source/stringcodec/iso2022jp.cpp

   1 /* $Id$
   2
   3 This file is part of libmspstrings
   4 Copyright © 2006-2007 Mikko Rasa
   5 Distributed under the LGPL
   6 */
   7 #include "ascii.h"
   8 #include "iso2022jp.h"
   9 #include "jisx0201.h"
  10 #include "jisx0208.h"
  11
  12 using namespace std;
  13
  14 namespace Msp {
  15 namespace Codecs {
  16
  17 void Iso2022Jp::Encoder::encode_char(UnicodeChar ch, string &buf)
  18 {
  19         if(ch>=0 && ch<=0x7F && ch!=0x5C && ch!=0x7E)
  20         {
  21                 if(mode!=ASCII && mode!=JISX0201)
  22                         switch_mode(ASCII, buf);
  23                 buf += ch;
  24         }
  25         else if(ch==0x5C || ch==0x7E)
  26         {
  27                 if(mode!=ASCII)
  28                         switch_mode(ASCII, buf);
  29                 buf += ch;
  30         }
  31         else if(ch==0xA5 || ch==0x203E)
  32         {
  33                 if(mode!=JISX0201)
  34                         switch_mode(JISX0201, buf);
  35                 if(ch==0xA5)
  36                         buf += 0x5C;
  37                 else if(ch==0x203E)
  38                         buf += 0x7E;
  39         }
  40         else
  41         {
  42                 Kuten jis = ucs_to_jisx0208(ch);
  43                 if(!jis)
  44                         return error(ch, buf, "Can't express character in ISO-2022-JP");
  45
  46                 if(mode!=JISX0208)
  47                         switch_mode(JISX0208, buf);
  48
  49                 char jbuf[2] = {jis.ku+0x20, jis.ten+0x20};
  50                 buf.append(jbuf, 2);
  51         }
  52 }
  53
  54 void Iso2022Jp::Encoder::sync(string &buf)
  55 {
  56         if(mode!=ASCII)
  57                 switch_mode(ASCII, buf);
  58 }
  59
  60 void Iso2022Jp::Encoder::reset()
  61 {
  62         mode = ASCII;
  63 }
  64
  65 void Iso2022Jp::Encoder::switch_mode(Mode m, string &buf)
  66 {
  67         mode = m;
  68         switch(mode)
  69         {
  70         case ASCII:    buf.append("\033(B", 3); break;
  71         case JISX0201: buf.append("\033(J", 3); break;
  72         case JISX0208: buf.append("\033$B", 3); break;
  73         default: throw CodecError("WTF?  Invalid mode in Iso2022Jp::Encoder::switch_mode");
  74         }
  75 }
  76
  77 void Iso2022Jp::Encoder::transliterate(UnicodeChar, string &buf)
  78 {
  79         if(mode!=ASCII)
  80                 switch_mode(ASCII, buf);
  81         buf += '?';
  82 }
  83
  84
  85 Iso2022Jp::Decoder::Decoder(ErrorMode em):
  86         Codec::Decoder(em),
  87         mode(ASCII),
  88         dec(new Ascii::Decoder)
  89 { }
  90
  91 UnicodeChar Iso2022Jp::Decoder::decode_char(const string &str, string::const_iterator &i)
  92 {
  93         if(i==str.end())
  94                 return error("No input");
  95
  96         while(i!=str.end())
  97         {
  98                 string::const_iterator j = i;
  99
 100                 UnicodeChar result = -1;
 101                 if(*j==033)
 102                 {
 103                         unsigned escape = 0;
 104                         for(++j; j!=str.end(); ++j)
 105                         {
 106                                 escape = escape<<8 | static_cast<unsigned char>(*j);
 107                                 if(*j>='@' && *j<='Z')
 108                                         break;
 109                         }
 110
 111                         bool ok = true;
 112                         switch(escape)
 113                         {
 114                         case 0x2842: switch_mode(ASCII); break;    // ESC ( B
 115                         case 0x284A: switch_mode(JISX0201); break; // ESC ( J
 116                         case 0x2440:                               // ESC $ @
 117                         case 0x2442: switch_mode(JISX0208); break; // ESC $ B
 118                         default: ok = false;
 119                         }
 120
 121                         if(ok)
 122                                 i = j;
 123                         else
 124                                 result = *i;
 125                         ++i;
 126                 }
 127                 else if(dec)
 128                         return dec->decode_char(str, i);
 129                 else
 130                         throw CodecError("WTF?  No sub-decoder for Iso2022Jp::Decoder");
 131
 132                 if(result>=0)
 133                         return result;
 134         }
 135
 136         return -1;
 137 }
 138
 139 void Iso2022Jp::Decoder::reset()
 140 {
 141         delete dec;
 142         mode = ASCII;
 143         dec = new Ascii::Decoder;
 144 }
 145
 146 void Iso2022Jp::Decoder::switch_mode(Mode m)
 147 {
 148         delete dec;
 149
 150         mode = m;
 151         switch(mode)
 152         {
 153         case ASCII: dec = new Ascii::Decoder; break;
 154         case JISX0201: dec = new JisX0201::Decoder; break;
 155         case JISX0208: dec = new JisX0208::Decoder; break;
 156         }
 157 }
 158
 159 } // namespace Codecs
 160 } // namespace Msp