Atlas Search Workshop

Session slides

Steps

1. Add sample data to your MongoDB Atlas cluster

If you haven't already sone so, create an Atlas cluster.

Make sure at least 1 user with read-write access has been added.

Make sure that IP Access List allows access (add Allow ACCESS FROM ANYWHERE(0.0.0.0\0) to be sure).

Add sample data:

• Click the ... next to your cluster in the Atlas UI
• Select Load Sample Dataset

It will take a few minutes for the sample data to be added.

2. Add default Search index

• Keep the name as default
• Databse: sample_mflix
• Collection: movies
• Leave Dynamic Mapping turned on

3. Add autocomplete index

• Set the name to autocomplete
• Turn off Dynamic Mapping
• Click Refine Your Index
• Add a new field mapping for the title field with a Data Type of Autocomplete

4. (OPTIONAL) Connect MongoDB Compass

You can get the URL from the Atlas UI - click the Connect button next to your cluster.

Paste the URL into the Compass connection box, replacing the username and password.

5. Create aggregation pipeline to search for movies

From Compass (or the Atlas UI), navigate to the Aggregation pane for sample_mflix.movies.

Create these stages in sequence:

a. $search

{
  index: 'default',
  text: {
    query: 'Indiana Ones',
    path: ['plot', 'fullplot', 'title'],
    fuzzy: {}
  },
  highlight: {path: 'fullplot'}
}

b. $limit

12

c. $project

{
  title: 1,
  year: 1,
  poster: 1,
  plot: 1, 
  imdb: 1,
  score: {$meta: 'searchScore'},
  highlights: {
    '$meta': 'searchHighlights'
  }
}

• Click EXPORT TO LANGUAGE
• Choose NODE as the export language
• Copy the code and keep it safe
• Save the pipeline if using Compass

6. Create and test a HTTPS endpoint to search for movies

• Create a new Atlas App Service and name it Movies.
• Select Functions and then Create New Function.
• Set Authentication to System, and then Save Draft.
• Use this template for the function code:

exports = async function({ query, headers, body}, response) {
    
    // GET A HANDLE TO THE MOVIES COLLECTION
    const moviesCollection = context.services.get("mongodb-atlas").db("sample_mflix").collection("movies");
    
    // GET SEARCHTERM FROM QUERY PARAMETER. IF NONE, RETURN EMPTY ARRAY
    let searchTerm = query.searchTerm;
    if (!query.searchTerm || searchTerm === ""){
         return [];
    }
  
    const searchAggregation = [];  
    return await moviesCollection.aggregate(searchAggregation).toArray();
};

• Replace the [] for searchAggregation with your aggregation pipeline
• Replace Indiana Ones with searchTerm
• Save Draft
• Select HTTPS Endpoints
• Add An Endpoint
• Route: `/movies'
• HTTP Method: GET
• Enable Respond With Result
• Function: the function you just created
• Save Draft
• REVIEW DRAFT & DEPLOY
• Copy the URL -> <baseURL>
• From a browser, paste the URL into the location bar (<baseURL>?searchTerm=blues). This will search for movies that mention blues - replace it with your own string (using %20 for spaces - e.g., blues%20brothers)

7. Create aggregation pipeline to autocomplete movie titles

From Compass (or the Atlas UI), navigate to the Aggregation pane for sample_mflix.movies.

Create these stages in sequence:

a. $search

{
  index: 'autocomplete',
  autocomplete: {
    query: 'Indiana J',
    path: 'title'
  }
}

b. $limit

10

c. $project

{
  _id: 0,
  title: 1
}

8. Create and test a HTTPS endpoint to autocomplete movie titles

• Open your Movies Atlas App Service.
• Select Functions and then Create New Function.
• Set Authentication to System, and then Save Draft.
• Use this template for the function code:

exports = async function({ query, headers, body}, response) {
  
  // GET A HANDLE TO THE MOVIES COLLECTION
  const moviesCollection = context.services.get("mongodb-atlas").db("sample_mflix").collection("movies");
  
  // GET SEARCHTERM FROM QUERY PARAMETER. IF NONE, RETURN EMPTY ARRAY
  let searchTerm = query.searchTerm;

  if (!query.searchTerm || searchTerm ===""){
    return [];
  }
  
  const searchAggregation =[];

  return await moviesCollection.aggregate(searchAggregation).toArray();
};

• Replace the [] for searchAggregation with your aggregation pipeline
• Replace Indiana Ones with searchTerm
• Save Draft
• Select HTTPS Endpoints
• Add An Endpoint
• Route: `/titles'
• HTTP Method: GET
• Enable Respond With Result
• Function: the function you just created
• Save Draft
• REVIEW DRAFT & DEPLOY
• Copy the URL -> <baseURL>
• From a browser, paste the URL into the location bar (<baseURL>?searchTerm=blues). This will search for movies that mention blues - replace it with your own string (using %20 for spaces - e.g., blues%20brothers)

OPTIONAL: Use your new API from a REACT web app

This repo includes a REACT movie-search web app which is able to use the two HTTP endpoints you've just created.

To download and start the app (if you're able to install Node.js on your machine):

git clone git@github.com:ClusterDB/AtlasSearchWorkshop.git
cd AtlasSearchWorkshop
npm install
npm start

To download and start the app (if you're not able to install Node.js on your machine):

Create a CodeSpace (you must be logged in to GitHub):

You should then wait for the installation to complete in the terminal window. Once complete, start the app by running npm start from the CodeSpace terminal.

The web app should open in a browser window. If you try searching for anything, you should get a message that you need to set up the HTTPS endpoints.

To make the app work, you need to add your endpoints to src/components/Home.js (lines 23 & 24):

// INSERT YOUR CREATED MOVIE ENDPOINTS
const [moviesEndpoint, setMoviesEndpoint] = useState('Your endpoint to fetch movies goes here');
const [titlesEndpoint, setTitlesEndpoint] = useState('Your endpoint to fetch titles goes here');

When you save the files, the app should rebuild - you may need to refresh the browser window.

OPTIONAL: Test your backend Atlas app using a shared frontend

If you're not able to build the frontend Node.js app locally or in CodeSpace, you can still test out your Atlas backend by pasting your endpoints into this shared frontend app: https://application-0-jdvvr.mongodbstitch.com/.

Final demo

A quick demo of some of the features of Atlas Search (including the queries) using Restaurant Finder. Code can be found here.