reader.c

将rft格式文件转换成html格式文件

/*
 * - Need to document error code meanings.
 * - Need to do something with \* on destinations.
 * - Make the parameter a long?
 *
 * reader.c - RTF file reader.  Release 1.11.
 *
 * ASCII 10 (\n) and 13 (\r) are ignored and silently discarded.
 * Nulls are also discarded.
 * (although the read hook will still get a look at them.)
 *
 * "\:" is not a ":", it's a control symbol.  But some versions of
 * Word seem to write "\:" for ":".  This reader treats "\:" as a
 * plain text ":"
 *
 * 19 Mar 93
 * - Add hack to skip "{\*\keycode ... }" group in stylesheet.
 * This is probably the wrong thing to do, but it's simple.
 * 13 Jul 93
 * - Add THINK C awareness to malloc() declaration.  Necessary so
 * compiler knows the malloc argument is 4 bytes.  Ugh.
 * 07 Sep 93
 * - Text characters are mapped onto standard codes, which are placed
 * in rtfMinor.
 * - Eliminated use of index() function.
 * 05 Mar 94
 * - Added zillions of new symbols (those defined in RTF spec 1.2).
 * 14 Mar 94
 * - Public functions RTFMsg() and RTFPanic() now take variable arguments.
 * This means RTFPanic() now is used in place of what was formerly the
 * internal function Error().
 * - 8-bit characters are now legal, so they're not converted to \'xx
 * hex char representation now.
 * 01 Apr 94
 * - Added public variables rtfLineNum and rtfLinePos.
 * - #include string.h or strings.h, avoiding strncmp() problem where
 * last argument is treated as zero when prototype isn't available.
 * 04 Apr 94
 * - Treat style numbers 222 and 0 properly as "no style" and "normal".
 * 08 Apr 94
 * - Control symbol table is no longer compiled in.  It's read in the
 * first time that RTFInit() is called from the library file rtf-ctrl.
 * This shrinks all the translator binaries.  rtf-ctrl is built by
 * mkrtf-ctrl.c.
 */

# ifndef STRING_H
# define STRING_H <string.h>
# endif

# include	<stdio.h>
# include	<ctype.h>
# include	STRING_H
# ifdef STDARG
# include	<stdarg.h>
# else
# ifdef	VARARGS
# include	<varargs.h>
# endif	/* VARARGS */
# endif	/* STDARG */

# include	"tokenscan.h"

# define	rtfInternal
# include	"rtf.h"
# undef		rtfInternal


/*
 * Return pointer to new element of type t, or NULL
 * if no memory available.
 */

# define	New(t)	((t *) RTFAlloc (sizeof (t)))


/* maximum number of character values representable in a byte */

# define	charSetSize		256

/* charset stack size */

# define	maxCSStack		40


#ifndef THINK_C
extern char	*malloc ();
#else
extern void	*malloc(size_t);
#endif

static void	_RTFGetToken ();
 void	_RTFGetToken2 ();
static int	GetChar ();
static void	ReadFontTbl ();
static void	ReadColorTbl ();
static void	ReadStyleSheet ();
static void	ReadInfoGroup ();
static void	ReadPictGroup ();
static void	ReadObjGroup ();
static void	LookupInit ();
static void	Lookup ();
static int	Hash ();

static void	CharSetInit ();
static void	ReadCharSetMaps ();


/*
 * Public variables (listed in rtf.h)
 */

int	rtfClass;
int	rtfMajor;
int	rtfMinor;
int	rtfParam;
char	*rtfTextBuf = (char *) NULL;
int	rtfTextLen;

long	rtfLineNum;
int	rtfLinePos;


/*
 * Private stuff
 */

static int	pushedChar;	/* pushback char if read too far */

static int	pushedClass;	/* pushed token info for RTFUngetToken() */
static int	pushedMajor;
static int	pushedMinor;
static int	pushedParam;
static char	*pushedTextBuf = (char *) NULL;

static int	prevChar;
static int	bumpLine;


static RTFFont	*fontList = (RTFFont *) NULL;	/* these lists MUST be */
static RTFColor	*colorList = (RTFColor *) NULL;	/* initialized to NULL */
static RTFStyle	*styleList = (RTFStyle *) NULL;


static FILE	*rtffp = stdin;

static char	*inputName = (char *) NULL;
static char	*outputName = (char *) NULL;


/*
 * This array is used to map standard character names onto their numeric codes.
 * The position of the name within the array is the code.
 * stdcharnames.h is generated in the ../h directory.
 */

static char	*stdCharName[] =
{
# include	"stdcharnames.h"
	(char *) NULL
};


/*
 * These arrays are used to map RTF input character values onto the standard
 * character names represented by the values.  Input character values are
 * used as indices into the arrays to produce standard character codes.
 */


static char	*genCharSetFile = (char *) NULL;
static int	genCharCode[charSetSize];	/* general */
static int	haveGenCharSet = 0;

static char	*symCharSetFile = (char *) NULL;
static int	symCharCode[charSetSize];	/* symbol */
static int	haveSymCharSet = 0;

static int	curCharSet = rtfCSGeneral;
static int	*curCharCode = genCharCode;

/*
 * By default, the reader is configured to handle charset mapping invisibly,
 * including reading the charset files and switching charset maps as necessary
 * for Symbol font.
 */

static int	autoCharSetFlags;

/*
 * Stack for keeping track of charset map on group begin/end.  This is
 * necessary because group termination reverts the font to the previous
 * value, which may implicitly change it.
 */

static int	csStack[maxCSStack];
int	csTop = 0;

/*
 * Initialize the reader.  This may be called multiple times,
 * to read multiple files.  The only thing not reset is the input
 * stream; that must be done with RTFSetStream().
 */

void
RTFInit ()
{
int	i;
RTFColor	*cp;
RTFFont		*fp;
RTFStyle	*sp;
RTFStyleElt	*eltList, *ep;

	rtfClass = -1;
	pushedClass = -1;
	pushedChar = EOF;

	rtfLineNum = 0;
	rtfLinePos = 0;
	prevChar = EOF;
	bumpLine = 0;

	if (rtfTextBuf == (char *) NULL)	/* initialize text buffers */
	{
		rtfTextBuf = RTFAlloc (rtfBufSiz);
		pushedTextBuf = RTFAlloc (rtfBufSiz);
		if (rtfTextBuf == (char *) NULL
			|| pushedTextBuf == (char *) NULL)
			RTFPanic ("Cannot allocate text buffers.");
		rtfTextBuf[0] = pushedTextBuf[0] = '\0';
	}

	RTFFree (inputName);
	RTFFree (outputName);
	inputName = outputName = (char *) NULL;
			
	/* initialize control symbol lookup table */
	LookupInit ();
	
	for (i = 0; i < rtfMaxClass; i++)
		RTFSetClassCallback (i, (RTFFuncPtr) NULL);
	for (i = 0; i < rtfMaxDestination; i++)
		RTFSetDestinationCallback (i, (RTFFuncPtr) NULL);

	/* install built-in destination readers */
	RTFSetDestinationCallback (rtfFontTbl, ReadFontTbl);
	RTFSetDestinationCallback (rtfColorTbl, ReadColorTbl);
	RTFSetDestinationCallback (rtfStyleSheet, ReadStyleSheet);
	RTFSetDestinationCallback (rtfInfo, ReadInfoGroup);
	RTFSetDestinationCallback (rtfPict, ReadPictGroup);
	RTFSetDestinationCallback (rtfObject, ReadObjGroup);


	RTFSetReadHook ((RTFFuncPtr) NULL);

	/* dump old lists if necessary */

	while (fontList != (RTFFont *) NULL)
	{
		fp = fontList->rtfNextFont;
		RTFFree (fontList->rtfFName);
		RTFFree ((char *) fontList);
		fontList = fp;
	}
	while (colorList != (RTFColor *) NULL)
	{
		cp = colorList->rtfNextColor;
		RTFFree ((char *) colorList);
		colorList = cp;
	}
	while (styleList != (RTFStyle *) NULL)
	{
		sp = styleList->rtfNextStyle;
		eltList = styleList->rtfSSEList;
		while (eltList != (RTFStyleElt *) NULL)
		{
			ep = eltList->rtfNextSE;
			RTFFree (eltList->rtfSEText);
			RTFFree ((char *) eltList);
			eltList = ep;
		}
		RTFFree (styleList->rtfSName);
		RTFFree ((char *) styleList);
		styleList = sp;
	}

	CharSetInit ();
	csTop = 0;
}


/*
 * Set the reader's input stream to the given stream.  Can
 * be used to redirect to other than the default (stdin).
 */

void
RTFSetStream (stream)
FILE	*stream;
{
	rtffp = stream;
}


/*
 * Set or get the input or output file name.  These are never guaranteed
 * to be accurate, only insofar as the calling program makes them so.
 */

void
RTFSetInputName (name)
char	*name;
{
	if ((inputName = RTFStrSave (name)) == (char *) NULL)
		RTFPanic ("RTFSetInputName: out of memory");
}


char *
RTFGetInputName ()
{
	return (inputName);
}


void
RTFSetOutputName (name)
char	*name;
{
	if ((outputName = RTFStrSave (name)) == (char *) NULL)
		RTFPanic ("RTFSetOutputName: out of memory");
}


char *
RTFGetOutputName ()
{
	return (outputName);
}



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

/*
 * Callback table manipulation routines
 */


/*
 * Install or return a writer callback for a token class
 */


static RTFFuncPtr	ccb[rtfMaxClass];		/* class callbacks */


void
RTFSetClassCallback (class, callback)
int		class;
RTFFuncPtr	callback;
{
	if (class >= 0 && class < rtfMaxClass)
		ccb[class] = callback;
}


RTFFuncPtr
RTFGetClassCallback (class)
int	class;
{
	if (class >= 0 && class < rtfMaxClass)
		return (ccb[class]);
	return ((RTFFuncPtr) NULL);
}


/*
 * Install or return a writer callback for a destination type
 */

static RTFFuncPtr	dcb[rtfMaxDestination];	/* destination callbacks */


void
RTFSetDestinationCallback (dest, callback)
int		dest;
RTFFuncPtr	callback;
{
	if (dest >= 0 && dest < rtfMaxDestination)
		dcb[dest] = callback;
}


RTFFuncPtr
RTFGetDestinationCallback (dest)
int	dest;
{
	if (dest >= 0 && dest < rtfMaxDestination)
		return (dcb[dest]);
	return ((RTFFuncPtr) NULL);
}


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

/*
 * Token reading routines
 */


/*
 * Read the input stream, invoking the writer's callbacks
 * where appropriate.
 */

void
RTFRead ()
{
	while (RTFGetToken () != rtfEOF)
		RTFRouteToken ();
}


/*
 * Route a token.  If it's a destination for which a reader is
 * installed, process the destination internally, otherwise
 * pass the token to the writer's class callback.
 */

void
RTFRouteToken ()
{
RTFFuncPtr	p;

	if (rtfClass < 0 || rtfClass >= rtfMaxClass)	/* watchdog */
	{
		RTFPanic ("Unknown class %d: %s (reader malfunction)",
							rtfClass, rtfTextBuf);
	}
	if (RTFCheckCM (rtfControl, rtfDestination))
	{
		/* invoke destination-specific callback if there is one */
		if ((p = RTFGetDestinationCallback (rtfMinor))
							!= (RTFFuncPtr) NULL)
		{
			(*p) ();
			return;
		}
	}
	/* invoke class callback if there is one */
	if ((p = RTFGetClassCallback (rtfClass)) != (RTFFuncPtr) NULL)
		(*p) ();
}


/*
 * Skip to the end of the current group.  When this returns,
 * writers that maintain a state stack may want to call their
 * state unstacker; global vars will still be set to the group's
 * closing brace.
 */

void
RTFSkipGroup ()
{
int	level = 1;
int savecsTop;
	if(csTop>0)savecsTop=csTop-1;
	else savecsTop=0;

	RTFGetToken ();
	while (1)
	{
		if (rtfClass == rtfEOF)
			break;
		if (rtfClass == rtfGroup)
		{
			if (rtfMajor == rtfBeginGroup)
				++level;
			else if (rtfMajor == rtfEndGroup)
			{
				if (--level < 1)
					break;	/* end of initial group */
			}
		}
		_RTFGetToken2();
	}
	csTop=savecsTop;
}


/*
 * Read one token.  Call the read hook if there is one.  The
 * token class is the return value.  Returns rtfEOF when there
 * are no more tokens.
 */

int
RTFGetToken ()
{
RTFFuncPtr	p;

	for (;;)
	{
		_RTFGetToken ();
		if ((p = RTFGetReadHook ()) != (RTFFuncPtr) NULL)
			(*p) ();	/* give read hook a look at token */

		/* Silently discard newlines, carriage returns, nulls.  */
		if (!(rtfClass == rtfText
			&& (rtfMajor == '\n' || rtfMajor == '\r'
						|| rtfMajor == '\0')))
			break;
	}
	return (rtfClass);
}


/*
 * Install or return a token reader hook.
 */

static RTFFuncPtr	readHook;


void
RTFSetReadHook (f)
RTFFuncPtr	f;
{
	readHook = f;
}


RTFFuncPtr
RTFGetReadHook ()
{
	return (readHook);
}


void
RTFUngetToken ()
{
	if (pushedClass >= 0)	/* there's already an ungotten token */
		RTFPanic ("cannot unget two tokens");
	if (rtfClass < 0)
		RTFPanic ("no token to unget");
	pushedClass = rtfClass;
	pushedMajor = rtfMajor;
	pushedMinor = rtfMinor;
	pushedParam = rtfParam;
	(void) strcpy (pushedTextBuf, rtfTextBuf);
}


int
RTFPeekToken ()
{
	_RTFGetToken ();
	RTFUngetToken ();
	return (rtfClass);
}


static void
_RTFGetToken ()
{
RTFFont	*fp;

	/* first check for pushed token from RTFUngetToken() */