The original pornolize.com is back up. It didn't bomb, I think it was shut down due to abuse. You will find a group, and one of the original creators on Facebook, if you wish to find out more about the history of the site.
A few months ago, while pornolize.com was down, I started work on my own site (
http://perversify.net/) which does essentially the same thing the original pornolize.com did, but is a bit more subtle in the output. There is also the Rinkworks Dialectizer (
http://www.rinkworks.com/dialect/), which simply does a search and replace (very well though).
Although I still have work to do in refining the handling of words which are both nouns and other types, I don't think the tricky part is the dictionary work or language parsing, but the HTML parsing. I expected language parsing to need a lot of ongoing work, so the code was designed to facilitate looking forward and rewinding the stream.
My language of choice was C, chosen for speed and in-memory data storage efficiency, but any language should do. Now that it's done, I can see CPU usage is so low as to be insignificant on the host. Memory usage is high because the database is memory resident. I looked in to file-backed access to the dictionaries, but it was significantly slower, and I expect unworkable in a hosted environment with a SAN back-end.
.