Acceptance Testing with Selenium
Last updated
Last updated
Puppet web apps are C# apps with a thin presentation layer rendered in HTML5. Because of that, acceptance tests need to be executed in a web browser. If you would like to add automated acceptance testing to your project, there is no better tool than Selenium.
This page teaches how to use NUnit to run Selenium acceptance tests of a Starcounter Puppet web app. It is assumed here that you are using Visual Studio 2015, though it should also work with the older editions.
Selenium is a free (open source) automated testing suite for web applications across different browsers and platforms.
Selenium is not just a single tool but a suite of tools, each catering to different testing needs. The components that we will use are:
Selenium WebDriver, which is an API that controls the web browser. It is implemented in all major programming languages, including C#.
Selenium RemoteWebDriver, a Selenium WebDriver implementation that we will use, because it allows to execute the tests on the same or different machine. It is implemented in all major programming languages, including C#.
Selenium Server Standalone, a Java-based server that runs specific web browser client drivers. RemoteWebDrivers runs the tests through it.
Selenium client drivers for web browsers such as Firefox, Chrome and Edge.
Selenium IDE, a Firefox extension that allows to record the tests, so you don't have to write them manually.
You can learn more about Selenium at their or read one of the interesting blog posts:
NUnit is a testing framework that has great integration with Visual Studio as well as continuous integration systems (TeamCity, etc). We will use NUnit to run our Selenium tests.
It is recommended to keep the testing project in the same solution as the tested project. Note that they are two different projects.
In the following example, we will add acceptance tests to Launcher. Let's create a new test project in Visual Studio, by clicking Add New Project → Visual C# → Test → Unit Test Project.
Call the new project "Launcher.AcceptanceTest". We will use Launcher\test\ as the project location.
Open the newly created test project. Now, we need to install a bunch of libraries mentioned above.
Open the package manager (Tools → NuGet Packet Manager → Packet Manager Console).
Important: In the console, choose your test project from the "Default project" dropdown.
Run the following commands in the console to install the required dependencies:
We don't want to test only Firefox. Because of that, you should use Selenium RemoteWebDriver, which adds a layer of abstraction that runs tests in multiple browsers, possibly on remote machines.
To make this happen, you will need to install some additional software.
Download and install Java, required by Selenium Standalone Server
Create a directory C:\Selenium
Selenium Standalone Server (selenium-server-standalone-3.*.jar)
Gecko Driver (geckodriver.exe)
Microsoft Edge Driver (MicrosoftWebDriver.exe)
Google Chrome Driver (chromedriver.exe)
Put the jar file as well as 3 exe files directly in C:\Selenium
Open your Properties in the Tests project. Go to Reference Paths, enter C:\Selenium and click Add Folder.
BaseTest is a helper class that makes it easier to test multiple browsers. The source code is available:
When you rebuild the test project now, you should see each test for every browser.
The final setup looks like this:
Before you can execute the tests, start Selenium Server Standalone by calling java -jar selenium-server-standalone-3.*.jar.
There is one common pitfall when writing Selenium tests. The test is executed with disregard of the asynchronously loaded content. This means that your tests need to explicitly wait for UI elements to appear before running any actions or assertions on them.
It is a good practice to always wait:
Wait for a text element to be present before you check the content of that element
An example can be found in the method TextareaPage_WriteToTextArea in KitchenSink
compare the text value of TextareaInfoLabel asynchronously. The assertion passes if the
text is found within 5 seconds, otherwise it fails.
Wait for a button to be present before you click on that button
An example can be found in the method ButtonPage_RegularButton in the KitchenSink
asynchronously check the state of the Displayed property of a button. It halts the test
for a default maximum time of 10 seconds. If the button is not displayed within that
time, it throws an exception.
Wait for presence of an input field before typing in it and wait for text to be present in label
An example can be found in the method TextPage_TextPropagationOnUnfocus in the
presented above. The method WaitUntil() is used here to asynchronously wait for the
input to be Displayed. Notice that the following response check is also done
asynchrously using WaitForText.
Build your test project. If it builds correctly, you should see this:
Now, the only thing left to do is to run that test! In the Test Explorer, click on the "Run All" button. If it works well, you should see your tests passing.
Some of the Starcounter's sample apps come with acceptance test suite. We run tests every night to make sure that we keep the good quality.
NUnit provides a (command line and built into Visual Studio), to define the classes and functions that carry the testing code, that test the actual values.
You can learn more about NUnit at their or read one of the interesting blog posts:
Download the following files from :
BaseTest helper class
Using of BaseTest class
Tests (see ). The method WaitForText() is used to
tests (see ). The method WaitUntil() is used to
KitchenSink tests (see ). This test mixes the other examples
Clone from the on GitHub.
Follow the steps that were presented at Install Selenium Standalone Server and browser drivers.
The sample app includes a Selenium test case for every UI pattern that is presented in that app. You can learn from the test project (in the test directory), how to achieve Selenium testing of particular actions, such as button clicks, page changing, typing in forms, etc.
