Visual Integration And Application Of Code Smell Detection Tools

Code smell detection is a new research field of software engineering in recent years.Code smell is the program code that is introduced into the program and affects the software structure due to design defects or bad coding habits.The existence of code smell may have a bad impact on the scalability,stability and practicability of the software system.However,there are many existing code smell detection tools which are different in use and detection performance.They need users to spend a lot of time learning and reduce the efficiency.Therefore,it is necessary to design a system to integrate code smell detection tools in order to provide users with a unified way of use and improve the usability and efficiency of related tools.In order to facilitate users to use the code smell detection tools,this thesis designs an online code smell detection system,so that users can directly detect the project without downloading or installing the tools.This system integrates WekaNose,Checkstyle and PMD.Users can use the three tools to detect the code smell of the project uploaded by users and generate the detection results of the three tools for users to query and download.Moreover,the system designed in this thesis supports users to upload labeled dataset to compare the detection performance of the three tools under the measurement indicators of accuracy,recall,precision and F-score.In the system interface,users can view the comparison chart of the detection performance indicators of the three tools.However,because of the inconsistency of the detection results of various tools,it is not convenient for users to understand the results.Therefore,this thesis uses voting mechanism to design a consistency recommendation method of detection results which allow users to understand the detection results better.Finally,this thesis selects 10 software systems in Qualitas Corpus as the dataset to verify the effectiveness of the code smell detection integrated system.According to the visual results,we analyze the performance differences and consistency of three code smell detection tools for different code smell.In addition,we also analyze the performance differences between the recommendation method and the three existing tools and verify the effectiveness of the voting method proposed in this thesis.These results show that the online code smell detection system designed in this thesis has certain practicability and research significance,which provides users with convenient,intuitive and visual code smell detection function.