BrainAPI is a knowledge graph–powered AI memory layer that transforms unstructured data into structured knowledge, enabling intelligent search, recommendations, and contextual memory for AI agents and applications.
Python ai-agentsai-memoryknowledge-graphknowledge-managementknowledge-representationreccomendation-system
Refactor vector search calls with facade and safe keywords
Cursor Agent480be1ac7207
diff --git a/src/core/agents/tools/kg_agent/KGAgentAddTripletsTool.py b/src/core/agents/tools/kg_agent/KGAgentAddTripletsTool.pyindex 0bda98a..d7befe5 100644--- a/src/core/agents/tools/kg_agent/KGAgentAddTripletsTool.py+++ b/src/core/agents/tools/kg_agent/KGAgentAddTripletsTool.py@@ -93,10 +93,16 @@ class KGAgentAddTripletsTool(BaseTool):v_obj = self.embeddings.embed_text(object_node.name)v_sim_sub = self.vector_store.search_vectors(- v_sub.embeddings, "nodes", k=5, brain_id=self.brain_id+ v_sub.embeddings,+ store="nodes",+ brain_id=self.brain_id,+ k=5,)v_sim_obj = self.vector_store.search_vectors(- v_obj.embeddings, "nodes", k=5, brain_id=self.brain_id+ v_obj.embeddings,+ store="nodes",+ brain_id=self.brain_id,+ k=5,)sim_sub = most_similar_name_with_labels_or_none(diff --git a/src/core/agents/tools/kg_agent/KGAgentSearchGraphTool.py b/src/core/agents/tools/kg_agent/KGAgentSearchGraphTool.pyindex 036f1fe..5c53348 100644--- a/src/core/agents/tools/kg_agent/KGAgentSearchGraphTool.py+++ b/src/core/agents/tools/kg_agent/KGAgentSearchGraphTool.py@@ -111,10 +111,15 @@ class KGAgentSearchGraphTool(BaseTool):if query:query_embedding = self.embeddings.embed_text(query)v_triplets_results = self.vector_store.search_vectors(- query_embedding.embeddings, "triplets", k=5, brain_id=self.brain_id+ query_embedding.embeddings,+ store="triplets",+ k=5,+ brain_id=self.brain_id,)v_nodes_results = self.vector_store.search_vectors(- query_embedding.embeddings, "nodes", brain_id=self.brain_id+ query_embedding.embeddings,+ store="nodes",+ brain_id=self.brain_id,)nodes.extend(self.kg.node_text_search(query, brain_id=self.brain_id))nodes.extend(diff --git a/src/core/search/entity_context.py b/src/core/search/entity_context.pyindex ec68e1f..52eed38 100644--- a/src/core/search/entity_context.py+++ b/src/core/search/entity_context.py@@ -15,6 +15,9 @@ from src.services.kg_agent.main import (vector_store_adapter,graph_adapter,)+from src.utils.vector_search import VectorSearchFacade++vector_search = VectorSearchFacade(vector_store_adapter)class EntityContext:@@ -50,8 +53,9 @@ class EntityContext:natural_language_web = list()target_embedding = embeddings_adapter.embed_text(self.target)- target_node_vs = vector_store_adapter.search_vectors(- target_embedding.embeddings, store="nodes", brain_id=self.brain_id+ target_node_vs = vector_search.search_nodes(+ target_embedding.embeddings,+ brain_id=self.brain_id,)if not target_node_vs:diff --git a/src/core/search/entity_info.py b/src/core/search/entity_info.pyindex 94a60b4..041833e 100644--- a/src/core/search/entity_info.py+++ b/src/core/search/entity_info.py@@ -19,9 +19,13 @@ from src.constants.kg import Node, Predicatefrom src.services.kg_agent.main import graph_adapterfrom src.services.kg_agent.main import embeddings_adapterfrom src.services.kg_agent.main import vector_store_adapter+from src.utils.vector_search import VectorSearchFacadefrom src.utils.similarity.vectors import cosine_similarity+vector_search = VectorSearchFacade(vector_store_adapter)++# ================================================================# NOTE Currently not fully supported by the Event-Centric v2 kg# Need to figure out how to handle this.@@ -164,8 +168,8 @@ class EventSynergyRetriever:query_embedding = embeddings_adapter.embed_text(query)target_embedding = embeddings_adapter.embed_text(target)- target_node_vs = vector_store_adapter.search_vectors(- target_embedding.embeddings, store="nodes", brain_id=self.brain_id+ target_node_vs = vector_search.search_nodes(+ target_embedding.embeddings, brain_id=self.brain_id)if not target_node_vs:diff --git a/src/core/search/entity_sibilings.py b/src/core/search/entity_sibilings.pyindex f8be810..f0240f8 100644--- a/src/core/search/entity_sibilings.py+++ b/src/core/search/entity_sibilings.py@@ -18,6 +18,7 @@ from src.services.kg_agent.main import (graph_adapter,vector_store_adapter,)+from src.utils.vector_search import VectorSearchFacadefrom src.utils.similarity.numbers import wmean, wsimfrom src.utils.similarity.vectors import cosine_similarity@@ -33,6 +34,8 @@ FACTORS_INCREMENTAL_WEIGHT = 0.3# Weight to strengthen the similarity by the comparison of the neighbor+target descriptionsNODE_SIM_DESC_INCREMENTAL_WEIGHT = 0.5+vector_search = VectorSearchFacade(vector_store_adapter)+class EntitySinergyRetriever:def __init__(self, brain_id: str = "default"):@@ -358,9 +361,8 @@ class EntitySinergyRetriever:_upsert_connection(seed_node, neighbor[1], neighbor[0])target_embedding = embeddings_adapter.embed_text(target)- target_node_vs = vector_store_adapter.search_vectors(+ target_node_vs = vector_search.search_nodes(target_embedding.embeddings,- store="nodes",brain_id=self.brain_id,)if not target_node_vs:@@ -406,9 +408,8 @@ class EntitySinergyRetriever:)if seed_embedding is None:continue- similar_seed_vs = vector_store_adapter.search_vectors(+ similar_seed_vs = vector_search.search_nodes(seed_embedding,- store="nodes",brain_id=self.brain_id,)diff --git a/src/services/api/controllers/entities.py b/src/services/api/controllers/entities.pyindex 65774f4..d6ebaf6 100644--- a/src/services/api/controllers/entities.py+++ b/src/services/api/controllers/entities.py@@ -24,6 +24,9 @@ from src.core.search.entity_sibilings import EntitySinergyRetrieverfrom src.services.data.main import data_adapterfrom src.services.input.agents import embeddings_adapterfrom src.services.kg_agent.main import graph_adapter, vector_store_adapter+from src.utils.vector_search import VectorSearchFacade++vector_search = VectorSearchFacade(vector_store_adapter)async def get_entity_info(@@ -149,8 +152,9 @@ async def get_entity_status(types = []target_embeddings = embeddings_adapter.embed_text(target)- target_node_vs = vector_store_adapter.search_vectors(- target_embeddings.embeddings, store="nodes", brain_id=brain_id+ target_node_vs = vector_search.search_nodes(+ target_embeddings.embeddings,+ brain_id=brain_id,)target_node = Nonediff --git a/src/services/api/controllers/kg.py b/src/services/api/controllers/kg.pyindex 2dfda6f..adcb840 100644--- a/src/services/api/controllers/kg.py+++ b/src/services/api/controllers/kg.py@@ -14,6 +14,9 @@ from src.services.kg_agent.main import (graph_adapter,vector_store_adapter,)+from src.utils.vector_search import VectorSearchFacade++vector_search = VectorSearchFacade(vector_store_adapter)async def get_hops(@@ -25,8 +28,9 @@ async def get_hops(Currently only supports 2nd degree hops."""text_embeddings = embeddings_adapter.embed_text(query)- data_vectors = vector_store_adapter.search_vectors(- text_embeddings.embeddings, store="nodes", brain_id=brain_id+ data_vectors = vector_search.search_nodes(+ text_embeddings.embeddings,+ brain_id=brain_id,)node_uuids = [v.metadata.get("uuid") for v in data_vectors]print("[+] Getting 2nd degree hops for nodes:", node_uuids)diff --git a/src/services/api/controllers/retrieve.py b/src/services/api/controllers/retrieve.pyindex 8784763..8fd2e0d 100644--- a/src/services/api/controllers/retrieve.py+++ b/src/services/api/controllers/retrieve.py@@ -19,6 +19,7 @@ from src.constants.embeddings import Vectorfrom src.constants.kg import IdentificationParams, Node, Predicatefrom src.core.search.entities import search_entitiesfrom src.core.search.relationships import search_relationships+from src.utils.vector_search import VectorSearchFacadefrom src.services.api.constants.requests import (GetContextRequestBody,GetContextResponse,@@ -33,6 +34,8 @@ from src.services.kg_agent.main import embeddings_adapter, vector_store_adapterfrom src.utils.similarity.vectors import cosine_similarityfrom src.utils.nlp.ner import _entity_extractor+vector_search = VectorSearchFacade(vector_store_adapter)+async def retrieve_data(text: str, limit: int, preferred_entities: str, brain_id: str = "default"@@ -50,11 +53,15 @@ async def retrieve_data(def _get_data():text_embeddings = embeddings_adapter.embed_text(text)- data_vectors = vector_store_adapter.search_vectors(- text_embeddings.embeddings, "data", brain_id, limit+ data_vectors = vector_search.search_data(+ text_embeddings.embeddings,+ brain_id=brain_id,+ k=limit,)- triple_vectors = vector_store_adapter.search_vectors(- text_embeddings.embeddings, "triplets", brain_id, limit+ triple_vectors = vector_search.search_triplets(+ text_embeddings.embeddings,+ brain_id=brain_id,+ k=limit,)search_result = data_adapter.search(text, brain_id)@@ -400,8 +407,9 @@ async def get_context(request: GetContextRequestBody) -> GetContextResponse:if text in embeddings_map:return embeddings_map[text]text_embeddings = embeddings_adapter.embed_text(text)- data_vectors = vector_store_adapter.search_vectors(- text_embeddings.embeddings, store="nodes", brain_id=request.brain_id+ data_vectors = vector_search.search_nodes(+ text_embeddings.embeddings,+ brain_id=request.brain_id,)neighbors = graph_adapter.get_event_centric_neighbors([v.metadata.get("uuid") for v in data_vectors], brain_id=request.brain_iddiff --git a/src/services/mcp/main.py b/src/services/mcp/main.pyindex 6c28c88..724d844 100644--- a/src/services/mcp/main.py+++ b/src/services/mcp/main.py@@ -23,8 +23,10 @@ from src.core.instances import ()from src.lib.neo4j.client import _neo4j_clientfrom src.services.mcp.utils import guard_brainpat+from src.utils.vector_search import VectorSearchFacadeauth_token_var: ContextVar[str | None] = ContextVar("auth_token", default=None)+vector_search = VectorSearchFacade(vector_store_adapter)mcp = FastMCP("brainapi-mcp", stateless_http=True, host="0.0.0.0")@@ -71,8 +73,10 @@ def _search_semantically_sync(query: str, brain_id: str) -> Any:if not guard_brainpat(auth_token_var.get(), brain_id):return "Unauthorized"query_embedding = embeddings_adapter.embed_text(query)- data_vectors = vector_store_adapter.search_vectors(- query_embedding.embeddings, store="nodes", brain_id=brain_id, k=5+ data_vectors = vector_search.search_nodes(+ query_embedding.embeddings,+ brain_id=brain_id,+ k=5,)triplets = graph_adapter.get_event_centric_neighbors([v.metadata.get("uuid") for v in data_vectors], brain_id=brain_iddiff --git a/src/utils/vector_search.py b/src/utils/vector_search.pynew file mode 100644index 0000000..4216938--- /dev/null+++ b/src/utils/vector_search.py@@ -0,0 +1,58 @@+from src.adapters.embeddings import VectorStoreAdapter+from src.constants.embeddings import Vector+++class VectorSearchFacade:+ def __init__(self, vector_store: VectorStoreAdapter):+ self._vector_store = vector_store++ def search(+ self,+ data_vector: list[float],+ *,+ store: str,+ brain_id: str = "default",+ k: int = 10,+ ) -> list[Vector]:+ return self._vector_store.search_vectors(+ data_vector,+ store=store,+ brain_id=brain_id,+ k=k,+ )++ def search_nodes(+ self,+ data_vector: list[float],+ *,+ brain_id: str = "default",+ k: int = 10,+ ) -> list[Vector]:+ return self.search(data_vector, store="nodes", brain_id=brain_id, k=k)++ def search_triplets(+ self,+ data_vector: list[float],+ *,+ brain_id: str = "default",+ k: int = 10,+ ) -> list[Vector]:+ return self.search(data_vector, store="triplets", brain_id=brain_id, k=k)++ def search_relationships(+ self,+ data_vector: list[float],+ *,+ brain_id: str = "default",+ k: int = 10,+ ) -> list[Vector]:+ return self.search(data_vector, store="relationships", brain_id=brain_id, k=k)++ def search_data(+ self,+ data_vector: list[float],+ *,+ brain_id: str = "default",+ k: int = 10,+ ) -> list[Vector]:+ return self.search(data_vector, store="data", brain_id=brain_id, k=k)diff --git a/tests/test_architecture_refactors.py b/tests/test_architecture_refactors.pyindex 102df56..6fcaccf 100644--- a/tests/test_architecture_refactors.py+++ b/tests/test_architecture_refactors.py@@ -38,8 +38,11 @@ from src.adapters.embeddings import ()from src.adapters.graph import GraphAdapterfrom src.constants.embeddings import Vector+from src.constants.data import Observation, TextChunkfrom src.core.agents.core.runtime_agent_factory import RuntimeAgentFactoryfrom src.services.api.controllers.entities import get_entity_status+from src.services.api.controllers.retrieve import retrieve_data+from src.utils.vector_search import VectorSearchFacadedef raise_embedding_error(*args, **kwargs):@@ -191,7 +194,7 @@ class EntityStatusControllerTests(unittest.IsolatedAsyncioTestCase):return_value=Vector(id="q", embeddings=[0.1, 0.2], metadata={}),),patch(- "src.services.api.controllers.entities.vector_store_adapter.search_vectors",+ "src.services.api.controllers.entities.vector_search.search_nodes",return_value=[fake_vector],),patch(@@ -206,5 +209,91 @@ class EntityStatusControllerTests(unittest.IsolatedAsyncioTestCase):self.assertEqual(response.observations, [])+class VectorSearchFacadeTests(unittest.TestCase):+ def test_facade_routes_search_arguments_with_keywords(self):+ class FakeVectorStore:+ def __init__(self):+ self.calls = []++ def search_vectors(self, data_vector, brain_id, store, k=10):+ self.calls.append(+ {+ "data_vector": data_vector,+ "brain_id": brain_id,+ "store": store,+ "k": k,+ }+ )+ return [Vector(id="v1", embeddings=[0.1], metadata={"uuid": "n1"})]++ fake = FakeVectorStore()+ facade = VectorSearchFacade(fake)+ result = facade.search_nodes([0.1, 0.2], brain_id="brain-a", k=7)+ self.assertEqual(result[0].id, "v1")+ self.assertEqual(+ fake.calls[0],+ {+ "data_vector": [0.1, 0.2],+ "brain_id": "brain-a",+ "store": "nodes",+ "k": 7,+ },+ )+++class RetrieveControllerVectorSearchTests(unittest.IsolatedAsyncioTestCase):+ async def test_retrieve_data_uses_store_specific_vector_search(self):+ fake_data_chunk = TextChunk(text="chunk", brain_version="1.0.0")+ fake_observation = Observation(text="obs", resource_id="r1")+ fake_search_result = type(+ "SearchResult",+ (),+ {+ "text_chunks": [fake_data_chunk],+ "observations": [fake_observation],+ },+ )()+ fake_data_vector = type("FakeVector", (), {"metadata": {"resource_id": "r1"}})()+ fake_triplet_vector = type(+ "FakeVector", (), {"metadata": {"node_ids": ["n1"], "predicate": "RELATED_TO"}}+ )()+ with (+ patch(+ "src.services.api.controllers.retrieve.embeddings_adapter.embed_text",+ return_value=Vector(id="q", embeddings=[0.1, 0.2], metadata={}),+ ),+ patch(+ "src.services.api.controllers.retrieve.vector_search.search_data",+ return_value=[fake_data_vector],+ ) as search_data,+ patch(+ "src.services.api.controllers.retrieve.vector_search.search_triplets",+ return_value=[fake_triplet_vector],+ ) as search_triplets,+ patch(+ "src.services.api.controllers.retrieve.data_adapter.search",+ return_value=fake_search_result,+ ),+ patch(+ "src.services.api.controllers.retrieve.data_adapter.get_text_chunks_by_ids",+ return_value=([], []),+ ),+ patch(+ "src.services.api.controllers.retrieve.graph_adapter.get_nodes_by_uuid",+ return_value=[],+ ),+ ):+ response = await retrieve_data(+ text="sample",+ limit=6,+ preferred_entities="Person,Company",+ brain_id="brain-a",+ )+ search_data.assert_called_once_with([0.1, 0.2], brain_id="brain-a", k=6)+ search_triplets.assert_called_once_with([0.1, 0.2], brain_id="brain-a", k=6)+ self.assertEqual(len(response.data), 1)+ self.assertEqual(len(response.observations), 1)++if __name__ == "__main__":unittest.main()