html.c

基于rtos开发的浏览器！

/*		Structured stream to Rich hypertext converter
**		============================================
**
**	This generates a hypertext object.  It converts from the
**	structured stream interface of HTML events into the style-
**	oriented interface of the HText.h interface.  This module is
**	only used in clients and should not be linked into servers.
**
**	Override this module if making a new GUI browser.
**
**   Being Overidden
**
*/
#include "HTUtils.h"
#include "tcp.h"

#define Lynx_HTML_Handler
#include "HTChunk.h"
#include "HText.h"
#include "HTStyle.h"
#include "HTML.h"

#include "HTCJK.h"
#include "HTAtom.h"
#include "HTAnchor.h"
#include "HTMLGen.h"
#include "HTParse.h"
#include "HTList.h"
#include "UCMap.h"
#include "UCDefs.h"
#include "UCAux.h"

#include "LYGlobalDefs.h"
#include "LYCharUtils.h"
#include "LYCharSets.h"

#include "HTAlert.h"
#include "HTFont.h"
#include "HTForms.h"
#include "HTNestedList.h"
#include "GridText.h"
#include "LYSignal.h"
#include "LYUtils.h"
#include "LYMap.h"
#include "LYList.h"
#include "LYBookmark.h"

#ifdef VMS
#include "LYCurses.h"
#include "HTVMSUtils.h"
#endif /* VMS */

#ifdef USE_COLOR_STYLE
#include "AttrList.h"
#include "LYHash.h"
#include "LYStyle.h"
#undef SELECTED_STYLES
#define pHText_changeStyle(X,Y,Z) {}
char Style_className[16384];
#endif

#include "LYexit.h"
#include "LYLeaks.h"

#define FREE(x) if (x) {free(x); x = NULL;}

extern BOOL HTPassEightBitRaw;
extern HTCJKlang HTCJK;

extern BOOLEAN HT_Is_Gopher_URL;

/* from Curses.h */
extern int LYcols;

struct _HTStream {
    CONST HTStreamClass *	isa;
    /* .... */
};

PRIVATE HTStyleSheet * styleSheet;	/* Application-wide */

/*	Module-wide style cache
*/
PUBLIC  HTStyle *styles[HTML_ELEMENTS+31]; /* adding 24 nested list styles  */
					   /* and 3 header alignment styles */
					   /* and 3 div alignment styles    */
PRIVATE HTStyle *default_style;

PUBLIC char *LYToolbarName = "LynxPseudoToolbar";

/* used to turn off a style if the HTML author forgot to
PRIVATE int i_prior_style = -1;
 */

/*
 *	Private function....
 */
PRIVATE void HTML_end_element PARAMS((HTStructured *me,
				      int element_number,
				      char **include));

/*		Forward declarations of routines
*/
PRIVATE void get_styles NOPARAMS;
PRIVATE void change_paragraph_style PARAMS((HTStructured * me,
					    HTStyle * style));

/*	Set an internal flag that the next call to a stack-affecting method
**	is only internal and the stack manipulation should be skipped. - kw
*/
#define SET_SKIP_STACK(el_num) if (HTML_dtd.tags[el_num].contents != SGML_EMPTY) \
						{ me->skip_stack++; }

extern int hash_code PARAMS((char* i));

PUBLIC void strtolower ARGS1(char*, i)
{
	if (!i) return;
	while (*i) { *i=tolower(*i); i++; }
}

/*		Flattening the style structure
**		------------------------------
**
On the NeXT, and on any read-only browser, it is simpler for the text to have
a sequence of styles, rather than a nested tree of styles. In this
case we have to flatten the structure as it arrives from SGML tags into
a sequence of styles.
*/

/*
**  If style really needs to be set, call this.
*/
PUBLIC void actually_set_style ARGS1(HTStructured *, me)
{
    if (!me->text) {			/* First time through */
	LYGetChartransInfo(me);
	UCSetTransParams(&me->T,
		     me->UCLYhndl, me->UCI,
			 HTAnchor_getUCLYhndl(me->node_anchor,
					      UCT_STAGE_HTEXT),
			 HTAnchor_getUCInfoStage(me->node_anchor,
						 UCT_STAGE_HTEXT));
	me->text = HText_new2(me->node_anchor, me->target);
	HText_beginAppend(me->text);
	HText_setStyle(me->text, me->new_style);
	me->in_word = NO;
	LYCheckForContentBase(me);
    } else {
	HText_setStyle(me->text, me->new_style);
    }

    me->old_style = me->new_style;
    me->style_change = NO;
}

/*
**  If you THINK you need to change style, call this.
*/
PRIVATE void change_paragraph_style ARGS2(HTStructured *, me, HTStyle *,style)
{
    if (me->new_style != style) {
	me->style_change = YES;
	me->new_style = style;
    }
    me->in_word = NO;
}

/*_________________________________________________________________________
**
**			A C T I O N	R O U T I N E S
*/

/*	Character handling
**	------------------
*/
PUBLIC void HTML_put_character ARGS2(HTStructured *, me, char, c)
{
    /*
     *	Ignore all non-MAP content when just
     *	scanning a document for MAPs. - FM
     */
    if (LYMapsOnly)
	return;

    /*
     *	Do EOL conversion if needed. - FM
     *
     *	Convert EOL styles:
     *	 macintosh:  cr    --> lf
     *	 ascii:      cr-lf --> lf
     *	 unix:	     lf    --> lf
     */
    if ((me->lastraw == '\r') && c == '\n') {
	me->lastraw = -1;
	return;
    }
    me->lastraw = c;
    if (c == '\r')
	c = '\n';

    /*
     *	Handle SGML_LITTERAL tags that have HTChunk elements. - FM
     */
    switch (me->sp[0].tag_number) {

    case HTML_COMMENT:
	return; /* Do Nothing */

    case HTML_TITLE:
	if (c == LY_SOFT_HYPHEN)
	    return;
	if (c != '\n' && c != '\t' && c != '\r')
	    HTChunkPutc(&me->title, c);
	else
	    HTChunkPutc(&me->title, ' ');
	return;

    case HTML_STYLE:
	HTChunkPutc(&me->style_block, c);
	return;

    case HTML_SCRIPT:
	HTChunkPutc(&me->script, c);
	return;

    case HTML_OBJECT:
	HTChunkPutc(&me->object, c);
	return;

    case HTML_TEXTAREA:
	HTChunkPutc(&me->textarea, c);
	return;

    case HTML_SELECT:
    case HTML_OPTION:
	HTChunkPutc(&me->option, c);
	return;

    case HTML_MATH:
	HTChunkPutc(&me->math, c);
	return;

    default:
	if (me->inSELECT) {
	    /*
	     *	If we are within a SELECT not caught by the cases
	     *	above - HTML_SELECT or HTML_OPTION may not be the
	     *	last element pushed on the style stack if there were
	     *	invalid markup tags within a SELECT element.  For error
	     *	recovery, treat text as part of the OPTION text, it is
	     *	probably meant to show up as user-visible text.
	     *	Having A as an open element while in SELECT is really sick,
	     *	don't make anchor text part of the option text in that case
	     *	since the option text will probably just be discarded. - kw
	     */
	    if (me->sp[0].tag_number == HTML_A)
		break;
	    HTChunkPutc(&me->option, c);
	    return;
	}
	break;
    } /* end first switch */

    /*
     *	Handle all other tag content. - FM
     */
    switch (me->sp[0].tag_number) {

    case HTML_PRE:				/* Formatted text */
	/*
	 *  We guarantee that the style is up-to-date in begin_litteral
	 *  But we still want to strip \r's
	 */
	if (c != '\r' &&
	    !(c == '\n' && me->inLABEL && !me->inP) &&
	    !(c == '\n' && !me->inPRE)) {
	    me->inP = TRUE;
	    me->inLABEL = FALSE;
	    HText_appendCharacter(me->text, c);
	}
	me->inPRE = TRUE;
	break;

    case HTML_LISTING:				/* Literal text */
    case HTML_XMP:
    case HTML_PLAINTEXT:
	/*
	 *  We guarantee that the style is up-to-date in begin_litteral
	 *  But we still want to strip \r's
	 */
	if (c != '\r')	{
	    me->inP = TRUE;
	    me->inLABEL = FALSE;
	    HText_appendCharacter(me->text, c);
	}
	break;

    default:
	/*
	 *  Free format text.
	 */
	if (!strcmp(me->sp->style->name,"Preformatted")) {
	    if (c != '\r' &&
		!(c == '\n' && me->inLABEL && !me->inP) &&
		!(c == '\n' && !me->inPRE)) {
		me->inP = TRUE;
		me->inLABEL = FALSE;
		HText_appendCharacter(me->text, c);
	    }
	    me->inPRE = TRUE;

	} else if (!strcmp(me->sp->style->name,"Listing") ||
		   !strcmp(me->sp->style->name,"Example")) {
	    if (c != '\r') {
		me->inP = TRUE;
		me->inLABEL = FALSE;
		HText_appendCharacter(me->text, c);
	    }

	} else {
	    if (me->style_change) {
		if ((c == '\n') || (c == ' '))
		    return;	/* Ignore it */
		UPDATE_STYLE;
	    }
	    if (c == '\n') {
		if (me->in_word) {
		    if (HText_getLastChar(me->text) != ' ') {
			me->inP = TRUE;
			me->inLABEL = FALSE;
			HText_appendCharacter(me->text, ' ');
		    }
		    me->in_word = NO;
		}

	    } else if (c == ' ' || c == '\t') {
		if (HText_getLastChar(me->text) != ' ') {
		    me->inP = TRUE;
		    me->inLABEL = FALSE;
		    HText_appendCharacter(me->text, ' ');
		}

	    } else if (c == '\r') {
	       /* ignore */

	    } else {
		me->inP = TRUE;
		me->inLABEL = FALSE;
		HText_appendCharacter(me->text, c);
		me->in_word = YES;
	    }
	}
    } /* end second switch */

    if (c == '\n' || c == '\t') {
	HText_setLastChar(me->text, ' '); /* set it to a generic seperater */

	/*
	 *  \r's are ignored.  In order to keep collapsing spaces
	 *  correctly we must default back to the previous
	 *  seperater if there was one
	 */
    } else if (c == '\r' && HText_getLastChar(me->text) == ' ') {
	HText_setLastChar(me->text, ' '); /* set it to a generic seperater */
    } else {
	HText_setLastChar(me->text, c);
    }
}

/*	String handling
**	---------------
**
**	This is written separately from put_character because the loop can
**	in some cases be promoted to a higher function call level for speed.
*/
PUBLIC void HTML_put_string ARGS2(HTStructured *, me, CONST char *, s)
{
   if (LYMapsOnly || s == NULL)
      return;

    switch (me->sp[0].tag_number) {

    case HTML_COMMENT:
	break;					/* Do Nothing */

    case HTML_TITLE:
	HTChunkPuts(&me->title, s);
	break;

    case HTML_STYLE:
	HTChunkPuts(&me->style_block, s);
	break;

    case HTML_SCRIPT:
	HTChunkPuts(&me->script, s);
	break;

    case HTML_PRE:				/* Formatted text */
    case HTML_LISTING:				/* Literal text */
    case HTML_XMP:
    case HTML_PLAINTEXT:
	/*
	 *  We guarantee that the style is up-to-date in begin_litteral
	 */
	HText_appendText(me->text, s);
	break;

    case HTML_OBJECT:
	HTChunkPuts(&me->object, s);
	break;

    case HTML_TEXTAREA:
	HTChunkPuts(&me->textarea, s);
	break;

    case HTML_SELECT:
    case HTML_OPTION:
	HTChunkPuts(&me->option, s);
	break;

    case HTML_MATH:
	HTChunkPuts(&me->math, s);
	break;

    default:					/* Free format text? */
	if (!me->sp->style->freeFormat) {
	    /*
	     *	If we are within a preformatted text style not caught
	     *	by the cases above (HTML_PRE or similar may not be the
	     *	last element pushed on the style stack). - kw
	     */
	    HText_appendText(me->text, s);
	    break;
	} else {
	    CONST char *p = s;
	    char c;
	    if (me->style_change) {
		for (; *p && ((*p == '\n') || (*p == '\r') ||
			      (*p == ' ') || (*p == '\t')); p++)
		    ;	/* Ignore leaders */
		if (!*p)
		    return;
		UPDATE_STYLE;
	    }
	    for (; *p; p++) {
		if (*p == 13 && p[1] != 10) {
		    /*
		     *	Treat any '\r' which is not followed by '\n'
		     *	as '\n', to account for macintosh lineend in
		     *	ALT attributes etc. - kw
		     */
		    c = '\n';
		} else {
		    c = *p;
		}