2003-08-29 23:09:00 +02:00
|
|
|
#ifndef ARABICA_UTF16UTF8_CODECVT_H
|
|
|
|
#define ARABICA_UTF16UTF8_CODECVT_H
|
2002-06-21 13:16:28 +02:00
|
|
|
//---------------------------------------------------------------------------
|
2003-08-29 23:09:00 +02:00
|
|
|
// class utf16utf8codecvt
|
2002-06-21 13:16:28 +02:00
|
|
|
// This facet converts from Unicode (UCS-2) wchar_ts to
|
|
|
|
// char using the UTF-8 encoding.
|
|
|
|
//
|
|
|
|
// For the full guff on codecvts see section 22.2.1.5 of
|
|
|
|
// The C++ Standard (ISO/IEC 14882 to be pedantic).
|
|
|
|
//
|
|
|
|
// I got my information about UTF-8 from RFC 2044.
|
|
|
|
//
|
|
|
|
// $Id$
|
|
|
|
//---------------------------------------------------------------------------
|
2007-09-05 00:55:47 +02:00
|
|
|
#include <SAX/ArabicaConfig.hpp>
|
2003-09-11 16:26:26 +02:00
|
|
|
|
2002-06-21 13:16:28 +02:00
|
|
|
#include <locale>
|
|
|
|
|
2003-09-10 12:39:01 +02:00
|
|
|
#ifndef ARABICA_NO_CODECVT_SPECIALISATIONS
|
2007-09-05 00:55:47 +02:00
|
|
|
#include <Utils/impl/codecvt_specialisations.hpp>
|
2003-09-10 12:39:01 +02:00
|
|
|
#endif
|
2003-09-10 17:15:55 +02:00
|
|
|
|
2003-09-10 12:39:01 +02:00
|
|
|
#ifdef ARABICA_VS6_WORKAROUND
|
2007-09-05 00:55:47 +02:00
|
|
|
#include <Utils/impl/VS6Workaround.hpp>
|
2002-06-21 13:16:28 +02:00
|
|
|
#endif
|
|
|
|
|
2003-09-13 01:15:14 +02:00
|
|
|
namespace Arabica
|
|
|
|
{
|
|
|
|
namespace convert
|
|
|
|
{
|
|
|
|
|
2003-08-29 23:09:00 +02:00
|
|
|
class utf16utf8codecvt : public std::codecvt<char, wchar_t, std::mbstate_t>
|
2002-06-21 13:16:28 +02:00
|
|
|
{
|
|
|
|
protected:
|
2003-09-13 01:15:14 +02:00
|
|
|
typedef std::codecvt_base::result result;
|
|
|
|
|
2003-08-29 23:09:00 +02:00
|
|
|
virtual ~utf16utf8codecvt() { }
|
2002-06-21 13:16:28 +02:00
|
|
|
|
|
|
|
virtual result do_out(std::mbstate_t&,
|
|
|
|
const char* from,
|
|
|
|
const char* from_end,
|
|
|
|
const char*& from_next,
|
|
|
|
wchar_t* to,
|
|
|
|
wchar_t* to_limit,
|
|
|
|
wchar_t*& to_next) const;
|
|
|
|
|
2003-08-29 23:09:00 +02:00
|
|
|
virtual result do_in(std::mbstate_t&,
|
|
|
|
const wchar_t* from,
|
|
|
|
const wchar_t* from_end,
|
|
|
|
const wchar_t*& from_next,
|
|
|
|
char* to,
|
|
|
|
char* to_limit,
|
|
|
|
char*& to_next) const;
|
|
|
|
|
2002-06-21 13:16:28 +02:00
|
|
|
virtual result do_unshift(std::mbstate_t&,
|
2003-08-29 23:09:00 +02:00
|
|
|
wchar_t*,
|
|
|
|
wchar_t*,
|
|
|
|
wchar_t*&) const;
|
2002-06-21 13:16:28 +02:00
|
|
|
|
2003-08-29 23:09:00 +02:00
|
|
|
virtual int do_encoding() const throw() { return 0; }
|
2002-06-21 13:16:28 +02:00
|
|
|
|
2003-08-29 23:09:00 +02:00
|
|
|
virtual bool do_always_noconv() const throw() { return false; }
|
2002-06-21 13:16:28 +02:00
|
|
|
|
|
|
|
virtual int do_length(const std::mbstate_t&,
|
2003-08-29 23:09:00 +02:00
|
|
|
const wchar_t* from,
|
|
|
|
const wchar_t* end,
|
2002-06-21 13:16:28 +02:00
|
|
|
size_t max) const;
|
|
|
|
|
2003-08-29 23:09:00 +02:00
|
|
|
virtual int do_max_length() const throw() { return 1; }
|
|
|
|
}; // class utf16utf8codecvt
|
2002-06-21 13:16:28 +02:00
|
|
|
|
2003-09-11 16:05:18 +02:00
|
|
|
} // namespace convert
|
|
|
|
} // namespace Arabica
|
2003-09-11 16:26:26 +02:00
|
|
|
|
2002-06-21 13:16:28 +02:00
|
|
|
#endif
|
2003-09-10 12:39:01 +02:00
|
|
|
|