基于分支函数线性逼近的测试数据自动生成算法

Algorithm for automated test data generation based on branch function linear approximation

软件测试是保证软件可靠性的一个重要手段。面向路径测试是软件测试中一种重要方法。提出了一种分支函数线性逼近的测试数据自动生成算法。结合赵瑞莲给出的谓词切片算法和程序DUC表达方式以及本文提出的算法,给出了一个基于程序执行的路径测试及测试数据自动生成新算法。由于算法采用DUC表达式,不仅可以从源端判断子路径是否可行,而且有效地降低了不可行路径对算法性能的影响。另外,与现有文献中单纯利用分支函数极小化方法的算法相比,新算法由于有机结合了分支函数线性逼近和极小化方法的长处,因此减少了测试用例的数量,提高了测试效率。

Software test is a key way to guarantee the software＇s reliability. Path oriented testing is a main approach in software test. A new innovative algorithm for automated test data generation based on branch function linear approximation was presented. A new algorithm for paths oriented tests of program execution and automated test data generation was given by combining the method of predicate slice of ZHAO Rui-lian and Definition Use Control （DUC） expression and the algorithm proposed by this paper. It can not only automatically find out all of the feasible paths from source in the program, but also decrease the impact of infeasible paths efficiently because of DUC expression of the algorithm. In addition, compared to those algorithms with minimization of branch function presented in some reference documentations, since the novel algorithm integrates advantages of linear approximation and the minimization of branch function, so it can reduce the quantity of test cases and improve the test efficiency.