cleaning article titles for URLs

I'm taking a user-supplied article title, and creating a search-engine and user friendly URL for that article. Obviously the titles need to be cleaned up and well-formatted (so no, urlencode() is not an option). Currently I'm: - converting it all to lowercase - replacing spaces with underscores - removing ?,&,#,% - removing multiple underscores But what I'm concerned about is other special characters, accented characters, etc. - Are accented characters (eg cafè) allowed/safe in URLs? - Would it be more appropriate to convert them down to ASCII (è > e, ü

u, etc) -- and has any one got a routine to do this? - Or would it be more appropriate to remove them altogether?

--- Justin French http://indent.com.au