addCollectionFunction()

Addedv2.6.x

This operation adds a custom function to an existing collection.

await milvusClient.addCollectionFunction(data: AddCollectionFunctionReq)

Request Syntax

await milvusClient.addCollectionFunction({
    collection_name: string,
    function: FunctionObject,
    db_name?: string,
    timeout?: number,
})

PARAMETERS:

• collection_name (string) -

  [REQUIRED]

  The name of the collection to add the function to.

• function (FunctionObject) -

  [REQUIRED]

  The function to add to the collection. For the full field reference, see the FunctionObject section below.

• db_name (string) -

  The name of the database where the collection resides.

• timeout (number) -

  The timeout duration in milliseconds for this operation.

RETURNS:

Promise<ResStatus>

EXCEPTIONS:

• MilvusError

  This exception will be raised when any error occurs during this operation.

FunctionObject

A FunctionObject defines a server-side function that automatically transforms data into vector embeddings during insert and search.

PARAMETERS:

• name (string) -

  [REQUIRED]

  The name of the function. Used to reference the function within queries and collections.

• type (FunctionType) -

  [REQUIRED]

  The function type. Possible values: FunctionType.BM25 (sparse embeddings from text using BM25), FunctionType.TEXTEMBEDDING (dense embeddings from text), FunctionType.RERANK (reranking function).

• input_field_names (string[]) -

  [REQUIRED]

  The names of the fields containing the raw data to transform. For FunctionType.BM25, exactly one field name is expected.

• output_field_names (string[]) -

  The names of the fields where the generated embeddings will be stored. For FunctionType.BM25, exactly one field name is expected.

• params (object) -

  Additional function parameters in key-value pairs.

• description (string) -

  A brief description of the function's purpose. Defaults to an empty string.

Example

import { MilvusClient, FunctionType } from '@zilliz/milvus2-sdk-node';

const milvusClient = new MilvusClient({
    address: 'YOUR_CLUSTER_ENDPOINT',
    token: 'YOUR_CLUSTER_TOKEN',
});

const resStatus = await milvusClient.addCollectionFunction({
    collection_name: 'my_collection',
    function: {
        name: 'my_bm25_function',
        description: 'BM25 sparse embedding function',
        type: FunctionType.BM25,
        input_field_names: ['text'],
        output_field_names: ['sparse_vector'],
        params: {},
    },
});