/Grace

Primary LanguagePython

Grace: Graph-based representation learning for fault localization

Grace is a coverage-based fault localization approach with graph-based representation. This repository includes data and code about the technique.

Execution

runtotal.py is main entry file. Using python runtotal.py subname (e.g., python runtotal.py Lang) would execute the run.py, sum.py, watch.py respectively.

  • run.py is for each buggy version of each project, which is repeatedly executed in runtotal.py.
  • sum.py merges the results for all the buggy version of one project.
  • watch.py prints the results.
  • Model.py is about the model.
  • Dataset.py is about the dataset.

The final results are logged in the directory result_final_XXXX while the third line is the number of Top-1 Value.

Environment

PyTorch: V1.7.1
OS: Ubuntu 16.04.6 LTS

Dataset

The preprocessed dataset could be download in link. Please ensure that the path to these .pkl file is correctly set in the code.