Function CharToWideMap

Libraries/unrar/unicode.cpp:183–228

Convert and map inconvertible Unicode characters. We use it for extended ASCII names in Unix.

// Convert and map inconvertible Unicode characters.
// We use it for extended ASCII names in Unix.
void CharToWideMap(const char *Src,wchar *Dest,size_t DestSize,bool &Success)
{
  // Map inconvertible characters to private use Unicode area 0xE000.
  // Mark such string by placing special non-character code before
  // first inconvertible character.
  Success=false;
  bool MarkAdded=false;
  uint SrcPos=0,DestPos=0;
  while (DestPos<DestSize)
  {
    if (Src[SrcPos]==0)
    {
      Success=true;
      break;
    }
    mbstate_t ps;
    memset(&ps,0,sizeof(ps));
    size_t res=mbrtowc(Dest+DestPos,Src+SrcPos,MB_CUR_MAX,&ps);
    if (res==(size_t)-1 || res==(size_t)-2)
    {
      // For security reasons we do not want to map low ASCII characters,
      // so we do not have additional .. and path separator codes.
      if (byte(Src[SrcPos])>=0x80)
      {
        if (!MarkAdded)
        {
          Dest[DestPos++]=MappedStringMark;
          MarkAdded=true;
          if (DestPos>=DestSize)
            break;
        }
        Dest[DestPos++]=byte(Src[SrcPos++])+MapAreaStart;
      }
      else
        break;
    }
    else
    {
      memset(&ps,0,sizeof(ps));
      int Length=mbrlen(Src+SrcPos,MB_CUR_MAX,&ps);
      SrcPos+=Max(Length,1);
      DestPos++;
    }
  }
  Dest[Min(DestPos,DestSize-1)]=0;
}
#endif