source/stringcodec/iso2022jp.cpp

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