Accurate Extraction of Bug Fix Pattern Occurrences using Abstract Syntax Tree Analysis

Matias Martinez (1) , Laurence Duchien (2, 3, 4) , Martin Monperrus (1, 2, 3)

1 SPIRALS - Self-adaptation for distributed services and large software systems
2 LIFL - Laboratoire d'Informatique Fondamentale de Lille
3 Université de Lille, Sciences et Technologies
4 Université Lille Nord de France (COMUE)

This manuscript presents an approach for studying the abundance of bug fix patterns. Bug fix patterns capture the knowledge on how to fix bugs, they are essential building blocks for research areas such as bug fix recommendation and automatic repair. In this paper, we focus on the problem of the accurate measurement of bug fix pattern abundance: how to reliably tell that one pattern is more common than another one? We propose an approach to formalizing bug fix patterns and an accurate instance pattern identification process that uses this formalization. Our technique is based on a tree differencing algorithm working with abstract syntax trees (AST). A comparative evaluation shows that our approach improves the accuracy of pattern instance identification by an order of magnitude.

Domaines: Génie logiciel [cs.SE]

Matias Martinez, Laurence Duchien, Martin Monperrus. Accurate Extraction of Bug Fix Pattern Occurrences using Abstract Syntax Tree Analysis. [Technical Report] hal-01075938, Inria. 2014. ⟨hal-01075938⟩