NODoctor

The artifact repository for paper "Can ChatGPT repair Non-Order-Dependent Tests?" in Flaky Test Workshop at ICSE2024, Lisbon, Portugal, April 2024.

Reproduce results

1. To set up the environments:

bash -x scripts/setup.sh

2. To clone all Java projects into a directory, one can run the following command:

bash -x clone.sh [InputCSV] [CloneDIR]

The arguments are as follows:

• InputCSV: An input csv files which includes the info of project,sha,module, such as:

https://github.com/apache/nifi,2bd752d868a8f3e36113b078bb576cf054e945e8,nifi-commons/nifi-record

• CloneDIR: the directory where all Java projects to be cloned

3. To reproduce the results, one can run the following command:

bash -x scripts/repair.sh [InputCSV] [CloneDIR] [APIKey] [ResultDir] NOD_repair.py

The arguments are as follows:

• InputCSV: An input csv files which includes the info of project,sha,module,test,type,status,pr,notes for each test (same information as in IDoFT, such as:

https://github.com/apache/nifi,2bd752d868a8f3e36113b078bb576cf054e945e8,nifi-commons/nifi-record,org.apache.nifi.serialization.record.TestDataTypeUtils.testInferTypeWithMapNonStringKeys,ID,,,,

• CloneDIR: the directory where all Java projects are located
• APIKey: OpenAI token
• ResultDir: the directory to save all results. Each run of the experiments will generate a directory with a unique SHA as the folder name, under the folder there are patches, detailed result information, and all logs