885,7 +1882,7 @@ def test_create_autoindex_with_invalid_params(self):
         method: create index with invalid params
         expected: raise exception
         """
-        collection_w = self.init_collection_general(prefix, is_index=True)[0]
+        collection_w = self.init_collection_general(prefix, is_index=False)[0]
         index_params = {"metric_type": "L2", "nlist": "1024", "M": "100"}
         collection_w.create_index(field_name, index_params,
                                   check_task=CheckTasks.err_res,
@@ -1900,7 +1897,7 @@ def test_create_autoindex_on_binary_vectors(self):
         method: create index on binary vectors
         expected: raise exception
         """
-        collection_w = self.init_collection_general(prefix, is_binary=True, is_index=True)[0]
+        collection_w = self.init_collection_general(prefix, is_binary=True, is_index=False)[0]
         collection_w.create_index(binary_field_name, {},
                                   check_task=CheckTasks.err_res,
                                   check_items={"err_code": 1,

tests/python_client/testcases/test_query.py

@@ -780,13 +780,11 @@ def test_query_output_fields_part_scale_wildcard(self):
         expected: verify query result
         """
         # init collection with fields: int64, float, float_vec
-        collection_w, vectors = self.init_collection_general(prefix, insert_data=True, is_index=True)[0:2]
+        collection_w, vectors = self.init_collection_general(prefix, insert_data=True)[0:2]
         df = vectors[0]
 
         # query with output_fields=["*", float_vector)
         res = df.iloc[:2, :4].to_dict('records')
-        collection_w.create_index(ct.default_float_vec_field_name, index_params=ct.default_flat_index)
-        collection_w.load()
         collection_w.query(default_term_expr, output_fields=["*", ct.default_float_vec_field_name],
                            check_task=CheckTasks.check_query_results,
                            check_items={exp_res: res, "with_vec": True})
@@ -1338,13 +1336,7 @@ def test_query_after_index(self):
                 3. query
         expected: query result is correct
         """
-        collection_w, vectors, binary_raw_vectors = self.init_collection_general(prefix, insert_data=True, is_index=True)[0:3]
-
-        default_field_name = ct.default_float_vec_field_name
-        collection_w.create_index(default_field_name, default_index_params)
-
-        collection_w.load()
-
+        collection_w, vectors, binary_raw_vectors = self.init_collection_general(prefix, insert_data=True)[0:3]
         int_values = [0]
         term_expr = f'{ct.default_int64_field_name} in {int_values}'
         check_vec = vectors[0].iloc[:, [0]][0:len(int_values)].to_dict('records')
@@ -1405,7 +1397,7 @@ def test_query_output_binary_vec_field_after_index(self):
         method: create index and specify vec field as output field
         expected: return primary field and vec field
         """
-        collection_w, vectors = self.init_collection_general(prefix, insert_data=True, is_binary=True, is_index=True)[0:2]
+        collection_w, vectors = self.init_collection_general(prefix, insert_data=True, is_binary=True, is_index=False)[0:2]
         fields = [ct.default_int64_field_name, ct.default_binary_vec_field_name]
         collection_w.create_index(ct.default_binary_vec_field_name, binary_index_params)
         assert collection_w.has_index()[0]
@@ -1589,7 +1581,7 @@ def test_query_string_expr_with_binary(self):
         method: query string expr with binary
         expected: verify query successfully
         """
-        collection_w, vectors= self.init_collection_general(prefix, insert_data=True, is_binary=True, is_index=True)[0:2]
+        collection_w, vectors = self.init_collection_general(prefix, insert_data=True, is_binary=True, is_index=False)[0:2]
         collection_w.create_index(ct.default_binary_vec_field_name, binary_index_params)
         collection_w.load()
         assert collection_w.has_index()[0]
@@ -1737,7 +1729,7 @@ def test_query_string_field_not_primary_is_empty(self):
         expected: query successfully
         """
         # 1.  create a collection
-        collection_w, vectors = self.init_collection_general(prefix, insert_data=False, is_index=True)[0:2]
+        collection_w, vectors = self.init_collection_general(prefix, insert_data=False, is_index=False)[0:2]
 
         nb = 3000
         df = cf.gen_default_list_data(nb)
@@ -1764,7 +1756,7 @@ def test_query_with_create_diskann_index(self):
         method: create a collection and build diskann index 
         expected: verify query result
         """
-        collection_w, vectors = self.init_collection_general(prefix, insert_data=True, is_index=True)[0:2]
+        collection_w, vectors = self.init_collection_general(prefix, insert_data=True, is_index=False)[0:2]
 
         collection_w.create_index(ct.default_float_vec_field_name, ct.default_diskann_index)
         assert collection_w.has_index()[0]
@@ -1783,7 +1775,7 @@ def test_query_with_create_diskann_with_string_pk(self):
         method: create a collection with string pk and build diskann index 
         expected: verify query result
         """
-        collection_w, vectors = self.init_collection_general(prefix, insert_data=True,  primary_field=ct.default_string_field_name, is_index=True)[0:2]
+        collection_w, vectors = self.init_collection_general(prefix, insert_data=True,  primary_field=ct.default_string_field_name, is_index=False)[0:2]
         collection_w.create_index(ct.default_float_vec_field_name, ct.default_diskann_index)
         assert collection_w.has_index()[0]
         collection_w.load()
@@ -1802,7 +1794,7 @@ def test_query_with_scalar_field(self):
         expected: query successfully
         """
         # 1.  create a collection
-        collection_w, vectors = self.init_collection_general(prefix, insert_data=False, is_index=True)[0:2]
+        collection_w, vectors = self.init_collection_general(prefix, insert_data=False, is_index=False)[0:2]
 
         nb = 3000
         df = cf.gen_default_list_data(nb)