February 18, 2016   |   6min read

The perfect recipe for iOS app automation

A couple of weeks ago we shared with you our selection of the best Pods which have been used in Polidea during the past year. But we know that having pods is surely not enough to build a functioning app, you have to test them too (and no you are not more of a male, if you wait production to do so). Having that in mind, today I would like to present in this post useful tutorials to implement iOS functional test automations. Because the automation of testing procedures free staff and resources, which in return maximize your productivity and minimize errors, you better start soon. So let’s get started!

Testing like a User

The main ingredient is an open-source testing framework, KIF which stands for Keep It Functional. KIF aims to operate on the app and attempts to imitate factual users actions and inputs on a touch screen. KIF localizes apps elements on the screen based on accessibilityLabels and accessibilityIdentifiers attributes - in most cases accessibilityLabel matches the visible text label. The most common actions in KIF are waitForViewWithAccessibilityLabel: and tapViewWithAccessibilityLabel:, but if you want to find out more what KIF is capable of, then take a look for a whole bunch of available test methods. Every test methods are accessed by using tester which is an instance of the main class KIFUITestActor to drive the app UI and make inputs and asserts:

- (void)enterUserCredentialsWithEmail:(NSString *)email
                          andPassword:(NSString *)password {
    [tester enterText:email intoViewWithAccessibilityLabel:kEmailTextField];
    [tester enterText:password intoViewWithAccessibilityLabel:kPasswordTextField];
    [tester tapViewWithAccessibilityLabel:kLoginButton];

To run KIF tests you don’t have to go outside Xcode IDE - tests are written as Apple’s XCTest subclasses and you can trigger them using Xcode’s test navigator. Unless you are a terminal fan or want to stick your test suite into a continuous integration - then no extra infrastructure is needed, because KIF uses command line xcodebuild or xctool tools to build and run tests:

 - workspace TestProject.xcworkspace
 - scheme TestProjectKIF
 - sdk iphonesimulator 9.2
 - destination 'name=iPhone 6'

Pixel perfect checks

KIF is super cool in terms of accessibility base functional testing of the application, but:

  1. What if there are UI artifacts and glitches on the application’s screen?
  2. What if a screen does not match the particular iOS device screen size?

Then you still need testers to manually verify, check and compare with the wireframes every screen in the app, because KIF does not tell you if app views have changed visually.

Here comes along with help another ingredient which is Lela framework. Lela stands for “Less Eyeballing Large Apps” and is a library which helps KIF if apps screens appear different from expected. To make it work during your tests, after checking if all elements with accessibilityLabel are on the screen, include additional Lela’s assert:

[tester expectScreenToMatchImageNamed:@"LoginScreen"];

Then test will fail and Lela will take a screenshot:

Test failed: Could not find expected image for LoginScreen.
Actual result: /Library/Developer/CoreSimulator/Devices/6D74A6B4-46BF-42EB-9007-106C0601AD24/data/Containers/Data/Application/6D74A6B4-46BF-42EB-9007-106C0601AD24/Documents/Lela Tests/2016-02-14T19:23:14/LoginScreen-414x736@3x-iphone,iOS9.2/LoginScreen-414x736@3x-iphone,iOS9.2.png

Add image to your test bundle, re-run the test and here you are - PASSED result. In case the reference image differs from the expected one, Lela does all the work - test will fail and in the log output you will be served with a snippet which contains expected, actual and differences between images:

Screen does not match expected image for LoginScreen.
Expected result: /Library/Developer/CoreSimulator/Devices/6D74A6B4-46BF-42EB-9007-106C0601AD24/data/Containers/Data/Application/52E86D74-D96E-4024-8EB2-FCC273F4430D/Documents/Lela Tests/2016-02-11T19:28:00/LoginScreen-414x736@3x-iphone,iOS9.2/Expected.png
Actual result:   /Library/Developer/CoreSimulator/Devices/6D74A6B4-46BF-42EB-9007-106C0601AD24/data/Containers/Data/Application/52E86D74-D96E-4024-8EB2-FCC273F4430D/Documents/Lela Tests/2016-02-11T19:28:00/LoginScreen-414x736@3x-iphone,iOS9.2/LoginScreen-414x736@3x-iphone,iOS9.2.png
Difference:      /Library/Developer/CoreSimulator/Devices/6D74A6B4-46BF-42EB-9007-106C0601AD24/data/Containers/Data/Application/52E86D74-D96E-4024-8EB2-FCC273F4430D/Documents/Lela Tests/2016-02-11T19:28:00/LoginScreen-414x736@3x-iphone,iOS9.2/Difference.png

For each iOS device screen size, density and OS version Lela distinguishes separated images so you can validate as many types of devices as you need.

Hacks & tricks

Have you ever experienced an issue when the system alert view requested you to grant an access to the address book or photo album during your tests? It is not necessary to reinvent the wheel as long as you have a JPSimulatorHacks . You can hack iOS Simulator without a need to run any additional scripts which handle clicks and dismiss system alert views. Just simply give an access from your test suite setUp() to your current application:

[JPSimulatorHacks grantAccessToPhotos];
[JPSimulatorHacks grantAccessToAddressBook];

If you need a specific image in the gallery for further tests simply add it:

NSURL *assetURL = [NSURL URLWithString:@""];
[JPSimulatorHacks addAssetWithURL:assetURL];

Then during the test use KIF method to grab it:

[tester choosePhotoInAlbum:@"Camera Roll" atRow:(NSInteger) column:(NSInteger)]

What’s more while using to JPSimulatorHacks is that you can test your application with various keyboard parameters and change those settings on the fly:

[JPSimulatorHacks editApplicationPreferences:^(NSMutableDictionary *preferences) {
        [preferences setValue:@NO forKey:@"KeyboardAutocapitalization"];
        [preferences setValue:@NO forKey:@"KeyboardAutocorrection"];
        [preferences setValue:@NO forKey:@"KeyboardCapsLock"];
        [preferences setValue:@NO forKey:@"KeyboardCheckSpelling"];

Network stubs

Most of our applications are using networks and communicate with the API services. Our automation test suites contain correct user login, correct submitting and receiving data and a lot of other “happy” paths. But the devil is in the “non-happy” path scenarios when it comes to the communication between application and API. Applications should handle errors and invalid API responses properly to ensure good user experience in all (especially bad) conditions.

By using Nocilla framework you can simulate and trick your network data very easily, right from the tests:

 [[LSNocilla sharedInstance] start];

             withBody(@{  @"email" : @"","
                          @"password" : @"qwerty1234"}).
             withBody(@{  @"status": @"error",
                          @"message" : @"Wrong email or password."});

[[LSNocilla sharedInstance] stop];
[[LSNocilla sharedInstance] clearStubs];

Don’t forget to test how your app behaves against long API response time as well - simply add .withDelay() with appropriate time interval into your stub.

I hope you found this recipe helpful. A great advantage of our selection of testing frameworks and helpers is that you can combine them to work together in order to boost test automation efficiency. And what about you? What is your approach and toolbox for a perfect iOS functional test automation?

Tomek Mnich

Director of Quality

Did you enjoy the read?

If you have any questions, don’t hesitate to ask!