Module weaviate
![](https://bcentral-packageicons.azureedge.net/images/ballerinax_weaviate_1.0.0.png)
ballerinax/weaviate Ballerina library
Overview
This is a generated connector for the Weaviate Vector Search Engine API OpenAPI specification. Weaviate is an open-source vector search engine, which allows storing data objects and vector embeddings from the ML models, including the LLMs offered by OpenAI, Hugging Face, and Cohere. Weaviate provides a powerful GraphQL API for querying the embeddings while looking at the similarity and can scale seamlessly into billions of data objects.
Prerequisites
Before using this connector in your Ballerina application, complete the following:
- Create a Weaviate Cluster using the Weaviate Cloud Service or deploy using Docker/Kubernetes.
- Obtain the OIDC Authentication key.
Quick start
To use the Weaviate connector in your Ballerina application, update the .bal
file as follows:
Step 1: Import the connector
First, import the ballerinax/weaviate
module into the Ballerina project.
import ballerinax/weaviate;
Step 2: Create a new connector instance
Create and initialize a weaviate:Client
with your Service URL
and the obtained Authentication key
.
weaviate:Client weaviateClient = check new ({ auth: { token: "sk-XXXXXXXXX" } }, serviceURL: "https://weaviate-server:port/v1");
Step 3: Invoke the connector operation
- Now, you can use the operations available within the connector.
Note: that they are in the form of remote operations.
Following is an example of inserting new objects to the Weaviate vector storage as a batch operation.
string className = "DocClass"; // weaviate class name string[] text; // list of text float[][] embeddings; // list of embbedings for the texts int len = text.length(); // Creates the batch of Weaviate objects. weaviate:Object[] objArr = []; foreach int i in 0...len { objArr.push( { 'class: className, vector: embeddings[i], properties: { "docs": text[i] } }); } weaviate:ObjectsGetResponse[] responseArray = check weaviateClient->/batch/objects.post({ objects: objArr });
Once the new records are inserted, you can query the Weaviate vector storage using the Weaviate GraphQL API, similar to the example below.
float[] embeddings; // This is the embedding for the text being searched for similar content. string graphQLQuery = string`{ Get { DocStore ( nearVector: { vector: ${embeddings.toString()} } limit: 5 ){ docs _additional { certainty, id } } } }`; weaviate:GraphQLResponse|error results = check weaviateClient->/graphql.post({ query: graphQLQuery });
- Use the
bal run
command to compile and run the Ballerina program.