You should watch your traffic with Fiddler in one of the blocking scenarios to see how they've implemented blocking. My guess
would be that the blocked resource results in a HTTP/302 redirect to the blocking page. If so, then yes, you could easily write a rule in Fiddler that captures the original URL for any response that is a 302 with a Location
header pointed at the blocking HTML page. Those captured URLs could then be written to a text file or whatever.
(Having said that, don't most content filtering proxies like websense themselves capture these URL requests and log them somewhere before returning the block to the client?)