Faster reachability analysis for LR(1) parsers

Abstract

We present a novel algorithm for reachability in an LR(1) automaton. For each transition in the automaton, the problem is to determine under what conditions this transition can be taken, that is, which (minimal) input fragment and which lookahead symbol allow taking this transition. Our algorithm outperforms Pottier's algorithm (2016) by up to three orders of magnitude on real-world grammars. Among other applications, this vastly improves the scalability of Jeffery's error reporting technique (2003), where a mapping of (reachable) error states to messages must be created and maintained.