html.c

this is the file used to browse web

		d++;
	}
	fp->parent = html_top.frameset;
	if (fp->n) html_top.frameset = special_f(ff, SP_FRAMESET, fp);
	mem_free(fp);
	f:
	mem_free(c);
}*/

void html_link(unsigned char *a)
{
	unsigned char *name, *url, *title;
	if ((name = get_attr_val(a, "type"))) {
		if (strcasecmp(name, "text/html")) {
			mem_free(name);
			return;
		}
		mem_free(name);
	}
	if (!(url = get_url_val(a, "href"))) return;
	if (!(name = get_attr_val(a, "rel")))
		if (!(name = get_attr_val(a, "rev")))
			if (!(name = get_attr_val(a, "ref")))
				name = stracpy(url);
	if (!strcasecmp(name, "stylesheet") ||
	    !strcasecmp(name, "alternate stylesheet") ||
	    !strcasecmp(name, "made") ||
	    !strcasecmp(name, "icon") ||
	    !strcasecmp(name, "shortcut icon") ||
	    !strcasecmp(name, "apple-touch-icon") ||
	    !strcasecmp(name, "meta") ||
	    !strcasecmp(name, "pingback") ||
	    !strcasecmp(name, "File-List") ||
	    !casecmp(name, "schema", 6)) goto skip;
	if ((title = get_attr_val(a, "title"))) {
		add_to_strn(&name, ": ");
		add_to_strn(&name, title);
		mem_free(title);
	}
	put_link_line("Link: ", name, url, format.target_base);
	skip:
	mem_free(name);
	mem_free(url);
}

struct element_info {
	char *name;
	void (*func)(unsigned char *);
	int linebreak;
	int nopair;
};

struct element_info elements[] = {
	{"SPAN",	html_span,	0, 0},
	{"B",		html_bold,	0, 0},
	{"STRONG",	html_bold,	0, 0},
	{"DFN",		html_bold,	0, 0},
	{"I",		html_italic,	0, 0},
	{"Q",		html_italic,	0, 0},
	{"CITE",	html_italic,	0, 0},
	{"EM",		html_italic,	0, 0},
	{"ABBR",	html_italic,	0, 0},
	{"U",		html_underline,	0, 0},
	{"S",		html_underline,	0, 0},
	{"STRIKE",	html_underline,	0, 0},
	{"FIXED",	html_fixed,	0, 0},
	{"CODE",	html_fixed,	0, 0},
	{"FONT",	html_font,	0, 0},
	{"A",		html_a,		0, 2},
	{"IMG",		html_img,	0, 1},

	{"BASE",	html_base,	0, 1},
	{"BASEFONT",	html_font,	0, 1},

	{"BODY",	html_body,	0, 0},

/*	{"HEAD",	html_skip,	0, 0},*/
	{"TITLE",	html_title,	0, 0},
	{"SCRIPT",	html_skip,	0, 0},
	{"STYLE",	html_skip,	0, 0},

	{"BR",		html_br,	1, 1},
	{"DIV",		html_linebrk,	1, 0},
	{"CENTER",	html_center,	1, 0},
	{"CAPTION",	html_center,	1, 0},
	{"P",		html_p,		2, 2},
	{"HR",		html_hr,	2, 1},
	{"H1",		html_center,	2, 2},
	{"H2",		html_h2,	2, 2},
	{"H3",		html_h3,	2, 2},
	{"H4",		html_h4,	2, 2},
	{"H5",		html_h5,	2, 2},
	{"H6",		html_h6,	2, 2},
	{"BLOCKQUOTE",	html_blockquote,2, 0},
	{"ADDRESS",	html_address,	2, 0},
	{"PRE",		html_pre,	2, 0},
	{"LISTING",	html_pre,	2, 0},

	{"UL",		html_ul,	1, 0},
	{"DIR",		html_ul,	1, 0},
	{"MENU",	html_ul,	1, 0},
	{"OL",		html_ol,	1, 0},
	{"LI",		html_li,	1, 3},
	{"DL",		html_dl,	1, 0},
	{"DT",		html_dt,	1, 1},
	{"DD",		html_dd,	1, 1},

	{"TABLE",	html_table,	2, 0},
	{"TR",		html_tr,	1, 0},
	{"TD",		html_td,	0, 0},
	{"TH",		html_th,	0, 0},

	{"FORM",	html_form,	1, 0},
	{"INPUT",	html_input,	0, 1},
	{"TEXTAREA",	html_textarea,	0, 1},
	{"SELECT",	html_select,	0, 0},
	{"OPTION",	html_option,	1, 1},
	{"BUTTON",	html_button,	0, 0},

	{"LINK",	html_link,	1, 1},
	{"IFRAME",	html_iframe,	1, 1},
	{"FRAME",	html_frame,	1, 1},
	{"FRAMESET",	html_frameset,	1, 0},
	{"NOFRAMES",	html_noframes,	0, 0},

	{NULL,		NULL, 0, 0},
};

unsigned char *skip_comment(unsigned char *html, unsigned char *eof)
{
	int comm = html + 4 <= eof && html[2] == '-' && html[3] == '-';
	html += comm ? 4 : 2;
	while (html < eof) {
		if (!comm && html[0] == '>') return html + 1;
		if (comm && html + 2 <= eof && html[0] == '-' && html[1] == '-') {
			html += 2;
			while (html < eof && *html == '-') html++;
			while (html < eof && WHITECHAR(*html)) html++;
			if (html >= eof) return eof;
			if (*html == '>') return html + 1;
			continue;
		}
		html++;
	}
	return eof;
}

void process_head(unsigned char *head)
{
	unsigned char *r, *p;
	if ((r = parse_http_header(head, "Refresh", NULL))) {
		if ((p = parse_header_param(r, "URL")) || (p = parse_header_param(r, ""))) {
			put_link_line("Refresh: ", p, p, d_opt->framename);
			mem_free(p);
		}
		mem_free(r);
	}
}

int qd(unsigned char *html, unsigned char *eof, int *len)
{
	int l;
	*len = 1;
	if (html >= eof) {
		internal("qd: out of data, html == %p, eof == %p", html, eof);
		return -1;
	}
	if (html[0] != '&') return html[0];
	if (html + 1 >= eof) return -1;
	if (html[1] != '#') return -1;
	for (l = 2; l < 10 && html + l < eof; l++) if (html[l] == ';') {
		int n = get_entity_number(html + 2, l - 2);
		if (n >= 0) {
			*len = l + 1;
			return n;
		}
		break;
	}
	return -1;
}

void parse_html(unsigned char *html, unsigned char *eof, int (*put_chars)(void *, unsigned char *, int), void (*line_break)(void *), void *(*special)(void *, int, ...), void *f, unsigned char *head)
{
	/*unsigned char *start = html;*/
	unsigned char *lt;
	putsp = -1;
	line_breax = table_level ? 2 : 1;
	pos = 0;
	was_br = 0;
	put_chars_f = put_chars;
	line_break_f = line_break;
	special_f = special;
	ff = f;
	eoff = eof;
	if (head) process_head(head);
	set_lt:
	put_chars_f = put_chars;
	line_break_f = line_break;
	special_f = special;
	ff = f;
	eoff = eof;
	lt = html;
	while (html < eof) {
		unsigned char *name, *attr, *end;
		int namelen;
		struct element_info *ei;
		int inv;
		if (WHITECHAR(*html) && par_format.align != AL_NO) {
			unsigned char *h = html;
			/*if (putsp == -1) {
				while (html < eof && WHITECHAR(*html)) html++;
				goto set_lt;
			}
			putsp = 0;*/
			while (h < eof && WHITECHAR(*h)) h++;
			if (h + 1 < eof && h[0] == '<' && h[1] == '/') {
				if (!parse_element(h, eof, &name, &namelen, &attr, &end)) {
					put_chrs(lt, html - lt, put_chars, f);
					lt = html = h;
					if (!html_top.invisible) putsp = 1;
					goto element;
				}
			}
			html++;
			if (!(pos + (html-lt-1))) goto skip_w; /* ??? */
			if (*(html - 1) == ' ') {
				if (html < eof && !WHITECHAR(*html)) continue;	/* BIG performance win; not sure if it doesn't cause any bug */
				put_chrs(lt, html - lt, put_chars, f);
			} else {
				put_chrs(lt, html - 1 - lt, put_chars, f);
				put_chrs(" ", 1, put_chars, f);
			}
			skip_w:
			while (html < eof && WHITECHAR(*html)) html++;
			/*putsp = -1;*/
			goto set_lt;
			put_sp:
			put_chrs(" ", 1, put_chars, f);
			/*putsp = -1;*/
		}
		if (par_format.align == AL_NO && (*html < 32 || *html == '&')) {
			int l;
			int q = qd(html, eof, &l);
			putsp = 0;
			if (q == 9) {
				put_chrs(lt, html - lt, put_chars, f);
				put_chrs("        ", 8 - pos % 8, put_chars, f);
				html += l;
				goto set_lt;
			} else if (q == 13 || q == 10) {
				put_chrs(lt, html - lt, put_chars, f);
				next_break:
				html += l;
				if (q == 13 && html < eof - 1 && qd(html, eof, &l) == 10) html += l;
				ln_break(1, line_break, f);
				if (html >= eof) goto set_lt;
				q = qd(html, eof, &l);
				if (q == 13 || q == 10) {
					line_breax = 0;
					goto next_break;
				}
				goto set_lt;
			}
		}
		if (*html < ' ') {
			/*if (putsp == 1) goto put_sp;
			putsp = 0;*/
			put_chrs(lt, html - lt, put_chars, f);
			put_chrs(".", 1, put_chars, f);
			html++;
			goto set_lt;
		}
		if (html + 2 <= eof && html[0] == '<' && (html[1] == '!' || html[1] == '?') && !d_opt->plain) {
			/*if (putsp == 1) goto put_sp;
			putsp = 0;*/
			put_chrs(lt, html - lt, put_chars, f);
			html = skip_comment(html, eof);
			goto set_lt;
		}
		if (*html != '<' || d_opt->plain || parse_element(html, eof, &name, &namelen, &attr, &end)) {
			/*if (putsp == 1) goto put_sp;
			putsp = 0;*/
			html++;
			continue;
		}
		element:
		inv = *name == '/'; name += inv; namelen -= inv;
		if (!inv && putsp == 1 && !html_top.invisible) goto put_sp;
		put_chrs(lt, html - lt, put_chars, f);
		if (par_format.align != AL_NO) if (!inv && !putsp) {
			unsigned char *ee = end;
			unsigned char *nm;
			while (!parse_element(ee, eof, &nm, NULL, NULL, &ee))
				if (*nm == '/') goto ng;
			if (ee < eof && WHITECHAR(*ee)) {
				/*putsp = -1;*/
				put_chrs(" ", 1, put_chars, f);
			}
			ng:;
		}
		html = end;
		for (ei = elements; ei->name; ei++) {
			if (ei->name &&
			   (strlen(ei->name) != (size_t)namelen || casecmp(ei->name, name, namelen)))
				continue;
			if (!inv) {
				char *a;
				ln_break(ei->linebreak, line_break, f);
				if (ei->name && ((a = get_attr_val(attr, "id")))) {
					special(f, SP_TAG, a);
					mem_free(a);
				}
				if (!html_top.invisible) {
					int a = par_format.align == AL_NO;
					struct par_attrib pa = par_format;
					if (ei->func == html_table && d_opt->tables && table_level < HTML_MAX_TABLE_LEVEL) {
						format_table(attr, html, eof, &html, f);
						ln_break(2, line_break, f);
						goto set_lt;
					}
					if (ei->func == html_select) {
						if (!do_html_select(attr, html, eof, &html, f))
							goto set_lt;
					}
					if (ei->func == html_textarea) {
						do_html_textarea(attr, html, eof, &html, f);
						goto set_lt;
					}
					if (ei->nopair == 2 || ei->nopair == 3) {
						struct html_element *e;
						if (ei->nopair == 2) {
							foreach(e, html_stack) {
								if (e->dontkill) break;
								if (e->linebreak || !ei->linebreak) break;
							}
						} else foreach(e, html_stack) {
							if (e->linebreak && !ei->linebreak) break;
							if (e->dontkill) break;
							if (e->namelen == namelen && !casecmp(e->name, name, e->namelen)) break;
						}
						if (e->namelen == namelen && !casecmp(e->name, name, e->namelen)) {
							while (e->prev != (void *)&html_stack) kill_html_stack_item(e->prev);
							if (e->dontkill != 2) kill_html_stack_item(e);
						}
					}
					if (ei->nopair != 1) {
						html_stack_dup();
						html_top.name = name;
						html_top.namelen = namelen;
						html_top.options = attr;
						html_top.linebreak = ei->linebreak;
					}
					if (ei->func) ei->func(attr);
					if (ei->func != html_br) was_br = 0;
					if (a) par_format = pa;
				}
			} else {
				struct html_element *e, *ff;
				int lnb = 0;
				int xxx = 0;
				was_br = 0;
				if (ei->nopair == 1 || ei->nopair == 3) break;
				/*debug_stack();*/
				foreach(e, html_stack) {
					if (e->linebreak && !ei->linebreak && ei->name) xxx = 1;
					if (e->namelen != namelen || casecmp(e->name, name, e->namelen)) {
						if (e->dontkill) break;
						else continue;
					}
					if (xxx) {
						kill_html_stack_item(e);
						break;
					}
					for (ff = e; ff != (void *)&html_stack; ff = ff->prev)
						if (ff->linebreak > lnb) lnb = ff->linebreak;
					ln_break(lnb, line_break, f);
					while (e->prev != (void *)&html_stack) kill_html_stack_item(e->prev);
					kill_html_stack_item(e);
					break;
				}
				/*debug_stack();*/
			}
			goto set_lt;
		}
		goto set_lt; /* unreachable */
	}
	put_chrs(lt, html - lt, put_chars, f);
	ln_break(1, line_break, f);
	putsp = -1;
	pos = 0;
	/*line_breax = 1;*/
	was_br = 0;
}

int get_image_map(unsigned char *head, unsigned char *s, unsigned char *eof, unsigned char *tag, struct menu_item **menu, struct memory_list **ml, unsigned char *href_base, unsigned char *target_base, int to, int def, int hdef)
{
	unsigned char *name, *attr, *al, *label, *href, *target;
	int namelen, lblen;
	struct link_def *ld;
	struct menu_item *nm;
	int nmenu = 0;
	int i;
	unsigned char *hd = init_str();
	int hdl = 0;
	struct conv_table *ct;
	if (head) add_to_str(&hd, &hdl, head);
	scan_http_equiv(s, eof, &hd, &hdl, NULL);
	ct = get_convert_table(hd, to, def, NULL, NULL, hdef);
	mem_free(hd);
	*menu = mem_alloc(sizeof(struct menu_item));
	memset(*menu, 0, sizeof(struct menu_item));
	/*(*menu)->text = NULL;*/
	se:
	while (s < eof && *s != '<') {
		sp:
		s++;
	}
	if (s >= eof) {
		mem_free(*menu);
		return -1;
	}
	if (s + 2 <= eof && (s[1] == '!' || s[1] == '?')) {
		s = skip_comment(s, eof);
		goto se;
	}
	if (parse_element(s, eof, &name, &namelen, &attr, &s)) goto sp;
	if (namelen != 3 || casecmp(name, "MAP", 3)) goto se;
	if (tag && *tag) {
		if (!(al = get_attr_val(attr, "name"))) goto se;
		if (strcasecmp(al, tag)) {
			mem_free(al);
			goto se;
		}
		mem_free(al);
	}
	*ml = getml(NULL);
	se2:
	while (s < eof && *s != '<') {
		sp2:
		s++;
	}
	if (s >= eof) {
		freeml(*ml);
		mem_free(*menu);
		return -1;
	}
	if (s + 2 <= eof && (s[1] == '!' || s[1] == '?')) {
		s = skip_comment(s, eof);
		goto se2;
	}
	if (parse_element(s, eof, &name, &namelen, &attr, &s)) goto sp2;
	if (namelen == 1 && !casecmp(name, "A", 1)) {
		unsigned char *ss;
		label = init_str();
		lblen = 0;
		se3:
		ss = s;
		while (ss < eof && *ss != '<') ss++;
		if (ss >= eof) {
			mem_free(label);
			freeml(*ml);
			mem_free(*menu);
			return -1;
		}
		add_bytes_to_str(&label, &lblen, s, ss - s);
		s = ss;
		if (s + 2 <= eof && (s[1] == '!' || s[1] == '?')) {
			s = skip_comment(s, eof);
			goto se3;
		}
		if (parse_element(s, eof, NULL, NULL, NULL, &ss)) goto se3;
		if (!((namelen == 1 && !casecmp(name, "A", 1)) ||
		      (namelen == 2 && !casecmp(name, "/A", 2)) ||
		      (namelen == 3 && !casecmp(name, "MAP", 3)) ||
		      (namelen == 4 && !casecmp(name, "/MAP", 4)) ||
		      (namelen == 4 && !casecmp(name, "AREA", 4)) ||
		      (namelen == 5 && !casecmp(name, "/AREA", 5)))) {
				s = ss;
				goto se3;
		}
	} else if (namelen == 4 && !casecmp(name, "AREA", 4)) {
		unsigned char *l = get_attr_val(attr, "alt");
		if (l) label = convert_string(ct, l, strlen(l)), mem_free(l);
		else label = NULL;
	} else if (namelen == 4 && !casecmp(nam