Use regular expression in your C++ program

Introduction

Regular expressions provide a convenient way to specify complicated string pattern for search, replace or validate the text input. Since it is very useful, many people wrote their own library. Many libraries I found are buggy and it takes a lot of time to debug source code. However, actually you do not need to search any more, since there is already one built in your computer for free � the regular expression parser written by Microsoft.

As many other utilities, Microsoft provides this functionality using COM interface. It is easy to find this COM server object, named Microsoft VBScript Regular Expressions 5.5, with the tool oleview:

One problem is that there is no type library associated with this DLL. Fortunately it is not a big deal since we have the IDL definition. From OleView, save the IDL definition into a file and use MIDL to compile it, you will get type library. After that, we can use this type library within our C++ program. Suppose we have this file named RegExp.tlb.

Using regular expression

You can find all the documentation in either MSDN or the URL http://msdn.microsoft.com/scripting/default.htm?/scripting/vbscript/doc/vsobjregexp.htm. Though it is for scripting, you can still use them directly with the help of the newly-added keyword #import since Visual C++ 6. Generally you define a pattern, then you can test this pattern against the input string. or exacute to see whether they are any matches.

To demonstrate its usage, we wrote a custome DDX routine to verify the input of a control in a dialog. The function prototype is listed as following:

void WINAPI DDX_RegExp(CDataExchange* pDX, int nIDC, LPCTSTR lpszPattern, CString& value);

If the control input exactly matches the specified pattern(lpszPattern), the validation is passed otherwise an message box will pop up.

#import "RegExp.tlb" no_namespace
...
void AFXAPI DDX_RegExp(CDataExchange* pDX, int nIDC, LPCTSTR lpszPattern, CString& value)
{
   try {
      static IRegExpPtr regExp( __uuidof(RegExp) );
      regExp->Pattern = _bstr_t(lpszPattern);
    
      HWND hWndCtrl = pDX->PrepareEditCtrl(nIDC);
      if (pDX->m_bSaveAndValidate)
      {
          int nLen = ::GetWindowTextLength(hWndCtrl);
          ::GetWindowText(hWndCtrl, value.GetBufferSetLength(nLen),nLen+1);
          value.ReleaseBuffer();

          //now we verify it

          if ( regExp->Test( (LPCTSTR)value) )
          {
              IMatchCollectionPtr matches=regExp->Execute((LPCTSTR)value);
              if ( matches->Count== 1)
              {
                  IMatchPtr match = matches->Item[0];
                  if ( match->FirstIndex==0 && match->Length == value.GetLength() )
                  {
                    return;
                  }
              }
          }
          CString strMsg = CString("The input does not exactly have the pattern ") + lpszPattern;
          pDX->m_pDlgWnd->MessageBox(strMsg);
          pDX->PrepareEditCtrl(nIDC);
          pDX->Fail();
      }
      else
      {
      }
    }
    catch (_com_error& e)
    {
        AfxMessageBox( e.ErrorMessage() );
    }
}

In the code above, we first use Test method to see whether there is a match. If there is one, we use Execute method to retrieve all the matches. There should be only one match.

After we define that, we can use this function in our MFC application. Note that you must initialize the COM library first in your application. The following will validate an input box to see if it matches the phone number format:

DDX_RegExp(pDX, IDC_INPUT, _T("\\d{3}-\\d{3}-\\d{4}"), m_strInput);

In this way, you can write validation code for more complicated pattern. The COM makes things a lot easier.