Detecting ambiguity in programming language grammars

Abstract

Ambiguous Context Free Grammars (CFGs) are problematic for programming languages, as they allow inputs to be parsed in more than one way. In this paper, we introduce a simple non-deterministic search-based approach to ambiguity detection which non-exhaustively explores a grammar in breadth for ambiguity. We also introduce two new techniques for generating random grammars - Boltzmann sampling and grammar mutation - allowing us to test ambiguity detection tools on much larger corpuses than previously possible. Our experiments show that our breadth-based approach to ambiguity detection performs as well as, and generally better, than extant tools. © 2013 Springer International Publishing.