libjava/gnu/gcj/convert/natInput_SJIS.cc

   1 /* Copyright (C) 1999  Free Software Foundation
   2
   3    This file is part of libgcj.
   4
   5 This software is copyrighted work licensed under the terms of the
   6 Libgcj License.  Please consult the file "LIBGCJ_LICENSE" for
   7 details.  */
   8
   9 #include <config.h>
  10 #include <gcj/cni.h>
  11 #include <gnu/gcj/convert/Input_SJIS.h>
  12
  13 #define ERROR_CHAR 0xFFFD
  14
  15 extern unsigned short JIS0208_to_Unicode[84][94];
  16 extern unsigned short JIS0212_to_Unicode[76][94];
  17
  18 jint
  19 gnu::gcj::convert::Input_SJIS::read(jcharArray outbuffer, jint outpos,
  20                                     jint count)
  21 {
  22   jint start_outpos = outpos;
  23   for (;;)
  24     {
  25       if (outpos - start_outpos >= count)
  26         break;
  27       if (inpos >= inlength)
  28         break;
  29       int b = ((unsigned char*) elements(inbuffer))[inpos++];
  30       if (first_byte == 0)
  31         {
  32           if (b < 128)
  33             {
  34 #if 1
  35               // Technically, we should translate 0x5c to Yen symbol;
  36               // in practice, it is not clear.
  37               if (b == 0x5c)
  38                 b = 0x00A5;  // Yen sign.
  39 #endif
  40               elements(outbuffer)[outpos++] = (char) b;
  41             }
  42           else if (b >= 0xA1 && b <= 0xDF)
  43             {
  44               b += 0xFF61 - 0xA1;
  45               elements(outbuffer)[outpos++] = b;
  46             }
  47           else
  48             first_byte = b;
  49         }
  50       else
  51         {
  52           // From Lunde: "CJKV Informatio Processing", O'Reilly, 1999, p 420:
  53           bool adjust = b < 159;
  54           int rowOffset = first_byte < 160 ? 112 : 176;
  55           int cellOffset = adjust ? (b > 127 ? 32 : 31) : 126;
  56           first_byte = ((first_byte - rowOffset) << 1) - adjust;
  57           b -= cellOffset;
  58
  59           first_byte -= 33;
  60           b -= 33;
  61
  62           if ((unsigned) first_byte >= 84 || (unsigned) b >= 94)
  63             b = ERROR_CHAR;
  64           else
  65             {
  66               b = JIS0208_to_Unicode[first_byte][b];
  67               if (b == 0)
  68                 b = ERROR_CHAR;
  69             }
  70           elements(outbuffer)[outpos++] = b;
  71
  72           first_byte = 0;
  73         }
  74     }
  75   return outpos - start_outpos;
  76 }