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 Codecs {
  16
  17 bool Codec::detect(const string &str) const
  18 {
  19         Decoder *dec = create_decoder();
  20         bool result = true;
  21         try
  22         {
  23                 for(string::const_iterator i=str.begin(); i!=str.end(); )
  24                         dec->decode_char(str, i);
  25         }
  26         catch(const CodecError &)
  27         {
  28                 result = false;
  29         }
  30
  31         delete dec;
  32
  33         return result;
  34 }
  35
  36 void Codec::Encoder::encode(const ustring &str, string &buf)
  37 {
  38         for(ustring::const_iterator i=str.begin(); i!=str.end(); ++i)
  39                 encode_char(*i, buf);
  40 }
  41
  42 string Codec::Encoder::encode(const ustring &str)
  43 {
  44         string buf;
  45         encode(str, buf);
  46         sync(buf);
  47         return buf;
  48 }
  49
  50 void Codec::Encoder::error(UnicodeChar ch, string &buf, const string &msg)
  51 {
  52         switch(err_mode)
  53         {
  54         case TRANSLITERATE:
  55                 transliterate(ch, buf);
  56         case IGNORE_ERRORS:
  57                 break;
  58         default:
  59                 throw CodecError(msg);
  60         }
  61 }
  62
  63
  64 void Codec::Decoder::decode(const string &str, ustring &buf)
  65 {
  66         for(string::const_iterator i=str.begin(); i!=str.end();)
  67         {
  68                 UnicodeChar c = decode_char(str, i);
  69                 if(c!=-1)
  70                         buf += c;
  71         }
  72 }
  73
  74 ustring Codec::Decoder::decode(const string &str)
  75 {
  76         ustring buf;
  77         decode(str, buf);
  78         return buf;
  79 }
  80
  81 UnicodeChar Codec::Decoder::error(const string &msg)
  82 {
  83         switch(err_mode)
  84         {
  85         case TRANSLITERATE:
  86                 return 0xFFFE;
  87         case IGNORE_ERRORS:
  88                 return -1;
  89         default:
  90                 throw CodecError(msg);
  91         }
  92 }
  93
  94 Codec *create_codec(const string &n)
  95 {
  96         string name;
  97         for(string::const_iterator i=n.begin(); i!=n.end(); ++i)
  98         {
  99                 if(isupper(*i))
 100                         name += tolower(*i);
 101                 else if(islower(*i) || isdigit(*i))
 102                         name += *i;
 103         }
 104
 105         if(name=="ascii") return new Ascii;
 106         if(name=="iso2022jp") return new Iso2022Jp;
 107         if(name=="iso646fi") return new Iso646Fi;
 108         if(name=="iso88591" || name=="latin1") return new Iso88591;
 109         if(name=="iso885915" || name=="latin9") return new Iso885915;
 110         if(name=="jisx0201") return new JisX0201;
 111         if(name=="jisx0208") return new JisX0208;
 112         if(name=="utf8") return new Utf8;
 113         if(name=="windows1252" || name=="cp1252") return new Windows1252;
 114         throw InvalidParameterValue("Unknown string codec");
 115 }
 116
 117 Codec *detect_codec(const string &str)
 118 {
 119         bool is_utf8 = true;
 120         bool is_ascii = true;
 121         bool is_latin1 = true;
 122         unsigned utf8_mb = 0;
 123
 124         for(string::const_iterator i=str.begin(); i!=str.end(); ++i)
 125         {
 126                 unsigned char c = *i;
 127                 if(c&0x80)
 128                 {
 129                         is_ascii = false;
 130                         if((c&0xC0)==0x80)
 131                         {
 132                                 if((c&0xE0)==0x80)
 133                                         is_latin1 = false;
 134                                 if(utf8_mb)
 135                                         --utf8_mb;
 136                                 else
 137                                         is_utf8 = false;
 138                         }
 139                         else if((c&0xC0)==0xC0)
 140                         {
 141                                 if(utf8_mb)
 142                                 {
 143                                         is_utf8 = false;
 144                                         utf8_mb = 0;
 145                                 }
 146                                 else
 147                                 {
 148                                         for(utf8_mb=1; (c>>(6-utf8_mb))&1; ++utf8_mb) ;
 149                                 }
 150                         }
 151                 }
 152                 else if(utf8_mb)
 153                 {
 154                         is_utf8 = false;
 155                         utf8_mb = 0;
 156                 }
 157         }
 158
 159         if(is_ascii)
 160                 return new Ascii;
 161         else if(is_utf8)
 162                 return new Utf8;
 163         else if(is_latin1)
 164                 return new Iso88591;
 165         else
 166                 return new Windows1252;
 167 }
 168
 169 } // namespace Codecs
 170 } // namespace Msp