source/utf8.cpp

   1 #include "utf8.h"
   2
   3 using namespace std;
   4
   5 namespace Msp {
   6
   7 void Utf8::Encoder::encode_char(wchar_t c)
   8 {
   9         unsigned code=c;
  10         if(code>0x10FFFF)
  11         {
  12                 error("Can't express character in UTF-8");
  13                 return;
  14         }
  15
  16         unsigned bytes=1;
  17         if(code>0xFFFF)
  18                 bytes=4;
  19         else if(code>0x7FF)
  20                 bytes=3;
  21         else if(code>0x7F)
  22                 bytes=2;
  23
  24         if(bytes==1)
  25                 append(code);
  26         else
  27         {
  28                 char buf[4];
  29
  30                 buf[0]=0xFF<<(8-bytes) | code>>(bytes*6-6);
  31                 for(unsigned j=bytes-1; j>0; --j)
  32                 {
  33                         buf[j]=0x80 | code&0x3F;
  34                         code>>=6;
  35                 }
  36
  37                 append(buf, bytes);
  38         }
  39 }
  40
  41
  42 void Utf8::Decoder::decode_char(const string &str, string::const_iterator &i)
  43 {
  44         while(i!=str.end())
  45         {
  46                 if(bytes==0)
  47                 {
  48                         if((*i&0xC0)==0x80)
  49                         {
  50                                 error("Invalid UTF-8 string (tail byte when expecting head)");
  51                                 ++i;
  52                                 break;
  53                         }
  54                         else if(*i&0x80)
  55                         {
  56                                 unsigned mask=0x40;
  57                                 for(; *i&mask; mask>>=1)
  58                                         ++bytes;
  59
  60                                 if(bytes>3)
  61                                 {
  62                                         error("Invalid UTF-8 string (overlong multibyte sequence)");
  63                                         ++i;
  64                                         break;
  65                                 }
  66                                 else
  67                                 {
  68                                         code=(*i++)&(mask-1);
  69                                         if(!code)
  70                                         {
  71                                                 error("Invalid UTF-8 string (denormalized multibyte sequence)");
  72                                                 break;
  73                                         }
  74                                 }
  75                         }
  76                         else
  77                         {
  78                                 append(*i++);
  79                                 break;
  80                         }
  81                 }
  82                 else
  83                 {
  84                         if((*i&0xC0)!=0x80)
  85                         {
  86                                 error("Invalid UTF-8 string (head byte when expecting tail)");
  87                                 ++i;
  88                                 break;
  89                         }
  90
  91                         code=code<<6 | (*i++)&0x3F;
  92                         --bytes;
  93
  94                         if(!bytes)
  95                         {
  96                                 if(code>0x10FFFF)
  97                                         error("Invalid UTF-8 string (character out of range)");
  98                                 else
  99                                         append(code);
 100                                 break;
 101                         }
 102                 }
 103         }
 104 }
 105
 106 void Utf8::Decoder::sync()
 107 {
 108         if(bytes)
 109         {
 110                 error("Sync in the middle of multibyte UTF-8 sequence");
 111                 bytes=0;
 112         }
 113 }
 114
 115 void Utf8::Decoder::reset()
 116 {
 117         bytes=0;
 118 }
 119
 120 } // namespace Msp