HTMLUnit has its own share of parsing gotchas.

I've used Selenium with Firefox and Xvfb to do headless scraping in the past; looks like my toolkit just got simpler.