The Research&Implementation Of Web Service Test Data Generation

Service Oriented Architecture (SOA) has become the major trends of web-based distributed systems. It provides a loosely coupled, standards-based and cross-platform paradigm to the distributed computing on the Internet. Web Service is a way to achieve SOA, It make use of the overall architecture of SOA to introduce a completely new web application development, deployment and integration mode. Web Service has become widely used nowadays for its property of cross-platform, distributed and easy integrated. Before integrate a web service to a system, the users o f the web service have to do a lot of testing works to make sure that the web service can achieve the expected quality because of the features of web service. To test a web service, one key premise is that how to generate a sufficient number of valid test data.In this paper, I have done a lot of research in-depth on recent techniques about web service test data generation. And then I discover that current techniques for the generation of web service test data are found to generate a lot of redundant test data. To solve this problem, I present an approach of constraint mutation to screen the test data. This approach first mutate the constraints in WSDL to generate a lot of constraint mutants, then it make each test data run the original constraints and constraint mutants to find out the ability of discovering faults of each test data. The approach screens the test data according to the ability of discovering faults of each test data. It screens those can discover more faults than others as the final test data. In this way we can eliminate a large number of redundant test data of web services.At last, by combining the current techniques about web service test data generation and the constraint mutation approach, I design and implement a web service test data automated generation tool. I use this tool to make some experiments and the results of the experiments show that the constraint mutation approach presented by this paper can achieve a significant reduction in redundant test data and eventually filter out adequate test data to test web service.