I’ll use this code as a basis, and also add a HTML parser that can parse Selenium HTML tests and invoke the com.thoughtworks.selenium.Selenium API. In practive the instance will be be.denali.test.SeleniumHTMLUnit.

For the HTML parsing I might base some of the code on http://code.google.com/p/selenium-htmlunit-adapter/

In order to be able to commit my work, I might put this up on a Git repository – probably at GitHub. Send me an email if you want to discuss this further.

Whoa – before doing that – consider that PushToTest already has a Selenese parser in it, you might be able to just use PushToTest out of the box – we don’t write our Selenium tests in Selenese so I had to dig into the source to get the part I needed – you might find PushToTest an even easier solution.

HTMLUnit 2.3 offers some utility APIs to handle the missing JavaScript files in the WebClient API. We find these all the time in our customer’s sites. We added the following commands to TestMaker. These are in the 5.3 code base:

• loglevel = “off”, “info”, “debug” controls the Selenium HTMLUnit logging
• throwExceptionOnFailingStatusCode, when set to true HTMLUnit will ignore errors when the source of an external script cannot be loaded
• throwExceptionOnScriptError, when set to true HTMLUnit will ignore JavaScript? execution errors

-Frank

Thanks Frank – now I have to do even less to improve the abstraction layer :)