Semantic Patches for Java Program Transformation (Artifact)

Hong Jin Kang, Ferdian Thung, Julia Lawall, Gilles Muller, Lingxiao Jiang, David LoDagstuhl Artifacts Ser 2019 · OpenReview

TL;DR

This paper introduces Coccinelle4J, an extension of the powerful C code transformation tool Coccinelle, enabling semantic patch application to Java. This artifact provides a reproducible environment and examples for applying these transformations, demonstrating its utility in tasks like updating deprecated Android API calls.

What Problem Does This Paper Address?

Maintaining large software projects often requires making the same change across numerous files and locations. Tools like Coccinelle have been highly successful in automating such tasks for C code, particularly within the Linux kernel. However, a similar robust and widely adopted tool for applying semantic transformations to Java code has been lacking. This gap makes it challenging and time-consuming to perform large-scale, context-aware code refactorings and updates in Java projects.

Methods

The core contribution is the extension of the Coccinelle framework to support Java. This involves adapting Coccinelle's semantic patching language, SmPL, to parse and transform Java source code. The artifact itself is a Docker image that encapsulates the Coccinelle4J tool, its dependencies, and pre-configured examples. This approach ensures a consistent and reproducible environment for users to experiment with the tool and its capabilities, including a case study on migrating deprecated Android API usage.

Datasets

Not discussed in the paper. The paper mentions using "source code extracted from real-world Java projects" and specifically a case study involving "deprecated Android API methods," but does not detail specific datasets or benchmarks.

Key Insights and Ideas

  • Bridging the Gap for Java: The success of semantic patching in C (via Coccinelle) highlights a significant need for similar capabilities in the Java ecosystem. Coccinelle4J aims to fill this void.
  • Reproducibility is Key: Providing a Docker image is a crucial step for making research artifacts usable and verifiable. It abstracts away complex setup issues and ensures that others can easily run the provided examples.
  • Practical Application: The case study on updating deprecated Android APIs demonstrates a concrete, real-world problem that Coccinelle4J can effectively address, showcasing its practical value beyond theoretical exploration.

Key Results

The paper's primary result is the availability of the Coccinelle4J artifact, which includes:

  • A Docker image (coccinelle4j/coccinelle4j:ecoop) for easy setup and execution.
  • The source code for Coccinelle4J.
  • A document detailing setup and usage instructions.
  • Examples of semantic patches written in SmPL for Java, along with corresponding Java source files and their transformation outputs.
  • The artifact is available under GNU GPL version 2. The zip file size is 101.1MB, and the Docker image is approximately 3.5GB.

Limitations and Future Work

The paper focuses on presenting the artifact and its setup. While it mentions a case study, detailed quantitative results on the effectiveness, performance, or scalability of Coccinelle4J across a broader range of Java projects are not presented. Future work would likely involve further development of SmPL for Java, expanding the library of semantic patches for common Java refactoring tasks, and conducting more extensive evaluations on diverse Java codebases.

Summary updated

Abstract

The program transformation tool Coccinelle is designed for making changes that is required in many locations within a software project. It has been shown to be useful for C code and has been been adopted for use in the Linux kernel by many developers. Over 6000 commits mentioning the use of Coccinelle have been made in the Linux kernel. Our artifact, Coccinelle4J, is an extension to Coccinelle in order for it to apply program transformations to Java source code. This artifact accompanies our experience report "Semantic Patches for Java Program Transformation", in which we show a case study of applying code transformations to upgrade usage of deprecated Android API methods to replacement API methods.
BibTeX
@inproceedings{kang2019semantic,
  author = {Hong Jin Kang and Ferdian Thung and Julia Lawall and Gilles Muller and Lingxiao Jiang and David Lo},
  title = {{Semantic Patches for Java Program Transformation (Artifact)}},
  booktitle = {Dagstuhl Artifacts Ser},
  year = {2019},
  url = {https://openreview.net/forum?id=MtJFJkvHAS},
}
Figure