Research On Several Issues Of Legacy System Reengineering

Ever-changing technology and business requirements turn more and more existing software into legacy systems. Due to the high cost of creating a system from scratch or modifications, legacy systems still occupy a relative large proportion. Therefore, generating software through settling problems of legacy systems has considerably practical significance. Software reengineering refers to the procedure of understanding, refactoring and redeveloping legacy systems. It is a primary solution to resolve problems of legacy systems, which is composed of three phases including reverse engineering, refactoring and forward engineering. In this thesis, several issues of legacy system reengineering are investigated. The specific contents are as follows:A solution of legacy system reengineering which covers all the research issues is investigated. The solution comprises a process model and several components which support the process model, and is able to be extended to resolve other problems of legacy systems.In the phase of reverse engineering, to solve the problem of lack of consideration of pertinence and flexibility in software clustering, an agglomerative information bottleneck based weighted hierarchical clustering (ABWHC) approach is proposed. Targeted entities and features are selected, weight scheme is proposed to generate feature vectors, information loss is considered as the similarity metric, cluster labels are created for entities and clusters, and objective and subjective evaluation measures are respectively chosen and given. Compared with superior agglomerative software clustering approaches, ABWHC is more cohesive, requires less arbitrary decisions, is flexible to adjust the focus of software clustering, and may assist to generate a more accurate system partition.In the phase of refactoring, problems of access control in legacy systems are investigated. In the aspect of role engineering, an approach for legacy access control policy reorganization is proposed by involving the wide spectrum language and transformation theory of FermaT Maintenance Environement. Relevant concepts of access control policy reorganization and WSL based access control policy language are defined, transformation rules for access control policy reorganization are designed and analyzed, a transformation method is proposed and management rules for the reorganized access control policies are given. Case study demonstrates the feasibility of the proposed approach. Compared with other primary role mining approaches, the process of role mining remains the external behavior of access control policies unchanged. The constructed roles are ease to be evolved. The role hierarchy is complete and irredundant. The approach is with the lowest cost in the condition of constrcting necessary roles. In the aspect of access control policy integration, a GUI based access control policy and corresponding integration approach are proposed. Case study reveals the proposed approach is feasible. It is suitable for GUI software, and is easy to be controlled and evolved.In the phase of forward engineering, to solve the problems of high reusing cost and combination explosion of test cases, an ontology based GUI testing approach is proposed on the basis of the knowledge intensive and sensitive characteristics of GUI testing. GUI Ontology which reflects the objective knowledge of GUI tesing is constrcted. Based on the subjective knowledge of GUI testing, basic relations of GUI components are given, test sequence generation rules and test case generation method are proposed. Case study reveals the proposed approach is feasible. Compared with other primary GUI testing approaches, the constructed GUI Ontology which provides a formal GUI model may improve the reusability and maintainability of GUI testing. The experience of testers which is reserved as test sequence generation rules is able to simplify test cases, and raises the intelligency of GUI testing.