pkuzqh/icse23repair

An implementation of Tare.

ICSE2023Repair

A PyTorch Implementation of "Tare: Type-Aware Neural Program Repair".

Introduction

Automated program repair (APR) aims to reduce the effort for software development. With the development of deep learning, lots of DL-based APR approaches have been proposed using an encoder-decoder architecture. Despite the promising performance, these models share one same limitation: generating lots of untypable patches. The main reason for this phenomenon is that the existing models do not consider the constraints of code captured by a set of typing rules.

In this paper, we propose, Tare, a type-aware model for neural program repair to learn the typing rules. To encode an individual typing rule, we introduces three novel components: (1) a novel type of grammars, T-Grammar that integrates the type information into a standard grammar, (2) a novel representation of code, T-Graph that integrates the key information needed for type checking in an AST, and (3) a novel type-aware neural program repair approach, Tare that encodes the T-Graph and generates the patches guided by T-Grammar.

The experiment was conducted on three benchmarks, 393 bugs from Defects4J v1.2, 444 additional bugs from Defects4J v2.0, and 40 bugs from QuixBugs. Our results show that Tare repairs 64, 32, and 27 bugs on these benchmarks respectively and outperforms the existing APR approaches on all benchmarks. The further analysis also shows that Tare tends to generate more compilable patches than the existing DL-based APR approaches with the typing rule information.

The Main File Tree

.
├── generation: code for patch generation
│   ├── bugs-QuixBugs
│ 	├── location
│ 	├── location2
│ 	└── tesetDefect4j.py
├── model: code for neural model
│   ├── data
│   ├── Model.py
│   └── relAttention.py
├── train: code for training the model
│   ├── run.py
│   ├── Dataset.py
│   └── SearchNode.py
├── validation: code for patch validation
│   ├── patches-all: plausible patches
│   ├── Dataset.py
│   └── SearchNode.py

Dataset

Train set

The raw data https://drive.google.com/drive/folders/1ECNX98qj9FMdRT2MXOUY6aQ6-sNT0b_a?usp=sharing from Recoder.

Test set

Defects4J

QuixBugs

Updated Result

❗❗❗ We test Tare with the perfect fault localization on Defects4J 2.0.

The left of table below is copied the original Table V in the paper, showing the results using Ochiai for fault localization, where "x/y" indicates the approach correctly repairs x bugs and produces plausible patches for y bugs (a plausible patch passes all the tests). The last column is newly added, showing the number of bugs Tare correctly repairs with perfect fault localization. Only the first plausible patch is considered in the experiment.

Project	Bugs	TBar	SimFix	Recoder	RewardRepair	Tare	Tare+PerfectLocation
	Original Result	Original Result	Original Result	Original Result	Original Result	Original Result	Updated Result
Cli	39	1/7	0/4	3/3	2/-	5/13	8
Closure	43	0/5	1/5	0/7	1/-	0/5	1
JacksonDatabind	112	0/0	0/0	0/0	3/-	0/4	0
Codec	18	2/6	0/2	2/2	3/-	3/7	5
Collections	4	0/1	0/1	0/0	0/-	0/0	0
Compress	47	1/13	0/6	3/9	0/-	4/13	4
Csv	16	0/2	1/5	4/4	2/-	5/7	5
JacksonCore	26	0/6	0/0	0/4	1/-	2/7	14
Jsoup	93	3/7	1/5	7/13	4/-	10/16	14
JxPath	22	0/0	0/0	0/4	3/-	2/10	3
Gson	18	0/0	0/0	0/0	1/-	1/1	1
JacksonXml	6	0/0	0/0	0/0	0/-	0/1	0
Total	444	8/50	2/25	19/46	20/-	32/84	55

Usage

Train a New Model

CUDA_VISIBLE_DEVICES=0,1 python3 train/run.py train

The saved model is checkModel/best_model.cpkt.

Test the Model

Generate Patches for Defects4J v1.2 with Ochiai by

CUDA_VISIBLE_DEVICES=0 python3 generation/testDefect4j.py bugid

The generated patches are in folder generation/patch-all/patch/ in json.

Generate Patches for Defects4J v1.2 with groundtruth by

CUDA_VISIBLE_DEVICES=0 python3 generation/testDefect4j1.py bugid

The generated patches are in folder generation/patch-all/patchground/ in json.

Generate Patches for Defects4J v2.0 by

CUDA_VISIBLE_DEVICES=0 python3 generation/testDefects4j2.py bugid

The generated patches are in folder generation/patch-all/patch2 in json.

Generate Patches for QuixBugs by

CUDA_VISIBLE_DEVICES=0 python3 generation/testQuixbug.py bugid

The generated patches are in folder generation/patch-all/patchQuix in json.

Validate Patches for Defects4J v1.2 with Ochiai

python3 validation/repair.py bugid

The results are in folder validation/patch-all/patches/ in json.

Validate Patches for Defects4J v1.2 with groundtruth

python3 validation/repair1.py bugid

The results are in folder validation/patch-all/patchesground/ in json.

Validate Patches for Defects4J v2.0 with Ochiai

python3 validation/repair2.py bugid

The results are in folder validation/patch-all/patches2/ in json.

Validate Patches for QuixBugs with Ochiai

python3 validation/repairqfix.py bugid

The results are in folder validation/patch-all/patchesqfix/ in json.

Gnerated Patches

The generated patches are in the folder patches-all.

Dependency

• Python 3.7
• PyTorch 1.3
• Defects4J
• Java 8
• docker
• nvidia-docker

icse23repair's Issues

This project's rule.pkl.is different from rule.pkl in Recoder

I meet this program:
assert('start -> copyword2' in self.ruledict)
AssertionError
and I check the rule in run.pkl of Recoder and not find 'start -> copyword2'.
Can you public new rule.pkl?

Can you provide the list of plausible patches?

Thanks for your excellent work, it is very impressive.
I'm developing an apr tool and want to set Tare as one of my baselines. But I need filter some bugs out to compare the precision value.

Can you provide the list of bugids that Tare will generate plausible patches?

Thank you for considering my request. I am looking forward to your reply.

Train dataset

Hello authors, may I ask for the codes to process the data from raw data (data0.pkl you mentioned in readme) into the mode input (data.pkl)? Thanks so much for your time.