Evolução da ferramenta RMT: uma proposta para integração de módulos externos

Abstract

The Refactoring and Measurement Tool (RMT) is a tool aimed at the analysis, refactoring, and quality measurement of Java source code, being used to identify candidates for the application of design patterns and to evaluate the effects of these transformations on software. In version 2.0,RMT still follows a predominantly deterministic flow, based on heuristic strategies for detecting and executing refactorings, without including a specific architecture for the future integration of Artificial Intelligence modules. Therefore, this work developed an architectural expansion of the tool, focusing on the decoupling of the discovery, selection, application, and measurement stages of refactorings. Stable contracts, persisted artifacts, and extension points were defined to enable communication with external services in a secure manner and decoupled from RMT’s main core. In this architecture, the tool remains responsible for orchestrating the flow, validating the received responses, deterministically applying the refactorings, and measuring quality attributes. An external module based on Artificial Intelligence was proposed to show how the refactoring process can be carried out. The communication foreseen between RMT and this module occurs through HTTP/JSON contracts, avoiding direct dependency between the components and preserving the separation between external recommendations and the tool’s deterministic guarantees. Thus, the proposed architecture establishes a secure and reproducible basis for future investigations into the use of Artificial Intelligence in the refactoring process, while keeping RMT responsible for the final decision and the analysis of results.