程序博客网 > 矢量地图数据下载

GBK与UTF-8转化

来源：互联网 发布：矢量地图数据下载 编辑：程序博客网 时间：2024/05/22 03:07

1、
1)tools.h
#include <windows.h>
#include <string>

/************************************************************************/
/* 将gb2312编码转为UTF-8字符                                             */
/************************************************************************/
std::string GBToUTF8(const char*);
/************************************************************************/
/* 将UTF-8编码转为gb2312字符                                             */
/************************************************************************/
std::string UTF8ToGB(const char*);

2)tools.cpp
#include "tools.h"

std::string GBToUTF8(const char* str)
{
     std::string result;
     WCHAR *strSrc;
     TCHAR *szRes;

     //获得临时变量的大小
     int i = MultiByteToWideChar(CP_ACP, 0, str, -1, NULL, 0);
     strSrc = new WCHAR[i+1];
     MultiByteToWideChar(CP_ACP, 0, str, -1, strSrc, i);

     //获得临时变量的大小
     i = WideCharToMultiByte(CP_UTF8, 0, strSrc, -1, NULL, 0, NULL, NULL);
     szRes = new TCHAR[i+1];
     int j=WideCharToMultiByte(CP_UTF8, 0, strSrc, -1, szRes, i, NULL, NULL);

     result = szRes;
     delete []strSrc;
     delete []szRes;

     return result;
}

std::string UTF8ToGB(const char* str)
{
     std::string result;
     WCHAR *strSrc;
     TCHAR *szRes;

     //获得临时变量的大小
     int i = MultiByteToWideChar(CP_UTF8, 0, str, -1, NULL, 0);
     strSrc = new WCHAR[i+1];
     MultiByteToWideChar(CP_UTF8, 0, str, -1, strSrc, i);

     //获得临时变量的大小
     i = WideCharToMultiByte(CP_ACP, 0, strSrc, -1, NULL, 0, NULL, NULL);
     szRes = new TCHAR[i+1];
     WideCharToMultiByte(CP_ACP, 0, strSrc, -1, szRes, i, NULL, NULL);

     result = szRes;
     delete []strSrc;
     delete []szRes;

     return result;
}

附注：
const char * = std::string.c_str

2、
1)tools.h
#include <string>
#include <afx.h>

/************************************************************************/
/* 将gb2312编码转为UTF-8字符                                             */
/************************************************************************/
std::string GBToUTF8(CString);
/************************************************************************/
/* 将UTF-8编码转为gb2312字符                                             */
/************************************************************************/
CString UTF8ToGB(std::string*);

2)tools.cpp
#include "tools.h"

std::string GBToUTF8(CString cstr)
{
     std::string result;
     WCHAR *strSrc;
     TCHAR *szRes;
     const char* str = (LPCTSTR)cstr;

     //获得临时变量的大小
     int i = MultiByteToWideChar(CP_ACP, 0, str, -1, NULL, 0);
     strSrc = new WCHAR[i+1];
     MultiByteToWideChar(CP_ACP, 0, str, -1, strSrc, i);

     //获得临时变量的大小
     i = WideCharToMultiByte(CP_UTF8, 0, strSrc, -1, NULL, 0, NULL, NULL);
     szRes = new TCHAR[i+1];
     int j=WideCharToMultiByte(CP_UTF8, 0, strSrc, -1, szRes, i, NULL, NULL);

     result = szRes;
     delete []strSrc;
     delete []szRes;

     return result;
}

CString UTF8ToGB(std::string* sstr)
{
     CString result;
     WCHAR *strSrc;
     TCHAR *szRes;
     const char* str = sstr->c_str();

     //获得临时变量的大小
     int i = MultiByteToWideChar(CP_UTF8, 0, str, -1, NULL, 0);
     strSrc = new WCHAR[i+1];
     MultiByteToWideChar(CP_UTF8, 0, str, -1, strSrc, i);

     //获得临时变量的大小
     i = WideCharToMultiByte(CP_ACP, 0, strSrc, -1, NULL, 0, NULL, NULL);
     szRes = new TCHAR[i+1];
     WideCharToMultiByte(CP_ACP, 0, strSrc, -1, szRes, i, NULL, NULL);

     result = szRes;
     delete []strSrc;
     delete []szRes;

     return result;
}


3、
CString EncodeToUTF8(LPCTSTR szSource)
{
     WORD ch;

     BYTE bt1, bt2, bt3, bt4, bt5, bt6;

     int n, nMax = _tcslen(szSource);

     CString sFinal, sTemp;

     for (n = 0; n < nMax; ++n)
     {
         ch = (WORD)szSource[n];

         if (ch == _T('='))
         {
             sTemp.Format(_T("=%02X"), ch);

             sFinal += sTemp;
         }
         else if (ch < 128)
         {
             sFinal += szSource[n];
         }
         else if (ch <= 2047)
         {
              bt1 = (BYTE)(192 + (ch / 64));
              bt2 = (BYTE)(128 + (ch % 64));

             sTemp.Format(_T("=%02X=%02X"), bt1, bt2);
                        
             sFinal += sTemp;
         }
         else if (ch <= 65535)
         {
              bt1 = (BYTE)(224 + (ch / 4096));
              bt2 = (BYTE)(128 + ((ch / 64) % 64));
              bt3 = (BYTE)(128 + (ch % 64));

             sTemp.Format(_T("=%02X=%02X=%02X"), bt1, bt2, bt3);
                        
             sFinal += sTemp;
         }
         else if (ch <= 2097151)
         {
              bt1 = (BYTE)(240 + (ch / 262144));
              bt2 = (BYTE)(128 + ((ch / 4096) % 64));
              bt3 = (BYTE)(128 + ((ch / 64) % 64));
              bt4 = (BYTE)(128 + (ch % 64));

             sTemp.Format(_T("=%02X=%02X=%02X=%02X"), bt1, bt2, bt3, bt4);
             sFinal += sTemp;
         }
         else if (ch <=67108863)
         {
             bt1 = (BYTE)(248 + (ch / 16777216));
             bt2 = (BYTE)(128 + ((ch / 262144) % 64));
             bt3 = (BYTE)(128 + ((ch / 4096) % 64));
             bt4 = (BYTE)(128 + ((ch / 64) % 64));
             bt5 = (BYTE)(128 + (ch % 64));

             sTemp.Format(_T("=%02X=%02X=%02X=%02X=%02X"), bt1, bt2, bt3, bt4, bt5);
             sFinal += sTemp;
         }
         else if (ch <=2147483647)
         {
              bt1 = (BYTE)(252 + (ch / 1073741824));
              bt2 = (BYTE)(128 + ((ch / 16777216) % 64));
              bt3 = (BYTE)(128 + ((ch / 262144) % 64));
              bt4 = (BYTE)(128 + ((ch / 4096) % 64));
              bt5 = (BYTE)(128 + ((ch / 64) % 64));
              bt6 = (BYTE)(128 + (ch % 64));

             sTemp.Format(_T("=%02X=%02X=%02X=%02X=%02X=%02X"), bt1, bt2, bt3, bt4, bt5, bt6);
             sFinal += sTemp;
         }

     }

     return sFinal;
}
CString DecodeFromUTF8(LPCTSTR szSource)
{
     TRACE(_T("\n"));

     int n, nMax = _tcslen(szSource);
     WORD ch;

     CString sFinal, sTemp;

     BYTE z, y, x, w, v, u;
    
     for (n = 0; n < nMax; ++n)
     {
         ch = (WORD)szSource[n];

         if (ch != _T('='))
         {
             sFinal += (TCHAR)ch;
             continue;
         }

         if (n >= nMax - 2) break; // something is wrong
         z = MakeByte(szSource[n+1], szSource[n+2]);

         if (z < 127)
         {
             sFinal += (TCHAR)z;
             n = n + 2;
         }
         else if (z >= 192 && z <= 223)
         {
             // character is two bytes
             if (n >= nMax - 5) break; // something is wrong
             y = MakeByte(szSource[n+4], szSource[n+5]);
             sFinal += (TCHAR)( (z-192)*64 + (y-128) );
             n = n + 5;
         }
         else if (z >= 224 && z <= 239)
         {
             // character is three bytes
             if (n >= nMax - 8) break; // something is wrong
             y = MakeByte(szSource[n+4], szSource[n+5]);
             x = MakeByte(szSource[n+7], szSource[n+8]);
             sFinal += (TCHAR)( (z-224)*4096 + (y-128)*64 + (x-128) );
             n = n + 8;
         }
         else if (z >= 240 && z <= 247)
         {
             // character is four bytes
             if (n >= nMax - 11) break; // something is wrong
             y = MakeByte(szSource[n+4], szSource[n+5]);
             x = MakeByte(szSource[n+7], szSource[n+8]);
             w = MakeByte(szSource[n+10], szSource[n+11]);
             sFinal += (TCHAR)( (z-240)*262144 + (y-128)*4096 + (x-128)*64 + (w-128) );
             n = n + 11;
         }
         else if (z >= 248 && z <= 251)
         {
             // character is four bytes
             if (n >= nMax - 14) break; // something is wrong
             y = MakeByte(szSource[n+4], szSource[n+5]);
             x = MakeByte(szSource[n+7], szSource[n+8]);
             w = MakeByte(szSource[n+10], szSource[n+11]);
             v = MakeByte(szSource[n+13], szSource[n+14]);
             sFinal += (TCHAR)( (z-248)*16777216 + (y-128)*262144 + (x-128)*4096 + (w-128)*64 + (v-128) );
             n = n + 14;
         }
         else if (z >= 252 && z <= 253)
         {
             // character is four bytes
             if (n >= nMax - 17) break; // something is wrong
             y = MakeByte(szSource[n+4], szSource[n+5]);
             x = MakeByte(szSource[n+7], szSource[n+8]);
             w = MakeByte(szSource[n+10], szSource[n+11]);
             v = MakeByte(szSource[n+13], szSource[n+14]);
             u = MakeByte(szSource[n+16], szSource[n+17]);
             sFinal += (TCHAR)( (z-252)*1073741824 + (y-128)*16777216 + (x-128)*262144 + (w-128)*4096 + (v-128)*64 + (u-128) );
             n = n + 17;
         }
        
     }

     return sFinal;
}
// helper function for decoding
BYTE MakeByte(TCHAR ch1, TCHAR ch2)
{
     BYTE bt1 = 0, bt2 = 0;

     switch (ch2)
     {
         case _T('0'):
             bt2 = 0x00;
             break;
         case _T('1'):
             bt2 = 0x01;
             break;
         case _T('2'):
             bt2 = 0x02;
             break;
         case _T('3'):
             bt2 = 0x03;
             break;
         case _T('4'):
             bt2 = 0x04;
             break;
         case _T('5'):
             bt2 = 0x05;
             break;
         case _T('6'):
             bt2 = 0x06;
             break;
         case _T('7'):
             bt2 = 0x07;
             break;
         case _T('8'):
             bt2 = 0x08;
             break;
         case _T('9'):
             bt2 = 0x09;
             break;
         case _T('A'):
             bt2 = 0x0A;
             break;
         case _T('B'):
             bt2 = 0x0B;
             break;
         case _T('C'):
             bt2 = 0x0C;
             break;
         case _T('D'):
             bt2 = 0x0D;
             break;
         case _T('E'):
             bt2 = 0x0E;
             break;
         case _T('F'):
             bt2 = 0x0F;
             break;
     }

     switch (ch1)
     {
         case _T('0'):
             bt1 = 0x00;
             break;
         case _T('1'):
             bt1 = 0x10;
             break;
         case _T('2'):
             bt1 = 0x20;
             break;
         case _T('3'):
             bt1 = 0x30;
             break;
         case _T('4'):
             bt1 = 0x40;
             break;
         case _T('5'):
             bt1 = 0x50;
             break;
         case _T('6'):
             bt1 = 0x60;
             break;
         case _T('7'):
             bt1 = 0x70;
             break;
         case _T('8'):
             bt1 = 0x80;
             break;
         case _T('9'):
             bt1 = 0x90;
             break;
         case _T('A'):
             bt1 = 0xA0;
             break;
         case _T('B'):
             bt1 = 0xB0;
             break;
         case _T('C'):
             bt1 = 0xC0;
             break;
         case _T('D'):
             bt1 = 0xD0;
             break;
         case _T('E'):
             bt1 = 0xE0;
             break;
         case _T('F'):
             bt1 = 0xF0;
             break;
     }

     BYTE btFinal = bt2 | bt1;

     return   btFinal;    

 

引用地址：http://hi.baidu.com/jhzlf/blog/item/54a1ea2595b7d322d5074215.html

附录：

网上看到的一点东东：

要实现 GB2312 （其实是GBK）转换为 UTF-8 其实很简单，先用 MultiByteToWideChar 把 GB2312 转换为 Unicode，再用 WideCharToMultiByte 把 Unicode 转换为 UTF-8 就可以了。

UTF-8 转换为 GB2312 是个相反的过程，先用 MultiByteToWideChar 把 UTF-8 转换为 Unicode，再用 WideCharToMultiByte 把 Unicode 转换为 GB2312 就可以了。 


矢量地图数据下载 矢量地图数据下载
原创粉丝点击
热门IT博客
在家编程赚钱在家编程赚钱
抢魅族软件下载抢魅族软件下载
java json日期格式化
windows编程
linux搭建lamp环境
ultraiso mac版下载
被网络诈骗了怎么报警
能耗模拟软件
js反混淆
将军绝命时刻单位数据
mac pro 重装系统教程
淘宝购物有发票吗
白金数据小说
note station windows
北明软件 中标
非线性优化算法有哪些
淘宝客服工资有6000吗
淘宝汉服元素店铺推荐
软件开发人日费用
360软件管家官方版
热门问题 老师的惩罚 人脸识别 我在镇武司摸鱼那些年 重生之率土为王 我在大康的咸鱼生活 盘龙之生命进化 天生仙种 凡人之先天五行 春回大明朝 姑娘不必设防，我是瞎子 女起名 韩国女团 sing女团 女团 女团舞蹈 kda女团 101女团 gidle女团 kda女团图片 优我女团 女团热舞 女团腿 女团名字 itzy女团 slng女团 韩女团开阴无内裤全集 韩国女团没内裤慢放放大 韩国女团舞蹈 韩国女团妆容 韩国女团应援色 韩国女团开阴高清慢放 韩国女团排名 男爱豆跳女团舞 巨星竟然女团开始 穿到古代带女团 巨星竟然从女团开始 韩国女团无内裤慢放版 韩国女团热舞 巨星女团开始 gidle女团成员介绍 巨星从女团开始 韩国女团没内裤超清版 gidle女团叶舒华 奶奶跳舞不输女团 中国队获亚锦赛女团七连冠 西凉女国生肖 西凉女国打一生肖 古穿今女国师日常 西凉女国什么生肖 西凉女国 西梁女国

程序博客网，程序员的互联网技术博客家园。csdn论坛精品 msdn技术资料都在这里