Efficient full LR(I) parser generation

  • Authors:
  • D. Spector

  • Affiliations:
  • MS 10C-17, Prime Computer, Inc., 500 Old Connecticut Path, Framingham, Massachusetts

  • Venue:
  • ACM SIGPLAN Notices
  • Year:
  • 1988

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper proposes that full LR(1) parser generators are easier to use than the LALR(1) parser generators commonly in use, and that minimal-state full LR(1) tables are not much larger than LALR(1) tables. A method is presented for the automatic construction of minimal-state full LR(1) parser tables. The method is fast and accurate because it starts with the LR(0) table and resolves lookahead ambiguities as necessary by tracing the LR(0) table and splitting the minimum number of states, and because there is no need for generalized set manipulation.