Research On Automatic Generation Of Input Domain Test Data Based On Program Understanding

Since the domain of software application is expanded and the software complexity and scale are increased continuously, there raises much higher requirements for software testing. Software testing automation is an effective method of improving the efficiency of testing. While automatic test data generation is the key procedure for software testing automation, it has a significant precondition that is automatic determination of input domain. Therefore, it is of great theoretical meanings and broad application prospects to research on methods of automatic input domain determination and test data generation.A method of automatic input domain determination based on double constraint propagation and a method of automatic test data generation are proposed in this paper. On the basis of that, an automatic test data generation system is designed and implemented.This paper starts with obtaining constraint set through program understanding procedure by analyzing the source code, and then solves the constraint set with double constraint propagation method, so that the program path input domain can be found. Moreover, a method of program input domain presented by multi-interval is proposed in this thesis and the arithmetic laws of interval are expanded, which lay the foundation for automatic test data generation.Automatic test data generation is based on input domain mainly contains stratagem selection and test data generation procedures. It reads the program input domain and test data generation stratagems from command script and drives test data automatic generation through the orders from command script.The automatic test data generation system based on program understanding has two key procedures, which are automatic input domain determination and automatic test data generation. The experiment results show that the system is feasible to determine input domain and generate test data based on input domain automatically.