SSTF: A novel automated test generation framework using software semantics and syntax

Test Automation saves time and cost by digitizing the process of test generation and execution. The automated test generation techniques in the literature do not always produce effective and compilable test cases. A test generation framework is proposed in this paper which uses the information extracted from UML diagrams and source code. The three layer architecture of the framework is responsible for this generation task. The first layer processes the user inputs i.e the UMLs as XMLs and the source code as source classes, which are used by the second layer. This layer identifies the application semantic from XMLs and extracts syntax from source. It combines this extracted information together and generates unit and integration test cases. This incorporation of syntax and semantics should make the generated tests less erroneous as it creates a better understanding of the application before the test construction. These two directional information collection should also mitigate the negative effects of inconsistent UMLs or source code on the test suites. Moreover, the generation of both unit and integration test case may increase the test coverage. A case study, conducted on a sample Java project, assessed the framework competence and has been successful to construct test scripts.