Build a Simple REST API with Node and OAuth 2.0

JavaScript is used everywhere on the web - nearly every web page will include at least some JavaScript, and even if it doesn’t, your browser probably has some sort of extension that injects bits of JavaScript code on to the page anyway. It’s hard to avoid in 2018.

JavaScript can also be used outside the context of a browser, for anything from hosting a web server to controlling an RC car or running a full-fledged operating system. Sometimes you want a couple of servers to talk to each other, whether on a local network or over the internet.

Today, I’ll show you how to create a REST API using Node.js, and secure it with OAuth 2.0 to prevent unwarranted requests. REST APIs are all over the web, but without the proper tools require a ton of boilerplate code. I’ll show you how to use a couple of amazing tools that make it all a breeze, including Okta to implement the Client Credentials Flow, which securely connects two machines together without the context of a user.

Build Your Node Server

Setting up a web server in Node is quite simple using the Express JavaScript library. Make a new folder that will contain your server.

$ mkdir rest-api

Node uses a package.json to manage dependencies and define your project. To create one, use npm init, which will ask you some questions to help you initialize the project. For now, you can use standard JS to enforce a coding standard, and use that as the tests.

$ cd rest-api

$ npm init
This utility will walk you through creating a package.json file.
It only covers the most common items, and tries to guess sensible defaults.

See `npm help json` for definitive documentation on these fields
and exactly what they do.

Use `npm install <pkg>` afterwards to install a package and
save it as a dependency in the package.json file.

Press ^C at any time to quit.
package name: (rest-api)
version: (1.0.0)
description: A parts catalog
entry point: (index.js)
test command: standard
git repository:
keywords:
author:
license: (ISC)
About to write to /Users/Braden/code/rest-api/package.json:

{
  "name": "rest-api",
  "version": "1.0.0",
  "description": "A parts catalog",
  "main": "index.js",
  "scripts": {
    "test": "standard"
  },
  "author": "",
  "license": "ISC"
}

Is this OK? (yes)

The default entry point is index.js, so you should create a new file by that name. The following code will get you a really basic server that doesn’t really do anything but listens on port 3000 by default.

index.js

const express = require('express')
const bodyParser = require('body-parser')
const { promisify } = require('util')

const app = express()
app.use(bodyParser.json())

const startServer = async () => {
  const port = process.env.SERVER_PORT || 3000
  await promisify(app.listen).bind(app)(port)
  console.log(`Listening on port ${port}`)
}

startServer()

The promisify function of util lets you take a function that expects a callback and instead will return a Promise, which is the new standard as far as handling asynchronous code. This also lets us use the relatively new async/await syntax and make our code look much prettier.

In order for this to work, you need to install the dependencies that you require at the top of the file. Add them using npm install. This will automatically save some metadata to your package.json file and install them locally in a node_modules folder.

Note : You should never commit node_modules to source control because it tends to become bloated quickly, and the package-lock.json file will keep track of the exact versions you used to that if you install this on another machine they get the same code.

$ npm install express@4.16.3 util@0.11.0

For some quick linting, install standard as a dev dependency, then run it to make sure your code is up to par.

$ npm install --save-dev standard@11.0.1
$ npm test

> rest-api@1.0.0 test /Users/bmk/code/okta/apps/rest-api
> standard

If all is well, you shouldn’t see any output past the > standard line. If there’s an error, it might look like this:

$ npm test

> rest-api@1.0.0 test /Users/bmk/code/okta/apps/rest-api
> standard

standard: Use JavaScript Standard Style (https://standardjs.com)
standard: Run `standard --fix` to automatically fix some problems.
  /Users/Braden/code/rest-api/index.js:3:7: Expected consistent spacing
  /Users/Braden/code/rest-api/index.js:3:18: Unexpected trailing comma.
  /Users/Braden/code/rest-api/index.js:3:18: A space is required after ','.
  /Users/Braden/code/rest-api/index.js:3:38: Extra semicolon.
npm ERR! Test failed. See above for more details.

Now that your code is ready and you have installed your dependencies, you can run your server with node . (the . says to look at the current directory, and then checks your package.json file to see that the main file to use in this directory is index.js):

$ node .

Listening on port 3000

To test that it’s working, you can use the curl command. There are no endpoints yet, so express will return an error:

$ curl localhost:3000 -i
HTTP/1.1 404 Not Found
X-Powered-By: Express
Content-Security-Policy: default-src 'self'
X-Content-Type-Options: nosniff
Content-Type: text/html; charset=utf-8
Content-Length: 139
Date: Thu, 16 Aug 2018 01:34:53 GMT
Connection: keep-alive

<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<title>Error</title>
</head>
<body>
<pre>Cannot GET /</pre>
</body>
</html>

Even though it says it’s an error, that’s good. You haven’t set up any endpoints yet, so the only thing for Express to return is a 404 error. If your server wasn’t running at all, you’d get an error like this:

$ curl localhost:3000 -i
curl: (7) Failed to connect to localhost port 3000: Connection refused

Build Your REST API with Express, Sequelize, and Epilogue

Now that you have a working Express server, you can add a REST API. This is actually much simpler than you might think. The easiest way I’ve seen is by using Sequelize to define your database schema, and Epilogue to create some REST API endpoints with near-zero boilerplate.

You’ll need to add those dependencies to your project. Sequelize also needs to know how to communicate with the database. For now, use SQLite as it will get us up and running quickly.

npm install sequelize@4.38.0 epilogue@0.7.1 sqlite3@4.0.2

Create a new file database.js with the following code. I’ll explain each part in more detail below.

database.js

const Sequelize = require('sequelize')
const epilogue = require('epilogue')

const database = new Sequelize({
  dialect: 'sqlite',
  storage: './test.sqlite',
  operatorsAliases: false
})

const Part = database.define('parts', {
  partNumber: Sequelize.STRING,
  modelNumber: Sequelize.STRING,
  name: Sequelize.STRING,
  description: Sequelize.TEXT
})

const initializeDatabase = async (app) => {
  epilogue.initialize({ app, sequelize: database })

  epilogue.resource({
    model: Part,
    endpoints: ['/parts', '/parts/:id']
  })

  await database.sync()
}

module.exports = initializeDatabase

Now you just need to import that file into your main app and run the initialization function. Make the following additions to your index.js file.

index.js

@@ -2,10 +2,14 @@ const express = require('express')
 const bodyParser = require('body-parser')
 const { promisify } = require('util')

+const initializeDatabase = require('./database')
+
 const app = express()
 app.use(bodyParser.json())

 const startServer = async () => {
+ await initializeDatabase(app)
+
   const port = process.env.SERVER_PORT || 3000
   await promisify(app.listen).bind(app)(port)
   console.log(`Listening on port ${port}`)

You can now test for syntax errors and run the app if everything seems good:

$ npm test && node .

> rest-api@1.0.0 test /Users/bmk/code/okta/apps/rest-api
> standard

Executing (default): CREATE TABLE IF NOT EXISTS `parts` (`id` INTEGER PRIMARY KEY AUTOINCREMENT, `partNumber` VARCHAR(255), `modelNu
mber` VARCHAR(255), `name` VARCHAR(255), `description` TEXT, `createdAt` DATETIME NOT NULL, `updatedAt` DATETIME NOT NULL);
Executing (default): PRAGMA INDEX_LIST(`parts`)
Listening on port 3000

In another terminal, you can test that this is actually working (to format the JSON response I use a json CLI, installed globally using npm install --global json):

$ curl localhost:3000/parts
[]

$ curl localhost:3000/parts -X POST -d '{
  "partNumber": "abc-123",
  "modelNumber": "xyz-789",
  "name": "Alphabet Soup",
  "description": "Soup with letters and numbers in it"
}' -H 'content-type: application/json' -s0 | json
{
  "id": 1,
  "partNumber": "abc-123",
  "modelNumber": "xyz-789",
  "name": "Alphabet Soup",
  "description": "Soup with letters and numbers in it",
  "updatedAt": "2018-08-16T02:22:09.446Z",
  "createdAt": "2018-08-16T02:22:09.446Z"
}

$ curl localhost:3000/parts -s0 | json
[
  {
    "id": 1,
    "partNumber": "abc-123",
    "modelNumber": "xyz-789",
    "name": "Alphabet Soup",
    "description": "Soup with letters and numbers in it",
    "createdAt": "2018-08-16T02:22:09.446Z",
    "updatedAt": "2018-08-16T02:22:09.446Z"
  }
]

What’s Going On Here?

Feel free to skip this section if you followed along with all that, but I did promise an explanation.

The Sequelize function creates a database. This is where you configure details, such as what dialect of SQL to use. For now, use SQLite to get up and running quickly.

const database = new Sequelize({
  dialect: 'sqlite',
  storage: './test.sqlite',
  operatorsAliases: false
})

Once you’ve created the database, you can define the schema for it using database.define for each table. Create a table called parts with a few useful fields to keep track of parts. By default, Sequelize also automatically creates and updates id, createdAt, and updatedAt fields when you create or update a row.

const Part = database.define('parts', {
  partNumber: Sequelize.STRING,
  modelNumber: Sequelize.STRING,
  name: Sequelize.STRING,
  description: Sequelize.TEXT
})

Epilogue requires access to your Express app in order to add endpoints. However, app is defined in another file. One way to deal with this is to export a function that takes the app and does something with it. In the other file when we import this script, you would run it like initializeDatabase(app).

Epilogue needs to initialize with both the app and the database. You then define which REST endpoints you would like to use. The resource function will include endpoints for the GET, POST, PUT, and DELETE verbs, mostly automagically.

To actually create the database, you need to run database.sync(), which returns a Promise. You’ll want to wait until it’s finished before starting your server.

The module.exports command says that the initializeDatabase function can be imported from another file.

const initializeDatabase = async (app) => {
  epilogue.initialize({ app, sequelize: database })

  epilogue.resource({
    model: Part,
    endpoints: ['/parts', '/parts/:id']
  })

  await database.sync()
}

module.exports = initializeDatabase

Secure Your Node + Express REST API with OAuth 2.0

Now that you have a REST API up and running, imagine you’d like a specific application to use this from a remote location. If you host this on the internet as is, then anybody can add, modify, or remove parts at their will.

To avoid this, you can use the OAuth 2.0 Client Credentials Flow. This is a way of letting two servers communicate with each other, without the context of a user. The two servers must agree ahead of time to use a third-party authorization server. Assume there are two servers, A and B, and an authorization server. Server A is hosting the REST API, and Server B would like to access the API.

• Server B sends a secret key to the authorization server to prove who they are and asks for a temporary token.
• Server B then consumes the REST API as usual but sends the token along with the request.
• Server A asks the authorization server for some metadata that can be used to verify tokens.
• Server A verifies the Server B’s request.
  • If it’s valid, a successful response is sent and Server B is happy.
  • If the token is invalid, an error message is sent instead, and no sensitive information is leaked.

Create an Authorization Server

This is where Okta comes into play. Okta can act as an authorization server to allow you to secure your data. You’re probably asking yourself “Why Okta? Well, it’s pretty cool to build a REST app, but it’s even cooler to build a secure one. To achieve that, you’ll want to add authentication so users have to log in before viewing/modifying groups. At Okta, our goal is to make identity management a lot easier, more secure, and more scalable than what you’re used to. Okta is a cloud service that allows developers to create, edit, and securely store user accounts and user account data, and connect them with one or multiple applications. Our API enables you to:

• Authenticate and authorize your users
• Store data about your users
• Perform password-based and social login
• Secure your application with multi-factor authentication
• And much more! Check out our product documentation

If you don’t already have one, sign up for a forever-free developer account, and let’s get started!

After creating your account, log in to your developer console, navigate to API , then to the Authorization Servers tab. Click on the link to your default server.

From this Settings tab, copy the Issuer field. You’ll need to save this somewhere that your Node app can read. In your project, create a file named .env that looks like this:

.env

ISSUER=https://{yourOktaDomain}/oauth2/default

The value for ISSUER should be the value from the Settings page’s Issuer URI field.

Note : As a general rule, you should not store this .env file in source control. This allows multiple projects to use the same source code without needing a separate fork. It also makes sure that your secure information is not public (especially if you’re publishing your code as open source).

Next, navigate to the Scopes tab. Click the Add Scope button and create a scope for your REST API. You’ll need to give it a name (e.g. parts_manager) and you can give it a description if you like.

You should add the scope name to your .env file as well so your code can access it.

.env

ISSUER=https://{yourOktaDomain}/oauth2/default
SCOPE=parts_manager

Now you need to create a client. Navigate to Applications , then click Add Application. Select Service , then click Next. Enter a name for your service, (e.g. Parts Manager), then click Done.

This will take you to a page that has your client credentials. These are the credentials that Server B (the one that will consume the REST API) will need in order to authenticate. For this example, the client and server code will be in the same repository, so go ahead and add this data to your .env file. Make sure to replace {yourClientId} and {yourClientSecret} with the values from this page.

CLIENT_ID={yourClientId}
CLIENT_SECRET={yourClientSecret}

Create Middleware to Verify Tokens in Express

In Express, you can add middleware that will run before each endpoint. You can then add metadata, set headers, log some information, or even cancel the request early and send an error message. In this case, you’ll want to create some middleware that verifies the token sent by the client. If the token is valid, it will continue to the REST API and return the appropriate response. If the token is invalid, it will instead respond with an error message so that only authorized machines have access.

To validate tokens, you can use Okta’s middleware. You’ll also need a tool called dotenv to load the environment variables:

npm install dotenv@6.0.0 @okta/jwt-verifier@0.0.12

Now create a file named auth.js that will export the middleware:

auth.js

const OktaJwtVerifier = require('@okta/jwt-verifier')

const oktaJwtVerifier = new OktaJwtVerifier({ issuer: process.env.ISSUER })

module.exports = async (req, res, next) => {
  try {
    const { authorization } = req.headers
    if (!authorization) throw new Error('You must send an Authorization header')

    const [authType, token] = authorization.trim().split(' ')
    if (authType !== 'Bearer') throw new Error('Expected a Bearer token')

    const { claims } = await oktaJwtVerifier.verifyAccessToken(token)
    if (!claims.scp.includes(process.env.SCOPE)) {
      throw new Error('Could not verify the proper scope')
    }
    next()
  } catch (error) {
    next(error.message)
  }
}

This function first checks that the authorization header is on the request and throws an error otherwise. If it exists, it should look like Bearer {token} where {token} is a JWT string. This will throw another error if the header doesn’t start with Bearer. Then we send the token to Okta’s JWT Verifier to validate the token. If the token is invalid, the JWT verifier will throw an error. Otherwise, it will return an object with some information. You can then verify that the claims include the scope that you’re expecting.

If everything is successful, it calls the next() function without any parameters, which tells Express that it’s OK to move on to the next function in the chain (either another middleware or the final endpoint). If you pass a string into the next function, Express treats it as an error that will be passed back to the client, and will not proceed in the chain.

You still need to import this function and add it as middleware to your app. You also need to load dotenv at the top of your index file to make sure that the environment variables from .env are loaded in your app. Make the following changes to index.js:

index.js

@@ -1,11 +1,14 @@
+require('dotenv').config()
 const express = require('express')
 const bodyParser = require('body-parser')
 const { promisify } = require('util')

+const authMiddleware = require('./auth')
 const initializeDatabase = require('./database')

 const app = express()
 app.use(bodyParser.json())
+app.use(authMiddleware)

 const startServer = async () => {
   await initializeDatabase(app)

To test that requests are properly blocked, try running it again…

$ npm test && node .

…then in another terminal run a few curl commands to test for:

1. An authorization header is required

$ curl localhost:3000/parts
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<title>Error</title>
</head>
<body>
<pre>You must send an Authorization header</pre>
</body>
</html>

1. A Bearer token is required in the authorization header

$ curl localhost:3000/parts -H 'Authorization: Basic asdf:1234'
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<title>Error</title>
</head>
<body>
<pre>Expected a Bearer token</pre>
</body>
</html>

1. The Bearer token is valid

$ curl localhost:3000/parts -H 'Authorization: Bearer asdf'
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<title>Error</title>
</head>
<body>
<pre>Jwt cannot be parsed</pre>
</body>
</html>

Create a Test Client in Node

You have now disabled access to the app for someone without a valid token, but how do you get a token and use it? I’ll show you how to write a simple client in Node, which will also help you test that a valid token works.

npm install btoa@1.2.1 request-promise@4.2.2

client.js

require('dotenv').config()
const request = require('request-promise')
const btoa = require('btoa')

const { ISSUER, CLIENT_ID, CLIENT_SECRET, SCOPE } = process.env

const [,, uri, method, body] = process.argv
if (!uri) {
  console.log('Usage: node client {url} [{method}] [{jsonData}]')
  process.exit(1)
}

const sendAPIRequest = async () => {
  const token = btoa(`${CLIENT_ID}:${CLIENT_SECRET}`)
  try {
    const auth = await request({
      uri: `${ISSUER}/v1/token`,
      json: true,
      method: 'POST',
      headers: {
        authorization: `Basic ${token}`
      },
      form: {
        grant_type: 'client_credentials',
        scope: SCOPE
      }
    })

    const response = await request({
      uri,
      method,
      body,
      headers: {
        authorization: `${auth.token_type} ${auth.access_token}`
      }
    })

    console.log(response)
  } catch (error) {
    console.log(`Error: ${error.message}`)
  }
}

sendAPIRequest()

Here the code is loading the variables from .env into the environment, then grabbing them from Node. Node stores environment variables in process.env (process is a global variable with a bunch of useful variables and functions).

require('dotenv').config()
// ...
const { ISSUER, CLIENT_ID, CLIENT_SECRET, SCOPE } = process.env
// ...

Next, since this will be run from the command line, you can use process again to grab the arguments passed in with process.argv. This gives you an array with all the arguments passed in. The first two commas are there without variable names in front of them because the first two are unimportant in this case; those will just be the path to node, and the name of the script (client or client.js).

The URL is required, which would include the endpoint, but the method and JSON data are optional. The default method is GET, so if you’re just fetching data you can leave that out. You also wouldn’t need any payload in that case. If the arguments don’t seem right, then this will exit the program with an error message and an exit code of 1, signifying an error.

const [,, uri, method, body] = process.argv
if (!uri) {
  console.log('Usage: node client {url} [{method}] [{jsonData}]')
  process.exit(1)
}

Node currently doesn’t allow for await in the main thread, so to make use of the cleaner async/await syntax, you have to create a function and then call it afterward.

If an error occurs in any of the awaited functions, the try/catch they’ll be printed out to the screen.

const sendAPIRequest = async () => {
  try {
    // ...
  } catch (error) {
    console.error(`Error: ${error.message}`)
  }
}

sendAPIRequest()

This is where the client sends a request to the authorization server for a token. For authorizing with the authorization server itself, you need to use Basic Auth. Basic Auth is the same thing a browser uses when you get one of those built-in pop-ups asking for a username and password. Say your username is AzureDiamond and your password is hunter2. Your browser would then concatenate them together with a colon (:) and then encode them with base64 (this is what the btoa function does) to get QXp1cmVEaWFtb25kOmh1bnRlcjI=. It then sends an authorization header of Basic QXp1cmVEaWFtb25kOmh1bnRlcjI=. The server can then decode the token with base64 to get the username and password.

Basic authorization isn’t inherently secure because it’s so easy to decode, which is why https is important, to prevent a man-in-the-middle attack. Here, the client ID and client secret are the username and password, respectively. That’s also why it’s important to keep your CLIENT_ID and CLIENT_SECRET private.

For OAuth 2.0, you also need to specify the grant type, which in this case is client_credentials since you’re planning to talk between two machines. You also need to specify the scope. There are a lot of other options that could be added here, but this is all we need for this demo.

const token = btoa(`${CLIENT_ID}:${CLIENT_SECRET}`)
const auth = await request({
  uri: `${ISSUER}/v1/token`,
  json: true,
  method: 'POST',
  headers: {
    authorization: `Basic ${token}`
  },
  form: {
    grant_type: 'client_credentials',
    scope: SCOPE
  }
})

Once you’re authenticated, you’ll get an access token that you can send along to your REST API that should look something like Bearer eyJra...HboUg (the actual token is much longer than that – likely somewhere around 800 characters). The token contains all the information needed for the REST API to verify who you are, when the token will expire, and all kinds of other information, like the scopes requested, the issuer, and the client ID used to request the token.

The response from the REST API is then printed to the screen.

const response = await request({
  uri,
  method,
  body,
  headers: {
    authorization: `${auth.token_type} ${auth.access_token}`
  }
})

console.log(response)

Go ahead and test it out now. Again, start the app with npm test && node ., then try out some commands like the following:

$ node client http://localhost:3000/parts | json
[
  {
    "id": 1,
    "partNumber": "abc-123",
    "modelNumber": "xyz-789",
    "name": "Alphabet Soup",
    "description": "Soup with letters and numbers in it",
    "createdAt": "2018-08-16T02:22:09.446Z",
    "updatedAt": "2018-08-16T02:22:09.446Z"
  }
]

$ node client http://localhost:3000/parts post '{
  "partNumber": "ban-bd",
  "modelNumber": 1,
  "name": "Banana Bread",
  "description": "Bread made from bananas"
}' | json
{
  "id": 2,
  "partNumber": "ban-bd",
  "modelNumber": "1",
  "name": "Banana Bread",
  "description": "Bread made from bananas",
  "updatedAt": "2018-08-17T00:23:23.341Z",
  "createdAt": "2018-08-17T00:23:23.341Z"
}

$ node client http://localhost:3000/parts | json
[
  {
    "id": 1,
    "partNumber": "abc-123",
    "modelNumber": "xyz-789",
    "name": "Alphabet Soup",
    "description": "Soup with letters and numbers in it",
    "createdAt": "2018-08-16T02:22:09.446Z",
    "updatedAt": "2018-08-16T02:22:09.446Z"
  },
  {
    "id": 2,
    "partNumber": "ban-bd",
    "modelNumber": "1",
    "name": "Banana Bread",
    "description": "Bread made from bananas",
    "createdAt": "2018-08-17T00:23:23.341Z",
    "updatedAt": "2018-08-17T00:23:23.341Z"
  }
]

$ node client http://localhost:3000/parts/1 delete | json
{}

$ node client http://localhost:3000/parts | json
[
  {
    "id": 2,
    "partNumber": "ban-bd",
    "modelNumber": "1",
    "name": "Banana Bread",
    "description": "Bread made from bananas",
    "createdAt": "2018-08-17T00:23:23.341Z",
    "updatedAt": "2018-08-17T00:23:23.341Z"
  }
]

Learn More About Node and OAuth 2.0 Client Credentials with Okta

Hopefully you’ve seen how easy it is to create a REST API in Node and secure it from unauthorized users. Now that you’ve had a chance to make your own sample project, check out some of these other great resources about Node, OAuth 2.0, and Okta. You can also browse the Okta developer blog for other excellent articles.

• Implementing the Client Credentials Flow
• Validating Access Tokens
• Customizing Your Authorization Server
• Tutorial: Build a Basic CRUD App with Node.js
• Secure a Node API with OAuth 2.0 Client Credentials

As always, you can hit us up in the comments below with feedback or questions, or on Twitter @oktadev. We look forward to hearing from you!