Large Scale Generation of Complex and Faulty PHP Test Cases

Abstract

Developing good test cases is an intellectually demanding and critical task, and it has a strong impact on the effectiveness and efficiency of the whole testing process. This paper presents an automated generator of test cases, which are designed to evaluate source code security analyzers. The generator produces PHP: Hypertext Preprocessor (PHP) programs with most common vulnerabilities embedded in various code complexities. It also produces programs without vulnerabilities to test for false positives. The generator is modular and extensible. We describe its internal design and how it works. The generated PHP test cases were added to the Software Assurance Reference Dataset (SARD) and will be used to assess the effectiveness of static analyzers. We conclude with the current state of the tool, its benefits and future work.