Um modelo para análise da árvore de sintaxe abstrata como apoio à extração de fatos e a compreensão de software

Abstract

Maintain software systems is a complex task, particularly over time, in which teams of developers of varying levels of experience contribute to a constant evolution of software. When new features are developed, many issues appear and are fixed. But this evolutionary process various problems can be generated in the design and source code, which hinder the understanding of the implemented architecture. The Software Visualization (SV) is presented as an approach to Software Comprehension (SC), using reverse engineering principles to analyze artifacts that provide relevant information to the study of architecture. The goal of SV is to represent software through graphics that stimulate the visual and cognitive sense of the developers, providing a better understanding of the characteristics implicit in the source code. For this, several techniques are used to extract facts by analyzing the code Abstract Syntax Tree (AST). However, there is a very large deficit of SV tools that provide a more dynamic and interactive development process, without abandoning the scalability of the complex combined analysis and extraction of thousands of lines of code. The biggest problem is the high cost involved in the process of such complex analysis, which depending on the amount of analyzing code can load even gigabytes of information in memory. Based on this problem, we propose the construction of a fact extraction model called PF FactFinder, which performs the analysis of the source code to store data in a relational database. The model should reuse principles of many consolidated extraction models in the area, indicating the use of Structured Query Language (SQL) as an approach that allows all kinds of queries to the development of software analysis. Separating responsibilities among processes of extraction, data analysis and SV can enhance the processing performed on the source code, enabling and enhancing it with the amplitude of the combined analysis software with minimal impact on the scalability of the tool.