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