Repairing syntax errors in LR parsers

Abstract

This article reports on an error-repair algorithm for LR parsers. It locally inserts, deletes or shifts symbols at the positions where errors are detected, thus modifying the right context in order to resume parsing on a valid piece of input. This method improves on others in that it does not require the user to provide additional information about the repair process, it does not require precalculation of auxiliary tables, and it can be easily integrated into existing LR parser generators. A Yacc-based implementation is presented along with some experimental results and comparisons with other well-known methods.

Syntax errors just aren't natural: improving error reporting with language models

“…Many modifications for particular parser algorithms have also been proposed to attempt to suppress spurious parse errors by repairing or resuming the parse after an error. Recent examples can be found by Kim et al [13] who apply the k-nearest neighbour algorithm to search for repairs, or Corchuelo et al [3] who present a modification that can be applied to parser generators and does not require user interaction. Other researchers have focused on type-based static analysis such as Heeren's Ph.D. thesis [6] which suggests implementing a constraint-based framework inside the compiler.…”

“…For example, the combination of this method with token prediction could automatically repair source code by statistically directing most-likely syntax repair searches based on a dynamic and project-specific corpus instead of statically defined least-cost repair searches such as those presented in Corchuelo et al [3] and Kim et al [13]. This approach should be much more efficient than the approach applied in Weimer et al [20].…”

“…Previous methods tend to rely on rules and heuristics [3,6,8]. In contrast to those methods, we seek to improve on error location reporting by exploiting recent research on the regularity and naturalness of written source code [7].…”

Natural and Flexible Error Recovery for Generated Modular Language Environments

“…Following these criteria, an excellent recovery is one that is exactly the same as the intended program, a good recovery is one that results in a reasonable program without spurious or missed errors, and a poor recovery is a recovery that introduces spurious errors or involves excessive token deletion. The Pennello and DeRemer criteria represent the state of the art evaluation method for syntactic error recovery applied in, amongst others, [Pennello and DeRemer 1978;Pai and Kieburtz 1980;Degano and Priami 1995;Corchuelo et al 2002].…”

“…The problem of handling syntax errors during parsing has been widely studied [Lévy 1971;Mauney and Fischer 1988;Pai and Kieburtz 1980;Barnard and Holt 1982;Tai 1978;Fischer et al 1980;Degano and Priami 1995;McKenzie et al 1995;Corchuelo et al 2002]. We focus on LR parsing for which there are several different error recovery techniques [Degano and Priami 1995].…”