prop8.c

这是一个同样来自贝尔实验室的和UNIX有着渊源的操作系统, 其简洁的设计和实现易于我们学习和理解

/*
 * prop8.c
 * Copyright (C) 1998-2005 A.J. van Os; Released under GNU GPL
 *
 * Description:
 * Read the property information from a MS Word 8, 9,10 or 11 file
 *
 * Word  8 is better known as Word 97 or as Word 98 for Mac
 * Word  9 is better known as Word 2000 or as Word 2001 for Mac
 * Word 10 is better known as Word 2002 or as Word XP
 * Word 11 is better known as Word 2003
 */

#include <stdlib.h>
#include <string.h>
#include "antiword.h"

#define DEFAULT_LISTCHAR	0x002e	/* A full stop */


/*
 * iGet8InfoLength - the length of the information for Word 8/9/10/11 files
 */
static int
iGet8InfoLength(int iByteNbr, const UCHAR *aucGrpprl)
{
	int	iTmp, iDel, iAdd;
	USHORT	usOpCode;

	usOpCode = usGetWord(iByteNbr, aucGrpprl);

	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, aucGrpprl);
		if (usOpCode == 0xc615 && iTmp == 255) {
			iDel = (int)ucGetByte(iByteNbr + 3, aucGrpprl);
			iAdd = (int)ucGetByte(
					iByteNbr + 4 + iDel * 4, aucGrpprl);
			iTmp = 2 + iDel * 4 + iAdd * 3;
		}
		return 3 + iTmp;
	default:
		DBG_HEX(usOpCode);
		DBG_FIXME();
		return 1;
	}
} /* end of iGet8InfoLength */

/*
 * aucFillInfoBuffer - fill the information buffer
 *
 * Returns the information buffer when successful, otherwise NULL
 */
static UCHAR *
aucFillInfoBuffer(FILE *pFile, const pps_type *pTable,
	const ULONG *aulBBD, size_t tBBDLen,
	const ULONG *aulSBD, size_t tSBDLen,
	ULONG ulBeginInfo, size_t tInfoLen)
{
	const ULONG	*aulBlockDepot;
	UCHAR	*aucBuffer;
	size_t	tBlockDepotLen, tBlockSize;

	fail(pFile == NULL || pTable == NULL);
	fail(aulBBD == NULL || aulSBD == NULL);
	fail(tInfoLen == 0);

	NO_DBG_DEC(pTable->ulSB);
	NO_DBG_HEX(pTable->ulSize);
	if (pTable->ulSize == 0) {
		DBG_MSG("No information");
		return NULL;
	}

	if (pTable->ulSize < MIN_SIZE_FOR_BBD_USE) {
		/* Use the Small Block Depot */
		aulBlockDepot = aulSBD;
		tBlockDepotLen = tSBDLen;
		tBlockSize = SMALL_BLOCK_SIZE;
	} else {
		/* Use the Big Block Depot */
		aulBlockDepot = aulBBD;
		tBlockDepotLen = tBBDLen;
		tBlockSize = BIG_BLOCK_SIZE;
	}
	aucBuffer = xmalloc(tInfoLen);
	if (!bReadBuffer(pFile, pTable->ulSB,
			aulBlockDepot, tBlockDepotLen, tBlockSize,
			aucBuffer, ulBeginInfo, tInfoLen)) {
		aucBuffer = xfree(aucBuffer);
		return NULL;
	}
	return aucBuffer;
} /* end of aucFillInfoBuffer */

/*
 * Build the lists with Document Property Information for Word 8/9/10/11 files
 */
void
vGet8DopInfo(FILE *pFile, const pps_type *pTable,
	const ULONG *aulBBD, size_t tBBDLen,
	const ULONG *aulSBD, size_t tSBDLen,
	const UCHAR *aucHeader)
{
	document_block_type	tDocument;
	UCHAR	*aucBuffer;
	ULONG	ulBeginDocpInfo, ulTmp;
	size_t	tDocpInfoLen;
	USHORT	usTmp;

	fail(pFile == NULL || pTable == NULL || aucHeader == NULL);
	fail(aulBBD == NULL || aulSBD == NULL);

	ulBeginDocpInfo = ulGetLong(0x192, aucHeader); /* fcDop */
	NO_DBG_HEX(ulBeginSectInfo);
	tDocpInfoLen = (size_t)ulGetLong(0x196, aucHeader); /* lcbDop */
	NO_DBG_DEC(tSectInfoLen);
	if (tDocpInfoLen < 28) {
		DBG_MSG("No Document information");
		return;
	}

	aucBuffer = aucFillInfoBuffer(pFile, pTable,
			aulBBD, tBBDLen, aulSBD, tSBDLen,
			ulBeginDocpInfo, tDocpInfoLen);
	if (aucBuffer == NULL) {
		return;
	}

	usTmp = usGetWord(0x00, aucBuffer);
	tDocument.ucHdrFtrSpecification = (UCHAR)(usTmp >> 8); /* grpfIhdt */
	tDocument.usDefaultTabWidth = usGetWord(0x0a, aucBuffer); /* dxaTab */
	ulTmp = ulGetLong(0x14, aucBuffer); /* dttmCreated */
	tDocument.tCreateDate = tConvertDTTM(ulTmp);
	ulTmp = ulGetLong(0x18, aucBuffer); /* dttmRevised */
	tDocument.tRevisedDate = tConvertDTTM(ulTmp);
	vCreateDocumentInfoList(&tDocument);

	aucBuffer = xfree(aucBuffer);
} /* end of vGet8DopInfo */

/*
 * Fill the section information block with information
 * from a Word 8/9/10/11 file.
 */
static void
vGet8SectionInfo(const UCHAR *aucGrpprl, size_t tBytes,
		section_block_type *pSection)
{
	UINT	uiIndex;
	int	iFodoOff, iInfoLen, iSize, iTmp;
	USHORT	usCcol;
	UCHAR	ucTmp;

	fail(aucGrpprl == NULL || pSection == NULL);

	iFodoOff = 0;
	while (tBytes >= (size_t)iFodoOff + 2) {
		iInfoLen = 0;
		switch (usGetWord(iFodoOff, aucGrpprl)) {
		case 0x3009:	/* bkc */
			ucTmp = ucGetByte(iFodoOff + 2, aucGrpprl);
			DBG_DEC(ucTmp);
			pSection->bNewPage = ucTmp != 0 && ucTmp != 1;
			break;
		case 0x3014:	/* grpfIhdt */
			pSection->ucHdrFtrSpecification =
					ucGetByte(iFodoOff + 2, aucGrpprl);
			break;
		case 0x500b:	/* ccolM1 */
			usCcol = 1 + usGetWord(iFodoOff + 2, aucGrpprl);
			DBG_DEC(usCcol);
			break;
		case 0xd202:	/* olstAnm */
			iSize = (int)ucGetByte(iFodoOff + 2, aucGrpprl);
			DBG_DEC_C(iSize != 212, iSize);
			for (uiIndex = 0, iTmp = iFodoOff + 3;
			     uiIndex < 9 && iTmp < iFodoOff + 3 + iSize - 15;
			     uiIndex++, iTmp += 16) {
				pSection->aucNFC[uiIndex] =
						ucGetByte(iTmp, aucGrpprl);
				DBG_DEC(pSection->aucNFC[uiIndex]);
				ucTmp = ucGetByte(iTmp + 3, aucGrpprl);
				DBG_HEX(ucTmp);
				if ((ucTmp & BIT(2)) != 0) {
					pSection->usNeedPrevLvl |=
							(USHORT)BIT(uiIndex);
				}
				if ((ucTmp & BIT(3)) != 0) {
					pSection->usHangingIndent |=
							(USHORT)BIT(uiIndex);
				}
			}
			DBG_HEX(pSection->usNeedPrevLvl);
			DBG_HEX(pSection->usHangingIndent);
			break;
		default:
			break;
		}
		if (iInfoLen <= 0) {
			iInfoLen = iGet8InfoLength(iFodoOff, aucGrpprl);
			fail(iInfoLen <= 0);
		}
		iFodoOff += iInfoLen;
	}
} /* end of vGet8SectionInfo */

/*
 * Build the lists with Section Property Information for Word 8/9/10/11 files
 */
void
vGet8SepInfo(FILE *pFile, const pps_info_type *pPPS,
	const ULONG *aulBBD, size_t tBBDLen,
	const ULONG *aulSBD, size_t tSBDLen,
	const UCHAR *aucHeader)
{
	section_block_type	tSection;
	ULONG	*aulSectPage, *aulCharPos;
	UCHAR	*aucBuffer, *aucFpage;
	ULONG	ulBeginOfText, ulTextOffset, ulBeginSectInfo;
	size_t	tSectInfoLen, tIndex, tOffset, tLen, tBytes;
	UCHAR	aucTmp[2];

	fail(pFile == NULL || pPPS == NULL || aucHeader == NULL);
	fail(aulBBD == NULL || aulSBD == NULL);

	ulBeginOfText = ulGetLong(0x18, aucHeader); /* fcMin */
	NO_DBG_HEX(ulBeginOfText);
	ulBeginSectInfo = ulGetLong(0xca, aucHeader); /* fcPlcfsed */
	NO_DBG_HEX(ulBeginSectInfo);
	tSectInfoLen = (size_t)ulGetLong(0xce, aucHeader); /* lcbPlcfsed */
	NO_DBG_DEC(tSectInfoLen);
	if (tSectInfoLen < 4) {
		DBG_DEC(tSectInfoLen);
		return;
	}

	aucBuffer = aucFillInfoBuffer(pFile, &pPPS->tTable,
			aulBBD, tBBDLen, aulSBD, tSBDLen,
			ulBeginSectInfo, tSectInfoLen);
	if (aucBuffer == NULL) {
		return;
	}
	NO_DBG_PRINT_BLOCK(aucBuffer, tSectInfoLen);

	/* Read the Section Descriptors */
	tLen = (tSectInfoLen - 4) / 16;
	/* Save the section offsets */
	aulCharPos = xcalloc(tLen, sizeof(ULONG));
	for (tIndex = 0, tOffset = 0;
	     tIndex < tLen;
	     tIndex++, tOffset += 4) {
		ulTextOffset = ulGetLong(tOffset, aucBuffer);
		NO_DBG_HEX(ulTextOffset);
		aulCharPos[tIndex] = ulBeginOfText + ulTextOffset;
		NO_DBG_HEX(aulCharPos[tIndex]);
	}
	/* Save the Sepx offsets */
	aulSectPage = xcalloc(tLen, sizeof(ULONG));
	for (tIndex = 0, tOffset = (tLen + 1) * 4;
	     tIndex < tLen;
	     tIndex++, tOffset += 12) {
		 aulSectPage[tIndex] = ulGetLong(tOffset + 2, aucBuffer);
		 NO_DBG_HEX(aulSectPage[tIndex]); /* fcSepx */
	}
	aucBuffer = xfree(aucBuffer);

	/* Read the Section Properties */
	for (tIndex = 0; tIndex < tLen; tIndex++) {
		if (aulSectPage[tIndex] == FC_INVALID) {
			vDefault2SectionInfoList(aulCharPos[tIndex]);
			continue;
		}
		/* Get the number of bytes to read */
		if (!bReadBuffer(pFile, pPPS->tWordDocument.ulSB,
				aulBBD, tBBDLen, BIG_BLOCK_SIZE,
				aucTmp, aulSectPage[tIndex], 2)) {
			continue;
		}
		tBytes = 2 + (size_t)usGetWord(0, aucTmp);
		NO_DBG_DEC(tBytes);
		/* Read the bytes */
		aucFpage = xmalloc(tBytes);
		if (!bReadBuffer(pFile, pPPS->tWordDocument.ulSB,
				aulBBD, tBBDLen, BIG_BLOCK_SIZE,
				aucFpage, aulSectPage[tIndex], tBytes)) {
			aucFpage = xfree(aucFpage);
			continue;
		}
		NO_DBG_PRINT_BLOCK(aucFpage, tBytes);
		/* Process the bytes */
		vGetDefaultSection(&tSection);
		vGet8SectionInfo(aucFpage + 2, tBytes - 2, &tSection);
		vAdd2SectionInfoList(&tSection, aulCharPos[tIndex]);
		aucFpage = xfree(aucFpage);
	}
	aulCharPos = xfree(aulCharPos);
	aulSectPage = xfree(aulSectPage);
} /* end of vGet8SepInfo */

/*
 * Build the list with Header/Footer Information for Word 8/9/10/11 files
 */
void
vGet8HdrFtrInfo(FILE *pFile, const pps_type *pTable,
	const ULONG *aulBBD, size_t tBBDLen,
	const ULONG *aulSBD, size_t tSBDLen,
	const UCHAR *aucHeader)
{
	ULONG	*aulCharPos;
	UCHAR	*aucBuffer;
	ULONG	ulHdrFtrOffset, ulBeginHdrFtrInfo;
	size_t	tHdrFtrInfoLen, tIndex, tOffset, tLen;

	fail(pFile == NULL || pTable == NULL || aucHeader == NULL);
	fail(aulBBD == NULL || aulSBD == NULL);

	ulBeginHdrFtrInfo = ulGetLong(0xf2, aucHeader); /* fcPlcfhdd */
	NO_DBG_HEX(ulBeginHdrFtrInfo);
	tHdrFtrInfoLen = (size_t)ulGetLong(0xf6, aucHeader); /* lcbPlcfhdd */
	NO_DBG_DEC(tHdrFtrInfoLen);
	if (tHdrFtrInfoLen < 8) {
		DBG_DEC_C(tHdrFtrInfoLen != 0, tHdrFtrInfoLen);
		return;
	}

	aucBuffer = aucFillInfoBuffer(pFile, pTable,
			aulBBD, tBBDLen, aulSBD, tSBDLen,
			ulBeginHdrFtrInfo, tHdrFtrInfoLen);
	if (aucBuffer == NULL) {
		return;
	}
	NO_DBG_PRINT_BLOCK(aucBuffer, tHdrFtrInfoLen);

	tLen = tHdrFtrInfoLen / 4 - 1;
	DBG_DEC_C(tLen % 12 != 1 && tLen % 12 != 7, tLen);
	/* Save the header/footer offsets */
	aulCharPos = xcalloc(tLen, sizeof(ULONG));
	for (tIndex = 0, tOffset = 0;
	     tIndex < tLen;
	     tIndex++, tOffset += 4) {
		ulHdrFtrOffset = ulGetLong(tOffset, aucBuffer);
		NO_DBG_HEX(ulHdrFtrOffset);
		aulCharPos[tIndex] = ulHdrFtrOffset2CharPos(ulHdrFtrOffset);
		NO_DBG_HEX(aulCharPos[tIndex]);
	}
	vCreat8HdrFtrInfoList(aulCharPos, tLen);
	/* Clean up and leave */
	aulCharPos = xfree(aulCharPos);
	aucBuffer = xfree(aucBuffer);
} /* end of vGet8HdrFtrInfo */

/*
 * Translate the rowinfo to a member of the row_info enumeration
 */
row_info_enum
eGet8RowInfo(int iFodo,
	const UCHAR *aucGrpprl, int iBytes, row_block_type *pRow)
{
	int	iFodoOff, iInfoLen;
	int	iIndex, iSize, iCol;
	int	iPosCurr, iPosPrev;
	USHORT	usTmp;
	BOOL	bFound2416_0, bFound2416_1, bFound2417_0, bFound2417_1;
	BOOL	bFound244b_0, bFound244b_1, bFound244c_0, bFound244c_1;
	BOOL	bFoundd608;

	fail(iFodo < 0 || aucGrpprl == NULL || pRow == NULL);

	iFodoOff = 0;
	bFound2416_0 = FALSE;
	bFound2416_1 = FALSE;
	bFound2417_0 = FALSE;
	bFound2417_1 = FALSE;
	bFound244b_0 = FALSE;
	bFound244b_1 = FALSE;
	bFound244c_0 = FALSE;
	bFound244c_1 = FALSE;
	bFoundd608 = FALSE;
	while (iBytes >= iFodoOff + 2) {
		iInfoLen = 0;
		switch (usGetWord(iFodo + iFodoOff, aucGrpprl)) {
		case 0x2416:	/* fInTable */
			if (odd(ucGetByte(iFodo + iFodoOff + 2, aucGrpprl))) {
				bFound2416_1 = TRUE;
			} else {
				bFound2416_0 = TRUE;
			}
			break;
		case 0x2417:	/* fTtp */
			if (odd(ucGetByte(iFodo + iFodoOff + 2, aucGrpprl))) {
				bFound2417_1 = TRUE;
			} else {
				bFound2417_0 = TRUE;
			}
			break;
		case 0x244b:	/* sub-table fInTable */
			if (odd(ucGetByte(iFodo + iFodoOff + 2, aucGrpprl))) {
				bFound244b_1 = TRUE;
			} else {
				bFound244b_0 = TRUE;
			}
			break;
		case 0x244c:	/* sub-table fTtp */
			if (odd(ucGetByte(iFodo + iFodoOff + 2, aucGrpprl))) {
				bFound244c_1 = TRUE;
			} else {
				bFound244c_0 = TRUE;
			}
			break;
		case 0x6424:	/* brcTop */
			usTmp = usGetWord(iFodo + iFodoOff + 2, aucGrpprl);
			usTmp &= 0xff00;
			NO_DBG_DEC(usTmp >> 8);
			if (usTmp == 0) {
				pRow->ucBorderInfo &= ~TABLE_BORDER_TOP;
			} else {
				pRow->ucBorderInfo |= TABLE_BORDER_TOP;
			}
			break;
		case 0x6425:	/* brcLeft */
			usTmp = usGetWord(iFodo + iFodoOff + 2, aucGrpprl);
			usTmp &= 0xff00;
			NO_DBG_DEC(usTmp >> 8);
			if (usTmp == 0) {
				pRow->ucBorderInfo &= ~TABLE_BORDER_LEFT;
			} else {
				pRow->ucBorderInfo |= TABLE_BORDER_LEFT;
			}
			break;
		case 0x6426:	/* brcBottom */
			usTmp = usGetWord(iFodo + iFodoOff + 2, aucGrpprl);
			usTmp &= 0xff00;
			NO_DBG_DEC(usTmp >> 8);
			if (usTmp == 0) {
				pRow->ucBorderInfo &= ~TABLE_BORDER_BOTTOM;
			} else {
				pRow->ucBorderInfo |= TABLE_BORDER_BOTTOM;
			}
			break;
		case 0x6427:	/* brcRight */
			usTmp = usGetWord(iFodo + iFodoOff + 2, aucGrpprl);
			usTmp &= 0xff00;
			NO_DBG_DEC(usTmp >> 8);
			if (usTmp == 0) {
				pRow->ucBorderInfo &= ~TABLE_BORDER_RIGHT;
			} else {
				pRow->ucBorderInfo |= TABLE_BORDER_RIGHT;
			}
			break;
		case 0xd606:	/* cDefTable10 */
			DBG_MSG("0xd606: sprmTDefTable10");
			iSize = (int)usGetWord(iFodo + iFodoOff + 2, aucGrpprl);
			DBG_DEC(iSize);
			break;
		case 0xd608:	/* cDefTable */
			iSize = (int)usGetWord(iFodo + iFodoOff + 2, aucGrpprl);
			if (iSize < 6 || iBytes < iFodoOff + 8) {
				DBG_DEC(iSize);
				DBG_DEC(iFodoOff);
				iInfoLen = 2;
				break;
			}
			iCol = (int)ucGetByte(iFodo + iFodoOff + 4, aucGrpprl);
			if (iCol < 1 ||
			    iBytes < iFodoOff + 4 + (iCol + 1) * 2) {
				DBG_DEC(iCol);
				DBG_DEC(iFodoOff);
				iInfoLen = 2;
				break;
			}
			if (iCol >= (int)elementsof(pRow->asColumnWidth)) {
				DBG_DEC(iCol);
				werr(1, "The number of columns is corrupt");
			}
			pRow->ucNumberOfColumns = (UCHAR)iCol;
			iPosPrev = (int)(short)usGetWord(
					iFodo + iFodoOff + 5,
					aucGrpprl);
			for (iIndex = 0; iIndex < iCol; iIndex++) {
				iPosCurr = (int)(short)usGetWord(
					iFodo + iFodoOff + 7 + iIndex * 2,
					aucGrpprl);
				pRow->asColumnWidth[iIndex] =
						(short)(iPosCurr - iPosPrev);
				iPosPrev = iPosCurr;
			}
			bFoundd608 = TRUE;
			break;
		default:
			break;