Tool Support for Refactoring Duplicated OO Code

Code duplication is an important problem in application maintenance. Tools exist that support code duplication detection. However, few of them propose a solution for the problem, i.e. refactorings. We propose an approach that uses the information given by code duplication detection to guide the refactorings of OO applications. 1 Code Duplication and Refactorings: Two Poles of a Magnet Code duplication is a well-known problem in software maintenance, bloating up source code and complicating the extermination of errors. Tool support exists that helps to identify the duplicated code [JP93, Bak92, BYM+98, DRD99]. Refactorings are a way to transform an application [OJ92] in a behavior preserving manner. Refactoring tools [RBJ97] support the elementary operations to transform code. Duplication is one important aspect of code where refactorings can be used to improve the source code. Very few tools, however, support corrections of systems plagued by duplicated code. In [Bak97, BYM+98], macro extraction from source code (C and C++) were proposed, but these approaches did not tackle the intricate problems of refactorings in an object-oriented context. In this paper, we propose to use a duplication detection tool to guide the refactoring process. We want to stress the fact that our proposition is not to fully automate the process but to help the reengineer by doing a first analysis of the situation and providing him with a solution if possible. This paper presents first an overview over the ingredients that we need for our idea and then gives a short overview how duplication in object-oriented code can be categorized and tackled. 2 Combining Duplication Information and a Static Object-Oriented Model The first ingredient for our approach is obviously a tool to detect duplication in source code. Secondly, the approach is based on the ability to represent a program in terms of object-oriented entities, like for example classes, methods and attributes (see section 2.2 To represent the program in such a model is crucial since the possible refactorings that can be carried out on duplication found in object-oriented code depend on the structural relationships of the program entities the duplication was found in. In this section, we first detail the two ingredients, then propose a step by step recipe which combines them for the goal of guiding refactorings. 2.1 Identifying Duplication DUPLOC is a tool for detecting duplicated code [RD98, DRD99]. DUPLOC has been developed in the context of the FAMOOS Esprit project and features: