The tokenizer needs a heavy upgrade or full rewrite atleast. The parser is perfectly fine. I might push a commit if possible.