Language specification by multi-axiom grammars

Abstract

Multiaxiom grammars and language, presented as generalizations of context-free grammars and languages, are defined and used as a mechanism for programming language specification and implementation. It is shown how to divide such a grammar into a sequence of subgrammars that generate inductively the language specified by the original grammar. Furthermore, it is shown how to use this sequence of subgrammars for inductive language recognition by a process of tokenizing. Two classes of algorithms for languages parsing based on multi-axiom grammars are developed: an algorithm obtained by generalizing context-free LR-parsers to multi-axiom grammars, and a pattern-matching algorithm that results from the ability to layer a multi-axiom language into levels such that each sublanguage is independent of the language that contains it. The implications of multi-axiom grammars for compiler code generation are briefly discussed.<>