Skip to content
Ryan Bush edited this page May 5, 2017 · 1 revision

One of the core goals of VIPER is to make unit testing easy. Many times developers are introduced to unit testing and end up hating it because it can be difficult. But why is it difficult? More often then not, the code was built without the thought of testing in mind. In other words, it wasn't written to be tested.

VIPER, by default, is built to be tested by abstracting the interfaces into protocols. This allows anything to conform to the protocol and thus the layers are easily injectable for unit testing.

Say we need to write code to capture a user tapping login. The user story would be something like this:

Given I am a user trying to login
When I tap the login button
Then I should login with my username and password

The tap on the button needs to cause the Presenter to be told that the user is trying to login in using the inputted username and password. Inside the View layer, the variable presenter is anything that conforms to the ViewToPresenterInterface. So what do we do if we want to test that, when the user taps login, the Presenter is told of the user event? The templates already abstract out the ViewToPresenterInterfaceMock. So in TDD we would start by writing our failing test, something like this:

//ViewTests.swift
@testable import Project
class ViewTests: XCTestCase {
    var view: JogsView!
    var presenterMock: JogsViewToPresenterInterfaceMock!

    override func setUp() {
        super.setUp()

        let sb = UIStoryboard(name: Jogs.storyboardIdentifier, bundle: Bundle(for: JogsView.self))
        view = sb.instantiateViewController(withIdentifier: .viewIdentifier) as! JogsView
        presenterMock = JogsViewToPresenterInterfaceMock()

        _ = view.view

        // This is where we easily inject a mocked interface into the view for testing
        view.presenter = presenterMock
    }
    
    func testLoginWasTappedWithUsernameAndPasswordShouldTellPresenterUserTappedLogin() {
         // Arrange
         view.usernameField.text = "testUserName"
         view.passwordField.text = "testPassword"

         // Act
         view.loginTapped(view.loginButton)

         // Assert
         XCTAssert(presenterMock.functionsCalled.contains("userTappedLogin(withUsername:andPassword:)"))
         XCTAssertEqual(presenterMock.withUsername, "testUserName")
         XCTAssertEqual(presenterMock.andPassword, "testPassword")
    }
}

Now we can run this test and see that it fails because we haven't implemented anything yet. Next, we define the interface function to call for the Presenter.

//PresenterProtocols.swift
protocol ViewToPresenterInterface {
    func userTappedLogin(withUsername username: String, andPassword password: String)
}

Pretty good, next we need to have our actual Presenter and our ViewToPresenterInterfaceMock conform to the protocols.

//Presenter.swift
class Presenter {
    weak var interactor: PresenterToInteractorInterface!
    weak var view: PresenterToViewInterface!
    weak var wireframe: PresenterToWireframeInterface!
}

extension Presenter: ViewToPresenterInterface {
    func userTappedLogin(withUsername username: String, andPassword password: String) {
         interactor.login(withUsername: username, andPassword: password)
    }
}

For the mocked interface, we just need to update the extension to conform to protocol, and save the input values so we can test them. Our mock turns into something like this:

//ViewToPresenterInterfaceMock.swift
class ViewToPresenterInterfaceMock {
    var functionsCalled = [String]()
    
    var andPassword: String?
    var withUsername: String?
}

extension ViewToPresenterInterfaceMock: ViewToPresenterInterface {
    func userTappedLogin(withUsername username: String, andPassword password: String) {
        functionsCalled.append(#function)
        withUsername = userName
        andPassword = password
    }
}

Great, everything conforms to the protocols we need them to, the View tests are injecting the mocked ViewToPresenterInterface, and everything compiles. However, we still have our failing test because we haven't implemented the functionality yet.

//View.swift
class View {
    weak var presenter: ViewToPresenterInterface!
    
    @IBAction func loginTapped(sender: AnyObject) {
        let username = usernameTextField.text
        let password = passwordTextField.text
        presenter.userTappedLogin(withUsername: username, andPassword: username)
    }
}

Now we are at a spot where we can run our tests and see if the presenter is told of the user event with the correct values being passed in. What happens when we run the tests?

func testLoginWasTappedWithUsernameAndPasswordShouldTellPresenterUserTappedLogin() {
    // Arrange
    view.usernameField.text = "testUserName"
    view.passwordField.text = "testPassword"

    // Act
    view.loginTapped(view.loginButton)

    // Assert
    XCTAssert(presenterMock.functionsCalled.contains("userTappedLogin(withUsername:andPassword:)"))
    XCTAssertEqual(presenterMock.withUsername, "testUserName")
    XCTAssertEqual(presenterMock.andPassword, "testPassword")
    // TEST FAILED: "testUserName" != "testPassword"
}

Oh dear, how could our test have failed?? But the message makes sense and we can easily find the error

//View.swift
class View {
    weak var presenter: ViewToPresenterInterface!
    
    @IBAction func loginTapped(sender: AnyObject) {
        let username = usernameTextField.text
        let password = passwordTextField.text
        presenter.userTappedLogin(withUsername: username, andPassword: username)
        // We passed the username as the password on accident!
    }
}

A quick change to the presenter call and we are good to go!

presenter.userTappedLogin(withUsername: username, andPassword: password)

Run the unit tests again and they should pass without problems.