The Research And Implementation Of Test Case Generation Based On UML Behavior Diagrams

As the core content of software testing, test cases are the important guarantee of software quality. UML model can be used to guide test procedure, check the consistency of model and code and verify the customer requirements. Reuse the models that designed in analyses and development phase would speed up the test case generation procedure. The UML based testing techniques can be classified into four major categories: specification based testing, graph theoretic testing, heuristic testing, and direct UML specification processing. Single UML model has been broadly discussed while few papers studied combination of variety of UML graph. In addition, these test case generation approaches has to consider the specific implementation, development process and have limited usefulness.The thesis proposed a two levels test scenario generation algorithm and can be used to guide system level test case generation. The algorithm is based on graph theory and combines the use case diagram and activity diagram. Firstly, the activity diagram is defined formally, and coverage rules are set down. Then, transfer the activity diagram into directed graph. The primary level test scenario generation is according to the branch backtracking depth-first traversal algorithm. The second level test scenario can be generated by sorting the concurrent modules and limiting the cycles of the loops. Three solutions: depth-first search, random filter and modified ant colony algorithm were analyzed to solve the concurrent modules. Thirdly, combine the two level test scenarios by the multiplication principle, and optimize the generated final test scenarios. Test data were added to produce executable test case. The key points of the algorithm are automation translate of the activity diagram and the solution of the explosion of paths in concurrent modules.The test case generation algorithm focuses on practical engineering applications and was applied to a real system. In practice, the generated test case are accuracy and modifiable and speed up the testing process.