CS 31003 Compilers Difference between SLR and LR1

CS 31003: Compilers Ø Difference between SLR and LR(1) Ø Construction of LR(1) parsing table Ø LALR parser Bandi Sumanth 11 CS 30006 Date : 9/10/2013

Differences between SLR and LR(1) ●

SLR , Non SLR and LR(1) grammar ● SLR grammar ● A grammar is said to be an SLR grammar if the SLR parsing table for the grammar has no conflicting entries

SLR , Non SLR and LR(1) grammar ● A → B +. C Sj + A → B. + C A → B. Si

SLR , Non SLR and LR(1) grammar ● LR(1) grammar A → B +. C Sj + A → B. + C , = A → B. , * Si If the current state is Si and + is the next token, a shift is made leading to the state Sj Ø Ø Reduction cannot be made because the lookahead symbol * and the next token + does not match

Construction of LR(1) parsing table ●

Constructing LR(1) parsing table ● Consider the following augmented grammar Ś → S S → C C C → c C | d

LR(1) sets of items

LR(1) parsing table for the grammar

LR(1) vs SLR parser ● LR(1) parser is better than SLR parser because LR(1) accepts more languages than SLR ● SLR(1) is weak because it has no lookahead information ● But the cost is LR(1) has more states than SLR and thus it consumes more space ● LR(1) is impractical because its lookahead information makes the automaton too big

Can we retain the LR(1) automaton's lookahead information without all its states?

Review of LR(1) ● Each state in an LR(1) automaton is a combination of an LR(0) state and lookahead information. ● Two LR(1) items have the same core if they are identical except for lookahead.

A Surprisingly Powerful Idea ● In an LR(1) automaton, we have multiple states with the same core but different Lookahead ● What if we merge all these states together? Ø This is called LALR parser Lookahead LR parser

Advantages of LALR(1) ● Maintains lookahead information ● Keeps automaton small ● Resulting automaton has same size as LR(0) automaton

The End