Circe: A grammar-based oracle for testing Cross-site scripting in web applications

Security is a crucial concern, especially for those applications, like web-based programs, that are constantly exposed to potentially malicious environments. Security testing aims at verifying the presence of security related defects. Security tests consist of two major parts, input values to run the application and the decision if the actual output matches the expected output, the latter is known as the “oracle”. In this paper, we present a process to build a security oracle for testing Cross-site scripting vulnerabilities in web applications. In the learning phase, we analyze web pages generated in safe conditions to learn a model of their syntactic structure. Then, in the testing phase, the model is used to classify new test cases either as “safe tests” or as “successful attacks”. This approach has been implemented in a tool, called Circe, and empirically assessed in classifying security test cases for two real world open source web applications.