// ConsoleApplication1.cpp : This file contains the 'main' function. Program execution begins and ends there.
//

#include "pch.h"
// utf8 convetreis.cpp : This file contains the 'main' function. Program execution begins and ends there.
//

#include <iostream>
#include <string>
#include <fstream>
#include <vector>
#include <map>
#include <locale>
#include <wchar.h>
#include <io.h>
#include <fcntl.h>
#include <windows.h>



using namespace std;
ofstream fr("utf8.txt");
string decToHex(int raide) {
	string hex;
	int liekana = raide;
	int sveikaDalis = raide;

	while (sveikaDalis >= 1) {          //skaicius dalinamas is 16 ir randama liekana is padalinto skaiciaus
		liekana = sveikaDalis % 16;
		sveikaDalis /= 16;
		if (liekana < 10)       hex.insert(0, to_string(liekana));
		else if (liekana == 10)   hex.insert(0, "A");
		else if (liekana == 11)   hex.insert(0, "B");
		else if (liekana == 12)   hex.insert(0, "C");
		else if (liekana == 13)   hex.insert(0, "D");
		else if (liekana == 14)   hex.insert(0, "E");
		else if (liekana == 15)   hex.insert(0, "F");
	}
	return hex;
}

string hexToBin(string sk) {
	string bin;
	for (int i = 0; i < sk.length(); i++) {
		if (sk[i] == '0')      bin.append("0000");
		else if (sk[i] == '1') bin.append("0001");
		else if (sk[i] == '2') bin.append("0010");
		else if (sk[i] == '3') bin.append("0011");
		else if (sk[i] == '4') bin.append("0100");
		else if (sk[i] == '5') bin.append("0101");
		else if (sk[i] == '6') bin.append("0110");
		else if (sk[i] == '7') bin.append("0111");
		else if (sk[i] == '8') bin.append("1000");
		else if (sk[i] == '9') bin.append("1001");
		else if (sk[i] == 'A') bin.append("1010");
		else if (sk[i] == 'B') bin.append("1011");
		else if (sk[i] == 'C') bin.append("1100");
		else if (sk[i] == 'D') bin.append("1101");
		else if (sk[i] == 'E') bin.append("1110");
		else if (sk[i] == 'F') bin.append("1111");
	}
	return bin;
}

string binToHex(string unicode) {
	string utf;
	for (int i = 0; i < unicode.length(); i += 4) {      //paimami 4 skaiciai is ivesto skaiciaus ir prilyginami
		string ch;
		ch = unicode.substr(i, 4);
		if (ch == "0000") utf.append("0");
		else if (ch == "0001") utf.append("1");
		else if (ch == "0010") utf.append("2");
		else if (ch == "0011") utf.append("3");
		else if (ch == "0100") utf.append("4");
		else if (ch == "0101") utf.append("5");
		else if (ch == "0110") utf.append("6");
		else if (ch == "0111") utf.append("7");
		else if (ch == "1000") utf.append("8");
		else if (ch == "1001") utf.append("9");
		else if (ch == "1010") utf.append("A");
		else if (ch == "1011") utf.append("B");
		else if (ch == "1100") utf.append("C");
		else if (ch == "1101") utf.append("D");
		else if (ch == "1110") utf.append("E");
		else if (ch == "1111") utf.append("F");
	}
	return utf;
}


string convert(int level, string bin) {
	string utf;
	while (1) {
		if (bin.length() > 0) {
			utf.insert(0, 1, bin.back());   //imama paskutinis string char
			bin = bin.substr(0, bin.size() - 1);  //nutrinamas paskutinis char
		}

		else utf.insert(0, "0");

		if (utf.length() == 13 && level == 2) {
			utf.insert(0, "110");
			break;
		}
		else if (utf.length() == 20 && level == 3) {
			utf.insert(0, "1110");
			break;
		}
		else if (utf.length() == 27 && level == 4) {
			utf.insert(0, "11110");
			break;
		}

		if (utf.length() == 6 || utf.length() == 14 || utf.length() == 22) utf.insert(0, "10");

	}
	return utf;
}

void library(map<int, int> &lib) {
	ifstream fd("librarynochar.txt");
	string blank;
	int unicode;
	int codePage;
	unsigned char ch;
	for (int i = 128; i < 256; i++) {
		fd >> blank >> codePage >> unicode;

		lib[codePage] = unicode;
	}
}

void hexToFullChar(string a) {

	for (int i = 0; i < a.length(); i+=2) {
		int b = 0;
		if (a[i] >= '0' && a[i] <= '9') {
			b += (a[i] - '0') * 16;
		}
		else {
			b += (a[i] - 'A' + 10) * 16;
		}

		if (a[i+1] >= '0' && a[i+1] <= '9') {
			b += (a[i+1] - '0');
		}
		else {
			b += (a[i+1] - 'A' + 10);
		}

		char ch = b;
		fr << ch;

	}
}


int main() {

	map<int, int> lib;
	library(lib);

	ifstream fd("386intel.txt");

	fd >> noskipws; //kad skaitytu ir- space

	while (!fd.eof()) {
		unsigned char ch;
		fd >> ch;
		int raide = ch;

		if (raide < 128) {
			fr << ch;
		}
		else {
			raide = lib[raide];
			string hex = decToHex(raide);
			string bin = hexToBin(hex);
			string utf;
			if (raide < 2048) {
				utf = binToHex(convert(2, bin));
			}
			else if (raide < 65536) {
				utf = binToHex(convert(3, bin));
			}
			else {
				utf = binToHex(convert(4, bin));
			}
			hexToFullChar(utf);


		}
	}

	return 0;
}