pinecone 0.1.0

Unofficial Dart client for Pinecone vector database

pinecone

pinecone is an unofficial Dart client for your managed Pinecone vector database instance.

Is this package actively maintained?

Yes! This package is used in production applications and is actively maintained!

API Reference

For detailed API information, please see the official Pinecone API reference. This package is a thin wrapper around the official API and aims to have full parity with the official API.

Client Instance

To create a client instance you will need your API key and the environment name of your Pinecone project. You can find your API key and project environment in the Pinecone console.

final client = PineconeClient(
  apiKey: '123-456-789',
  environment: 'us-west1-gcp-free',
);

Once the client is created, you can access the index and vector operations APIs respectively:

// Access the index operations API
client.index;

// Access the vector operations API
client.vector(host:host);

Index Operations

List Indexes

Official Documentation: list_indexes

List<String> names = await client.index.listIndexes();

Create Index

Official Documentation: create_index

await client.index.createIndex(
  body: IndexDefinition(
    name: 'my-index',
    dimension: 1536,
  ),
);

Describe Index

Official Documentation: describe_index

final Index index = await client.index.describeIndex(
  indexName: 'my-index',
);

Delete Index

Official Documentation: delete_index

await client.index.deleteIndex(
  indexName: 'my-index',
);

Configure Index

Official Documentation: configure_index

await client.index.configureIndex(
  body: IndexConfiguration(
    podType: 's1.x2',
    replicas: 3,
  ),
);

List Collections

Official Documentation: list_collections

List<String> names = await client.index.listCollections();

Create Collection

Official Documentation: create_collection

await client.index.createCollection(
  body: CollectionDefinition(
    name: 'my-collection',
    source: 'my-index'
  ),
);

Describe Collection

Official Documentation: describe_collection

final Collection collection = await client.index.describeCollection(
  collectionName: 'my-collection'
);

Delete Collection

Official Documentation: delete_collection

await client.index.deleteCollection(
  collectionName: 'my-collection'
);

Vector Operations

In order to use the vector operations API, you must provide the host name of your specified index. The host name will look something like the following:

my-index-e345e78.svc.us-west1-gcp-free.pinecone.io

To retrieve the host name, you can use the describeIndex method:

final Index index = await client.index.describeIndex(
  indexName: 'my-index',
);

// The host name is the `status.host` property
final String host = index.status.host;

Describe Index Stats

Official Documentation: describe_index_stats_post

final DescribeIndexStatsResponse res =
    await client.vector(host: host).describeIndexStats(
          body: DescribeIndexStatsRequest(),
        );

Query

Official Documentation: query

final QueryResponse res = await client.vector(host: host).query(
      body: QueryRequest(
        topK: 5,
        namespace: 'my-namespace',
        includeValues: false,
        includeMetadata: false,
        vector: [1, 2, 3],
      ),
    );

Delete

Official Documentation: delete

final DeleteResponse res = await client.vector(host: host).delete(
      body: DeleteRequest(
        ids: ['id-1', 'id-2', 'id-3'],
        namespace: 'my-namespace',
      ),
    );

Fetch

Official Documentation: fetch

final FetchResponse res = await client.vector(host: host).fetch(
      body: FetchRequest(
        ids: [],
        namespace: 'my-namespace',
      ),
    );

Update

Official Documentation: update

final UpdateResponse res = await client.vector(host: host).update(
      body: UpdateRequest(
        id: 'id-1',
        namespace: 'my-namespace',
        values: [1, 2, 3],
      ),
    );

Upsert

Official Documentation: upsert

final UpsertResponse res = await client.vector(host: host).upsert(
      body: UpsertRequest(
        vectors: [
          UpsertVector(
            id: 'id-1',
            values: [1, 2, 3],
          ),
          UpsertVector(
            id: 'id-2',
            values: [4, 5, 6],
          ),
        ],
        namespace: 'my-namespace',
      ),
    );

Contributing

Please see the pinecone Github repository.