stdiofileex.cpp

Encode/Decode UTF-8 format from/to text format

			{
				int	nChars = lstrlen(lpsz) + 1;	 // Why plus 1? Because yes

				pszUnicodeString	= new wchar_t[nChars]; 

				// Initialise to something safe
				memset(pszUnicodeString, 0, sizeof(wchar_t) * nChars);
				
				// Convert to Unicode using the locale code page (the code page we are using in memory)
				nChars = GetUnicodeStringFromMultiByteString((LPCSTR)(const char*)lpsz, pszUnicodeString, nChars, nLocaleCodePage);

				// Convert back to multibyte using the file code page
				// (Note that you can't reliably read a non-Unicode file written in code page A on a system using a code page B,
				// modify the file and write it back using code page A, unless you disable all this double-conversion code.
				// In effect, you have to choose between a mangled character display and mangled file writing).
				if (nChars > 0)
				{
					// Calculate how much we need for the MB buffer (it might be larger)
					nChars = GetRequiredMultiByteLengthForUnicodeString(pszUnicodeString, m_nFileCodePage);

					pszMultiByteString= new char[nChars];  
					memset(pszMultiByteString, 0, sizeof(char) * nChars);

					nChars = GetMultiByteStringFromUnicodeString(pszUnicodeString, pszMultiByteString, nChars, m_nFileCodePage);

					// Do byte-mode write. This avoids annoying "interpretation" of \n's as \r\n
					CFile::Write((const void*)pszMultiByteString, nChars * sizeof(char));
				}
			}
			else
			{
				// Do byte-mode write. This avoids annoying "interpretation" of \n's as \r\n
				CFile::Write((const void*)lpsz, lstrlen(lpsz)*sizeof(char));
			}
		}

#endif
	}
	// Ensure we always clean up
	catch(...)
	{
		if (pszUnicodeString)	delete [] pszUnicodeString;
		if (pszMultiByteString)	delete [] pszMultiByteString;
		throw;
	}

	if (pszUnicodeString)	delete [] pszUnicodeString;
	if (pszMultiByteString)	delete [] pszMultiByteString;
}

UINT CStdioFileEx::ProcessFlags(const CString& sFilePath, UINT& nOpenFlags)
{
	m_bIsUnicodeText = false;

	// If we have writeUnicode we must have write or writeRead as well
#ifdef _DEBUG
	if (nOpenFlags & CStdioFileEx::modeWriteUnicode)
	{
		ASSERT(nOpenFlags & CFile::modeWrite || nOpenFlags & CFile::modeReadWrite);
	}
#endif

	// If reading in text mode and not creating... ; fixed by Dennis Jeryd 6/8/03
	if (nOpenFlags & CFile::typeText && !(nOpenFlags & CFile::modeCreate) && !(nOpenFlags & CFile::modeWrite ))
	{
		m_bIsUnicodeText = IsFileUnicode(sFilePath);

		// If it's Unicode, switch to binary mode
		if (m_bIsUnicodeText)
		{
			nOpenFlags ^= CFile::typeText;
			nOpenFlags |= CFile::typeBinary;
		}
	}

	m_nFlags = nOpenFlags;

	return nOpenFlags;
}

// --------------------------------------------------------------------------------------------
//
//	CStdioFileEx::IsFileUnicode()
//
// --------------------------------------------------------------------------------------------
// Returns:    bool
// Parameters: const CString& sFilePath
//
// Purpose:		Determines whether a file is Unicode by reading the first character and detecting
//					whether it's the Unicode byte marker.
// Notes:		None.
// Exceptions:	None.
//
/*static*/ bool CStdioFileEx::IsFileUnicode(const CString& sFilePath)
{
	CFile				file;
	bool				bIsUnicode = false;
	wchar_t			cFirstChar;
	CFileException	exFile;

	// Open file in binary mode and read first character
	if (file.Open(sFilePath, CFile::typeBinary | CFile::modeRead, &exFile))
	{
		// If byte is Unicode byte-order marker, let's say it's Unicode
		if (file.Read(&cFirstChar, sizeof(wchar_t)) > 0 && cFirstChar == (wchar_t)nUNICODE_BOM)
		{
			bIsUnicode = true;
		}

		file.Close();
	}
	else
	{
		// Handle error here if you like
	}

	return bIsUnicode;
}

unsigned long CStdioFileEx::GetCharCount()
{
	int				nCharSize;
	unsigned long	nByteCount, nCharCount = 0;

	if (m_pStream)
	{
		// Get size of chars in file
		nCharSize = m_bIsUnicodeText ? sizeof(wchar_t): sizeof(char);

		// If Unicode, remove byte order mark from count
		nByteCount = GetLength();
		
		if (m_bIsUnicodeText)
		{
			nByteCount = nByteCount - sizeof(wchar_t);
		}

		// Calc chars
		nCharCount = (nByteCount / nCharSize);
	}

	return nCharCount;
}

// Get the current user磗 code page
UINT CStdioFileEx::GetCurrentLocaleCodePage()
{
	_TCHAR	szLocalCodePage[10];
	UINT		nLocaleCodePage = 0;
	int		nLocaleChars = ::GetLocaleInfo(LOCALE_USER_DEFAULT, LOCALE_IDEFAULTANSICODEPAGE, szLocalCodePage, 10);

	// If OK
	if (nLocaleChars > 0)
	{
		nLocaleCodePage = (UINT)_ttoi(szLocalCodePage);
		ASSERT(nLocaleCodePage > 0);
	}
	else
	{
		ASSERT(false);
	}

	// O means either: no ANSI code page (Unicode-only locale?) or failed to get locale
	// In the case of Unicode-only locales, what do multibyte apps do? Answers on a postcard.
	return nLocaleCodePage;
}

// --------------------------------------------------------------------------------------------
//
//	CStdioFileEx::GetUnicodeStringFromMultiByteString()
//
// --------------------------------------------------------------------------------------------
// Returns:    int - num. of chars written (0 means error)
// Parameters: char *		szMultiByteString		(IN)		Multi-byte input string
//					wchar_t*		szUnicodeString		(OUT)		Unicode outputstring
//					int			nUnicodeBufferSize	(IN)		Size of Unicode output buffer in chars(IN)
//					UINT			nCodePage				(IN)		Code page used to perform conversion
//																			Default = -1 (Get local code page).
//
// Purpose:		Gets a Unicode string from a MultiByte string.
// Notes:		None.
// Exceptions:	None.
//
int CStdioFileEx::GetUnicodeStringFromMultiByteString(IN LPCSTR szMultiByteString, OUT wchar_t* szUnicodeString, IN int nUnicodeBufferSize, IN int nCodePage)
{
	bool		bOK = true;
	int		nCharsWritten = 0;
		
	if (szUnicodeString && szMultiByteString)
	{
		// If no code page specified, take default for system
		if (nCodePage == -1)
		{
			nCodePage = GetACP();
		}

		try 
		{
			// Zero out buffer first. NB: nUnicodeBufferSize is NUMBER OF CHARS, NOT BYTES!
			memset((void*)szUnicodeString, '\0', sizeof(wchar_t) *
				nUnicodeBufferSize);

			// When converting to UTF8, don't set any flags (see Q175392).
			nCharsWritten = MultiByteToWideChar((UINT)nCodePage,
				(nCodePage==CP_UTF8 ? 0:MB_PRECOMPOSED), // Flags
				szMultiByteString,-1,szUnicodeString,nUnicodeBufferSize);
		}
		catch(...)
		{
			TRACE(_T("Controlled exception in MultiByteToWideChar!\n"));
		}
	}

	// Now fix nCharsWritten
	if (nCharsWritten > 0)
	{
		nCharsWritten--;
	}
	
	ASSERT(nCharsWritten > 0);
	return nCharsWritten;
}

// --------------------------------------------------------------------------------------------
//
//	CStdioFileEx::GetMultiByteStringFromUnicodeString()
//
// --------------------------------------------------------------------------------------------
// Returns:    int - number of characters written. 0 means error
// Parameters: wchar_t *	szUnicodeString			(IN)	Unicode input string
//					char*			szMultiByteString			(OUT)	Multibyte output string
//					int			nMultiByteBufferSize		(IN)	Multibyte buffer size (chars)
//					UINT			nCodePage					(IN)	Code page used to perform conversion
//																			Default = -1 (Get local code page).
//
// Purpose:		Gets a MultiByte string from a Unicode string
// Notes:		Added fix by Andy Goodwin: make buffer into int.
// Exceptions:	None.
//
int CStdioFileEx::GetMultiByteStringFromUnicodeString(wchar_t * szUnicodeString, char* szMultiByteString, 
																			int nMultiByteBufferSize, int nCodePage)
{
	BOOL		bUsedDefChar	= FALSE;
	int		nCharsWritten = 0;

	// Fix by Andy Goodwin: don't do anything if buffer is 0
	if ( nMultiByteBufferSize > 0 )
	{
		if (szUnicodeString && szMultiByteString) 
		{
			// Zero out buffer first
			memset((void*)szMultiByteString, '\0', nMultiByteBufferSize);
			
			// If no code page specified, take default for system
			if (nCodePage == -1)
			{
				nCodePage = GetACP();
			}

			try 
			{
				// If writing to UTF8, flags, default char and boolean flag must be NULL
				nCharsWritten = WideCharToMultiByte((UINT)nCodePage, 
					(nCodePage==CP_UTF8 ? 0 : WC_COMPOSITECHECK | WC_SEPCHARS), // Flags
					szUnicodeString,-1, 
					szMultiByteString, 
					nMultiByteBufferSize, 
					(nCodePage==CP_UTF8 ? NULL:sDEFAULT_UNICODE_FILLER_CHAR),	// Filler char
					(nCodePage==CP_UTF8 ? NULL: &bUsedDefChar));						// Did we use filler char?

				// If no chars were written and the buffer is not 0, error!
				if (nCharsWritten == 0 && nMultiByteBufferSize > 0)
				{
					TRACE1("Error in WideCharToMultiByte: %d\n", ::GetLastError());
				}
			}
			catch(...) 
			{
				TRACE(_T("Controlled exception in WideCharToMultiByte!\n"));
			}
		} 
	}

	// Now fix nCharsWritten 
	if (nCharsWritten > 0)
	{
		nCharsWritten--;
	}
	
	return nCharsWritten;
}

//---------------------------------------------------------------------------------------------------
//
//	CStdioFileEx::GetRequiredMultiByteLengthForUnicodeString()
//
//---------------------------------------------------------------------------------------------------
// Returns:    int
// Parameters: wchar_t * szUnicodeString,int nCodePage=-1
//
// Purpose:		Obtains the multi-byte buffer size needed to accommodate a converted Unicode string.
//	Notes:		We can't assume that the buffer length is simply equal to the number of characters
//					because that wouldn't accommodate multibyte characters!
//
/*static*/ int CStdioFileEx::GetRequiredMultiByteLengthForUnicodeString(wchar_t * szUnicodeString,int nCodePage /*=-1*/)
{
	int nCharsNeeded = 0;

	try 
	{
		// If no code page specified, take default for system
		if (nCodePage == -1)
		{
			nCodePage = GetACP();
		}

		// If writing to UTF8, flags, default char and boolean flag must be NULL
		nCharsNeeded = WideCharToMultiByte((UINT)nCodePage, 
			(nCodePage==CP_UTF8 ? 0 : WC_COMPOSITECHECK | WC_SEPCHARS), // Flags
			szUnicodeString,-1, 
			NULL, 
			0,	// Calculate required buffer, please! 
			(nCodePage==CP_UTF8 ? NULL:sDEFAULT_UNICODE_FILLER_CHAR),	// Filler char
			NULL);
	}
	catch(...) 
	{
		TRACE(_T("Controlled exception in WideCharToMultiByte!\n"));
	}

	return nCharsNeeded;
}