chore: Making embedding optional in retrieve_online_documents (feast-dev#5104)

Author: franciscojavierarceo

sdk/python/feast/feature_store.py

@@ -1863,11 +1863,11 @@ def retrieve_online_documents(
 
     def retrieve_online_documents_v2(
         self,
-        query: Union[str, List[float]],
-        top_k: int,
         features: List[str],
-        distance_metric: Optional[str] = "L2",
+        top_k: int,
+        query: Optional[List[float]] = None,
         query_string: Optional[str] = None,
+        distance_metric: Optional[str] = "L2",
     ) -> OnlineResponse:
         """
         Retrieves the top k closest document features. Note, embeddings are a subset of features.
@@ -1876,15 +1876,14 @@ def retrieve_online_documents_v2(
             features: The list of features that should be retrieved from the online document store. These features can be
                 specified either as a list of string document feature references or as a feature service. String feature
                 references must have format "feature_view:feature", e.g, "document_fv:document_embeddings".
-            query: The query to retrieve the closest document features for.
+            query: The embeded query to retrieve the closest document features for (optional)
             top_k: The number of closest document features to retrieve.
             distance_metric: The distance metric to use for retrieval.
             query_string: The query string to retrieve the closest document features using keyword search (bm25).
         """
-        if isinstance(query, str):
-            raise ValueError(
-                "Using embedding functionality is not supported for document retrieval. Please embed the query before calling retrieve_online_documents."
-            )
+        assert query is not None or query_string is not None, (
+            "Either query or query_string must be provided."
+        )
 
         (
             available_feature_views,
@@ -1988,7 +1987,7 @@ def _retrieve_from_online_store_v2(
         provider: Provider,
         table: FeatureView,
         requested_features: List[str],
-        query: List[float],
+        query: Optional[List[float]],
         top_k: int,
         distance_metric: Optional[str],
         query_string: Optional[str],

sdk/python/feast/infra/online_stores/milvus_online_store/milvus.py

@@ -460,7 +460,7 @@ def retrieve_online_documents_v2(
         config: RepoConfig,
         table: FeatureView,
         requested_features: List[str],
-        embedding: List[float],
+        embedding: Optional[List[float]],
         top_k: int,
         distance_metric: Optional[str] = None,
         query_string: Optional[str] = None,
@@ -471,6 +471,7 @@ def retrieve_online_documents_v2(
             Optional[Dict[str, ValueProto]],
         ]
     ]:
+        assert embedding is not None, "Key Word Search not yet implemented for Milvus"
         entity_name_feast_primitive_type_map = {
             k.name: k.dtype for k in table.entity_columns
         }

sdk/python/feast/infra/online_stores/online_store.py

@@ -436,7 +436,7 @@ def retrieve_online_documents_v2(
         config: RepoConfig,
         table: FeatureView,
         requested_features: List[str],
-        embedding: List[float],
+        embedding: Optional[List[float]],
         top_k: int,
         distance_metric: Optional[str] = None,
         query_string: Optional[str] = None,
@@ -455,7 +455,7 @@ def retrieve_online_documents_v2(
             config: The config for the current feature store.
             table: The feature view whose feature values should be read.
             requested_features: The list of features whose embeddings should be used for retrieval.
-            embedding: The embeddings to use for retrieval.
+            embedding: The embeddings to use for retrieval (optional)
             top_k: The number of documents to retrieve.
             query_string: The query string to search for using keyword search (bm25) (optional)
 
@@ -464,6 +464,9 @@ def retrieve_online_documents_v2(
             where the first item is the event timestamp for the row, and the second item is a dict of feature
             name to embeddings.
         """
+        assert embedding is not None or query_string is not None, (
+            "Either embedding or query_string must be specified"
+        )
         raise NotImplementedError(
             f"Online store {self.__class__.__name__} does not support online retrieval"
         )

sdk/python/feast/infra/online_stores/sqlite.py

@@ -439,7 +439,7 @@ def retrieve_online_documents_v2(
         config: RepoConfig,
         table: FeatureView,
         requested_features: List[str],
-        query: List[float],
+        query: Optional[List[float]],
         top_k: int,
         distance_metric: Optional[str] = None,
         query_string: Optional[str] = None,
@@ -456,7 +456,7 @@ def retrieve_online_documents_v2(
             config: Feast configuration object
             table: FeatureView object as the table to search
             requested_features: List of requested features to retrieve
-            query: Query embedding to search for
+            query: Query embedding to search for (optional)
             top_k: Number of items to return
             distance_metric: Distance metric to use (optional)
             query_string: The query string to search for using keyword search (bm25) (optional)

sdk/python/feast/infra/passthrough_provider.py

@@ -318,7 +318,7 @@ def retrieve_online_documents_v2(
         config: RepoConfig,
         table: FeatureView,
         requested_features: Optional[List[str]],
-        query: List[float],
+        query: Optional[List[float]],
         top_k: int,
         distance_metric: Optional[str] = None,
         query_string: Optional[str] = None,

sdk/python/feast/infra/provider.py

@@ -456,7 +456,7 @@ def retrieve_online_documents_v2(
         config: RepoConfig,
         table: FeatureView,
         requested_features: List[str],
-        query: List[float],
+        query: Optional[List[float]],
         top_k: int,
         distance_metric: Optional[str] = None,
         query_string: Optional[str] = None,
@@ -475,7 +475,7 @@ def retrieve_online_documents_v2(
             config: The config for the current feature store.
             table: The feature view whose embeddings should be searched.
             requested_features: the requested document feature names.
-            query: The query embedding to search for.
+            query: The query embedding to search for (optional).
             top_k: The number of documents to return.
             query_string: The query string to search for using keyword search (bm25) (optional)
 

sdk/python/tests/foo_provider.py

@@ -169,7 +169,7 @@ def retrieve_online_documents_v2(
         config: RepoConfig,
         table: FeatureView,
         requested_features: List[str],
-        query: List[float],
+        query: Optional[List[float]],
         top_k: int,
         distance_metric: Optional[str] = None,
         query_string: Optional[str] = None,

sdk/python/tests/unit/online_store/test_online_retrieval.py

@@ -979,14 +979,13 @@ def test_sqlite_get_online_documents_v2_search() -> None:
         )
 
         # Test vector similarity search
-        query_embedding = [float(x) for x in np.random.random(vector_length)]
+        # query_embedding = [float(x) for x in np.random.random(vector_length)]
         result = store.retrieve_online_documents_v2(
             features=[
                 "document_embeddings:Embeddings",
                 "document_embeddings:content",
                 "document_embeddings:title",
             ],
-            query=query_embedding,
             query_string="(content: 5) OR (title: 1) OR (title: 3)",
             top_k=3,
         ).to_dict()
@@ -1279,12 +1278,12 @@ def test_milvus_native_from_feast_data() -> None:
     search_res = client.search(
         collection_name=COLLECTION_NAME,
         data=[query_embedding],
-        limit=3,  # Top 3 results
+        limit=5,  # Top 3 results
         output_fields=["item_id", "author_id", "sentence_chunks"],
     )
 
     # Validate the search results
-    assert len(search_res[0]) == 3
+    assert len(search_res[0]) == 5
     print("Search Results:", search_res[0])
 
     # Clean up the collection