Vectors/deprecate legacy introduce named vectors only syntax (#1659)

* Introduce `Configure.Vectors` and `vector_config` in `.create`/`.update`

* Update CI images to latest build on each branch

* Change prop descr test to handle props sorted by name

* Make changes to pass unit tests of legacy vectorization

* Alter specific tests for old versions with new syntax

* Add skip to test with resolved issue absent from previous vers

* - Add `MultiVectors` factory
- Allow `quantizer` at root of all `Vectors` and `MultiVectors` methods
- Allow `encoding` also at root of all `MultiVectors` methods

* Update CI tags to latest patch

* Rename `.text2colbert_jinaa()` to `text2vec_jinaai()`

* Add warnings for all deprecated syntax

* Response to review comments
- Add `multi_vector_config` arg to `MultiVectors` methods
- Add deprs warnings if `multi_vector`
- Update tests to use new syntax

* Update tests to reflect single vectors usage

* Fix vector config name passing when creating collections

* Skip parsing vector to obj if it's empty

* Update CI images to latest patches

* Remove unneeded added parsing

Author: tsmith023

.github/workflows/main.yaml

@@ -20,12 +20,12 @@ concurrency:
 env:
   WEAVIATE_125: 1.25.34
   WEAVIATE_126: 1.26.17
-  WEAVIATE_127: 1.27.14
-  WEAVIATE_128: 1.28.8
-  WEAVIATE_129: 1.29.8
-  WEAVIATE_130: 1.30.9
+  WEAVIATE_127: 1.27.27
+  WEAVIATE_128: 1.28.16
+  WEAVIATE_129: 1.29.9
+  WEAVIATE_130: 1.30.11
   WEAVIATE_131: 1.31.5
-  WEAVIATE_132: 1.32.0-rc.0-6eb89d6.amd64
+  WEAVIATE_132: 1.32.0-rc.1-098c594
 
 jobs:
   lint-and-format:

.pre-commit-config.yaml

@@ -49,4 +49,4 @@ repos:
       language: node
       pass_filenames: false
       types: [python]
-      additional_dependencies: [pyright@1.1.398]
+      additional_dependencies: [pyright@1.1.400]

docs/conf.py

@@ -6,7 +6,6 @@
 import contextlib
 import os
 import sys
-
 from importlib.metadata import PackageNotFoundError
 from importlib.metadata import version as version_func
 

integration/conftest.py

@@ -28,6 +28,7 @@
     _ReferencePropertyBase,
     _ReplicationConfigCreate,
     _RerankerProvider,
+    _VectorConfigCreate,
     _VectorIndexConfigCreate,
     _VectorizerConfigCreate,
 )
@@ -58,6 +59,9 @@ def __call__(
         vector_index_config: Optional[_VectorIndexConfigCreate] = None,
         description: Optional[str] = None,
         reranker_config: Optional[_RerankerProvider] = None,
+        vector_config: Optional[
+            Optional[Union[_VectorConfigCreate, List[_VectorConfigCreate]]]
+        ] = None,
     ) -> Collection[Any, Any]:
         """Typing for fixture."""
         ...
@@ -129,6 +133,9 @@ def _factory(
         vector_index_config: Optional[_VectorIndexConfigCreate] = None,
         description: Optional[str] = None,
         reranker_config: Optional[_RerankerProvider] = None,
+        vector_config: Optional[
+            Optional[Union[_VectorConfigCreate, List[_VectorConfigCreate]]]
+        ] = None,
     ) -> Collection[Any, Any]:
         try:
             nonlocal client_fixture, name_fixtures, call_counter  # noqa: F824
@@ -148,7 +155,8 @@ def _factory(
             collection: Collection[Any, Any] = client_fixture.collections.create(
                 name=name_fixture,
                 description=description,
-                vectorizer_config=vectorizer_config or Configure.Vectorizer.none(),
+                vectorizer_config=vectorizer_config
+                or (Configure.Vectorizer.none() if vector_config is None else None),
                 properties=properties,
                 references=references,
                 inverted_index_config=inverted_index_config,
@@ -159,6 +167,7 @@ def _factory(
                 replication_config=replication_config,
                 vector_index_config=vector_index_config,
                 reranker_config=reranker_config,
+                vector_config=vector_config,
             )
             return collection
         except Exception as e:
@@ -315,6 +324,9 @@ def __call__(
         vectorizer_config: Optional[
             Union[_VectorizerConfigCreate, List[_NamedVectorConfigCreate]]
         ] = None,
+        vector_config: Optional[
+            Optional[Union[_VectorConfigCreate, List[_VectorConfigCreate]]]
+        ] = None,
     ) -> Collection[Any, Any]:
         """Typing for fixture."""
         ...
@@ -329,17 +341,18 @@ def _factory(
         vectorizer_config: Optional[
             Union[_VectorizerConfigCreate, List[_NamedVectorConfigCreate]]
         ] = None,
+        vector_config: Optional[
+            Optional[Union[_VectorConfigCreate, List[_VectorConfigCreate]]]
+        ] = None,
     ) -> Collection[Any, Any]:
         api_key = os.environ.get("OPENAI_APIKEY")
         if api_key is None:
             pytest.skip("No OpenAI API key found.")
 
-        if vectorizer_config is None:
-            vectorizer_config = Configure.Vectorizer.none()
-
         collection = collection_factory(
             name=name,
-            vectorizer_config=vectorizer_config or Configure.Vectorizer.none(),
+            vectorizer_config=vectorizer_config,
+            vector_config=vector_config or Configure.Vectors.self_provided(),
             properties=[
                 Property(name="text", data_type=DataType.TEXT),
                 Property(name="content", data_type=DataType.TEXT),

integration/test_client.py

@@ -1,4 +1,4 @@
-from typing import Callable, Generator, Tuple, Union
+from typing import Callable, Generator, Optional, Tuple, Union
 
 import pytest
 from _pytest.fixtures import SubRequest
@@ -28,7 +28,7 @@
 
 @pytest.fixture(scope="module")
 def client_factory() -> Generator[Callable[[int, int], weaviate.WeaviateClient], None, None]:
-    client: weaviate.WeaviateClient = None
+    client: Optional[weaviate.WeaviateClient] = None
 
     def maker(http: int, grpc: int) -> weaviate.WeaviateClient:
         nonlocal client
@@ -593,7 +593,7 @@ async def test_async_client_with_extra_options() -> None:
 
 def test_client_error_for_wcs_without_auth() -> None:
     with pytest.raises(weaviate.exceptions.AuthenticationFailedError) as e:
-        weaviate.connect_to_wcs(cluster_url=WCS_URL, auth_credentials=None)
+        weaviate.connect_to_wcs(cluster_url=WCS_URL, auth_credentials=None)  # pyright: ignore
         assert "wvc.init.Auth.api_key" in e.value.message