add text2vec-weaviate vectorizer

Author: augustas1

src/collections/config/types/vectorizer.ts

@@ -40,6 +40,7 @@ export type Vectorizer =
   | 'text2vec-google'
   | 'text2vec-transformers'
   | 'text2vec-voyageai'
+  | 'text2vec-weaviate'
   | 'none';
 
 /** The configuration for image vectorization using a neural network module.
@@ -406,6 +407,20 @@ export type Text2VecVoyageAIConfig = {
   vectorizeCollectionName?: boolean;
 };
 
+/**
+ * The configuration for text vectorization using Weaviate's self-hosted text-based embedding models.
+ *
+ * TODO: add documentation reference once available.
+ */
+export interface Text2VecWeaviateConfig {
+  /** The base URL to use where API requests should go. */
+  baseURL?: string;
+  /** The model to use. */
+  model?: 'Snowflake/snowflake-arctic-embed-m-v1.5' | string;
+  /** Whether to vectorize the collection name. */
+  vectorizeCollectionName?: boolean;
+};
+
 export type NoVectorizerConfig = {};
 
 export type VectorizerConfig =
@@ -428,6 +443,7 @@ export type VectorizerConfig =
   | Text2VecPalmConfig
   | Text2VecTransformersConfig
   | Text2VecVoyageAIConfig
+  | Text2VecWeaviateConfig
   | NoVectorizerConfig;
 
 export type VectorizerConfigType<V> = V extends 'img2vec-neural'
@@ -474,6 +490,8 @@ export type VectorizerConfigType<V> = V extends 'img2vec-neural'
   ? Text2VecTransformersConfig | undefined
   : V extends 'text2vec-voyageai'
   ? Text2VecVoyageAIConfig | undefined
+  : V extends 'text2vec-weaviate'
+  ? Text2VecWeaviateConfig | undefined
   : V extends 'none'
   ? {}
   : V extends undefined

src/collections/configure/types/vectorizer.ts

@@ -17,6 +17,7 @@ import {
   Text2VecOpenAIConfig,
   Text2VecTransformersConfig,
   Text2VecVoyageAIConfig,
+  Text2VecWeaviateConfig,
   VectorIndexType,
   Vectorizer,
   VectorizerConfigType,
@@ -182,6 +183,8 @@ export type Text2VecTransformersConfigCreate = Text2VecTransformersConfig;
 
 export type Text2VecVoyageAIConfigCreate = Text2VecVoyageAIConfig;
 
+export type Text2VecWeaviateConfigCreate = Text2VecWeaviateConfig;
+
 export type VectorizerConfigCreateType<V> = V extends 'img2vec-neural'
   ? Img2VecNeuralConfigCreate | undefined
   : V extends 'multi2vec-clip'
@@ -226,6 +229,8 @@ export type VectorizerConfigCreateType<V> = V extends 'img2vec-neural'
   ? Text2VecTransformersConfigCreate | undefined
   : V extends 'text2vec-voyageai'
   ? Text2VecVoyageAIConfigCreate | undefined
+  : V extends 'text2vec-weaviate'
+  ? Text2VecWeaviateConfigCreate | undefined
   : V extends 'none'
   ? {}
   : V extends undefined

src/collections/configure/unit.test.ts

@@ -1342,6 +1342,45 @@ describe('Unit testing of the vectorizer factory class', () => {
       },
     });
   });
+
+  it('should create the correct Text2VecWeaviateConfig type with defaults', () => {
+    const config = configure.vectorizer.text2VecWeaviate();
+    expect(config).toEqual<VectorConfigCreate<never, undefined, 'hnsw', 'text2vec-weaviate'>>({
+      name: undefined,
+      vectorIndex: {
+        name: 'hnsw',
+        config: undefined,
+      },
+      vectorizer: {
+        name: 'text2vec-weaviate',
+        config: undefined,
+      },
+    });
+  });
+
+  it('should create the correct Text2VecWeaviateConfig type with all values', () => {
+    const config = configure.vectorizer.text2VecWeaviate({
+      name: 'test',
+      baseURL: 'base-url',
+      model: 'model',
+      vectorizeCollectionName: true,
+    });
+    expect(config).toEqual<VectorConfigCreate<never, 'test', 'hnsw', 'text2vec-weaviate'>>({
+      name: 'test',
+      vectorIndex: {
+        name: 'hnsw',
+        config: undefined,
+      },
+      vectorizer: {
+        name: 'text2vec-weaviate',
+        config: {
+          baseURL: 'base-url',
+          model: 'model',
+          vectorizeCollectionName: true,
+        },
+      },
+    });
+  });
 });
 
 describe('Unit testing of the generative factory class', () => {

src/collections/configure/vectorizer.ts

@@ -600,4 +600,26 @@ export const vectorizer = {
       },
     });
   },
+
+  /**
+   * Create a `VectorConfigCreate` object with the vectorizer set to `'text2vec-weaviate'`.
+   *
+   * TODO: add documentation reference once available.
+   *
+   * @param {ConfigureTextVectorizerOptions<T, N, I, 'text2vec-weaviate'>} [opts] The configuration for the `text2vec-weaviate` vectorizer.
+   * @returns {VectorConfigCreate<PrimitiveKeys<T>, N, I, 'text2vec-weaviate'>} The configuration object.
+   */
+  text2VecWeaviate: <T, N extends string | undefined = undefined, I extends VectorIndexType = 'hnsw'>(
+    opts?: ConfigureTextVectorizerOptions<T, N, I, 'text2vec-weaviate'>
+  ): VectorConfigCreate<PrimitiveKeys<T>, N, I, 'text2vec-weaviate'> => {
+    const { name, sourceProperties, vectorIndexConfig, ...config } = opts || {};
+    return makeVectorizer(name, {
+      sourceProperties,
+      vectorIndexConfig,
+      vectorizerConfig: {
+        name: 'text2vec-weaviate',
+        config: Object.keys(config).length === 0 ? undefined : config,
+      },
+    });
+  },
 };