[dtunkelang]

Actually, my original implementation used Dijkstra's algorithm. But I think the linear table of suffixes (or prefixes if you prefer) is a cleaner approach.