It’s apparently being worked on currently as a built-in feature of the product.
In particular, take note of the posts by Rick E, a Google employee working in that area. It’s worth reading back at least a few pages too (or all of it like I did), to get a better understanding of the discussion.
Otherwise I’d suggest looking at Fake, a really useful scriptable browser app. I’d say it’s a but more complicated to use than Automator, but I use it for some scheduled browser stuff that I would have to do manually.
My overall advice however would be to see what Google have coming unless you are desperate to dive in. I suspect it will be a higher quality output than anything that could be in effect snagged from the browser into a PDF.
Hope that helps.