michael@0: // UTFConvert.cpp
michael@0: 
michael@0: #include "StdAfx.h"
michael@0: 
michael@0: #include "UTFConvert.h"
michael@0: #include "Types.h"
michael@0: 
michael@0: static Byte kUtf8Limits[5] = { 0xC0, 0xE0, 0xF0, 0xF8, 0xFC };
michael@0: 
michael@0: // These functions are for UTF8 <-> UTF16 conversion.
michael@0: 
michael@0: bool ConvertUTF8ToUnicode(const AString &src, UString &dest)
michael@0: {
michael@0:   dest.Empty();
michael@0:   for(int i = 0; i < src.Length();)
michael@0:   {
michael@0:     Byte c = (Byte)src[i++];
michael@0:     if (c < 0x80)
michael@0:     {
michael@0:       dest += (wchar_t)c;
michael@0:       continue;
michael@0:     }
michael@0:     if(c < 0xC0)
michael@0:       return false;
michael@0:     int numAdds;
michael@0:     for (numAdds = 1; numAdds < 5; numAdds++)
michael@0:       if (c < kUtf8Limits[numAdds])
michael@0:         break;
michael@0:     UInt32 value = (c - kUtf8Limits[numAdds - 1]);
michael@0:     do
michael@0:     {
michael@0:       if (i >= src.Length())
michael@0:         return false;
michael@0:       Byte c2 = (Byte)src[i++];
michael@0:       if (c2 < 0x80 || c2 >= 0xC0)
michael@0:         return false;
michael@0:       value <<= 6;
michael@0:       value |= (c2 - 0x80);
michael@0:       numAdds--;
michael@0:     }
michael@0:     while(numAdds > 0);
michael@0:     if (value < 0x10000)
michael@0:       dest += (wchar_t)(value);
michael@0:     else
michael@0:     {
michael@0:       value -= 0x10000;
michael@0:       if (value >= 0x100000)
michael@0:         return false;
michael@0:       dest += (wchar_t)(0xD800 + (value >> 10));
michael@0:       dest += (wchar_t)(0xDC00 + (value & 0x3FF));
michael@0:     }
michael@0:   }
michael@0:   return true; 
michael@0: }
michael@0: 
michael@0: bool ConvertUnicodeToUTF8(const UString &src, AString &dest)
michael@0: {
michael@0:   dest.Empty();
michael@0:   for(int i = 0; i < src.Length();)
michael@0:   {
michael@0:     UInt32 value = (UInt32)src[i++];
michael@0:     if (value < 0x80)
michael@0:     {
michael@0:       dest += (char)value;
michael@0:       continue;
michael@0:     }
michael@0:     if (value >= 0xD800 && value < 0xE000)
michael@0:     {
michael@0:       if (value >= 0xDC00)
michael@0:         return false;
michael@0:       if (i >= src.Length())
michael@0:         return false;
michael@0:       UInt32 c2 = (UInt32)src[i++];
michael@0:       if (c2 < 0xDC00 || c2 >= 0xE000)
michael@0:         return false;
michael@0:       value = ((value - 0xD800) << 10) | (c2 - 0xDC00);
michael@0:     }
michael@0:     int numAdds;
michael@0:     for (numAdds = 1; numAdds < 5; numAdds++)
michael@0:       if (value < (((UInt32)1) << (numAdds * 5 + 6)))
michael@0:         break;
michael@0:     dest += (char)(kUtf8Limits[numAdds - 1] + (value >> (6 * numAdds)));
michael@0:     do
michael@0:     {
michael@0:       numAdds--;
michael@0:       dest += (char)(0x80 + ((value >> (6 * numAdds)) & 0x3F));
michael@0:     }
michael@0:     while(numAdds > 0);
michael@0:   }
michael@0:   return true;
michael@0: }