Function WideToCharMap

Libraries/unrar/unicode.cpp:134–176

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

// Convert and restore mapped inconvertible Unicode characters. 
// We use it for extended ASCII names in Unix.
bool WideToCharMap(const wchar *Src,char *Dest,size_t DestSize,bool &Success)
{
  // String with inconvertible characters mapped to private use Unicode area
  // must have the mark code somewhere.
  if (wcschr(Src,(wchar)MappedStringMark)==NULL)
    return false;

  // Seems to be that wcrtomb in some memory analyzing libraries
  // can produce uninitilized output while reporting success on garbage input.
  // So we clean the destination to calm analyzers.
  memset(Dest,0,DestSize);
  
  Success=true;
  uint SrcPos=0,DestPos=0;
  while (Src[SrcPos]!=0 && DestPos<DestSize-MB_CUR_MAX)
  {
    if (uint(Src[SrcPos])==MappedStringMark)
    {
      SrcPos++;
      continue;
    }
    // For security reasons do not restore low ASCII codes, so mapping cannot
    // be used to hide control codes like path separators.
    if (uint(Src[SrcPos])>=MapAreaStart+0x80 && uint(Src[SrcPos])<MapAreaStart+0x100)
      Dest[DestPos++]=char(uint(Src[SrcPos++])-MapAreaStart);
    else
    {
      mbstate_t ps;
      memset(&ps,0,sizeof(ps));
      if (wcrtomb(Dest+DestPos,Src[SrcPos],&ps)==(size_t)-1)
      {
        Dest[DestPos]='_';
        Success=false;
      }
      SrcPos++;
      memset(&ps,0,sizeof(ps));
      int Length=mbrlen(Dest+DestPos,MB_CUR_MAX,&ps);
      DestPos+=Max(Length,1);
    }
  }
  Dest[Min(DestPos,DestSize-1)]=0;
  return true;
}
#endif