link.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

/* HTML parser */

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

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

#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

#include "elinks.h"

#include "bfu/listmenu.h"
#include "bfu/menu.h"
#include "bookmarks/bookmarks.h"
#include "config/options.h"
#include "config/kbdbind.h"
#include "document/html/frames.h"
#include "document/html/parser/link.h"
#include "document/html/parser/parse.h"
#include "document/html/parser/stack.h"
#include "document/html/parser.h"
#include "document/html/renderer.h"
#include "globhist/globhist.h"
#include "mime/mime.h"
#include "protocol/uri.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_a(struct html_context *html_context, unsigned char *a,
       unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	unsigned char *href;

	href = get_url_val(a, "href", html_context->options);
	if (href) {
		unsigned char *target;

		mem_free_set(&format.link,
			     join_urls(html_context->base_href,
				       trim_chars(href, ' ', 0)));

		mem_free(href);

		target = get_target(html_context->options, a);
		if (target) {
			mem_free_set(&format.target, target);
		} else {
			mem_free_set(&format.target, stracpy(html_context->base_target));
		}

		if (0) {
			; /* Shut up compiler */
#ifdef CONFIG_GLOBHIST
		} else if (get_global_history_item(format.link)) {
			format.style.fg = format.vlink;
			html_top.pseudo_class &= ~ELEMENT_LINK;
			html_top.pseudo_class |= ELEMENT_VISITED;
#endif
#ifdef CONFIG_BOOKMARKS
		} else if (get_bookmark(format.link)) {
			format.style.fg = format.bookmark_link;
			html_top.pseudo_class &= ~ELEMENT_VISITED;
			/* XXX: Really set ELEMENT_LINK? --pasky */
			html_top.pseudo_class |= ELEMENT_LINK;
#endif
		} else {
			format.style.fg = format.clink;
			html_top.pseudo_class &= ~ELEMENT_VISITED;
			html_top.pseudo_class |= ELEMENT_LINK;
		}

		mem_free_set(&format.title,
		             get_attr_val(a, "title", html_context->options));

		html_focusable(html_context, a);

	} else {
		kill_html_stack_item(html_context, &html_top);
	}

	set_fragment_identifier(html_context, a, "name");
}

/* Returns an allocated string made after @label
 * but limited to @max_len length, by truncating
 * the middle of @label string, which is replaced
 * by an asterisk ('*').
 * If @max_len < 0 it returns NULL.
 * If @max_len == 0 it returns an unmodified copy
 * of @label string.
 * In either case, it may return NULL if a memory
 * allocation failure occurs.
 * Example:
 * truncate_label("some_string", 5) => "so*ng" */
static unsigned char *
truncate_label(unsigned char *label, int max_len)
{
	unsigned char *new_label;
	int len = strlen(label);
	int left_part_len;
	int right_part_len;

	if (max_len < 0) return NULL;
	if (max_len == 0 || len <= max_len)
		return stracpy(label);

	right_part_len = left_part_len = max_len / 2;

	if (left_part_len + right_part_len + 1 > max_len)
		right_part_len--;

	new_label = mem_alloc(max_len + 1);
	if (!new_label) return NULL;

	if (left_part_len)
		memcpy(new_label, label, left_part_len);

	new_label[left_part_len] = '*';

	if (right_part_len)
		memcpy(new_label + left_part_len + 1,
		       label + len - right_part_len, right_part_len);

	new_label[max_len] = '\0';

	return new_label;
}

/* Get image filename from its src attribute. */
static unsigned char *
get_image_filename_from_src(int max_len, unsigned char *src)
{
	unsigned char *text = NULL;
	unsigned char *start, *filename;
	int len;

	if (!src) return NULL;
	/* We can display image as [foo.gif]. */

	len = strcspn(src, "?");

	for (start = src + len; start > src; start--)
		if (dir_sep(start[-1])) {
			break;
		}

	len -= start - src;

	filename = memacpy(start, len);
	if (filename) {
		/* XXX: Due to a compatibility alias (added: 2004-12-15 in
		 * 0.10pre3.CVS for document.browse.images.file_tags) this can
		 * return a negative @max_len. */
		text = truncate_label(filename, max_len);
		mem_free(filename);
	}

	return text;
}


/* Returns an allocated string containing formatted @label. */
static unsigned char *
get_image_label(int max_len, unsigned char *label)
{
	unsigned char *formatted_label;

	if (!label) return NULL;

	formatted_label = truncate_label(label, max_len);
	mem_free(label);

	return formatted_label;
}

static void
put_image_label(unsigned char *a, unsigned char *label,
                struct html_context *html_context)
{
	color_T fg;

	/* This is not 100% appropriate for <img>, but well, accepting
	 * accesskey and tabindex near <img> is just our little
	 * extension to the standard. After all, it makes sense. */
	html_focusable(html_context, a);

	fg = format.style.fg;
	format.style.fg = format.image_link;
	put_chrs(html_context, label, strlen(label));
	format.style.fg = fg;
}

static void
html_img_do(unsigned char *a, unsigned char *object_src,
            struct html_context *html_context)
{
	int ismap, usemap = 0;
	int add_brackets = 0;
	unsigned char *src = NULL;
	unsigned char *label = NULL;
	unsigned char *usemap_attr;
	struct document_options *options = html_context->options;
	int display_style = options->image_link.display_style;

	/* Note about display_style:
	 * 0     means always display IMG
	 * 1     means always display filename
	 * 2     means display alt/title attribute if possible, IMG if not
	 * 3     means display alt/title attribute if possible, filename if not */

	usemap_attr = get_attr_val(a, "usemap", options);
	if (usemap_attr) {
		unsigned char *joined_urls = join_urls(html_context->base_href,
						       usemap_attr);
		unsigned char *map_url;

		mem_free(usemap_attr);
		if (!joined_urls) return;
		map_url = straconcat("MAP@", joined_urls, NULL);
		mem_free(joined_urls);
		if (!map_url) return;

		html_stack_dup(html_context, ELEMENT_KILLABLE);
		mem_free_set(&format.link, map_url);
		format.form = NULL;
		format.style.attr |= AT_BOLD;
		usemap = 1;
 	}

	ismap = format.link
	        && has_attr(a, "ismap", options)
	        && !usemap;

	if (display_style == 2 || display_style == 3) {
		label = get_attr_val(a, "alt", options);
		if (!label)
			label = get_attr_val(a, "title", options);

		/* Little hack to preserve rendering of [   ], in directories listing,
		 * but we still want to drop extra spaces in alt or title attribute
		 * to limit display width on certain websites. --Zas */
		if (label && strlen(label) > 5) clr_spaces(label);
	}

	src = null_or_stracpy(object_src);
	if (!src) src = get_url_val(a, "src", options);
	if (!src) src = get_url_val(a, "dynsrc", options);

	/* If we have no label yet (no title or alt), so
	 * just use default ones, or image filename. */
	if (!label || !*label) {
		mem_free_set(&label, NULL);
		/* Do we want to display images with no alt/title and with no
		 * link on them ?
		 * If not, just exit now. */
		if (!options->images && !format.link) {
			mem_free_if(src);
			if (usemap) kill_html_stack_item(html_context, &html_top);
			return;
		}

		add_brackets = 1;

		if (usemap) {
			label = stracpy("USEMAP");
		} else if (ismap) {
			label = stracpy("ISMAP");
		} else {
			if (display_style == 3)
				label = get_image_filename_from_src(options->image_link.filename_maxlen, src);
		}

	} else {
		label = get_image_label(options->image_link.label_maxlen, label);
	}

	if (!label || !*label) {
		mem_free_set(&label, NULL);
		add_brackets = 1;
		if (display_style == 1)
			label = get_image_filename_from_src(options->image_link.filename_maxlen, src);
		if (!label || !*label)
			mem_free_set(&label, stracpy("IMG"));
	}

	mem_free_set(&format.image, NULL);
	mem_free_set(&format.title, NULL);

	if (label) {
		int img_link_tag = options->image_link.tagging;

		if (img_link_tag && (img_link_tag == 2 || add_brackets)) {
			unsigned char *img_link_prefix = options->image_link.prefix;
			unsigned char *img_link_suffix = options->image_link.suffix;
			unsigned char *new_label = straconcat(img_link_prefix, label, img_link_suffix, NULL);

			if (new_label) mem_free_set(&label, new_label);
		}

		if (!options->image_link.show_any_as_links) {
			put_image_label(a, label, html_context);

		} else {
			if (src) {
				format.image = join_urls(html_context->base_href, src);
			}

			format.title = get_attr_val(a, "title", options);

			if (ismap) {
				unsigned char *new_link;

				html_stack_dup(html_context, ELEMENT_KILLABLE);
				new_link = straconcat(format.link, "?0,0", NULL);
				if (new_link)
					mem_free_set(&format.link, new_link);
			}

			put_image_label(a, label, html_context);

			if (ismap) kill_html_stack_item(html_context, &html_top);
			mem_free_set(&format.image, NULL);
			mem_free_set(&format.title, NULL);
		}

		mem_free(label);
	}

	mem_free_if(src);
	if (usemap) kill_html_stack_item(html_context, &html_top);
}

void
html_img(struct html_context *html_context, unsigned char *a,
         unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	html_img_do(a, NULL, html_context);
}

void
put_link_line(unsigned char *prefix, unsigned char *linkname,
	      unsigned char *link, unsigned char *target,
	      struct html_context *html_context)
{
	html_context->has_link_lines = 1;
	html_stack_dup(html_context, ELEMENT_KILLABLE);
	ln_break(html_context, 1);
	mem_free_set(&format.link, NULL);
	mem_free_set(&format.target, NULL);
	mem_free_set(&format.title, NULL);
	format.form = NULL;
	put_chrs(html_context, prefix, strlen(prefix));
	format.link = join_urls(html_context->base_href, link);
	format.target = stracpy(target);
	format.style.fg = format.clink;
	put_chrs(html_context, linkname, strlen(linkname));
	ln_break(html_context, 1);
	kill_html_stack_item(html_context, &html_top);
}


void
html_applet(struct html_context *html_context, unsigned char *a,
            unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	unsigned char *code, *alt;

	code = get_url_val(a, "code", html_context->options);
	if (!code) return;

	alt = get_attr_val(a, "alt", html_context->options);

	html_focusable(html_context, a);

	if (alt && *alt) {
		put_link_line("Applet: ", alt, code,
			      html_context->options->framename, html_context);
	} else {
		put_link_line("", "Applet", code,
			      html_context->options->framename, html_context);
	}

	mem_free_if(alt);
	mem_free(code);
}

static void
html_iframe_do(unsigned char *a, unsigned char *object_src,
               struct html_context *html_context)
{
	unsigned char *name, *url = NULL;

	url = null_or_stracpy(object_src);
	if (!url) url = get_url_val(a, "src", html_context->options);
	if (!url) return;

	name = get_attr_val(a, "name", html_context->options);
	if (!name) name = get_attr_val(a, "id", html_context->options);
	if (!name) name = stracpy("");
	if (!name) {
		mem_free(url);
		return;
	}

	html_focusable(html_context, a);

	if (*name) {
		put_link_line("IFrame: ", name, url,
			      html_context->options->framename, html_context);
	} else {
		put_link_line("", "IFrame", url,
			      html_context->options->framename, html_context);
	}

	mem_free(name);
	mem_free(url);
}

void
html_iframe(struct html_context *html_context, unsigned char *a,
            unsigned char *xxx3, unsigned char *xxx4, unsigned char **xxx5)
{
	html_iframe_do(a, NULL, html_context);
}

void