Comment by larodi

In place replacing the text while parsing is very powerful approach. Fingers crosses this flies.

This, combined with probabilistic approach can be even more interesting. Probabilistic approaches regexes exist since 70s if memory serves right.