_source/core/htmlparser/fragment.js

   1 /*\r
   2 Copyright (c) 2003-2011, 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         // Block-level elements whose internal structure should be respected during\r
  41         // parser fixing.\r
  42         var nonBreakingBlocks = CKEDITOR.tools.extend( { table:1,ul:1,ol:1,dl:1 }, CKEDITOR.dtd.table, CKEDITOR.dtd.ul, CKEDITOR.dtd.ol, CKEDITOR.dtd.dl );\r
  43 \r
  44         // IE < 8 don't output the close tag on definition list items. (#6975)\r
  45         var optionalCloseTags = CKEDITOR.env.ie && CKEDITOR.env.version < 8 ? { dd : 1, dt :1 } : {};\r
  46 \r
  47         var listBlocks = { ol:1, ul:1 };\r
  48 \r
  49         // Dtd of the fragment element, basically it accept anything except for intermediate structure, e.g. orphan <li>.\r
  50         var rootDtd = CKEDITOR.tools.extend( {}, { html: 1 }, CKEDITOR.dtd.html, CKEDITOR.dtd.body, CKEDITOR.dtd.head, { style:1,script:1 } );\r
  51 \r
  52         /**\r
  53          * Creates a {@link CKEDITOR.htmlParser.fragment} from an HTML string.\r
  54          * @param {String} fragmentHtml The HTML to be parsed, filling the fragment.\r
  55          * @param {Number} [fixForBody=false] Wrap body with specified element if needed.\r
  56          * @param {CKEDITOR.htmlParser.element} contextNode Parse the html as the content of this element.\r
  57          * @returns CKEDITOR.htmlParser.fragment The fragment created.\r
  58          * @example\r
  59          * var fragment = CKEDITOR.htmlParser.fragment.fromHtml( '<b>Sample</b> Text' );\r
  60          * alert( fragment.children[0].name );  "b"\r
  61          * alert( fragment.children[1].value );  " Text"\r
  62          */\r
  63         CKEDITOR.htmlParser.fragment.fromHtml = function( fragmentHtml, fixForBody, contextNode )\r
  64         {\r
  65                 var parser = new CKEDITOR.htmlParser(),\r
  66                         fragment = contextNode || new CKEDITOR.htmlParser.fragment(),\r
  67                         pendingInline = [],\r
  68                         pendingBRs = [],\r
  69                         currentNode = fragment,\r
  70                     // Indicate we're inside a <pre> element, spaces should be touched differently.\r
  71                         inPre = false;\r
  72 \r
  73                 function checkPending( newTagName )\r
  74                 {\r
  75                         var pendingBRsSent;\r
  76 \r
  77                         if ( pendingInline.length > 0 )\r
  78                         {\r
  79                                 for ( var i = 0 ; i < pendingInline.length ; i++ )\r
  80                                 {\r
  81                                         var pendingElement = pendingInline[ i ],\r
  82                                                 pendingName = pendingElement.name,\r
  83                                                 pendingDtd = CKEDITOR.dtd[ pendingName ],\r
  84                                                 currentDtd = currentNode.name && CKEDITOR.dtd[ currentNode.name ];\r
  85 \r
  86                                         if ( ( !currentDtd || currentDtd[ pendingName ] ) && ( !newTagName || !pendingDtd || pendingDtd[ newTagName ] || !CKEDITOR.dtd[ newTagName ] ) )\r
  87                                         {\r
  88                                                 if ( !pendingBRsSent )\r
  89                                                 {\r
  90                                                         sendPendingBRs();\r
  91                                                         pendingBRsSent = 1;\r
  92                                                 }\r
  93 \r
  94                                                 // Get a clone for the pending element.\r
  95                                                 pendingElement = pendingElement.clone();\r
  96 \r
  97                                                 // Add it to the current node and make it the current,\r
  98                                                 // so the new element will be added inside of it.\r
  99                                                 pendingElement.parent = currentNode;\r
 100                                                 currentNode = pendingElement;\r
 101 \r
 102                                                 // Remove the pending element (back the index by one\r
 103                                                 // to properly process the next entry).\r
 104                                                 pendingInline.splice( i, 1 );\r
 105                                                 i--;\r
 106                                         }\r
 107                                 }\r
 108                         }\r
 109                 }\r
 110 \r
 111                 function sendPendingBRs()\r
 112                 {\r
 113                         while ( pendingBRs.length )\r
 114                                 currentNode.add( pendingBRs.shift() );\r
 115                 }\r
 116 \r
 117                 /*\r
 118                 * Beside of simply append specified element to target, this function also takes\r
 119                 * care of other dirty lifts like forcing block in body, trimming spaces at\r
 120                 * the block boundaries etc.\r
 121                 *\r
 122                 * @param {Element} element  The element to be added as the last child of {@link target}.\r
 123                 * @param {Element} target The parent element to relieve the new node.\r
 124                 * @param {Boolean} [moveCurrent=false] Don't change the "currentNode" global unless\r
 125                 * there's a return point node specified on the element, otherwise move current onto {@link target} node.\r
 126                  */\r
 127                 function addElement( element, target, moveCurrent )\r
 128                 {\r
 129                         // Ignore any element that has already been added.\r
 130                         if ( element.previous !== undefined )\r
 131                                 return;\r
 132 \r
 133                         target = target || currentNode || fragment;\r
 134 \r
 135                         // Current element might be mangled by fix body below,\r
 136                         // save it for restore later.\r
 137                         var savedCurrent = currentNode;\r
 138 \r
 139                         // If the target is the fragment and this inline element can't go inside\r
 140                         // body (if fixForBody).\r
 141                         if ( fixForBody && ( !target.type || target.name == 'body' ) )\r
 142                         {\r
 143                                 var elementName, realElementName;\r
 144                                 if ( element.attributes\r
 145                                          && ( realElementName =\r
 146                                                   element.attributes[ 'data-cke-real-element-type' ] ) )\r
 147                                         elementName = realElementName;\r
 148                                 else\r
 149                                         elementName =  element.name;\r
 150 \r
 151                                 if ( elementName && !( elementName in CKEDITOR.dtd.$body || elementName == 'body' || element.isOrphan ) )\r
 152                                 {\r
 153                                         // Create a <p> in the fragment.\r
 154                                         currentNode = target;\r
 155                                         parser.onTagOpen( fixForBody, {} );\r
 156 \r
 157                                         // The new target now is the <p>.\r
 158                                         element.returnPoint = target = currentNode;\r
 159                                 }\r
 160                         }\r
 161 \r
 162                         // Rtrim empty spaces on block end boundary. (#3585)\r
 163                         if ( element._.isBlockLike\r
 164                                  && element.name != 'pre' )\r
 165                         {\r
 166 \r
 167                                 var length = element.children.length,\r
 168                                         lastChild = element.children[ length - 1 ],\r
 169                                         text;\r
 170                                 if ( lastChild && lastChild.type == CKEDITOR.NODE_TEXT )\r
 171                                 {\r
 172                                         if ( !( text = CKEDITOR.tools.rtrim( lastChild.value ) ) )\r
 173                                                 element.children.length = length -1;\r
 174                                         else\r
 175                                                 lastChild.value = text;\r
 176                                 }\r
 177                         }\r
 178 \r
 179                         target.add( element );\r
 180 \r
 181                         if ( element.returnPoint )\r
 182                         {\r
 183                                 currentNode = element.returnPoint;\r
 184                                 delete element.returnPoint;\r
 185                         }\r
 186                         else\r
 187                                 currentNode = moveCurrent ? target : savedCurrent;\r
 188                 }\r
 189 \r
 190                 parser.onTagOpen = function( tagName, attributes, selfClosing, optionalClose )\r
 191                 {\r
 192                         var element = new CKEDITOR.htmlParser.element( tagName, attributes );\r
 193 \r
 194                         // "isEmpty" will be always "false" for unknown elements, so we\r
 195                         // must force it if the parser has identified it as a selfClosing tag.\r
 196                         if ( element.isUnknown && selfClosing )\r
 197                                 element.isEmpty = true;\r
 198 \r
 199                         // Check for optional closed elements, including browser quirks and manually opened blocks.\r
 200                         element.isOptionalClose = tagName in optionalCloseTags || optionalClose;\r
 201 \r
 202                         // This is a tag to be removed if empty, so do not add it immediately.\r
 203                         if ( CKEDITOR.dtd.$removeEmpty[ tagName ] )\r
 204                         {\r
 205                                 pendingInline.push( element );\r
 206                                 return;\r
 207                         }\r
 208                         else if ( tagName == 'pre' )\r
 209                                 inPre = true;\r
 210                         else if ( tagName == 'br' && inPre )\r
 211                         {\r
 212                                 currentNode.add( new CKEDITOR.htmlParser.text( '\n' ) );\r
 213                                 return;\r
 214                         }\r
 215 \r
 216                         if ( tagName == 'br' )\r
 217                         {\r
 218                                 pendingBRs.push( element );\r
 219                                 return;\r
 220                         }\r
 221 \r
 222                         while( 1 )\r
 223                         {\r
 224                                 var currentName = currentNode.name;\r
 225 \r
 226                                 var currentDtd = currentName ? ( CKEDITOR.dtd[ currentName ]\r
 227                                                 || ( currentNode._.isBlockLike ? CKEDITOR.dtd.div : CKEDITOR.dtd.span ) )\r
 228                                                 : rootDtd;\r
 229 \r
 230                                 // If the element cannot be child of the current element.\r
 231                                 if ( !element.isUnknown && !currentNode.isUnknown && !currentDtd[ tagName ] )\r
 232                                 {\r
 233                                         // Current node doesn't have a close tag, time for a close\r
 234                                         // as this element isn't fit in. (#7497)\r
 235                                         if ( currentNode.isOptionalClose )\r
 236                                                 parser.onTagClose( currentName );\r
 237                                         // Fixing malformed nested lists by moving it into a previous list item. (#3828)\r
 238                                         else if ( tagName in listBlocks\r
 239                                                 && currentName in listBlocks )\r
 240                                         {\r
 241                                                 var children = currentNode.children,\r
 242                                                         lastChild = children[ children.length - 1 ];\r
 243 \r
 244                                                 // Establish the list item if it's not existed.\r
 245                                                 if ( !( lastChild && lastChild.name == 'li' ) )\r
 246                                                         addElement( ( lastChild = new CKEDITOR.htmlParser.element( 'li' ) ), currentNode );\r
 247 \r
 248                                                 !element.returnPoint && ( element.returnPoint = currentNode );\r
 249                                                 currentNode = lastChild;\r
 250                                         }\r
 251                                         // Establish new list root for orphan list items.\r
 252                                         else if ( tagName in CKEDITOR.dtd.$listItem && currentName != tagName )\r
 253                                                 parser.onTagOpen( tagName == 'li' ? 'ul' : 'dl', {}, 0, 1 );\r
 254                                         // We're inside a structural block like table and list, AND the incoming element\r
 255                                         // is not of the same type (e.g. <td>td1<td>td2</td>), we simply add this new one before it,\r
 256                                         // and most importantly, return back to here once this element is added,\r
 257                                         // e.g. <table><tr><td>td1</td><p>p1</p><td>td2</td></tr></table>\r
 258                                         else if ( currentName in nonBreakingBlocks && currentName != tagName )\r
 259                                         {\r
 260                                                 !element.returnPoint && ( element.returnPoint = currentNode );\r
 261                                                 currentNode = currentNode.parent;\r
 262                                         }\r
 263                                         else\r
 264                                         {\r
 265                                                 // The current element is an inline element, which\r
 266                                                 // need to be continued even after the close, so put\r
 267                                                 // it in the pending list.\r
 268                                                 if ( currentName in CKEDITOR.dtd.$inline )\r
 269                                                         pendingInline.unshift( currentNode );\r
 270 \r
 271                                                 // The most common case where we just need to close the\r
 272                                                 // current one and append the new one to the parent.\r
 273                                                 if ( currentNode.parent )\r
 274                                                         addElement( currentNode, currentNode.parent, 1 );\r
 275                                                 // We've tried our best to fix the embarrassment here, while\r
 276                                                 // this element still doesn't find it's parent, mark it as\r
 277                                                 // orphan and show our tolerance to it.\r
 278                                                 else\r
 279                                                 {\r
 280                                                         element.isOrphan = 1;\r
 281                                                         break;\r
 282                                                 }\r
 283                                         }\r
 284                                 }\r
 285                                 else\r
 286                                         break;\r
 287                         }\r
 288 \r
 289                         checkPending( tagName );\r
 290                         sendPendingBRs();\r
 291 \r
 292                         element.parent = currentNode;\r
 293 \r
 294                         if ( element.isEmpty )\r
 295                                 addElement( element );\r
 296                         else\r
 297                                 currentNode = element;\r
 298                 };\r
 299 \r
 300                 parser.onTagClose = function( tagName )\r
 301                 {\r
 302                         // Check if there is any pending tag to be closed.\r
 303                         for ( var i = pendingInline.length - 1 ; i >= 0 ; i-- )\r
 304                         {\r
 305                                 // If found, just remove it from the list.\r
 306                                 if ( tagName == pendingInline[ i ].name )\r
 307                                 {\r
 308                                         pendingInline.splice( i, 1 );\r
 309                                         return;\r
 310                                 }\r
 311                         }\r
 312 \r
 313                         var pendingAdd = [],\r
 314                                 newPendingInline = [],\r
 315                                 candidate = currentNode;\r
 316 \r
 317                         while ( candidate != fragment && candidate.name != tagName )\r
 318                         {\r
 319                                 // If this is an inline element, add it to the pending list, if we're\r
 320                                 // really closing one of the parents element later, they will continue\r
 321                                 // after it.\r
 322                                 if ( !candidate._.isBlockLike )\r
 323                                         newPendingInline.unshift( candidate );\r
 324 \r
 325                                 // This node should be added to it's parent at this point. But,\r
 326                                 // it should happen only if the closing tag is really closing\r
 327                                 // one of the nodes. So, for now, we just cache it.\r
 328                                 pendingAdd.push( candidate );\r
 329 \r
 330                                 // Make sure return point is properly restored.\r
 331                                 candidate = candidate.returnPoint || candidate.parent;\r
 332                         }\r
 333 \r
 334                         if ( candidate != fragment )\r
 335                         {\r
 336                                 // Add all elements that have been found in the above loop.\r
 337                                 for ( i = 0 ; i < pendingAdd.length ; i++ )\r
 338                                 {\r
 339                                         var node = pendingAdd[ i ];\r
 340                                         addElement( node, node.parent );\r
 341                                 }\r
 342 \r
 343                                 currentNode = candidate;\r
 344 \r
 345                                 if ( currentNode.name == 'pre' )\r
 346                                         inPre = false;\r
 347 \r
 348                                 if ( candidate._.isBlockLike )\r
 349                                         sendPendingBRs();\r
 350 \r
 351                                 addElement( candidate, candidate.parent );\r
 352 \r
 353                                 // The parent should start receiving new nodes now, except if\r
 354                                 // addElement changed the currentNode.\r
 355                                 if ( candidate == currentNode )\r
 356                                         currentNode = currentNode.parent;\r
 357 \r
 358                                 pendingInline = pendingInline.concat( newPendingInline );\r
 359                         }\r
 360 \r
 361                         if ( tagName == 'body' )\r
 362                                 fixForBody = false;\r
 363                 };\r
 364 \r
 365                 parser.onText = function( text )\r
 366                 {\r
 367                         // Trim empty spaces at beginning of text contents except <pre>.\r
 368                         if ( ( !currentNode._.hasInlineStarted || pendingBRs.length ) && !inPre )\r
 369                         {\r
 370                                 text = CKEDITOR.tools.ltrim( text );\r
 371 \r
 372                                 if ( text.length === 0 )\r
 373                                         return;\r
 374                         }\r
 375 \r
 376                         sendPendingBRs();\r
 377                         checkPending();\r
 378 \r
 379                         if ( fixForBody\r
 380                                  && ( !currentNode.type || currentNode.name == 'body' )\r
 381                                  && CKEDITOR.tools.trim( text ) )\r
 382                         {\r
 383                                 this.onTagOpen( fixForBody, {}, 0, 1 );\r
 384                         }\r
 385 \r
 386                         // Shrinking consequential spaces into one single for all elements\r
 387                         // text contents.\r
 388                         if ( !inPre )\r
 389                                 text = text.replace( /[\t\r\n ]{2,}|[\t\r\n]/g, ' ' );\r
 390 \r
 391                         currentNode.add( new CKEDITOR.htmlParser.text( text ) );\r
 392                 };\r
 393 \r
 394                 parser.onCDATA = function( cdata )\r
 395                 {\r
 396                         currentNode.add( new CKEDITOR.htmlParser.cdata( cdata ) );\r
 397                 };\r
 398 \r
 399                 parser.onComment = function( comment )\r
 400                 {\r
 401                         sendPendingBRs();\r
 402                         checkPending();\r
 403                         currentNode.add( new CKEDITOR.htmlParser.comment( comment ) );\r
 404                 };\r
 405 \r
 406                 // Parse it.\r
 407                 parser.parse( fragmentHtml );\r
 408 \r
 409                 // Send all pending BRs except one, which we consider a unwanted bogus. (#5293)\r
 410                 sendPendingBRs( !CKEDITOR.env.ie && 1 );\r
 411 \r
 412                 // Close all pending nodes, make sure return point is properly restored.\r
 413                 while ( currentNode != fragment )\r
 414                         addElement( currentNode, currentNode.parent, 1 );\r
 415 \r
 416                 return fragment;\r
 417         };\r
 418 \r
 419         CKEDITOR.htmlParser.fragment.prototype =\r
 420         {\r
 421                 /**\r
 422                  * Adds a node to this fragment.\r
 423                  * @param {Object} node The node to be added. It can be any of of the\r
 424                  *              following types: {@link CKEDITOR.htmlParser.element},\r
 425                  *              {@link CKEDITOR.htmlParser.text} and\r
 426                  *              {@link CKEDITOR.htmlParser.comment}.\r
 427                  *      @param {Number} [index] From where the insertion happens.\r
 428                  * @example\r
 429                  */\r
 430                 add : function( node, index )\r
 431                 {\r
 432                         isNaN( index ) && ( index = this.children.length );\r
 433 \r
 434                         var previous = index > 0 ? this.children[ index - 1 ] : null;\r
 435                         if ( previous )\r
 436                         {\r
 437                                 // If the block to be appended is following text, trim spaces at\r
 438                                 // the right of it.\r
 439                                 if ( node._.isBlockLike && previous.type == CKEDITOR.NODE_TEXT )\r
 440                                 {\r
 441                                         previous.value = CKEDITOR.tools.rtrim( previous.value );\r
 442 \r
 443                                         // If we have completely cleared the previous node.\r
 444                                         if ( previous.value.length === 0 )\r
 445                                         {\r
 446                                                 // Remove it from the list and add the node again.\r
 447                                                 this.children.pop();\r
 448                                                 this.add( node );\r
 449                                                 return;\r
 450                                         }\r
 451                                 }\r
 452 \r
 453                                 previous.next = node;\r
 454                         }\r
 455 \r
 456                         node.previous = previous;\r
 457                         node.parent = this;\r
 458 \r
 459                         this.children.splice( index, 0, node );\r
 460 \r
 461                         this._.hasInlineStarted = node.type == CKEDITOR.NODE_TEXT || ( node.type == CKEDITOR.NODE_ELEMENT && !node._.isBlockLike );\r
 462                 },\r
 463 \r
 464                 /**\r
 465                  * Writes the fragment HTML to a CKEDITOR.htmlWriter.\r
 466                  * @param {CKEDITOR.htmlWriter} writer The writer to which write the HTML.\r
 467                  * @example\r
 468                  * var writer = new CKEDITOR.htmlWriter();\r
 469                  * var fragment = CKEDITOR.htmlParser.fragment.fromHtml( '&lt;P&gt;&lt;B&gt;Example' );\r
 470                  * fragment.writeHtml( writer )\r
 471                  * alert( writer.getHtml() );  "&lt;p&gt;&lt;b&gt;Example&lt;/b&gt;&lt;/p&gt;"\r
 472                  */\r
 473                 writeHtml : function( writer, filter )\r
 474                 {\r
 475                         var isChildrenFiltered;\r
 476                         this.filterChildren = function()\r
 477                         {\r
 478                                 var writer = new CKEDITOR.htmlParser.basicWriter();\r
 479                                 this.writeChildrenHtml.call( this, writer, filter, true );\r
 480                                 var html = writer.getHtml();\r
 481                                 this.children = new CKEDITOR.htmlParser.fragment.fromHtml( html ).children;\r
 482                                 isChildrenFiltered = 1;\r
 483                         };\r
 484 \r
 485                         // Filtering the root fragment before anything else.\r
 486                         !this.name && filter && filter.onFragment( this );\r
 487 \r
 488                         this.writeChildrenHtml( writer, isChildrenFiltered ? null : filter );\r
 489                 },\r
 490 \r
 491                 writeChildrenHtml : function( writer, filter )\r
 492                 {\r
 493                         for ( var i = 0 ; i < this.children.length ; i++ )\r
 494                                 this.children[i].writeHtml( writer, filter );\r
 495                 }\r
 496         };\r
 497 })();\r