Reverse Engineering by Visualizing and Querying

The automatic extraction of high-level structural information from code is important for both software maintenance and reuse. Instead of using specialpurpose tools, we explore the use of a general-purpose data visualization system called Hy+ for querying and visualizing information about object-oriented software systems. Hy+ supports visualization and visual querying of arbitrary graph-like databases. We store information about software systems in a database and use Hy+ for analyzing the source code and visualizing various relationships. In this paper we demonstrate the use of Hy+ for evaluating software metrics, verifying constraints, and identifying design patterns. Software metrics can be used to find components with low reusability or components that are hard to understand. Checking the source code against constraints can help bring design flaws to light, eliminate sources of errors, and guarantee consistent style. Identifying design patterns in a software system can reveal design decisions and facilitate understanding the code. We conclude that the flexibility achieved by using a generalpurpose system like Hy+ gives this approach advantages over special-purpose reverse-engineering tools, although specialized tools will have better performance and more knowledge of specific software engineering tasks. Combining the advantages of the two approaches is an interesting challenge.