Named vectors

.github/workflows/main.yaml

@@ -15,7 +15,7 @@ on:
 
 env:
   OLD_WEAVIATE_VERSION: 1.23.10
-  NEW_WEAVIATE_VERSION: preview--f005cb8
+  NEW_WEAVIATE_VERSION: 1.24.0-rc.1
 
 jobs:
   lint-and-format:

integration/conftest.py

@@ -1,5 +1,5 @@
 import os
-from typing import Any, Optional, List, Generator, Protocol, Type, Dict, Tuple
+from typing import Any, Optional, List, Generator, Protocol, Type, Dict, Tuple, Union
 
 import pytest
 from _pytest.fixtures import SubRequest
@@ -22,6 +22,8 @@
 from weaviate.collections.classes.types import Properties
 from weaviate.config import AdditionalConfig
 
+from weaviate.collections.classes.config_named_vectors import _NamedVectorConfigCreate
+
 
 class CollectionFactory(Protocol):
     """Typing for fixture."""
@@ -31,7 +33,9 @@ def __call__(
         name: str = "",
         properties: Optional[List[Property]] = None,
         references: Optional[List[_ReferencePropertyBase]] = None,
-        vectorizer_config: Optional[_VectorizerConfigCreate] = None,
+        vectorizer_config: Optional[
+            Union[_VectorizerConfigCreate, List[_NamedVectorConfigCreate]]
+        ] = None,
         inverted_index_config: Optional[_InvertedIndexConfigCreate] = None,
         multi_tenancy_config: Optional[_MultiTenancyConfigCreate] = None,
         generative_config: Optional[_GenerativeConfigCreate] = None,
@@ -57,7 +61,9 @@ def _factory(
         name: str = "",
         properties: Optional[List[Property]] = None,
         references: Optional[List[_ReferencePropertyBase]] = None,
-        vectorizer_config: Optional[_VectorizerConfigCreate] = None,
+        vectorizer_config: Optional[
+            Union[_VectorizerConfigCreate, List[_NamedVectorConfigCreate]]
+        ] = None,
         inverted_index_config: Optional[_InvertedIndexConfigCreate] = None,
         multi_tenancy_config: Optional[_MultiTenancyConfigCreate] = None,
         generative_config: Optional[_GenerativeConfigCreate] = None,
@@ -109,7 +115,11 @@ class OpenAICollection(Protocol):
     """Typing for fixture."""
 
     def __call__(
-        self, name: str = "", vectorizer_config: Optional[_VectorizerConfigCreate] = None
+        self,
+        name: str = "",
+        vectorizer_config: Optional[
+            Union[_VectorizerConfigCreate, List[_NamedVectorConfigCreate]]
+        ] = None,
     ) -> Collection[Any, Any]:
         """Typing for fixture."""
         ...
@@ -120,7 +130,10 @@ def openai_collection(
     collection_factory: CollectionFactory,
 ) -> Generator[OpenAICollection, None, None]:
     def _factory(
-        name: str = "", vectorizer_config: Optional[_VectorizerConfigCreate] = None
+        name: str = "",
+        vectorizer_config: Optional[
+            Union[_VectorizerConfigCreate, List[_NamedVectorConfigCreate]]
+        ] = None,
     ) -> Collection[Any, Any]:
         api_key = os.environ.get("OPENAI_APIKEY")
         if api_key is None:

integration/test_batch_v4.py

@@ -1,6 +1,6 @@
 import uuid
 from dataclasses import dataclass
-from typing import Generator, List, Optional, Sequence, Protocol, Tuple, Callable
+from typing import Generator, List, Optional, Protocol, Tuple, Callable
 
 import pytest
 from _pytest.fixtures import SubRequest
@@ -20,7 +20,7 @@
     ReferenceToMulti,
 )
 from weaviate.collections.classes.tenants import Tenant
-from weaviate.types import UUID
+from weaviate.types import UUID, VECTORS
 
 UUID1 = uuid.UUID("806827e0-2b31-43ca-9269-24fa95a221f9")
 UUID2 = uuid.UUID("8ad0d33c-8db1-4437-87f3-72161ca2a51a")
@@ -146,7 +146,7 @@ def test_flushing(client_factory: ClientFactory) -> None:
 def test_add_object(
     client_factory: ClientFactory,
     uid: Optional[UUID],
-    vector: Optional[Sequence],
+    vector: Optional[VECTORS],
 ) -> None:
     client, name = client_factory()
     with client.batch.fixed_size() as batch:

integration/test_client.py

@@ -264,6 +264,45 @@ def test_create_export_and_recreate(client: weaviate.WeaviateClient, request: Su
     assert not client.collections.exists(name2)
 
 
+def test_create_export_and_recreate_named_vectors(
+    client: weaviate.WeaviateClient, request: SubRequest
+) -> None:
+    if client._connection._weaviate_version.is_lower_than(1, 24, 0):
+        pytest.skip("Named vectors are not supported in versions lower than 1.24.0")
+
+    name1 = request.node.name
+    name2 = request.node.name + "2"
+    client.collections.delete([name1, name2])
+
+    col = client.collections.create(
+        name=name1,
+        properties=[
+            Property(
+                name="name",
+                data_type=DataType.TEXT,
+                vectorize_property_name=True,
+            ),
+        ],
+        vectorizer_config=[
+            Configure.NamedVectors.text2vec_contextionary(
+                "name",
+                source_properties=["name"],
+                vectorize_collection_name=False,
+            ),
+            Configure.NamedVectors.none("custom", vector_index_config=Configure.VectorIndex.flat()),
+        ],
+    )
+    conf = col.config.get()
+    conf.name = name2
+
+    col2 = client.collections.create_from_config(conf)
+
+    conf2 = col2.config.get()
+    assert conf2.vector_config == conf.vector_config
+
+    client.collections.delete([name1, name2])
+
+
 def test_collection_name_capitalization(
     client: weaviate.WeaviateClient, request: SubRequest
 ) -> None:

integration/test_collection_batch.py

@@ -1,6 +1,6 @@
 import uuid
 from dataclasses import dataclass
-from typing import Generator, Optional, Sequence, Union, Any, Protocol
+from typing import Generator, Optional, Union, Any, Protocol
 
 import pytest
 
@@ -16,6 +16,8 @@
 from weaviate.collections.classes.internal import _CrossReference, ReferenceToMulti
 from weaviate.collections.classes.tenants import Tenant
 
+from weaviate.types import VECTORS
+
 UUID = Union[str, uuid.UUID]
 
 
@@ -83,7 +85,7 @@ def _factory(name: str = "", multi_tenancy: bool = False) -> Collection[Any, Any
 )
 @pytest.mark.parametrize("uuid", [None, UUID1, str(UUID2), UUID3.hex])
 def test_add_object(
-    batch_collection: BatchCollection, uuid: Optional[UUID], vector: Optional[Sequence]
+    batch_collection: BatchCollection, uuid: Optional[UUID], vector: Optional[VECTORS]
 ) -> None:
     collection = batch_collection()
 

integration/test_collection_config.py

@@ -357,7 +357,7 @@ def test_collection_config_update(collection_factory: CollectionFactory) -> None
             stopwords_removals=["the"],
         ),
         replication_config=Reconfigure.replication(factor=2),
-        vector_index_config=Reconfigure.VectorIndex.hnsw(
+        vectorizer_config=Reconfigure.VectorIndex.hnsw(
             vector_cache_max_objects=2000000,
             quantizer=Reconfigure.VectorIndex.Quantizer.pq(
                 bit_compression=True,
@@ -405,7 +405,7 @@ def test_collection_config_update(collection_factory: CollectionFactory) -> None
     assert config.vector_index_type == VectorIndexType.HNSW
 
     collection.config.update(
-        vector_index_config=Reconfigure.VectorIndex.hnsw(
+        vectorizer_config=Reconfigure.VectorIndex.hnsw(
             quantizer=Reconfigure.VectorIndex.Quantizer.pq(enabled=False),
         )
     )
@@ -449,6 +449,7 @@ def test_hnsw_with_bq(collection_factory: CollectionFactory) -> None:
 
     config = collection.config.get()
     assert config.vector_index_type == VectorIndexType.HNSW
+    assert config.vector_index_config is not None
     assert isinstance(config.vector_index_config.quantizer, _BQConfig)
 
 
@@ -462,25 +463,27 @@ def test_update_flat(collection_factory: CollectionFactory) -> None:
 
     config = collection.config.get()
     assert config.vector_index_type == VectorIndexType.FLAT
+    assert config.vector_index_config is not None
     assert config.vector_index_config.vector_cache_max_objects == 5
     assert isinstance(config.vector_index_config.quantizer, _BQConfig)
     assert config.vector_index_config.quantizer.rescore_limit == 10
 
     collection.config.update(
-        vector_index_config=Reconfigure.VectorIndex.flat(
+        vectorizer_config=Reconfigure.VectorIndex.flat(
             vector_cache_max_objects=10,
             quantizer=Reconfigure.VectorIndex.Quantizer.bq(rescore_limit=20),
         ),
     )
     config = collection.config.get()
     assert config.vector_index_type == VectorIndexType.FLAT
+    assert config.vector_index_config is not None
     assert config.vector_index_config.vector_cache_max_objects == 10
     assert isinstance(config.vector_index_config.quantizer, _BQConfig)
     assert config.vector_index_config.quantizer.rescore_limit == 20
 
     # Cannot currently disabled BQ after it has been enabled
     # collection.config.update(
-    #     vector_index_config=Reconfigure.VectorIndex.flat(
+    #     vectorizer_config=Reconfigure.VectorIndex.flat(
     #         quantizer=Reconfigure.VectorIndex.Quantizer.bq(enabled=False),
     #     )
     # )
@@ -567,6 +570,7 @@ def test_config_vector_index_flat_and_quantizer_bq(collection_factory: Collectio
 
     conf = collection.config.get()
     assert conf.vector_index_type == VectorIndexType.FLAT
+    assert conf.vector_index_config is not None
     assert conf.vector_index_config.vector_cache_max_objects == 234
     assert isinstance(conf.vector_index_config.quantizer, _BQConfig)
     assert conf.vector_index_config.quantizer.rescore_limit == 456
@@ -583,6 +587,7 @@ def test_config_vector_index_hnsw_and_quantizer_pq(collection_factory: Collectio
 
     conf = collection.config.get()
     assert conf.vector_index_type == VectorIndexType.HNSW
+    assert conf.vector_index_config is not None
     assert conf.vector_index_config.vector_cache_max_objects == 234
     assert isinstance(conf.vector_index_config, _VectorIndexConfigHNSW)
     assert conf.vector_index_config.ef_construction == 789