Read time: 6 minutes
Today I’ll show how to write integration tests for an ASP.NET Core API, step by step.
Integration tests are the best way to test an API, because they allow us to test the API in the context of the whole ASP.NET Core pipeline, which is something that unit tests can’t do.
In the past, writing integration tests was a bit challenging, because you had to figure out a way to standup and setup your ASP.NET Core app for test purposes, which was a lot of work.
But with the help of a handy NuGet package, you can now bootstrap your full ASP.NET Core app in memory, making the setup needed for integration tests very straightforward.
Let’s dive in.
The API to test
Let’s say we have this ASP.NET Core API controller, which is part of our Match Making system:
[ApiController]
[Route("api/matches")]
public class MatchesController : ControllerBase
{
private readonly GameMatcher matcher;
public MatchesController(GameMatcher matcher)
{
this.matcher = matcher;
}
[HttpPost]
[Authorize]
public async Task<GameMatchResponse> JoinMatch(JoinMatchRequest request)
{
var match = await matcher.MatchPlayerAsync(request.PlayerId);
return match.ToGameMatchResponse();
}
}
And let’s say we already have an extensive unit test suite for the GameMatcher.MatchPlayerAsync() method, as well as for the ToGameMatchResponse() extension method. So that’s all covered.
Now we want to test the JoinMatch controller action, which is the entry point to the whole match making process.
Write unit tests for the controller action?
Not really. Unit tests are not the best option to test controller actions, because:
-
This is a thin controller, with very little logic. GameMatcher is doing the heavy lifting here and is already unit tested.
-
ToGameMatchResponse can also be unit tested in isolation. No need to test it again here.
-
JoinMatch requires authorization, but a unit test that just calls JoinMatch would skip that completely.
What we really want to test here is that the controller action can do its job in the context of the whole ASP.NET Core pipeline, which is something that unit tests can’t do.
Therefore, it’s best to use a different type of test here: an Integration Test.
Let’s see how to write an integration test for this controller action, step by step.
Step 1: Create the test project and add dependencies
Let’s add a new test project for our integration tests:
dotnet new xunit -n MatchMaker.Api.IntegrationTests
Then, add the following NuGet packages to your new test project:
- Microsoft.AspNetCore.Mvc.Testing. Adds support for writing integration tests for ASP.NET Core apps.
- FluentAssertions. Adds support for writing assertions in a more readable way.
Also, add a project reference from your new test project to your API project:
dotnet add reference ../MatchMaker.Api/MatchMaker.Api.csproj
Step 2: Allow your test project to view your API internals
In preparation for next step, we’ll need to allow our test project to view the internal classes defined in our API project.
In particular, we need access from the test project to the Program class that is the entry point to our API.
It is via the Program class that our integration tests will be able to register all services and middleware as if we were running the API in real life.
To allow this, we need to add an InternalsVisibleTo element to our API project’s .csproj file (not the test project file):
Step 3: Add the test database connection string
We could use an in-memory database for our integration tests, but since our application uses SQL Server, why not use a real SQL Server instance for our tests?
That should give us the highest confidence that the entire stack is working properly.
Now, I’ll assume you already have a SQL Server instance running somewhere in your box, so I won’t get into how to stand up one.
But we still need to come up with the DB used for the tests and also how to make our test project aware of the corresponding connection string.
And, since storing things like that in your code is not a great idea (even for tests), you can just store the connection string as a user secret.
So, in your terminal, switch to your test project directory, and run this (replace YOUR-CONNECTION-STRING-HERE with your actual connection string):
dotnet user-secrets init
dotnet user-secrets set "ConnectionStrings:MatchMaker" "YOUR-CONNECTION-STRING-HERE"
For instance, since I’m using a SQL Server docker container, this is how my connection string looks like:
Server=localhost; Database=MatchMaker-Tests; User Id=sa; Password=MY-SA-PASSWORD-HERE;TrustServerCertificate=True
Important: Here you want to use a completely different database name, meant only for your tests. Please don’t use your real DB name!
Notice how my database is named MatchMaker-Tests here, not MatchMaker.
Step 4: Create a WebApplicationFactory
A WebApplicationFactory is a class that can bootstrap an entire application in memory, which is what we need to do in our integration tests.
Add this class to your test project:
internal class MatchMakerWebApplicationFactory : WebApplicationFactory<Program>
{
override protected void ConfigureWebHost(IWebHostBuilder builder)
{
builder.ConfigureTestServices(services =>
{
// Remove the existing DbContextOptions
services.RemoveAll(typeof(DbContextOptions<MatchMakerDbContext>));
// Register a new DBContext that will use our test connection string
string? connString = GetConnectionString();
services.AddSqlServer<MatchMakerDbContext>(connString);
// Delete the database (if exists) to ensure we start clean
MatchMakerDbContext dbContext = CreateDbContext(services);
dbContext.Database.EnsureDeleted();
});
}
private static string? GetConnectionString()
{
var configuration = new ConfigurationBuilder()
.AddUserSecrets<MatchMakerWebApplicationFactory>()
.Build();
var connString = configuration.GetConnectionString("MatchMaker");
return connString;
}
private static MatchMakerDbContext CreateDbContext(IServiceCollection services)
{
var serviceProvider = services.BuildServiceProvider();
var scope = serviceProvider.CreateScope();
var dbContext = scope.ServiceProvider.GetRequiredService<MatchMakerDbContext>();
return dbContext;
}
}
A few key things about this class:
-
The class has to be internal in order to be able to access the internal Program class of our API project.
-
ConfigureWebHost will be invoked by the WebApplicationFactory when it’s time to bootstrap our API in memory.
-
ConfigureTestServices method is where we register all the services that will be used exclusively during our tests. Gives you a chance to replace any service with a test version.
-
We use ConfigureTestServices to replace the real DBContext that our API uses with a new one that will use our test database.
-
GetConnectionString is a helper method that will read the connection string from our user secrets.
-
We also delete the test database via EnsureDeleted to ensure each test starts with a clean slate.
Let’s now go ahead and add the actual test.
Step 5: Create the integration test
This is the easy part. Now that we have our WebApplicationFactory in place, writing the integration test is as easy as writing any program that talks to an HTTP API.
Replace the default UnitTest1 class with this test class in your test project:
public class MatchesControllerTests
{
[Fact]
public async Task JoinMatchRequest_AddsPlayerToMatch()
{
// Arrange
var application = new MatchMakerWebApplicationFactory();
JoinMatchRequest request = new("P1");
var client = application.CreateClient();
// Act
var response = await client.PostAsJsonAsync("/api/matches", request);
// Assert
response.EnsureSuccessStatusCode(); // Status Code 200-299
var matchResponse = await response.Content.ReadFromJsonAsync<GameMatchResponse>();
matchResponse?.Id.Should().BePositive();
matchResponse?.Player1.Should().Be("P1");
matchResponse?.State.Should().Be(nameof(GameMatchState.WaitingForOpponent));
}
}
Our test is doing basically the following:
-
It creates a new instance of our WebApplicationFactory, which will bootstrap our API in memory and create our test database.
-
It creates a new JoinMatchRequest object, which is the payload that our API expects.
-
It creates a standard HTTP client out of the WebApplicationFactory.
-
It uses the HttpClient to post the JoinMatchRequest to the API, as any standard client would do.
-
It asserts that the response is a 200 OK, and that the response body contains the expected values.
Now, go ahead and run this test:
dotnet test
Unfortunately, you will get something like this:
Which is expected, because our API requires authorization, and we are not providing it as part of our test.
Let’s fix that.
Step 6: Add an authentication handler
An AuthenticationHandler is a class that can be used to provide authentication to our API.
Let’s add this class to our test project:
public class TestAuthHandler : AuthenticationHandler<AuthenticationSchemeOptions>
{
public TestAuthHandler(IOptionsMonitor<AuthenticationSchemeOptions> options,
ILoggerFactory logger, UrlEncoder encoder, ISystemClock clock)
: base(options, logger, encoder, clock)
{
}
protected override Task<AuthenticateResult> HandleAuthenticateAsync()
{
var identity = new ClaimsIdentity(Array.Empty<Claim>(), "Test");
var principal = new ClaimsPrincipal(identity);
var ticket = new AuthenticationTicket(principal, "TestScheme");
var result = AuthenticateResult.Success(ticket);
return Task.FromResult(result);
}
}
The key in this class is the HandleAuthenticateAsync method, which is where we create an AuthenticationTicket that will be used to authenticate our test request.
We now need to tell our WebApplicationFactory to take advantage of it, so let’s make an AddAuthentication call to ConfigureTestServices in our MatchMakerWebApplicationFactory:
Notice that here we use both the TestAuthHandler we just created and the TestScheme we used there to create the AuthenticationTicket.
Now run your test again:
dotnet test
And you should get a green light:
Mission accomplished!
Now you have a great way to verify that your API is working as expected, from the entry point all the way to the database, and back.
I hope this was useful.
Whenever you’re ready, there are 4 ways I can help you:
-
.NET Cloud Developer Bootcamp: Everything you need to build production ready .NET applications for the Azure cloud at scale.
-
All-Access Pass: A complete catalog of premium courses, with continuous access to new training and updates.
-
Patreon Community: Join for exclusive discounts on all my in-depth courses and access my Discord server for community support and discussions.
-
Promote yourself to 19,000+ subscribers by sponsoring this newsletter.