Pyref: a refactoring detection tool for python projects

Refactoring, the process of improving internal code structure of a software system without altering its external behaviors, is widely applied during software development. Understanding how developers refactor source code can help us gain better understandings of the software development process and the relationship between various versions of software systems. Currently, many refactoring detection tools (e.g., REFACTORINGMINER and REF-FINDER) have been proposed and have received considerable attention. However, most of these tools focus on Java program, and are not able to detect refactorings applied throughout the history of a Python project, although the popularity of Python is rapidly magnifying. Developing a refactoring detection tool for Python projects would fill this gap and help extend the language boundary of the analysis in variant software engineering tasks. In this work, we present PYREF, a tool that automatically detect 11 different types of refactoring operations in Python projects. Our tool is inspired by REFACTORING MINER, the state-of-the-art refactoring detection tool for Java projects. With that said, while our core algorithms and heuristics largely inherit those of REFACTORING MINER, considerable changes are made due to the different language characteristics between Python and Java. PYREF is evaluated against oracles collected from various online resources. Meanwhile, we also examine the reliability of PYREF by manually inspecting the detected refactorings from real Python projects. Our results indicate that PYREF can achieve satisfactory precision. Our work not only presents a tool for researchers to conduct refactoring-related studies with Python projects, but also provides the first dataset of refactoring operations in the Python language.