Back in 2008 I wrote some articles about the proposed Selectors API for the DOM. Now, in 2009, it’s still “only” a working draft of the W3C. I’ve been exploring this topic for a long time, now I will finally have time to write about what I’ve been working on for several months.
Jacqueline is a library which will help you to find unused CSS Selectors in multiple HTML documents. This is a very important feature when you want to check a complete website: it will be almost certain that you have some CSS selectors used on page “A” but not on page “B” whereas “B” uses some CSS selectors which “A” doesn’t use.
Jacqueline is completely written in Java – this was necessary because we wanted to run it in our Java based infrastructure. This brings some other problems Jacqueline has to face: the DOM implementation in Java doesn’t feature the Selectors API. So I wrote a CSS-to-XPath engine – inspired by DOMAssistant. What this basically does is: it transforms CSS2.1 selectors into XPath queries which can then be used for node tests (since XPath support is included in the JDK) on the DOM tree of the HTML document.
Because our documents are still using HTML 4.0 (due to several problems when switching to XHTML 1.0) I had to use JTidy for building DOMs.
Jacqueline also uses Apache Batik to retrieve all the CSS rules and CSS selectors from a document – it was the only implementation of the SAC (Simple API for CSS) I got to run – and which had some special features. These include locator information of the found tokens for example.
At the moment Jacqueline only works single-threaded, but because this includes many cpu-intensive operations which could definitely be fastened by using the power of multiple cores I’m also dreaming of a multi-threaded version. Since most operations of Jacqueline could be executed in a parallel fashion this could bring a major speed-up.
The main thing of Jacqueline is that it’s one of the first pieces to the lately mentioned Project Mess Tool. We’ll be using the power of Jacqueline to detect unused CSS selectors in our web projects.
Finally, that stupid name: Jacqueline is named after Jacqueline Bouvier from my favorite TV show the Simpsons. See this poster in our office:
Jacqueline - Poster