The article mentions screenshots. I am interested in running the browser as a rendering tool, but ouputing not just a bitmap, but something closer to PDF. The goal here would be to crawl a bunch of pages and spin up the CPU to run all the needed JavaScript to render things, but to obtain an identical version of the page that runs about as fast viewing a screenshot, but also includes text. An offline "web OCR", if you will.
NeWS, Display PostScript, Quartz all come to mind. I know you can save webpages as PDF documents, but I am thinking of something that is like PDF but closer to HTML-just more closely coupled to what is being displayed, and not whatever arbitrary style or JavaScript abuse the developer decided on.
Even just outputing some neat, canonical HTML based on the state of the page once everything is loaded would be helpful, so that the bitmap could later be combined in some kind of new document format.