The final goal of my current project is to write a library that is able to parse a CSS file via Ajax. Wow! Nothing less? Actually, I'm still in beta version. I use a lot of regular expressions to extract comments, properties, values, selectors, etc. When I switch to the final version, I'll get rid of this approach to embrace a more practical, token-based, stream-based way of coding. My inspiration comes from the nsCSSScanner of Firefox (see it here).
jQuery has a lot of helping features. $.grep(), for example, allows you to filter an arrat based on certain criteria. Anyway, I need to use them more. Another source of inspiration is the JSON parser written by Douglas Crockford. That's beatiful. I'm going to publish the beta version soon (probably here). So... stay tuned!