Example #1
0
nuiHTMLAttrib::nuiHTMLAttrib(const void* _tattrib, nglTextEncoding encoding)
{
  TidyAttr tattr = (TidyAttr)_tattrib;
  mType = (AttributeType)tidyAttrGetId(tattr);
  mName.Import(tidyAttrName(tattr), encoding);
  mValue.Import(tidyAttrValue(tattr), encoding);
}
Example #2
0
File: parse.c Project: ASpade/mulk
static void parse_html(TidyDoc tdoc, TidyNode tnod, const url_list_t *elem, int indent, FILE *outfile)
{
	TidyNode child;
	TidyAttr attr;
	TidyAttrId attr_id = TidyAttr_UNKNOWN;
	TidyNodeType node_type;
	TidyTagId node_id;
	ctmbstr name;
	char *url, *relative_url = NULL;
	int found = 0;
	int get_html_link = (!option_values.depth || elem->level < option_values.depth);
	int get_int_html_link = (!option_values.depth || elem->level < option_values.depth+1);
	int get_ext_depends = ((!option_values.depth || elem->level < option_values.depth+1)
		&& !option_values.no_html_dependencies);

	for (child = tidyGetChild(tnod); child; child = tidyGetNext(child)) {
		node_type = tidyNodeGetType(child);

		switch (node_type) {
			case TidyNode_Start:
			case TidyNode_StartEnd:
				node_id = tidyNodeGetId(child);
				if (get_html_link && (node_id == TidyTag_A || node_id == TidyTag_AREA || node_id == TidyTag_MAP)) {
					found = 1;
					attr_id = TidyAttr_HREF;
				}
				else if (get_int_html_link && (node_id == TidyTag_FRAME || node_id == TidyTag_IFRAME)) {
					found = 1;
					attr_id = TidyAttr_SRC; 
				}
				else if (get_ext_depends) {
					if (node_id == TidyTag_LINK) {
						found = 1;
						attr_id = TidyAttr_HREF;
					}
					else if (node_id == TidyTag_IMG || node_id == TidyTag_SCRIPT) {
						found = 1;
						attr_id = TidyAttr_SRC; 
					}
					else {
						found = 0;
						attr_id = TidyAttr_UNKNOWN;
					}
				}
				else {
					found = 0;
					attr_id = TidyAttr_UNKNOWN;
				}

				if (found && (attr = tidyAttrGetById(child, attr_id)) != NULL) {
					url = (char *) tidyAttrValue(attr);

					string_free(relative_url);
					if (url && *url)
						add_new_url_and_check(elem, url, outfile ? &relative_url : NULL);
				}

				if (outfile && (name = tidyNodeGetName(child)) != NULL) {
					fprintf(outfile, "%*.*s%s", indent, indent, "<", name);
					for (attr = tidyAttrFirst(child); attr; attr = tidyAttrNext(attr)) {
						fprintf(outfile, " %s", tidyAttrName(attr));
						if (relative_url && (tidyAttrGetId(attr) == attr_id))
							fprintf(outfile, "=\"%s\"", relative_url);
						else if (tidyAttrValue(attr))
							fprintf(outfile, "=\"%s\"", tidyAttrValue(attr) ? tidyAttrValue(attr) : "");
						else
							fprintf(outfile, "=\"\"");
					}
					string_free(relative_url);

					if (node_type == TidyNode_StartEnd)
						fprintf(outfile, "/>\n");
					else {
						fprintf(outfile, ">\n");
						parse_html(tdoc, child, elem, indent + 1, outfile);
						fprintf(outfile, "%*.*s%s>\n", indent + 1, indent + 1, "</", name);
					}
				}
				else {
					string_free(relative_url);
					parse_html(tdoc, child, elem, indent + 1, outfile);
				}
				break;
			case TidyNode_End:
				if (outfile) {
					if ((name = tidyNodeGetName(child)) != NULL)
						fprintf(outfile, "%*.*s/%s>\n", indent, indent, "<", name);
				}
				break;
			case TidyNode_Text:
				if (outfile) {
					TidyBuffer buf;
					TidyTagId parent_node_id = tidyNodeGetId(tnod);

					tidyBufInit(&buf);
					if (parent_node_id == TidyTag_SCRIPT || parent_node_id == TidyTag_STYLE)
						tidyNodeGetValue(tdoc, child, &buf);
					else
						tidyNodeGetText(tdoc, child, &buf);
					if (buf.bp)
						fprintf(outfile, "%s", (char *)buf.bp);
					tidyBufFree(&buf);
				}
				break;
			case TidyNode_Comment:
				if (outfile) {
					TidyBuffer buf;

					tidyBufInit(&buf);
					tidyNodeGetValue(tdoc, child, &buf);
					if (buf.bp)
						fprintf(outfile, "<!--%s-->\n", (char *)buf.bp);
					tidyBufFree(&buf);
				}
				break;
			case TidyNode_CDATA:
				if (outfile) {
					TidyBuffer buf;

					tidyBufInit(&buf);
					tidyNodeGetValue(tdoc, child, &buf);
					if (buf.bp)
						fprintf(outfile, "<![CDATA[%s]]>\n", (char *)buf.bp);
					tidyBufFree(&buf);
				}
				break;
			case TidyNode_DocType:
				if (outfile) {
					int pub = 0;

					fprintf(outfile, "<!DOCTYPE %s", tidyNodeGetName(child));
					for (attr = tidyAttrFirst(child); attr; attr = tidyAttrNext(attr)) {
						if (!pub) {
							fprintf(outfile, " %s", tidyAttrName(attr));
							if (!string_casecmp(tidyAttrName(attr), "PUBLIC"))
								pub = 1;
						}
						if (tidyAttrValue(attr))
							fprintf(outfile, " \"%s\"", tidyAttrValue(attr));
					}
					fprintf(outfile, ">\n");
				}
				break;
			default:
				if (outfile) {
					TidyBuffer buf;

					tidyBufInit(&buf);
					tidyNodeGetValue(tdoc, child, &buf);
					if (buf.bp)
						fprintf(outfile, "%s", (char *)buf.bp);
					tidyBufFree(&buf);
				}
				break;
		}
	}
}