Okay, then I'll just parse out everything that appeas after (what seems to be) "Content-Type: text/plain" followed by two CRLF pairs.

Does that sound right?