How to Decode HTML String ?

void HTMLDecode(wxString& s)
    {
      wxString sRet;

      for(size_t i = 0; i < s.length(); ++i)
      {
      if(s.c_str()[i] == wxT('&'))
      {
      if(s.length() - i >= 7)
      {
      if(s.c_str()[i] == wxT('&') &&
      s.c_str()[i+1] == wxT('#') &&
      wxIsdigit(s.c_str()[i+2]) &&
      wxIsdigit(s.c_str()[i+3]) &&
      wxIsdigit(s.c_str()[i+4]) &&
      s.c_str()[i+5] == wxT(';'))
      {
      sRet += (wxChar) ((s.c_str()[i+2] - wxT('0')) * 100) +
      ((s.c_str()[i+3] - wxT('0')) * 10) +
      ((s.c_str()[i+4] - wxT('0')) );
      i += 6;
      continue;
      }
      }
      if(s.length() - i >= 6)
      {
      if(wxTmemcmp(&s.c_str()[i], wxT("'"), 6) == 0)
      {
      sRet += wxT(''');
      i += 5;
      continue;
      }
      if(wxTmemcmp(&s.c_str()[i], wxT(" "), 6) == 0)
      {
      sRet += wxT(' ');
      i += 5;
      continue;
      }
      if(wxTmemcmp(&s.c_str()[i], wxT("""), 6) == 0)
      {
      sRet += wxT('"');
      i += 5;
      continue;
      }
      }
      if(s.length() - i >= 5 && wxTmemcmp(&s.c_str()[i], wxT("&"), 5) == 0)
      {
      sRet += wxT('&');
      i += 4; //todo: We should re-scan like this:
// wxTmemmove((wxChar*)&s.c_str()[i+1],
// (wxChar*)&s.c_str()[i+5],
// s.length() - (i+5) );
// s.c_str()[i+5]
// i -= 1;
continue;
      }
      if(s.length() - i >= 4)
      {
      if(wxTmemcmp(&s.c_str()[i], wxT("<"), 4) == 0)
      {
      sRet += wxT('<');
      i += 3;
      continue;
      }
      if(wxTmemcmp(&s.c_str()[i], wxT(">"), 4) == 0)
      {
      sRet += wxT('>');
      i += 3;
      continue;
      }
      }
      }

      sRet += s[i];
      }

      s = sRet;
}

No comments are included for this article. Submit a comment with your feedback/suggestions.

Click to View Printer Friendly Version