面向设计层次优化的软件自动化重构

摘要

目前许多研究人员对自动化软件重构进行了探索并开发了一系列重构工具,旨在帮助程序员更高效地完成软件重构任务、提升代码质量.然而,现有的软件重构工具多侧重于局部的设计或编码问题,而非设计层面的问题.另一方面,基于搜索的重构方法往往将改进某一项代码度量指标作为重构目标,而非面向软件的层次化设计.针对这种情况,提出一种新的基于搜索的软件自动化重构方法,该方法使用了基于设计结构矩阵(DSM)的软件模块层次化度量方法,能够自动生成可以得到最优软件模块化设计的重构建议.在此基础上,实现了自动化重构工具DSMRefactoring,并将DSMRefactoring应用于开源系统进行案例研究,初步验证了方法和工具的有效性.%At present,many researchers have explored automated software refactoring and developed a series of refactoring tools designed to help developers conduct refactoring tasks with more efficiency and improve the code quality accordingly.However,existing software refactoring tools mainly focus on improving the code quality from a local perspective instead of an overall design perspective.On the other hand,search-based refactoring approaches usually aim at improving some specific code metrics instead of modularized and layered software design.This paper proposes a novel search-based automatic software refactoring approach,which leverages DSM-based code metric to modularize code.This approach is able to generate refactoring suggestions to achieve an optimal modularized and layered software design.This paper also introduces a proof-of-concept tool,DSMRefactoring,and applies the tool on an open-source system.The results validate the effectiveness of both the approach and its proof-of-concept tool.