Automated Testing Of Web Applications Based On Heuristic Exploration Strtegy

Testing is an important mean to assure the quality of web applications.Automated testing of web applications can be performed with dynamic explortaion techniques,which try to exercise possible user interface elements and dynamically explore user interface state spaces of web applications.Despite a high automation level,most existing exploration techniques exhaustively explore state spaces of web applications,which tend to become mired in local or irrelevent states in a limited test resouce and make it difficult to test web applications effectively.In this paper,we propose a page element feature-guided exploration strategy,which adopts a machine learning algorithm to analyze clickable page elements that are capable of triggering main functionalities in web applications.After extracting thier spatial features(e.g.position and size)and content features(e.g.the number of images and links),we train and build a model for preidcting the importance of each clickable-page element.Given a limited amount of time,during exploration,our strategy estimates the importance of each clickable-page element,prefers to explore states of main functionalities in web applications.Finally,a subset of the state flow graph is constructed as a test model,with which test cases will be generated automatically.We have implemented a tool of testing web applications automatically called PefjaxTest,and conducted an empirical study on eight open source web applications.Experimental results show that PefjaxTest is capable of covering main functionalities more effectively in a limited amount of time,compared with other exploration strategies,it covers roughly 10%more states and 16.6%more edges than DFS,BFS and Feedex for main functionalities.Yet,it generates a smaller test model and less test cases than other tools without the loss of main functionality coverage.