renderer.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

}


/* DOM Source Renderer */

#define check_dom_node_source(renderer, str, len)	\
	((renderer)->source <= (str) && (str) + (len) <= (renderer)->end)

#define assert_source(renderer, str, len) \
	assertm(check_dom_node_source(renderer, str, len), "renderer[%p : %p] str[%p : %p]", \
		(renderer)->source, (renderer)->end, (str), (str) + (len))

static inline void
render_dom_flush(struct dom_renderer *renderer, unsigned char *string)
{
	struct screen_char *template = &renderer->styles[DOM_NODE_TEXT];
	int length = string - renderer->position;

	assert_source(renderer, renderer->position, 0);
	assert_source(renderer, string, 0);

	if (length <= 0) return;
	render_dom_text(renderer, template, renderer->position, length);
	renderer->position = string;

	assert_source(renderer, renderer->position, 0);
}

static inline void
render_dom_node_text(struct dom_renderer *renderer, struct screen_char *template,
		     struct dom_node *node)
{
	unsigned char *string = node->string.string;
	int length = node->string.length;

	if (node->type == DOM_NODE_ENTITY_REFERENCE) {
		string -= 1;
		length += 2;
	}

	if (check_dom_node_source(renderer, string, length)) {
		render_dom_flush(renderer, string);
		renderer->position = string + length;
		assert_source(renderer, renderer->position, 0);
	}

	render_dom_text(renderer, template, string, length);
}

#ifdef HAVE_REGEX_H
static inline void
render_dom_node_enhanced_text(struct dom_renderer *renderer, struct dom_node *node)
{
	regex_t *regex = &renderer->url_regex;
	regmatch_t regmatch;
	unsigned char *string = node->string.string;
	int length = node->string.length;
	struct screen_char *template = &renderer->styles[node->type];
	unsigned char *alloc_string;

	if (check_dom_node_source(renderer, string, length)) {
		render_dom_flush(renderer, string);
		renderer->position = string + length;
		assert_source(renderer, renderer->position, 0);
	}

	alloc_string = memacpy(string, length);
	if (alloc_string)
		string = alloc_string;

	while (length > 0 && !regexec(regex, string, 1, &regmatch, 0)) {
		int matchlen = regmatch.rm_eo - regmatch.rm_so;
		int offset = regmatch.rm_so;

		if (!matchlen || offset < 0 || regmatch.rm_eo > length)
			break;

		if (offset > 0)
			render_dom_text(renderer, template, string, offset);

		string += offset;
		length -= offset;

		add_dom_link(renderer, string, matchlen);

		length -= matchlen;
		string += matchlen;
	}

	if (length > 0)
		render_dom_text(renderer, template, string, length);

	mem_free_if(alloc_string);
}
#endif

static void
render_dom_node_source(struct dom_stack *stack, struct dom_node *node, void *data)
{
	struct dom_renderer *renderer = stack->current->data;

	assert(node && renderer && renderer->document);

#ifdef HAVE_REGEX_H
	if (renderer->find_url
	    && (node->type == DOM_NODE_TEXT
		|| node->type == DOM_NODE_CDATA_SECTION
		|| node->type == DOM_NODE_COMMENT)) {
		render_dom_node_enhanced_text(renderer, node);
		return;
	}
#endif

	render_dom_node_text(renderer, &renderer->styles[node->type], node);
}

/* This callback is also used for rendering processing instruction nodes.  */
static void
render_dom_element_source(struct dom_stack *stack, struct dom_node *node, void *data)
{
	struct dom_renderer *renderer = stack->current->data;

	assert(node && renderer && renderer->document);

	render_dom_node_text(renderer, &renderer->styles[node->type], node);
}

static void
render_dom_element_end_source(struct dom_stack *stack, struct dom_node *node, void *data)
{
	struct dom_renderer *renderer = stack->current->data;
	struct dom_stack_state *state = get_dom_stack_top(stack);
	struct sgml_parser_state *pstate = get_dom_stack_state_data(stack->contexts[0], state);
	struct dom_scanner_token *token = &pstate->end_token;
	unsigned char *string = token->string.string;
	int length = token->string.length;

	assert(node && renderer && renderer->document);

	if (!string || !length)
		return;

	if (check_dom_node_source(renderer, string, length)) {
		render_dom_flush(renderer, string);
		renderer->position = string + length;
		assert_source(renderer, renderer->position, 0);
	}

	render_dom_text(renderer, &renderer->styles[node->type], string, length);
}

static void
set_base_uri(struct dom_renderer *renderer, unsigned char *value, size_t valuelen)
{
	unsigned char *href = memacpy(value, valuelen);
	unsigned char *uristring;
	struct uri *uri;

	if (!href) return;
	uristring = join_urls(renderer->base_uri, href);
	mem_free(href);

	if (!uristring) return;
	uri = get_uri(uristring, 0);
	mem_free(uristring);

	if (!uri) return;

	done_uri(renderer->base_uri);
	renderer->base_uri = uri;
}

static void
render_dom_attribute_source(struct dom_stack *stack, struct dom_node *node, void *data)
{
	struct dom_renderer *renderer = stack->current->data;
	struct screen_char *template = &renderer->styles[node->type];

	assert(node && renderer->document);

	render_dom_node_text(renderer, template, node);

	if (is_dom_string_set(&node->data.attribute.value)) {
		int quoted = node->data.attribute.quoted == 1;
		unsigned char *value = node->data.attribute.value.string - quoted;
		int valuelen = node->data.attribute.value.length + quoted * 2;

		if (check_dom_node_source(renderer, value, 0)) {
			render_dom_flush(renderer, value);
			renderer->position = value + valuelen;
			assert_source(renderer, renderer->position, 0);
		}

		if (node->data.attribute.reference
		    && valuelen - quoted * 2 > 0) {
			int skips;

			/* Need to flush the first quoting delimiter and any
			 * leading whitespace so that the renderers x position
			 * is at the start of the value string. */
			for (skips = 0; skips < valuelen; skips++) {
				if ((quoted && skips == 0)
				    || isspace(value[skips])
				    || value[skips] < ' ')
					continue;

				break;
			}

			if (skips > 0) {
				render_dom_text(renderer, template, value, skips);
				value    += skips;
				valuelen -= skips;
			}

			/* Figure out what should be skipped after the actual
			 * link text. */
			for (skips = 0; skips < valuelen; skips++) {
				if ((quoted && skips == 0)
				    || isspace(value[valuelen - skips - 1])
				    || value[valuelen - skips - 1] < ' ')
					continue;

				break;
			}

			if (renderer->doctype == SGML_DOCTYPE_HTML
			    && node->data.attribute.type == HTML_ATTRIBUTE_HREF
			    && node->parent->data.element.type == HTML_ELEMENT_BASE) {
				set_base_uri(renderer, value, valuelen - skips);
			}

			add_dom_link(renderer, value, valuelen - skips);

			if (skips > 0) {
				value += valuelen - skips;
				render_dom_text(renderer, template, value, skips);
			}
		} else {
			render_dom_text(renderer, template, value, valuelen);
		}
	}
}

static void
render_dom_cdata_source(struct dom_stack *stack, struct dom_node *node, void *data)
{
	struct dom_renderer *renderer = stack->current->data;
	unsigned char *string = node->string.string;

	assert(node && renderer && renderer->document);

	/* Highlight the 'CDATA' part of <![CDATA[ if it is there. */
	if (check_dom_node_source(renderer, string - 6, 6)) {
		render_dom_flush(renderer, string - 6);
		render_dom_text(renderer, &renderer->styles[DOM_NODE_ATTRIBUTE], string - 6, 5);
		renderer->position = string - 1;
		assert_source(renderer, renderer->position, 0);
	}

	render_dom_node_text(renderer, &renderer->styles[node->type], node);
}

static void
render_dom_document_end(struct dom_stack *stack, struct dom_node *node, void *data)
{
	struct dom_renderer *renderer = stack->current->data;

	/* If there are no non-element nodes after the last element node make
	 * sure that we flush to the end of the cache entry source including
	 * the '>' of the last element tag if it has one. (bug 519) */
	if (check_dom_node_source(renderer, renderer->position, 0)) {
		render_dom_flush(renderer, renderer->end);
	}
}

static struct dom_stack_context_info dom_source_renderer_context_info = {
	/* Object size: */			0,
	/* Push: */
	{
		/*				*/ NULL,
		/* DOM_NODE_ELEMENT		*/ render_dom_element_source,
		/* DOM_NODE_ATTRIBUTE		*/ render_dom_attribute_source,
		/* DOM_NODE_TEXT		*/ render_dom_node_source,
		/* DOM_NODE_CDATA_SECTION	*/ render_dom_cdata_source,
		/* DOM_NODE_ENTITY_REFERENCE	*/ render_dom_node_source,
		/* DOM_NODE_ENTITY		*/ render_dom_node_source,
		/* DOM_NODE_PROC_INSTRUCTION	*/ render_dom_element_source,
		/* DOM_NODE_COMMENT		*/ render_dom_node_source,
		/* DOM_NODE_DOCUMENT		*/ NULL,
		/* DOM_NODE_DOCUMENT_TYPE	*/ render_dom_node_source,
		/* DOM_NODE_DOCUMENT_FRAGMENT	*/ render_dom_node_source,
		/* DOM_NODE_NOTATION		*/ render_dom_node_source,
	},
	/* Pop: */
	{
		/*				*/ NULL,
		/* DOM_NODE_ELEMENT		*/ render_dom_element_end_source,
		/* DOM_NODE_ATTRIBUTE		*/ NULL,
		/* DOM_NODE_TEXT		*/ NULL,
		/* DOM_NODE_CDATA_SECTION	*/ NULL,
		/* DOM_NODE_ENTITY_REFERENCE	*/ NULL,
		/* DOM_NODE_ENTITY		*/ NULL,
		/* DOM_NODE_PROC_INSTRUCTION	*/ NULL,
		/* DOM_NODE_COMMENT		*/ NULL,
		/* DOM_NODE_DOCUMENT		*/ render_dom_document_end,
		/* DOM_NODE_DOCUMENT_TYPE	*/ NULL,
		/* DOM_NODE_DOCUMENT_FRAGMENT	*/ NULL,
		/* DOM_NODE_NOTATION		*/ NULL,
	}
};


/* Shared multiplexor between renderers */
void
render_dom_document(struct cache_entry *cached, struct document *document,
		    struct string *buffer)
{
	unsigned char *head = empty_string_or_(cached->head);
	struct dom_node *root;
	struct dom_renderer renderer;
	struct conv_table *convert_table;
	struct sgml_parser *parser;
	unsigned char *string = struri(cached->uri);
	size_t length = strlen(string);
	struct dom_string uri = INIT_DOM_STRING(string, length);
	struct dom_string source = INIT_DOM_STRING(buffer->source, buffer->length);

	assert(document->options.plain);

	convert_table = get_convert_table(head, document->options.cp,
					  document->options.assume_cp,
					  &document->cp,
					  &document->cp_status,
					  document->options.hard_assume);

	init_dom_renderer(&renderer, document, buffer, convert_table);

	document->bgcolor = document->options.default_bg;

	if (!strcasecmp("application/rss+xml", cached->content_type)) {
		renderer.doctype = SGML_DOCTYPE_RSS;

	} else if (!strcasecmp("application/xbel+xml", cached->content_type)
		   || !strcasecmp("application/x-xbel", cached->content_type)
		   || !strcasecmp("application/xbel", cached->content_type)) {
		renderer.doctype = SGML_DOCTYPE_XBEL;

	} else {
		assertm(!strcasecmp("text/html", cached->content_type)
			|| !strcasecmp("application/xhtml+xml", cached->content_type),
			"Couldn't resolve doctype '%s'", cached->content_type);

		renderer.doctype = SGML_DOCTYPE_HTML;
	}

	parser = init_sgml_parser(SGML_PARSER_STREAM, renderer.doctype, &uri);
	if (!parser) return;

	add_dom_stack_context(&parser->stack, &renderer,
			      &dom_source_renderer_context_info);

	root = parse_sgml(parser, &source);
	if (root) {
		assert(parser->stack.depth == 1);

		get_dom_stack_top(&parser->stack)->immutable = 0;
		/* For SGML_PARSER_STREAM this will free the DOM
		 * root node. */
		pop_dom_node(&parser->stack);
	}

#ifdef HAVE_REGEX_H
	if (renderer.find_url)
		regfree(&renderer.url_regex);
#endif
	done_uri(renderer.base_uri);
	done_sgml_parser(parser);
}