Canonical LR parser

A canonical LR parser (also called a LR(1) parser) is a type of bottom-up parsing algorithm used in computer science to analyze and process programming languages. It is based on the LR parsing technique, which stands for "left-to-right, rightmost derivation in reverse."

Formally, a canonical LR parser is an LR(k) parser for k=1, i.e. with a single lookahead terminal. The special attribute of this parser is that any LR(k) grammar with k>1 can be transformed into an LR(1) grammar. However, back-substitutions are required to reduce k and as back-substitutions increase, the grammar can quickly become large, repetitive and hard to understand. LR(k) can handle all deterministic context-free languages. In the past this LR(k) parser has been avoided because of its huge memory requirements in favor of less powerful alternatives such as the LALR and the LL(1) parser. Recently, however, a "minimal LR(1) parser" whose space requirements are close to LALR parsers, is being offered by several parser generators.

Like most parsers, the LR(1) parser is automatically generated by compiler-compilers like GNU Bison, MSTA, Menhir, HYACC, and LRSTAR.