Approach To Find Duplicate & Near-Duplicate Clones in High-Level Code

Code duplication is one of the most troublesome hurdles in software maintenance. Clone detection must be adopted as a pro-active measure in software development in order to avoid the problems faced by clones later in the phases like testing and maintenance.The main focus of this work is the detection of duplication in the source code which is one of the major reasons for poor structure in real programs. The novel approach of our clone- detection approach is the detection of near-miss clones at the file level as an aid to software maintenance. Near-miss clones are the ones that are not exactly akin to each other and may indulge changes in variable names, literals, whitespace, etc. but as far as the functionality is concerned, they yield a similar output. Our work is based on metrics and Abstract Syntax Tree-based approaches and it takes into consideration the Levenshtein distance. For experimental results, we have developed a tool based on the method proposed, and the results obtained show clearly that it effectively detects the near-miss clones at file-level means between different files.