source/stringcodec/codec.cpp

   1 #include "ascii.h"
   2 #include "codec.h"
   3 #include "iso2022jp.h"
   4 #include "iso646fi.h"
   5 #include "iso88591.h"
   6 #include "iso885915.h"
   7 #include "jisx0201.h"
   8 #include "jisx0208.h"
   9 #include "utf8.h"
  10 #include "windows1252.h"
  11
  12 using namespace std;
  13
  14 namespace Msp {
  15 namespace StringCodec {
  16
  17 bool Codec::detect(const string &str) const
  18 {
  19         Decoder *dec = create_decoder(IGNORE_ERRORS);
  20
  21         bool result = true;
  22         for(string::const_iterator i=str.begin(); (result && i!=str.end()); )
  23                 result = (dec->decode_char(str, i)!=-1);
  24
  25         delete dec;
  26
  27         return result;
  28 }
  29
  30 void Codec::Encoder::encode(const ustring &str, string &buf)
  31 {
  32         for(ustring::const_iterator i=str.begin(); i!=str.end(); ++i)
  33                 encode_char(*i, buf);
  34 }
  35
  36 string Codec::Encoder::encode(const ustring &str)
  37 {
  38         string buf;
  39         encode(str, buf);
  40         sync(buf);
  41         return buf;
  42 }
  43
  44 void Codec::Encoder::error(unichar ch, string &buf, const string &msg)
  45 {
  46         switch(err_mode)
  47         {
  48         case TRANSLITERATE:
  49                 transliterate(ch, buf);
  50         case IGNORE_ERRORS:
  51                 break;
  52         default:
  53                 throw CodecError(msg);
  54         }
  55 }
  56
  57
  58 void Codec::Decoder::decode(const string &str, ustring &buf)
  59 {
  60         for(string::const_iterator i=str.begin(); i!=str.end();)
  61         {
  62                 unichar c = decode_char(str, i);
  63                 if(c!=-1)
  64                         buf += c;
  65         }
  66 }
  67
  68 ustring Codec::Decoder::decode(const string &str)
  69 {
  70         ustring buf;
  71         decode(str, buf);
  72         return buf;
  73 }
  74
  75 unichar Codec::Decoder::error(const string &msg)
  76 {
  77         switch(err_mode)
  78         {
  79         case TRANSLITERATE:
  80                 return 0xFFFD;
  81         case IGNORE_ERRORS:
  82                 return -1;
  83         default:
  84                 throw CodecError(msg);
  85         }
  86 }
  87
  88 Codec *create_codec(const string &n)
  89 {
  90         string name;
  91         for(string::const_iterator i=n.begin(); i!=n.end(); ++i)
  92         {
  93                 if(isupper(*i))
  94                         name += tolower(*i);
  95                 else if(islower(*i) || isdigit(*i))
  96                         name += *i;
  97         }
  98
  99         if(name=="ascii") return new Ascii;
 100         if(name=="iso2022jp") return new Iso2022Jp;
 101         if(name=="iso646fi") return new Iso646Fi;
 102         if(name=="iso88591" || name=="latin1") return new Iso88591;
 103         if(name=="iso885915" || name=="latin9") return new Iso885915;
 104         if(name=="jisx0201") return new JisX0201;
 105         if(name=="jisx0208") return new JisX0208;
 106         if(name=="utf8") return new Utf8;
 107         if(name=="windows1252" || name=="cp1252") return new Windows1252;
 108         throw InvalidParameterValue("Unknown string codec");
 109 }
 110
 111 Codec *detect_codec(const string &str)
 112 {
 113         bool is_utf8 = true;
 114         bool is_ascii = true;
 115         bool is_latin1 = true;
 116         unsigned utf8_mb = 0;
 117
 118         for(string::const_iterator i=str.begin(); i!=str.end(); ++i)
 119         {
 120                 unsigned char c = *i;
 121                 if(c&0x80)
 122                 {
 123                         is_ascii = false;
 124                         if((c&0xC0)==0x80)
 125                         {
 126                                 if((c&0xE0)==0x80)
 127                                         is_latin1 = false;
 128                                 if(utf8_mb)
 129                                         --utf8_mb;
 130                                 else
 131                                         is_utf8 = false;
 132                         }
 133                         else if((c&0xC0)==0xC0)
 134                         {
 135                                 if(utf8_mb)
 136                                 {
 137                                         is_utf8 = false;
 138                                         utf8_mb = 0;
 139                                 }
 140                                 else
 141                                 {
 142                                         for(utf8_mb=1; (c>>(6-utf8_mb))&1; ++utf8_mb) ;
 143                                 }
 144                         }
 145                 }
 146                 else if(utf8_mb)
 147                 {
 148                         is_utf8 = false;
 149                         utf8_mb = 0;
 150                 }
 151         }
 152
 153         if(is_ascii)
 154                 return new Ascii;
 155         else if(is_utf8)
 156                 return new Utf8;
 157         else if(is_latin1)
 158                 return new Iso88591;
 159         else
 160                 return new Windows1252;
 161 }
 162
 163 } // namespace StringCodec
 164 } // namespace Msp