While working on a web spider I encountered the following page that causes the read() call of a urllib2 response to fail. It uses 100% of the CPU and does not seem to ever return. I have this behaviour on Python 2.4.4, but several people on 2.5.1 have tried the code below and reported the same behaviour. 

By the way, the page it uses is a porn site, but please don't get hung up on that fact. This is a data processing issue, not a subject matter issue. 

This test case is attached as a file, but is also available at http://pastebin.com/d6f98618f . Please note that the user-agent masquerading is present to rule out any issues with the server returning different data to different clients; commenting out the line so Python sends the standard headers still results in the issue occuring.