ELECTRONICS, cilt.11, sa.17, 2022 (SCI-Expanded)
With the increase in the number of APIs and interconnected applications, API testing has become a critical part of the software testing process. Particularly considering the busines-scritical systems using API messages, the importance of repetitive API tests increases. Successfully performing repetitive manual API testing for a large number of test scenarios in large business enterprise applications becomes even more difficult due to the fact that human errors may prevent performing thousands of human-written tests with high precision every time. Furthermore, the existing API test automation tools used in the market cannot be integrated into all business domains due to their dependence on applications. These tools generally support web APIs over the HTTP protocol. Hence, this study is motivated by the fact that there is a lack of API message-driven regression testing frameworks in a particular area in which API messages are used in client-server communication. This study has been prepared to close the gap in a specific domain which uses business domain APIs, rather than HTTP, in client-server communication. We propose a novel approach based on the use of network packets for regression testing. We developed a proof-of-concept test automation tool implementing our approach and evaluated it in a financial domain. Unlike prior studies, our approach can provide the use of real data packets in software testing. The use of network packets increases the generalization of the framework. Overall, our study reports remarkable reuse capacity and makes a significant impact on a real-world business-critical system by reducing effort and increasing the automation level of API regression testing.