source/utf8.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 "utf8.h"
   9
  10 using namespace std;
  11
  12 namespace Msp {
  13
  14 void Utf8::Encoder::encode_char(wchar_t c)
  15 {
  16         unsigned code=c;
  17         if(code>0x10FFFF)
  18         {
  19                 error("Can't express character in UTF-8");
  20                 return;
  21         }
  22
  23         unsigned bytes=1;
  24         if(code>0xFFFF)
  25                 bytes=4;
  26         else if(code>0x7FF)
  27                 bytes=3;
  28         else if(code>0x7F)
  29                 bytes=2;
  30
  31         if(bytes==1)
  32                 append(code);
  33         else
  34         {
  35                 char buf[4];
  36
  37                 buf[0]=0xFF<<(8-bytes) | code>>(bytes*6-6);
  38                 for(unsigned j=bytes-1; j>0; --j)
  39                 {
  40                         buf[j]=0x80 | code&0x3F;
  41                         code>>=6;
  42                 }
  43
  44                 append(buf, bytes);
  45         }
  46 }
  47
  48
  49 void Utf8::Decoder::decode_char(const string &str, string::const_iterator &i)
  50 {
  51         while(i!=str.end())
  52         {
  53                 if(bytes==0)
  54                 {
  55                         if((*i&0xC0)==0x80)
  56                         {
  57                                 error("Invalid UTF-8 string (tail byte when expecting head)");
  58                                 ++i;
  59                                 break;
  60                         }
  61                         else if(*i&0x80)
  62                         {
  63                                 unsigned mask=0x40;
  64                                 for(; *i&mask; mask>>=1)
  65                                         ++bytes;
  66
  67                                 if(bytes>3)
  68                                 {
  69                                         error("Invalid UTF-8 string (overlong multibyte sequence)");
  70                                         ++i;
  71                                         break;
  72                                 }
  73                                 else
  74                                 {
  75                                         code=(*i++)&(mask-1);
  76                                         if(!code)
  77                                         {
  78                                                 error("Invalid UTF-8 string (denormalized multibyte sequence)");
  79                                                 break;
  80                                         }
  81                                 }
  82                         }
  83                         else
  84                         {
  85                                 append(*i++);
  86                                 break;
  87                         }
  88                 }
  89                 else
  90                 {
  91                         if((*i&0xC0)!=0x80)
  92                         {
  93                                 error("Invalid UTF-8 string (head byte when expecting tail)");
  94                                 ++i;
  95                                 break;
  96                         }
  97
  98                         code=code<<6 | (*i++)&0x3F;
  99                         --bytes;
 100
 101                         if(!bytes)
 102                         {
 103                                 if(code>0x10FFFF)
 104                                         error("Invalid UTF-8 string (character out of range)");
 105                                 else
 106                                         append(code);
 107                                 break;
 108                         }
 109                 }
 110         }
 111 }
 112
 113 void Utf8::Decoder::sync()
 114 {
 115         if(bytes)
 116         {
 117                 error("Sync in the middle of multibyte UTF-8 sequence");
 118                 bytes=0;
 119         }
 120 }
 121
 122 void Utf8::Decoder::reset()
 123 {
 124         bytes=0;
 125 }
 126
 127 } // namespace Msp