A character-level n-gram tokenizer using mixed n sizes, for text files encoded in latin1 (work in progress) - View it on GitHub