I have a vague memory of someone posting an archive of (mini.)opera.com on HN, but I cannot find the post anymore.
Do you know of something similar?
The purpose is to use this large "database" to test HTTP parsers against some HTTP transactions recorded in real traffic.