/Artifact-of-Assertion-ICSE22

Primary LanguagePython

Artifact of "Automated Assertion Generation via Information Retrieval and Its Integration with Deep Learning"

Usage

Depedency:

  • pytorch
  • javalang

Step1: Dataset

unzip Dataset in ./Result and create a config file to indicate data path like:

SomeWhere/TrainMethod.txt
SomeWhere/TestMethod.txt
SomeWhere/TrainAssertion.txt
SomeWhere/TestAssertion.txt
SomeWhere/ValMethod.txt
SomeWhere/ValMethod.txt

Step2: Retrieval

python ./Retrieval/IR.py $input_config $result_path

Step3: Adaption by Heuristic

New DataSet

python ./Retrieval/IR.py  $result_path 
New

Old DataSet

python ./Retrieval/IR.py  $result_path Old

Step4: Training Neural Models

Train:

Data Preparation

python ./NeuralModel/DataPrepration $input_config $result_path $neural_data_path_train train

Model Training

python ./NeuralModel/main $neural_data_path_train $neural_result_path train

Inference:

Data Preparation

python ./NeuralModel/DataPrepration.py $input_config $result_path $neural_data_path_evaluate evaluate

Model Evaluating

python ./NeuralModel/main.py $neural_data_path_evaluate  $neural_result_path evaluate

Step5: Evaluating

  1. Before Evaluate Integrated Approach, Use a deep learning generative model (i.e. ATLAS) to generate result.

  2. Generate Result from Adapt NN and Integration

python AdaptionIntegration.py $input_config $result_path $neural_result_path $integration_threshold
  1. Evaluate Result

Old Dataset

python countMultiOldDataSet.py $result_path

New Dataset

python countMultiNewDataSet.py $result_path