The Research On GUI Testing Based On Weighted Events

Software testing is the key to guarantee software quality and reliability. Cost of testing often takes large percentage of the total project budget. With the widely using of graphical user interface (GUI) in today's software, GUI testing has gradually become an important part of software testing. However, the research of GUI testing is far from being extensively performed. Since GUI software has characteristics that are different from those of conventional software, techniques developed to test conventional software cannot be directly applied to GUI testing. More difficulties and challenges were faced to the work of GUI testing than those of conventional software testing. How to generate test case for GUI software is a crucial research subject involved in.Firstly, this thesis analyzes the theoretical bases and testing phases of GUI test. And then discusses the construction of GUI testing framework and the related research work in the each phase of GUI Testing.Aiming at the representation of GUI that were employed by all the other components of the GUI testing framework, a new Weighted Events-based GUI representation is presented in this thesis. It distinguishes the importance of GUI events by assigns weight values to them, then divides a classification of events and separates the events into different subsets. The weighted event-flow graph and integration tree of GUIs can be constructed followed above.With analyses of GUI test case generation based on AI planning, a new method of Weight-based automatic GUI test cases generation is proposed in this thesis, and the Weight-based GUI test flow is introduced also. It obtains the set of hierarchical abstracted planning operators from GUI weighted events, reduces the number of GUI event states. The method uses the AI planning to generate the set of planning operator steps for the given GUI goal state. Then it determines the order to generate test cases in different subsets according to the weight values of events when decomposes the planning operator steps. With more reasonable test cases generated, it makes the GUI testing more efficient. The results of experiment show that more errors are detected apply this new method under the identical number of test cases. The technique of Weighted Events-based GUI regression testing is discussed in the last of this thesis. It uses the selective retest strategy. Firstly, adjusts weight values of events according to test results and the modifications of GUI. Thenanalyzes the GUI modifications, repairs the part of original test cases and appends some new test cases if necessary. Both of them comprised the GUI regression test case suit.