Local Independence Transformation and Its Application to Removing Nonduplicate Statements in Code Clones

This paper proposes a new program transformation technique and describes its application to removing nonduplicate statements in code clones while preserving the semantics of the program. Allowing nonduplicate statements in code clones helps in finding larger code clones. However, the nonduplicate statements in code clones may induce unexpected difficulty and complexity when we analyze the software systems. In this paper, we propose a new technique of program transformation, called local independence transformation. The transformation analyzes the definitions of variables, and replaces the uses of the variables with their definitions in terms of constants and nonlocal variables, when possible. With the proposed transformation, some of the nonduplicate statements in code clones may be transformed into locally independent statements and hence may be moved out of the code clones while preserving the semantics of the program. Based on the new technique, we develop a threestep framework for moving the nonduplicate statements out of code clones. In the experiments on the wget source code, our approach successfully moves more than 77.9% of the nonduplicate statements out of the code clones.