_source/core/htmlparser/fragment.js

   1 /*\r
   2 Copyright (c) 2003-2009, CKSource - Frederico Knabben. All rights reserved.\r
   3 For licensing, see LICENSE.html or http://ckeditor.com/license\r
   4 */\r
   5 \r
   6 /**\r
   7  * A lightweight representation of an HTML DOM structure.\r
   8  * @constructor\r
   9  * @example\r
  10  */\r
  11 CKEDITOR.htmlParser.fragment = function()\r
  12 {\r
  13         /**\r
  14          * The nodes contained in the root of this fragment.\r
  15          * @type Array\r
  16          * @example\r
  17          * var fragment = CKEDITOR.htmlParser.fragment.fromHtml( '<b>Sample</b> Text' );\r
  18          * alert( fragment.children.length );  "2"\r
  19          */\r
  20         this.children = [];\r
  21 \r
  22         /**\r
  23          * Get the fragment parent. Should always be null.\r
  24          * @type Object\r
  25          * @default null\r
  26          * @example\r
  27          */\r
  28         this.parent = null;\r
  29 \r
  30         /** @private */\r
  31         this._ =\r
  32         {\r
  33                 isBlockLike : true,\r
  34                 hasInlineStarted : false\r
  35         };\r
  36 };\r
  37 \r
  38 (function()\r
  39 {\r
  40         // Elements which the end tag is marked as optional in the HTML 4.01 DTD\r
  41         // (expect empty elements).\r
  42         var optionalClose = {colgroup:1,dd:1,dt:1,li:1,option:1,p:1,td:1,tfoot:1,th:1,thead:1,tr:1};\r
  43 \r
  44         // Block-level elements whose internal structure should be respected during\r
  45         // parser fixing.\r
  46         var nonBreakingBlocks = CKEDITOR.tools.extend(\r
  47                         {table:1,ul:1,ol:1,dl:1},\r
  48                         CKEDITOR.dtd.table, CKEDITOR.dtd.ul, CKEDITOR.dtd.ol, CKEDITOR.dtd.dl );\r
  49 \r
  50         /**\r
  51          * Creates a {@link CKEDITOR.htmlParser.fragment} from an HTML string.\r
  52          * @param {String} fragmentHtml The HTML to be parsed, filling the fragment.\r
  53          * @param {Number} [fixForBody=false] Wrap body with specified element if needed.\r
  54          * @returns CKEDITOR.htmlParser.fragment The fragment created.\r
  55          * @example\r
  56          * var fragment = CKEDITOR.htmlParser.fragment.fromHtml( '<b>Sample</b> Text' );\r
  57          * alert( fragment.children[0].name );  "b"\r
  58          * alert( fragment.children[1].value );  " Text"\r
  59          */\r
  60         CKEDITOR.htmlParser.fragment.fromHtml = function( fragmentHtml, fixForBody )\r
  61         {\r
  62                 var parser = new CKEDITOR.htmlParser(),\r
  63                         html = [],\r
  64                         fragment = new CKEDITOR.htmlParser.fragment(),\r
  65                         pendingInline = [],\r
  66                         currentNode = fragment,\r
  67                     // Indicate we're inside a <pre> element, spaces should be touched differently.\r
  68                         inPre = false,\r
  69                         returnPoint;\r
  70 \r
  71                 function checkPending( newTagName )\r
  72                 {\r
  73                         if ( pendingInline.length > 0 )\r
  74                         {\r
  75                                 for ( var i = 0 ; i < pendingInline.length ; i++ )\r
  76                                 {\r
  77                                         var pendingElement = pendingInline[ i ],\r
  78                                                 pendingName = pendingElement.name,\r
  79                                                 pendingDtd = CKEDITOR.dtd[ pendingName ],\r
  80                                                 currentDtd = currentNode.name && CKEDITOR.dtd[ currentNode.name ];\r
  81 \r
  82                                         if ( ( !currentDtd || currentDtd[ pendingName ] ) && ( !newTagName || !pendingDtd || pendingDtd[ newTagName ] || !CKEDITOR.dtd[ newTagName ] ) )\r
  83                                         {\r
  84                                                 // Get a clone for the pending element.\r
  85                                                 pendingElement = pendingElement.clone();\r
  86 \r
  87                                                 // Add it to the current node and make it the current,\r
  88                                                 // so the new element will be added inside of it.\r
  89                                                 pendingElement.parent = currentNode;\r
  90                                                 currentNode = pendingElement;\r
  91 \r
  92                                                 // Remove the pending element (back the index by one\r
  93                                                 // to properly process the next entry).\r
  94                                                 pendingInline.splice( i, 1 );\r
  95                                                 i--;\r
  96                                         }\r
  97                                 }\r
  98                         }\r
  99                 }\r
 100 \r
 101                 function addElement( element, target, enforceCurrent )\r
 102                 {\r
 103                         target = target || currentNode || fragment;\r
 104 \r
 105                         // If the target is the fragment and this element can't go inside\r
 106                         // body (if fixForBody).\r
 107                         if ( fixForBody && !target.type )\r
 108                         {\r
 109                                 var elementName, realElementName;\r
 110                                 if ( element.attributes\r
 111                                          && ( realElementName =\r
 112                                                   element.attributes[ '_cke_real_element_type' ] ) )\r
 113                                         elementName = realElementName;\r
 114                                 else\r
 115                                         elementName =  element.name;\r
 116                                 if ( !( elementName in CKEDITOR.dtd.$body ) )\r
 117                                 {\r
 118                                         var savedCurrent = currentNode;\r
 119 \r
 120                                         // Create a <p> in the fragment.\r
 121                                         currentNode = target;\r
 122                                         parser.onTagOpen( fixForBody, {} );\r
 123 \r
 124                                         // The new target now is the <p>.\r
 125                                         target = currentNode;\r
 126 \r
 127                                         if ( enforceCurrent )\r
 128                                                 currentNode = savedCurrent;\r
 129                                 }\r
 130                         }\r
 131 \r
 132                         // Rtrim empty spaces on block end boundary. (#3585)\r
 133                         if ( element._.isBlockLike\r
 134                                  && element.name != 'pre' )\r
 135                         {\r
 136 \r
 137                                 var length = element.children.length,\r
 138                                         lastChild = element.children[ length - 1 ],\r
 139                                         text;\r
 140                                 if ( lastChild && lastChild.type == CKEDITOR.NODE_TEXT )\r
 141                                 {\r
 142                                         if ( !( text = CKEDITOR.tools.rtrim( lastChild.value ) ) )\r
 143                                                 element.children.length = length -1;\r
 144                                         else\r
 145                                                 lastChild.value = text;\r
 146                                 }\r
 147                         }\r
 148 \r
 149                         target.add( element );\r
 150 \r
 151                         if ( element.returnPoint )\r
 152                         {\r
 153                                 currentNode = element.returnPoint;\r
 154                                 delete element.returnPoint;\r
 155                         }\r
 156                 }\r
 157 \r
 158                 parser.onTagOpen = function( tagName, attributes, selfClosing )\r
 159                 {\r
 160                         var element = new CKEDITOR.htmlParser.element( tagName, attributes );\r
 161 \r
 162                         // "isEmpty" will be always "false" for unknown elements, so we\r
 163                         // must force it if the parser has identified it as a selfClosing tag.\r
 164                         if ( element.isUnknown && selfClosing )\r
 165                                 element.isEmpty = true;\r
 166 \r
 167                         // This is a tag to be removed if empty, so do not add it immediately.\r
 168                         if ( CKEDITOR.dtd.$removeEmpty[ tagName ] )\r
 169                         {\r
 170                                 pendingInline.push( element );\r
 171                                 return;\r
 172                         }\r
 173                         else if ( tagName == 'pre' )\r
 174                                 inPre = true;\r
 175                         else if ( tagName == 'br' && inPre )\r
 176                         {\r
 177                                 currentNode.add( new CKEDITOR.htmlParser.text( '\n' ) );\r
 178                                 return;\r
 179                         }\r
 180 \r
 181                         var currentName = currentNode.name,\r
 182                                 currentDtd = ( currentName && CKEDITOR.dtd[ currentName ] ) || ( currentNode._.isBlockLike ? CKEDITOR.dtd.div : CKEDITOR.dtd.span );\r
 183 \r
 184                         // If the element cannot be child of the current element.\r
 185                         if ( !element.isUnknown && !currentNode.isUnknown && !currentDtd[ tagName ] )\r
 186                         {\r
 187                                 // If this is the fragment node, just ignore this tag and add\r
 188                                 // its children.\r
 189                                 if ( !currentName )\r
 190                                         return;\r
 191 \r
 192                                 var reApply = false;\r
 193 \r
 194                                 // If the element name is the same as the current element name,\r
 195                                 // then just close the current one and append the new one to the\r
 196                                 // parent. This situation usually happens with <p>, <li>, <dt> and\r
 197                                 // <dd>, specially in IE. Do not enter in this if block in this case.\r
 198                                 if ( tagName == currentName )\r
 199                                 {\r
 200                                         addElement( currentNode, currentNode.parent );\r
 201                                 }\r
 202                                 else\r
 203                                 {\r
 204                                         if ( nonBreakingBlocks[ currentName ] )\r
 205                                         {\r
 206                                                 if ( !returnPoint )\r
 207                                                         returnPoint = currentNode;\r
 208                                         }\r
 209                                         else\r
 210                                         {\r
 211                                                 addElement( currentNode, currentNode.parent, true );\r
 212 \r
 213                                                 if ( !optionalClose[ currentName ] )\r
 214                                                 {\r
 215                                                         // The current element is an inline element, which\r
 216                                                         // cannot hold the new one. Put it in the pending list,\r
 217                                                         // and try adding the new one after it.\r
 218                                                         pendingInline.unshift( currentNode );\r
 219                                                 }\r
 220                                         }\r
 221 \r
 222                                         reApply = true;\r
 223                                 }\r
 224 \r
 225                                 // In any of the above cases, we'll be adding, or trying to\r
 226                                 // add it to the parent.\r
 227                                 currentNode = currentNode.returnPoint || currentNode.parent;\r
 228 \r
 229                                 if ( reApply )\r
 230                                 {\r
 231                                         parser.onTagOpen.apply( this, arguments );\r
 232                                         return;\r
 233                                 }\r
 234                         }\r
 235 \r
 236                         checkPending( tagName );\r
 237 \r
 238                         element.parent = currentNode;\r
 239                         element.returnPoint = returnPoint;\r
 240                         returnPoint = 0;\r
 241 \r
 242                         if ( element.isEmpty )\r
 243                                 addElement( element );\r
 244                         else\r
 245                                 currentNode = element;\r
 246                 };\r
 247 \r
 248                 parser.onTagClose = function( tagName )\r
 249                 {\r
 250                         var index = 0,\r
 251                                 pendingAdd = [],\r
 252                                 candidate = currentNode;\r
 253 \r
 254                         while ( candidate.type && candidate.name != tagName )\r
 255                         {\r
 256                                 // If this is an inline element, add it to the pending list, so\r
 257                                 // it will continue after the closing tag.\r
 258                                 if ( !candidate._.isBlockLike )\r
 259                                 {\r
 260                                         pendingInline.unshift( candidate );\r
 261 \r
 262                                         // Increase the index, so it will not get checked again in\r
 263                                         // the pending list check that follows.\r
 264                                         index++;\r
 265                                 }\r
 266 \r
 267                                 // This node should be added to it's parent at this point. But,\r
 268                                 // it should happen only if the closing tag is really closing\r
 269                                 // one of the nodes. So, for now, we just cache it.\r
 270                                 pendingAdd.push( candidate );\r
 271 \r
 272                                 candidate = candidate.parent;\r
 273                         }\r
 274 \r
 275                         if ( candidate.type )\r
 276                         {\r
 277                                 // Add all elements that have been found in the above loop.\r
 278                                 for ( var i = 0 ; i < pendingAdd.length ; i++ )\r
 279                                 {\r
 280                                         var node = pendingAdd[ i ];\r
 281                                         addElement( node, node.parent );\r
 282                                 }\r
 283 \r
 284                                 currentNode = candidate;\r
 285 \r
 286                                 if( currentNode.name == 'pre' )\r
 287                                         inPre = false;\r
 288 \r
 289                                 addElement( candidate, candidate.parent );\r
 290 \r
 291                                 // The parent should start receiving new nodes now, except if\r
 292                                 // addElement changed the currentNode.\r
 293                                 if ( candidate == currentNode )\r
 294                                         currentNode = currentNode.parent;\r
 295                         }\r
 296                         // The tag is not actually closing anything, thus we need invalidate\r
 297                         // the pending elements.(#3862)\r
 298                         else\r
 299                         {\r
 300                                 pendingInline.splice( 0, index );\r
 301                                 index = 0;\r
 302                         }\r
 303 \r
 304                         // Check if there is any pending tag to be closed.\r
 305                         for ( ; index < pendingInline.length ; index++ )\r
 306                         {\r
 307                                 // If found, just remove it from the list.\r
 308                                 if ( tagName == pendingInline[ index ].name )\r
 309                                 {\r
 310                                         pendingInline.splice( index, 1 );\r
 311 \r
 312                                         // Decrease the index so we continue from the next one.\r
 313                                         index--;\r
 314                                 }\r
 315                         }\r
 316                 };\r
 317 \r
 318                 parser.onText = function( text )\r
 319                 {\r
 320                         // Trim empty spaces at beginning of element contents except <pre>.\r
 321                         if ( !currentNode._.hasInlineStarted && !inPre )\r
 322                         {\r
 323                                 text = CKEDITOR.tools.ltrim( text );\r
 324 \r
 325                                 if ( text.length === 0 )\r
 326                                         return;\r
 327                         }\r
 328 \r
 329                         checkPending();\r
 330 \r
 331                         if ( fixForBody && !currentNode.type )\r
 332                                 this.onTagOpen( fixForBody, {} );\r
 333 \r
 334                         // Shrinking consequential spaces into one single for all elements\r
 335                         // text contents.\r
 336                         if ( !inPre )\r
 337                                 text = text.replace( /[\t\r\n ]{2,}|[\t\r\n]/g, ' ' );\r
 338 \r
 339                         currentNode.add( new CKEDITOR.htmlParser.text( text ) );\r
 340                 };\r
 341 \r
 342                 parser.onCDATA = function( cdata )\r
 343                 {\r
 344                         currentNode.add( new CKEDITOR.htmlParser.cdata( cdata ) );\r
 345                 };\r
 346 \r
 347                 parser.onComment = function( comment )\r
 348                 {\r
 349                         currentNode.add( new CKEDITOR.htmlParser.comment( comment ) );\r
 350                 };\r
 351 \r
 352                 // Parse it.\r
 353                 parser.parse( fragmentHtml );\r
 354 \r
 355                 // Close all pending nodes.\r
 356                 while ( currentNode.type )\r
 357                 {\r
 358                         var parent = currentNode.parent,\r
 359                                 node = currentNode;\r
 360 \r
 361                         if ( fixForBody && !parent.type && !CKEDITOR.dtd.$body[ node.name ] )\r
 362                         {\r
 363                                 currentNode = parent;\r
 364                                 parser.onTagOpen( fixForBody, {} );\r
 365                                 parent = currentNode;\r
 366                         }\r
 367 \r
 368                         parent.add( node );\r
 369                         currentNode = parent;\r
 370                 }\r
 371 \r
 372                 return fragment;\r
 373         };\r
 374 \r
 375         CKEDITOR.htmlParser.fragment.prototype =\r
 376         {\r
 377                 /**\r
 378                  * Adds a node to this fragment.\r
 379                  * @param {Object} node The node to be added. It can be any of of the\r
 380                  *              following types: {@link CKEDITOR.htmlParser.element},\r
 381                  *              {@link CKEDITOR.htmlParser.text} and\r
 382                  *              {@link CKEDITOR.htmlParser.comment}.\r
 383                  * @example\r
 384                  */\r
 385                 add : function( node )\r
 386                 {\r
 387                         var len = this.children.length,\r
 388                                 previous = len > 0 && this.children[ len - 1 ] || null;\r
 389 \r
 390                         if ( previous )\r
 391                         {\r
 392                                 // If the block to be appended is following text, trim spaces at\r
 393                                 // the right of it.\r
 394                                 if ( node._.isBlockLike && previous.type == CKEDITOR.NODE_TEXT )\r
 395                                 {\r
 396                                         previous.value = CKEDITOR.tools.rtrim( previous.value );\r
 397 \r
 398                                         // If we have completely cleared the previous node.\r
 399                                         if ( previous.value.length === 0 )\r
 400                                         {\r
 401                                                 // Remove it from the list and add the node again.\r
 402                                                 this.children.pop();\r
 403                                                 this.add( node );\r
 404                                                 return;\r
 405                                         }\r
 406                                 }\r
 407 \r
 408                                 previous.next = node;\r
 409                         }\r
 410 \r
 411                         node.previous = previous;\r
 412                         node.parent = this;\r
 413 \r
 414                         this.children.push( node );\r
 415 \r
 416                         this._.hasInlineStarted = node.type == CKEDITOR.NODE_TEXT || ( node.type == CKEDITOR.NODE_ELEMENT && !node._.isBlockLike );\r
 417                 },\r
 418 \r
 419                 /**\r
 420                  * Writes the fragment HTML to a CKEDITOR.htmlWriter.\r
 421                  * @param {CKEDITOR.htmlWriter} writer The writer to which write the HTML.\r
 422                  * @example\r
 423                  * var writer = new CKEDITOR.htmlWriter();\r
 424                  * var fragment = CKEDITOR.htmlParser.fragment.fromHtml( '&lt;P&gt;&lt;B&gt;Example' );\r
 425                  * fragment.writeHtml( writer )\r
 426                  * alert( writer.getHtml() );  "&lt;p&gt;&lt;b&gt;Example&lt;/b&gt;&lt;/p&gt;"\r
 427                  */\r
 428                 writeHtml : function( writer, filter )\r
 429                 {\r
 430                         for ( var i = 0, len = this.children.length ; i < len ; i++ )\r
 431                                 this.children[i].writeHtml( writer, filter );\r
 432                 }\r
 433         };\r
 434 })();\r