//clang
#include <iostream>
#include <string>
#include <locale>
#include <fstream>
#include <codecvt>

using namespace std;

int main(int argc, char **argv) {
	// in C++03:
	const char *s1 = "Standard string";
	const wchar_t *s2 = L"Wide character string";
	cout << "s1: " << s1 << endl;
	cout << "s2: " << s2 << endl; // pointer written b/c cout is a char stream
	wcout << "s2: " << s1 << endl; 
	wcout << "s2: " << s2 << endl; 
	// (neither are unicode encoded!)
	
	
	// C++11 Unicode support:
	// char is now wide enough to store eight-bit coding of UTF-8
	// char16_t is a 16-bit character (to store UTF-16)
	// char32_t is a 32-bit character (to store UCS-4/UTF-32)
	// NOTE: UTF-32: always 32 bits per code point (character)
	//       UTF-16: 1 or 2 16-bit code units per code point (character)
	//       UTF-8:  1-4 bytes per code point (character)
	//               [no endianness problems]
	
	char c1 = 'A';
	wchar_t c2 = L'A';
	// below literals are UCS-encoded
	char16_t c3 = u'A'; // use 'u' prefix to denote char16_t
	char32_t c4 = U'A'; // use 'U' prefix to denote char32_t
	cout << c1 << ' ' << c2 << ' ' << c3 << ' ' << c4 << endl;
	// only c1 written as character, because cout is a char stream
	wcout << c1 << ' ' << c2 << ' ' << c3 << ' ' << c4 << endl;
	// c1 and c2 written above, but char16_t and char32_t are not
	
	// strings also available:
			// UTF-8 encoded string:
	const char *s3 = u8"Euros \u20AC and G clef \U0001D11E";
			// UTF-16 encoded string:
	const char16_t *s4 = u"Euros \u20AC and G clef \U0001D11E";
			// UTF-32 encoded string:
	const char32_t *s5 = U"Euros \u20AC and G clef \U0001D11E";

	// note: u8 not allowed for characters
	//       u allowed for characters, but cannot produce 2 16-bit code units
	//       (u"   " can produce multiple 16-bit units per "character")
	
	std::u16string s6 = s4; // u16string = basic_string<char16_t>	
	std::u32string s7 = s5; // u32string = basic_string<char32_t>	

	cout << endl;
	// write out the code units:
	cout << string(s3).length() << ": ";
	for(char c : string(s3))
		cout << c << ' ';
	cout << endl;

	cout << s6.length() << ": ";
	for(char16_t c : s6)
		cout << hex << showbase << c << ' ';
	cout << dec << endl;

	cout << s7.length() << ": ";
	for(char32_t c : s7)
		cout << hex << showbase << c << ' ';
	cout << dec << endl;


	// UTF-8 encoded should work to wcout if 
	//wcout << s3 << endl; // doesn't on my system... :(

	// this should create a method to write a char32_t string to cout
	// but it doesn't work with any STL I can find... :(
	/*
	wbuffer_convert<codecvt_utf8<char32_t>,char32_t> u32buffer(cout.rdbuf(),new codecvt_utf8<char32_t>());
	std::basic_ostream<char32_t> u32out(&u32buffer);
	u32out << s5 << endl;
	*/

	// Raw String Literals:
	// ====================
	// use "R" (after "L" "u" "U" or "u8") for a raw string
	cout << endl;
	cout << "standard string: " <<   "a string\t\"quote\"" << endl;
	cout << "raw string:      " << R"(a string\t"quote")" << endl;
	// can change "( )" into "<something>( )<something>"
	cout << "another raw str: " << R"#(a string\t(paren))#" << endl;

}