Enhanced technique to identify higher level clones in software

Abstract

Code copy and reuse are themost common way of programming practice. Code duplication occurs in every software program. A function, a module, or a file is duplicated for various reasons. The copied part of the source code with or without modification is called a code clone. Several tools have been designed to detect duplicated code fragments. These simple code clones assists to identify the design level similarities. Recurring patterns of simple clones indicate the presence of design level similarities called higher level clones. In this work we describe a new technique using fingerprinting to find higher level clones in software. Initially the simple clones are found, and then using LSH, we compare the fingerprints to find recurring patterns of method level, file level, and directory level clones. Finally, experiments and results shows that the proposed method finds all higher level clones in the software.