2 * @class Roo.htmleditor.FilterWord
3 * try and clean up all the mess that Word generates.
5 * This is the 'nice version' - see 'Heavy' that white lists a very short list of elements, and multi-filters
8 * Run a new Span Filter
9 * @param {Object} config Configuration options
12 Roo.htmleditor.FilterWord = function(cfg)
14 // no need to apply config.
15 this.replaceDocBullets(cfg.node);
17 // this.walk(cfg.node);
22 Roo.extend(Roo.htmleditor.FilterWord, Roo.htmleditor.Filter,
28 * Clean up MS wordisms...
30 replaceTag : function(node)
33 // no idea what this does - span with text, replaceds with just text.
35 node.nodeName == 'SPAN' &&
36 !node.hasAttributes() &&
37 node.childNodes.length == 1 &&
38 node.firstChild.nodeName == "#text"
40 var textNode = node.firstChild;
41 node.removeChild(textNode);
42 if (node.getAttribute('lang') != 'zh-CN') { // do not space pad on chinese characters..
43 node.parentNode.insertBefore(node.ownerDocument.createTextNode(" "), node);
45 node.parentNode.insertBefore(textNode, node);
46 if (node.getAttribute('lang') != 'zh-CN') { // do not space pad on chinese characters..
47 node.parentNode.insertBefore(node.ownerDocument.createTextNode(" ") , node);
50 node.parentNode.removeChild(node);
51 return false; // dont do chidren - we have remove our node - so no need to do chdhilren?
56 if (node.tagName.toLowerCase().match(/^(style|script|applet|embed|noframes|noscript)$/)) {
57 node.parentNode.removeChild(node);
58 return false; // dont do chidlren
60 //Roo.log(node.tagName);
61 // remove - but keep children..
62 if (node.tagName.toLowerCase().match(/^(meta|link|\\?xml:|st1:|o:|v:|font)/)) {
63 //Roo.log('-- removed');
64 while (node.childNodes.length) {
65 var cn = node.childNodes[0];
67 node.parentNode.insertBefore(cn, node);
68 // move node to parent - and clean it..
69 if (cn.nodeType == 1) {
74 node.parentNode.removeChild(node);
75 /// no need to iterate chidlren = it's got none..
76 //this.iterateChildren(node, this.cleanWord);
77 return false; // no need to iterate children.
80 if (node.className.length) {
82 var cn = node.className.split(/\W+/);
84 Roo.each(cn, function(cls) {
85 if (cls.match(/Mso[a-zA-Z]+/)) {
90 node.className = cna.length ? cna.join(' ') : '';
92 node.removeAttribute("class");
96 if (node.hasAttribute("lang")) {
97 node.removeAttribute("lang");
100 if (node.hasAttribute("style")) {
102 var styles = node.getAttribute("style").split(";");
104 Roo.each(styles, function(s) {
108 var kv = s.split(":");
109 if (kv[0].match(/^(mso-|line|font|background|margin|padding|color)/)) {
112 // what ever is left... we allow.
115 node.setAttribute("style", nstyle.length ? nstyle.join(';') : '');
116 if (!nstyle.length) {
117 node.removeAttribute('style');
120 return true; // do children
126 styleToObject: function(node)
128 var styles = (node.getAttribute("style") || '').split(";");
130 Roo.each(styles, function(s) {
134 var kv = s.split(":");
136 // what ever is left... we allow.
143 replaceDocBullets : function(doc)
145 // this is a bit odd - but it appears some indents use ql-indent-1
147 var listpara = doc.getElementsByClassName('ql-indent-1');
148 while(listpara.length) {
149 this.replaceDocBullet(listpara.item(0));
152 var listpara = doc.getElementsByClassName('MsoListParagraph');
153 while(listpara.length) {
154 this.replaceDocBullet(listpara.item(0));
158 replaceDocBullet : function(p)
160 // gather all the siblings.
162 parent = p.parentNode,
163 doc = parent.ownerDocument,
166 if (ns.nodeType != 1) {
170 if (!ns.className.match(/(MsoListParagraph|ql-indent-1)/i)) {
177 var ul = parent.ownerDocument.createElement('ul'); // what about number lists...
178 parent.insertBefore(ul, p);
182 items.forEach(function(n) {
183 //Roo.log("got innertHMLT=" + n.innerHTML);
185 var spans = n.getElementsByTagName('span');
187 //Roo.log("No spans found");
189 parent.removeChild(n);
190 return; // skip it...
196 for(var i = 0; i < spans.length; i++) {
198 style = this.styleToObject(spans[i]);
199 if (typeof(style['mso-list']) == 'undefined') {
203 spans[i].parentNode.removeChild(spans[i]); // remove the fake bullet.
206 //Roo.log("NOW GOT innertHMLT=" + n.innerHTML);
207 style = this.styleToObject(n); // mo-list is from the parent node.
208 if (typeof(style['mso-list']) == 'undefined') {
209 //Roo.log("parent is missing level");
210 parent.removeChild(n);
214 var nlvl = (style['mso-list'].split(' ')[1].replace(/level/,'') *1) - 1;
217 var nul = doc.createElement('ul'); // what about number lists...
218 last_li.appendChild(nul);
223 var nli = stack[nlvl].appendChild(doc.createElement('li'));
225 nli.innerHTML = n.innerHTML;
226 //Roo.log("innerHTML = " + n.innerHTML);
227 parent.removeChild(n);
229 // copy children of p into nli
230 /*while(n.firstChild) {
231 var fc = n.firstChild;