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

							iWordVersion,
							pStyleTmp);
			}
			if (bInList && bFirstLine) {
				vStartOfListItem(pDiag, bNoMarks);
			}
			vPutIndentation(pDiag, pAnchor, bNoMarks, bFirstLine,
					usListNumber, ucNFC, szListChar,
					lLeftIndentation, lLeftIndentation1);
			bFirstLine = FALSE;
			/* One number or mark per paragraph will do */
			bNoMarks = TRUE;
		}

		if (bWasEndOfParagraph) {
			vStartOfParagraph2(pDiag);
			bWasEndOfParagraph = FALSE;
		}

		switch (ulChar) {
		case PICTURE:
			(void)memset(&tImage, 0, sizeof(tImage));
			eRes = eExamineImage(pFile, ulFileOffsetImage, &tImage);
			switch (eRes) {
			case image_no_information:
				bSuccess = FALSE;
				break;
			case image_minimal_information:
			case image_full_information:
#if 0
				if (bOutputContainsText(pAnchor)) {
					OUTPUT_LINE();
				} else {
					RESET_LINE();
				}
#endif
				bSuccess = bTranslateImage(pDiag, pFile,
					eRes == image_minimal_information,
					ulFileOffsetImage, &tImage);
				break;
			default:
				DBG_DEC(eRes);
				bSuccess = FALSE;
				break;
			}
			if (!bSuccess) {
				vStoreString("[pic]", 5, pOutput);
			}
			break;
		case FOOTNOTE_CHAR:
			uiFootnoteNumber++;
			if (tOptions.eConversionType == conversion_xml) {
				vStoreCharacter((ULONG)FOOTNOTE_OR_ENDNOTE,
								pOutput);
				break;
			}
			vStoreCharacter((ULONG)'[', pOutput);
			vStoreNumberAsDecimal(uiFootnoteNumber, pOutput);
			vStoreCharacter((ULONG)']', pOutput);
			break;
		case ENDNOTE_CHAR:
			uiEndnoteNumber++;
			vStoreCharacter((ULONG)'[', pOutput);
			vStoreNumberAsRoman(uiEndnoteNumber, pOutput);
			vStoreCharacter((ULONG)']', pOutput);
			break;
		case UNKNOWN_NOTE_CHAR:
			vStoreString("[?]", 3, pOutput);
			break;
		case PAR_END:
			if (bIsTableRow) {
				vStoreCharacter((ULONG)'\n', pOutput);
				break;
			}
			if (bOutputContainsText(pAnchor)) {
				OUTPUT_LINE();
			} else {
				vMove2NextLine(pDiag,
					pOutput->tFontRef, pOutput->usFontSize);
				RESET_LINE();
			}
			vEndOfParagraph(pDiag,
					pOutput->tFontRef,
					pOutput->usFontSize,
					lAfterIndentation);
			bWasEndOfParagraph = TRUE;
			break;
		case HARD_RETURN:
			if (bIsTableRow) {
				vStoreCharacter((ULONG)'\n', pOutput);
				break;
			}
			if (bOutputContainsText(pAnchor)) {
				OUTPUT_LINE();
			} else {
				vMove2NextLine(pDiag,
					pOutput->tFontRef, pOutput->usFontSize);
				RESET_LINE();
			}
			break;
		case PAGE_BREAK:
		case COLUMN_FEED:
			pSection = pSectionNext;
			break;
		case TABLE_SEPARATOR:
			if (bIsTableRow) {
				vStoreCharacter(ulChar, pOutput);
				break;
			}
			vStoreCharacter((ULONG)' ', pOutput);
			vStoreCharacter((ULONG)TABLE_SEPARATOR_CHAR, pOutput);
			break;
		case TAB:
			if (bIsTableRow ||
			    tOptions.eConversionType == conversion_xml) {
				vStoreCharacter((ULONG)' ', pOutput);
				break;
			}
			if (tOptions.iParagraphBreak == 0 &&
			    (tOptions.eConversionType == conversion_text ||
			     tOptions.eConversionType == conversion_fmt_text)) {
				/* No logical lines, so no tab expansion */
				vStoreCharacter(TAB, pOutput);
				break;
			}
			lHalfSpaceWidth = (lComputeSpaceWidth(
					pOutput->tFontRef,
					pOutput->usFontSize) + 1) / 2;
			lTmp = lTotalStringWidth(pAnchor);
			lTmp += lDrawUnits2MilliPoints(pDiag->lXleft);
			lTmp /= lDefaultTabWidth;
			do {
				vStoreCharacter((ULONG)FILLER_CHAR, pOutput);
				lWidthCurr = lTotalStringWidth(pAnchor);
				lWidthCurr +=
					lDrawUnits2MilliPoints(pDiag->lXleft);
			} while (lTmp == lWidthCurr / lDefaultTabWidth &&
				 lWidthCurr < lWidthMax + lRightIndentation);
			break;
		default:
			if (bHiddenText && tOptions.bHideHiddenText) {
				continue;
			}
			if (bMarkDelText && tOptions.bRemoveRemovedText) {
				continue;
			}
			if (ulChar == UNICODE_ELLIPSIS &&
			    tOptions.eEncoding != encoding_utf_8) {
				vStoreString("...", 3, pOutput);
			} else {
				if (bAllCapitals) {
					ulChar = ulToUpper(ulChar);
				}
				vStoreCharacter(ulChar, pOutput);
			}
			break;
		}

		if (bWasTableRow && !bIsTableRow) {
			/* End of a table */
			vEndOfTable(pDiag);
			/* Resume normal font */
			NO_DBG_MSG("End of table font");
			vCloseFont();
			bTableFontClosed = TRUE;
			pOutput->ucFontColor = ucFontColor;
			pOutput->usFontStyle = usFontStyle;
			pOutput->usFontSize = usFontSize;
			pOutput->tFontRef = tOpenFont(
					ucFontNumber, usFontStyle, usFontSize);
		}
		bWasTableRow = bIsTableRow;

		if (bIsTableRow) {
			fail(pAnchor != pOutput);
			if (!bEndRowNorm && !bEndRowFast) {
				continue;
			}
			/* End of a table row */
			if (bEndRowNorm) {
				fail(pRowInfo == NULL);
				vTableRow2Window(pDiag, pAnchor, pRowInfo,
						tOptions.eConversionType,
						tOptions.iParagraphBreak);
			} else {
				fail(!bEndRowFast);
			}
			/* Reset */
			pAnchor = pStartNewOutput(pAnchor, NULL);
			pOutput = pAnchor;
			if (bEndRowNorm) {
				pRowInfo = pGetNextRowInfoListItem();
			}
			bIsTableRow = FALSE;
			bEndRowNorm = FALSE;
			bEndRowFast = FALSE;
			NO_DBG_HEX_C(pRowInfo != NULL,
						pRowInfo->ulFileOffsetStart);
			NO_DBG_HEX_C(pRowInfo != NULL,
						pRowInfo->ulFileOffsetEnd);
			continue;
		}
		lWidthCurr = lTotalStringWidth(pAnchor);
		lWidthCurr += lDrawUnits2MilliPoints(pDiag->lXleft);
		if (lWidthCurr < lWidthMax + lRightIndentation) {
			continue;
		}
		pLeftOver = pSplitList(pAnchor);
		vJustify2Window(pDiag, pAnchor,
				lWidthMax, lRightIndentation, ucAlignment);
		pAnchor = pStartNewOutput(pAnchor, pLeftOver);
		for (pOutput = pAnchor;
		     pOutput->pNext != NULL;
		     pOutput = pOutput->pNext)
			;	/* EMPTY */
		fail(pOutput == NULL);
		if (lTotalStringWidth(pAnchor) > 0) {
			vSetLeftIndentation(pDiag, lLeftIndentation);
		}
	}

	pAnchor = pStartNewOutput(pAnchor, NULL);
	pAnchor->szStorage = xfree(pAnchor->szStorage);
	pAnchor = xfree(pAnchor);
	vCloseFont();
	vFreeDocument();
	Hourglass_Off();
	return TRUE;
} /* end of bWordDecryptor */

/*
 * lLastStringWidth - compute the width of the last part of the output string
 */
static long
lLastStringWidth(const output_type *pAnchor)
{
	const output_type	*pCurr, *pStart;

	pStart = NULL;
	for (pCurr = pAnchor; pCurr != NULL; pCurr = pCurr->pNext) {
		if (pCurr->tNextFree == 1 &&
		    (pCurr->szStorage[0] == PAR_END ||
		     pCurr->szStorage[0] == HARD_RETURN)) {
			/* Found a separator. Start after the separator */
			pStart = pCurr->pNext;
		}
	}
	if (pStart == NULL) {
		/* No separators. Use the whole output string */
		pStart = pAnchor;
	}
	return lTotalStringWidth(pStart);
} /* end of lLastStringWidth */

/*
 * pHdrFtrDecryptor - turn a header/footer list element to something useful
 */
output_type *
pHdrFtrDecryptor(FILE *pFile, ULONG ulCharPosStart, ULONG ulCharPosNext)
{
	output_type	*pAnchor, *pOutput, *pLeftOver;
	ULONG	ulChar, ulFileOffset, ulCharPos;
	long	lWidthCurr, lWidthMax;
	long	lRightIndentation;
	USHORT	usChar;
	UCHAR	ucAlignment;
	BOOL	bSkip;

	fail(iWordVersion < 0);
	fail(tOptions.eConversionType == conversion_unknown);
	fail(tOptions.eEncoding == 0);

	if (ulCharPosStart == ulCharPosNext) {
		/* There are no bytes to decrypt */
		return NULL;
	}

	lRightIndentation = 0;
	ucAlignment = ALIGNMENT_LEFT;
	bSkip = FALSE;
	lWidthMax = lGetWidthMax(tOptions.iParagraphBreak);
	pAnchor = pStartNewOutput(NULL, NULL);
	pOutput = pAnchor;
	pOutput->tFontRef = tOpenFont(0, FONT_REGULAR, DEFAULT_FONT_SIZE);
	usChar = usToHdrFtrPosition(pFile, ulCharPosStart);
	ulCharPos = ulCharPosStart;
	ulFileOffset = ulCharPos2FileOffset(ulCharPos);
	while (usChar != (USHORT)EOF && ulCharPos != ulCharPosNext) {
		/* Skip embedded characters */
		if (usChar == START_EMBEDDED) {
			bSkip = TRUE;
		} else if (usChar == END_IGNORE || usChar == END_EMBEDDED) {
			bSkip = FALSE;
		}
		/* Translate character */
		if (bSkip || usChar == END_IGNORE || usChar == END_EMBEDDED) {
			ulChar = IGNORE_CHARACTER;
		} else {
			ulChar = ulTranslateCharacters(usChar,
					ulFileOffset,
					iWordVersion,
					tOptions.eConversionType,
					tOptions.eEncoding,
					bOldMacFile);
		}
		/* Process character */
		if (ulChar != IGNORE_CHARACTER) {
			switch (ulChar) {
			case PICTURE:
				vStoreString("[pic]", 5, pOutput);
				break;
			case PAR_END:
			case HARD_RETURN:
			case PAGE_BREAK:
			case COLUMN_FEED:
				/* To the next substring */
				pOutput = pStartNextOutput(pOutput);
				vCloseFont();
				pOutput->tFontRef = tOpenFont(0,
					FONT_REGULAR, DEFAULT_FONT_SIZE);
				/* A substring with just one character */
				if (ulChar == HARD_RETURN) {
					vStoreCharacter(HARD_RETURN, pOutput);
				} else {
					vStoreCharacter(PAR_END, pOutput);
				}
				/* To the next substring */
				pOutput = pStartNextOutput(pOutput);
				vCloseFont();
				pOutput->tFontRef = tOpenFont(0,
					FONT_REGULAR, DEFAULT_FONT_SIZE);
				fail(!bCheckDoubleLinkedList(pAnchor));
				break;
			case TABLE_SEPARATOR:
				vStoreCharacter((ULONG)' ', pOutput);
				vStoreCharacter((ULONG)TABLE_SEPARATOR_CHAR,
							pOutput);
				break;
			case TAB:
				vStoreCharacter((ULONG)FILLER_CHAR, pOutput);
				break;
			default:
				vStoreCharacter(ulChar, pOutput);
				break;
			}
		}
		lWidthCurr = lLastStringWidth(pAnchor);
		if (lWidthCurr >= lWidthMax + lRightIndentation) {
			pLeftOver = pSplitList(pAnchor);
			for (pOutput = pAnchor;
			     pOutput->pNext != NULL;
			     pOutput = pOutput->pNext)
				;	/* EMPTY */
			fail(pOutput == NULL);
			/* To the next substring */
			pOutput = pStartNextOutput(pOutput);
			/* A substring with just one HARD_RETURN */
			vStoreCharacter(HARD_RETURN, pOutput);
			/* Put the leftover piece(s) at the end */
			pOutput->pNext = pLeftOver;
			if (pLeftOver != NULL) {
				pLeftOver->pPrev = pOutput;
			}
			fail(!bCheckDoubleLinkedList(pAnchor));
			for (pOutput = pAnchor;
			     pOutput->pNext != NULL;
			     pOutput = pOutput->pNext)
				;	/* EMPTY */
			fail(pOutput == NULL);
		}
		usChar = usNextChar(pFile, hdrftr_list,
					&ulFileOffset, &ulCharPos, NULL);
	}
	vCloseFont();
	if (bOutputContainsText(pAnchor)) {
		return pAnchor;
	}
	pAnchor = pStartNewOutput(pAnchor, NULL);
	pAnchor->szStorage = xfree(pAnchor->szStorage);
	pAnchor = xfree(pAnchor);
	return NULL;
} /* end of pHdrFtrDecryptor */

/*
 * pFootnoteDecryptor - turn a footnote text list element into text
 */
char *
szFootnoteDecryptor(FILE *pFile, ULONG ulCharPosStart, ULONG ulCharPosNext)
{
	char	*szText;
	ULONG	ulChar, ulFileOffset, ulCharPos;
	USHORT	usChar;
	size_t	tLen, tIndex, tNextFree, tStorageSize;
	char	szResult[6];
	BOOL	bSkip;

	fail(iWordVersion < 0);
	fail(tOptions.eConversionType == conversion_unknown);
	fail(tOptions.eEncoding == 0);

	if (ulCharPosStart == ulCharPosNext) {
		/* There are no bytes to decrypt */
		return NULL;
	}

	if (tOptions.eConversionType != conversion_xml) {
		/* Only implemented for XML output */
		return NULL;
	}

	bSkip = FALSE;

	/* Initialise the text buffer */
	tStorageSize = INITIAL_SIZE;
	szText = xmalloc(tStorageSize);
	tNextFree = 0;
	szText[tNextFree] = '\0';

	/* Goto the start */
	usChar = usToFootnotePosition(pFile, ulCharPosStart);
	ulCharPos = ulCharPosStart;
	ulFileOffset = ulCharPos2FileOffset(ulCharPos);
	/* Skip the unwanted starting characters */
	while (usChar != (USHORT)EOF && ulCharPos != ulCharPosNext &&
	       (usChar == FOOTNOTE_OR_ENDNOTE ||
		usChar == PAR_END ||
		usChar == TAB ||
		usChar == (USHORT)' ')) {
		usChar = usNextChar(pFile, footnote_list,
					&ulFileOffset, &ulCharPos, NULL);
	}
	/* Process the footnote text */
	while (usChar != (USHORT)EOF && ulCharPos != ulCharPosNext) {
		/* Skip embedded characters */
		if (usChar == START_EMBEDDED) {
			bSkip = TRUE;
		} else if (usChar == END_IGNORE || usChar == END_EMBEDDED) {
			bSkip = FALSE;
		}
		/* Translate character */
		if (bSkip ||
		    usChar == END_IGNORE ||
		    usChar == END_EMBEDDED ||
		    usChar == FOOTNOTE_OR_ENDNOTE) {
			ulChar = IGNORE_CHARACTER;
		} else {
			ulChar = ulTranslateCharacters(usChar,
					ulFileOffset,
					iWordVersion,
					tOptions.eConversionType,
					tOptions.eEncoding,
					bOldMacFile);
		}
		/* Process character */
		if (ulChar == PICTURE) {
			tLen = 5;
			strcpy(szResult, "[pic]");
		} else if (ulChar == IGNORE_CHARACTER) {
			tLen = 0;
			szResult[0] = '\0';
		} else {
			switch (ulChar) {
			case PAR_END:
			case HARD_RETURN:
			case PAGE_BREAK:
			case COLUMN_FEED:
				ulChar = (ULONG)PAR_END;
				break;
			case TAB:
				ulChar = (ULONG)' ';
				break;
			default:
				break;
			}
			tLen = tUcs2Utf8(ulChar, szResult, sizeof(szResult));
		}
		/* Add the results to the text */
		if (tNextFree + tLen + 1 > tStorageSize) {
			tStorageSize += EXTENTION_SIZE;
			szText = xrealloc(szText, tStorageSize);
		}
		for (tIndex = 0; tIndex < tLen; tIndex++) {
			szText[tNextFree++] = szResult[tIndex];
		}
		szText[tNextFree] = '\0';
		/* Next character */
		usChar = usNextChar(pFile, footnote_list,
					&ulFileOffset, &ulCharPos, NULL);
	}
	/* Remove redundant spaces */
	while (tNextFree != 0 && szText[tNextFree - 1] == ' ') {
		szText[tNextFree - 1] = '\0';
		tNextFree--;
	}
	if (tNextFree == 0) {
		/* No text */
		szText = xfree(szText);
		return NULL;
	}
	return szText;
} /* end of szFootnoteDecryptor */