source/codec.cpp

   1 /* $Id$
   2
   3 This file is part of libmspstrings
   4 Copyright © 2006-2007 Mikko Rasa
   5 Distributed under the LGPL
   6 */
   7
   8 #include "ascii.h"
   9 #include "codec.h"
  10 #include "iso2022jp.h"
  11 #include "iso646fi.h"
  12 #include "iso88591.h"
  13 #include "iso885915.h"
  14 #include "jisx0201.h"
  15 #include "jisx0208.h"
  16 #include "utf8.h"
  17 #include "windows1252.h"
  18
  19 using namespace std;
  20
  21 namespace Msp {
  22 namespace Codecs {
  23
  24 bool Codec::detect(const string &str) const
  25 {
  26         Decoder *dec = create_decoder();
  27         bool result = true;
  28         try
  29         {
  30                 for(string::const_iterator i=str.begin(); i!=str.end(); )
  31                         dec->decode_char(str, i);
  32         }
  33         catch(const CodecError &)
  34         {
  35                 result = false;
  36         }
  37
  38         delete dec;
  39
  40         return result;
  41 }
  42
  43 void Codec::Encoder::encode(const ustring &str, string &buf)
  44 {
  45         for(ustring::const_iterator i=str.begin(); i!=str.end(); ++i)
  46                 encode_char(*i, buf);
  47 }
  48
  49 string Codec::Encoder::encode(const ustring &str)
  50 {
  51         string buf;
  52         encode(str, buf);
  53         sync(buf);
  54         return buf;
  55 }
  56
  57 void Codec::Encoder::error(UnicodeChar ch, string &buf, const string &msg)
  58 {
  59         switch(err_mode)
  60         {
  61         case TRANSLITERATE:
  62                 transliterate(ch, buf);
  63         case IGNORE_ERRORS:
  64                 break;
  65         default:
  66                 throw CodecError(msg);
  67         }
  68 }
  69
  70
  71 void Codec::Decoder::decode(const string &str, ustring &buf)
  72 {
  73         for(string::const_iterator i=str.begin(); i!=str.end();)
  74         {
  75                 UnicodeChar c = decode_char(str, i);
  76                 if(c!=-1)
  77                         buf += c;
  78         }
  79 }
  80
  81 ustring Codec::Decoder::decode(const string &str)
  82 {
  83         ustring buf;
  84         decode(str, buf);
  85         return buf;
  86 }
  87
  88 UnicodeChar Codec::Decoder::error(const string &msg)
  89 {
  90         switch(err_mode)
  91         {
  92         case TRANSLITERATE:
  93                 return 0xFFFE;
  94         case IGNORE_ERRORS:
  95                 return -1;
  96         default:
  97                 throw CodecError(msg);
  98         }
  99 }
 100
 101 Codec *create_codec(const string &n)
 102 {
 103         string name;
 104         for(string::const_iterator i=n.begin(); i!=n.end(); ++i)
 105         {
 106                 if(isupper(*i))
 107                         name += tolower(*i);
 108                 else if(islower(*i) || isdigit(*i))
 109                         name += *i;
 110         }
 111
 112         if(name=="ascii") return new Ascii;
 113         if(name=="iso2022jp") return new Iso2022Jp;
 114         if(name=="iso646fi") return new Iso646Fi;
 115         if(name=="iso88591" || name=="latin1") return new Iso88591;
 116         if(name=="iso885915" || name=="latin9") return new Iso885915;
 117         if(name=="jisx0201") return new JisX0201;
 118         if(name=="jisx0208") return new JisX0208;
 119         if(name=="utf8") return new Utf8;
 120         if(name=="windows1252" || name=="cp1252") return new Windows1252;
 121         throw InvalidParameterValue("Unknown string codec");
 122 }
 123
 124 Codec *detect_codec(const string &str)
 125 {
 126         bool is_utf8 = true;
 127         bool is_ascii = true;
 128         bool is_latin1 = true;
 129         unsigned utf8_mb = 0;
 130
 131         for(string::const_iterator i=str.begin(); i!=str.end(); ++i)
 132         {
 133                 unsigned char c = *i;
 134                 if(c&0x80)
 135                 {
 136                         is_ascii = false;
 137                         if((c&0xC0)==0x80)
 138                         {
 139                                 if((c&0xE0)==0x80)
 140                                         is_latin1 = false;
 141                                 if(utf8_mb)
 142                                         --utf8_mb;
 143                                 else
 144                                         is_utf8 = false;
 145                         }
 146                         else if((c&0xC0)==0xC0)
 147                         {
 148                                 if(utf8_mb)
 149                                 {
 150                                         is_utf8 = false;
 151                                         utf8_mb = 0;
 152                                 }
 153                                 else
 154                                 {
 155                                         for(utf8_mb=1; (c>>(6-utf8_mb))&1; ++utf8_mb) ;
 156                                 }
 157                         }
 158                 }
 159                 else if(utf8_mb)
 160                 {
 161                         is_utf8 = false;
 162                         utf8_mb = 0;
 163                 }
 164         }
 165
 166         if(is_ascii)
 167                 return new Ascii;
 168         else if(is_utf8)
 169                 return new Utf8;
 170         else if(is_latin1)
 171                 return new Iso88591;
 172         else
 173                 return new Windows1252;
 174 }
 175
 176 } // namespace Codecs
 177 } // namespace Msp