This directory contains an implementation of an html context scanner with no lookahead. Its purpose is to scan an html stream and provide context information at any point within the input stream. This is used in the auto escaping functionality of the templating system, which would require html context information at very specific points within the html stream. The implementation is based on a simplified state machine of HTML4.1. The implementation is written in C, with the idea of making is usable in other contexts besides this html paresr; however, we compile all the files as c++ files in order to be able to use a namespace (and avoid global namespace pollution). There's also a c++ shim to give an object-oriented look at the html parser state, which is what the ctemplate system actually uses.