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.replaceAname(cfg.node);
18 // this is disabled as the removal is done by other filters;
19 // this.walk(cfg.node);
24 Roo.extend(Roo.htmleditor.FilterWord, Roo.htmleditor.Filter,
30 * Clean up MS wordisms...
32 replaceTag : function(node)
35 // no idea what this does - span with text, replaceds with just text.
37 node.nodeName == 'SPAN' &&
38 !node.hasAttributes() &&
39 node.childNodes.length == 1 &&
40 node.firstChild.nodeName == "#text"
42 var textNode = node.firstChild;
43 node.removeChild(textNode);
44 if (node.getAttribute('lang') != 'zh-CN') { // do not space pad on chinese characters..
45 node.parentNode.insertBefore(node.ownerDocument.createTextNode(" "), node);
47 node.parentNode.insertBefore(textNode, node);
48 if (node.getAttribute('lang') != 'zh-CN') { // do not space pad on chinese characters..
49 node.parentNode.insertBefore(node.ownerDocument.createTextNode(" ") , node);
52 node.parentNode.removeChild(node);
53 return false; // dont do chidren - we have remove our node - so no need to do chdhilren?
58 if (node.tagName.toLowerCase().match(/^(style|script|applet|embed|noframes|noscript)$/)) {
59 node.parentNode.removeChild(node);
60 return false; // dont do chidlren
62 //Roo.log(node.tagName);
63 // remove - but keep children..
64 if (node.tagName.toLowerCase().match(/^(meta|link|\\?xml:|st1:|o:|v:|font)/)) {
65 //Roo.log('-- removed');
66 while (node.childNodes.length) {
67 var cn = node.childNodes[0];
69 node.parentNode.insertBefore(cn, node);
70 // move node to parent - and clean it..
71 if (cn.nodeType == 1) {
76 node.parentNode.removeChild(node);
77 /// no need to iterate chidlren = it's got none..
78 //this.iterateChildren(node, this.cleanWord);
79 return false; // no need to iterate children.
82 if (node.className.length) {
84 var cn = node.className.split(/\W+/);
86 Roo.each(cn, function(cls) {
87 if (cls.match(/Mso[a-zA-Z]+/)) {
92 node.className = cna.length ? cna.join(' ') : '';
94 node.removeAttribute("class");
98 if (node.hasAttribute("lang")) {
99 node.removeAttribute("lang");
102 if (node.hasAttribute("style")) {
104 var styles = node.getAttribute("style").split(";");
106 Roo.each(styles, function(s) {
110 var kv = s.split(":");
111 if (kv[0].match(/^(mso-|line|font|background|margin|padding|color)/)) {
114 // what ever is left... we allow.
117 node.setAttribute("style", nstyle.length ? nstyle.join(';') : '');
118 if (!nstyle.length) {
119 node.removeAttribute('style');
122 return true; // do children
128 styleToObject: function(node)
130 var styles = (node.getAttribute("style") || '').split(";");
132 Roo.each(styles, function(s) {
136 var kv = s.split(":");
138 // what ever is left... we allow.
139 ret[kv[0].trim()] = kv[1];
145 replaceAname : function (doc)
147 // replace all the a/name without..
148 var aa = Array.from(doc.getElementsByTagName('a'));
149 for (var i = 0; i < aa.length; i++) {
151 if (a.hasAttribute("name")) {
152 a.removeAttribute("name");
154 if (a.hasAttribute("href")) {
157 // reparent children.
158 this.removeNodeKeepChildren(a);
168 replaceDocBullets : function(doc)
170 // this is a bit odd - but it appears some indents use ql-indent-1
171 //Roo.log(doc.innerHTML);
173 var listpara = Array.from(doc.getElementsByClassName('MsoListParagraphCxSpFirst'));
174 for( var i = 0; i < listpara.length; i ++) {
175 listpara[i].className = "MsoListParagraph";
178 listpara = Array.from(doc.getElementsByClassName('MsoListParagraphCxSpMiddle'));
179 for( var i = 0; i < listpara.length; i ++) {
180 listpara[i].className = "MsoListParagraph";
182 listpara = Array.from(doc.getElementsByClassName('MsoListParagraphCxSpLast'));
183 for( var i = 0; i < listpara.length; i ++) {
184 listpara[i].className = "MsoListParagraph";
186 listpara = Array.from(doc.getElementsByClassName('ql-indent-1'));
187 for( var i = 0; i < listpara.length; i ++) {
188 listpara[i].className = "MsoListParagraph";
191 // this is a bit hacky - we had one word document where h2 had a miso-list attribute.
192 var htwo = Array.from(doc.getElementsByTagName('h2'));
193 for( var i = 0; i < htwo.length; i ++) {
194 if (htwo[i].hasAttribute('style') && htwo[i].getAttribute('style').match(/mso-list:/)) {
195 htwo[i].className = "MsoListParagraph";
198 listpara = Array.from(doc.getElementsByClassName('MsoNormal'));
199 for( var i = 0; i < listpara.length; i ++) {
200 if (listpara[i].hasAttribute('style') && listpara[i].getAttribute('style').match(/mso-list:/)) {
201 listpara[i].className = "MsoListParagraph";
203 listpara[i].className = "MsoNormalx";
207 listpara = doc.getElementsByClassName('MsoListParagraph');
208 // Roo.log(doc.innerHTML);
212 while(listpara.length) {
214 this.replaceDocBullet(listpara.item(0));
221 replaceDocBullet : function(p)
223 // gather all the siblings.
225 parent = p.parentNode,
226 doc = parent.ownerDocument,
231 if (ns.nodeType != 1) {
235 if (!ns.className.match(/(MsoListParagraph|ql-indent-1)/i)) {
238 var spans = ns.getElementsByTagName('span');
239 if (ns.hasAttribute('style') && ns.getAttribute('style').match(/mso-list/)) {
243 if (spans.length && spans[0].hasAttribute('style')) {
244 var style = this.styleToObject(spans[0]);
245 if (typeof(style['font-family']) != 'undefined' && !style['font-family'].match(/Symbol/)) {
252 var spans = ns.getElementsByTagName('span');
256 var has_list = false;
257 for(var i = 0; i < spans.length; i++) {
258 if (spans[i].hasAttribute('style') && spans[i].getAttribute('style').match(/mso-list/)) {
276 var ul = parent.ownerDocument.createElement(listtype); // what about number lists...
277 parent.insertBefore(ul, p);
282 var margin_to_depth = {};
285 items.forEach(function(n, ipos) {
286 //Roo.log("got innertHMLT=" + n.innerHTML);
288 var spans = n.getElementsByTagName('span');
290 //Roo.log("No spans found");
292 parent.removeChild(n);
295 return; // skip it...
301 for(var i = 0; i < spans.length; i++) {
303 style = this.styleToObject(spans[i]);
304 if (typeof(style['mso-list']) == 'undefined') {
307 if (listtype == 'ol') {
308 num = spans[i].innerText.replace(/[^0-9]+]/g,'') * 1;
310 spans[i].parentNode.removeChild(spans[i]); // remove the fake bullet.
313 //Roo.log("NOW GOT innertHMLT=" + n.innerHTML);
314 style = this.styleToObject(n); // mo-list is from the parent node.
315 if (typeof(style['mso-list']) == 'undefined') {
316 //Roo.log("parent is missing level");
318 parent.removeChild(n);
323 var margin = style['margin-left'];
324 if (typeof(margin_to_depth[margin]) == 'undefined') {
326 margin_to_depth[margin] = max_margins;
328 nlvl = margin_to_depth[margin] ;
332 var nul = doc.createElement(listtype); // what about number lists...
334 last_li = doc.createElement('li');
335 stack[lvl].appendChild(last_li);
337 last_li.appendChild(nul);
343 // not starting at 1..
344 if (!stack[nlvl].hasAttribute("start") && listtype == "ol") {
345 stack[nlvl].setAttribute("start", num);
348 var nli = stack[nlvl].appendChild(doc.createElement('li'));
350 nli.innerHTML = n.innerHTML;
351 //Roo.log("innerHTML = " + n.innerHTML);
352 parent.removeChild(n);