big_clone_tailor_evaluator.hpp

Go to the documentation of this file.

 
#pragma once
 
#include <cassert>
#include <iomanip>
#include <iostream>
#include <map>
#include <set>
#include <string>
#include <tuple>
#include <vector>
 
#include <arkanjo/base/path.hpp>
#include <arkanjo/base/similarity_table.hpp>
#include <arkanjo/utils/utils.hpp>
 
#include <arkanjo/commands/command_base.hpp>
 
using namespace std;
 
class BigCloneTailorEvaluator : public CommandBase<BigCloneTailorEvaluator> {
  private:
    string CLONE_LABELS_FILE_PATH = "/home/luan/linux/test-duplication-code/parser/bigclonebench/clone_labels.txt"; 
    string EXTENSION = ".java";                                                                                     
    int NUMBER_OF_TYPES = 6;                                                                                        
    int NOT_CLONE_TYPE_ID = 5;                                                                                      
    string RECALL_PER_TYPE_PRINT = "Recall Value of each type";                                                     
    double MINIMUM_SIMILARITY_TEMP = 0;                                                                             
 
    vector<string> ID_TO_TYPE_LABEL = {
        "T1",
        "T2",
        "ST3",
        "MT3",
        "WT3/T4",
        "Negative",
    };
 
    Similarity_Table* similarity_table;       
    map<pair<int, int>, int> id_pair_to_type; 
    vector<int> count_of_samples_by_type;     
 
    void read_clone_labels();
 
    int path_to_id(Path path);
 
    vector<tuple<double, int, int>> similar_path_pairs_formated_with_id();
 
    bool is_relevant_pair(int id0, int id1);
 
    set<pair<int, int>> filter_similar_id_pairs_only_relevant_ones(
        vector<pair<int, int>> similar_id_pairs);
 
    vector<pair<int, int>> filter_similar_path_pairs_by_similarity(
        vector<tuple<double, int, int>> similar_id_pairs,
        double minimum_similarity);
 
    vector<int> build_frequency_corrected_guessed_by_type(
        vector<pair<int, int>> similar_id_pairs);
 
    double calc_recall(vector<int> frequency, int type);
 
    void print_recall_per_type(vector<int> frequency);
 
    void evaluate(double minimum_similarity);
 
  public:
    COMMAND_DESCRIPTION(
        "Evaluate clone detection results using the BigCloneBench dataset "
        "methodology. The command compares detected similarities against "
        "ground truth labels and reports recall metrics per clone type.")
 
    
    explicit BigCloneTailorEvaluator(Similarity_Table* _similarity_table);
 
    bool validate(const ParsedOptions& options) override;
 
    bool run(const ParsedOptions& options) override;
};