Understanding Web applications through dynamic analysis

The relevance and pervasiveness of Web applications as a vital part of modern enterprise systems has significantly increased in recent years. However, the lack of adequate documentation promotes the need for reverse engineering tools aiming at supporting Web application maintenance and evolution tasks. A nontrivial Web application is a complex artifact integrating technologies such as scripting languages, middleware, Web services, data warehouses and databases. The task to recover abstractions requires the adoption of dynamic analyses to complement the information gathered with static analyses. This paper presents an approach and a tool, named WANDA, that instruments Web applications and combines static and dynamic information to recover the as-is architecture and, in general, the UML documentation of the application itself. To this aim we propose an extension of the Conallen UML diagrams to account for detailed dynamic information. The tool has been implemented and tested on several Web applications. Its architecture has been conceived to allow easy customization and extension. The paper presents our tool in the context of a program understanding task; however, it can be usefully applied to many other tasks such as profiling, security and dependability verification and application restructuring.