general.c

elinks下lynx是最重要的二个文本浏览器, 在linux下非常实用, elinks也是gentoo安装过程中默认使用的浏览器, 这是elinks源代码

/* General element handlers */

#ifdef HAVE_CONFIG_H
#include "config.h"
#endif

#ifndef _GNU_SOURCE
#define _GNU_SOURCE /* strcasestr() */
#endif

#include <ctype.h>
#include <errno.h>
#include <stdlib.h>
#include <string.h>

#include "elinks.h"

#include "config/options.h"
#include "document/css/apply.h"
#include "document/html/frames.h"
#include "document/html/parser/general.h"
#include "document/html/parser/link.h"
#include "document/html/parser/stack.h"
#include "document/html/parser/parse.h"
#include "document/html/parser.h"
#include "document/html/renderer.h"
#include "document/html/tables.h"
#include "document/options.h"
#include "intl/charsets.h"
#include "protocol/uri.h"
#include "terminal/draw.h"
#include "util/align.h"
#include "util/box.h"
#include "util/color.h"
#include "util/conv.h"
#include "util/error.h"
#include "util/memdebug.h"
#include "util/memory.h"
#include "util/string.h"

/* Unsafe macros */
#include "document/html/internal.h"


void
html_span(struct html_context *html_context, unsigned char *a,
          unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
}

void
html_bold(struct html_context *html_context, unsigned char *a,
          unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	format.style.attr |= AT_BOLD;
}

void
html_italic(struct html_context *html_context, unsigned char *a,
            unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	format.style.attr |= AT_ITALIC;
}

void
html_underline(struct html_context *html_context, unsigned char *a,
               unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	format.style.attr |= AT_UNDERLINE;
}

void
html_fixed(struct html_context *html_context, unsigned char *a,
           unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	format.style.attr |= AT_FIXED;
}

void
html_subscript(struct html_context *html_context, unsigned char *a,
               unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	format.style.attr |= AT_SUBSCRIPT | AT_UPDATE_SUB;
}

void
html_superscript(struct html_context *html_context, unsigned char *a,
                 unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	format.style.attr |= AT_SUPERSCRIPT | AT_UPDATE_SUP;
}

void
html_font(struct html_context *html_context, unsigned char *a,
          unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	unsigned char *al = get_attr_val(a, "size", html_context->options);

	if (al) {
		int p = 0;
		unsigned s;
		unsigned char *nn = al;
		unsigned char *end;

		if (*al == '+') p = 1, nn++;
		else if (*al == '-') p = -1, nn++;

		errno = 0;
		s = strtoul(nn, (char **) &end, 10);
		if (!errno && *nn && !*end) {
			if (s > 7) s = 7;
			if (!p) format.fontsize = s;
			else format.fontsize += p * s;
			if (format.fontsize < 1) format.fontsize = 1;
			else if (format.fontsize > 7) format.fontsize = 7;
		}
		mem_free(al);
	}
	get_color(html_context, a, "color", &format.style.fg);
}

void
html_body(struct html_context *html_context, unsigned char *a,
          unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	get_color(html_context, a, "text", &format.style.fg);
	get_color(html_context, a, "link", &format.clink);
	get_color(html_context, a, "vlink", &format.vlink);

	if (get_bgcolor(html_context, a, &format.style.bg) != -1)
		html_context->was_body_background = 1;

	html_context->was_body = 1; /* this will be used by "meta inside body" */
	html_apply_canvas_bgcolor(html_context);
}

void
html_apply_canvas_bgcolor(struct html_context *html_context)
{
#ifdef CONFIG_CSS
	/* If there are any CSS twaks regarding bgcolor, make sure we will get
	 * it _and_ prefer it over bgcolor attribute. */
	if (html_context->options->css_enable)
		css_apply(html_context, &html_top, &html_context->css_styles,
		          &html_context->stack);
#endif

	if (par_format.bgcolor != format.style.bg) {
		/* Modify the root HTML element - format_html_part() will take
		 * this from there. */
		struct html_element *e = html_context->stack.prev;

		html_context->was_body_background = 1;
		e->parattr.bgcolor = e->attr.style.bg = par_format.bgcolor = format.style.bg;
	}

	if (html_context->has_link_lines
	    && par_format.bgcolor != html_context->options->default_bg
	    && !search_html_stack(html_context, "BODY")) {
		html_context->special_f(html_context, SP_COLOR_LINK_LINES);
	}
}

void
html_script(struct html_context *html_context, unsigned char *a,
            unsigned char *html, unsigned char *eof, unsigned char **end)
{
#ifdef CONFIG_ECMASCRIPT
	/* TODO: <noscript> processing. Well, same considerations apply as to
	 * CSS property display: none processing. */
	/* TODO: Charsets for external scripts. */
	unsigned char *type, *language, *src;
	int in_comment = 0;
#endif

	html_skip(html_context, a);

#ifdef CONFIG_ECMASCRIPT
	/* We try to process nested <script> if we didn't process the parent
	 * one. That's why's all the fuzz. */
	/* Ref:
	 * http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
	 */
	type = get_attr_val(a, "type", html_context->options);
	if (type) {
		unsigned char *pos = type;

		if (!strncasecmp(type, "text/", 5)) {
			pos += 5;

		} else if (!strncasecmp(type, "application/", 12)) {
			pos += 12;

		} else {
			mem_free(type);
not_processed:
			/* Permit nested scripts and retreat. */
			html_top.invisible++;
			return;
		}

		if (!strncasecmp(pos, "javascript", 10)) {
			int len = strlen(pos);

			if (len > 10 && !isdigit(pos[10])) {
				mem_free(type);
				goto not_processed;
			}

		} else if (strcasecmp(pos, "ecmascript")
		    && strcasecmp(pos, "jscript")
		    && strcasecmp(pos, "livescript")
		    && strcasecmp(pos, "x-javascript")
		    && strcasecmp(pos, "x-ecmascript")) {
			mem_free(type);
			goto not_processed;
		}

		mem_free(type);
	}

	/* Check that the script content is ecmascript. The value of the
	 * language attribute can be JavaScript with optional version digits
	 * postfixed (like: ``JavaScript1.1'').
	 * That attribute is deprecated in favor of type by HTML 4.01 */
	language = get_attr_val(a, "language", html_context->options);
	if (language) {
		int languagelen = strlen(language);

		if (languagelen < 10
		    || (languagelen > 10 && !isdigit(language[10]))
		    || strncasecmp(language, "javascript", 10)) {
			mem_free(language);
			goto not_processed;
		}

		mem_free(language);
	}

	if (html_context->part->document
	    && (src = get_attr_val(a, "src", html_context->options))) {
		/* External reference. */

		unsigned char *import_url;
		struct uri *uri;

		if (!get_opt_bool("ecmascript.enable")) {
			mem_free(src);
			goto not_processed;
		}

		/* HTML <head> urls should already be fine but we can.t detect them. */
		import_url = join_urls(html_context->base_href, src);
		mem_free(src);
		if (!import_url) goto imported;

		uri = get_uri(import_url, URI_BASE);
		if (!uri) goto imported;

		/* Request the imported script as part of the document ... */
		html_context->special_f(html_context, SP_SCRIPT, uri);
		done_uri(uri);

		/* Create URL reference onload snippet. */
		insert_in_string(&import_url, 0, "^", 1);
		add_to_string_list(&html_context->part->document->onload_snippets,
		                   import_url, -1);

imported:
		/* Retreat. Do not permit nested scripts, tho'. */
		if (import_url) mem_free(import_url);
		return;
	}

	/* Positive, grab the rest and interpret it. */

	/* First position to the real script start. */
	while (html < eof && *html <= ' ') html++;
	if (eof - html > 4 && !strncmp(html, "<!--", 4)) {
		in_comment = 1;
		/* We either skip to the end of line or to -->. */
		for (; *html != '\n' && *html != '\r' && eof - html >= 3; html++) {
			if (!strncmp(html, "-->", 3)) {
				/* This means the document is probably broken.
				 * We will now try to process the rest of
				 * <script> contents, which is however likely
				 * to be empty. Should we try to process the
				 * comment too? Currently it seems safer but
				 * less tolerant to broken pages, if there are
				 * any like this. */
				html += 3;
				in_comment = 0;
				break;
			}
		}
	}

	*end = html;

	/* Now look ahead for the script end. The <script> contents is raw
	 * CDATA, so we just look for the ending tag and need not care for
	 * any quote marks counting etc - YET, we are more tolerant and permit
	 * </script> stuff inside of the script if the whole <script> element
	 * contents is wrapped in a comment. See i.e. Mozilla bug 26857 for fun
	 * reading regarding this. */
	for (; *end < eof; (*end)++) {
		unsigned char *name;
		int namelen;

		if (in_comment) {
			/* TODO: If we ever get some standards-quirk mode
			 * distinction, this should be disabled in the
			 * standards mode (and we should just look for CDATA
			 * end, which is "</"). --pasky */
			if (eof - *end >= 3 && !strncmp(*end, "-->", 3)) {
				/* Next iteration will jump passed the ending '>' */
				(*end) += 2;
				in_comment = 0;
			}
			continue;
			/* XXX: Scan for another comment? That's admittelly
			 * already stretching things a little bit to an
			 * extreme ;-). */
		}

		if (**end != '<')
			continue;
		/* We want to land before the closing element, that's why we
		 * don't pass @end also as the appropriate parse_element()
		 * 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;
	}

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

void
html_style(struct html_context *html_context, unsigned char *a,
           unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	html_skip(html_context, a);
}

void
html_html(struct html_context *html_context, unsigned char *a,
          unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	/* 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(struct html_context *html_context, unsigned char *a,
          unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	/* This makes sure it gets to the stack and helps tame down unclosed
	 * <title>. */
}

void
html_meta(struct html_context *html_context, unsigned char *a,
          unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	/* html_handle_body_meta() does all the work. */
}

/* Handles meta tags in the HTML body. */
void
html_handle_body_meta(struct html_context *html_context, unsigned char *meta,
		      unsigned char *eof)
{
	struct string head;

	if (!init_string(&head)) return;

	scan_http_equiv(meta, eof, &head, NULL, html_context->options);
	process_head(html_context, head.source);
	done_string(&head);
}

void
html_title(struct html_context *html_context, unsigned char *a,
           unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	html_top.invisible = 1;
	html_top.type = ELEMENT_WEAK;
}

void
html_center(struct html_context *html_context, unsigned char *a,
            unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	par_format.align = ALIGN_CENTER;
	if (!html_context->table_level)
		par_format.leftmargin = par_format.rightmargin = 0;
}

void
html_linebrk(struct html_context *html_context, unsigned char *a,
             unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	unsigned char *al = get_attr_val(a, "align", html_context->options);

	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(struct html_context *html_context, unsigned char *a,
        unsigned char *html, unsigned char *eof, unsigned char **end)
{
	html_linebrk(html_context, a, html, eof, end);
	if (html_context->was_br)
		ln_break(html_context, 2);
	else
		html_context->was_br = 1;
}

void
html_p(struct html_context *html_context, unsigned char *a,
       unsigned char *html, unsigned char *eof, unsigned char **end)
{
	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(html_context, a, html, eof, end);
}

void
html_address(struct html_context *html_context, unsigned char *a,
             unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	par_format.leftmargin++;
	par_format.align = ALIGN_LEFT;
}

void
html_blockquote(struct html_context *html_context, unsigned char *a,
                unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	par_format.leftmargin += 2;
	par_format.align = ALIGN_LEFT;
}

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

	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(struct html_context *html_context, unsigned char *a,
	unsigned char *html, unsigned char *eof, unsigned char **end)
{
	format.style.attr |= AT_BOLD;
	html_h(1, a, ALIGN_CENTER, html_context, html, eof, end);
}

void
html_h2(struct html_context *html_context, unsigned char *a,
        unsigned char *html, unsigned char *eof, unsigned char **end)
{
	html_h(2, a, ALIGN_LEFT, html_context, html, eof, end);
}

void
html_h3(struct html_context *html_context, unsigned char *a,
        unsigned char *html, unsigned char *eof, unsigned char **end)
{
	html_h(3, a, ALIGN_LEFT, html_context, html, eof, end);
}