the unreasonable effectiveness of suffix trees

I needed to find the longest substrings of two separate pages, for some clustering we’re doing on meanpath. Sadly, there seemed to be no package for it on Hackage, so I decided to contribute one.

There may be cleverer ways of doing this, and I’d welcome pull requests showing better solutions. For now, benchmarks are showing it running in 1.2ms for ~200 character strings (where the naive version takes 800ms), 14ms for 2000 characters, and 160ms for 20000 characters, which (extremely unscientifically) looks roughly O(n log n).

Source on github, package is on Hackage. Thanks to Bryan O’Sullivan for the excellent suffix tree library.