I have several documents in Sanskrit, encoded in UTF-8, and I'm trying to develop a search engine for the contents. The thing is the queries would be in roman characters, without diacritics (for ease of use). Do I need to resort to some kind of internal mapping and database or is there another way to do this?