KernelLand/Kernel/utf16.c

   1 /*
   2  * Acess2 Kernel
   3  * - By John Hodge (thePowersGang)
   4  *
   5  * utf16.c
   6  * - UTF-16 Translation/Manipulation
   7  */
   8 #define DEBUG   1
   9 #include <acess.h>
  10 #include <utf16.h>
  11
  12 int ReadUTF16(const Uint16 *Str16, Uint32 *Codepoint)
  13 {
  14         if( 0xD800 < *Str16 && *Str16 <= 0xDFFF )
  15         {
  16                 // UTF-16 surrogate pair
  17                 // > 0xDC00 is the second word
  18                 if( Str16[0] > 0xDC00 ) {
  19                         *Codepoint = 0;
  20                         return 1;
  21                 }
  22                 if( Str16[1] < 0xD800 || Str16[1] >= 0xDC00 ) {
  23                         *Codepoint = 0;
  24                         return 2;
  25                 }
  26                 // 2^16 + 20-bit
  27                 *Codepoint = 0x10000 + (((Str16[0] & 0x3FF) << 10) | (Str16[1] & 0x3FF));
  28                 return 2;
  29         }
  30         else {
  31                 *Codepoint = *Str16;
  32                 return 1;
  33         }
  34 }
  35
  36 size_t UTF16_ConvertToUTF8(size_t DestLen, char *Dest, size_t SrcLen, const Uint16 *Source)
  37 {
  38          int    len = 0;
  39         for( ; *Source && SrcLen --; Source ++ )
  40         {
  41                 // TODO: Decode/Reencode
  42                 if( Dest && len < DestLen )
  43                         Dest[len] = *Source;
  44                 len += 1;
  45         }
  46         if( Dest && len < DestLen )
  47                 Dest[len] = 0;
  48         return len;
  49 }
  50
  51 int UTF16_CompareWithUTF8(size_t Str16Len, const Uint16 *Str16, const char *Str8)
  52 {
  53          int    pos16 = 0, pos8 = 0;
  54         const Uint8     *str8 = (const Uint8 *)Str8;
  55
  56         while( pos16 < Str16Len && Str16[pos16] && str8[pos8] )
  57         {
  58                 Uint32  cp8, cp16;
  59                 pos16 += ReadUTF16(Str16+pos16, &cp16);
  60                 pos8 += ReadUTF8(str8 + pos8, &cp8);
  61
  62                 LOG("cp16 = %x, cp8 = %x", cp16, cp8);
  63                 if(cp16 == cp8) continue ;
  64
  65                 if(cp16 < cp8)
  66                         return -1;
  67                 else
  68                         return 1;
  69         }
  70         if( pos16 == Str16Len )
  71                 return 0;
  72         if( Str16[pos16] && str8[pos8] )
  73                 return 0;
  74         if( Str16[pos16] )
  75                 return 1;
  76         else
  77                 return -1;
  78 }
  79