what i would do is to download the site and build a c program (or any other
language) that will scan the documents and write specific words to a text
file.