# Full-text Fuzzy Search with MongoDB and Typesense

This walk-through will show you how to ingest data from MongoDB into Typesense, and then use Typesense to search through the data with typo-tolerance, filtering, faceting, etc.

At a high-level we'll be setting up a trigger using MongoDB's Change Streams and pushing the data into Typesense on each change event.

Typesense MongoDB Integration Chart

UPDATE

We've published a Node.js CLI that you can install to automatically sync MongoDB documents into Typesense.

Here's how to set it up: typesense/typesense-mongodb (opens new window).

# Step 1: Install and Run Typesense

To install and start Typesense using docker run the following Docker command:

Now, we can check if our Typesense server is ready to accept requests.

curl http://localhost:8108/health
{"ok":true}

You can also run Typesense in other ways. Check out Typesense Installation and Typesense Cloud (opens new window) for more details.

# Step 2: Start a MongoDB Replica Set

MongoDB Replica Sets provide redundancy and high availability, and are the basis for all production deployments.

If you have a standalone MongoDB instance, you can convert it to a replica set by following steps:

Shutdown already running MongoDB server.
Start the MongoDB server by specifying -- replSet option

mongod --port "PORT" --dbpath "YOUR_DB_DATA_PATH" --replSet "REPLICA_SET_INSTANCE_NAME"

Check the status of replica set issuing the command rs.status() in mongo shell.

# Step 3: Open a Change Stream

Now let's open a change stream to listen for any changes to data in our MongoDB cluster. We'll later push these changes to Typesense.

We can open a change stream for MongoDB Replica Set from any of the data-bearing members. For detailed explanation check out MongoDB Change Streams (opens new window)

Here's an example:

const uri = '<MongoDB-URI>'
const mongodbOptions = {
  useNewUrlParser: true,
  useUnifiedTopology: true,
}
const client = new MongoClient(uri, mongodbOptions)
await client.connect()
const collection = client.db('sample').collection('books')
const changeStream = collection.watch()
changeStream.on('change', next => {
  // process next document
})

# Step 4: Create a Typesense Collection

To use Typesense, we first need to create a client. Typesense supports multiple API clients including Javascript, Python, Ruby, PHP etc.

To initialize the Javascript client, you need the API key of the Typesense server:

import Typesense from 'typesense'

let typesense = new Typesense.Client({
  nodes: [
    {
      host: 'localhost',
      port: '8108',
      protocol: 'http',
    },
  ],
  apiKey: '<API_KEY>',
  connectionTimeoutSeconds: 2,
})

Next, we will create a collection. A collection needs a schema, that represents how a document would look like.

let schema = {
  name: 'books',
  fields: [
    { name: 'id', type: 'string', facet: false },
    { name: 'name', type: 'string', facet: false },
    { name: 'author', type: 'string', facet: false },
    { name: 'year', type: 'int32', facet: true },
  ],
  default_sorting_field: 'year',
}
await typesense.collections().create(schema)

# Step 5: Index documents to Typesense

Next, we'll create a function to listen to change streams from MongoDB and write the changes to Typesense.

Here's an example MongoDB change streams response:

{
  _id: {
    _data: '826062978E000000012B022C0100296E5'
  },
  operationType: 'insert',
    clusterTime: Timestamp { _bsontype: 'Timestamp', low_: 1, high_: 1617074062 },
  fullDocument: {
    _id: 6062978e06e4444ef0c7f16a,
      name: 'Davinci Code',
      author: 'Dan Brown',
      year: 2003
  },
  ns: { db: 'sample', coll: 'books' },
  documentKey: { _id: 6062978e06e4444ef0c7f16a }
}
{
  _id: {
    _data: '826062978E000000032B022C0100296E5'
  },
  operationType: 'update',
    clusterTime: Timestamp { _bsontype: 'Timestamp', low_: 3, high_: 1617074062 },
  ns: { db: 'sample', coll: 'books' },
  documentKey: { _id: 6062978e06e4444ef0c7f16a },
  updateDescription: { updatedFields: { year: 2000 }, removedFields: [] }
}
{
  _id: {
    _data: '826062978E000000072B022C0100296E5'
  },
  operationType: 'delete',
    clusterTime: Timestamp { _bsontype: 'Timestamp', low_: 7, high_: 1617074062 },
  ns: { db: 'sample', coll: 'books' },
  documentKey: { _id: 6062978e06e4444ef0c7f16c }
}

async function index(next, typesense) {
  if (next.operationType == 'delete') {
    await typesense.collections('books').documents(next.documentKey._id).delete()
  } else if (next.operationType == 'update') {
    let data = JSON.stringify(next.updateDescription.updatedFields)
    await typesense.collections('books').documents(next.documentKey._id).update(data)
  } else {
    next.fullDocument.id = next.fullDocument['_id']
    delete next.fullDocument._id
    let data = JSON.stringify(next.fullDocument)
    await typesense.collections('books').documents().upsert(data)
  }
}

# Full Example

Here is the full code example:

const { MongoClient } = require('mongodb')
const Typesense = require('typesense')

async function listDatabases(client) {
  databasesList = await client.db().admin().listDatabases()

  console.log('Databases:')
  databasesList.databases.forEach(db => console.log(` - ${db.name}`))
}

function closeChangeStream(timeInMs = 60000, changeStream) {
  return new Promise(resolve => {
    setTimeout(() => {
      console.log('Closing the change stream')
      changeStream.close()
      resolve()
    }, timeInMs)
  })
}

async function index(next, typesense) {
  console.log(next)
  if (next.operationType == 'delete') {
    await typesense.collections('books').documents(next.documentKey._id).delete()
    console.log(next.documentKey._id)
  } else if (next.operationType == 'update') {
    let data = JSON.stringify(next.updateDescription.updatedFields)
    await typesense.collections('books').documents(next.documentKey._id).update(data)
    console.log(data)
  } else {
    next.fullDocument.id = next.fullDocument['_id']
    delete next.fullDocument._id
    let data = JSON.stringify(next.fullDocument)
    await typesense.collections('books').documents().upsert(data)
    console.log(data)
  }
}

async function monitorListingsUsingEventEmitter(client, typesense, timeInMs = 60000) {
  const collection = client.db('sample').collection('books')
  const changeStream = collection.watch()
  changeStream.on('change', next => {
    index(next, typesense)
  })
  await closeChangeStream(timeInMs, changeStream)
}

async function createSchema(schema, typesense) {
  const collectionsList = await typesense.collections().retrieve()
  var toCreate = collectionsList.find((value, index, array) => {
    return value['name'] == schema['name']
  })

  if (!toCreate) {
    await typesense.collections().create(schema)
  }
}

async function main() {
  const typesense = new Typesense.Client({
    nodes: [
      {
        host: 'localhost',
        port: '8108',
        protocol: 'http',
      },
    ],
    apiKey: '<API_KEY>',
    connectionTimeoutSeconds: 2,
  })
  let schema = {
    name: 'books',
    fields: [
      {
        name: 'id',
        type: 'string',
        facet: false,
      },
      {
        name: 'name',
        type: 'string',
        facet: false,
      },
      {
        name: 'author',
        type: 'string',
        facet: false,
      },
      {
        name: 'year',
        type: 'int32',
        facet: true,
      },
    ],
    default_sorting_field: 'year',
  }
  createSchema(schema, typesense)
  const mongodbOptions = {
    useNewUrlParser: true,
    useUnifiedTopology: true,
  }
  const uri = '<Mongo-URI>'
  const client = new MongoClient(uri, mongodbOptions)
  try {
    await client.connect()
    await listDatabases(client)
    await monitorListingsUsingEventEmitter(client, typesense)
  } catch (e) {
    console.error(e)
  } finally {
    await client.close()
  }
}

main().catch(console.error)

That's it 😊! Now you can easily search through your MongoDB documents using Typsense. You can even use Typesense Cloud (opens new window) and MongoDB Atlas (opens new window) for hosted versions of Typesense and MongoDB.

# References

← AWS DynamoDB Full Text Search Supabase Full Text Search →