Replication package for the paper: Harnessing Test Call Structures for Improved Fault Localization Effectiveness

This repository contains the replication package for the paper "Harnessing Test Call Structures for Improved Fault Localization Effectiveness". The package includes the necessary scripts, data, and instructions to reproduce the results presented in the study, focusing on the effectiveness of Spectrum-Based Fault Localization (SBFL) using the Barinel algorithm.

Description of Folders and Files:

algorithms/: Contains the SBFL algorithms implemented for this study. Note that only the Barinel algorithm was used in the presented results.

base/: Contains the scripts needed for calculating spectra for SBFL.

D4J/: Contains the Defects4J projects. These need to be unpacked using the appropriate scripts provided in this folder.

changeset.json: Contains metadata about changesets for the projects analyzed in this study.

main.py: The main script for calculating ranks and metrics for the selected projects and heuristics.

SBFL ranks.csv: Contains the SBFL ranks calculated from the analysis.

Requirements

To run the scripts, you will need:

Python 3.9

Defects4J: The D4J folder should contain unpacked Defects4J projects. Please ensure you have followed the instructions in the D4J folder to unpack the projects accordingly.

Required Python packages (install using pip): pip install -r requirements.txt

Prepare the Defects4J Projects:

Navigate to the D4J directory.

Run the provided scripts to unpack the necessary Defects4J projects.

Ensure that each project folder is structured properly to be used by the main.py script.

Running the Main Script

The main analysis script is main.py, which calculates the SBFL ranks using the Barinel algorithm based on the selected heuristics.

Usage

To run the main script:

python main.py

Parameters and Settings:

Projects and Ranges: The projects are defined in the projects list, with their respective bug ranges in the ranges list. The script iterates over these projects and bug IDs to calculate SBFL ranks.

Outputs:

The script outputs the ranks and coverage metrics directly to the console. Results can be redirected or saved as needed.

It uses the Ranks.RankContainer class to calculate and print the minimum suspiciousness ranks for the selected metrics.