Java quality assurance by detecting code smells

TLDR

Software inspection improves quality by examining code, design, and documentation for problematic patterns, and code smells serve as indicators of bad design that help identify refactoring opportunities. This study investigates automatic code quality assessment through code smell detection to support automated code inspection. The authors propose an automatic detection and visualization approach, demonstrated by the jCOSMO prototype that identifies and displays Java code smells for use in inspection tools. A case study applying jCOSMO confirmed its feasibility for detecting code smells in real Java projects.

Abstract

Software inspection is a known technique for improving software quality. It involves carefully examining the code, the design, and the documentation of software and checking these for aspects that are known to be potentially problematic based on past experience. Code smells are a metaphor to describe patterns that are generally associated with bad design and bad programming practices. Originally, code smells are used to find the places in software that could benefit from refactoring. In this paper we investigate how the quality of code can be automatically assessed by checking for the presence of code smells and how this approach can contribute to automatic code inspection. We present an approach for the automatic detection and visualization of code smells and discuss how this approach can be used in the design of a software inspection tool. We illustrate the feasibility of our approach with the development of jCOSMO, a prototype code smell browser that detects and visualizes code smells in JAVA source code. Finally, we show how this tool was applied in a case study.