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 void Codec::Encoder::error(UnicodeChar ch, string &buf, const string &msg)
  50 {
  51         switch(err_mode)
  52         {
  53         case TRANSLITERATE:
  54                 transliterate(ch, buf);
  55         case IGNORE_ERRORS:
  56                 break;
  57         default:
  58                 throw CodecError(msg);
  59         }
  60 }
  61
  62
  63 void Codec::Decoder::decode(const string &str, ustring &buf)
  64 {
  65         for(string::const_iterator i=str.begin(); i!=str.end();)
  66         {
  67                 UnicodeChar c=decode_char(str, i);
  68                 if(c!=-1)
  69                         buf+=c;
  70         }
  71 }
  72
  73 UnicodeChar Codec::Decoder::error(const string &msg)
  74 {
  75         switch(err_mode)
  76         {
  77         case TRANSLITERATE:
  78                 return 0xFFFE;
  79         case IGNORE_ERRORS:
  80                 return -1;
  81         default:
  82                 throw CodecError(msg);
  83         }
  84 }
  85
  86 Codec *create_codec(const string &n)
  87 {
  88         string name;
  89         for(string::const_iterator i=n.begin(); i!=n.end(); ++i)
  90         {
  91                 if(isupper(*i))
  92                         name+=tolower(*i);
  93                 else if(islower(*i) || isdigit(*i))
  94                         name+=*i;
  95         }
  96
  97         if(name=="ascii") return new Ascii;
  98         if(name=="iso2022jp") return new Iso2022Jp;
  99         if(name=="iso646fi") return new Iso646Fi;
 100         if(name=="iso88591" || name=="latin1") return new Iso88591;
 101         if(name=="iso885915" || name=="latin9") return new Iso885915;
 102         if(name=="jisx0201") return new JisX0201;
 103         if(name=="jisx0208") return new JisX0208;
 104         if(name=="utf8") return new Utf8;
 105         if(name=="windows1252" || name=="cp1252") return new Windows1252;
 106         throw InvalidParameterValue("Unknown string codec");
 107 }
 108
 109 Codec *detect_codec(const string &str)
 110 {
 111         bool is_utf8=true;
 112         bool is_ascii=true;
 113         bool is_latin1=true;
 114         unsigned utf8_mb=0;
 115
 116         for(string::const_iterator i=str.begin(); i!=str.end(); ++i)
 117         {
 118                 unsigned char c=*i;
 119                 if(c&0x80)
 120                 {
 121                         is_ascii=false;
 122                         if((c&0xC0)==0x80)
 123                         {
 124                                 if((c&0xE0)==0x80)
 125                                         is_latin1=false;
 126                                 if(utf8_mb)
 127                                         --utf8_mb;
 128                                 else
 129                                         is_utf8=false;
 130                         }
 131                         else if((c&0xC0)==0xC0)
 132                         {
 133                                 if(utf8_mb)
 134                                         is_utf8=false;
 135                                 else
 136                                 {
 137                                         for(utf8_mb=1; (c>>(6-utf8_mb))&1; ++utf8_mb) ;
 138                                 }
 139                         }
 140                 }
 141         }
 142
 143         if(is_ascii)
 144                 return new Ascii;
 145         else if(is_utf8)
 146                 return new Utf8;
 147         else if(is_latin1)
 148                 return new Iso88591;
 149         else
 150                 return new Windows1252;
 151 }
 152
 153 } // namespace Codecs
 154 } // namespace Msp