MsoParagraphList is sometimes used when there is no list.
[roojs1] / Roo / htmleditor / FilterWord.js
index 3af0cd3..d3d6cc4 100644 (file)
 Roo.htmleditor.FilterWord = function(cfg)
 {
     // no need to apply config.
-    this.walk(cfg.node);
+    this.replaceDocBullets(cfg.node);
+    
+    // this is disabled as the removal is done by other filters;
+   // this.walk(cfg.node);
+    
+    
 }
 
 Roo.extend(Roo.htmleditor.FilterWord, Roo.htmleditor.Filter,
@@ -62,7 +67,10 @@ Roo.extend(Roo.htmleditor.FilterWord, Roo.htmleditor.Filter,
                 node.removeChild(cn);
                 node.parentNode.insertBefore(cn, node);
                 // move node to parent - and clean it..
-                this.replaceTag(cn);
+                if (cn.nodeType == 1) {
+                    this.replaceTag(cn);
+                }
+                
             }
             node.parentNode.removeChild(node);
             /// no need to iterate chidlren = it's got none..
@@ -114,5 +122,173 @@ Roo.extend(Roo.htmleditor.FilterWord, Roo.htmleditor.Filter,
         
         
         
+    },
+    
+    styleToObject: function(node)
+    {
+        var styles = (node.getAttribute("style") || '').split(";");
+        var ret = {};
+        Roo.each(styles, function(s) {
+            if (!s.match(/:/)) {
+                return;
+            }
+            var kv = s.split(":");
+             
+            // what ever is left... we allow.
+            ret[kv[0].trim()] = kv[1];
+        });
+        return ret;
+    },
+    
+    
+    replaceDocBullets : function(doc)
+    {
+        // this is a bit odd - but it appears some indents use ql-indent-1
+         //Roo.log(doc.innerHTML);
+        
+        var listpara = doc.getElementsByClassName('MsoListParagraphCxSpFirst');
+        for( var i = 0; i < listpara.length; i ++) {
+            listpara.item(i).className = "MsoListParagraph";
+        }
+        // this is a bit hacky - we had one word document where h2 had a miso-list attribute.
+        var htwo = doc.getElementsByTagName('h2');
+        for( var i = 0; i < htwo.length; i ++) {
+            if (htwo.item(i).getAttribute('style').match(/mso-list:/)) {
+                htwo.item(i).className = "MsoListParagraph";
+            }
+        }
+        
+        listpara = doc.getElementsByClassName('ql-indent-1');
+        while(listpara.length) {
+            this.replaceDocBullet(listpara.item(0));
+        }
+        listpara = doc.getElementsByClassName('MsoListParagraph');
+        while(listpara.length) {
+            
+            this.replaceDocBullet(listpara.item(0));
+        }
+      
+    },
+    
+     
+    
+    replaceDocBullet : function(p)
+    {
+        // gather all the siblings.
+        var ns = p,
+            parent = p.parentNode,
+            doc = parent.ownerDocument,
+            items = [];
+            
+            
+        while (ns) {
+            if (ns.nodeType != 1) {
+                ns = ns.nextSibling;
+                continue;
+            }
+            if (!ns.className.match(/(MsoListParagraph|ql-indent-1)/i)) {
+                break;
+            }
+            var spans = ns.getElementsByTagName('span');
+            if (!spans.length) {
+                break;
+            }
+            var has_list  = false;
+            for(var i = 0; i < spans.length; i++) {
+                if (spans[i].getAttribute('style').match(/mso-list/)) {
+                    has_list = true;
+                    break;
+                }
+            }
+            if (!has_list) {
+                break;
+            }
+            
+            
+            items.push(ns);
+            ns = ns.nextSibling;
+        }
+        if (!items.length) {
+            ns.className = "";
+            return;
+        }
+        
+        var ul = parent.ownerDocument.createElement('ul'); // what about number lists...
+        parent.insertBefore(ul, p);
+        var lvl = 0;
+        var stack = [ ul ];
+        var last_li = false;
+        
+         
+        items.forEach(function(n, ipos) {
+            //Roo.log("got innertHMLT=" + n.innerHTML);
+            
+            var spans = n.getElementsByTagName('span');
+            if (!spans.length) {
+                //Roo.log("No spans found");
+                 
+                parent.removeChild(n);
+                
+                
+                return; // skip it...
+            }
+           
+                
+            
+            var style = {};
+            for(var i = 0; i < spans.length; i++) {
+            
+                style = this.styleToObject(spans[i]);
+                if (typeof(style['mso-list']) == 'undefined') {
+                    continue;
+                }
+                
+                spans[i].parentNode.removeChild(spans[i]); // remove the fake bullet.
+                break;
+            }
+            //Roo.log("NOW GOT innertHMLT=" + n.innerHTML);
+            style = this.styleToObject(n); // mo-list is from the parent node.
+            if (typeof(style['mso-list']) == 'undefined') {
+                //Roo.log("parent is missing level");
+                
+                 
+                parent.removeChild(n);
+                 
+                return;
+            }
+            
+            var nlvl =   (style['mso-list'].split(' ')[1].replace(/level/,'') *1) - 1  ;
+            
+             
+            if (nlvl > lvl) {
+                //new indent
+                var nul = doc.createElement('ul'); // what about number lists...
+                if (!last_li) {
+                    last_li = doc.createElement('li');
+                    stack[lvl].appendChild(last_li);
+                }
+                last_li.appendChild(nul);
+                stack[nlvl] = nul;
+                
+            }
+            lvl = nlvl;
+            
+            var nli = stack[nlvl].appendChild(doc.createElement('li'));
+            last_li = nli;
+            nli.innerHTML = n.innerHTML;
+            //Roo.log("innerHTML = " + n.innerHTML);
+            parent.removeChild(n);
+            
+             
+             
+            
+        },this);
+        
+        
+        
+        
     }
+    
+    
+    
 });
\ No newline at end of file