preprocessor_build.hpp

Go to the documentation of this file.

 
#pragma once
 
#include "duplication_finder_diff.hpp"
#include "duplication_finder_tool.hpp"
#include "function_breaker.hpp"
#include "parser.hpp"
#include <arkanjo/base/config.hpp>
#include <arkanjo/commands/pre/preprocessor.hpp>
#include <arkanjo/commands/command_base.hpp>
#include <string>
#include <tuple>
#include <filesystem>
using namespace std;
namespace fs = std::filesystem;
 
class PreprocessorBuild : public Preprocessor, public CommandBase<PreprocessorBuild> {
  private:
    static constexpr const char* TEMPLATE_COUNT =
      "The total number of functions that are "
      "similar to the found one is {count}. "
      "More info about them are listed below.";
    // User interaction messages
    string PROJECT_PATH_MESSAGE = "Enter your project path:";                                  
    string MINIMUM_SIMILARITY_MESSAGE = "Enter minimum similarity desired on using the tool:"; 
 
    // Processing stage messages
    string INITIAL_MESSAGE = "Initiating Preprocessing";                                           
    string BREAKER_MESSAGE = "Reading codebase... (this may take a while)";                        
    string DUPLICATION_MESSAGE = "Finding duplication in the codebase... (this may take a while)"; 
    string SAVING_MESSAGE = "Saving results...";                                                   
    string END_MESSAGE = "Finished preprocessing";                                                 
 
    // Duplication finder selection messages
    string MESSAGE_DUPLICATION_FINDER_TYPE_1 = "Enter the number of the duplication finder technique you want to use:";
    string MESSAGE_DUPLICATION_FINDER_TYPE_2 = "1) NLP text similarity using gensim";
    string MESSAGE_DUPLICATION_FINDER_TYPE_3 = "2) Count proportion of equal lines using diff command";
    string INVALID_CODE_DUPLICATION_FINDER = "Valid options are '1' or '2' only. Stopping Program...";
 
    tuple<string, double, bool> read_parameters();
 
    void preprocess(const fs::path& path, double similarity, bool use_duplication_finder_by_tool);
 
  public:
    PreprocessorBuild();
 
    PreprocessorBuild(bool force_preprocess);
 
    PreprocessorBuild(bool force_preprocess, const fs::path& path, double similarity);
 
    COMMAND_DESCRIPTION("Creates the foundation for fast query responses during the main operation phase.")
 
    bool validate(const ParsedOptions& options) override;
 
    bool run(const ParsedOptions& options) override;
};