cmemory.cpp

日本的开源编辑器源码

					cmemBuf.Append( (char*)uchDecode, iByteIndex );
				}
			}
			delete [] pBuf;
		}
		nBgn = nPos;
		while( nBgn < nSrcLen && ( pszSrc[nBgn] == CR || pszSrc[nBgn] == LF ) ){
			nBgn++;
		}
	}
	SetData( cmemBuf.GetPtr(), cmemBuf.m_nDataLen );
	return;
}



/* Quoted-Printableデコード */
long CMemory::QuotedPrintable_Decode( char* pszSrc, long nSrcLen )
{
	int			i;
	char*		pszDes;
	long		lDesSize;
	char		szHex[3];
	int			nHex;

	memset( szHex, 0, 3 );
	pszDes = new char [nSrcLen + 1];
	memset( pszDes, 0, nSrcLen + 1 );
	lDesSize = 0;
	for( i = 0; i < nSrcLen; i++ ){
		if( pszSrc[i] == '=' ){
			szHex[0] = pszSrc[i + 1];
			szHex[1] = pszSrc[i + 2];
			sscanf( szHex, "%x", &nHex );
			pszDes[lDesSize] = (char)nHex;
			lDesSize++;
			i += 2;
		}else{
			pszDes[lDesSize] = pszSrc[i];
			lDesSize++;
		}
	}
	pszDes[lDesSize] = 0;
	memcpy( (char *)pszSrc, (const char *)pszDes, lDesSize );
	delete [] pszDes;
	return lDesSize;
}





/* JIS→SJIS変換 */
long CMemory::MemJIStoSJIS( unsigned char* pszSrc, long nSrcLen )
{
	int				i, j;
	char*			pszDes;
	unsigned short	sCode;

	pszDes = new char [nSrcLen + 1];
	memset( pszDes, 0, nSrcLen + 1 );
	j = 0;
	for( i = 0; i < nSrcLen - 1; i++ ){
		sCode = (unsigned short)_mbcjistojms(
			(unsigned int)
			(((unsigned short)pszSrc[i	  ] << 8) |
			 ((unsigned short)pszSrc[i + 1]))
		);
		if( sCode != 0 ){
			pszDes[j	] = (unsigned char)(sCode >> 8);
			pszDes[j + 1] = (unsigned char)(sCode);
			++i;
			j+=2;
		}else{
			pszDes[j] = pszSrc[i];
			j++;
		}
	}
	pszDes[j] = 0;
	memcpy( (char *)pszSrc, (const char *)pszDes, j );
	delete [] pszDes;
	return j;
}



/************************************************************************
*
*【関数名】
*	EUCToSJIS
*
*【機能】
*	指定範囲のバッファ内にEUC漢字コード
*	があればSJIS全角コードに変換する。	//Sept. 1, 2000 jepro 'シフト'を'S'に変更
*	半角文字は変換せずにそのまま残す。
*
*	制御文字CRLF以外のバイナリコードが混入している場合には結果が不定と
*	なることがあるので注意。
*	バッファの最後に漢字コードの1バイト目だけがあると困る
*
*【入力】	なし
*
*【戻り値】	なし
*
************************************************************************/
/* EUC→SJISコード変換 */
void CMemory::EUCToSJIS( void )
{
	char*			pBuf = m_pData;
	int				nBufLen = m_nDataLen;
	int				nPtr = 0L;
	int				nPtrDes = 0L;
	char*			pszDes = new char[nBufLen];
	unsigned int	sCode;

	while( nPtr < nBufLen ){
		if( (unsigned char)pBuf[nPtr] == (unsigned char)0x8e && nPtr < nBufLen - 1 ){
			/* 半角カタカナ */
			pszDes[nPtrDes] = pBuf[nPtr + 1];
			nPtrDes++;
			nPtr += 2;
		}else
		/* EUC漢字コードか? */
		if( nPtr < nBufLen - 1 && IsEUCKan1(pBuf[nPtr]) && IsEUCKan2(pBuf[nPtr + 1L]) ){
			/* 通常のJISコードに変換 */
			pBuf[nPtr	  ] &= 0x7f;
			pBuf[nPtr + 1L] &= 0x7f;

			/* SJISコードに変換 */	//Sept. 1, 2000 jepro 'シフト'を'S'に変更
			sCode = (unsigned short)_mbcjistojms(
				(unsigned int)
				(((unsigned short)pBuf[nPtr] << 8) |
				 ((unsigned short)pBuf[nPtr + 1]))
			);
			if( sCode != 0 ){
				pszDes[nPtrDes	  ] = (unsigned char)(sCode >> 8);
				pszDes[nPtrDes + 1] = (unsigned char)(sCode);
				nPtrDes += 2;;
				nPtr += 2;
			}else{
				pszDes[nPtrDes] = pBuf[nPtr];
				nPtrDes++;
				nPtr++;
			}
		}else{
			pszDes[nPtrDes] = pBuf[nPtr];
			nPtrDes++;
			nPtr++;
		}
	}
	SetData( pszDes, nPtrDes );
	delete [] pszDes;
	return;
}




/* SJIS→EUCコード変換 */
void CMemory::SJISToEUC( void )
{
	unsigned char*	pBuf = (unsigned char*)m_pData;
	int				nBufLen = m_nDataLen;
	int				i;
	int				nCharChars;
	unsigned char*	pDes;
	int				nDesIdx;
	unsigned short	sCode;

	pDes = new unsigned char[nBufLen * 2];
	nDesIdx = 0;
	for( i = 0; i < nBufLen; ++i ){
		nCharChars = CMemory::MemCharNext( (const char *)pBuf, nBufLen, (const char *)&(pBuf[i]) ) - (const char*)&(pBuf[i]);
		if( nCharChars == 1 ){
			if( pBuf[i] >= (unsigned char)0x80 ){
				/* 半角カタカナ */
				pDes[nDesIdx	] = (unsigned char)0x8e;
				pDes[nDesIdx + 1] = pBuf[i];
				nDesIdx += 2;
			}else{
				pDes[nDesIdx] = pBuf[i];
				nDesIdx++;
			}
		}else
		if( nCharChars == 2 ){
			/* 全角文字 */
			//	Oct. 3, 2002 genta IBM拡張文字対応
			sCode =	(unsigned short)_mbcjmstojis_ex( pBuf + i );
			if(sCode != 0){
				pDes[nDesIdx	] = (unsigned char)0x80 | (unsigned char)(sCode >> 8);
				pDes[nDesIdx + 1] = (unsigned char)0x80 | (unsigned char)(sCode);
				nDesIdx += 2;
				++i;
			}else{
				pDes[nDesIdx	] = pBuf[i];
				pDes[nDesIdx + 1] = pBuf[i + 1];
				nDesIdx += 2;
				++i;
			}
		}else
		if( nCharChars > 0 ){
			i += nCharChars - 1;
		}
	}
	SetData( (const char *)pDes, nDesIdx );
	delete [] pDes;
	return;
}

/****
UTF-8のコード
	ビット列		内容
	0xxx xxxx	1バイトコード
	10xx xxxx	2バイトコード、3バイトコードの2, 3文字目
	110x xxxx	2バイトコードの先頭バイト
	1110 xxxx	3バイトコードの先頭バイト

UTF-8のエンコーディング
	フォーマット	Unicode			Unicodeビット列						UTFビット列				備考
	1バイトコード	\u0～\u7F		0000 0000 0aaa bbbb					0aaa bbbb
	2バイトコード	\u80～\u7FF		0000 0aaa bbbb cccc					110a aabb 10bb cccc
	3バイトコード	\u800～\uFFFF	aaaa bbbb cccc dddd					1110 aaaa 10bb bbcc 10cc dddd
	4バイトコード	Surrogate		110110wwwwzzzzyy + 110111yyyyxxxxxx 1111 0uuu 10uu zzzz 10yy yyyy 10xx xxxx Javaでは未使用
***/

/* UTF-8の文字か */
int CMemory::IsUTF8( const unsigned char* pData, int nDataLen )
{
	if( 3 <= nDataLen
	 && (unsigned char)0xe0 == (unsigned char)(pData[0] & (unsigned char)0xf0)
	 && (unsigned char)0x80 == (unsigned char)(pData[1] & (unsigned char)0xc0)
	 && (unsigned char)0x80 == (unsigned char)(pData[2] & (unsigned char)0xc0)
	){
		return 3;
	}
	if( 2 <= nDataLen
	 && (unsigned char)0xc0 == (unsigned char)(pData[0] & (unsigned char)0xe0)
	 && (unsigned char)0x80 == (unsigned char)(pData[1] & (unsigned char)0xc0)
	){
		return 2;
	}
//	if( 1 <= nDataLen
//	 && (unsigned char)0 == (unsigned char)(pData[0] & (unsigned char)0x80)
//	){
//		return 1;
//	}
	return 0;

}

int CMemory::DecodeUTF8toUnicode( const unsigned char* pData, int nUTF8Bytes, unsigned char* pDes )
{
	switch( nUTF8Bytes ){
	case 1:
		pDes[1] = 0x00;
		pDes[0] = pData[0];
		return 2;
	case 2:
		pDes[1] = ( pData[0] & 0x1c ) >> 2;
		pDes[0] = ( pData[0] << 6 ) | ( pData[1] & 0x3f );
		return 2;
	case 3:
		pDes[1] =  ( ( pData[0] & 0x0f ) << 4 )
				 | ( ( pData[1] & 0x3c ) >> 2 );
		pDes[0] =  (   pData[2] & 0x3f )
				 | ( ( pData[1] & 0x03 ) << 6 );
		return 2;
	}
	return 0;
}

/* UTF-8→SJISコード変換 */
void CMemory::UTF8ToSJIS( void )
{
	unsigned char*	pBuf = (unsigned char*)m_pData;
	int				nBufLen = m_nDataLen;
	int				i;
	int				j;
//	int				nCharChars;
	unsigned char*	pDes;
	int				k;
//	unsigned short	sCode;
	int				nUTF8Bytes;
	int				nUNICODEBytes;
	unsigned char	pWork[100];

	setlocale( LC_ALL, "Japanese" );
	pDes = new unsigned char[nBufLen * 2];
	k = 0;
	for( i = 0; i < nBufLen; ){
		/* UTF-8の文字か */
		nUTF8Bytes = IsUTF8( &pBuf[i], nBufLen - i );
		switch( nUTF8Bytes ){
		case 1:
		case 2:
		case 3:
			nUNICODEBytes = DecodeUTF8toUnicode( &pBuf[i], nUTF8Bytes, pWork );
//			memcpy( &pDes[k], pWork, nUNICODEBytes );
//			k += nUNICODEBytes;

			if( 2 == nUNICODEBytes ){
				j = wctomb( (char*)&(pDes[k]), ((wchar_t*)pWork)[0] );
				if( -1 == j ){
					memcpy( &pDes[k], &pBuf[i], nUTF8Bytes );
					k += nUTF8Bytes;
				}else{
					k += j;
				}
			}else{
				memcpy( &pDes[k], &pBuf[i], nUTF8Bytes );
				k += nUTF8Bytes;
			}

			i += nUTF8Bytes;
			break;
		default:
			pDes[k] = pBuf[i];
			++i;
			k++;
			break;
		}
	}
	SetData( (const char *)pDes, k );
	delete [] pDes;
	return;

}





/* コード変換 SJIS→UTF-8 */
void CMemory::SJISToUTF8( void )
{
	/* コード変換 SJIS→Unicode */
	SJISToUnicode();

	/* コード変換 Unicode→UTF-8 */
	UnicodeToUTF8();
	return;
}

/* コード変換 SJIS→UTF-7 */
void CMemory::SJISToUTF7( void )
{
	/* コード変換 SJIS→Unicode */
	SJISToUnicode();

	/* コード変換 Unicode→UTF-7 */
	UnicodeToUTF7();
	return;
}




/* コード変換 Unicode→UTF-8 */
void CMemory::UnicodeToUTF8( void )
{
	wchar_t*		pUniBuf = (wchar_t*)m_pData;
//	char*			pBuf = (char*)m_pData;
	int				nBufLen = m_nDataLen;
	int				i;
//	int				j;
//	int				nCharChars;
	unsigned char*	pWork;
	unsigned char*	pDes;
	int				k;
//	unsigned short	sCode;
//	int				nUTF8Bytes;
//	int				nUNICODEBytes;
//	unsigned char	pWork[100];
//	unsigned char*	pBufUTF8;
//	wchar_t*		pUniBuf;


	setlocale( LC_ALL, "Japanese" );
	k = 0;
	for( i = 0; i < (int)(nBufLen / sizeof( wchar_t )); ++i ){
		if( 0x0000 <= pUniBuf[i] && 0x007f >= pUniBuf[i] ){
			k += 1;
		}else
		if( 0x0080 <= pUniBuf[i] && 0x07ff >= pUniBuf[i] ){
			k += 2;
		}else
		if( 0x0800 <= pUniBuf[i] && 0xffff >= pUniBuf[i] ){
			k += 3;
		}
	}
	pDes = new unsigned char[k + 1];
	memset( pDes, 0, k + 1 );
	k = 0;
	for( i = 0; i < (int)(nBufLen / sizeof( wchar_t )); ++i ){
		pWork = (unsigned char*)&((pUniBuf[i]));
//		pWork = (unsigned char*)pUniBuf;
//		pWork += sizeof( wchar_t ) * i;

		if( 0x0000 <= pUniBuf[i] && 0x007f >= pUniBuf[i] ){
			pDes[k] = pWork[0];
			k += 1;
		}else
		if( 0x0080 <= pUniBuf[i] && 0x07ff >= pUniBuf[i] ){
			pDes[k] = ( ( pWork[0] & 0xc0 ) >> 6 )
					| ( ( pWork[1] & 0x07 ) << 2 )
					| 0xc0;
			pDes[k + 1] = ( pWork[0] & 0x3f ) | 0x80;
			k += 2;
		}else
		if( 0x0800 <= pUniBuf[i] && 0xffff >= pUniBuf[i] ){
			pDes[k] = ( ( pWork[1] & 0xf0 ) >> 4 )
						| 0xe0;
			pDes[k + 1] = 0x80
						| ( ( pWork[1] & 0x0f ) << 2 )
						| ( ( pWork[0] & 0xc0 ) >> 6 );
			pDes[k + 2] = 0x80
						| ( pWork[0] & 0x3f );

			k += 3;
		}else{
		}
	}
	SetData( (const char *)pDes, k );
	delete [] pDes;
	return;
}

/* コード変換 UTF-7→SJIS

	UTF-7についてはRFC 2152参照．
*/
void CMemory::UTF7ToSJIS( void )
{
	char*		pBuf = (char*)m_pData;
	int			nBufLen = m_nDataLen;
	int			i;
	char*		pszWork;
	int			nWorkLen;
	char*		pDes;
	int			k;
	BOOL		bBASE64;
	int			nBgn;
//	CMemory		cmemWork;
	CMemory*	pcmemWork;

	pcmemWork = new CMemory;

	pDes = new char[nBufLen + 1];
	setlocale( LC_ALL, "Japanese" );
	k = 0;
	bBASE64 = FALSE;
	// 2002.10.05 Moca < を<= に変更 ただし，pBuf[nBufLen]は存在しないことに注意
	for( i = 0; i <= nBufLen; ++i ){
		if( !bBASE64 ){
			if( i < nBufLen - 1
			  &&  '+' == pBuf[i]
			  &&  '-' == pBuf[i + 1]
			){
				pDes[k] = '+';
				k++;
				++i;
			}else
			if( i < nBufLen - 2
			  &&  '+' == pBuf[i]
			  &&  -1 != IsBASE64Char( pBuf[i + 1] ) /* 文字がBase64のデータか */
			){
				nBgn = i + 1;
				bBASE64 = TRUE;
			}else
			if( i < nBufLen ){
				pDes[k] = pBuf[i];
				k++;
			}
		}else{
			if( i == nBufLen ||				//	2002.10.25 Moca データ終端時もデコード
				-1 == IsBASE64Char( pBuf[i] ) ){	//	Oct. 10, 2000 genta
				nWorkLen = i - nBgn;
				BOOL bSuccess;
				bSuccess = TRUE;

				if( 3 <= nWorkLen ){