word2text.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

static ULONG
ulGetChar(FILE *pFile, list_id_enum eListID)
{
	const font_block_type	*pCurr;
	ULONG		ulChar, ulFileOffset, ulCharPos;
	row_info_enum	eRowInfo;
	USHORT		usChar, usPropMod;
	BOOL		bSkip;

	fail(pFile == NULL);

	pCurr = pFontInfo;
	bSkip = FALSE;
	for (;;) {
		usChar = usNextChar(pFile, eListID,
				&ulFileOffset, &ulCharPos, &usPropMod);
		if (usChar == (USHORT)EOF) {
			return (ULONG)EOF;
		}

		vUpdateCounters();

		eRowInfo = ePropMod2RowInfo(usPropMod, iWordVersion);
		if (!bStartRow) {
#if 0
			bStartRow = eRowInfo == found_a_cell ||
				(pRowInfo != NULL &&
				 ulFileOffset == pRowInfo->ulFileOffsetStart &&
				 eRowInfo != found_not_a_cell);
#else
			bStartRow = pRowInfo != NULL &&
				ulFileOffset == pRowInfo->ulFileOffsetStart;
#endif
			NO_DBG_HEX_C(bStartRow, pRowInfo->ulFileOffsetStart);
		}
		if (!bEndRowNorm) {
#if 0
			bEndRow = eRowInfo == found_end_of_row ||
				(pRowInfo != NULL &&
				 ulFileOffset == pRowInfo->ulFileOffsetEnd &&
				 eRowInfo != found_not_end_of_row);
#else
			bEndRowNorm = pRowInfo != NULL &&
				ulFileOffset == pRowInfo->ulFileOffsetEnd;
#endif
			NO_DBG_HEX_C(bEndRowNorm, pRowInfo->ulFileOffsetEnd);
		}
		if (!bEndRowFast) {
			bEndRowFast = eRowInfo == found_end_of_row;
			NO_DBG_HEX_C(bEndRowFast, pRowInfo->ulFileOffsetEnd);
		}

		if (!bStartStyle) {
			bStartStyle = pStyleInfo != NULL &&
				ulFileOffset == pStyleInfo->ulFileOffset;
			NO_DBG_HEX_C(bStartStyle, ulFileOffset);
		}
		if (pCurr != NULL && ulFileOffset == pCurr->ulFileOffset) {
			bStartFont = TRUE;
			NO_DBG_HEX(ulFileOffset);
			pFontInfo = pCurr;
			pCurr = pGetNextFontInfoListItem(pCurr);
		}

		/* Skip embedded characters */
		if (usChar == START_EMBEDDED) {
			bSkip = TRUE;
			continue;
		}
		if (usChar == END_IGNORE || usChar == END_EMBEDDED) {
			bSkip = FALSE;
			continue;
		}
		if (bSkip) {
			continue;
		}
		ulChar = ulTranslateCharacters(usChar,
					ulFileOffset,
					iWordVersion,
					tOptions.eConversionType,
					tOptions.eEncoding,
					bOldMacFile);
		if (ulChar == IGNORE_CHARACTER) {
			continue;
		}
		if (ulChar == PICTURE) {
			ulFileOffsetImage = ulGetPictInfoListItem(ulFileOffset);
		} else {
			ulFileOffsetImage = FC_INVALID;
		}
		if (ulChar == PAR_END) {
			/* End of paragraph seen, prepare for the next */
			vFillStyleFromStylesheet(usIstdNext, &tStyleNext);
			vCorrectStyleValues(&tStyleNext);
			bStartStyleNext = TRUE;
			vFillFontFromStylesheet(usIstdNext, &tFontNext);
			vCorrectFontValues(&tFontNext);
			bStartFontNext = TRUE;
		}
		if (ulChar == PAGE_BREAK) {
			/* Might be the start of a new section */
			pSectionNext = pGetSectionInfo(pSection, ulCharPos);
		}
		return ulChar;
	}
} /* end of ulGetChar */

/*
 * lGetWidthMax - get the maximum line width from the paragraph break value
 *
 * Returns the maximum line width in millipoints
 */
static long
lGetWidthMax(int iParagraphBreak)
{
	fail(iParagraphBreak < 0);

	if (iParagraphBreak == 0) {
		return LONG_MAX;
	}
	if (iParagraphBreak < MIN_SCREEN_WIDTH) {
		return lChar2MilliPoints(MIN_SCREEN_WIDTH);
	}
	if (iParagraphBreak > MAX_SCREEN_WIDTH) {
		return lChar2MilliPoints(MAX_SCREEN_WIDTH);
	}
	return lChar2MilliPoints(iParagraphBreak);
} /* end of lGetWidthMax */

/*
 * bWordDecryptor - turn Word to something more useful
 *
 * returns TRUE when succesful, otherwise FALSE
 */
BOOL
bWordDecryptor(FILE *pFile, long lFilesize, diagram_type *pDiag)
{
	imagedata_type	tImage;
	const style_block_type	*pStyleTmp;
	const font_block_type	*pFontTmp;
	const char	*szListChar;
	output_type	*pAnchor, *pOutput, *pLeftOver;
	ULONG	ulChar;
	long	lBeforeIndentation, lAfterIndentation;
	long	lLeftIndentation, lLeftIndentation1, lRightIndentation;
	long	lWidthCurr, lWidthMax, lDefaultTabWidth, lHalfSpaceWidth, lTmp;
	list_id_enum 	eListID;
	image_info_enum	eRes;
	UINT	uiFootnoteNumber, uiEndnoteNumber, uiTmp;
	int	iListSeqNumber;
	BOOL	bWasTableRow, bTableFontClosed, bWasEndOfParagraph;
	BOOL	bInList, bWasInList, bNoMarks, bFirstLine;
	BOOL	bAllCapitals, bHiddenText, bMarkDelText, bSuccess;
	USHORT	usListNumber;
	USHORT	usFontStyle, usFontStyleMinimal, usFontSize, usTmp;
	UCHAR	ucFontNumber, ucFontColor;
	UCHAR	ucNFC, ucAlignment;

	fail(pFile == NULL || lFilesize <= 0 || pDiag == NULL);

	TRACE_MSG("bWordDecryptor");

	iWordVersion = iInitDocument(pFile, lFilesize);
	if (iWordVersion < 0) {
		DBG_DEC(iWordVersion);
		return FALSE;
	}

	vGetOptions(&tOptions);
	bOldMacFile = bIsOldMacFile();
	vPrepareHdrFtrText(pFile);
	vPrepareFootnoteText(pFile);

	vPrologue2(pDiag, iWordVersion);

	/* Initialisation */
#if defined(__riscos)
	ulCharCounter = 0;
	iCurrPct = 0;
	iPrevPct = -1;
	ulDocumentLength = ulGetDocumentLength();
#endif /* __riscos */
	pSection = pGetSectionInfo(NULL, 0);
	pSectionNext = pSection;
	lDefaultTabWidth = lGetDefaultTabWidth();
	DBG_DEC_C(lDefaultTabWidth != 36000, lDefaultTabWidth);
	pRowInfo = pGetNextRowInfoListItem();
	DBG_HEX_C(pRowInfo != NULL, pRowInfo->ulFileOffsetStart);
	DBG_HEX_C(pRowInfo != NULL, pRowInfo->ulFileOffsetEnd);
	DBG_MSG_C(pRowInfo == NULL, "No rows at all");
	bStartRow = FALSE;
	bEndRowNorm = FALSE;
	bEndRowFast = FALSE;
	bIsTableRow = FALSE;
	bWasTableRow = FALSE;
	vResetStyles();
	pStyleInfo = pGetNextTextStyle(NULL);
	bStartStyle = FALSE;
	bInList = FALSE;
	bWasInList = FALSE;
	iListSeqNumber = 0;
	usIstdNext = ISTD_NORMAL;
	pAnchor = NULL;
	pFontInfo = pGetNextFontInfoListItem(NULL);
	DBG_HEX_C(pFontInfo != NULL, pFontInfo->ulFileOffset);
	DBG_MSG_C(pFontInfo == NULL, "No fonts at all");
	bStartFont = FALSE;
	ucFontNumber = 0;
	usFontStyleMinimal = FONT_REGULAR;
	usFontStyle = FONT_REGULAR;
	usFontSize = DEFAULT_FONT_SIZE;
	ucFontColor = FONT_COLOR_DEFAULT;
	pAnchor = pStartNewOutput(pAnchor, NULL);
	pOutput = pAnchor;
	pOutput->ucFontColor = ucFontColor;
	pOutput->usFontStyle = usFontStyle;
	pOutput->tFontRef = tOpenFont(ucFontNumber, usFontStyle, usFontSize);
	pOutput->usFontSize = usFontSize;
	bTableFontClosed = TRUE;
	lBeforeIndentation = 0;
	lAfterIndentation = 0;
	lLeftIndentation = 0;
	lLeftIndentation1 = 0;
	lRightIndentation = 0;
	bWasEndOfParagraph = TRUE;
	bNoMarks = TRUE;
	bFirstLine = TRUE;
	ucNFC = LIST_BULLETS;
	if (pStyleInfo != NULL) {
		szListChar = pStyleInfo->szListChar;
		pStyleTmp = pStyleInfo;
	} else {
		if (tStyleNext.szListChar[0] == '\0') {
			vGetBulletValue(tOptions.eConversionType,
				tOptions.eEncoding, tStyleNext.szListChar, 4);
		}
		szListChar = tStyleNext.szListChar;
		pStyleTmp = &tStyleNext;
	}
	usListNumber = 0;
	ucAlignment = ALIGNMENT_LEFT;
	bAllCapitals = FALSE;
	bHiddenText = FALSE;
	bMarkDelText = FALSE;
	lWidthMax = lGetWidthMax(tOptions.iParagraphBreak);
	NO_DBG_DEC(lWidthMax);

	Hourglass_On();

	uiFootnoteNumber = 0;
	uiEndnoteNumber = 0;
	eListID = text_list;
	for(;;) {
		ulChar = ulGetChar(pFile, eListID);
		if (ulChar == (ULONG)EOF) {
			if (bOutputContainsText(pAnchor)) {
				OUTPUT_LINE();
			} else {
				RESET_LINE();
			}
			switch (eListID) {
			case text_list:
				if (tOptions.eConversionType !=
							conversion_xml) {
					eListID = footnote_list;
					if (uiFootnoteNumber != 0) {
						vPutSeparatorLine(pAnchor);
						OUTPUT_LINE();
						uiFootnoteNumber = 0;
					}
					break;
				}
				/* No break or return */
			case footnote_list:
				eListID = endnote_list;
				if (uiEndnoteNumber != 0) {
					vPutSeparatorLine(pAnchor);
					OUTPUT_LINE();
					uiEndnoteNumber = 0;
				}
				break;
			case endnote_list:
				eListID = textbox_list;
				if (bExistsTextBox()) {
					vPutSeparatorLine(pAnchor);
					OUTPUT_LINE();
				}
				break;
			case textbox_list:
				eListID = hdrtextbox_list;
				if (bExistsHdrTextBox()) {
					vPutSeparatorLine(pAnchor);
					OUTPUT_LINE();
				}
				break;
			case hdrtextbox_list:
			default:
				eListID = end_of_lists;
				break;
			}
			if (eListID == end_of_lists) {
				break;
			}
			continue;
		}

		if (ulChar == UNKNOWN_NOTE_CHAR) {
			switch (eListID) {
			case footnote_list:
				ulChar = FOOTNOTE_CHAR;
				break;
			case endnote_list:
				ulChar = ENDNOTE_CHAR;
				break;
			default:
				break;
			}
		}

		if (bStartRow) {
			/* Begin of a tablerow found */
			if (bOutputContainsText(pAnchor)) {
				OUTPUT_LINE();
			} else {
				RESET_LINE();
			}
			fail(pAnchor != pOutput);
			if (bTableFontClosed) {
				/* Start special table font */
				vCloseFont();
				/*
				 * Compensate for the fact that Word uses
				 * proportional fonts for its tables and we
				 * only one fixed-width font
				 */
				uiTmp = ((UINT)usFontSize * 5 + 3) / 6;
				if (uiTmp < MIN_TABLEFONT_SIZE) {
					uiTmp = MIN_TABLEFONT_SIZE;
				} else if (uiTmp > MAX_TABLEFONT_SIZE) {
					uiTmp = MAX_TABLEFONT_SIZE;
				}
				pOutput->usFontSize = (USHORT)uiTmp;
				pOutput->tFontRef =
					tOpenTableFont(pOutput->usFontSize);
				pOutput->usFontStyle = FONT_REGULAR;
				pOutput->ucFontColor = FONT_COLOR_BLACK;
				bTableFontClosed = FALSE;
			}
			bIsTableRow = TRUE;
			bStartRow = FALSE;
		}

		if (bWasTableRow &&
		    !bIsTableRow &&
		    ulChar != PAR_END &&
		    ulChar != HARD_RETURN &&
		    ulChar != PAGE_BREAK &&
		    ulChar != COLUMN_FEED) {
			/*
			 * The end of a table should be followed by an
			 * empty line, like the end of a paragraph
			 */
			OUTPUT_LINE();
			vEndOfParagraph(pDiag,
					pOutput->tFontRef,
					pOutput->usFontSize,
					(long)pOutput->usFontSize * 600);
		}

		switch (ulChar) {
		case PAGE_BREAK:
		case COLUMN_FEED:
			if (bIsTableRow) {
				/* Ignore when in a table */
				break;
			}
			if (bOutputContainsText(pAnchor)) {
				OUTPUT_LINE();
			} else {
				RESET_LINE();
			}
			if (ulChar == PAGE_BREAK) {
				vEndOfPage(pDiag, lAfterIndentation,
						pSection != pSectionNext);
			} else {
				vEndOfParagraph(pDiag,
					pOutput->tFontRef,
					pOutput->usFontSize,
					lAfterIndentation);
			}
			break;
		default:
			break;
		}

		if (bStartFont || (bStartFontNext && ulChar != PAR_END)) {
			/* Begin of a font found */
			if (bStartFont) {
				/* bStartFont takes priority */
				fail(pFontInfo == NULL);
				pFontTmp = pFontInfo;
			} else {
				pFontTmp = &tFontNext;
			}
			bAllCapitals = bIsCapitals(pFontTmp->usFontStyle);
			bHiddenText = bIsHidden(pFontTmp->usFontStyle);
			bMarkDelText = bIsMarkDel(pFontTmp->usFontStyle);
			usTmp = pFontTmp->usFontStyle &
				(FONT_BOLD|FONT_ITALIC|FONT_UNDERLINE|
				 FONT_STRIKE|FONT_MARKDEL|
				 FONT_SUPERSCRIPT|FONT_SUBSCRIPT);
			if (!bIsTableRow &&
			    (usFontSize != pFontTmp->usFontSize ||
			     ucFontNumber != pFontTmp->ucFontNumber ||
			     usFontStyleMinimal != usTmp ||
			     ucFontColor != pFontTmp->ucFontColor)) {
				pOutput = pStartNextOutput(pOutput);
				vCloseFont();
				pOutput->ucFontColor = pFontTmp->ucFontColor;
				pOutput->usFontStyle = pFontTmp->usFontStyle;
				pOutput->usFontSize = pFontTmp->usFontSize;
				pOutput->tFontRef = tOpenFont(
						pFontTmp->ucFontNumber,
						pFontTmp->usFontStyle,
						pFontTmp->usFontSize);
				fail(!bCheckDoubleLinkedList(pAnchor));
			}
			ucFontNumber = pFontTmp->ucFontNumber;
			usFontSize = pFontTmp->usFontSize;
			ucFontColor = pFontTmp->ucFontColor;
			usFontStyle = pFontTmp->usFontStyle;
			usFontStyleMinimal = usTmp;
			if (bStartFont) {
				/* Get the next font info */
				pFontInfo = pGetNextFontInfoListItem(pFontInfo);
				NO_DBG_HEX_C(pFontInfo != NULL,
						pFontInfo->ulFileOffset);
				DBG_MSG_C(pFontInfo == NULL, "No more fonts");
			}
			bStartFont = FALSE;
			bStartFontNext = FALSE;
		}

		if (bStartStyle || (bStartStyleNext && ulChar != PAR_END)) {
			bFirstLine = TRUE;
			/* Begin of a style found */
			if (bStartStyle) {
				/* bStartStyle takes priority */
				fail(pStyleInfo == NULL);
				pStyleTmp = pStyleInfo;
			} else {
				pStyleTmp = &tStyleNext;
			}
			if (!bIsTableRow) {
				vStoreStyle(pDiag, pOutput, pStyleTmp);
			}
			usIstdNext = pStyleTmp->usIstdNext;
			lBeforeIndentation =
				lTwips2MilliPoints(pStyleTmp->usBeforeIndent);
			lAfterIndentation =
				lTwips2MilliPoints(pStyleTmp->usAfterIndent);
			lLeftIndentation =
				lTwips2MilliPoints(pStyleTmp->sLeftIndent);
			lLeftIndentation1 =
				lTwips2MilliPoints(pStyleTmp->sLeftIndent1);
			lRightIndentation =
				lTwips2MilliPoints(pStyleTmp->sRightIndent);
			bInList = bStyleImpliesList(pStyleTmp, iWordVersion);
			bNoMarks = !bInList || pStyleTmp->bNumPause;
			ucNFC = pStyleTmp->ucNFC;
			szListChar = pStyleTmp->szListChar;
			ucAlignment = pStyleTmp->ucAlignment;
			if (bInList && !bWasInList) {
				/* Start of a list */
				iListSeqNumber++;
				vStartOfList(pDiag, ucNFC,
						bWasTableRow && !bIsTableRow);
			}
			if (!bInList && bWasInList) {
				/* End of a list */
				vEndOfList(pDiag);
			}
			bWasInList = bInList;
			if (bStartStyle) {
				pStyleInfo = pGetNextTextStyle(pStyleInfo);
				NO_DBG_HEX_C(pStyleInfo != NULL,
						pStyleInfo->ulFileOffset);
				DBG_MSG_C(pStyleInfo == NULL,
						"No more styles");
			}
			bStartStyle = FALSE;
			bStartStyleNext = FALSE;
		}

		if (bWasEndOfParagraph) {
			vStartOfParagraph1(pDiag, lBeforeIndentation);
		}

		if (!bIsTableRow &&
		    lTotalStringWidth(pAnchor) == 0) {
			if (!bNoMarks) {
				usListNumber = usGetListValue(iListSeqNumber,