Click here to Skip to main content
65,938 articles
CodeProject is changing. Read more.
Articles
(untagged)

Differences in the behaviour of CString >> and << using MFC 7 and double '0' ended strings (like REG_MULTI_SZ)

0.00/5 (No votes)
3 Jun 2003 1  
Differences in the behaviour of CString >> and << using MFC 7 and double '0' ended strings (like REG_MULTI_SZ) and a workaround

Introduction

This article explain the behaviour diference using CString with binary buffers (GetBuffer or GetBufferSetLength using REG_MULTI_SZ) and the serialization (more precisally the read operation, the write works fine). The problem occurs when your buffer contains '\0' at the middle of the buffer. The CArchive << writes all buffer to the file but the CArchive >> reads all data from the file as expected in a temporary buffer and performs an simple assignment to the CString using the CString operator=. This is the problem, the CString operator = stops in a '\0'. Below is the MFC code with the bug and a workaround to try solve the problem.

Peoples can say, why this ? Use an array. Yes I agree with all of then, but, I my case I can't use an array. This is the first time in 5 years of MFC development that I must made this "heresy" like some people consider it.

I have changed the title of the article to behaviour diference because the opinion of the people.

Using the code

This is the MFC code:

template< typename BaseType, class StringTraits >
    CArchive& AFXAPI operator>>(CArchive& ar, ATL::CStringT& str)
{
    int nCharSize;  // 1 = char, 2 = wchar_t

    UINT nLength = UINT( AfxReadStringLength(ar, nCharSize) );
    if (nCharSize == sizeof(char))
    {
        ATL::CTempBuffer< char > pszBufferA(nLength+1);

        pszBufferA[nLength] = '\0';
        UINT nBytesRead = ar.Read(pszBufferA, nLength*sizeof(char));
        if (nBytesRead != (nLength*sizeof(char)))
            AfxThrowArchiveException(CArchiveException::endOfFile);
        str = pszBufferA;
    }
    else
    {
        ASSERT(nCharSize == sizeof(wchar_t));

        ATL::CTempBuffer< wchar_t > pszBufferW( nLength+1 );

        pszBufferW[nLength] = L'\0';
        UINT nBytesRead = ar.Read(pszBufferW, nLength*sizeof(wchar_t));
        if (nBytesRead != (nLength*sizeof(wchar_t)))
            AfxThrowArchiveException(CArchiveException::endOfFile);
        str = pszBufferW;
    }

    return ar;
}

This is a workaround:

EXPORT_LCC UINT LCC_ReadBinaryCStringFromArchive(CArchive& ar, CString& str)
{
    ASSERT(ar.IsLoading());

    UINT nBytesRead;
    int nCharSize;  // 1 = char, 2 = wchar_t

    UINT nLength = UINT( AfxReadStringLength(ar, nCharSize) );
    if (nCharSize == sizeof(char))
    {
        // is UNICODE ? if nLength is 10 chars we must 5 wchar_t

        // to hold 10 chars

        if (nCharSize != sizeof(TCHAR))
            nBytesRead = ar.Read(str.GetBufferSetLength(nLength / sizeof 
                                (TCHAR) + (nLength % sizeof(TCHAR))),
                                nLength * sizeof(char));
        else
            nBytesRead = ar.Read(str.GetBufferSetLength(nLength),
                                 nLength*sizeof(char));

        if (nBytesRead != (nLength*sizeof(char)))
            AfxThrowArchiveException(CArchiveException::endOfFile);
    }
    else
    {
        ASSERT(nCharSize == sizeof(wchar_t));

        // not is UNICODE ? if nLength is 10 w_chars we must

        // 20 chars to hold 10 w_chars

        if (nCharSize != sizeof(TCHAR)) 
            nBytesRead = ar.Read(str.GetBufferSetLength(nLength * 
                         sizeof(TCHAR)), nLength * sizeof(wchar_t));
        else
            nBytesRead = ar.Read(str.GetBufferSetLength(nLength),
                                 nLength*sizeof(wchar_t));

        if (nBytesRead != (nLength*sizeof(wchar_t)))
            AfxThrowArchiveException(CArchiveException::endOfFile);
    }
    return nBytesRead;
}

I know that not is the same facility of CArchive overloaded operator >> but works for me. Use CString only if you need. If not, use an array.

History

License

This article has no explicit license attached to it but may contain usage terms in the article text or the download files themselves. If in doubt please contact the author via the discussion board below.

A list of licenses authors might use can be found here