docdecoder.cpp

此文件为Word 格式文件.DOC转换为TXT文件的源代码。

{
	int	iBytes, iFodoOff, iInfoLen;
	int	iIndex, iSize, iCol;
	int	iPosCurr, iPosPrev;
	bool	bFound18, bFound19, bFoundBE;


	iBytes = 2 * (int)ucGetByte(iFodo, aucFpage);
	iFodoOff = 3;
	bFound18 = false;
	bFound19 = false;
	bFoundBE = false;
	while (iBytes >= iFodoOff + 1) {
		iInfoLen = 0;
		switch (ucGetByte(iFodo + iFodoOff, aucFpage)) {
		case 0x18:
			if (ucGetByte(iFodo + iFodoOff + 1,
					aucFpage) == 0x01) {
				bFound18 = true;
			}
			break;
		case 0x19:
			if (ucGetByte(iFodo + iFodoOff + 1,
					aucFpage) == 0x01) {
				bFound19 = true;
			}
			break;
		case 0xbe:
			iSize = (int)ucGetByte(iFodo + iFodoOff + 1, aucFpage);
			if (iSize < 6 || iBytes < iFodoOff + 7) {
				iInfoLen = 1;
				break;
			}
			iCol = (int)ucGetByte(iFodo + iFodoOff + 3, aucFpage);
			if (iCol < 1 ||
			    iBytes < iFodoOff + 3 + (iCol + 1) * 2) {
				iInfoLen = 1;
				break;
			}
			if (iCol >= elementsof(pRow->asColumnWidth)) {
				return found_nothing;
//				werr(1, "The number of columns is corrupt");
			}
			pRow->ucNumberOfColumns = iCol;
			pRow->iColumnWidthSum = 0;
			iPosPrev = (int)(short)usGetWord(
					iFodo + iFodoOff + 4,
					aucFpage);
			for (iIndex = 0; iIndex < iCol; iIndex++) {
				iPosCurr = (int)(short)usGetWord(
					iFodo + iFodoOff + 6 + iIndex * 2,
					aucFpage);
				pRow->asColumnWidth[iIndex] =
							iPosCurr - iPosPrev;
				pRow->iColumnWidthSum +=
					pRow->asColumnWidth[iIndex];
				iPosPrev = iPosCurr;
			}
			bFoundBE = true;
			break;
		default:
			break;
		}
		if (iInfoLen <= 0) {
			iInfoLen =
				iGet6InfoLength(iFodo + iFodoOff, aucFpage);
		}
		iFodoOff += iInfoLen;
	}
	if (bFound18 && bFound19 && bFoundBE) {
		return found_end_of_row;
	}
	if (bFound18) {
		return found_a_cell;
	}
	return found_nothing;
} /* end of eGet6RowInfo */


bool
CDocDecoder::bGet6StyleInfo(int iFodo, const unsigned char *aucFpage,
		style_block_type *pStyle)
{
	int	iBytes, iFodoOff, iInfoLen;
	int	iTmp, iDel, iAdd;


	iBytes = 2 * (int)ucGetByte(iFodo, aucFpage);
	iFodoOff = 3;
	if (iBytes < 1) {
		return false;
	}
	(void)memset(pStyle, 0, sizeof(*pStyle));
	pStyle->ucStyle = ucGetByte(iFodo + 1, aucFpage);
	while (iBytes >= iFodoOff + 1) {
		iInfoLen = 0;
		switch (ucGetByte(iFodo + iFodoOff, aucFpage)) {
		case 0x05:
			pStyle->ucAlignment = ucGetByte(
					iFodo + iFodoOff + 1, aucFpage);
			break;
		case 0x0c:
			iTmp = (int)ucGetByte(
					iFodo + iFodoOff + 1, aucFpage);
			if (iTmp >= 1) {
				pStyle->ucListType = ucGetByte(
					iFodo + iFodoOff + 2, aucFpage);
				pStyle->bInList = true;
			}
			if (iTmp >= 21) {
				pStyle->ucListCharacter = ucGetByte(
					iFodo + iFodoOff + 22, aucFpage);
			}
			break;
		case 0x0d:
			iTmp = (int)ucGetByte(
					iFodo + iFodoOff + 1, aucFpage);
			switch (iTmp) {
			case 0x0c:
				pStyle->bUnmarked = true;
				/* No break or return */
			case 0x0a:
			case 0x0b:
				pStyle->bInList = true;
				break;
			default:
				break;
			}
			break;
		case 0x0f:
			iTmp = (int)ucGetByte(iFodo + iFodoOff + 1, aucFpage);
			if (iTmp < 2) {
				iInfoLen = 1;
				break;
			}
			iDel = (int)ucGetByte(iFodo + iFodoOff + 2, aucFpage);
			if (iTmp < 2 + 2 * iDel) {
				iInfoLen = 1;
				break;
			}
			iAdd = (int)ucGetByte(
				iFodo + iFodoOff + 3 + 2 * iDel, aucFpage);
			if (iTmp < 2 + 2 * iDel + 2 * iAdd) {
				iInfoLen = 1;
				break;
			}
			break;
		case 0x10:
			pStyle->sRightIndent = (short)usGetWord(
					iFodo + iFodoOff + 1, aucFpage);
			break;
		case 0x11:
			pStyle->sLeftIndent = (short)usGetWord(
					iFodo + iFodoOff + 1, aucFpage);
			break;
		case 0x12:
			iTmp = (short)usGetWord(iFodo + iFodoOff + 1, aucFpage);
			pStyle->sLeftIndent += iTmp;
			if (pStyle->sLeftIndent < 0) {
				pStyle->sLeftIndent = 0;
			}
			break;
		default:
			break;
		}
		if (iInfoLen <= 0) {
			iInfoLen =
				iGet6InfoLength(iFodo + iFodoOff, aucFpage);
		}
		iFodoOff += iInfoLen;
	}
	return true;
} /* end of bGet6StyleInfo */


void
CDocDecoder::vGet6ParInfo(unsigned char *pFile, int iStartBlock,
	const int *aiBBD, int iBBDLen,
	const unsigned char *aucHeader)
{
	row_block_type		tRow;
	style_block_type	tStyle;
	unsigned short	*ausParfPage;
	unsigned char	*aucBuffer;
	size_t	tSize;
	int	iLenOld, iLen;
	int	iParmOffset, iParmOffsetFirst, iParmOffsetLast;
	int	iIndex, iIndex2, iCount, iOffset, iNbr, iFodo;
	int	iParfFirstPage, iBeginParfInfo,iParfInfoLen, iParfPageNum;
	unsigned char	aucFpage[BIG_BLOCK_SIZE];


	iBeginParfInfo = (int)ulGetLong(0xc0, aucHeader);
	iParfInfoLen = (int)ulGetLong(0xc4, aucHeader);

	aucBuffer = (unsigned char *)xmalloc(iParfInfoLen);
	if (!bReadBuffer(pFile, iStartBlock,
			aiBBD, iBBDLen, BIG_BLOCK_SIZE,
			aucBuffer, iBeginParfInfo, iParfInfoLen)) {
		aucBuffer = (unsigned char *)xfree(aucBuffer);
		return;
	}

	iLen = (iParfInfoLen - 4) / 6;
	tSize = iLen * sizeof(unsigned short);
	ausParfPage = (unsigned short *)xmalloc(tSize);
	for (iIndex = 0, iOffset = (iLen + 1) * 4;
	     iIndex < iLen;
	     iIndex++, iOffset += 2) {
		 ausParfPage[iIndex] = usGetWord(iOffset, aucBuffer);
	}
	aucBuffer = (unsigned char *)xfree(aucBuffer);
	aucBuffer = NULL;
	iParfPageNum = (int)usGetWord(0x190, aucHeader);
	if (iLen < iParfPageNum) {
		/* Replace ParfPage by a longer version */
		iLenOld = iLen;
		iParfFirstPage = (int)usGetWord(0x18c, aucHeader);
		iLen += iParfPageNum - 1;
		tSize = iLen * sizeof(unsigned short);
		ausParfPage = (unsigned short *)xrealloc(ausParfPage, tSize);
		/* Add new values */
		iCount = iParfFirstPage + 1;
		for (iIndex = iLenOld; iIndex < iLen; iIndex++) {
			ausParfPage[iIndex] = iCount;
			iCount++;
		}
	}

	(void)memset(&tStyle, 0, sizeof(tStyle));
	(void)memset(&tRow, 0, sizeof(tRow));
	iParmOffsetFirst = -1;
	for (iIndex = 0; iIndex < iLen; iIndex++) {
		if (!bReadBuffer(pFile, iStartBlock,
				aiBBD, iBBDLen, BIG_BLOCK_SIZE,
				aucFpage,
				(int)ausParfPage[iIndex] * BIG_BLOCK_SIZE,
				BIG_BLOCK_SIZE)) {
			break;
		}
		iNbr = (int)ucGetByte(0x1ff, aucFpage);
		for (iIndex2 = 0; iIndex2 < iNbr; iIndex2++) {
			iFodo = 2 * (int)ucGetByte(
				(iNbr + 1) * 4 + iIndex2 * 7, aucFpage);
			if (iFodo <= 0) {
				continue;
			}
			if (bGet6StyleInfo(iFodo, aucFpage, &tStyle)) {
				iParmOffset = (int)ulGetLong(
						iIndex2 * 4, aucFpage);
				tStyle.iOffset = iTextOffset2FileOffset(
							iParmOffset);
				vAdd2StyleInfoList(&tStyle);
				(void)memset(&tStyle, 0, sizeof(tStyle));
			}
			switch (eGet6RowInfo(iFodo, aucFpage, &tRow)) {
			case found_a_cell:
				if (iParmOffsetFirst >= 0) {
					break;
				}
				iParmOffsetFirst = (int)ulGetLong(
						iIndex2 * 4, aucFpage);
				tRow.iOffsetStart = iTextOffset2FileOffset(
							iParmOffsetFirst);
				break;
			case found_end_of_row:
				iParmOffsetLast = (int)ulGetLong(
						iIndex2 * 4, aucFpage);
				tRow.iOffsetEnd = iTextOffset2FileOffset(
							iParmOffsetLast);
				vAdd2RowInfoList(&tRow);
				(void)memset(&tRow, 0, sizeof(tRow));
				iParmOffsetFirst = -1;
				break;
			default:
				break;
			}
		}
	}
	ausParfPage = (unsigned short *)xfree(ausParfPage);
} /* end of vGet6ParInfo */


int
CDocDecoder::iGet8InfoLength(int iByteNbr, const unsigned char *aucFpage)
{
	int	iTmp, iDel, iAdd;
	unsigned short	usOpCode;

	usOpCode = usGetWord(iByteNbr, aucFpage);

	switch (usOpCode & 0xe000) {
	case 0x0000: case 0x2000:
		return 3;
	case 0x4000: case 0x8000: case 0xa000:
		return 4;
	case 0xe000:
		return 5;
	case 0x6000:
		return 6;
	case 0xc000:
		iTmp = (int)ucGetByte(iByteNbr + 2, aucFpage);
		if (usOpCode == 0xc615 && iTmp == 255) {
			iDel = (int)ucGetByte(iByteNbr + 3, aucFpage);
			iAdd = (int)ucGetByte(
					iByteNbr + 4 + iDel * 4, aucFpage);
			iTmp = 2 + iDel * 4 + iAdd * 3;
		}
		return 3 + iTmp;
	default:
		return 1;
	}
} /* end of iGet8InfoLength */


row_info_enum
CDocDecoder::eGet8RowInfo(int iFodo, const unsigned char *aucFpage, row_block_type *pRow)
{
	int	iBytes, iFodoOff, iInfoLen;
	int	iIndex, iSize, iCol;
	int	iPosCurr, iPosPrev;
	bool	bFound2416, bFound2417, bFoundd608;


	iBytes = 2 * (int)ucGetByte(iFodo, aucFpage);
	iFodoOff = 3;
	if (iBytes == 0) {
		iFodo++;
		iBytes = 2 * (int)ucGetByte(iFodo, aucFpage);
	}
	bFound2416 = false;
	bFound2417 = false;
	bFoundd608 = false;
	while (iBytes >= iFodoOff + 2) {
		iInfoLen = 0;
		switch (usGetWord(iFodo + iFodoOff, aucFpage)) {
		case 0x2416:
			if (ucGetByte(iFodo + iFodoOff + 2,
					aucFpage) == 0x01) {
				bFound2416 = true;
			}
			break;
		case 0x2417:
			if (ucGetByte(iFodo + iFodoOff + 2,
					aucFpage) == 0x01) {
				bFound2417 = true;
			}
			break;
		case 0xd608:
			iSize = (int)ucGetByte(iFodo + iFodoOff + 2, aucFpage);
			if (iSize < 6 || iBytes < iFodoOff + 8) {
				iInfoLen = 2;
				break;
			}
			iCol = (int)ucGetByte(iFodo + iFodoOff + 4, aucFpage);
			if (iCol < 1 ||
			    iBytes < iFodoOff + 4 + (iCol + 1) * 2) {
				iInfoLen = 2;
				break;
			}
			if (iCol >= elementsof(pRow->asColumnWidth)) {
				return found_nothing;
//				werr(1, "The number of columns is corrupt");
			}
			pRow->ucNumberOfColumns = iCol;
			pRow->iColumnWidthSum = 0;
			iPosPrev = (int)(short)usGetWord(
					iFodo + iFodoOff + 5,
					aucFpage);
			for (iIndex = 0; iIndex < iCol; iIndex++) {
				iPosCurr = (int)(short)usGetWord(
					iFodo + iFodoOff + 7 + iIndex * 2,
					aucFpage);
				pRow->asColumnWidth[iIndex] =
							iPosCurr - iPosPrev;
				pRow->iColumnWidthSum +=
					pRow->asColumnWidth[iIndex];
				iPosPrev = iPosCurr;
			}
			bFoundd608 = true;
			break;
		default:
			break;
		}
		if (iInfoLen <= 0) {
			iInfoLen =
				iGet8InfoLength(iFodo + iFodoOff, aucFpage);
		}
		iFodoOff += iInfoLen;
	}
	if (bFound2416 && bFound2417 && bFoundd608) {
		return found_end_of_row;
	}
	if (bFound2416) {
		return found_a_cell;
	}
	return found_nothing;
} /* end of eGet8RowInfo */


bool
CDocDecoder::bGet8StyleInfo(int iFodo, const unsigned char *aucFpage,
		style_block_type *pStyle)
{
	int	iBytes, iFodoOff, iInfoLen;
	int	iTmp, iDel, iAdd;


	iBytes = 2 * (int)ucGetByte(iFodo, aucFpage);
	iFodoOff = 3;
	if (iBytes == 0) {
		iFodo++;
		iBytes = 2 * (int)ucGetByte(iFodo, aucFpage);
	}
	if (iBytes < 2) {
		return false;
	}
	(void)memset(pStyle, 0, sizeof(*pStyle));
	iTmp = usGetWord(iFodo + 1, aucFpage);
	if (iTmp >= 0 && iTmp <= UCHAR_MAX) {
		pStyle->ucStyle = (unsigned char)iTmp;
	}
	while (iBytes >= iFodoOff + 2) {
		iInfoLen = 0;
		switch (usGetWord(iFodo + iFodoOff, aucFpage)) {
		case 0x2403:
			pStyle->ucAlignment = ucGetByte(
					iFodo + iFodoOff + 2, aucFpage);
			break;
		case 0x260a:
			pStyle->bInList = true;
			pStyle->ucListLevel =
				ucGetByte(iFodo + iFodoOff + 2, aucFpage);
			break;
		case 0x460b:
			break;
		case 0x4610:
			iTmp = (short)usGetWord(iFodo + iFodoOff + 2, aucFpage);
			pStyle->sLeftIndent += iTmp;
			if (pStyle->sLeftIndent < 0) {
				pStyle->sLeftIndent = 0;
			}
			break;
		case 0x6c0d:
			iTmp = (int)ucGetByte(iFodo + iFodoOff + 2, aucFpage);
			if (iTmp < 2) {
				iInfoLen = 1;
				break;
			}
			iDel = (int)ucGetByte(iFodo + iFodoOff + 3, aucFpage);
			if (iTmp < 2 + 2 * iDel) {
				iInfoLen = 1;
				break;
			}
			iAdd = (int)ucGetByte(
				iFodo + iFodoOff + 4 + 2 * iDel, aucFpage);
			if (iTmp < 2 + 2 * iDel + 2 * iAdd) {
				iInfoLen = 1;
				break;
			}
			break;
		case 0x840e:
			pStyle->sRightIndent = (short)usGetWord(
					iFodo + iFodoOff + 2, aucFpage);
			break;
		case 0x840f:
			pStyle->sLeftIndent = (short)usGetWord(
					iFodo + iFodoOff + 2, aucFpage);
			break;
		default:
			break;
		}
		if (iInfoLen <= 0) {
			iInfoLen =
				iGet8InfoLength(iFodo + iFodoOff, aucFpage);
		}
		iFodoOff += iInfoLen;
	}
	return true;
} /* end of bGet8StyleInfo */


void
CDocDecoder::vGet8ParInfo(unsigned char *pFile, const pps_info_type *pPPS,
	const int *aiBBD, int iBBDLen, const int *aiSBD, int iSBDLen,
	const unsigned char *aucHeader)
{
	row_block_type		tRow;
	style_block_type	tStyle;
	const int	*aiBlockDepot;
	int		*aiParfPage;
	unsigned char	*aucBuffer;
	size_t		tSize;
	int	iTableStartBlock, iTableSize, iBlockDepotLen, iBlockSize;
	int	iLen, iOffset;
	int	iParmOffset, iParmOffsetFirst, iParmOffsetLast;
	int	iIndex, iIndex2, iNbr, iFodo;
	int	iBeginParfInfo, iParfInfoLen;
	unsigned short	usDocStatus;
	unsigned char	aucFpage[BIG_BLOCK_SI