Leveraging Existing Tests in Automated Test Generation for Web Applications
Authors: Amin Milani Fard, Mehdi Mirzaaghaei, Ali Mesbah
Note : This data is only summery of above paper.comment for any suggestion
We want to automate testing for web application because today software application is written as a WWW based application to be hunted down on internet browsers. The usefulness of these testing applications is varied from company to company. Thither are many advantage to test automation. One of related to reputability of the trials and execution speed of test. Mostly web application’s test based on a crawler to designate the active states of the application, this approach is automated and smooth, and simply there is lack of field knowledge required for its winner. Many developers currently write test cases in frameworks like Selenium, but there is also disadvantage for that this way needs much manual effort.Crawling base techniques automate the testing in a great direction, but they are defined in three fields. 1-Input values, good values are essential for proper reporting of test space of the diligence 2-paths to explore, market web apps have a big space volume, and it is difficult to handle every section of the application. Not knowing which routes are important to explore the results in obtaining a partial reporting of a specific part of the application 3- Assertion, any generated test case needs to affirm the application behavior.
Notwithstanding the automatic generating assertion without human knowledge is experienced to be challenging, as a result many web applications used standard rule to avoid these troubles.In this paper author proposed that mine the human knowledge standing in yourself written test cases, and combine that incidental information with the power of automated crawling and cover the test suit for the unchecked part of web applications beneath test. Author present its technique and tool called Testilizer, Testilizer used a Selenium test case Tc and the Url of the application, that automatically supposes a model from Tc and feeds that model to a crawler to enlarge by discovering exposed paths and states, makes assertions for newly noticed states based on the pattern learn from Tc, and finally generates new test instances.
Testilizer can generate test cases, however, failure may be happening due to requirements like efficiency, effectiveness and threats to validity. That work is interesting by the fact that human written test cases is appreciated source of domain information, which can be utilized to accommodate some of the challenges in automated web application test generation.The Dom state is fundamentally an abstracted version of the Dom tree of a web application that displayed in a web browser at runtime. This abstraction conducted through classification function, the performance of which is discussed in the report.Our tool utilizes the given test suite to generate effective test cases by exploring alternative paths of the application and regenerate assertions for new detected stats. For upcoming work, we plot to evaluate the success of other states space exploring approaches like a variation of test paths and analyze the relationship between the strength of the original test suite and generated test suite.
...................................................................................................................................................................
0 comments:
Post a Comment