Sifter is a client and server-side library (via UMD) for textually searching arrays and hashes of objects by property – or multiple properties. It's designed specifically for autocomplete. The process is three-step: score, filter, sort. Seaching will provide back meta information and an "items" array that contains objects with the index (or key, if searching a hash) and a score that represents how good of a match the item was. Items that did not match will not be returned.