project-codenet / model-experiments /token-based-similarity-classification /src /Clustering /ClusterClassifier.py
| """ | |
| Program for code classification by Bag or Tokens technique | |
| using cluster analysis | |
| Implements nearest cluster simple classifier of sorce code: | |
| The programm processes all files in the given directory | |
| Each file there contains all samples of one class of samples, | |
| i.e. tokenised source code | |
| Number of classes (lables) is defined automatically | |
| """ | |
| import sys | |
| import os | |
| import argparse | |
| sys.path.extend(["../BagOfTokens"]) | |
| from DatasetLoader import BagOfTokensLoader | |
| from NearestCluster import NearestClusterClassifier | |
| def main(args): | |
| """ | |
| Main function of program for classifying source code | |
| Parameters: | |
| - args -- Parsed command line arguments | |
| as object returned by ArgumentParser | |
| """ | |
| ds_loader = BagOfTokensLoader(args.samples, args.tokens) | |
| all_samples, problem_samples = ds_loader.getPartitionedSampes() | |
| val_train_split = args.valpart | |
| val_samples = \ | |
| [_solutions[: int(float(len(_solutions)) * val_train_split)] | |
| for _solutions in problem_samples] | |
| train_samples = \ | |
| [_solutions[int(float(len(_solutions)) * val_train_split) :] | |
| for _solutions in problem_samples] | |
| classifier = NearestClusterClassifier(train_samples, val_samples) | |
| #classifier.nClustersTrain(args.clusters) | |
| classifier.scaledClustersTrain(args.clusters) | |
| accuracy = classifier.validate() | |
| print("Accuracy: ", accuracy) | |
| ####################################################################### | |
| # Command line arguments of are described below | |
| ####################################################################### | |
| if __name__ == '__main__': | |
| print("\nCODE CLASSIFICATION WITH BAG OF TOKENS CLUTERING CLASSIFIER") | |
| #Command-line arguments | |
| parser = argparse.ArgumentParser( | |
| description = "Bag of Tokens program source code classifier") | |
| parser.add_argument('samples', | |
| type=str, help='path to data directory') | |
| parser.add_argument('--tokens', default=17, type=int, | |
| help='number of tokens') | |
| parser.add_argument('--valpart', default=0.2, type=float, | |
| help='fraction of samples for validation') | |
| parser.add_argument('--clusters', default=8, type=int, | |
| help='number of clusters') | |
| args = parser.parse_args() | |
| print("Parameter settings used:") | |
| for k,v in sorted(vars(args).items()): | |
| print("{}: {}".format(k,v)) | |
| main(args) | |