No abstract

Abstract-Web applications have become significantly more complex and have begun to be used in wide variety of areas including social networks, shopping, online banking and other safety critical systems. We present an approach for automated white-box test generation for web applications. Our approach is to convert problem of high branch coverage test suite generation into a reachability problem and to utilize existing software verification techniques to generate test data for each execution branch. Set of Abstract Syntax Trees (ASTs) is built that describes web application as a whole, both client-and serverside code, by analyzing JavaScript code, its AJAX server calls and callbacks. Constructed AST is converted into an C# function with similar behavior and a set of arguments that represent user inputs. Existing test automation tools are used to discover test data that covers all the possible execution branches in a C# function. Web application test cases are generated with the discovered values, Selenium toolkit is used to emulate user actions and to automatically run the program under test against test cases.

Extended Finite State Machines (EFSMs) are often used in model-based development and for modeling VHDL specifications. This paper proposes an approach for automated test generation for EFSM models. Design by contract approach is applied to formalize specification requirements. Genetic algorithm is proposed to find set of values that triggers given path in the EFSM and reveals inconsistensies with the specification.

No abstract

Abstract-Automata-based approach is often used for developing complex systems. Model Checking is commonly used to check conformance of the system against its specification. However, verification techniques don't allow checking the system in whole, as system consists of not only the model, but also control objects, which are not suitable for model checking. In this paper we propose an approach for testing of automata-based programs. We use EFSM and contracts to extend model with specification requirements and we demonstrate how genetic algorithms could be used to automate generation of tests to find faults in the system in whole.