bayesian-spam-filter/src/filter.hpp

#ifndef FILTER_HPP
#define FILTER_HPP

#include <map>
#include <string>
#include <vector>

// C++ maps don't allow specifying default value
struct DoubleDefaultedToHalf {
    double value = 0.5;
};

struct ReportData {
    double spam_precision;
    double spam_recall;
    double ham_precision;
    double ham_recall;
    double spam_f_score;
    double ham_f_score;
    double accuracy;
};

struct SMSMessage {
    SMSMessage(bool given_type, std::string given_message);
    bool is_ham;
    std::string message;
    bool is_ham_filter;
};

class SMSMessageFilter {
public:
    SMSMessageFilter(void) = default;
    ~SMSMessageFilter(void) = default;
    bool is_generator_defined = false;
    bool is_input_defined = false;
    void GenerateProbability();
    void Prepare();
    void Filter(void);
    void Report(void);
    void ReadArguments(int argc, char* argv[]);

private:
    double sentence_probability_ham = 0.2; // Sentence is spam if < this value 
    std::map<std::string, DoubleDefaultedToHalf> probability_dictionary;
    std::string generation_file_path;
    std::string filter_file_path;
    std::vector<SMSMessage> filtered_messages;
    ReportData GenerateReport(void);
    void PrintReport(ReportData report);
    
};

std::string SanitizeToken(std::string token);
void PrintHelp(void);

#endif // !FILTER_HPP