/*============================================================================== HTML2XHTML Converter 1.0 ======================== Copyright (c) 2004, 2005 Vyacheslav Smolin Author: ------- Vyacheslav Smolin (http://www.richarea.com, http://html2xhtml.richarea.com, re@richarea.com) About the script: ----------------- HTML2XHTML Converter (H2X) generates a well formed XHTML string from a HTML DOM object. Requirements: ------------- H2X works in MS IE 5.0 for Windows or above, in Netscape 7.1, Mozilla 1.3 or above. It should work in all Mozilla based browsers. Usage: ------ Please see description of function get_xhtml below. Demo: ----- http://html2xhtml.richarea.com/, http://www.richarea.com/demo/ License: -------- Free for non-commercial using. Please contact author for commercial licenses. ==============================================================================*/ //add \n before opening tag var need_nl_before = '|div|p|table|tbody|tr|td|th|title|head|body|script|comment|li|meta|h1|h2|h3|h4|h5|h6|hr|ul|ol|option|'; //add \n after opening tag var need_nl_after = '|html|head|body|p|th|style|'; var re_comment = new RegExp(); re_comment.compile("^$"); var re_hyphen = new RegExp(); re_hyphen.compile("-$"); // Convert inner text of node to xhtml // Call: get_xhtml(node); // get_xhtml(node, lang, encoding) -- to convert whole page // other parameters are for inner usage and should be omitted // Parameters: // node - dom node to convert // lang - document lang (need it if whole page converted) // encoding - document charset (need it if whole page converted) // need_nl - if true, add \n before a tag if it is in list need_nl_before // inside_pre - if true, do not change content, as it is inside a

function get_xhtml(node, lang, encoding, need_nl, inside_pre) {

var i;
var text = '';
var children = node.childNodes;
var child_length = children.length;
var tag_name;
var do_nl = need_nl?true:false;
var page_mode = true;

	for (i=0;i tags parsed incorrectly by ie-dom
				//so take their code and lowercase names of tags and attributes
				if (document.all && tag_name == 'object') {
					text += fix_object_code(child.outerHTML);
					continue;
				}

				if (!need_nl && tag_name == 'body') { //html fragment mode
					page_mode = false;
				}

				if (tag_name == '!') { //COMMENT_NODE in IE 5.0/5.5
					//get comment inner text
					var parts = re_comment.exec(child.text);

					if (parts) {
						//the last char of the comment text must not be a hyphen
						var inner_text = parts[1];
						text += fix_comment(inner_text);
					}
				} else {
					if (tag_name == 'html'){
						text = '\n\n';
					}

					//inset \n to make code more neat
					if (need_nl_before.indexOf('|'+tag_name+'|') != -1) {
						if ((do_nl || text != '') && !inside_pre) text += '\n';
							else do_nl = true;
					}

					text += '<'+tag_name;

					//add attributes
					var attr = child.attributes;
					var attr_length = attr.length;
					var attr_value;

					var attr_lang = false;
					var attr_xml_lang = false;
					var attr_xmlns = false;

					var is_alt_attr = false;

					for (j=0;j';
					} else {

						//these tags must have closing tags
						//'a' included as otherwise Mozilla extends  links
						//on content coming after the link, that is wrong
						if (tag_name == 'style' || tag_name == 'title' ||
							tag_name == 'script' || tag_name == 'textarea' ||
							tag_name == 'a') {

							text += '>';
							var inner_text;
							if (tag_name == 'script') {
								inner_text = child.text;
							}else inner_text = child.innerHTML;

							if (tag_name == 'style') {
								inner_text = String(inner_text).replace(/[\n]+/g,'\n');
							}

							text += inner_text+'';

						} else {
							text += ' />';
						}
					}

				}
				break;
			}
			case 3: { //TEXT_NODE
				if (!inside_pre) { //do not change text inside  tag
					if (child.nodeValue != '\n') {
						text += fix_text(child.nodeValue);
					}
				} else text += child.nodeValue;
				break;
			}
			case 8: { //COMMENT_NODE
				text += fix_comment(child.nodeValue);
				break;
			}
			default:
				break;
		}
	}

	if (!need_nl && !page_mode) { //delete head and body tags from html fragment
			text = text.replace(/<\/?head>[\n]*/gi, "");
			text = text.replace(/[\n]*/gi, "");
			text = text.replace(/<\/?body>[\n]*/gi, "");
	}

	return text;
}

//fix inner text of a comment
function fix_comment(text){

	//delete double hyphens from the comment text
	text = text.replace(/--/g, "__");

	if(re_hyphen.exec(text)){ //last char must not be a hyphen
		text += " ";
	}

	return "";
}

//fix content of a text node
function fix_text(text) {
	//convert <,> and & to the corresponding entities

	//change < and > or the next string convert their & chars
	var temp_text = String(text).replace(/\</g, "#h2x_lt").replace(/\>/g, "#h2x_gt");
	temp_text = temp_text.replace(/\n{2,}/g, "\n").replace(/\&/g, "&").replace(//g, ">").replace(/\u00A0/g, " ");
	return temp_text.replace(/#h2x_lt/g, "<").replace(/#h2x_gt/g, ">");
}

//fix content of attributes href, src or background
function fix_attribute(text) {
	//convert <,>, & and " to the corresponding entities

	//change < and > or the next string convert their & chars
	var temp_text = String(text).replace(/\</g, "#h2x_lt").replace(/\>/g, "#h2x_gt");
	temp_text = temp_text.replace(/\&/g, "&").replace(//g, ">").replace(/\"/g, """);
	return temp_text.replace(/#h2x_lt/g, "<").replace(/#h2x_gt/g, ">");
}

//lowercase names of tags and its attributes - the best we can do for
//flash objects in IE
function fix_object_code(text) {
var temp = String(text);

	temp = temp.replace(/ style=/gi, ' style=');
	temp = temp.replace(/ codeBase=/gi, ' codebase=');
	temp = temp.replace(/ height=/gi, ' height=');
	temp = temp.replace(/ width=/gi, ' width=');
	temp = temp.replace(/ align=/gi, ' align=');
	temp = temp.replace(/ classid=/gi, ' classid=');
	temp = temp.replace(/ src=/gi, ' src=');
	temp = temp.replace(/ NAME=/gi, ' name=');
	temp = temp.replace(/ VALUE=/gi, ' value=');
	temp = temp.replace(/ quality=/gi, ' quality=');
	temp = temp.replace(/ TYPE=/gi, ' type=');
	temp = temp.replace(/ PLUGINSPAGE=/gi, ' pluginspage=');
	temp = temp.replace(/');
	temp = temp.replace(//gi, '');
	temp = temp.replace(//gi, '');

	return temp;
}