[titzer]

Not all parsers do a lexing step. Such ones are called "scannerless parsers" and can be much, much faster than parsers with a separate lexing step.