docdecoder.cpp

此文件为Word 格式文件.DOC转换为TXT文件的源代码。

#include "StdAfx.h"
#include ".\docdecoder.h"


CDocDecoder::CDocDecoder(CPublicResource *ps):
		CDecoderBase(ps)
{
	reset();
}

CDocDecoder::~CDocDecoder(void)
{
}

void
CDocDecoder::vDestroyTextBlockList(void)
{
	list_mem_type	*apAnchor[5];
	list_mem_type	*pCurr, *pNext;
	int	iIndex;

	apAnchor[0] = pTextAnchor;
	apAnchor[1] = pFootAnchor;
	apAnchor[2] = pUnused1Anchor;
	apAnchor[3] = pEndAnchor;
	apAnchor[4] = pUnused2Anchor;

	for (iIndex = 0; iIndex < 5; iIndex++) {
		pCurr = apAnchor[iIndex];
		while (pCurr != NULL) {
			pNext = pCurr->pNext;
			pCurr = (list_mem_type *)xfree(pCurr);
			pCurr = pNext;
		}
	}
	/* Show that there are no lists any more */
	pTextAnchor = NULL;
	pFootAnchor = NULL;
	pUnused1Anchor = NULL;
	pEndAnchor = NULL;
	pUnused2Anchor = NULL;
	/* Reset all the controle variables */
	pBlockLast = NULL;
	pTextBlockCurrent = NULL;
	pFootBlockCurrent = NULL;
} /* end of vDestroyTextBlockList */


bool
CDocDecoder::bAdd2TextBlockList(text_block_type *pTextBlock)
{
	list_mem_type	*pListMember;

	if (pTextBlock->iFileOffset < 0 ||
		pTextBlock->iTextOffset < 0 ||
		pTextBlock->iLength <= 0) {
//			werr(0, "Software (textblock) error");
			return false;
		}
		/* Check for continuous blocks of the same character size */
		if (pBlockLast != NULL &&
			pBlockLast->tInfo.iFileOffset +
			pBlockLast->tInfo.iLength == pTextBlock->iFileOffset &&
			pBlockLast->tInfo.iTextOffset +
			pBlockLast->tInfo.iLength == pTextBlock->iTextOffset &&
			pBlockLast->tInfo.bUsesUnicode == pTextBlock->bUsesUnicode) {
				/* These are continous blocks */
				pBlockLast->tInfo.iLength += pTextBlock->iLength;
				return TRUE;
			}
			/* Make a new block */
			pListMember = (list_mem_type *)xmalloc(sizeof(list_mem_type));
			/* Add the block to the list */
			pListMember->tInfo = *pTextBlock;
			pListMember->pNext = NULL;
			if (pTextAnchor == NULL) {
				pTextAnchor = pListMember;
			} else {
				pBlockLast->pNext = pListMember;
			}
			pBlockLast = pListMember;
			return TRUE;
} /* end of bAdd2TextBlockList */



void
CDocDecoder::vSplitBlockList(int iTextLen, int iFootnoteLen,
				int iUnused1Len, int iEndnoteLen, int iUnused2Len,
				bool bMustExtend)
{
	list_mem_type	*apAnchors[5];
	list_mem_type	*pGarbageAnchor, *pCurr, *pNext;
	int		iIndex, iCharsToGo, iBytesTooFar;

	/* Text block list */
	if (iTextLen > 0) {
		iCharsToGo = iTextLen;
		iBytesTooFar = -1;
		for (pCurr = pTextAnchor;
			pCurr != NULL;
			pCurr = pCurr->pNext) {
				if (pCurr->tInfo.bUsesUnicode) {
					iCharsToGo -= pCurr->tInfo.iLength / 2;
					if (iCharsToGo < 0) {
						iBytesTooFar = -2 * iCharsToGo;
					}
				} else {
					iCharsToGo -= pCurr->tInfo.iLength;
					if (iCharsToGo < 0) {
						iBytesTooFar = -iCharsToGo;
					}
				}
				if (iCharsToGo <= 0) {
					break;
				}
			}
	}
	/* Split the list */
	if (iTextLen <= 0) {
		/* Empty text blocks list */
		pFootAnchor = pTextAnchor;
		pTextAnchor = NULL;
	} else if (pCurr == NULL) {
		/* No footnote blocks */
		pFootAnchor = NULL;
	} else if (iCharsToGo == 0) {
		/* Move the integral number of footnote blocks */
		pFootAnchor = pCurr->pNext;
		pCurr->pNext = NULL;
	} else {
		/* Split the part-text block, part-footnote block */
		pFootAnchor = (list_mem_type *)xmalloc(sizeof(list_mem_type));
		pFootAnchor->tInfo.iFileOffset =
			pCurr->tInfo.iFileOffset +
			pCurr->tInfo.iLength -
			iBytesTooFar;
		pFootAnchor->tInfo.iTextOffset =
			pCurr->tInfo.iTextOffset +
			pCurr->tInfo.iLength -
			iBytesTooFar;
		pFootAnchor->tInfo.iLength = iBytesTooFar;
		pCurr->tInfo.iLength -= iBytesTooFar;
		pFootAnchor->tInfo.bUsesUnicode = pCurr->tInfo.bUsesUnicode;
		/* Move the integral number of footnote blocks */
		pFootAnchor->pNext = pCurr->pNext;
		pCurr->pNext = NULL;
	}
	/* Footnote block list */
	if (iFootnoteLen > 0) {
		iCharsToGo = iFootnoteLen;
		iBytesTooFar = -1;
		for (pCurr = pFootAnchor;
			pCurr != NULL;
			pCurr = pCurr->pNext) {
				if (pCurr->tInfo.bUsesUnicode) {
					iCharsToGo -= pCurr->tInfo.iLength / 2;
					if (iCharsToGo < 0) {
						iBytesTooFar = -2 * iCharsToGo;
					}
				} else {
					iCharsToGo -= pCurr->tInfo.iLength;
					if (iCharsToGo < 0) {
						iBytesTooFar = -iCharsToGo;
					}
				}
				if (iCharsToGo <= 0) {
					break;
				}
			}
	}
	/* Split the list */
	if (iFootnoteLen <= 0) {
		/* Empty footnote list */
		pUnused1Anchor = pFootAnchor;
		pFootAnchor = NULL;
	} else if (pCurr == NULL) {
		/* No unused1 blocks */
		pUnused1Anchor = NULL;
	} else if (iCharsToGo == 0) {
		/* Move the integral number of unused1-list blocks */
		pUnused1Anchor = pCurr->pNext;
		pCurr->pNext = NULL;
	} else {
		/* Split the part-footnote block, part-unused1 block */
		pUnused1Anchor = (list_mem_type *)xmalloc(sizeof(list_mem_type));
		pUnused1Anchor->tInfo.iFileOffset =
			pCurr->tInfo.iFileOffset +
			pCurr->tInfo.iLength -
			iBytesTooFar;
		pUnused1Anchor->tInfo.iTextOffset =
			pCurr->tInfo.iTextOffset +
			pCurr->tInfo.iLength -
			iBytesTooFar;
		pUnused1Anchor->tInfo.iLength = iBytesTooFar;
		pCurr->tInfo.iLength -= iBytesTooFar;
		pUnused1Anchor->tInfo.bUsesUnicode =
			pCurr->tInfo.bUsesUnicode;
		/* Move the integral number of unused1 blocks */
		pUnused1Anchor->pNext = pCurr->pNext;
		pCurr->pNext = NULL;
	}
	/* Unused1 block list */
	if (iUnused1Len > 0) {
		iCharsToGo = iUnused1Len;
		iBytesTooFar = -1;
		for (pCurr = pUnused1Anchor;
			pCurr != NULL;
			pCurr = pCurr->pNext) {
				if (pCurr->tInfo.bUsesUnicode) {
					iCharsToGo -= pCurr->tInfo.iLength / 2;
					if (iCharsToGo < 0) {
						iBytesTooFar = -2 * iCharsToGo;
					}
				} else {
					iCharsToGo -= pCurr->tInfo.iLength;
					if (iCharsToGo < 0) {
						iBytesTooFar = -iCharsToGo;
					}
				}
				if (iCharsToGo <= 0) {
					break;
				}
			}
	}
	/* Split the list */
	if (iUnused1Len <= 0) {
		/* Empty unused1 list */
		pEndAnchor = pUnused1Anchor;
		pUnused1Anchor = NULL;
	} else if (pCurr == NULL) {
		/* No endnote blocks */
		pEndAnchor = NULL;
	} else if (iCharsToGo == 0) {
		/* Move the intergral number of endnote blocks */
		pEndAnchor = pCurr->pNext;
		pCurr->pNext = NULL;
	} else {
		/* Split the part-unused1-list block, part-endnote block */
		pEndAnchor = (list_mem_type *)xmalloc(sizeof(list_mem_type));
		pEndAnchor->tInfo.iFileOffset =
			pCurr->tInfo.iFileOffset +
			pCurr->tInfo.iLength -
			iBytesTooFar;
		pEndAnchor->tInfo.iTextOffset =
			pCurr->tInfo.iTextOffset +
			pCurr->tInfo.iLength -
			iBytesTooFar;
		pEndAnchor->tInfo.iLength = iBytesTooFar;
		pCurr->tInfo.iLength -= iBytesTooFar;
		pEndAnchor->tInfo.bUsesUnicode = pCurr->tInfo.bUsesUnicode;
		/* Move the integral number of endnote blocks */
		pEndAnchor->pNext = pCurr->pNext;
		pCurr->pNext = NULL;
	}
	/* Endnote block list */
	if (iEndnoteLen > 0) {
		iCharsToGo = iEndnoteLen;
		iBytesTooFar = -1;
		for (pCurr = pEndAnchor;
			pCurr != NULL;
			pCurr = pCurr->pNext) {
				if (pCurr->tInfo.bUsesUnicode) {
					iCharsToGo -= pCurr->tInfo.iLength / 2;
					if (iCharsToGo <= 0) {
						iBytesTooFar = -2 * iCharsToGo;
					}
				} else {
					iCharsToGo -= pCurr->tInfo.iLength;
					if (iCharsToGo <= 0) {
						iBytesTooFar = -iCharsToGo;
					}
				}
				if (iCharsToGo <= 0) {
					break;
				}
			}
	}
	/* Split the list */
	if (iEndnoteLen <= 0) {
		/* Empty endnote list */
		pUnused2Anchor = pEndAnchor;
		pEndAnchor = NULL;
	} else if (pCurr == NULL) {
		/* No unused2 blocks */
		pUnused2Anchor = NULL;
	} else if (iCharsToGo == 0) {
		/* Move the intergral number of unused2 blocks */
		pUnused2Anchor = pCurr->pNext;
		pCurr->pNext = NULL;
	} else {
		/* Split the part-endnote block, part-unused2 block */
		pUnused2Anchor = (list_mem_type *)xmalloc(sizeof(list_mem_type));
		pUnused2Anchor->tInfo.iFileOffset =
			pCurr->tInfo.iFileOffset +
			pCurr->tInfo.iLength -
			iBytesTooFar;
		pUnused2Anchor->tInfo.iTextOffset =
			pCurr->tInfo.iTextOffset +
			pCurr->tInfo.iLength -
			iBytesTooFar;
		pUnused2Anchor->tInfo.iLength = iBytesTooFar;
		pCurr->tInfo.iLength -= iBytesTooFar;
		pUnused2Anchor->tInfo.bUsesUnicode =
			pCurr->tInfo.bUsesUnicode;
		/* Move the integral number of unused2 blocks */
		pUnused2Anchor->pNext = pCurr->pNext;
		pCurr->pNext = NULL;
	}
	/* Unused2 block list */
	if (iUnused2Len > 0) {
		iCharsToGo = iUnused2Len;
		iBytesTooFar = -1;
		for (pCurr = pUnused2Anchor;
			pCurr != NULL;
			pCurr = pCurr->pNext) {
				if (pCurr->tInfo.bUsesUnicode) {
					iCharsToGo -= pCurr->tInfo.iLength / 2;
					if (iCharsToGo < 0) {
						iBytesTooFar = -2 * iCharsToGo;
					}
				} else {
					iCharsToGo -= pCurr->tInfo.iLength;
					if (iCharsToGo < 0) {
						iBytesTooFar = -iCharsToGo;
					}
				}
				if (iCharsToGo <= 0) {
					break;
				}
			}
	}
	if (iUnused2Len <= 0) {
		/* Empty unused2 list */
		pGarbageAnchor = pUnused2Anchor;
		pUnused2Anchor = NULL;
	} else if (pCurr == NULL) {
		/* No garbage block list */
		pGarbageAnchor = NULL;
	} else if (iCharsToGo == 0) {
		/* Move the intergral number of garbage blocks */
		pGarbageAnchor = pCurr->pNext;
	} else {
		/* Reduce the part-endnote block */
		pCurr->tInfo.iLength -= iBytesTooFar;
		/* Move the integral number of garbage blocks */
		pGarbageAnchor = pCurr->pNext;
		pCurr->pNext = NULL;
	}
	/* Free the garbage block list, this should never be needed */
	pCurr = pGarbageAnchor;
	while (pCurr != NULL) {
		pNext = pCurr->pNext;
		pCurr = (list_mem_type *)xfree(pCurr);
		pCurr = pNext;
	}


	if (!bMustExtend) {
		return;
	}
	/*
	* All blocks (except the last one) must have a length that
	* is a multiple of the Big Block Size
	*/

	apAnchors[0] = pTextAnchor;
	apAnchors[1] = pFootAnchor;
	apAnchors[2] = pUnused1Anchor;
	apAnchors[3] = pEndAnchor;
	apAnchors[4] = pUnused2Anchor;

	for (iIndex = 0; iIndex < 5; iIndex++) {
		for (pCurr = apAnchors[iIndex];
			pCurr != NULL;
			pCurr = pCurr->pNext) {
				if (pCurr->pNext != NULL &&
					pCurr->tInfo.iLength % BIG_BLOCK_SIZE != 0) {
						pCurr->tInfo.iLength /= BIG_BLOCK_SIZE;
						pCurr->tInfo.iLength++;
						pCurr->tInfo.iLength *= BIG_BLOCK_SIZE;
					}
			}
	}
} /* end of vSplitBlockList */


unsigned int
CDocDecoder::uiGetDocumentLength(void)
{
	list_mem_type	*apAnchors[3];
	list_mem_type	*pCurr;
	int	iTotal, iIndex;

	apAnchors[0] = pTextAnchor;
	apAnchors[1] = pFootAnchor;
	apAnchors[2] = pEndAnchor;

	iTotal = 0;
	for (iIndex = 0; iIndex < 3; iIndex++) {
		for (pCurr = apAnchors[iIndex];
			pCurr != NULL;
			pCurr = pCurr->pNext) {
				if (pCurr->tInfo.bUsesUnicode) {
					iTotal += pCurr->tInfo.iLength / 2;
				} else {
					iTotal += pCurr->tInfo.iLength;
				}
			}
	}
	return iTotal;
} /* end of uiGetDocumentLength */


int
CDocDecoder::iNextTextByte(unsigned char *pFile, int *piOffset)
{
	static int	iByteNext = 0;
	static int	iBlockOffset = 0;
	int	iReadLen, iReadOff;

	if (pTextBlockCurrent == NULL ||
		iByteNext >= sizeof(aucBlock) ||
		iBlockOffset + iByteNext >= pTextBlockCurrent->tInfo.iLength) {
			if (pTextBlockCurrent == NULL) {
				/* First block, first part */
				pTextBlockCurrent = pTextAnchor;
				iBlockOffset = 0;
			} else if ((iBlockOffset + (signed)sizeof(aucBlock)) <
				pTextBlockCurrent->tInfo.iLength) {
					/* Same block, next part */
					iBlockOffset += sizeof(aucBlock);
				} else {
					/* Next block, first part */
					pTextBlockCurrent = pTextBlockCurrent->pNext;
					iBlockOffset = 0;
				}
				if (pTextBlockCurrent == NULL) {
					/* Past the last part of the last block */
					return EOF;
				}
				iReadLen = pTextBlockCurrent->tInfo.iLength - iBlockOffset;
				if (iReadLen > sizeof(aucBlock)) {
					iReadLen = sizeof(aucBlock);
				}
				iReadOff = pTextBlockCurrent->tInfo.iFileOffset +
					iBlockOffset;
				if (!bReadBytes(aucBlock, iReadLen, iReadOff, pFile)) {
					return EOF;
				}
				iByteNext = 0;
		}
		if (piOffset != NULL) {
			*piOffset = pTextBlockCurrent->tInfo.iFileOffset +
				iBlockOffset +
				iByteNext;
		}
		return aucBlock[iByteNext++];
} /* end of iNextTextByte */


int
CDocDecoder::iNextFootByte(unsigned char *pFile, int *piOffset)
{
	static int	iByteNext = 0;
	static int	iBlockOffset = 0;
	int	iReadLen, iReadOff;

	if (pFootBlockCurrent == NULL ||
		iByteNext >= sizeof(aucBlock) ||
		iBlockOffset + iByteNext >= pFootBlockCurrent->tInfo.iLength) {
			if (pFootBlockCurrent == NULL) {
				/* First block, first part */
				pFootBlockCurrent = pFootAnchor;
				iBlockOffset = 0;
			} else if (iBlockOffset + (signed)sizeof(aucBlock) <
				pFootBlockCurrent->tInfo.iLength) {
					/* Same block, next part */
					iBlockOffset += sizeof(aucBlock);
				} else {
					/* Next block, first part */
					pFootBlockCurrent = pFootBlockCurrent->pNext;
					iBlockOffset = 0;
				}
				if (pFootBlockCurrent == NULL) {
					/* Past the last part of the last block */
					return EOF;
				}
				iReadLen = pFootBlockCurrent->tInfo.iLength - iBlockOffset;
				if (iReadLen > sizeof(aucBlock)) {
					iReadLen = sizeof(aucBlock);
				}
				iReadOff = pFootBlockCurrent->tInfo.iFileOffset +
					iBlockOffset;
				if (!bReadBytes(aucBlock, iReadLen, iReadOff, pFile)) {
					return EOF;
				}
				iByteNext = 0;
		}
		if (piOffset != NULL) {