Skip to content
Dynamic test framework for web and mobile applications
Branch: master
Clone or download
Type Name Latest commit message Commit time
Failed to load latest commit information.
justtestlah-awsdevicefarm Set version to 1.5 May 14, 2019
justtestlah-browserstack Set version to 1.5 May 14, 2019
justtestlah-core Set version to 1.5 May 14, 2019
justtestlah-demos Set version to 1.5 May 14, 2019
justtestlah-mobile-tools Set version to 1.5 May 14, 2019
src/site Fix travis build status link Jun 7, 2019
.gitignore Release 1.3 Oct 25, 2018
.travis.yml Release 1.5 (#26) May 7, 2019
LICENSE Create LICENSE Apr 15, 2018 Use symlink for Oct 13, 2018
_config.yml Initial commit Apr 11, 2018
pom.xml Set version to 1.5 May 14, 2019

JustTestLah! 🇸🇬

Build status Maven Central

JustTestLah! is a JAVA test framework targeting projects that support multiple platforms, in particular Web, Android and iOS. It follows a BDD approach and allows testing against all platforms using the same feature files. JustTestLah's main aim is to make the configuration as easy and the test code as simple and readable as possible.

Getting started

Pull the repo and run the example. It includes automated tests for Stack Overflow and Carousell.

git clone
cd justtestlah-demos
mvn test -Dtest=TestRunner

The default platform is web. To test one of the mobile apps you need to setup Appium and start an Appium server. You also need at least one physical or emulated device connected. Then simply execute the tests by setting platform=android or platform=ios in Please note that the Stackoverflow demo is only available for web and android (upvote this question to help change this). For the Carousell demo, you need to have a Carousell account (it's free). Configure username and password in

mvn test -Dtest=TestRunner -Dtestusers.file=/absolute/path/to/your/

Both parameters are optional; the default configuration files can be found under justtestlah-demos/src/test/resources. The is only required for tests which make use of io.github.martinschneider.justtestlah.user.UserService.

Use in your own projects

Add the following Maven dependency to your pom.xml.


Page objects, steps and feature files

There are three main ingredients for tests in JustTestLah!:

  • Page objects are a representation of a UI element (a page, a dialog, a screen etc.).
  • Step definitions use page objects to define the actions of a test. They form the building blocks to write
  • feature files which represent the test cases.

Steps and page objects are designed to be highly re-usable.

Demo of a feature file:

Feature: Search and tags 

Scenario: Filter by tags 
  Given I am on the homepage 
  When I go to the tags page 
  And I filter for "selenium" 
  And I select the tag "selenium" 
  And I select the first question 
  Then the question is tagged with "selenium" 
@web @android 
Scenario: Use the search function 
  Given I am on the homepage 
  When I search for "selenium" 
  And I select the first question 
  Then the question is tagged with "selenium"

Demo of a step definition class:

public class HomeSteps extends BaseSteps {
  private HomePage home;

  @Given("^I am on the homepage$")
  public void homepage() {

  @When("^I go to the tags page")
  public void goToTags() {

  @When("I search for \"([^\"]*)\"")
  public void search(String query) {;

Demo of a page object:

public class HomePage extends BasePage<HomePage> {

  private QuestionsPage questions;

  private TagsPage tags;

  public HomePage load() {
    return this;

  public TagsPage navigateToTagsPage() {
    return tags;

  public QuestionsPage search(String query) {
    return questions;

You can inject page objects in steps by declaring a private field:

private HomePage home;

As long as the page object class extends io.github.martinschneider.justtestlah.base.BasePage JustTestLah! (and Spring) will take care of the rest. In the same way you can also use page objects inside other page objects.


All configuration goes in a file called

# GENERAL settings

# optional

# WEB settings

# MOBILE settings

# ANDROID settings
android.deviceName=Google Nexus 6

# IOS settings
ios.deviceName=iPhone 6

# BROWSERSTACK settings (requires cloudprovider=browserstack)

You can specify the location of on start-up by providing it as a system property: -DjusttestlahProperties=/path/to/ If no path is specified it will be loaded from the classpath.

Test runner

JustTestLah! uses JUnit to run the tests. All you need to do is add an empty class which extends io.github.martinschneider.justtestlah.JustTestLahTest:

public class TestRunner extends JustTestLahTest {}

Alternatively, you can also use the JUnit test runner directly:

public class SomeTestClass {}

The feature files and steps are automatically picked up from the locations provided in


Elements can be identified by a unique id, a css or an xpath expression. AccesibilityId (for iOS) and UIAutomator (for Android) are supported as well. Each element has a unique key (e.g. SEARCH_FIELD) which is mapped to its corresponding locator expression in a .yaml file. For example, let's say the page object for the home page is demoproject.pages.HomePage (under /src/main/java). Then the corresponding locators are expected in /demoproject/pages/HomePage.yaml (under /src/main/resources).

Example of a locator YAML file:

    type: xpath
    value: "//BUTTON//SPAN[text()='Log in']"
    type: accesibilityId
    value: login_page_login_button
    type: id
    value: com.thecarousell.Carousell:id/login_page_login_button

The correct locator will be automatically resolved for the current platform. Taking the above example, the search field can be accessed in the HomePage page object by calling $("SEARCH_BUTTON"). This will return an instance of com.codeborne.selenide.SelenideElement. See the Selenide quick start to learn about all the cool ways you can interact with it. Two caveats to take note of:

  1. It is not possible to directly use elements in step definitions (only in page objects). This is by design as UI elements are meant to be encapsulated in the page objects.
  2. While we wrap Selenide's $ method for the locator handling the methods you can call on the returned SelenideElement instances remains the same.

If omitted the default type of locators is css.


Locators can include placeholders which will be replaced by variables passed to the $ method. For example:

    type: xpath
    value: "//A[contains(@class,'post-tag') and contains(text(),'%s')]"

Calling $("POST_TAG", "selenium") will return an element matching the following Xpath expression: //A[contains(@class,'post-tag') and contains(text(),'selenium').

Template matching

JustTestLah! allows locating elements using a template image:

boolean isImagePresent = homePage.hasImage("questionIcon.png");

Match image = homePage.findImage("questionIcon.png");

The images are expected under /src/test/resources/images.

The Match object contains the x and y coordinate of the matched image (more precisely, the center of the rectangle representing the match). These can be used to interact with an element located this way. For example, we can tap on an element like this:

new TouchAction((PerformsTouchActions) WebDriverRunner.getWebDriver())
.tap(PointOption.point(questionIcon.getX(), questionIcon.getY()))

Note, that future versions of JustTestLah! will include wrappers to perform these actions more conveniently.

The TemplateMatcher is scale-invariant (to some extent). The algorithm used to achieve this scales the target image (a screenshot of the device) up and down until either a match is found or a minimum (320) or maximum (3200) image width is reached.

Note, that the closer the size of the template matches the size of the image on the screen the faster and more accurate the matching will be.

Matching threshold

Both the hasImage and findImage method take an optional threshold parameter which can be used to define the accuracy of a match. The possible values range from 0 (no match) to 1 (pixel-perfect match). The default is 0.9.

Client and server-mode matching

There are two modes to use template matching which can be configured in

opencv.mode=client performs the image matching on the client (i.e. the machine running the test code). It requires OpenCV which is imported as a Maven dependency (

opencv.mode=server utilises the image matching feature of Appium. This requires OpenCV to be installed on the machine which runs the Appium server.

Note, that not all cloud providers (see below) support this.

Test user management

JustTestLah! implements a basic test user configuration. Test users are stored in a properties file passed as a VM argument: -Dtestusers.file=/absolute/path/to/ The content of the file looks as follows:


You can then call io.github.martinschneider.justtestlah.user.UserService.get(String) with the key as a parameter to retrieve the test user data as an object of type io.github.martinschneider.justtestlah.user.User.

For example:

public class LoginSteps extends BaseSteps {
  @Autowired private UserService userService;
  private HomePage homePage;
   * Login the given user.
   * @param userKey userKey of the user to log in
  @When("^I login as \"([^\"]*)\"$")
  public void loginAs(String userKey) {
@Profile({Platform.ANDROID, Platform.WEB})
public class LoginPage extends BasePage<LoginPage> {
  private HomePage home;

   * Log in the given user.
   * @param user the user to log in
   * @return {@link HomePage}
  public HomePage login(User user) {
    return home;


JustTestLah! includes a proof-of-concept integration of the Galen framework. It can be enabled by setting galen.enabled=true in

Similar to properties-file holding the locator information, there is an (optional) spec file for each page object (in the same package as the Java class under src/main/resources).

Checks can be triggered by calling checkLayout() on any page object class. An HTML report is generated in the directory defined in in (the default is target/galen-reports/).

  username_field  id  com.thecarousell.Carousell:id/login_page_username_text_field
  password_field  id  com.thecarousell.Carousell:id/login_page_password_text_field
  login_button    id  com.thecarousell.Carousell:id/login_page_login_button

= Login =

      above password_field
      aligned vertically all password_field
      width 100 % of password_field/width
      width 100 % of login_button/width
      below username_field
      aligned vertically all username_field
      width 100 % of username_field/width
      width 100 % of login_button/width
      below password_field
      below username_field
      width 100 % of username_field/width
      width 100 % of password_field/width
      text is "Log In" 

See the Galen documentation for more examples.

Cloud service integrations

JustTestLah! supports integration with various cloud service provides. Most of them are in PoC state. Please feel free to contribute.


There is a proof-of-concept integration of Applitools. It can be enabled by setting eyes.enabled=true in In addition a valid API key must be specified: eyes.apiKey=....

Checks can then be triggered by calling checkWindow() on any page object class (the initial run will create baseline images). Please note that Applitools is a paid service.

AWS Devicefarm

You can run tests against AWS Devicefarm by adding the following configuration in


# The arn of your AWS Devicefarm project (mandatory)

# App package to use. If this value is empty it will be created and uploaded to AWS Devicefarm before the test execution

# Test package to use. If this value is empty it will be created and uploaded to AWS Devicefarm before the test execution

# Optional extra data

# Fully-qualified path to the justtestlah-demos project (required to build the test package)

# Name for the test package (must match <finalName> in the justtestlah-demos pom.xml)

# Device filters (optional)

# Optional device configuration
# set this to true if you use device slots 

# Additional AWS Devicefarm configuration

Make sure justtestlah-awsdevicefarm is on your classpath:


Please note that AWS Devicefarm is a paid service.


You can run tests against BrowserStack by adding the following configuration in


# Browserstack username
# Browserstack access key

# Optional settings, see

Make sure justtestlah-browserstack is on your classpath:


Please note that BrowserStack is a paid service.

Used frameworks (selection)

JustTestLah! makes use of a variety of frameworks to make writing and executing tests as transparent and simple as possible.

  • Selenium, the main test framework used by JustTestLah!
  • Appium, an extension of Selenium for native mobile app testing
  • Cucumber, the BDD framework
  • JUnit, the unit testing framework (mostly used as the runner for the tests)
  • Selenide, a convenience mapper around Selenium
  • AssertJ, fluent assertions for unit tests
  • OpenCV, used for image comparison
  • Galen, used for layout based testing
  • Applitools, used for visual regression testing
  • BrowserStack, cloud provider for automated tests
  • Spring, IoC container for some added "magic" behind the scenes


This framework (under the name YaSeW) started as a PoC for the 2nd Singapore Appium Meet-up.

It has been showcased and mentioned in various presentations:

  • Martin Schneider: Android, iOS and Web testing in a single framework & Image-based testing with Appium and OpenCV. 2018-04-12, 2nd Singapore Appium Meet-up

  • Martin Schneider: A single framework for Android, IOS and Web testing. 2018-11-09, Testingmind Software Testing Symposium, Manila

  • Abhijeet Vaikar: Breaking free from static abuse in test automation frameworks. 2018-11-28, 6th Singapore Appium Meet-up

  • Martin Schneider: Re-use automated test scenarios across different platforms. 2019-01-08, Testing Corner, Taipei

  • Martin Schneider: testDevices.scaleUp(); Thoughts about mobile testing on the cloud. 2019-05-07, Testing Corner, Taipei


This article explains how to use JustTestLah! (amongst other options) to query meta information from APK, IPA and APP packages.

This article describes how to utilize AWS Devicefarm with JustTestLah!

Known issues & limitations

  • JustTestLah! requires Java 10 or higher (and has been tested on Java 10, 11 and 12).

  • Java 9 support has been dropped because of JDK-8193802 which isn't fixed on Java below 10.

  • The OpenCV integration (used for client-side template matching) doesn't work with Java 12 yet.

  • The Galen PoC only works on Appium 1.7. Newer versions are not yet supported. Please feel free to contribute an update for this feature.

Contact and support

Please let me know about any feedback, questions or ideas for improvement.

Martin Schneider -

Buy me a coffee

You can’t perform that action at this time.