Windows only. Linux needn't.
//UVi Soft ( 2008 )
//Long Fei ( lf426 ), E-mail: zbln426@163.com
//Laboratory of ZaiBieLiunNian
//http://www.cppblog.com/lf426/
//FileName: gb2312_to_Unicode.h
//For Windows only
#ifndef GB2312_TO_UNICODE_H_
#define GB2312_TO_UNICODE_H_
#include <iostream>
#include <vector>
#include "GNU/iconv.h"
//if not include "SDL/SDL.h"
#ifndef _SDL_H
typedef unsigned short int Uint16;
#endif
bool getUnicode(const std::string& str, std::vector<Uint16>& unicodeVectorArray);
#endif
//UVi Soft ( 2008 )
//Long Fei ( lf426 ), E-mail: zbln426@163.com
//Laboratory of ZaiBieLiunNian
//http://www.cppblog.com/lf426/
//FileName: gb2312_to_Unicode.cpp
//For Windows only
#include "gb2312_to_Unicode.h"
int myUTF8_to_UNICODE(Uint16* unicode, unsigned char* utf8, int len);
bool getUnicode(const std::string& str, std::vector<Uint16>& unicodeVectorArray)
{
const int CHAR_SIZE = 256;
//GB2312 src
const unsigned char* src = (const unsigned char*)(str.c_str());
size_t src_len = strlen((char*)src);
//Unicode dst to get
unsigned char dst[CHAR_SIZE] = {0};
size_t dst_len = sizeof(dst);
//iconv arg
const unsigned char* in = src;
unsigned char* out = dst;
iconv_t cd;
//GB2312 to UTF-8
cd = iconv_open("UTF-8", "GB2312");
if ((iconv_t)-1 == cd){
return false;
}
//conversion
iconv(cd, (const char**)&in, &src_len, (char**)&out, &dst_len);
//UTF-8 to Unicode
int utf8Len = strlen((char*)dst);
Uint16 unicodeData[CHAR_SIZE] = {0};
int unicodeLen = myUTF8_to_UNICODE(unicodeData, dst, utf8Len);
for (int i = 0; i < unicodeLen; i++) {
unicodeVectorArray.push_back(unicodeData[i]);
}
iconv_close(cd);
return true;
}
int myUTF8_to_UNICODE(Uint16* unicode, unsigned char* utf8, int len)
{
int length;
unsigned char* t = utf8;
length = 0;
while (utf8 - t < len){
//one byte.ASCII as a, b, c, 1, 2, 3 ect
if ( *(unsigned char *) utf8 <= 0x7f ) {
//expand with 0s.
*unicode++ = *utf8++;
}
//2 byte.
else if ( *(unsigned char *) utf8 <= 0xdf ) {
*unicode++ = ((*(unsigned char *) utf8 & 0x1f) << 6) + ((*(unsigned char *) (utf8 + 1)) & 0x3f);
utf8 += 2;
}
//3 byte.Chinese may use 3 byte.
else {
*unicode++ = ((int) (*(unsigned char *) utf8 & 0x0f) << 12) +
((*(unsigned char *) (utf8 + 1) & 0x3f) << 6) +
(*(unsigned char *) (utf8 + 2) & 0x3f);
utf8 += 3;
}
length++;
}
*unicode = 0;
return (length);
}
lib: iconv.lib
dll: iconv.dll
last update: 2008-05-12
posted on 2008-04-14 13:23
lf426 阅读(2544)
评论(0) 编辑 收藏 引用 所属分类:
mySDL_GameEngine