_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         function isRemoveEmpty( node )\r
  53         {\r
  54                 // Empty link is to be removed when empty but not anchor. (#7894)\r
  55                 return node.name == 'a' && node.attributes.href\r
  56                         || CKEDITOR.dtd.$removeEmpty[ node.name ];\r
  57         }\r
  58 \r
  59         /**\r
  60          * Creates a {@link CKEDITOR.htmlParser.fragment} from an HTML string.\r
  61          * @param {String} fragmentHtml The HTML to be parsed, filling the fragment.\r
  62          * @param {Number} [fixForBody=false] Wrap body with specified element if needed.\r
  63          * @param {CKEDITOR.htmlParser.element} contextNode Parse the html as the content of this element.\r
  64          * @returns CKEDITOR.htmlParser.fragment The fragment created.\r
  65          * @example\r
  66          * var fragment = CKEDITOR.htmlParser.fragment.fromHtml( '<b>Sample</b> Text' );\r
  67          * alert( fragment.children[0].name );  "b"\r
  68          * alert( fragment.children[1].value );  " Text"\r
  69          */\r
  70         CKEDITOR.htmlParser.fragment.fromHtml = function( fragmentHtml, fixForBody, contextNode )\r
  71         {\r
  72                 var parser = new CKEDITOR.htmlParser(),\r
  73                         fragment = contextNode || new CKEDITOR.htmlParser.fragment(),\r
  74                         pendingInline = [],\r
  75                         pendingBRs = [],\r
  76                         currentNode = fragment,\r
  77                     // Indicate we're inside a <textarea> element, spaces should be touched differently.\r
  78                         inTextarea = false,\r
  79                     // Indicate we're inside a <pre> element, spaces should be touched differently.\r
  80                         inPre = false;\r
  81 \r
  82                 function checkPending( newTagName )\r
  83                 {\r
  84                         var pendingBRsSent;\r
  85 \r
  86                         if ( pendingInline.length > 0 )\r
  87                         {\r
  88                                 for ( var i = 0 ; i < pendingInline.length ; i++ )\r
  89                                 {\r
  90                                         var pendingElement = pendingInline[ i ],\r
  91                                                 pendingName = pendingElement.name,\r
  92                                                 pendingDtd = CKEDITOR.dtd[ pendingName ],\r
  93                                                 currentDtd = currentNode.name && CKEDITOR.dtd[ currentNode.name ];\r
  94 \r
  95                                         if ( ( !currentDtd || currentDtd[ pendingName ] ) && ( !newTagName || !pendingDtd || pendingDtd[ newTagName ] || !CKEDITOR.dtd[ newTagName ] ) )\r
  96                                         {\r
  97                                                 if ( !pendingBRsSent )\r
  98                                                 {\r
  99                                                         sendPendingBRs();\r
 100                                                         pendingBRsSent = 1;\r
 101                                                 }\r
 102 \r
 103                                                 // Get a clone for the pending element.\r
 104                                                 pendingElement = pendingElement.clone();\r
 105 \r
 106                                                 // Add it to the current node and make it the current,\r
 107                                                 // so the new element will be added inside of it.\r
 108                                                 pendingElement.parent = currentNode;\r
 109                                                 currentNode = pendingElement;\r
 110 \r
 111                                                 // Remove the pending element (back the index by one\r
 112                                                 // to properly process the next entry).\r
 113                                                 pendingInline.splice( i, 1 );\r
 114                                                 i--;\r
 115                                         }\r
 116                                         else\r
 117                                         {\r
 118                                                 // Some element of the same type cannot be nested, flat them,\r
 119                                                 // e.g. <a href="#">foo<a href="#">bar</a></a>. (#7894)\r
 120                                                 if ( pendingName == currentNode.name )\r
 121                                                         addElement( currentNode, currentNode.parent, 1 ), i--;\r
 122                                         }\r
 123                                 }\r
 124                         }\r
 125                 }\r
 126 \r
 127                 function sendPendingBRs()\r
 128                 {\r
 129                         while ( pendingBRs.length )\r
 130                                 currentNode.add( pendingBRs.shift() );\r
 131                 }\r
 132 \r
 133                 /*\r
 134                 * Beside of simply append specified element to target, this function also takes\r
 135                 * care of other dirty lifts like forcing block in body, trimming spaces at\r
 136                 * the block boundaries etc.\r
 137                 *\r
 138                 * @param {Element} element  The element to be added as the last child of {@link target}.\r
 139                 * @param {Element} target The parent element to relieve the new node.\r
 140                 * @param {Boolean} [moveCurrent=false] Don't change the "currentNode" global unless\r
 141                 * there's a return point node specified on the element, otherwise move current onto {@link target} node.\r
 142                  */\r
 143                 function addElement( element, target, moveCurrent )\r
 144                 {\r
 145                         // Ignore any element that has already been added.\r
 146                         if ( element.previous !== undefined )\r
 147                                 return;\r
 148 \r
 149                         target = target || currentNode || fragment;\r
 150 \r
 151                         // Current element might be mangled by fix body below,\r
 152                         // save it for restore later.\r
 153                         var savedCurrent = currentNode;\r
 154 \r
 155                         // If the target is the fragment and this inline element can't go inside\r
 156                         // body (if fixForBody).\r
 157                         if ( fixForBody && ( !target.type || target.name == 'body' ) )\r
 158                         {\r
 159                                 var elementName, realElementName;\r
 160                                 if ( element.attributes\r
 161                                          && ( realElementName =\r
 162                                                   element.attributes[ 'data-cke-real-element-type' ] ) )\r
 163                                         elementName = realElementName;\r
 164                                 else\r
 165                                         elementName =  element.name;\r
 166 \r
 167                                 if ( elementName && !( elementName in CKEDITOR.dtd.$body || elementName == 'body' || element.isOrphan ) )\r
 168                                 {\r
 169                                         // Create a <p> in the fragment.\r
 170                                         currentNode = target;\r
 171                                         parser.onTagOpen( fixForBody, {} );\r
 172 \r
 173                                         // The new target now is the <p>.\r
 174                                         element.returnPoint = target = currentNode;\r
 175                                 }\r
 176                         }\r
 177 \r
 178                         // Rtrim empty spaces on block end boundary. (#3585)\r
 179                         if ( element._.isBlockLike\r
 180                                  && element.name != 'pre' && element.name != 'textarea' )\r
 181                         {\r
 182 \r
 183                                 var length = element.children.length,\r
 184                                         lastChild = element.children[ length - 1 ],\r
 185                                         text;\r
 186                                 if ( lastChild && lastChild.type == CKEDITOR.NODE_TEXT )\r
 187                                 {\r
 188                                         if ( !( text = CKEDITOR.tools.rtrim( lastChild.value ) ) )\r
 189                                                 element.children.length = length -1;\r
 190                                         else\r
 191                                                 lastChild.value = text;\r
 192                                 }\r
 193                         }\r
 194 \r
 195                         target.add( element );\r
 196 \r
 197                         if ( element.returnPoint )\r
 198                         {\r
 199                                 currentNode = element.returnPoint;\r
 200                                 delete element.returnPoint;\r
 201                         }\r
 202                         else\r
 203                                 currentNode = moveCurrent ? target : savedCurrent;\r
 204                 }\r
 205 \r
 206                 parser.onTagOpen = function( tagName, attributes, selfClosing, optionalClose )\r
 207                 {\r
 208                         var element = new CKEDITOR.htmlParser.element( tagName, attributes );\r
 209 \r
 210                         // "isEmpty" will be always "false" for unknown elements, so we\r
 211                         // must force it if the parser has identified it as a selfClosing tag.\r
 212                         if ( element.isUnknown && selfClosing )\r
 213                                 element.isEmpty = true;\r
 214 \r
 215                         // Check for optional closed elements, including browser quirks and manually opened blocks.\r
 216                         element.isOptionalClose = tagName in optionalCloseTags || optionalClose;\r
 217 \r
 218                         // This is a tag to be removed if empty, so do not add it immediately.\r
 219                         if ( isRemoveEmpty( element ) )\r
 220                         {\r
 221                                 pendingInline.push( element );\r
 222                                 return;\r
 223                         }\r
 224                         else if ( tagName == 'pre' )\r
 225                                 inPre = true;\r
 226                         else if ( tagName == 'br' && inPre )\r
 227                         {\r
 228                                 currentNode.add( new CKEDITOR.htmlParser.text( '\n' ) );\r
 229                                 return;\r
 230                         }\r
 231                         else if ( tagName == 'textarea' )\r
 232                                 inTextarea = true;\r
 233 \r
 234                         if ( tagName == 'br' )\r
 235                         {\r
 236                                 pendingBRs.push( element );\r
 237                                 return;\r
 238                         }\r
 239 \r
 240                         while( 1 )\r
 241                         {\r
 242                                 var currentName = currentNode.name;\r
 243 \r
 244                                 var currentDtd = currentName ? ( CKEDITOR.dtd[ currentName ]\r
 245                                                 || ( currentNode._.isBlockLike ? CKEDITOR.dtd.div : CKEDITOR.dtd.span ) )\r
 246                                                 : rootDtd;\r
 247 \r
 248                                 // If the element cannot be child of the current element.\r
 249                                 if ( !element.isUnknown && !currentNode.isUnknown && !currentDtd[ tagName ] )\r
 250                                 {\r
 251                                         // Current node doesn't have a close tag, time for a close\r
 252                                         // as this element isn't fit in. (#7497)\r
 253                                         if ( currentNode.isOptionalClose )\r
 254                                                 parser.onTagClose( currentName );\r
 255                                         // Fixing malformed nested lists by moving it into a previous list item. (#3828)\r
 256                                         else if ( tagName in listBlocks\r
 257                                                 && currentName in listBlocks )\r
 258                                         {\r
 259                                                 var children = currentNode.children,\r
 260                                                         lastChild = children[ children.length - 1 ];\r
 261 \r
 262                                                 // Establish the list item if it's not existed.\r
 263                                                 if ( !( lastChild && lastChild.name == 'li' ) )\r
 264                                                         addElement( ( lastChild = new CKEDITOR.htmlParser.element( 'li' ) ), currentNode );\r
 265 \r
 266                                                 !element.returnPoint && ( element.returnPoint = currentNode );\r
 267                                                 currentNode = lastChild;\r
 268                                         }\r
 269                                         // Establish new list root for orphan list items.\r
 270                                         else if ( tagName in CKEDITOR.dtd.$listItem && currentName != tagName )\r
 271                                                 parser.onTagOpen( tagName == 'li' ? 'ul' : 'dl', {}, 0, 1 );\r
 272                                         // We're inside a structural block like table and list, AND the incoming element\r
 273                                         // is not of the same type (e.g. <td>td1<td>td2</td>), we simply add this new one before it,\r
 274                                         // and most importantly, return back to here once this element is added,\r
 275                                         // e.g. <table><tr><td>td1</td><p>p1</p><td>td2</td></tr></table>\r
 276                                         else if ( currentName in nonBreakingBlocks && currentName != tagName )\r
 277                                         {\r
 278                                                 !element.returnPoint && ( element.returnPoint = currentNode );\r
 279                                                 currentNode = currentNode.parent;\r
 280                                         }\r
 281                                         else\r
 282                                         {\r
 283                                                 // The current element is an inline element, which\r
 284                                                 // need to be continued even after the close, so put\r
 285                                                 // it in the pending list.\r
 286                                                 if ( currentName in CKEDITOR.dtd.$inline )\r
 287                                                         pendingInline.unshift( currentNode );\r
 288 \r
 289                                                 // The most common case where we just need to close the\r
 290                                                 // current one and append the new one to the parent.\r
 291                                                 if ( currentNode.parent )\r
 292                                                         addElement( currentNode, currentNode.parent, 1 );\r
 293                                                 // We've tried our best to fix the embarrassment here, while\r
 294                                                 // this element still doesn't find it's parent, mark it as\r
 295                                                 // orphan and show our tolerance to it.\r
 296                                                 else\r
 297                                                 {\r
 298                                                         element.isOrphan = 1;\r
 299                                                         break;\r
 300                                                 }\r
 301                                         }\r
 302                                 }\r
 303                                 else\r
 304                                         break;\r
 305                         }\r
 306 \r
 307                         checkPending( tagName );\r
 308                         sendPendingBRs();\r
 309 \r
 310                         element.parent = currentNode;\r
 311 \r
 312                         if ( element.isEmpty )\r
 313                                 addElement( element );\r
 314                         else\r
 315                                 currentNode = element;\r
 316                 };\r
 317 \r
 318                 parser.onTagClose = function( tagName )\r
 319                 {\r
 320                         // Check if there is any pending tag to be closed.\r
 321                         for ( var i = pendingInline.length - 1 ; i >= 0 ; i-- )\r
 322                         {\r
 323                                 // If found, just remove it from the list.\r
 324                                 if ( tagName == pendingInline[ i ].name )\r
 325                                 {\r
 326                                         pendingInline.splice( i, 1 );\r
 327                                         return;\r
 328                                 }\r
 329                         }\r
 330 \r
 331                         var pendingAdd = [],\r
 332                                 newPendingInline = [],\r
 333                                 candidate = currentNode;\r
 334 \r
 335                         while ( candidate != fragment && candidate.name != tagName )\r
 336                         {\r
 337                                 // If this is an inline element, add it to the pending list, if we're\r
 338                                 // really closing one of the parents element later, they will continue\r
 339                                 // after it.\r
 340                                 if ( !candidate._.isBlockLike )\r
 341                                         newPendingInline.unshift( candidate );\r
 342 \r
 343                                 // This node should be added to it's parent at this point. But,\r
 344                                 // it should happen only if the closing tag is really closing\r
 345                                 // one of the nodes. So, for now, we just cache it.\r
 346                                 pendingAdd.push( candidate );\r
 347 \r
 348                                 // Make sure return point is properly restored.\r
 349                                 candidate = candidate.returnPoint || candidate.parent;\r
 350                         }\r
 351 \r
 352                         if ( candidate != fragment )\r
 353                         {\r
 354                                 // Add all elements that have been found in the above loop.\r
 355                                 for ( i = 0 ; i < pendingAdd.length ; i++ )\r
 356                                 {\r
 357                                         var node = pendingAdd[ i ];\r
 358                                         addElement( node, node.parent );\r
 359                                 }\r
 360 \r
 361                                 currentNode = candidate;\r
 362 \r
 363                                 if ( currentNode.name == 'pre' )\r
 364                                         inPre = false;\r
 365 \r
 366                                 if ( currentNode.name == 'textarea' )\r
 367                                         inTextarea = false;\r
 368 \r
 369                                 if ( candidate._.isBlockLike )\r
 370                                         sendPendingBRs();\r
 371 \r
 372                                 addElement( candidate, candidate.parent );\r
 373 \r
 374                                 // The parent should start receiving new nodes now, except if\r
 375                                 // addElement changed the currentNode.\r
 376                                 if ( candidate == currentNode )\r
 377                                         currentNode = currentNode.parent;\r
 378 \r
 379                                 pendingInline = pendingInline.concat( newPendingInline );\r
 380                         }\r
 381 \r
 382                         if ( tagName == 'body' )\r
 383                                 fixForBody = false;\r
 384                 };\r
 385 \r
 386                 parser.onText = function( text )\r
 387                 {\r
 388                         // Trim empty spaces at beginning of text contents except <pre> and <textarea>.\r
 389                         if ( ( !currentNode._.hasInlineStarted || pendingBRs.length ) && !inPre && !inTextarea )\r
 390                         {\r
 391                                 text = CKEDITOR.tools.ltrim( text );\r
 392 \r
 393                                 if ( text.length === 0 )\r
 394                                         return;\r
 395                         }\r
 396 \r
 397                         sendPendingBRs();\r
 398                         checkPending();\r
 399 \r
 400                         if ( fixForBody\r
 401                                  && ( !currentNode.type || currentNode.name == 'body' )\r
 402                                  && CKEDITOR.tools.trim( text ) )\r
 403                         {\r
 404                                 this.onTagOpen( fixForBody, {}, 0, 1 );\r
 405                         }\r
 406 \r
 407                         // Shrinking consequential spaces into one single for all elements\r
 408                         // text contents.\r
 409                         if ( !inPre && !inTextarea )\r
 410                                 text = text.replace( /[\t\r\n ]{2,}|[\t\r\n]/g, ' ' );\r
 411 \r
 412                         currentNode.add( new CKEDITOR.htmlParser.text( text ) );\r
 413                 };\r
 414 \r
 415                 parser.onCDATA = function( cdata )\r
 416                 {\r
 417                         currentNode.add( new CKEDITOR.htmlParser.cdata( cdata ) );\r
 418                 };\r
 419 \r
 420                 parser.onComment = function( comment )\r
 421                 {\r
 422                         sendPendingBRs();\r
 423                         checkPending();\r
 424                         currentNode.add( new CKEDITOR.htmlParser.comment( comment ) );\r
 425                 };\r
 426 \r
 427                 // Parse it.\r
 428                 parser.parse( fragmentHtml );\r
 429 \r
 430                 // Send all pending BRs except one, which we consider a unwanted bogus. (#5293)\r
 431                 sendPendingBRs( !CKEDITOR.env.ie && 1 );\r
 432 \r
 433                 // Close all pending nodes, make sure return point is properly restored.\r
 434                 while ( currentNode != fragment )\r
 435                         addElement( currentNode, currentNode.parent, 1 );\r
 436 \r
 437                 return fragment;\r
 438         };\r
 439 \r
 440         CKEDITOR.htmlParser.fragment.prototype =\r
 441         {\r
 442                 /**\r
 443                  * Adds a node to this fragment.\r
 444                  * @param {Object} node The node to be added. It can be any of of the\r
 445                  *              following types: {@link CKEDITOR.htmlParser.element},\r
 446                  *              {@link CKEDITOR.htmlParser.text} and\r
 447                  *              {@link CKEDITOR.htmlParser.comment}.\r
 448                  *      @param {Number} [index] From where the insertion happens.\r
 449                  * @example\r
 450                  */\r
 451                 add : function( node, index )\r
 452                 {\r
 453                         isNaN( index ) && ( index = this.children.length );\r
 454 \r
 455                         var previous = index > 0 ? this.children[ index - 1 ] : null;\r
 456                         if ( previous )\r
 457                         {\r
 458                                 // If the block to be appended is following text, trim spaces at\r
 459                                 // the right of it.\r
 460                                 if ( node._.isBlockLike && previous.type == CKEDITOR.NODE_TEXT )\r
 461                                 {\r
 462                                         previous.value = CKEDITOR.tools.rtrim( previous.value );\r
 463 \r
 464                                         // If we have completely cleared the previous node.\r
 465                                         if ( previous.value.length === 0 )\r
 466                                         {\r
 467                                                 // Remove it from the list and add the node again.\r
 468                                                 this.children.pop();\r
 469                                                 this.add( node );\r
 470                                                 return;\r
 471                                         }\r
 472                                 }\r
 473 \r
 474                                 previous.next = node;\r
 475                         }\r
 476 \r
 477                         node.previous = previous;\r
 478                         node.parent = this;\r
 479 \r
 480                         this.children.splice( index, 0, node );\r
 481 \r
 482                         this._.hasInlineStarted = node.type == CKEDITOR.NODE_TEXT || ( node.type == CKEDITOR.NODE_ELEMENT && !node._.isBlockLike );\r
 483                 },\r
 484 \r
 485                 /**\r
 486                  * Writes the fragment HTML to a CKEDITOR.htmlWriter.\r
 487                  * @param {CKEDITOR.htmlWriter} writer The writer to which write the HTML.\r
 488                  * @example\r
 489                  * var writer = new CKEDITOR.htmlWriter();\r
 490                  * var fragment = CKEDITOR.htmlParser.fragment.fromHtml( '&lt;P&gt;&lt;B&gt;Example' );\r
 491                  * fragment.writeHtml( writer )\r
 492                  * alert( writer.getHtml() );  "&lt;p&gt;&lt;b&gt;Example&lt;/b&gt;&lt;/p&gt;"\r
 493                  */\r
 494                 writeHtml : function( writer, filter )\r
 495                 {\r
 496                         var isChildrenFiltered;\r
 497                         this.filterChildren = function()\r
 498                         {\r
 499                                 var writer = new CKEDITOR.htmlParser.basicWriter();\r
 500                                 this.writeChildrenHtml.call( this, writer, filter, true );\r
 501                                 var html = writer.getHtml();\r
 502                                 this.children = new CKEDITOR.htmlParser.fragment.fromHtml( html ).children;\r
 503                                 isChildrenFiltered = 1;\r
 504                         };\r
 505 \r
 506                         // Filtering the root fragment before anything else.\r
 507                         !this.name && filter && filter.onFragment( this );\r
 508 \r
 509                         this.writeChildrenHtml( writer, isChildrenFiltered ? null : filter );\r
 510                 },\r
 511 \r
 512                 writeChildrenHtml : function( writer, filter )\r
 513                 {\r
 514                         for ( var i = 0 ; i < this.children.length ; i++ )\r
 515                                 this.children[i].writeHtml( writer, filter );\r
 516                 }\r
 517         };\r
 518 })();\r