feat: add support for multi2multivec-jinaai vectorizer (#317)

* feat: add support for multi2multivec-jinaai vectorizer

* chore: deprecate vectorizeCollectionName parameter for multi2vec-jinaai

* refactor: move multi2MultivecJinaAI static factory to multiVectors namespace

* chore: lint & format

* chore: rename static factory to muli2vecJinaAI for consistency with text2VecJinaAI

Author: bevzzz

src/collections/config/types/vectorizer.ts

@@ -25,6 +25,7 @@ export type Vectorizer =
   | Multi2VecPalmVectorizer
   | 'multi2vec-google'
   | 'multi2vec-jinaai'
+  | 'multi2multivec-jinaai'
   | 'multi2vec-voyageai'
   | 'ref2vec-centroid'
   | 'text2vec-aws'
@@ -190,6 +191,19 @@ export type Multi2VecGoogleConfig = {
   };
 };
 
+/** The configuration for multi-media-to-multi-vector vectorization using
+ * the jina-embeddings-v4 model
+ *
+ * See the [documentation](https://weaviate.io/developers/weaviate/model-providers/jinaai/embeddings-multimodal) for detailed usage.
+ */
+export type Multi2MultivecJinaAIConfig = {
+  /** The image fields used when vectorizing. */
+  imageFields?: string[];
+
+  /** The text fields used when vectorizing. */
+  textFields?: string[];
+};
+
 /** The configuration for multi-media vectorization using the Jina module.
  *
  * See the [documentation](https://weaviate.io/developers/weaviate/model-providers/jinaai/embeddings-multimodal) for detailed usage.
@@ -205,7 +219,11 @@ export type Multi2VecJinaAIConfig = {
   model?: string;
   /** The text fields used when vectorizing. */
   textFields?: string[];
-  /** Whether the collection name is vectorized. */
+  /**
+   * Whether the collection name is vectorized.
+   *
+   * @deprecated This parameter is not applicable and has no effect on the underlying module.
+   * */
   vectorizeCollectionName?: boolean;
   /** The weights of the fields used for vectorization. */
   weights?: {
@@ -525,6 +543,7 @@ export type VectorizerConfig =
   | Multi2VecBindConfig
   | Multi2VecGoogleConfig
   | Multi2VecJinaAIConfig
+  | Multi2MultivecJinaAIConfig
   | Multi2VecPalmConfig
   | Multi2VecVoyageAIConfig
   | Ref2VecCentroidConfig
@@ -556,6 +575,8 @@ export type VectorizerConfigType<V> = V extends 'img2vec-neural'
   ? Multi2VecGoogleConfig
   : V extends 'multi2vec-jinaai'
   ? Multi2VecJinaAIConfig | undefined
+  : V extends 'multi2multivec-jinaai'
+  ? Multi2MultivecJinaAIConfig | undefined
   : V extends Multi2VecPalmVectorizer
   ? Multi2VecPalmConfig
   : V extends 'multi2vec-voyageai'

src/collections/configure/types/vectorizer.ts

@@ -160,6 +160,13 @@ export type Multi2VecCohereConfigCreate = {
   vectorizeCollectionName?: boolean;
 };
 
+export type Multi2MultivecJinaAIConfigCreate = {
+  /** The image fields to use in vectorization. */
+  imageFields?: string[];
+  /** The text fields to use in vectorization. */
+  textFields?: string[];
+};
+
 export type Multi2VecJinaAIConfigCreate = {
   /** The base URL to use where API requests should go. */
   baseURL?: string;
@@ -262,6 +269,8 @@ export type VectorizerConfigCreateType<V> = V extends 'img2vec-neural'
   ? Multi2VecBindConfigCreate | undefined
   : V extends 'multi2vec-jinaai'
   ? Multi2VecJinaAIConfigCreate | undefined
+  : V extends 'multi2multivec-jinaai'
+  ? Multi2MultivecJinaAIConfigCreate | undefined
   : V extends 'multi2vec-palm'
   ? Multi2VecPalmConfigCreate
   : V extends 'multi2vec-google'

src/collections/configure/unit.test.ts

@@ -1,3 +1,4 @@
+import { requireAtLeast } from '../../../test/version.js';
 import {
   GenerativeAWSConfig,
   GenerativeAnthropicConfig,
@@ -702,6 +703,28 @@ describe('Unit testing of the vectorizer factory class', () => {
     });
   });
 
+  requireAtLeast(1, 32, 0).it('should create the correct Multi2MultivecJinaAIConfig with values', () => {
+    const config = configure.multiVectors.multi2VecJinaAI({
+      name: 'multi-jina',
+      imageFields: ['field1', 'field2'],
+      textFields: ['field3', 'field4'],
+    });
+    expect(config).toEqual<VectorConfigCreate<never, string, 'hnsw', 'multi2multivec-jinaai'>>({
+      name: 'multi-jina',
+      vectorIndex: {
+        name: 'hnsw',
+        config: undefined,
+      },
+      vectorizer: {
+        name: 'multi2multivec-jinaai',
+        config: {
+          imageFields: ['field1', 'field2'],
+          textFields: ['field3', 'field4'],
+        },
+      },
+    });
+  });
+
   it('should create the correct Multi2VecPalmConfig type using deprecated method with defaults', () => {
     const config = configure.vectors.multi2VecPalm({
       projectId: 'project-id',

src/collections/configure/vectorizer.ts

@@ -265,6 +265,7 @@ export const vectors = {
       },
     });
   },
+
   /**
    * Create a `VectorConfigCreate` object with the vectorizer set to `'multi2vec-jinaai'`.
    *
@@ -848,4 +849,25 @@ export const multiVectors = {
       true
     );
   },
+
+  /**
+   * Create a `VectorConfigCreate` object with the vectorizer set to `'multi2multivec-jinaai'`.
+   *
+   * See the [documentation](https://weaviate.io/developers/weaviate/model-providers/jinaai/embeddings-multimodal) for detailed usage.
+   *
+   * @param {ConfigureNonTextVectorizerOptions<N, I, 'multi2multivec-jinaai'>} [opts] The configuration options for the `multi2multivec-jinaai` vectorizer.
+   * @returns {VectorConfigCreate<PrimitiveKeys<T>[], N, I, 'multi2multivec-jinaai'>} The configuration object.
+   */
+  multi2VecJinaAI: <N extends string | undefined = undefined, I extends VectorIndexType = 'hnsw'>(
+    opts?: ConfigureNonTextVectorizerOptions<N, I, 'multi2multivec-jinaai'>
+  ): VectorConfigCreate<never, N, I, 'multi2multivec-jinaai'> => {
+    const { name, vectorIndexConfig, ...config } = opts || {};
+    return makeVectorizer(name, {
+      vectorIndexConfig,
+      vectorizerConfig: {
+        name: 'multi2multivec-jinaai',
+        config,
+      },
+    });
+  },
 };