msvc.txt

把html转成txt 把html转成txt

FYI: My String class is the STL wide string class.

String CPresentation::cleanRTF(String p_rtfString)
{
	String cleanString = NP_STRING;
	String convertedString = NP_STRING;
	String tempString = NP_STRING;

	IUnknown* pDoxCom;

    HRESULT hr = ::CoCreateInstance(CLSID_DoxLib,
                                    NULL,
                                    CLSCTX_INPROC_SERVER,
                                    IID_IUnknown,
                                    (LPVOID*)&pDoxCom);

	if (FAILED(hr))
	{
		cleanString = NP_STRING;
		m_errorMessage += L"RTF Conversion failed: Server DoxCom DLL not
available.";

		return cleanString;
	}

	IDoxLibPtr pDoxLib;

	hr = pDoxCom->QueryInterface(IID_IDoxLib,
		                         (VOID**)&pDoxLib);

	if (hr == S_OK)
	{
		// Convert our string to an LPSTR
		LPSTR narrowString = "";
	    char s1[MAX_STR + 1] = "";

        if (!WideCharToMultiByte(CP_ACP, 0, p_rtfString.c_str(), -1, s1,
MAX_STR, NULL, NULL))
		{
    		cleanString = NP_STRING;
			DWORD errorCode = GetLastError();

			if (errorCode == ERROR_INSUFFICIENT_BUFFER)
	    	    m_errorMessage += L"RTF Conversion failed: Source text exceeded
buffer size.\r\n";
			else
	    	    m_errorMessage += L"RTF Conversion failed: Unknown
WideCharToMultiByte Failure.\r\n";
		}
		else
		{
	    	pDoxLib->SetInString(s1);

			// if we're rendering WML, then we want to convert the RTF to
			// plain un-formatted text (otherwise set to HTML)
		    if ((m_renderingStyle == RNDR_WML) || (m_renderingStyle ==
RNDR_PLAIN_TEXT))
                pDoxLib->SetOutFormat("txt");
			else
				// we have to explicitly set to HTML even though its the default
				// because the DLL is state-ful - it will stay plain text once
				// it has been set that way.
                pDoxLib->SetOutFormat("html");

			BSTR newString = pDoxLib->StringConvert();

		    convertedString = newString;

			// Now remove all occurrences within the string of CRLF
			long index;
			index = convertedString.find(L"\r\n");
			while ( index != std::string::npos)
			{
				// Erase that character and the next one
				tempString = convertedString.erase(index, 2);

				convertedString = tempString;

				// Try and find another occurrence
    			index = convertedString.find(L"\r\n");
			}
			index = convertedString.find(L"\"");
			while ( index != std::string::npos)
			{
				// Now insert a reverse slash at the index
				tempString = convertedString.insert(index, L"\\");

				// Erase that character and the next one
				//tempString = convertedString.erase(index, 1);

				convertedString = tempString;

				// Try and find another occurrence
    			index = convertedString.find(L"\"", index+2);
			}

			// Now prepend all quotation marks with a forward slash

		    // Free the memory allocated by the COM object
		    ::SysFreeString(newString);

		    if ((m_renderingStyle == RNDR_WML) || (m_renderingStyle ==
RNDR_PLAIN_TEXT))
			{
		        // no additional formatting needed for plain text
				cleanString = convertedString;
			}
			else
			{
    		    // Now remove the html tags from the converted string
		        tempString = convertedString.erase(0,6);
			    long size = tempString.length();
			    cleanString = tempString.erase(size-7, 7);
			}
		}

		pDoxLib->Release();
	}

	return cleanString;
}