cnvchar.cpp

This software performs code conversion of Chinese characters, including GB2312/GBK and BIG5. It a

// cnvchar.cpp

#include <windows.h>
#include "msc.h"
#include "cnvchar.h"
#include "codepage.h"


#define IS_ALLOW_FOURBYTE  0     // 係僶僀僩暥帤偵懳墳偡傞偐

#define BUFFER_SIZE     8




// Unicode傪巊偭偰暥帤傪曄姺偡傞
// 摼傜傟偨暥帤傪曉偡
// 僄儔乕側傜 0 傪曉偡
int ConvertCharacter ( int C, int nSrcCodePage, int nDstCodePage ) {

   char szBufferA [ BUFFER_SIZE ] ;
   wchar_t szBufferW [ BUFFER_SIZE ] ;
   int nLen ;


   if ( nSrcCodePage == CP_UTF16 ) {
      if ( ! ( C & 0xFFFF0000 ) ) {
         szBufferW [ 0 ] = C & 0xFFFF ;
         nLen = 1 ;
      }
      else {
         if ( IS_ALLOW_FOURBYTE ) {
            szBufferW [ 0 ] = ( C >> 16 ) & 0xFFFF ;
            szBufferW [ 1 ] = C & 0xFFFF ;
            nLen = 2 ;
         }
         else return 0 ;
      }
   }
   else {

      if ( nSrcCodePage == CP_CHINESE_HKSCS ) nSrcCodePage = CP_CHINESE_TRADITIONAL ;

      if ( ! ( C & 0xFFFFFF00 ) ) {
         szBufferA [ 0 ] = C & 0xFF ;
         nLen = 1 ;
      }
      else if ( ! ( C & 0xFFFF0000 ) ) {
         szBufferA [ 0 ] = ( C >> 8 ) & 0xFF ;
         szBufferA [ 1 ] = C & 0xFF ;
         nLen = 2 ;
      }
      else {
         if ( IS_ALLOW_FOURBYTE ) {
            if ( ! ( C & 0xFF0000 ) ) {
               szBufferA [ 0 ] = ( C >> 16 ) & 0xFF ;
               szBufferA [ 1 ] = ( C >> 8 ) & 0xFF ;
               szBufferA [ 2 ] = C & 0xFF ;
               nLen = 3 ;
            }
            else {
               szBufferA [ 0 ] = ( C >> 24 ) & 0xFF ;
               szBufferA [ 1 ] = ( C >> 16 ) & 0xFF ;
               szBufferA [ 2 ] = ( C >> 8 ) & 0xFF ;
               szBufferA [ 3 ] = C & 0xFF ;
               nLen = 4 ;
            }
         }
         else return 0 ;
      }

      nLen = MultiByteToWideChar ( nSrcCodePage, 0, szBufferA, nLen, szBufferW, BUFFER_SIZE ) ;
      if ( nLen == 0 ) return 0 ;
      if ( nLen > ( IS_ALLOW_FOURBYTE ? 2 : 1 ) ) return 0 ;

      if ( *szBufferW == L'?' && C != '?' && ( nSrcCodePage != CP_US_ASCII || ( C & 0x7F ) != '?' ) ) return 0 ;
      if ( *szBufferW == 0x30FB ) {
         if ( nSrcCodePage == CP_JAPANESE && C != 0x8145 ) return 0 ;
         if ( nSrcCodePage == CP_EUC_JAPANESE_MS && C != 0xA1A6 ) return 0 ;
      }
   }


   if ( ! IS_ALLOW_FOURBYTE || nLen == 1 ) {
      if ( IsSurrogate ( *szBufferW ) ) return 0 ;
   }


   if ( nDstCodePage == CP_UTF16 ) {
      if ( IS_ALLOW_FOURBYTE ) {
         if ( nLen == 2 ) return ( szBufferW [ 0 ] << 16 ) + szBufferW [ 1 ] ;
      }
      return szBufferW [ 0 ] ;
   }


   if ( nDstCodePage == CP_CHINESE_HKSCS ) nDstCodePage = CP_CHINESE_TRADITIONAL ;

   int IsUsedDefaultChar ;
   nLen = WideCharToMultiByte ( nDstCodePage, 0, szBufferW, nLen, szBufferA, BUFFER_SIZE, NULL, & IsUsedDefaultChar ) ;
   if ( IsUsedDefaultChar ) return 0 ;

   if ( nLen == 1 ) return szBufferA [ 0 ] & 0xFF ;
   if ( nLen == 2 ) return ( ( szBufferA [ 0 ] & 0xFF ) << 8 ) + ( szBufferA [ 1 ] & 0xFF ) ;
   if ( IS_ALLOW_FOURBYTE ) {
      if ( nLen == 3 ) return ( ( szBufferA [ 0 ] & 0xFF ) << 16 ) + ( ( szBufferA [ 1 ] & 0xFF ) << 8 ) + ( szBufferA [ 2 ] & 0xFF ) ;
      if ( nLen == 4 ) return ( ( szBufferA [ 0 ] & 0xFF ) << 24 ) + ( ( szBufferA [ 1 ] & 0xFF ) << 16 ) + ( ( szBufferA [ 2 ] & 0xFF ) << 8 ) + ( szBufferA [ 3 ] & 0xFF ) ;
   }


   return 0 ;
}