Cache anonymous HTTP requests (based on cookies)

I would like to add a HTTP (or WSGI) cache to my Python web application, for all requests which

• are either GET or HEAD
• do not have a cookie called "session_id" (known users get specialized, non-cacheable pages)
• do not have a cookie called "_flash" (this cookie is used to show the user some short messages like "article created successfully", even after a redirect)
• if there are cookies which start with "__utm" (created by Google Analytics) they should be ignored (not stored in the cache), but the site should still be cached

If additional common caching directives (like cache-control: private) or even ETAGS are taken into account it would be nice too, but that is currently not a requirement.

Preferable, I do not want to run any additional piece of software like Squid (but I will use it if there a good reasons to do so), so I will prefer a mod_apache based solution, or even some kind of memcached powered WSGI-Middleware; Other solutions are also welcome.

Many thanks,
Christoph

I'd recommend checking out varnish. It's going to be a lot faster than any WSGI-Middleware you write/find, and you should be able to do everything you want with it. Here's the tutorial for a taste of it's many features, options: http://www.varnish-cache.org/docs/trunk/tutorial/index.html