XRFix

This repository contains the generated code and script framework used to generate the results in the 'XRFix: Exploring Bug Repair of Extended Reality Applications with Large Language Models' manuscript.

The folders are organized as follows.

• ./Customized_codeql contains our specially designed queries for XR-related errors.
• ./Customized_UnityLint conatins our customized UnityLint.
• ./code_repair_framework contains our LLM-assisted code repair framework.

Environment Requirement

1. CodeQL-cli version: 2.16.3, official guidance
2. Python 3.12.3
3. Make sure your XR projects are compilable. Compile them in Unity Editor and keep all the files in the same project directory after compilation.

Usage

UnityLint for asset files

Please first enter the repository and use the following command line to analyse asset files. Windows

ShellStarter.exe -d <dirPath>

MacOS / Linux

mono ShellStarter.exe -d <dirPath>

where the Required Argument is:

-p, --project       Project directory.

Furthermore, you can use the following Optional Arguments:

-v, --verbose      Display Log on the standard output.

--help             Display this help screen.

--version          Display version information.

CodeQL for Unity files

Please download the CodeQL Github repository to your workspace consisting to your codeql-cli version. Then, add the queries in CodeQL repository. CodeQL Repository

Set-up

1. Add files in ./Customized_codeql/codeql-suites to ./codeql/csharp/ql/src/codeql-suites
2. Add files in ./Customized_codeql/suite-helpers to ./codeql/misc/suite-helpers
3. Add ./Customized_codeql/unityCheck to ./codeql/csharp/ql/src

Run the following command to perform static analysis

codeql database analyze <database> ./codeql/csharp/ql/src/codeql-suites/special_select_check.qls
codeql database analyze <database> ./codeql/csharp/ql/src/codeql-suites/unity-check.qls

code repair framework

Please see README.md files in ./code_repair_framework to see more details.

Detailed Information of our Collected Open-source XR Projects

Project_ID (a-w)	Project_Name	Source	LoC (C#)
a	VR-AR-Game	GitHub	651k
b	Passthrough-Neuroanatomy	SideQuest	188k
c	Unity-XR-NetworkMapper-Project	GitHub	182k
d	N.a.N Industry	SideQuest	154k
e	quest-jp	GitHub	145k
f	UAFinder	GitHub	100k
g	MRTK Samples with Passthrough API	SideQuest	99k
h	Surfaces in Passthrough	SideQuest	95k
i	BricksVR Rebuilt	SideQuest	84k
j	Extended-Reality-XR---Building-AR-VR-MR-Projects	GitHub	69k
k	Hand Posing Tool: Pirates Demo	SideQuest	74k
l	Piha rip simulator	SideQuest	46k
m	VR Gallery for Google Photos	SideQuest	45k
n	SurgeryQuest	SideQuest	42k
o	SoundStage	SideQuest	26k
p	6DOF-Mobile-VR-Using-GVR-and-ARCore	GitHub	23k
q	VRsus-guARdian	GitHub	23k
r	GoogleVR_Unity_Endless-Runner	GitHub	17k
s	VR-Archery	GitHub	14k
t	SimulariumXR	GitHub	3k
u	VR_ArcheryChallenge	GitHub	2k
v	IntVisWorkshop	GitHub	296
w	unity-playground	GitHub	148

Reference

In our paper, we have implemented and compared XRFix with AlphaRepair [1], Finetuned CodeT5 [2], Self_Repair[3]:

[1] Xia, Chunqiu Steven, and Lingming Zhang. "Less training, more repairing please: revisiting automated program repair via zero-shot learning." Proceedings of the 30th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 2022.

[2] Huang, Kai, et al. "Comprehensive Fine-Tuning Large Language Models of Code for Automated Program Repair." IEEE Transactions on Software Engineering (2025).

[3] Olausson, Theo X., et al. "Demystifying gpt self-repair for code generation." CoRR (2023).

Publication

This Repository was proposed in the paper accepted by ICSE'26. And we really appreciate your citation if this Repo helps you!