UTF-8与GB2312之间的互换

最新推荐文章于 2023-08-23 10:34:08 发布

zhanglidn013

最新推荐文章于 2023-08-23 10:34:08 发布

阅读量672

点赞数

分类专栏： c++编程

c++编程专栏收录该内容

92 篇文章 0 订阅

订阅专栏

相信一定有不少的程序开发人员时常会遇到字符编码的问题，而这个问题也是非常让人头痛的。因为这些都是潜在的错误，要找出这些错误也得要有这方面的开发经验才行。特别是在处理xml文档时，该问题的出现就更加的频繁了，有一次用java写服务器端程序，用vc写客户端与之交互。交互的协议都是用xml写的。结果在通讯时老是发现数据接受不正确。纳闷！于是用抓取网络数据包工具抓取数据，后来才发现原来是java上xml的头是这样的,而vc上默认的是GB2312。所以一遇到汉字数据就不正确了。去网上找资料，这方面的文章好象特别少，针对像这样的问题，下面我介绍一下我自己写的一个转换程序。当然，程序很简单。如果有画蛇添足的地方，还望各位高手一笑了之。

如果您对UTF-8、Unicode、GB2312等还是很陌生的话，请查看http://www.linuxforum.net/books/UTF-8-Unicode.html，我这里就不浪费口舌了。下面介绍一下WinAPI的两个函数：WideCharToMultiByte、MultiByteToWideChar。

函数原型：

 
   01. 
   int WideCharToMultiByte( 
  
   02. 
   UINT CodePage, // code page 
  
   03. 
   DWORD dwFlags, // performance and mapping flags 
  
   04. 
   LPCWSTR lpWideCharStr, // wide-character string 
  
   05. 
   int cchWideChar, // number of chars in string 
  
   06. 
   LPSTR lpMultiByteStr, // buffer for new string 
  
   07. 
   int cbMultiByte, // size of buffer 
  
   08. 
   LPCSTR lpDefaultChar, // default for unmappable chars 
  
   09. 
   LPBOOL lpUsedDefaultChar // set when default char used 
  
   10. 
   ); //将宽字符转换成多个窄字符 
  
   11. 
     
   12. 
   int MultiByteToWideChar( 
  
   13. 
   UINT CodePage, // code page 
  
   14. 
   DWORD dwFlags, // character-type options 
  
   15. 
   LPCSTR lpMultiByteStr, // string to map 
  
   16. 
   int cbMultiByte, // number of bytes in string 
  
   17. 
   LPWSTR lpWideCharStr, // wide-character buffer 
  
   18. 
   int cchWideChar // size of buffer 
  
   19. 
   );//将多个窄字符转换成宽字符

需要用到的一些函数：

 
   01. 
   CString CXmlProcess::HexToBin(CString string)//将16进制数转换成2进制 
  
   02. 
   { 
  
   03. 
   if( string == "0") return "0000"; 
  
   04. 
   if( string == "1") return "0001"; 
  
   05. 
   if( string == "2") return "0010"; 
  
   06. 
   if( string == "3") return "0011"; 
  
   07. 
   if( string == "4") return "0100"; 
  
   08. 
   if( string == "5") return "0101"; 
  
   09. 
   if( string == "6") return "0110"; 
  
   10. 
   if( string == "7") return "0111"; 
  
   11. 
   if( string == "8") return "1000"; 
  
   12. 
   if( string == "9") return "1001"; 
  
   13. 
   if( string == "a") return "1010"; 
  
   14. 
   if( string == "b") return "1011"; 
  
   15. 
   if( string == "c") return "1100"; 
  
   16. 
   if( string == "d") return "1101"; 
  
   17. 
   if( string == "e") return "1110"; 
  
   18. 
   if( string == "f") return "1111"; 
  
   19. 
   return ""; 
  
   20. 
   } 
  
   21. 
     
   22. 
     
   23. 
   CString CXmlProcess::BinToHex(CString BinString)//将2进制数转换成16进制 
  
   24. 
   { 
  
   25. 
   if( BinString == "0000") return "0"; 
  
   26. 
   if( BinString == "0001") return "1"; 
  
   27. 
   if( BinString == "0010") return "2"; 
  
   28. 
   if( BinString == "0011") return "3"; 
  
   29. 
   if( BinString == "0100") return "4"; 
  
   30. 
   if( BinString == "0101") return "5"; 
  
   31. 
   if( BinString == "0110") return "6"; 
  
   32. 
   if( BinString == "0111") return "7"; 
  
   33. 
   if( BinString == "1000") return "8"; 
  
   34. 
   if( BinString == "1001") return "9"; 
  
   35. 
   if( BinString == "1010") return "a"; 
  
   36. 
   if( BinString == "1011") return "b"; 
  
   37. 
   if( BinString == "1100") return "c"; 
  
   38. 
   if( BinString == "1101") return "d"; 
  
   39. 
   if( BinString == "1110") return "e"; 
  
   40. 
   if( BinString == "1111") return "f"; 
  
   41. 
   return ""; 
  
   42. 
   } 
  
   43. 
     
   44. 
   int CXmlProcess::BinToInt(CString string)//2进制字符数据转换成10进制整型 
  
   45. 
   { 
  
   46. 
   int len =0; 
  
   47. 
   int tempInt = 0; 
  
   48. 
   int strInt = 0; 
  
   49. 
   for(int i =0 ;i < string.GetLength() ;i ++) 
  
   50. 
   { 
  
   51. 
   tempInt = 1; 
  
   52. 
   strInt = (int)string.GetAt(i)-48; 
  
   53. 
   for(int k =0 ;k < 7-i ; k++) 
  
   54. 
   { 
  
   55. 
   tempInt = 2*tempInt; 
  
   56. 
   } 
  
   57. 
   len += tempInt*strInt; 
  
   58. 
   } 
  
   59. 
   return len; 
  
   60. 
   }

UTF-8转换成GB2312先把UTF-8转换成Unicode.然后再把Unicode通过函数WideCharToMultiByte转换成GB2312

 
   01. 
   WCHAR* CXmlProcess::UTF_8ToUnicode(char *ustart)  //把UTF-8转换成Unicode 
  
   02. 
   { 
  
   03. 
   char char_one; 
  
   04. 
   char char_two; 
  
   05. 
   char char_three; 
  
   06. 
   int Hchar; 
  
   07. 
   int Lchar; 
  
   08. 
   char uchar[2]; 
  
   09. 
   WCHAR *unicode; 
  
   10. 
   CString string_one; 
  
   11. 
   CString string_two; 
  
   12. 
   CString string_three; 
  
   13. 
   CString combiString; 
  
   14. 
   char_one = *ustart; 
  
   15. 
   char_two = *(ustart+1); 
  
   16. 
   char_three = *(ustart+2); 
  
   17. 
   string_one.Format("%x",char_one); 
  
   18. 
   string_two.Format("%x",char_two); 
  
   19. 
   string_three.Format("%x",char_three); 
  
   20. 
   string_three = string_three.Right(2); 
  
   21. 
   string_two = string_two.Right(2); 
  
   22. 
   string_one = string_one.Right(2); 
  
   23. 
   string_three = HexToBin(string_three.Left(1))+HexToBin(string_three.Right(1)); 
  
   24. 
   string_two = HexToBin(string_two.Left(1))+HexToBin(string_two.Right(1)); 
  
   25. 
   string_one = HexToBin(string_one.Left(1))+HexToBin(string_one.Right(1)); 
  
   26. 
   combiString = string_one +string_two +string_three; 
  
   27. 
   combiString = combiString.Right(20); 
  
   28. 
   combiString.Delete(4,2); 
  
   29. 
   combiString.Delete(10,2); 
  
   30. 
   Hchar = BinToInt(combiString.Left(8)); 
  
   31. 
   Lchar = BinToInt(combiString.Right(8)); 
  
   32. 
   uchar[1] = (char)Hchar; 
  
   33. 
   uchar[0] = (char)Lchar; 
  
   34. 
   unicode = (WCHAR *)uchar; 
  
   35. 
   return unicode; 
  
   36. 
   } 
  
   37. 
     
   38. 
   char * CXmlProcess::UnicodeToGB2312(unsigned short uData)  //把Unicode 转换成 GB2312 
  
   39. 
   { 
  
   40. 
   char *buffer ; 
  
   41. 
   buffer = new char[sizeof(WCHAR)]; 
  
   42. 
   WideCharToMultiByte(CP_ACP,NULL,&uData,1,buffer,sizeof(WCHAR),NULL,NULL); 
  
   43. 
   return buffer; 
  
   44. 
   }

GB2312转换成UTF-8：先把GB2312通过函数MultiByteToWideChar转换成Unicode.然后再把Unicode通过拆开Unicode后拼装成UTF-8。

 
   01. 
   WCHAR * CXmlProcess::Gb2312ToUnicode(char *gbBuffer)  //GB2312 转换成　Unicode 
  
   02. 
   { 
  
   03. 
   WCHAR *uniChar; 
  
   04. 
   uniChar = new WCHAR[1]; 
  
   05. 
   ::MultiByteToWideChar(CP_ACP,MB_PRECOMPOSED,gbBuffer,2,uniChar,1); 
  
   06. 
   return uniChar; 
  
   07. 
   } 
  
   08. 
   char * CXmlProcess::UnicodeToUTF_8(WCHAR *UniChar) // Unicode 转换成UTF-8 
  
   09. 
   { 
  
   10. 
   char *buffer; 
  
   11. 
   CString strOne; 
  
   12. 
   CString strTwo; 
  
   13. 
   CString strThree; 
  
   14. 
   CString strFour; 
  
   15. 
   CString strAnd; 
  
   16. 
   buffer = new char[3]; 
  
   17. 
   int hInt,lInt; 
  
   18. 
   hInt = (int)((*UniChar)/256); 
  
   19. 
   lInt = (*UniChar)%256; 
  
   20. 
   CString string ; 
  
   21. 
   string.Format("%x",hInt); 
  
   22. 
   strTwo = HexToBin(string.Right(1)); 
  
   23. 
   string = string.Left(string.GetLength() - 1); 
  
   24. 
   strOne = HexToBin(string.Right(1)); 
  
   25. 
   string.Format("%x",lInt); 
  
   26. 
   strFour = HexToBin(string.Right(1)); 
  
   27. 
   string = string.Left(string.GetLength() -1); 
  
   28. 
   strThree = HexToBin(string.Right(1)); 
  
   29. 
   strAnd = strOne +strTwo + strThree + strFour; 
  
   30. 
   strAnd.Insert(0,"1110"); 
  
   31. 
   strAnd.Insert(8,"10"); 
  
   32. 
   strAnd.Insert(16,"10"); 
  
   33. 
   strOne = strAnd.Left(8); 
  
   34. 
   strAnd = strAnd.Right(16); 
  
   35. 
   strTwo = strAnd.Left(8); 
  
   36. 
   strThree = strAnd.Right(8); 
  
   37. 
   *buffer = (char)BinToInt(strOne); 
  
   38. 
   buffer[1] = (char)BinToInt(strTwo); 
  
   39. 
   buffer[2] = (char)BinToInt(strThree); 
  
   40. 
   return buffer; 
  
   41. 
   }

例子：将GB2312转换成UTF-8的调用：

 
   01. 
   char * CXmlProcess::translateCharToUTF_8(char *xmlStream, int len) 
  
   02. 
   { 
  
   03. 
   int newCharLen =0 ; 
  
   04. 
   int oldCharLen = 0; 
  
   05. 
   int revCharLen = len; 
  
   06. 
   char* newCharBuffer; 
  
   07. 
   char* finalCharBuffer; 
  
   08. 
   char *buffer ; 
  
   09. 
   CString string; 
  
   10. 
   buffer  = new char[sizeof(WCHAR)]; 
  
   11. 
   newCharBuffer = new char[int(1.5*revCharLen)];//设置最大的一个缓冲区 
  
   12. 
   while(oldCharLen < revCharLen) 
  
   13. 
   { 
  
   14. 
   if( *(xmlStream + oldCharLen) >= 0) 
  
   15. 
   { 
  
   16. 
   *(newCharBuffer+newCharLen) = *(xmlStream +oldCharLen); 
  
   17. 
   newCharLen ++; 
  
   18. 
   oldCharLen ++; 
  
   19. 
   }//如果是英文直接复制就可以 
  
   20. 
   else 
  
   21. 
   { 
  
   22. 
   WCHAR *pbuffer = this->Gb2312ToUnicode(xmlStream+oldCharLen); 
  
   23. 
   buffer = this->UnicodeToUTF_8(pbuffer); 
  
   24. 
   *(newCharBuffer+newCharLen) = *buffer; 
  
   25. 
   *(newCharBuffer +newCharLen +1) = *(buffer + 1); 
  
   26. 
   *(newCharBuffer +newCharLen +2) = *(buffer + 2); 
  
   27. 
   newCharLen += 3; 
  
   28. 
   oldCharLen += 2; 
  
   29. 
   } 
  
   30. 
   } 
  
   31. 
   newCharBuffer[newCharLen] = ''\0''; 
  
   32. 
   CString string1 ; 
  
   33. 
   string1.Format("%s",newCharBuffer); 
  
   34. 
   finalCharBuffer = new char[newCharLen+1]; 
  
   35. 
   memcpy(finalCharBuffer,newCharBuffer,newCharLen+1); 
  
   36. 
   return finalCharBuffer; 
  
   37. 
   }