The Research Of GUI Test Cases Generation Based On Models

The extensive use of graphic user interface(GUI) in applications brings greatchallenge to today’s sotfware development and testing. GUI has a portion of60%intotal codes of a single application today. The quality of Graphical User Interfaces(GUIs) directly affects the reliability and usability of the entire sotfware system, so thetesting of GUI is an active area.When testing the graphical user interface, we generally adopt the method ofextracting model. Model-based algorithm is modeled as flow chart with nodesrepresented as events and the path represented as event sequences, and event sequencesof the path can be used to generate the test cases in GUI. The frequent models based onthe graphical user interface are Event Flow Graph (EFG) and Event Interaction Graph(EIG). Event lfow graph is a directed graph, to simulate that all event interactions maybe executed on the GUI, which is used directly for rapid testing. Event interactiongraph is improved on Event flow graph，and divides all events for a GUI programaccording to modality of its windows, and depicts the execution order of the eventsusing graphs. Both of the two kinds of models give a certain presentation of the eventsinteracting relationships. It is provided the GUI events with a specific division througha simple GUI instance and also improves the MX algorithm. Finally, according to thedeifnition of EIQ the paper proposes a new method to make the EFG converted intoEIG: GEIG algorithm. Some examples show the new method easier.Automated test case generation(ATCG) has become increasingly popular due toits potential to reduce testing cost and improve sotfware quality. A few researcherspresent a model-dirven technique to generate test cases, and the technique usesfeedback from execution of a ‘‘seed test suite," which is generated automatically usingEIG model. During its execution, the run-time effect of each GUI event on all otherevents pinpoints event-semantic interaction(ESI) relationship, which are used togenerate new test cases. In this paper, the deifnition of the ESI relationship is more specific according to the feedback of GUI state, and proposes an algorithm to new testcases:GLES algorithm. About "seed test suites", using the GEIG algorithm proposed inthis paper instead of MX alogrithm to obtain EIG models. Finally, it indicates the newalgorithm can be used for GUI and the obtained data is more accurate and moreefficient with the real cases in GUI.