Skip to content

pinecone-io/pinecone-ts-client

Repository files navigation

Pinecone Node.js Client · License npm npm GitHub Workflow Status (with event)

This is the official Node.js client for Pinecone, written in TypeScript.

Documentation

Example code

Many of the brief examples shown in this README are using very small vectors to keep this documentation concise, but most real world usage will involve much larger embedding vectors. To see some more realistic examples of how this client can be used, explore these examples:

Prerequisites

The Pinecone TypeScript client is compatible with TypeScript 4.1 and greater.

Installation

npm install @pinecone-database/pinecone

Usage

Initializing the client

There are two pieces of configuration required to use the Pinecone client: an API key and environment value. These values can be passed using environment variables or in code through a configuration object. Find your configuration values in the console dashboard at https://app.pinecone.io.

Using environment variables

The environment variables used to configure the client are the following:

PINECONE_API_KEY="your_api_key"
PINECONE_ENVIRONMENT="your_environment"

When these environment variables are set, the client constructor does not require any additional arguments.

import { Pinecone } from '@pinecone-database/pinecone';

const pinecone = new Pinecone();

Using a configuration object

If you prefer to pass configuration in code, the constructor accepts a config object containing the apiKey and environment values. This could be useful if your application needs to interact with multiple projects, each with a different configuration.

const pinecone = new Pinecone({
  apiKey: 'your_api_key',
  environment: 'your_environment',
});

Indexes

Create Index

Create an index with minimal configuration

At a minimum, to create an index you must specify a name and dimension. The dimension indicates the size of the records you intend to store in the index. For example, if your intention was to store and query embeddings generated with OpenAI's textembedding-ada-002 model, you would need to create an index with dimension 1536 to match the output of that model.

const pinecone = new Pinecone();
await pinecone.createIndex({
  name: 'sample-index',
  dimension: 1536,
});

Create an index with other optional configurations

Many optional configuration fields allow greater control over hardware resources and availability. To learn more about the purpose of these fields, see Understanding indexes and Scaling indexes.

await pinecone.createIndex({
  name: 'sample-index-2',
  dimension: 1536,
  metric: 'dotproduct',
  pods: 2,
  replicas: 2,
  shards: 2,
  podType: 'p1.x2',
  metadataConfig: {
    indexed: ['product_type'],
  },

  // This option tells the client not to throw if the index already exists.
  suppressConflicts: true,

  // This option tells the client not to resolve the promise until the
  // index is ready.
  waitUntilReady: true,
});

Checking the status of a newly created index

The createIndex method issues a create request to the API that returns quickly, but the resulting index is not immediately ready for upserting, querying, or performing other data operations. You can use the describeIndex method to find out the status of an index and see whether it is ready for use.

> await pinecone.describeIndex('sample-index')
{
  database: {
    name: 'sample-index',
    dimension: 1536,
    metric: 'cosine',
    pods: 1,
    replicas: 1,
    shards: 1,
    podType: 'p1.x1'
  },
  status: {
    ready: false,
    state: 'Initializing',
    host: 'sample-index-c01b9b5.svc.us-east1-gcp.pinecone.io',
    port: 433
  }
}
> await pinecone.describeIndex('sample-index')
{
  database: {
    name: 'sample-index',
    dimension: 1536,
    metric: 'cosine',
    pods: 1,
    replicas: 1,
    shards: 1,
    podType: 'p1.x1'
  },
  status: {
    ready: true,
    state: 'Ready',
    host: 'sample-index-c01b9b5.svc.us-east1-gcp.pinecone.io',
    port: 433
  }
}

Waiting until the index is ready

If you pass the waitUntilReady option, the client will handle polling for status updates on a newly created index. The promise returned by createIndex will not be resolved until the index status indicates it is ready to handle data operations. This can be especially useful for integration testing, where index creation in a setup step will be immediately followed by data operations.

const pinecone = new Pinecone();
await pinecone.createIndex({
  name: 'sample-index',
  dimension: 1536,
  waitUntilReady: true,
});

Create an index from a Pinecone collection

As you use Pinecone for more things, you may wish to explore different index configurations with the same vector data. Collections provide an easy way to do this. See other client methods for working with collections here.

Given that you have an existing collection:

> await pinecone.describeCollection('product-description-embeddings')
{
  name: 'product-description-embeddings',
  size: 543427063,
  status: 'Ready',
  dimension: 2,
  recordCount: 10001498
}

You can specify a sourceCollection along with other configuration in your createIndex options:

await pinecone.createIndex({
  name: 'product-description-p1x1',
  sourceCollection: 'product-description-embeddings',
  dimension: 256,
  metric: 'cosine'
  podType: 'p1.x1'
})

When the new index is ready, it should contain all the data that was in the collection, ready to be queried.

> await pinecone.index('product-description-p2x2').describeIndexStats()
{
  namespaces: { '': { recordCount: 78000 } },
  dimension: 256,
  indexFullness: 0.9,
  totalRecordCount: 78000
}

Describe Index

You can fetch the description of any index by name using describeIndex.

> await pinecone.describeIndex('sample-index')
{
  database: {
    name: 'sample-index',
    dimension: 1536,
    metric: 'cosine',
    pods: 1,
    replicas: 1,
    shards: 1,
    podType: 'p1.x1'
  },
  status: {
    ready: true,
    state: 'Ready',
    host: 'sample-index-c01b9b5.svc.us-east1-gcp.pinecone.io',
    port: 433
  }
}

Configure Index

You can adjust the number of replicas or scale to a larger pod size (specified with podType). See Pod types and sizes. You cannot downgrade pod size or change the base pod type.

> await pinecone.configureIndex('my-index', { replicas: 3 })
> const config = await pinecone.describeIndex('my-index')
{
  database: {
    name: 'my-index',
    dimension: 2,
    metric: 'cosine',
    pods: 2,
    replicas: 2,
    shards: 1,
    podType: 'p1.x1'
  },
  status: {
    ready: true,
    state: 'ScalingUpPodSize',
    host: 'my-index-c01b9b5.svc.us-east1-gcp.pinecone.io',
    port: 433
  }
}

Delete Index

Indexes are deleted by name.

await pinecone.deleteIndex('sample-index');

List Indexes

The listIndexes command returns an array of index names.

> await pinecone.listIndexes()
[{ name: 'sample-index' }, { name: 'sample-index-2' }]

Collections

A collection is a static copy of an index that may be used to create backups, to create copies of indexes, or to perform experiments with different index configurations. To learn more about Pinecone collections, see Understanding collections.

Create Collection

await pinecone.createCollection({
  name: 'collection-name',
  source: 'index-name',
});

This API call should return quickly, but the creation of a collection can take from minutes to hours depending on the size of the source index and the index's configuration. Use describeCollection to check the status of a collection.

Delete Collection

await pinecone.deleteCollection('collection-name');

You can use listIndexes to confirm the deletion.

Describe Collection

> const describeCollection = await pinecone.describeCollection('collection3');
{
  name: 'collection3',
  size: 3126700,
  status: 'Ready',
  dimension: 3,
  recordCount: 99
}

List Collections

> const list = await pinecone.listCollections();
[{ name: "collection1" }, { name: "collection2" }]

Index operations

Pinecone indexes support operations for working with vector data using operations such as upsert, query, fetch, and delete.

Targeting an index

To perform data operations on an index, you target it using the index method.

const pinecone = new Pinecone();
const index = pinecone.index('test-index');

// Now perform index operations
await index.fetch(['1']);

Targeting an index, with metadata typing

If you are storing metadata alongside your vector values, you can pass a type parameter to index() in order to get proper TypeScript typechecking.

const pinecone = new Pinecone();

type MovieMetadata = {
  title: string,
  runtime: numbers,
  genre: 'comedy' | 'horror' | 'drama' | 'action'
}

// Specify a custom metadata type while targeting the index
const index = pinecone.index<MovieMetadata>('test-index');

// Now you get type errors if upserting malformed metadata
await index.upsert([{
  id: '1234',
  values: [
    .... // embedding values
  ],
  metadata: {
    genre: 'Gone with the Wind',
    runtime: 238,
    genre: 'drama',
    // @ts-expect-error because category property not in MovieMetadata
    category: 'classic'
  }
}])

const results = await index.query({
  vector: [
    ... // query embedding
  ],
  filter: { genre: { '$eq': 'drama' }}
})
const movie = results.matches[0];

if (movie.metadata) {
  // Since we passed the MovieMetadata type parameter above,
  // we can interact with metadata fields without having to
  // do any typecasting.
  const { title, runtime, genre } = movie.metadata;
  console.log(`The best match in drama was ${title}`)
}

Targeting a namespace

Note

Indexes in the gcp-starter environment do not support namespaces.

By default, all data operations take place inside the default namespace of ''. If you are working with other non-default namespaces, you can target the namespace by chaining a call to namespace().

const pinecone = new Pinecone();
const index = pinecone.index('test-index').namespace('ns1');

// Now perform index operations in the targeted index and namespace
await index.fetch(['1']);

If needed, you can check the currently targeted index and namespace by inspecting the target property of an index object.

const pinecone = new Pinecone();
const index = pinecone.index('test-index').namespace('ns1');

console.log(index.target); // { index: 'test-index', namespace: 'ns1' }

See Using namespaces for more information.

Upsert records

Pinecone expects records inserted into indexes to have the following form:

type PineconeRecord = {
  id: string;
  values: Array<number>;
  sparseValues?: Array<number>;
  metadata?: object;
};

To upsert some records, you can use the client like so:

const pinecone = new Pinecone();

// Target an index
const index = pinecone.index('sample-index');

// Prepare your data. The length of each array
// of vector values must match the dimension of
// the index where you plan to store them.
const records = [
  {
    id: '1',
    values: [0.236, 0.971, 0.559],
  },
  {
    id: '2',
    values: [0.685, 0.111, 0.857],
  },
];

// Upsert the data into your index
await index.upsert(records);

Seeing index statistics

When experimenting with data operations, it's sometimes helpful to know how many records are stored in each namespace. In that case, target the index and use the describeIndexStats() command.

> await pinecone.index('example-index').describeIndexStats()
{
  namespaces: {
    '': { recordCount: 10 }
    foo: { recordCount: 2000 },
    bar: { recordCount: 2000 }
  },
  dimension: 1536,
  indexFullness: 0,
  totalRecordCount: 4010
}

Querying

Querying with vector values

The query method accepts a large number of options. The dimension of the query vector must match the dimension of your index.

type QueryOptions = {
  topK: number; // number of results desired
  vector?: Array<number>; // must match dimension of index
  sparseVector?: {
    indices: Array<integer>; // indices must fall within index dimension
    values: Array<number>; // indices and values arrays must have same length
  };
  id?: string;
  includeMetadata: boolean;
  includeValues: boolean;
};

For example, to query by vector values you would pass the vector param in the options configuration. For brevity sake this example query vector is tiny (dimension 2), but in a more realistic use case this query vector would be an embedding outputted by a model. Look at the Example code to see more realistic examples of how to use query.

> await pinecone.index('my-index').query({ topK: 3, vector: [ 0.22, 0.66 ]})
{
  matches: [
    {
      id: '556',
      score: 1.00000012,
      values: [],
      sparseValues: undefined,
      metadata: undefined
    },
    {
      id: '137',
      score: 1.00000012,
      values: [],
      sparseValues: undefined,
      metadata: undefined
    },
    {
      id: '129',
      score: 1.00000012,
      values: [],
      sparseValues: undefined,
      metadata: undefined
    }
  ],
  namespace: ''
}

You include options to includeMetadata: true or includeValues: true if you need this information. By default these are not returned to keep the response payload small.

Remember that data operations take place within the context of a namespace, so if you are working with namespaces and do not see expected results you should check that you are targeting the correct namespace with your query.

const results = await pinecone
  .index('my-index')
  .namespace('my-namespace')
  .query({ topK: 3, vector: [0.22, 0.66] });

Querying by record id

You can query using the vector values of an existing record in the index by passing a record id.

const results = await pinecone.index('my-index').query({ topK: 10, id: '1' });

Hybrid search with sparseVector

If you are working with sparse-dense vectors, you can add sparse vector values to perform a hybrid search.

const pinecone = new Pinecone()

await pinecone.createIndex({
  name: 'hyrbid-image-search',
  metric: 'dotproduct',
  dimension: 512,
  podType: 's1',
  waitUntilReady: true
});
const index = pinecone.index('hybrid-image-search');

// Create some vector embeddings using your model of choice.
const records = [...]

// Upsert data
await index.upsert(records)

// Prepare query values. In a more realistic example, these would both come out of a model.
const vector = [
  // The dimension of this index needs to match the index dimension.
  // Pretend this is a 512 dimension vector.
]
const sparseVector = {
  indices: [23, 399, 251, 17],
  values: [ 0.221, 0.967, 0.016, 0.572]
}

// Execute the query
const results = await index.query({ topK: 10, vector, sparseVector, includeMetadata: true })

Update a record

You may want to update vector values, sparseValues, or metadata. Specify the id and the attribute value you want to update.

await pinecone.index('imdb-movies').update({
  id: '18593',
  metadata: { genre: 'romance' },
});

Fetch records by their IDs

const fetchResult = await index.fetch(['id-1', 'id-2']);

Delete records

For convenience there are several delete-related methods. You can verify the results of a delete operation by trying to fetch() a record or looking at the index summary with describeIndexStats()

Delete one

const index = pinecone.index('my-index');
await index.deleteOne('id-to-delete');

Delete many by id

const index = pinecone.index('my-index');
await index.deleteMany(['id-1', 'id-2', 'id-3']);

Delete many by metadata filter

await client.index('albums-database').deleteMany({ genre: 'rock' });

Delete all records in a namespace

Note

Indexes in the gcp-starter environment do not support namespaces.

To nuke everything in the targeted namespace, use the deleteAll method.

const index = pinecone.index('my-index');
await index.namespace('foo-namespace').deleteAll();

If you do not specify a namespace, the records in the default namespace '' will be deleted.

Productionizing

If you are ready to take a JavaScript application to production where raw performance is the overriding concern, you can set the environment variable PINECONE_DISABLE_RUNTIME_VALIDATIONS="true" to disable runtime argument validation in the Pinecone client. Runtime validations are used to provide feedback when incorrect method options are provided, for example if you attempt to create an index without specifying a required dimension property.

These runtime validations are most helpful for users who are not developing with Typescript or who are experimenting in a REPL or notebook-type setting. But once you've tested an application and have gained confidence things are working as expected, you can disable these checks to gain a small improvement in performance. This will have the most impact if your workload is upserting very large amounts of data.

Legacy exports

For information about the legacy PineconeClient export, see the old README.