继续编码转换
Unicode 有两套编码集,UCS-2 和 UCS-4。Windows 的内部其实是用的 UCS-2 标准,并用 UTF-16 来实现。而非 Windows 系统大多采用了 UTF-8 实现。
大家都知道在windows上wchar_t是2个字节表示,而在Linux上wchar_t是4个字节表示的。这样写跨平台的程序时,就会不统一。
下面是我用到的几个函数,wchar_t与UTF-8编码之间的转换和UTF-16和wchar_t之间的转换。
view plaincopy to clipboardprint?
#ifdef WINDOWS
#include <windows.h>
#include <stdio.h>
#include <ctype.h>
#else
#include <iconv.h>
#include <wctype.h>
#include <wchar.h>
#include <errno.h>
#endif
//wchar_t转成UTF-8
int FW2UTF8Convert( const wchar_t* a_szSrc, int a_nSrcSize, char* a_szDest, int a_nDestSize )
{
#ifdef WINDOWS
return WideCharToMultiByte( CP_UTF8, 0, a_szSrc, -1, a_szDest, a_nDestSize, NULL, NULL );
#else
size_t result;
iconv_t env;
env = iconv_open("UTF-8","WCHAR_T");
if (env==(iconv_t)-1)
{
printf("iconv_open WCHAR_T->UTF8 error%s %d/n",strerror(errno),errno) ;
return -1;
}
result = iconv(env,(char**)&a_szSrc,(size_t*)&a_nSrcSize,(char**)&a_szDest,(size_t*)&a_nDestSize);
if (result==(size_t)-1)
{
printf("iconv WCHAR_T->UTF8 error %d/n",errno) ;
return -1;
}
iconv_close(env);
return (int)result;
#endif
}
//UTF-8转成wchar_t
int FUTF82WConvert( const char* a_szSrc, wchar_t* a_szDest, int a_nDestSize )
{
#ifdef WINDOWS
return MultiByteToWideChar( CP_UTF8, 0, a_szSrc, -1, a_szDest, a_nDestSize );
#else
size_t result;
iconv_t env;
int size = strlen(a_szSrc)+1 ;
env = iconv_open("WCHAR_T","UTF-8");
if (env==(iconv_t)-1)
{
printf("iconv_open UTF8->WCHAR_T error %d/n",errno) ;
return -1;
}
result = iconv(env,(char**)&a_szSrc,(size_t*)&size,(char**)&a_szDest,(size_t*)&a_nDestSize);
if (result==(size_t)-1)
{
printf("iconv UTF8->WCHAR_T error %d/n",errno) ;
return -1;
}
iconv_close(env);
return (int)result;
#endif
}
//wchar_t转成utf16
int FW2UConvert( const wchar_t* a_szSrc, int a_nSize,char* a_szDest, int a_nDestSize )
{
#ifdef WINDOWS
memcpy_s((wchar_t*)a_szDest,a_nDestSize,a_szSrc,a_nSize);
return a_nSize ;
#else
size_t result;
iconv_t env;
env = iconv_open("UCS-2-INTERNAL","UCS-4-INTERNAL");
if (env==(iconv_t)-1)
{
printf("iconv_open WCHAR_T->UTF16 error%s %d/n", strerror(errno),errno);
return -1;
}
result = iconv(env,(char**)&a_szSrc,(size_t*)&a_nSize,(char**)&a_szDest,(size_t*)&a_nDestSize);
if (result==(size_t)-1)
{
printf("iconv WCHAR_T->UTF16 error %s %d/n", strerror(errno), errno);
return -1;
}
iconv_close(env);
return (int)result;
#endif
}
//utf16转成wchar_t
int FU2WConvert( const char* a_szSrc, int a_nSize, wchar_t* a_szDest, int a_nDestSize )
{
#ifdef WINDOWS
memcpy_s(a_szDest,a_nDestSize,(const wchar_t*)a_szSrc,a_nSize);
return a_nSize ;
#else
size_t result;
iconv_t env;
env = iconv_open("UCS-4-INTERNAL","UCS-2-INTERNAL");
if (env==(iconv_t)-1)
{
printf("iconv_open error %d/n",errno) ;
return -1;
}
result = iconv(env,(char**)&a_szSrc,(size_t*)&a_nSize,(char**)&a_szDest,(size_t*)&a_nDestSize);
if (result==(size_t)-1)
{
printf("UTF16 -> WCHAR_T conv error %d/n",errno) ;
return -1;
}
iconv_close(env);
return (int)result;
#endif
}