Automatic Test-Case Generation

This chapter is an introduction to the theory, techniques, and tool support for automatic test-case generation. We discuss how test models can be generated, for instance, from requirements specifications, and present different criteria and strategies for generating and selecting test cases from these models. The TARGET tool is presented and used in this the chapter for illustrating test-case generation techniques along with a case study in the domain of mobile-phone applications. TARGET generates abstract test cases from use-case specifications presented as templates whose contents are described using a controlled natural language. Section 2 presents a generic model-based testing (MBT) process model, briefly discusses models that are suitable for test generation, and introduces MBT in the mobile-phone applications domain. Section 3 presents models that are commonly used for test-case generation along with test-case generation strategies and algorithms. Section 4 presents different strategies for test-case selection from test models such as labelled transition systems. Test selection allows for a customised yet automatic choice of test cases among the possible ones to be generated according to a given coverage goal. The focus is on test-purpose selection, random selection and similarity-based selection. Section 5 presents a strategy for generating test models from use-case specifications. The specifications are formalised according to use-case templates that have been designed with two main goals: to be used by analysts for requirements documentation and also for test designers who can complement requirements with testing goals and concerns. The use-case specification is translated into a labelled transition system from which test cases can be generated. Section 6 presents the TARGET tool. This tool generates abstract test cases from use-case specifications with controlled natural language according to the templates presented in Section 5. TARGET has been developed for supporting feature and feature-interaction testing of mobile-phone applications. However, it can be largely applied for functional testing of reactive systems whose behavior can be expressed by case scenarios. The use of the TARGET tool is illustrated in a case study. Section 7 presents perspectives on the use MBT and test generation in practice and the obstacles yet to be faced.