Artifacts of the ESEC/FSE 2021 Submisssion Cognac

This repository contains the artifacts of Cognac, a paper under review by ESEC/FSE 2021.

Datasets: All datasets used in our study are open-sourced. We provide the links to each of them below.

Empirical dataset (here)

MNR task datasets: Java-small, Java-med, Java-large (here)

MNR task dataset: MNire's (here)

MCC task dataset (here)  To extract this dataset, we excuted the code from Liu et al. on a high-performance server for an extremely long tome. To facilitate future researches, we release our parsed data in RenamedMethods.zip. 

Source code: We release two source files in this repository.

dataextractor.py: This is a project-level call graph builder implemented in Python language. One can use it to analyze the call relations within specific projects. Furthermore, one can also use it to extract the input of Cognac, which is a sequence of tokens extracted from totally six sources as we have introduced in the paper.

model.py: This is the pointer-generator model of Cognac which is implemented in Pytorch.

We will build a homepage for Cognac on Github and release the whole project upon acceptance.