Quex
A Mode Oriented Directly Coded Lexical Analyser Generator

Important: For benchmarking compile with option '-DNDEBUG'. Only then the generated engine runs at full speed!

Abstract:
Quex is a lexical analyzer generator that produces a directly coded lexical analyzer engine. Those engines are much faster than the table driven engines of the lex/flex family. For convenience, quex parses regular expressions in the traditional lex/flex style. This way switching from flex to quex is made very easy. In addition to the fast analyzer engine, quex provides many advanced features, such a 'lexer modes' that can be inherited and that provide events for mode transitions. Mode transitions can be allowed and disallowed, one can trigger on indentation events, and many parts of the generated lexical analyzer class can be adorned with own code. For easier handling of token sequences a fast token queue is implemented that allows to implement lexical analysis directly from sequence diagrams. Also, quex is based on a dedicated buffer handling strategy that is webbed into the lexical analyzer to provide optimal performance. The established tool for character code conversions 'inconv' can also be included into the buffer handling. This way many different kinds of coding schemes are supported. Quex produces a complete C++ environment containing the lexical analyser engine, token classes and token queues. With the provided sample applications and documentation it should be possible to get a lexical analyser running within a couple of hours. In essence, quex allows complicated lexical analysis to be described very elegantly. At the same time, simple applications can be written much faster than with traditional lexical analyser generators.

Dowloads:   
Documentation (HTML)
Downloads
Project on Sourceforge