The World Wide Web Consortium has a tool it calls a "Semantic Data Extractor." It basically parses a web page and shows you how a page would be outlined according to your <h1>, <h2> , <h3> ... structure. I found it earlier today when I was using their Markup Validation Service to test my markup. Their tip of the day referenced it. I found it extraordinarily useful in analyizing the structure portion of the content + structure + presentation Webstandards equation.