parser.c

一个很有名的浏览器

		 * argument. */
		if (parse_element(*end, eof, &name, &namelen, NULL, NULL))
			continue;
		if (strlcasecmp(name, namelen, "/script", 7))
			continue;
		/* We have won! */
		break;
	}
	if (*end >= eof) {
		/* Either the document is not completely loaded yet or it's
		 * broken. At any rate, run away screaming. */
		*end = eof; /* Just for sanity. */
		return 1;
	}

	if (part->document && *html != '^') {
		add_to_string_list(&part->document->onload_snippets,
		                   html, *end - html);
	}
	return 0;
}
#endif

void
html_script(unsigned char *a)
{
#ifdef CONFIG_ECMASCRIPT
	/* We did everything (even possibly html_skip()) in do_html_script(). */
#else
	html_skip(a);
#endif
}

void
html_style(unsigned char *a)
{
	html_skip(a);
}

void
html_html(unsigned char *a)
{
	/* This is here just to get CSS stuff applied. */

	/* Modify the root HTML element - format_html_part() will take
	 * this from there. */
	struct html_element *e = html_context.stack.prev;

	if (par_format.bgcolor != format.style.bg)
		e->parattr.bgcolor = e->attr.style.bg = par_format.bgcolor = format.style.bg;
}

void
html_head(unsigned char *a)
{
	/* This makes sure it gets to the stack and helps tame down unclosed
	 * <title>. */
}

void
html_title(unsigned char *a)
{
	html_top.invisible = 1;
	html_top.type = ELEMENT_WEAK;
}

void
html_center(unsigned char *a)
{
	par_format.align = ALIGN_CENTER;
	if (!html_context.table_level)
		par_format.leftmargin = par_format.rightmargin = 0;
}

void
html_linebrk(unsigned char *a)
{
	unsigned char *al = get_attr_val(a, "align");

	if (al) {
		if (!strcasecmp(al, "left")) par_format.align = ALIGN_LEFT;
		else if (!strcasecmp(al, "right")) par_format.align = ALIGN_RIGHT;
		else if (!strcasecmp(al, "center")) {
			par_format.align = ALIGN_CENTER;
			if (!html_context.table_level)
				par_format.leftmargin = par_format.rightmargin = 0;
		} else if (!strcasecmp(al, "justify")) par_format.align = ALIGN_JUSTIFY;
		mem_free(al);
	}
}

void
html_br(unsigned char *a)
{
	html_linebrk(a);
	if (html_context.was_br)
		ln_break(2, html_context.line_break_f, html_context.part);
	else
		html_context.was_br = 1;
}

void
html_p(unsigned char *a)
{
	int_lower_bound(&par_format.leftmargin, html_context.margin);
	int_lower_bound(&par_format.rightmargin, html_context.margin);
	/*par_format.align = ALIGN_LEFT;*/
	html_linebrk(a);
}

void
html_address(unsigned char *a)
{
	par_format.leftmargin++;
	par_format.align = ALIGN_LEFT;
}

void
html_blockquote(unsigned char *a)
{
	par_format.leftmargin += 2;
	par_format.align = ALIGN_LEFT;
}

void
html_h(int h, unsigned char *a,
       enum format_align default_align)
{
	if (!par_format.align) par_format.align = default_align;
	html_linebrk(a);

	h -= 2;
	if (h < 0) h = 0;

	switch (par_format.align) {
		case ALIGN_LEFT:
			par_format.leftmargin = h * 2;
			par_format.rightmargin = 0;
			break;
		case ALIGN_RIGHT:
			par_format.leftmargin = 0;
			par_format.rightmargin = h * 2;
			break;
		case ALIGN_CENTER:
			par_format.leftmargin = par_format.rightmargin = 0;
			break;
		case ALIGN_JUSTIFY:
			par_format.leftmargin = par_format.rightmargin = h * 2;
			break;
	}
}

void
html_h1(unsigned char *a)
{
	format.style.attr |= AT_BOLD;
	html_h(1, a, ALIGN_CENTER);
}

void
html_h2(unsigned char *a)
{
	html_h(2, a, ALIGN_LEFT);
}

void
html_h3(unsigned char *a)
{
	html_h(3, a, ALIGN_LEFT);
}

void
html_h4(unsigned char *a)
{
	html_h(4, a, ALIGN_LEFT);
}

void
html_h5(unsigned char *a)
{
	html_h(5, a, ALIGN_LEFT);
}

void
html_h6(unsigned char *a)
{
	html_h(6, a, ALIGN_LEFT);
}

void
html_pre(unsigned char *a)
{
	format.style.attr |= AT_PREFORMATTED;
	par_format.leftmargin = (par_format.leftmargin > 1);
	par_format.rightmargin = 0;
}

void
html_xmp(unsigned char *a)
{
	html_context.was_xmp = 1;
	html_pre(a);
}

void
html_hr(unsigned char *a)
{
	int i/* = par_format.width - 10*/;
	unsigned char r = (unsigned char) BORDER_DHLINE;
	int q = get_num(a, "size");

	if (q >= 0 && q < 2) r = (unsigned char) BORDER_SHLINE;
	html_stack_dup(ELEMENT_KILLABLE);
	par_format.align = ALIGN_CENTER;
	mem_free_set(&format.link, NULL);
	format.form = NULL;
	html_linebrk(a);
	if (par_format.align == ALIGN_JUSTIFY) par_format.align = ALIGN_CENTER;
	par_format.leftmargin = par_format.rightmargin = html_context.margin;

	i = get_width(a, "width", 1);
	if (i == -1) i = get_html_max_width();
	format.style.attr = AT_GRAPHICS;
	html_context.special_f(html_context.part, SP_NOWRAP, 1);
	while (i-- > 0) {
		put_chrs(&r, 1, html_context.put_chars_f, html_context.part);
	}
	html_context.special_f(html_context.part, SP_NOWRAP, 0);
	ln_break(2, html_context.line_break_f, html_context.part);
	kill_html_stack_item(&html_top);
}

void
html_table(unsigned char *a)
{
	par_format.leftmargin = par_format.rightmargin = html_context.margin;
	par_format.align = ALIGN_LEFT;
	html_linebrk(a);
	format.style.attr = 0;
}

void
html_tr(unsigned char *a)
{
	html_linebrk(a);
}

void
html_th(unsigned char *a)
{
	/*html_linebrk(a);*/
	kill_html_stack_until(1, "TD", "TH", "", "TR", "TABLE", NULL);
	format.style.attr |= AT_BOLD;
	put_chrs(" ", 1, html_context.put_chars_f, html_context.part);
}

void
html_td(unsigned char *a)
{
	/*html_linebrk(a);*/
	kill_html_stack_until(1, "TD", "TH", "", "TR", "TABLE", NULL);
	format.style.attr &= ~AT_BOLD;
	put_chrs(" ", 1, html_context.put_chars_f, html_context.part);
}

void
html_base(unsigned char *a)
{
	unsigned char *al;

	al = get_url_val(a, "href");
	if (al) {
		unsigned char *base = join_urls(html_context.base_href, al);
		struct uri *uri = base ? get_uri(base, 0) : NULL;

		mem_free(al);
		mem_free_if(base);

		if (uri) {
			done_uri(html_context.base_href);
			html_context.base_href = uri;
		}
	}

	al = get_target(a);
	if (al) mem_free_set(&html_context.base_target, al);
}

void
html_ul(unsigned char *a)
{
	unsigned char *al;

	/* dump_html_stack(); */
	par_format.list_level++;
	par_format.list_number = 0;
	par_format.flags = P_STAR;

	al = get_attr_val(a, "type");
	if (al) {
		if (!strcasecmp(al, "disc") || !strcasecmp(al, "circle"))
			par_format.flags = P_O;
		else if (!strcasecmp(al, "square"))
			par_format.flags = P_PLUS;
		mem_free(al);
	}
	par_format.leftmargin += 2 + (par_format.list_level > 1);
	if (!html_context.table_level)
		int_upper_bound(&par_format.leftmargin, par_format.width / 2);

	par_format.align = ALIGN_LEFT;
	html_top.type = ELEMENT_DONT_KILL;
}

void
html_ol(unsigned char *a)
{
	unsigned char *al;
	int st;

	par_format.list_level++;
	st = get_num(a, "start");
	if (st == -1) st = 1;
	par_format.list_number = st;
	par_format.flags = P_NUMBER;

	al = get_attr_val(a, "type");
	if (al) {
		if (*al && !al[1]) {
			if (*al == '1') par_format.flags = P_NUMBER;
			else if (*al == 'a') par_format.flags = P_alpha;
			else if (*al == 'A') par_format.flags = P_ALPHA;
			else if (*al == 'r') par_format.flags = P_roman;
			else if (*al == 'R') par_format.flags = P_ROMAN;
			else if (*al == 'i') par_format.flags = P_roman;
			else if (*al == 'I') par_format.flags = P_ROMAN;
		}
		mem_free(al);
	}

	par_format.leftmargin += (par_format.list_level > 1);
	if (!html_context.table_level)
		int_upper_bound(&par_format.leftmargin, par_format.width / 2);

	par_format.align = ALIGN_LEFT;
	html_top.type = ELEMENT_DONT_KILL;
}

void
html_li(unsigned char *a)
{
	/* When handling the code <li><li> @was_li will be 1 and it means we
	 * have to insert a line break since no list item content has done it
	 * for us. */
	if (html_context.was_li) {
		html_context.line_breax = 0;
		ln_break(1, html_context.line_break_f, html_context.part);
	}

	/*kill_html_stack_until(0, "", "UL", "OL", NULL);*/
	if (!par_format.list_number) {
		unsigned char x[7] = "*&nbsp;";
		int t = par_format.flags & P_LISTMASK;

		if (t == P_O) x[0] = 'o';
		if (t == P_PLUS) x[0] = '+';
		put_chrs(x, 7, html_context.put_chars_f, html_context.part);
		par_format.leftmargin += 2;
		par_format.align = ALIGN_LEFT;

	} else {
		unsigned char c = 0;
		unsigned char n[32];
		int nlen;
		int t = par_format.flags & P_LISTMASK;
		int s = get_num(a, "value");

		if (s != -1) par_format.list_number = s;

		if (t == P_ALPHA || t == P_alpha) {
			put_chrs("&nbsp;", 6, html_context.put_chars_f, html_context.part);
			c = 1;
			n[0] = par_format.list_number
			       ? (par_format.list_number - 1) % 26
			         + (t == P_ALPHA ? 'A' : 'a')
			       : 0;
			n[1] = 0;

		} else if (t == P_ROMAN || t == P_roman) {
			roman(n, par_format.list_number);
			if (t == P_ROMAN) {
				unsigned char *x;

				for (x = n; *x; x++) *x = toupper(*x);
			}

		} else {
			if (par_format.list_number < 10) {
				put_chrs("&nbsp;", 6, html_context.put_chars_f, html_context.part);
				c = 1;
			}

			ulongcat(n, NULL, par_format.list_number, (sizeof(n) - 1), 0);
		}

		nlen = strlen(n);
		put_chrs(n, nlen, html_context.put_chars_f, html_context.part);
		put_chrs(".&nbsp;", 7, html_context.put_chars_f, html_context.part);
		par_format.leftmargin += nlen + c + 2;
		par_format.align = ALIGN_LEFT;
		html_top.next->parattr.list_number = par_format.list_number + 1;

		{
			struct html_element *element;

			element = search_html_stack("ol");
			if (element)
				element->parattr.list_number = par_format.list_number + 1;
		}

		par_format.list_number = 0;
	}

	html_context.putsp = -1;
	html_context.line_breax = 2;
	html_context.was_li = 1;
}

void
html_dl(unsigned char *a)
{
	par_format.flags &= ~P_COMPACT;
	if (has_attr(a, "compact")) par_format.flags |= P_COMPACT;
	if (par_format.list_level) par_format.leftmargin += 5;
	par_format.list_level++;
	par_format.list_number = 0;
	par_format.align = ALIGN_LEFT;
	par_format.dd_margin = par_format.leftmargin;
	html_top.type = ELEMENT_DONT_KILL;
	if (!(par_format.flags & P_COMPACT)) {
		ln_break(2, html_context.line_break_f, html_context.part);
		html_top.linebreak = 2;
	}
}

void
html_dt(unsigned char *a)
{
	kill_html_stack_until(0, "", "DL", NULL);
	par_format.align = ALIGN_LEFT;
	par_format.leftmargin = par_format.dd_margin;
	if (!(par_format.flags & P_COMPACT) && !has_attr(a, "compact"))
		ln_break(2, html_context.line_break_f, html_context.part);
}

void
html_dd(unsigned char *a)
{
	kill_html_stack_until(0, "", "DL", NULL);

	par_format.leftmargin = par_format.dd_margin
				+ (html_context.table_level ? 3 : 8);
	if (!html_context.table_level)
		int_upper_bound(&par_format.leftmargin, par_format.width / 2);
	par_format.align = ALIGN_LEFT;
}



void
html_noframes(unsigned char *a)
{
	struct html_element *element;

	if (!global_doc_opts->frames) return;

	element = search_html_stack("frameset");
	if (element && !element->frameset) return;

	html_skip(a);
}

void
html_frame(unsigned char *a)
{
	unsigned char *name, *src, *url;

	src = get_url_val(a, "src");
	if (!src) {
		url = stracpy("about:blank");
	} else {
		url = join_urls(html_context.base_href, src);
		mem_free(src);
	}
	if (!url) return;

	name = get_attr_val(a, "name");
	if (!name) {
		name = stracpy(url);
	} else if (!name[0]) {
		/* When name doesn't have a value */
		mem_free(name);
		name = stracpy(url);
	}
	if (!name) return;

	if (!global_doc_opts->frames || !html_top.frameset) {
		html_focusable(a);
		put_link_line("Frame: ", name, url, "");

	} else {
		if (html_context.special_f(html_context.part, SP_USED, NULL)) {
			html_context.special_f(html_context.part, SP_FRAME,
					       html_top.frameset, name, url);
		}
	}

	mem_free(name);
	mem_free(url);
}

void
html_frameset(unsigned char *a)
{
	struct frameset_param fp;
	unsigned char *cols, *rows;
	int width, height;

	/* XXX: This is still not 100% correct. We should also ignore the
	 * frameset when we encountered anything 3v1l (read as: non-whitespace
	 * text/element/anything) in the document outside of <head>. Well, this
	 * is still better than nothing and it should heal up the security
	 * concerns at least because sane sites should enclose the documents in
	 * <body> elements ;-). See also bug 171. --pasky */
	if (search_html_stack("BODY")
	    || !global_doc_opts->frames
	    || !html_context.special_f(html_context.part, SP_USED, NULL))
		return;

	cols = get_attr_val(a, "cols");
	if (!cols) {
		cols = stracpy("100%");
		if (!cols) return;
	}