reader.c

将rft格式文件转换成html格式文件

 * Set the current character set.  If csId is illegal, uses general charset.
 */

void
RTFSetCharSet (csId)
int	csId;
{
	switch (csId)
	{
	default:		/* use general if csId unknown */
	case rtfCSGeneral:
		curCharCode = genCharCode;
		curCharSet = csId;
		break;
	case rtfCSSymbol:
		curCharCode = symCharCode;
		curCharSet = csId;
		break;
	}
}


int
RTFGetCharSet ()
{
	return (curCharSet);
}


/* ---------------------------------------------------------------------- */

/*
 * Special destination readers.  They gobble the destination so the
 * writer doesn't have to deal with them.  That's wrong for any
 * translator that wants to process any of these itself.  In that
 * case, these readers should be overridden by installing a different
 * destination callback.
 *
 * NOTE: The last token read by each of these reader will be the
 * destination's terminating '}', which will then be the current token.
 * That '}' token is passed to RTFRouteToken() - the writer has already
 * seen the '{' that began the destination group, and may have pushed a
 * state; it also needs to know at the end of the group that a state
 * should be popped.
 *
 * It's important that rtf.h and the control token lookup table list
 * as many symbols as possible, because these destination readers
 * unfortunately make strict assumptions about the input they expect,
 * and a token of class rtfUnknown will throw them off easily.
 */


/*
 * Read { \fonttbl ... } destination.  Old font tables don't have
 * braces around each table entry; try to adjust for that.
 */

static void
ReadFontTbl ()
{
RTFFont	*fp;
char	buf[rtfBufSiz], *bp;
int	old = -1;
char	*fn = "ReadFontTbl";

	for (;;)
	{
		(void) RTFGetToken ();
		if (RTFCheckCM (rtfGroup, rtfEndGroup))
			break;
		if (old < 0)		/* first entry - determine tbl type */
		{
			if (RTFCheckCMM (rtfControl, rtfCharAttr, rtfFontNum))
				old = 1;	/* no brace */
			else if (RTFCheckCM (rtfGroup, rtfBeginGroup))
				old = 0;	/* brace */
			else			/* can't tell! */
				RTFPanic ("%s: Cannot determine format", fn);
		}
		if (old == 0)		/* need to find "{" here */
		{
			if (!RTFCheckCM (rtfGroup, rtfBeginGroup))
				RTFPanic ("%s: missing \"{\"", fn);
			(void) RTFGetToken ();	/* yes, skip to next token */
		}
		if ((fp = New (RTFFont)) == (RTFFont *) NULL)
			RTFPanic ("%s: cannot allocate font entry", fn);

		fp->rtfNextFont = fontList;
		fontList = fp;

		fp->rtfFName = (char *) NULL;
		fp->rtfFAltName = (char *) NULL;
		fp->rtfFNum = -1;
		fp->rtfFFamily = 0;
		fp->rtfFCharSet = 0;
		fp->rtfFPitch = 0;
		fp->rtfFType = 0;
		fp->rtfFCodePage = 0;

		while (rtfClass != rtfEOF && !RTFCheckCM (rtfText, ';'))
		{
			if (rtfClass == rtfControl)
			{
				switch (rtfMajor)
				{
				default:
					/* ignore token but announce it */
					RTFMsg ("%s: unknown token \"%s\"\n",
							fn, rtfTextBuf);
				case rtfFontFamily:
					fp->rtfFFamily = rtfMinor;
					break;
				case rtfCharAttr:
					switch (rtfMinor)
					{
					default:
						break;	/* ignore unknown? */
					case rtfFontNum:
						fp->rtfFNum = rtfParam;
						break;
					}
					break;
				case rtfFontAttr:
					switch (rtfMinor)
					{
					default:
						break;	/* ignore unknown? */
					case rtfFontCharSet:
						fp->rtfFCharSet = rtfParam;
						break;
					case rtfFontPitch:
						fp->rtfFPitch = rtfParam;
						break;
					case rtfFontCodePage:
						fp->rtfFCodePage = rtfParam;
						break;
					case rtfFTypeNil:
					case rtfFTypeTrueType:
						fp->rtfFType = rtfParam;
						break;
					}
					break;
				}
			}
			else if (RTFCheckCM (rtfGroup, rtfBeginGroup))	/* dest */
			{
				RTFSkipGroup ();	/* ignore for now */
			}
			else if (rtfClass == rtfText)	/* font name */
			{
				bp = buf;
				while (rtfClass != rtfEOF
					&& !RTFCheckCM (rtfText, ';'))
				{
					*bp++ = rtfMajor;
					(void) RTFGetToken ();
				}
				*bp = '\0';
				fp->rtfFName = RTFStrSave (buf);
				if (fp->rtfFName == (char *) NULL)
					RTFPanic ("%s: cannot allocate font name", fn);
				/* already have next token; don't read one */
				/* at bottom of loop */
				continue;
			}
			else
			{
				/* ignore token but announce it */
				RTFMsg ("%s: unknown token \"%s\"\n",
							fn, rtfTextBuf);
			}
			(void) RTFGetToken ();
		}
		if (old == 0)	/* need to see "}" here */
		{
			(void) RTFGetToken ();
			if (!RTFCheckCM (rtfGroup, rtfEndGroup))
				RTFPanic ("%s: missing \"}\"", fn);
		}
	}
	if (fp->rtfFNum == -1)
		RTFPanic ("%s: missing font number", fn);
/*
 * Could check other pieces of structure here, too, I suppose.
 */
	RTFRouteToken ();	/* feed "}" back to router */
}


/*
 * The color table entries have color values of -1 if
 * the default color should be used for the entry (only
 * a semi-colon is given in the definition, no color values).
 * There will be a problem if a partial entry (1 or 2 but
 * not 3 color values) is given.  The possibility is ignored
 * here.
 */

static void
ReadColorTbl ()
{
RTFColor	*cp;
int		cnum = 0;
char		*fn = "ReadColorTbl";

	for (;;)
	{
		(void) RTFGetToken ();
		if (RTFCheckCM (rtfGroup, rtfEndGroup))
			break;
		if ((cp = New (RTFColor)) == (RTFColor *) NULL)
			RTFPanic ("%s: cannot allocate color entry", fn);
		cp->rtfCNum = cnum++;
		cp->rtfCRed = cp->rtfCGreen = cp->rtfCBlue = -1;
		cp->rtfNextColor = colorList;
		colorList = cp;
		while (RTFCheckCM (rtfControl, rtfColorName))
		{
			switch (rtfMinor)
			{
			case rtfRed:	cp->rtfCRed = rtfParam; break;
			case rtfGreen:	cp->rtfCGreen = rtfParam; break;
			case rtfBlue:	cp->rtfCBlue = rtfParam; break;
			}
			RTFGetToken ();
		}
		if (!RTFCheckCM (rtfText, (int) ';'))
			RTFPanic ("%s: malformed entry", fn);
	}
	RTFRouteToken ();	/* feed "}" back to router */
}


/*
 * The "Normal" style definition doesn't contain any style number,
 * all others do.  Normal style is given style rtfNormalStyleNum.
 */

static void
ReadStyleSheet ()
{
RTFStyle	*sp;
RTFStyleElt	*sep, *sepLast;
char		buf[rtfBufSiz], *bp;
char		*fn = "ReadStyleSheet";

	for (;;)
	{
		(void) RTFGetToken ();
		if (RTFCheckCM (rtfGroup, rtfEndGroup))
			break;
		if ((sp = New (RTFStyle)) == (RTFStyle *) NULL)
			RTFPanic ("%s: cannot allocate stylesheet entry", fn);
		sp->rtfSName = (char *) NULL;
		sp->rtfSNum = -1;
		sp->rtfSType = rtfParStyle;
		sp->rtfSAdditive = 0;
		sp->rtfSBasedOn = rtfNoStyleNum;
		sp->rtfSNextPar = -1;
		sp->rtfSSEList = sepLast = (RTFStyleElt *) NULL;
		sp->rtfNextStyle = styleList;
		sp->rtfExpanding = 0;
		styleList = sp;
		if (!RTFCheckCM (rtfGroup, rtfBeginGroup))
			RTFPanic ("%s: missing \"{\"", fn);
		for (;;)
		{
			(void) RTFGetToken ();
			if (rtfClass == rtfEOF
				|| RTFCheckCM (rtfText, ';'))
				break;
			if (rtfClass == rtfControl)
			{
				if (RTFCheckMM (rtfSpecialChar, rtfOptDest))
					continue;	/* ignore "\*" */
				if (RTFCheckMM (rtfParAttr, rtfStyleNum))
				{
					sp->rtfSNum = rtfParam;
					sp->rtfSType = rtfParStyle;
					continue;
				}
				if (RTFCheckMM (rtfCharAttr, rtfCharStyleNum))
				{
					sp->rtfSNum = rtfParam;
					sp->rtfSType = rtfCharStyle;
					continue;
				}
				if (RTFCheckMM (rtfSectAttr, rtfSectStyleNum))
				{
					sp->rtfSNum = rtfParam;
					sp->rtfSType = rtfSectStyle;
					continue;
				}
				if (RTFCheckMM (rtfStyleAttr, rtfBasedOn))
				{
					sp->rtfSBasedOn = rtfParam;
					continue;
				}
				if (RTFCheckMM (rtfStyleAttr, rtfAdditive))
				{
					sp->rtfSAdditive = 1;
					continue;
				}
				if (RTFCheckMM (rtfStyleAttr, rtfNext))
				{
					sp->rtfSNextPar = rtfParam;
					continue;
				}
				if ((sep = New (RTFStyleElt)) == (RTFStyleElt *) NULL)
					RTFPanic ("%s: cannot allocate style element", fn);
				sep->rtfSEClass = rtfClass;
				sep->rtfSEMajor = rtfMajor;
				sep->rtfSEMinor = rtfMinor;
				sep->rtfSEParam = rtfParam;
				if ((sep->rtfSEText = RTFStrSave (rtfTextBuf))
								== (char *) NULL)
					RTFPanic ("%s: cannot allocate style element text", fn);
				if (sepLast == (RTFStyleElt *) NULL)
					sp->rtfSSEList = sep;	/* first element */
				else				/* add to end */
					sepLast->rtfNextSE = sep;
				sep->rtfNextSE = (RTFStyleElt *) NULL;
				sepLast = sep;
			}
			else if (RTFCheckCM (rtfGroup, rtfBeginGroup))
			{
				/*
				 * This passes over "{\*\keycode ... }, among
				 * other things. A temporary (perhaps) hack.
				 */
				RTFSkipGroup ();
				continue;
			}
			else if (rtfClass == rtfText)	/* style name */
			{
				bp = buf;
				while (rtfClass == rtfText)
				{
					if (rtfMajor == ';')
					{
						/* put back for "for" loop */
						(void) RTFUngetToken ();
						break;
					}
					*bp++ = rtfMajor;
					(void) RTFGetToken ();
				}
				*bp = '\0';
				if ((sp->rtfSName = RTFStrSave (buf)) == (char *) NULL)
					RTFPanic ("%s: cannot allocate style name", fn);
			}
			else		/* unrecognized */
			{
				/* ignore token but announce it */
				RTFMsg ("%s: unknown token \"%s\"\n",
							fn, rtfTextBuf);
			}
		}
		(void) RTFGetToken ();
		if (!RTFCheckCM (rtfGroup, rtfEndGroup))
			RTFPanic ("%s: missing \"}\"", fn);

		/*
		 * Check over the style structure.  A name is a must.
		 * If no style number was specified, check whether it's the
		 * Normal style (in which case it's given style number
		 * rtfNormalStyleNum).  Note that some "normal" style names
		 * just begin with "Normal" and can have other stuff following,
		 * e.g., "Normal,Times 10 point".  Ugh.
		 *
		 * Some German RTF writers use "Standard" instead of "Normal".
		 */
		if (sp->rtfSName == (char *) NULL)
			RTFPanic ("%s: missing style name", fn);
		if (sp->rtfSNum < 0)
		{
			if (strncmp (buf, "Normal", 6) != 0
				&& strncmp (buf, "Standard", 8) != 0)
				RTFPanic ("%s: missing style number", fn);
			sp->rtfSNum = rtfNormalStyleNum;
		}
		if (sp->rtfSNextPar == -1)	/* if \snext not given, */
			sp->rtfSNextPar = sp->rtfSNum;	/* next is itself */
	}
	RTFRouteToken ();	/* feed "}" back to router */
}


static void
ReadInfoGroup ()
{
	RTFSkipGroup ();
	RTFRouteToken ();	/* feed "}" back to router */
}


static void
ReadPictGroup ()
{
	RTFSkipGroup ();
	RTFRouteToken ();	/* feed "}" back to router */
}


static void
ReadObjGroup ()
{
	RTFSkipGroup ();
	RTFRouteToken ();	/* feed "}" back to router */
}


/* ---------------------------------------------------------------------- */

/*
 * Routines to return pieces of stylesheet, or font or color tables.
 * References to style 0 are mapped onto the Normal style.
 */


RTFStyle *
RTFGetStyle (num)
int	num;
{
RTFStyle	*s;

	if (num == -1)
		return (styleList);
	for (s = styleList; s != (RTFStyle *) NULL; s = s->rtfNextStyle)
	{
		if (s->rtfSNum == num)
			break;
	}
	return (s);		/* NULL if not found */
}


RTFFont *
RTFGetFont (num)
int	num;
{
RTFFont	*f;

	if (num == -1)
		return (fontList);
	for (f = fontList; f != (RTFFont *) NULL; f = f->rtfNextFont)
	{
		if (f->rtfFNum == num)
			break;
	}
	return (f);		/* NULL if not found */
}


RTFColor *
RTFGetColor (num)
int	num;
{
RTFColor	*c;

	if (num == -1)
		return (colorList);
	for (c = colorList; c != (RTFColor *) NULL; c = c->rtfNextColor)
	{
		if (c->rtfCNum == num)
			break;
	}
	return (c);		/* NULL if not found */
}


/* ---------------------------------------------------------------------- */


/*
 * Expand style n, if there is such a style.
 */

void
RTFExpandStyle (n)
int	n;
{
RTFStyle	*s;
RTFStyleElt	*se;

	if (n == -1 || (s = RTFGetStyle (n)) == (RTFStyle *) NULL)
		return;
	if (s->rtfExpanding != 0)
		RTFPanic ("Style expansion loop, style %d", n);
	s->rtfExpanding = 1;	/* set expansion flag for loop detection */
	/*
	 * Expand "based-on" style (unless it's the same as the current
	 * style -- Normal style usually gives itself as its own based-on
	 * style).  Based-on style expansion is done by synthesizing
	 * the token that the writer needs to see in order to trigger
	 * another style expansion, and feeding to token back through
	 * the router so the writer sees it.
	 */
	if (n != s->rtfSBasedOn)
	{
		RTFSetToken (rtfControl, rtfParAttr, rtfStyleNum,
							s->rtfSBasedOn, "\\s");
		RTFRouteToken ();
	}
	/*
	 * Now route the tokens unique to this style.  RTFSetToken()
	 * isn't used because it would add the param value to the end
	 * of the token text, which already has it in.
	 */
	for (se = s->rtfSSEList; se != (RTFStyleElt *) NULL; se = se->rtfNextSE)
	{
		rtfClass = se->rtfSEClass;
		rtfMajor = se->rtfSEMajor;
		rtfMinor = se->rtfSEMinor;
		rtfParam = se->rtfSEParam;
		(void) strcpy (rtfTextBuf, se->rtfSEText);
		rtfTextLen = strlen (rtfTextBuf);
		RTFRouteToken ();
	}
	s->rtfExpanding = 0;	/* done - clear expansion flag */
}


/* ---------------------------------------------------------------------- */

/*
 * Control symbol lookup routines
 */


typedef struct RTFCtrl	RTFCtrl;

struct RTFCtrl
{
	int	major;	/* major number */
	int	minor;	/* minor number */
	char	*str;	/* symbol name */
	int	hash;	/* symbol name hash value */
};

/*
 * A minor number of -1 means the token has no minor number
 * (all valid minor numbers are >= 0).
 */

static RTFCtrl	**rtfCtrl = (RTFCtrl **) NULL;
static int	nKeys;


/*
 * Initialize lookup table hash values.  Only need to do this once.
 */

# define        ctrlFileName    "rtf-ctrl"

static void
LookupInit ()
{
FILE    *f;
RTFCtrl *rp;
char    buf[rtfBufSiz];
int     line = 0, i;
char    *p1, *p2, *p3, c;
TSScanner       scanner;
char            *scanEscape;
char    *fn = "LookupInit";

        if (rtfCtrl != (RTFCtrl **) NULL)       /* already initialized */
                return;

        if ((f = RTFOpenLibFile (ctrlFileName, "r")) == (FILE *) NULL)