python-ctokenizer
A c-extension module for python that tokenizes unicode strings
Notes
This is a fairly special-purpose tokenizer written in c. Requires input to be unicode. All characters are defined as alpha, numeric, or punctuation. Consecutive characters of the same type are kept together as a single token. There are several "special" token sequences, that if found, will be kept together (e.g '.' , to handle values like '23.4'; and "'" , keeping "don't" together as one word).
Future
Allow special sequences to be passed in Support regular strings as well as unicode