[unlinkr]

> Maybe you can detect tags with regex, but maybe you shouldn't...

So what do you think would be a more appropriate choice for writing a tokenizer?