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
  45
  46 void Codec::Decoder::decode(const string &str, ustring &buf)
  47 {
  48         for(string::const_iterator i=str.begin(); i!=str.end();)
  49         {
  50                 unichar c = decode_char(str, i);
  51                 if(c!=-1)
  52                         buf += c;
  53         }
  54 }
  55
  56 ustring Codec::Decoder::decode(const string &str)
  57 {
  58         ustring buf;
  59         decode(str, buf);
  60         return buf;
  61 }
  62
  63 Codec *create_codec(const string &n)
  64 {
  65         string name;
  66         for(string::const_iterator i=n.begin(); i!=n.end(); ++i)
  67         {
  68                 if(isupper(*i))
  69                         name += tolower(*i);
  70                 else if(islower(*i) || isdigit(*i))
  71                         name += *i;
  72         }
  73
  74         if(name=="ascii") return new Ascii;
  75         if(name=="iso2022jp") return new Iso2022Jp;
  76         if(name=="iso646fi") return new Iso646Fi;
  77         if(name=="iso88591" || name=="latin1") return new Iso88591;
  78         if(name=="iso885915" || name=="latin9") return new Iso885915;
  79         if(name=="jisx0201") return new JisX0201;
  80         if(name=="jisx0208") return new JisX0208;
  81         if(name=="utf8") return new Utf8;
  82         if(name=="windows1252" || name=="cp1252") return new Windows1252;
  83         throw invalid_argument("unknown string codec");
  84 }
  85
  86 Codec *detect_codec(const string &str)
  87 {
  88         bool is_utf8 = true;
  89         bool is_ascii = true;
  90         bool is_latin1 = true;
  91         unsigned utf8_mb = 0;
  92
  93         for(string::const_iterator i=str.begin(); i!=str.end(); ++i)
  94         {
  95                 unsigned char c = *i;
  96                 if(c&0x80)
  97                 {
  98                         is_ascii = false;
  99                         if((c&0xC0)==0x80)
 100                         {
 101                                 if((c&0xE0)==0x80)
 102                                         is_latin1 = false;
 103                                 if(utf8_mb)
 104                                         --utf8_mb;
 105                                 else
 106                                         is_utf8 = false;
 107                         }
 108                         else if((c&0xC0)==0xC0)
 109                         {
 110                                 if(utf8_mb)
 111                                 {
 112                                         is_utf8 = false;
 113                                         utf8_mb = 0;
 114                                 }
 115                                 else
 116                                 {
 117                                         for(utf8_mb=1; (c>>(6-utf8_mb))&1; ++utf8_mb) ;
 118                                 }
 119                         }
 120                 }
 121                 else if(utf8_mb)
 122                 {
 123                         is_utf8 = false;
 124                         utf8_mb = 0;
 125                 }
 126         }
 127
 128         if(is_ascii)
 129                 return new Ascii;
 130         else if(is_utf8)
 131                 return new Utf8;
 132         else if(is_latin1)
 133                 return new Iso88591;
 134         else
 135                 return new Windows1252;
 136 }
 137
 138 } // namespace StringCodec
 139 } // namespace Msp