[WIP] Add hybrid diversity support

integration/test_collection_diversity_hybrid.py

@@ -0,0 +1,149 @@
+"""Integration tests for hybrid search + MMR diversity selection.
+
+``DiversitySelection`` passed inside ``HybridVector.near_vector`` /
+``HybridVector.near_text`` is applied by the server as a post-fusion MMR pass
+(Weaviate >= 1.39.0). These tests assert that ``balance=0`` (pure diversity)
+produces a different ordering than ``balance=1`` (pure relevance), and that
+``mmr.limit`` caps the result count.
+
+The equivalent ``near_vector`` behaviour is covered in
+``test_collection_diversity.py``.
+"""
+
+import pytest
+
+from integration.conftest import CollectionFactory
+from weaviate.classes.query import Diversity, HybridVector
+from weaviate.collections.classes.config import Configure, DataType, Property
+from weaviate.collections.classes.data import DataObject
+
+MIN_VERSION = (1, 39, 0)
+
+
+def _skip_if_unsupported(collection) -> None:
+    if collection._connection._weaviate_version.is_lower_than(*MIN_VERSION):
+        pytest.skip("Hybrid diversity selection requires Weaviate >= 1.39.0")
+
+
+def _create_clustered_collection(collection_factory: CollectionFactory):
+    """Create a collection with 3 tight clusters (a, b, c) of vectors in 3D."""
+    collection = collection_factory(
+        properties=[Property(name="text", data_type=DataType.TEXT)],
+        vectorizer_config=Configure.Vectorizer.none(),
+    )
+    _skip_if_unsupported(collection)
+    collection.data.insert_many(
+        [
+            DataObject(properties={"text": "a1"}, vector=[1.0, 0.0, 0.0]),
+            DataObject(properties={"text": "a2"}, vector=[0.95, 0.05, 0.0]),
+            DataObject(properties={"text": "a3"}, vector=[0.9, 0.1, 0.0]),
+            DataObject(properties={"text": "b1"}, vector=[0.0, 1.0, 0.0]),
+            DataObject(properties={"text": "b2"}, vector=[0.05, 0.95, 0.0]),
+            DataObject(properties={"text": "c1"}, vector=[0.0, 0.0, 1.0]),
+        ]
+    )
+    return collection
+
+
+def _create_large_collection(collection_factory: CollectionFactory, n_items: int = 50):
+    """Create a collection with enough items (>25) that a small mmr.limit is distinguishable from the server's default limit."""
+    collection = collection_factory(
+        properties=[Property(name="text", data_type=DataType.TEXT)],
+        vectorizer_config=Configure.Vectorizer.none(),
+    )
+    _skip_if_unsupported(collection)
+    collection.data.insert_many(
+        [
+            DataObject(properties={"text": f"t{i}"}, vector=[1.0 - 0.001 * i, 0.0, 0.0])
+            for i in range(n_items)
+        ]
+    )
+    return collection
+
+
+def test_hybrid_near_vector_balance_0_differs_from_balance_1(
+    collection_factory: CollectionFactory,
+) -> None:
+    """Hybrid near-vector: balance=0 (diversity) must reorder vs balance=1 (relevance)."""
+    collection = _create_clustered_collection(collection_factory)
+    balance_0 = collection.query.hybrid(
+        query=None,
+        vector=HybridVector.near_vector(
+            vector=[1.0, 0.0, 0.0],
+            diversity_selection=Diversity.mmr(limit=3, balance=0.0),
+        ),
+        limit=3,
+    ).objects
+    balance_1 = collection.query.hybrid(
+        query=None,
+        vector=HybridVector.near_vector(
+            vector=[1.0, 0.0, 0.0],
+            diversity_selection=Diversity.mmr(limit=3, balance=1.0),
+        ),
+        limit=3,
+    ).objects
+    assert [o.uuid for o in balance_0] != [o.uuid for o in balance_1]
+
+
+def test_hybrid_near_vector_balance_1_matches_baseline(
+    collection_factory: CollectionFactory,
+) -> None:
+    """Hybrid near-vector with MMR balance=1 (pure relevance) matches the plain baseline."""
+    collection = _create_clustered_collection(collection_factory)
+    baseline = collection.query.hybrid(
+        query=None,
+        vector=HybridVector.near_vector(vector=[1.0, 0.0, 0.0]),
+        limit=3,
+    ).objects
+    mmr_balance_1 = collection.query.hybrid(
+        query=None,
+        vector=HybridVector.near_vector(
+            vector=[1.0, 0.0, 0.0],
+            diversity_selection=Diversity.mmr(limit=3, balance=1.0),
+        ),
+        limit=3,
+    ).objects
+    assert [o.uuid for o in baseline] == [o.uuid for o in mmr_balance_1]
+
+
+def test_hybrid_alpha_1_balance_0_differs_from_balance_1(
+    collection_factory: CollectionFactory,
+) -> None:
+    """Hybrid with explicit alpha=1.0 (pure vector) applies MMR like near_vector."""
+    collection = _create_clustered_collection(collection_factory)
+    balance_0 = collection.query.hybrid(
+        query="irrelevant",
+        alpha=1.0,
+        vector=HybridVector.near_vector(
+            vector=[1.0, 0.0, 0.0],
+            diversity_selection=Diversity.mmr(limit=3, balance=0.0),
+        ),
+        limit=3,
+    ).objects
+    balance_1 = collection.query.hybrid(
+        query="irrelevant",
+        alpha=1.0,
+        vector=HybridVector.near_vector(
+            vector=[1.0, 0.0, 0.0],
+            diversity_selection=Diversity.mmr(limit=3, balance=1.0),
+        ),
+        limit=3,
+    ).objects
+    assert [o.uuid for o in balance_0] != [o.uuid for o in balance_1]
+
+
+def test_hybrid_respects_mmr_limit(
+    collection_factory: CollectionFactory,
+) -> None:
+    """Hybrid respects mmr.limit as the result-count cap when no outer limit is set."""
+    mmr_limit = 5
+    collection = _create_large_collection(collection_factory, n_items=50)
+
+    result = collection.query.hybrid(
+        query=None,
+        vector=HybridVector.near_vector(
+            vector=[1.0, 0.0, 0.0],
+            diversity_selection=Diversity.mmr(limit=mmr_limit, balance=0.5),
+        ),
+    ).objects
+    assert len(result) == mmr_limit

test/collection/test_hybrid_diversity.py

@@ -0,0 +1,62 @@
+"""Unit tests: hybrid search wires diversity_selection into the gRPC request.
+
+Hybrid diversity is a post-fusion, hybrid-level operation, so the top-level
+``query.hybrid`` / ``generate.hybrid`` ``diversity_selection`` argument must
+populate the top-level ``Hybrid.selection.mmr`` in the SearchRequest proto (not
+the nested ``near_vector`` / ``near_text`` selection).
+"""
+
+from weaviate.collections.grpc.query import _QueryGRPC
+from weaviate.classes.query import Diversity, HybridVector
+from weaviate.util import _ServerVersion
+
+
+_DEFAULT_VERSION = _ServerVersion(1, 38, 0)
+
+
+def _builder(version: _ServerVersion = _DEFAULT_VERSION) -> _QueryGRPC:
+    return _QueryGRPC(
+        weaviate_version=version,
+        name="Dummy",
+        tenant=None,
+        consistency_level=None,
+        validate_arguments=True,
+        uses_125_api=True,
+        uses_127_api=True,
+    )
+
+
+def test_hybrid_near_vector_sets_top_level_selection() -> None:
+    req = _builder().hybrid(
+        query=None,
+        vector=HybridVector.near_vector(vector=[1.0, 0.0, 0.0]),
+        diversity_selection=Diversity.mmr(limit=7, balance=0.0),
+        limit=7,
+    )
+    # Canonical location: top-level Hybrid.selection, not the nested near_vector.
+    mmr = req.hybrid_search.selection.mmr
+    assert mmr.limit == 7
+    assert mmr.balance == 0.0
+    assert not req.hybrid_search.near_vector.HasField("selection")
+
+
+def test_hybrid_near_text_sets_top_level_selection() -> None:
+    req = _builder().hybrid(
+        query=None,
+        vector=HybridVector.near_text(query="cats"),
+        diversity_selection=Diversity.mmr(limit=3, balance=0.5),
+        limit=3,
+    )
+    mmr = req.hybrid_search.selection.mmr
+    assert mmr.limit == 3
+    assert mmr.balance == 0.5
+    assert not req.hybrid_search.near_text.HasField("selection")
+
+
+def test_hybrid_without_selection_leaves_it_unset() -> None:
+    req = _builder().hybrid(
+        query=None,
+        vector=HybridVector.near_vector(vector=[1.0, 0.0, 0.0]),
+        limit=5,
+    )
+    assert not req.hybrid_search.HasField("selection")

weaviate/collections/grpc/query.py

@@ -173,6 +173,7 @@ def hybrid(
         generative: Optional[_Generative] = None,
         rerank: Optional[Rerank] = None,
         boost: Optional[_Boost] = None,
+        diversity_selection: Optional[MMR] = None,
         target_vector: Optional[TargetVectorJoinType] = None,
     ) -> search_get_pb2.SearchRequest:
         return self.__create_request(
@@ -196,6 +197,7 @@ def hybrid(
                 fusion_type,
                 distance,
                 target_vector,
+                diversity_selection,
             ),
         )
 

weaviate/collections/grpc/shared.py

@@ -600,6 +600,7 @@ def _parse_hybrid(
         fusion_type: Optional[HybridFusion],
         distance: Optional[NUMBER],
         target_vector: Optional[TargetVectorJoinType],
+        diversity_selection: Optional[MMR] = None,
     ) -> Union[base_search_pb2.Hybrid, None]:
         if self._validate_arguments:
             _validate_input(
@@ -739,6 +740,7 @@ def _parse_hybrid(
                 vector_bytes=vector_bytes,
                 vector_distance=distance,
                 vectors=vectors,
+                selection=self._diversity_selection_to_grpc(diversity_selection),
                 bm25_search_operator=base_search_pb2.SearchOperatorOptions(
                     operator=bm25_operator.operator,
                     minimum_or_tokens_match=bm25_operator.minimum_should_match

weaviate/collections/queries/hybrid/generate/async_.pyi

@@ -3,6 +3,7 @@ from typing import Generic, List, Literal, Optional, Type, Union, overload
 from weaviate.collections.classes.filters import FilterReturn
 from weaviate.collections.classes.grpc import (
     METADATA,
+    MMR,
     PROPERTIES,
     REFERENCES,
     BM25OperatorOptions,
@@ -56,6 +57,7 @@ class _HybridGenerateAsync(
         group_by: Literal[None] = None,
         rerank: Optional[Rerank] = None,
         boost: Optional[_Boost] = None,
+        diversity_selection: Optional[MMR] = None,
         target_vector: Optional[TargetVectorJoinType] = None,
         include_vector: INCLUDE_VECTOR = False,
         return_metadata: Optional[METADATA] = None,
@@ -84,6 +86,7 @@ class _HybridGenerateAsync(
         group_by: Literal[None] = None,
         rerank: Optional[Rerank] = None,
         boost: Optional[_Boost] = None,
+        diversity_selection: Optional[MMR] = None,
         target_vector: Optional[TargetVectorJoinType] = None,
         include_vector: INCLUDE_VECTOR = False,
         return_metadata: Optional[METADATA] = None,
@@ -112,6 +115,7 @@ class _HybridGenerateAsync(
         group_by: Literal[None] = None,
         rerank: Optional[Rerank] = None,
         boost: Optional[_Boost] = None,
+        diversity_selection: Optional[MMR] = None,
         target_vector: Optional[TargetVectorJoinType] = None,
         include_vector: INCLUDE_VECTOR = False,
         return_metadata: Optional[METADATA] = None,
@@ -140,6 +144,7 @@ class _HybridGenerateAsync(
         group_by: Literal[None] = None,
         rerank: Optional[Rerank] = None,
         boost: Optional[_Boost] = None,
+        diversity_selection: Optional[MMR] = None,
         target_vector: Optional[TargetVectorJoinType] = None,
         include_vector: INCLUDE_VECTOR = False,
         return_metadata: Optional[METADATA] = None,
@@ -168,6 +173,7 @@ class _HybridGenerateAsync(
         group_by: Literal[None] = None,
         rerank: Optional[Rerank] = None,
         boost: Optional[_Boost] = None,
+        diversity_selection: Optional[MMR] = None,
         target_vector: Optional[TargetVectorJoinType] = None,
         include_vector: INCLUDE_VECTOR = False,
         return_metadata: Optional[METADATA] = None,
@@ -196,6 +202,7 @@ class _HybridGenerateAsync(
         group_by: Literal[None] = None,
         rerank: Optional[Rerank] = None,
         boost: Optional[_Boost] = None,
+        diversity_selection: Optional[MMR] = None,
         target_vector: Optional[TargetVectorJoinType] = None,
         include_vector: INCLUDE_VECTOR = False,
         return_metadata: Optional[METADATA] = None,
@@ -224,6 +231,7 @@ class _HybridGenerateAsync(
         group_by: GroupBy,
         rerank: Optional[Rerank] = None,
         boost: Optional[_Boost] = None,
+        diversity_selection: Optional[MMR] = None,
         target_vector: Optional[TargetVectorJoinType] = None,
         include_vector: INCLUDE_VECTOR = False,
         return_metadata: Optional[METADATA] = None,
@@ -252,6 +260,7 @@ class _HybridGenerateAsync(
         group_by: GroupBy,
         rerank: Optional[Rerank] = None,
         boost: Optional[_Boost] = None,
+        diversity_selection: Optional[MMR] = None,
         target_vector: Optional[TargetVectorJoinType] = None,
         include_vector: INCLUDE_VECTOR = False,
         return_metadata: Optional[METADATA] = None,
@@ -280,6 +289,7 @@ class _HybridGenerateAsync(
         group_by: GroupBy,
         rerank: Optional[Rerank] = None,
         boost: Optional[_Boost] = None,
+        diversity_selection: Optional[MMR] = None,
         target_vector: Optional[TargetVectorJoinType] = None,
         include_vector: INCLUDE_VECTOR = False,
         return_metadata: Optional[METADATA] = None,
@@ -308,6 +318,7 @@ class _HybridGenerateAsync(
         group_by: GroupBy,
         rerank: Optional[Rerank] = None,
         boost: Optional[_Boost] = None,
+        diversity_selection: Optional[MMR] = None,
         target_vector: Optional[TargetVectorJoinType] = None,
         include_vector: INCLUDE_VECTOR = False,
         return_metadata: Optional[METADATA] = None,
@@ -336,6 +347,7 @@ class _HybridGenerateAsync(
         group_by: GroupBy,
         rerank: Optional[Rerank] = None,
         boost: Optional[_Boost] = None,
+        diversity_selection: Optional[MMR] = None,
         target_vector: Optional[TargetVectorJoinType] = None,
         include_vector: INCLUDE_VECTOR = False,
         return_metadata: Optional[METADATA] = None,
@@ -364,6 +376,7 @@ class _HybridGenerateAsync(
         group_by: GroupBy,
         rerank: Optional[Rerank] = None,
         boost: Optional[_Boost] = None,
+        diversity_selection: Optional[MMR] = None,
         target_vector: Optional[TargetVectorJoinType] = None,
         include_vector: INCLUDE_VECTOR = False,
         return_metadata: Optional[METADATA] = None,
@@ -392,6 +405,7 @@ class _HybridGenerateAsync(
         group_by: Optional[GroupBy] = None,
         rerank: Optional[Rerank] = None,
         boost: Optional[_Boost] = None,
+        diversity_selection: Optional[MMR] = None,
         target_vector: Optional[TargetVectorJoinType] = None,
         include_vector: INCLUDE_VECTOR = False,
         return_metadata: Optional[METADATA] = None,