I totally agree that munging files is a less than ideal solution. Not an insult to S_Shrum but when I saw that he was working with flat files and planning on searching flat files I geared the answer towards the question.

When possible, I think that indexing is best part of the publishing process. If you're working with the data files that generate a site, you're in a much better position to know what data is relevant and in what context.

Most of our clients have their data piped through us at one stage or another and it makes it much easier. For instance, if you have a global footer on every page that contains a keyword, you probably don't want to return every page back in the results. To try and detect things like that looking from the outside would be pretty difficult.

-Lee

"To be civilized is to deny one's nature."