source/stringcodec/iso2022jp.cpp

   1 #include <msp/core/except.h>
   2 #include "ascii.h"
   3 #include "iso2022jp.h"
   4 #include "jisx0201.h"
   5 #include "jisx0208.h"
   6
   7 using namespace std;
   8
   9 namespace Msp {
  10 namespace StringCodec {
  11
  12 void Iso2022Jp::Encoder::encode_char(unichar ch, string &buf)
  13 {
  14         if(ch>=0 && ch<=0x7F && ch!=0x5C && ch!=0x7E)
  15         {
  16                 if(mode!=ASCII && mode!=JISX0201)
  17                         switch_mode(ASCII, buf);
  18                 buf += ch;
  19         }
  20         else if(ch==0x5C || ch==0x7E)
  21         {
  22                 if(mode!=ASCII)
  23                         switch_mode(ASCII, buf);
  24                 buf += ch;
  25         }
  26         else if(ch==0xA5 || ch==0x203E)
  27         {
  28                 if(mode!=JISX0201)
  29                         switch_mode(JISX0201, buf);
  30                 if(ch==0xA5)
  31                         buf += 0x5C;
  32                 else if(ch==0x203E)
  33                         buf += 0x7E;
  34         }
  35         else
  36         {
  37                 Kuten jis = ucs_to_jisx0208(ch);
  38                 if(!jis)
  39                         return error(ch, buf, invalid_character(ch, "ISO-2022-JP"));
  40
  41                 if(mode!=JISX0208)
  42                         switch_mode(JISX0208, buf);
  43
  44                 char jbuf[2];
  45                 jbuf[0] = jis.ku+0x20;
  46                 jbuf[1] = jis.ten+0x20;
  47                 buf.append(jbuf, 2);
  48         }
  49 }
  50
  51 void Iso2022Jp::Encoder::sync(string &buf)
  52 {
  53         if(mode!=ASCII)
  54                 switch_mode(ASCII, buf);
  55 }
  56
  57 void Iso2022Jp::Encoder::reset()
  58 {
  59         mode = ASCII;
  60 }
  61
  62 void Iso2022Jp::Encoder::switch_mode(Mode m, string &buf)
  63 {
  64         mode = m;
  65         switch(mode)
  66         {
  67         case ASCII:    buf.append("\033(B", 3); break;
  68         case JISX0201: buf.append("\033(J", 3); break;
  69         case JISX0208: buf.append("\033$B", 3); break;
  70         default: throw invalid_argument("Iso2022Jp::Encoder::switch_mode");
  71         }
  72 }
  73
  74 void Iso2022Jp::Encoder::transliterate(unichar, string &buf)
  75 {
  76         if(mode!=ASCII)
  77                 switch_mode(ASCII, buf);
  78         buf += '?';
  79 }
  80
  81
  82 Iso2022Jp::Decoder::Decoder(ErrorMode em):
  83         Codec::Decoder(em),
  84         dec(new Ascii::Decoder)
  85 { }
  86
  87 unichar Iso2022Jp::Decoder::decode_char(const string &str, string::const_iterator &i)
  88 {
  89         if(i==str.end())
  90                 return -1;
  91
  92         while(i!=str.end())
  93         {
  94                 auto j = i;
  95
  96                 unichar result = -1;
  97                 if(*j==033)
  98                 {
  99                         unsigned escape = 0;
 100                         for(++j; j!=str.end(); ++j)
 101                         {
 102                                 escape = escape<<8 | static_cast<unsigned char>(*j);
 103                                 if(*j>='@' && *j<='Z')
 104                                         break;
 105                         }
 106
 107                         bool ok = true;
 108                         switch(escape)
 109                         {
 110                         case 0x2842: switch_mode(ASCII); break;    // ESC ( B
 111                         case 0x284A: switch_mode(JISX0201); break; // ESC ( J
 112                         case 0x2440:                               // ESC $ @
 113                         case 0x2442: switch_mode(JISX0208); break; // ESC $ B
 114                         default: ok = false;
 115                         }
 116
 117                         if(ok)
 118                                 i = j;
 119                         else
 120                                 result = *i;
 121                         ++i;
 122                 }
 123                 else if(dec)
 124                         return dec->decode_char(str, i);
 125                 else
 126                         throw internal_error("no sub-decoder");
 127
 128                 if(result>=0)
 129                         return result;
 130         }
 131
 132         return -1;
 133 }
 134
 135 void Iso2022Jp::Decoder::reset()
 136 {
 137         delete dec;
 138         mode = ASCII;
 139         dec = new Ascii::Decoder;
 140 }
 141
 142 void Iso2022Jp::Decoder::switch_mode(Mode m)
 143 {
 144         delete dec;
 145
 146         mode = m;
 147         switch(mode)
 148         {
 149         case ASCII: dec = new Ascii::Decoder; break;
 150         case JISX0201: dec = new JisX0201::Decoder; break;
 151         case JISX0208: dec = new JisX0208::Decoder; break;
 152         }
 153 }
 154
 155 } // namespace StringCodec
 156 } // namespace Msp