Categories
Coding

Creating an OAuth 2 server in AWS Lambda with C#

Google Smart Home and Alexa skills need to impersonate the user when calling private APIs, so an Account Linking is required.

Like the majority of the API economy, Google and Amazon support the OAuth 2 standard.

I will implement a OAuth server that supports the authorization code flow. This OAuth service will provide 2 endpoints:

  • Web login page
  • Token endpoint

It consists of a web login page that authenticates the user in my private API authentication system, and the provide back to Google/Amazon a grant code. With this grant code the Google/Amazon app can obtain from the token endpoint an access token and a refresh token. I will develop it in ASP.NET Core using a serverless AWS Lambda. For small number of calls (under 1 milion/month) it’s almost free. Most AWS services used in this tutorial are paid services, you have been warned!

This is the list of AWS services needed:

  • IAM: authentication service
  • CloudFormation
  • S3 bucket for CloudFormation files
  • APIGateway to get the Lambda public URL

Prerequisites

We need an AWS account and Visual Studio 2019.

Visual Studio ▶Extensions▶Manage extensions: install AWS Toolkit for Visual Studio 2019 (version now is 1.15.2.1).

The VS solution

Create a new AWS Serverless Application (.NET Core C#):

In the next dialog select the ASP.NET Core Web App blueprint:

We will find a full ASP.NET core project, based on .NET Core 2.1 and a few Razor pages.

The OAuth authorization code flow will redirect the user browser to our login page using such an URL: https://myservice.example.com/auth along with 4 parameters:

  • client_id
  • redirect_uri
  • state
  • response_type

So let’s add an auth login page that can receive these four parameters.

The login page

The login page is a razor page with a login/password form:

public async Task<IActionResult> OnPostAsync()
        {
                //TODO: check url params
                var login = await _ApiService.LoginAsync(Username, Password);
                if (login) {
                    //TODO: generate a code
                    string uri = RedirectUri + "?code=XXXXX&state=" + State;
                return Redirect(rup);
            }

The base html will be:

<form method="post">
    <input asp-for="Username" class="form-control" />
    <input asp-for="Password" class="form-control" type="password" />
    <input type="submit" value="Login" class="btn btn-primary" />
</form>

Summary: the login page is just a username/password form that authenticate against my private API backend, create an auth code, and redirect back to the caller, passing the auth code.

Token endpoint

The token endpoint is a Web Api controller:

  • Create an Api subfolder
  • Right click and select Add Items▶Api Controller Class
  • Call it ‘tokenController’

The token endpoint is responsible for two post action:

  • receive a grant auth code and return refresh+access tokens
  • receive a refresh token and return an access token
 [Route("api/v1/token")]
    [ApiController]
    public class tokenController : Controller
    {
        // POST api/v1/token
        [HttpPost]
        public IActionResult Post(string client_secret, string client_id, string grant_type, string code, string refresh_token) {
    if(authCodeRequest)
    {
        //generate refresh + access tokens
        return Content("{\"token_type\": \"Bearer\",\"access_token\": \"" + accessToken + "\",\"refresh_token\": \"" + refreshToken + "\",\"expires_in\": " + 3600 + "}");
    }
    else if (refreshTokenRequest)
    {
        return Content("{\"token_type\": \"Bearer\",\"access_token\": \"" + accessToken + "\",\"expires_in\": " + 3600 + "}");
    }

Publishing to AWS

Now that we have a working ASP.NET razor login page, and the token POST endpoint, we need to publish it in an AWS lambda, and it’s where the fun starts!

Visual Studio, with the ASW toolkit, provide a Publish to AWS Lambda… shortcut:

Publish to AWS Lambda…

Before publishing we need to understand what Publish to AWS Lambda means. In particular we need to know which AWS services will be created (also to understand what we will pay!).

AWS CloudFormation: think of it as a script that configure and starts all AWS services needed. It’s free.

Lambda function: the serverless function that will run the ASP.NET solution. Free for the first few millions calls.

S3 Bucket: the storage where the cloudformation script and the asp.net zipped solution will be stored. You pay for the disk usage.

API Gateway: the public endpoint that expose the ASP.NET site. Free for the first million calls monthly.

The Visual Studio project is already provided with a CloudFormation serverless.template configuration file that will create automatically all the AWS services required. This collection of services is called a stack.

Prerequisite: Install the AWS CLI to do the advanced administative work. You can test it in PowerShell, enter aws help:

IAM: create the administrative account in AWS to upload and run the lambda

To run the publish wizard Visual Studio needs an AWS user with the proper permissions. Go to AWS console, select IAM service and go to Users▶Add.

Visual Studio will use AWS SDK, so a programmatic access it’s enough:

Step 2: Set Permission. Select the Attach existing policies directly and select

  • AWSLambdaBasicExecutionRole
  • AmazonS3FullAccess
  • AWSCloudFormationFullAccess

Step 3: Tags. Skip this step safely.

Step 4: Review. The review page should look like this:

Click Create user. And be sure to download the CSV file!

This is your only chance to download the CSV file, make sure to backup it safely!

WARNING: The permissions set are not enough. The quickest way to configure the IAM user would be to give all permissions, but this could be a security issue. So we will select only the permissions we really need to create a ASP.NET stack.

Enter the IAM user and add a custom policy:

Add these permissions to the policy:

  • API Gateway
    • GET
    • DELETE
    • PATCH
    • POST
  • Lambda
    • GetFunction
    • GetFunctionConfiguration
    • CreateFunction
    • DeleteFunction
    • UpdateFunctionConfiguration
    • UpdateFunctionCode
    • AddPermission
    • RemovePermission
    • ListTags
    • TagResources
    • UntagResources
  • IAM
    • ListRolePolicies
    • ListRoles
    • ListRoleTags
    • ListUserTags
    • GetRole
    • CreateRole
    • DeleteRole
    • PassRole
    • AttachRolePolicy
    • DeleteRolePolicy
    • DetachRolePolicy
    • PutRolePolicy

Save the policy!

Let’s go back to Visual Studio

The CSV file is all we need to make Visual Studio publish the Lambda. Right click on the ASP.NET project and select Publish to AWS Lambda…:

Click on the ‘+’ icon to upload the CSV:

Enter a Profile name and select the CSV file:

Create a bucket, and give a CloudFormation stack name:

This is the final screen of the configured wizard:

When clicking Publish the wizard will start uploading the lambda and then creating the infrastructure with CloudFormation. At the end the status window will show the progresses:

? Congratulations! Your OAuth server is ready on-air!

Leave a Reply

Your email address will not be published. Required fields are marked *