var prev_line_req = '|div|p|table|tbody|tr|td|th|title|head|body|script|comment|li|meta|h1|h2|h3|h4|h5|h6|hr|ul|ol|option|'; var next_line_req = '|html|head|body|p|th|style|'; var req_remark = new RegExp(); req_remark.compile("^$"); var hypen_req = new RegExp(); hypen_req.compile("-$"); function receive_xhtml(node, lang, encoding, need_nl, inside_pre) { var i; var txt = ''; var inherit = node.childNodes; var inherit_length = inherit.length; var hold_tag; var prnt_line = need_nl ? true : false; var page_mode = true; for (i = 0; i < inherit_length; i++) { var inh = inherit[i]; switch (inh.nodeType) { case 1: { var hold_tag = String(inh.tagName).toLowerCase(); if (hold_tag == '') break; if (hold_tag == 'meta') { var meta_name = String(inh.name).toLowerCase(); if (meta_name == 'generator') break; } if (!need_nl && hold_tag == 'body') { page_mode = false; } if (hold_tag == '!') { var splts = req_remark.exec(inh.text); if (splts) { var inerTxt = splts[1]; txt += set_comm(inerTxt); } } else { if (hold_tag == 'html') { txt = '\n\n'; } if (prev_line_req.indexOf('|'+hold_tag+'|') != -1) { if ((prnt_line || txt != '') && !inside_pre) txt += '\n'; } else { prnt_line = true; } txt += '<'+hold_tag; var xprop = inh.attributes; var xprop_length = xprop.length; var xprop_value; var xprop_lang = false; var xprop_xml_lang = false; var xprop_xmlns = false; var is_alt_xprop = false; for (j = 0; j < xprop_length; j++) { var xprop_name = xprop[j].nodeName.toLowerCase(); if (!xprop[j].specified && (xprop_name != 'selected' || !inh.selected) && (xprop_name != 'style' || inh.style.cssText == '') && xprop_name != 'value') continue; if (xprop_name == '_moz_dirty' || xprop_name == '_moz_resizing' || hold_tag == 'br' && xprop_name == 'type' && inh.getAttribute('type') == '_moz') continue; var chk_vald_xprop = true; switch (xprop_name) { case "style": xprop_value = inh.style.cssText; break; case "class": xprop_value = inh.className; break; case "http-equiv": xprop_value = inh.httpEquiv; break; case "noshade": break; case "checked": break; case "selected": break; case "multiple": break; case "nowrap": break; case "disabled": break; xprop_value = xprop_name; break; default: try { xprop_value = inh.getAttribute(xprop_name, 2); } catch (e) { chk_vald_xprop = false; } break; } if (xprop_name == 'lang') { xprop_lang = true; xprop_value = lang; } if (xprop_name == 'xml:lang') { xprop_xml_lang = true; xprop_value = lang; } if (xprop_name == 'xmlns') xprop_xmlns = true; if (chk_vald_xprop) { if (!(hold_tag == 'li' && xprop_name == 'value')) { txt += ' '+xprop_name+'="'+setAttr(xprop_value)+'"'; } } if (xprop_name == 'alt') is_alt_xprop = true; } if (hold_tag == 'img' && !is_alt_xprop) { txt += ' alt=""'; } if (hold_tag == 'html') { if (!xprop_lang) txt += ' lang="'+lang+'"'; if (!xprop_xml_lang) txt += ' xml:lang="'+lang+'"'; if (!xprop_xmlns) txt += ' xmlns="http://www.w3.org/1999/xhtml"'; } if (inh.canHaveChildren || inh.hasChildNodes()){ txt += '>'; txt += receive_xhtml(inh, lang, encoding, true, inside_pre || hold_tag == 'pre' ? true : false); txt += ''; } else { if (hold_tag == 'style' || hold_tag == 'title' || hold_tag == 'script') { txt += '>'; var inerTxt; if (hold_tag == 'script') { inerTxt = inh.text; } else { inerTxt = inh.innerHTML; } if (hold_tag == 'style') { inerTxt = String(inerTxt).replace(/[\n]+/g,'\n'); } txt += inerTxt+''; } else { txt += ' />'; } } } break; } case 3: { if (!inside_pre) { if (inh.nodeValue != '\n') { txt += setTxt(inh.nodeValue); } } else { txt += inh.nodeValue; } break; } case 8: { txt += set_comm(inh.nodeValue); break; } default: break; } } if (!need_nl && !page_mode) { txt = txt.replace(/<\/?head>[\n]*/gi, ""); txt = txt.replace(/[\n]*/gi, ""); txt = txt.replace(/<\/?body>[\n]*/gi, ""); } return txt; } function set_comm(txt) { txt = txt.replace(/--/g, "__"); if(hypen_req.exec(txt)) { txt += " "; } return ""; } function setTxt(txt) { return String(txt).replace(/\n{2,}/g, "\n").replace(/\&/g, "&").replace(//g, ">").replace(/\u00A0/g, " "); } function setAttr(txt) { return String(txt).replace(/\&/g, "&").replace(//g, ">").replace(/\"/g, """); }