Pattern-based refactoring of legacy software systems

Abstract

Rearchitecturing large software systems becomes more and more complex after years of development and a growing size of the code base. Nonetheless, a constant adaptation of software in production is needed to cope with new requirements. Thus, refactoring legacy code requires tool support to help developers performing this demanding task. Since the code base of legacy software systems is far beyond the size that developers can handle manually we present an approach to perform refactoring tasks automatically. In the pattern-based transformation the abstract syntax tree of a legacy software system is scanned for a particular software pattern. If the pattern is found it is automatically substituted by a target pattern. In particular, we focus on software refactorings to move methods or groups of methods and dependent member variables. The main objective of this refactoring is to reduce the number of dependencies within a software architecture which leads to a less coupled architecture. We demonstrate the effectiveness of our approach in a case study. © 2009 Springer Berlin Heidelberg.