// codecvt facet for UTF-8 multibyte code, UTF-16 wide-character code
 #pragma once
#ifndef _CVT_UTF8_UTF16_
#define _CVT_UTF8_UTF16_
#ifndef RC_INVOKED
#include <stdexcept>
#include <locale>
#include <codecvt>
#include <cwchar>

 #pragma pack(push,_CRT_PACKING)
 #pragma warning(push,_STL_WARNING_LEVEL)
 #pragma warning(disable: _STL_DISABLED_WARNINGS)
 _STL_DISABLE_CLANG_WARNINGS
 #pragma push_macro("new")
 #undef new

#if !_HAS_IF_CONSTEXPR
 #pragma warning(disable: 4127)	// conditional expression is constant
 #pragma warning(disable: 6326)	// Potential comparison of a constant with another constant.
#endif /* ^^^ original code ^^^ */

namespace stdext {
	namespace cvt {

typedef _CSTD mbstate_t _Mbstatype;

#pragma warning(push)
#pragma warning(disable: 4996)	// was declared deprecated
		// CLASS codecvt_utf8_utf16
template<class _Elem,
	unsigned long _Maxcode = 0x10ffff,
	_STD codecvt_mode _Mode = (_STD codecvt_mode)0,
	class _Statype = _Mbstatype>
	class codecvt_utf8_utf16
	: public _STD codecvt<_Elem, char, _Statype>
	{	// facet for converting between UTF-16 _Elem and UTF-8 byte sequences
public:
	typedef _STD codecvt<_Elem, char, _Statype> _Mybase;
	typedef typename _Mybase::result result;
	typedef char _Byte;
	typedef _Elem intern_type;
	typedef _Byte extern_type;
	typedef _Statype state_type;

	static_assert(sizeof (unsigned short) <= sizeof (state_type),
		"state_type too small");

	explicit codecvt_utf8_utf16(size_t _Refs = 0)
		: _Mybase(_Refs)
		{	// construct with ref count
		}


	virtual ~codecvt_utf8_utf16() noexcept
		{	// destroy the object
		}

protected:
	virtual result do_in(_Statype& _State,
		const _Byte *_First1, const _Byte *_Last1, const _Byte *& _Mid1,
		_Elem *_First2, _Elem *_Last2, _Elem *& _Mid2) const
		{	// convert bytes [_First1, _Last1) to [_First2, _Last2)
		unsigned short *_Pstate = (unsigned short *)&_State;
		_Mid1 = _First1;
		_Mid2 = _First2;

		while (_Mid1 != _Last1 && _Mid2 != _Last2)
			{	// convert a multibyte sequence
			unsigned long _By = static_cast<unsigned char>(*_Mid1);
			unsigned long _Ch;
			int _Nextra, _Nskip;

			if (*_Pstate > 1)
				{	// leftover word
				if (_By < 0x80 || 0xc0 <= _By)
					{
					return (_Mybase::error);	// not continuation byte
					}

				// deliver second half of two-word value
				++_Mid1;
				*_Mid2++ = (_Elem)(*_Pstate | (_By & 0x3f));
				*_Pstate = 1;
				continue;
				}

			if (_By < 0x80)
				{
				_Ch = _By;
				_Nextra = 0;
				}
			else if (_By < 0xc0)
				{	// 0x80-0xdf not first byte
				++_Mid1;
				return (_Mybase::error);
				}
			else if (_By < 0xe0)
				{
				_Ch = _By & 0x1f;
				_Nextra = 1;
				}
			else if (_By < 0xf0)
				{
				_Ch = _By & 0x0f;
				_Nextra = 2;
				}
			else if (_By < 0xf8)
				{
				_Ch = _By & 0x07;
				_Nextra = 3;
				}
			else
				{
				_Ch = _By & 0x03;
				_Nextra = _By < 0xfc ? 4 : 5;
				}

			_Nskip = _Nextra < 3 ? 0 : 1;	// leave a byte for 2nd word
			_First1 = _Mid1;	// roll back point

			if (_Nextra == 0)
				{
				++_Mid1;
				}
			else if (_Last1 - _Mid1 < _Nextra + 1 - _Nskip)
				{
				break;	// not enough input
				}
			else
				{
				for (++_Mid1; _Nskip < _Nextra; --_Nextra, ++_Mid1)
					{
					if ((_By = (unsigned char)*_Mid1) < 0x80 || 0xc0 <= _By)
						{
						return (_Mybase::error);	// not continuation byte
						}
					else
						{
						_Ch = _Ch << 6 | (_By & 0x3f);
						}
					}
				}

			if (0 < _Nskip)
				{
				_Ch <<= 6;	// get last byte on next call
				}

			if ((_Maxcode < 0x10ffff ? _Maxcode : 0x10ffff) < _Ch)
				{
				return (_Mybase::error);	// value too large
				}
			else if (0xffff < _Ch)
				{	// deliver first half of two-word value, save second word
				unsigned short _Ch0 =
					(unsigned short)(0xd800 | (_Ch >> 10) - 0x0040);

				*_Mid2++ = (_Elem)_Ch0;
				*_Pstate = (unsigned short)(0xdc00 | (_Ch & 0x03ff));
				continue;
				}

			if (_Nskip != 0)
				{
				if (_Mid1 == _Last1)
					{	// not enough bytes, noncanonical value
					_Mid1 = _First1;
					break;
					}

				if ((_By = (unsigned char)*_Mid1++) < 0x80 || 0xc0 <= _By)
					{
					return (_Mybase::error);	// not continuation byte
					}

				_Ch |= _By & 0x3f;	// complete noncanonical value
				}

			if (*_Pstate == 0)
				{	// first time, maybe look for and consume header
				*_Pstate = 1;

				if _CONSTEXPR_IF ((_Mode & _STD consume_header) != 0)
					{
					if (_Ch == 0xfeff)
						{	// drop header and retry
						result _Ans = do_in(_State, _Mid1, _Last1, _Mid1,
							_First2, _Last2, _Mid2);

						if (_Ans == _Mybase::partial)
							{	// roll back header determination
							*_Pstate = 0;
							_Mid1 = _First1;
							}
						return (_Ans);
						}
					}
				}

			*_Mid2++ = (_Elem)_Ch;
			}

		return (_First1 == _Mid1 ? _Mybase::partial : _Mybase::ok);
		}

	virtual result do_out(_Statype& _State,
		const _Elem *_First1, const _Elem *_Last1, const _Elem *& _Mid1,
		_Byte *_First2, _Byte *_Last2, _Byte *& _Mid2) const
		{	// convert [_First1, _Last1) to bytes [_First2, _Last)
		unsigned short *_Pstate = (unsigned short *)&_State;
		_Mid1 = _First1;
		_Mid2 = _First2;

		while (_Mid1 != _Last1 && _Mid2 != _Last2)
			{	// convert and put a wide char
			unsigned long _Ch;
			unsigned short _Ch1 = (unsigned short)*_Mid1;
			bool _Save = false;

			if (1 < *_Pstate)
				{	// get saved MS 11 bits from *_Pstate
				if (_Ch1 < 0xdc00 || 0xe000 <= _Ch1)
					{
					return (_Mybase::error);	// bad second word
					}
				_Ch = static_cast<unsigned long>((*_Pstate << 10) | (_Ch1 - 0xdc00));
				}
			else if (0xd800 <= _Ch1 && _Ch1 < 0xdc00)
				{	// get new first word
				_Ch = static_cast<unsigned long>((_Ch1 - 0xd800 + 0x0040) << 10);
				_Save = true;	// put only first byte, rest with second word
				}
			else
				{
				_Ch = _Ch1;	// not first word, just put it
				}

			_Byte _By;
			int _Nextra;

			if (_Ch < 0x0080)
				{
				_By = (_Byte)_Ch;
				_Nextra = 0;
				}
			else if (_Ch < 0x0800)
				{
				_By = (_Byte)(0xc0 | _Ch >> 6);
				_Nextra = 1;
				}
			else if (_Ch < 0x10000)
				{
				_By = (_Byte)(0xe0 | _Ch >> 12);
				_Nextra = 2;
				}
			else
				{
				_By = (_Byte)(0xf0 | _Ch >> 18);
				_Nextra = 3;
				}

			int _Nput = _Nextra < 3 ? _Nextra + 1 : _Save ? 1 : 3;

			if (_Last2 - _Mid2 < _Nput)
				{
				break;	// not enough room, even without header
				}

			if _CONSTEXPR_IF ((_Mode & _STD generate_header) != 0)
				{	// maybe header to put
				if (*_Pstate == 0)
					{	// header to put
					if (_Last2 - _Mid2 < 3 + _Nput)
						{
						break;	// not enough room for both
						}

					// prepend header
					*_Mid2++ = (_Byte)(unsigned char)0xef;
					*_Mid2++ = (_Byte)(unsigned char)0xbb;
					*_Mid2++ = (_Byte)(unsigned char)0xbf;
					}
				}

			++_Mid1;
			if (_Save || _Nextra < 3)
				{	// put first byte of sequence, if not already put
				*_Mid2++ = _By;
				--_Nput;
				}

			for (; 0 < _Nput; --_Nput)
				{
				*_Mid2++ = (_Byte)((_Ch >> 6 * --_Nextra & 0x3f) | 0x80);
				}

			*_Pstate = (unsigned short)(_Save ? _Ch >> 10 : 1);
			}

		return (_First1 == _Mid1 ? _Mybase::partial : _Mybase::ok);
		}

	virtual result do_unshift(_Statype& _State,
		_Byte *_First2, _Byte *, _Byte *& _Mid2) const
		{	// generate bytes to return to default shift state
		unsigned short *_Pstate = (unsigned short *)&_State;
		_Mid2 = _First2;

		return (1 < *_Pstate
			? _Mybase::error : _Mybase::ok);	// fail if trailing first word
		}

	virtual int do_length(_Statype& _State, const _Byte *_First1,
		const _Byte *_Last1, size_t _Count) const noexcept
		{	// return min(_Count, converted length of bytes [_First1, _Last1))
		size_t _Wchars = 0;
		_Statype _Mystate = _State;

		while (_Wchars < _Count && _First1 != _Last1)
			{	// convert another wide character
			const _Byte *_Mid1;
			_Elem *_Mid2;
			_Elem _Ch;

			switch (do_in(_Mystate, _First1, _Last1, _Mid1,
				&_Ch, &_Ch + 1, _Mid2))
				{	// test result of single wide-char conversion
			case _Mybase::noconv:
				return ((int)(_Wchars + (_Last1 - _First1)));

			case _Mybase::ok:
				if (_Mid2 == &_Ch + 1)
					++_Wchars;	// replacement do_in might not convert one
				_First1 = _Mid1;
				break;

			default:
				return ((int)_Wchars);	// error or partial
				}
			}

		return ((int)_Wchars);
		}

	virtual bool do_always_noconv() const noexcept
		{	// return true if conversions never change input
		return (false);
		}

	virtual int do_max_length() const noexcept
		{	// return maximum length required for a conversion
		if _CONSTEXPR_IF ((_Mode & _STD consume_header) != 0)
			{
			return (9);	// header + max input
			}
		else if _CONSTEXPR_IF ((_Mode & _STD generate_header) != 0)
			{
			return (7);	// header + max output
			}
		else
			{
			return (6);	// 6-byte max input sequence, no 3-byte header
			}
		}

	virtual int do_encoding() const noexcept
		{	// return length of code sequence (from codecvt)
		return (0);	// 0 => varying length
		}
	};
#pragma warning(pop)
		}	// namespace cvt
	}	// namespace stdext

 #ifdef _TEST_IT
  #define NCHARS	0x10000
  #define MYWC_MAX	0x10ffff
  #define MYFILE	"utf8_utf16"
  #define MYNAME	stdext::cvt::codecvt_utf8_utf16<Mywchar>
  #define MYMAKE	make_string

  #include <string>

_STD wstring make_string()
	{	/* make a valid UTF-16 sequence */
	_STD wstring mystring;
	unsigned long ch = 0;

	for (; ch < 0xd800; ++ch)
		mystring.push_back((wchar_t)ch);
	for (; ch < 0xdc00; ++ch)
		{	// add a valid pair
		mystring.push_back((wchar_t)ch);
		mystring.push_back((wchar_t)(ch + 0x400));
		}
	for (; ch < 0xffff; ++ch)
		mystring.push_back((wchar_t)ch);
	if (ch != (unsigned long)WEOF)
		mystring.push_back((wchar_t)ch);
	return (mystring);
	}

  #include <cvt/xtest>
 #endif /* _TEST_IT */

 #pragma pop_macro("new")
 _STL_RESTORE_CLANG_WARNINGS
 #pragma warning(pop)
 #pragma pack(pop)

#endif /* RC_INVOKED */
#endif /* _CVT_UTF8_UTF16_ */

/*
 * Copyright (c) by P.J. Plauger. All rights reserved.
 * Consult your license regarding permissions and restrictions.
V6.50:0009 */