Last week, as part of my work with Threepress building HTML5-based ebook reading software, I had to debug a problem with inaccurate cache manifests. Like any good engineer, rather than spend three hours doing it by hand, I spent six hours building a tool to do it.
The result is Caveman, a Python tool to validate HTML5 cache manifests. It scrapes the HTML page you specify, finding resources, then compares them to the cache manifest and reports problems.
When run as a command-line tool, it writes problems to its standard output. When run as a module, it's encapsulated enough that you can define how it gets HTTP data, and where it writes its results.
It worked well for me, and it might also for you. Try it and let me know.