Research Method Of White-box Testing

This thesis is set focus on the white-box testing method in the software testing and the problems in the current white-box testing is analysed such as huge testing data, uncompleted testing, and low efficiency. To solve the problems, the conception of DD graph and procedure tree is introduced detailedly and the method model based on procedure tree is presented. The model based on procedure tree describes white-box testing in five steps. In the first step, slice of program codes is selected with analysing the important degrees of programs and slicing the program. If in the integrated testing, it will be necessary to analysing the important degrees of programs to select the most important program to test. If in the unit testing, the only program is the most important program. Then the complicated program is sliced with program slicing technology. In the second step, the program is transformed into the procedure graph. In the third step, free edges assemble in the DD graph is analogized to simplify testing paths. In the fourth step, the procedure tree is created according to logical meaning in the procedure graph. After that, the circle paths in the procedure graph are transformed into branches in the procedure tree. Each path whose starting point is the root and terminal point is a leaf in the tree, is a testing path. The whole program is covered by all the paths. In the fifth step, the best testing paths are selected according to covering more free edges. Furthermore, the conception is clarified from the white-box testing method based on procedure tree, and traditional white-box testing method. The theory is checked practically by software.