arff-mining/src/arff/arff.cpp

#include "arff.hpp"
#include <iostream>
#include <fstream>
#include <sstream>
#include <algorithm>

namespace ARFF {
    bool isVerbose = false;

    void ParseArguments(int argc, char* argv[]) {
        std::string argument_string;
        for (int i = 0; i < argc; ++i) {
            argument_string.assign(argv[i]);
            if (argument_string == "-v" || argument_string == "--verbose") { 
                isVerbose = true;
            }
        }
    }

    std::string GetDataFilename(void) {
        std::string filename;
        std::cout << "Please enter name of the data file:\t";
        std::cin >> filename;
        if (filename.empty()) {
            LogError("ARFF/Setup", "No data filename provided, exiting...");
            exit(1);
        }
        std::cout << std::endl;
        return filename;
    }

    void LogInfo(const std::string location, const std::string message) {
        if (!isVerbose) { return; }
        std::cout << '[' << location << " - INFO] ";
        std::cout << message << std::endl;
    }

    void LogError(const std::string location, const std::string message) {
        if (!isVerbose) { return; }
        std::cerr << '[' << location << " - ERROR] ";
        std::cerr << message << std::endl;
    }

    AttributeType::AttributeType(std::string attribute) {
        this->attribute = attribute;
    }

    void AttributeType::AddValue(std::string value) {
        values.emplace_back(value);
    }

    Instance::Instance(const int id, const int size) {
        this->id = id;
        this->values.resize(size);
    }

    // Read entire data file and parse it
    void Arff::Read(std::string filename) {
        std::ifstream dataFile(filename);
        if (!dataFile.is_open()) {
            LogError("ARFF/Read", "Unable to open file with name `" 
                    + filename + ", exiting...");
            exit(1);
        }
        std::string line;
        while (std::getline(dataFile, line)) {
            if (line.size() == 1) { continue; }
            switch (line.at(0)) {
                case '%':
                    // Comment line in data
                    continue;
                    break;
                case '@':
                    AddAttribute(line);
                    break;
                default:
                    AddData(line);
                    break;
            }
        }
        TestIntegrity();
    }

    void Arff::Print(void) {
        std::cout << attributeList.size() << " attributes\n";
        std::cout << database.size() << " examples\n";
        std::cout << std::endl;

        std::cout << "Attribute (#): values\n";
        for (AttributeType type : attributeList) {
            std::cout << type.attribute << " (" << type.values.size() << "):";
            for (std::string value : type.values) {
                std::cout << " " << value;
            }
            std::cout << '\n';
        }
        std::cout << std::endl;

        std::cout << relation << '\n';
        for (Instance instance: database) {
            for (std::string value : instance.values) {
                std::cout << '\t' << value;
            }
            std::cout << '\n';
        }
        std::cout << std::endl;
    }

    // Add the attribute to the list
    void Arff::AddAttribute(std::string line) {
        std::stringstream parser(line);
        std::string token;
        parser >> token;
        // Signifies beginning of data
        // Might add a boolean later to mark this
        if (token == "@data") {
            return;
        }
        if (token == "@relation") {
            parser >> token;
            relation = token;
            LogInfo("ARFF/Attribute", "Relation set: " + relation);
            return;
        }
        parser >> token;
        attributeList.emplace_back(token);
        LogInfo("ARFF/Attribute", "Added attribute: " + token);
        while (parser >> token) {
            // Clean token from outside pieces
            token.erase(std::remove(token.begin(), token.end(), '{'), token.end());
            token.erase(std::remove(token.begin(), token.end(), '}'), token.end());
            token.erase(std::remove(token.begin(), token.end(), ','), token.end());
            token.erase(std::remove(token.begin(), token.end(), '\r'), token.end());
            token.erase(std::remove(token.begin(), token.end(), '\n'), token.end());
            attributeList.back().AddValue(token);
            LogInfo("ARFF/Attribute", "Added value: " + token);
        }
    }

    // Add data to runtime database
    void Arff::AddData(std::string line) {
        std::istringstream parser(line);
        std::string token;
        int id = 0;
        if (!database.empty()) { id = database.back().id + 1; }
        database.emplace_back(id, attributeList.size());
        LogInfo("ARFF/Data", "Added id: " + std::to_string(database.back().id));
        for (int i = 0; i < attributeList.size(); ++i) {
            std::getline(parser, token, ',');
            token.erase(std::remove(token.begin(), token.end(), '\r'), token.end());
            token.erase(std::remove(token.begin(), token.end(), '\n'), token.end());
            database.back().values.at(i) = token;
            LogInfo("ARFF/Data", "Added instance value: " + token);
        }
    }

    void Arff::TestIntegrity(void) {
        for (Instance instance : database) {
            int successCheck = 0;
            for (int i = 0; i < attributeList.size(); ++i) {
                LogInfo("ARFF/Integrity", "Instance value tested: '" 
                        + instance.values.at(i) + "'");
                for (std::string value : attributeList.at(i).values) {
                    LogInfo("ARFF/Integrity", "attributeList value: '" 
                            + value + "'");
                    if (instance.values.at(i) == value) {
                        LogInfo("ARFF/Integrity", "Value found: " + value);
                        ++successCheck;
                        break;
                    }
                }
            }
            if (successCheck != attributeList.size()) {
                LogError("ARFF/Integrity", "Value size mismatch: "
                        + std::to_string(successCheck) + " out of " 
                        + std::to_string(attributeList.size()));
                exit(1);
            }
        }
        LogInfo("ARFF/Integrity", "All values exist, continuing...");
    }
}