Java-oriented Design Patterns Detection Research And Tool Development

Design patterns detection is the processes of identifying the design pattern instances in existing source code.Use recoverd design patterns instances to reconstruct the documents of software sysytems helping us comprehend complex designs, optimize the code, improve the documentation and maintenance of existing systems. Many Java developers used GoF design patterns consciously and provided an abundance of data sources for design patterns detection.The purpose of this thesis is to study design patterns detection and to develop a Java oriented design patterns detection tool.By studying the theory of design patterns detection, we provided a design pattern abstract model and a Java oriented implementation model based on the methods of design pattern formalization. Some of the classic design patterns were selected as cases and Java programs became the target for detect. We designed an architecture of the JDPD tool which implemented design patterns detection based on static analysis.The JDPD tool consists of four modules covering information extraction, information organization, design patterns specification and matching, design patterns instances presentation in accordance with the canonical activities of reverse engineering. The JDPD extracts the information from AST of Java program and uses Java information organization model to denote the information. According to the solution parts of design patterns description by GoF, we built models to provide the precise specification of some design patterns,and wrote the algorithms of matching design patterns specification with program information to detect the instances of design patterns.The models of Program information and design patterns specification improve the effect of the matching algorithms,and the design of separating the design patterns presentation from the matching algorithms get the design patterns detection processes more normative and customizable.