Hcdetector: A Hybrid Approach to Detect Code Clones in Java Programs

Abstract

In software programs, substantial amount of existing code has been reused by simply copying and pasting mechanism. This process is called software cloning which leads to introduction of similar codes called code clone in system. The impact of these clones are very severe and of great concern from maintenance and development point of view. Due to this, software code clone detection is an active research area and many code clone detection techniques and tools have been proposed to detect them based on the syntactic or semantic similarity existing between code fragments. These techniques are mainly categorized into five types. Among them text based, token based, tree based, and metrics based clone detection techniques are used for detecting syntactically similar code fragments whereas program dependence graph based technique is able to detect both syntactically and semantically similar code fragments. Moreover there are certain tools which combine these techniques in order to detect code clones more efficiently and are termed as hybrid code clone detection tools. This thesis presents such kind of hybrid code clone detection tool called HCDetector which combines PDG based and metrics based code clone detection technique in order to detect both structural as well as behavioral similar code fragments using java byte code. This tool is language dependent and works only on java programs.