One year ago I was involved with Selenium as part of a web automated testing project. I discovered multiple limitations with the Selenium framework, mainly with its integration with existing CruiseControl architecture. Having some time lately, I explored the idea of Selenium4j, a framework translating Selenium HTML scripts to Java Junit tests. I hosted to project at

http://code.google.com/p/selenium4j/

Problems with Selenium IDE HTML tests

  • While Selenium IDE is great tool for creating HTML tests, the HTML scripts are not Java tests and as such are not usable outside of the Selenium IDE environment
  • Using the IDE to create the Java JUnit test works one way only – from HTML to Java. Re-using the java test within Selenium IDE proved to be unreliable and buggy
  • Using Selenium RC to invoke the HTML files is not practical solution as it allows only a limited configuration such as browser, server and a single suite. E.g.
java -jar selenium-server.jar -multiwindow -htmlSuite "*iexplore"
"https://www.website.com" "C:\suite.html" "C:\results.html"

Motivation for Selenium4j

In reality we would like to:
  • Reuse the Selenium HTML suites for regression testing
  • To be able to integrate them within an automated environment (think CruiseControl f.i.)
  • Make the Selenium HTML tests configurable

WebDriver (Selenium2) solution?

Looking into WebDriver there is no analog of the Selenium IDE. The IDE is an amazing tool for recording the user clicks and it should be great if WebDriver solve the HTML automation integration. It seems that the later version of the IDE (1.0.5 as of writing) provides better support for the new Selenium2 API. Yet, the problem with HTML test integration is still on the table. As long as you record your script and export it to HTML,  you are on your own. You need to manually use the IDE to translate the HTML scripts to the desired language (Java in my case), then you need to copy the source, paste it to you Eclipse/Netbeans, make sure compiles, configure it properly. In other words, it’s not fun.  WebDriver should provide a solution that simplifies that process and at the end the recorded Selenium IDE script should be seamlessly integrated within an automated environment. In that context, Selenium4j is an example how HTML could be easily translated to a Java JUnit sources.

How does Selenium4j work?

In short Selenium4j translates the HTML tests to Java JUnit tests. It iterates through the HTML suite and tests, digest them and smartly discovers their Selenium commands. Each command is being subsequently transformed to a Java JUnit method. At the end of the transformation the HTML scripts are being translated to Java JUnit sources following same directory(package) structure as the suite and tests. In addition, Selenium4j have suite setup/teardown utilities as well as external configuration.

Once again, for all the sources and ideas refer to http://code.google.com/p/selenium4j/

Have fun and lazy automated web regression testing!