Deep Learning-Based Erlang Refactoring

This repository contains a proof-of-concept method for refactoring nonidiomatic Erlang code.

This is the implementation of the presented method in the following paper: Balázs Szalontai, Péter Bereczky and Dániel Horpácsi, "Deep Learning-Based Refactoring with Formally Verified Training Data", Infocommunications Journal, Special Issue on Applied Informatics, 2023, pp. 2-8, https://doi.org/10.36244/ICJ.2023.5.1

Repository Structure

Experiments were conducted on Google Colaboratory, thus all Python files are notebooks. To run the notebooks, make sure to adjust paths in notebooks, if necessary.

• Finder.ipynb: Implementation of the localizer.
• Idiomatizer.ipynb: Implementation of the idiomatizer.
• Prototype.ipynb: Refactoring approach, evaluations, and experiments.
• Finder_func.ipynb, Idiomatizer_func.ipynb, Common_functions.ipynb: Utility functions for localization and idiomatization.
• pprint.erl, tok.erl: Pretty printer and tokenizer implemented in Erlang.

Dataset

The two training datasets can be accessed on HuggingFace here. These csv files should be moved to src/dataset.