_source/core/htmlparser/fragment.js

   1 /*\r
   2 Copyright (c) 2003-2010, 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                 listBlocks = CKEDITOR.dtd.$list, listItems = CKEDITOR.dtd.$listItem;\r
  50 \r
  51         /**\r
  52          * Creates a {@link CKEDITOR.htmlParser.fragment} from an HTML string.\r
  53          * @param {String} fragmentHtml The HTML to be parsed, filling the fragment.\r
  54          * @param {Number} [fixForBody=false] Wrap body with specified element if needed.\r
  55          * @returns CKEDITOR.htmlParser.fragment The fragment created.\r
  56          * @example\r
  57          * var fragment = CKEDITOR.htmlParser.fragment.fromHtml( '<b>Sample</b> Text' );\r
  58          * alert( fragment.children[0].name );  "b"\r
  59          * alert( fragment.children[1].value );  " Text"\r
  60          */\r
  61         CKEDITOR.htmlParser.fragment.fromHtml = function( fragmentHtml, fixForBody )\r
  62         {\r
  63                 var parser = new CKEDITOR.htmlParser(),\r
  64                         html = [],\r
  65                         fragment = new CKEDITOR.htmlParser.fragment(),\r
  66                         pendingInline = [],\r
  67                         pendingBRs = [],\r
  68                         currentNode = fragment,\r
  69                     // Indicate we're inside a <pre> element, spaces should be touched differently.\r
  70                         inPre = false,\r
  71                         returnPoint;\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( brsToIgnore )\r
 112                 {\r
 113                         while ( pendingBRs.length - ( brsToIgnore || 0 ) > 0 )\r
 114                                 currentNode.add( pendingBRs.shift() );\r
 115                 }\r
 116 \r
 117                 function addElement( element, target, enforceCurrent )\r
 118                 {\r
 119                         target = target || currentNode || fragment;\r
 120 \r
 121                         // If the target is the fragment and this element can't go inside\r
 122                         // body (if fixForBody).\r
 123                         if ( fixForBody && !target.type )\r
 124                         {\r
 125                                 var elementName, realElementName;\r
 126                                 if ( element.attributes\r
 127                                          && ( realElementName =\r
 128                                                   element.attributes[ '_cke_real_element_type' ] ) )\r
 129                                         elementName = realElementName;\r
 130                                 else\r
 131                                         elementName =  element.name;\r
 132                                 if ( elementName\r
 133                                                 && !( elementName in CKEDITOR.dtd.$body )\r
 134                                                 && !( elementName in CKEDITOR.dtd.$nonBodyContent )  )\r
 135                                 {\r
 136                                         var savedCurrent = currentNode;\r
 137 \r
 138                                         // Create a <p> in the fragment.\r
 139                                         currentNode = target;\r
 140                                         parser.onTagOpen( fixForBody, {} );\r
 141 \r
 142                                         // The new target now is the <p>.\r
 143                                         target = currentNode;\r
 144 \r
 145                                         if ( enforceCurrent )\r
 146                                                 currentNode = savedCurrent;\r
 147                                 }\r
 148                         }\r
 149 \r
 150                         // Rtrim empty spaces on block end boundary. (#3585)\r
 151                         if ( element._.isBlockLike\r
 152                                  && element.name != 'pre' )\r
 153                         {\r
 154 \r
 155                                 var length = element.children.length,\r
 156                                         lastChild = element.children[ length - 1 ],\r
 157                                         text;\r
 158                                 if ( lastChild && lastChild.type == CKEDITOR.NODE_TEXT )\r
 159                                 {\r
 160                                         if ( !( text = CKEDITOR.tools.rtrim( lastChild.value ) ) )\r
 161                                                 element.children.length = length -1;\r
 162                                         else\r
 163                                                 lastChild.value = text;\r
 164                                 }\r
 165                         }\r
 166 \r
 167                         target.add( element );\r
 168 \r
 169                         if ( element.returnPoint )\r
 170                         {\r
 171                                 currentNode = element.returnPoint;\r
 172                                 delete element.returnPoint;\r
 173                         }\r
 174                 }\r
 175 \r
 176                 parser.onTagOpen = function( tagName, attributes, selfClosing )\r
 177                 {\r
 178                         var element = new CKEDITOR.htmlParser.element( tagName, attributes );\r
 179 \r
 180                         // "isEmpty" will be always "false" for unknown elements, so we\r
 181                         // must force it if the parser has identified it as a selfClosing tag.\r
 182                         if ( element.isUnknown && selfClosing )\r
 183                                 element.isEmpty = true;\r
 184 \r
 185                         // This is a tag to be removed if empty, so do not add it immediately.\r
 186                         if ( CKEDITOR.dtd.$removeEmpty[ tagName ] )\r
 187                         {\r
 188                                 pendingInline.push( element );\r
 189                                 return;\r
 190                         }\r
 191                         else if ( tagName == 'pre' )\r
 192                                 inPre = true;\r
 193                         else if ( tagName == 'br' && inPre )\r
 194                         {\r
 195                                 currentNode.add( new CKEDITOR.htmlParser.text( '\n' ) );\r
 196                                 return;\r
 197                         }\r
 198 \r
 199                         if ( tagName == 'br' )\r
 200                         {\r
 201                                 pendingBRs.push( element );\r
 202                                 return;\r
 203                         }\r
 204 \r
 205                         var currentName = currentNode.name;\r
 206 \r
 207                         var currentDtd = currentName\r
 208                                 && ( CKEDITOR.dtd[ currentName ]\r
 209                                         || ( currentNode._.isBlockLike ? CKEDITOR.dtd.div : CKEDITOR.dtd.span ) );\r
 210 \r
 211                         // If the element cannot be child of the current element.\r
 212                         if ( currentDtd   // Fragment could receive any elements.\r
 213                                  && !element.isUnknown && !currentNode.isUnknown && !currentDtd[ tagName ] )\r
 214                         {\r
 215 \r
 216                                 var reApply = false,\r
 217                                         addPoint;   // New position to start adding nodes.\r
 218 \r
 219                                 // Fixing malformed nested lists by moving it into a previous list item. (#3828)\r
 220                                 if ( tagName in listBlocks\r
 221                                         && currentName in listBlocks )\r
 222                                 {\r
 223                                         var children = currentNode.children,\r
 224                                                 lastChild = children[ children.length - 1 ];\r
 225 \r
 226                                         // Establish the list item if it's not existed.\r
 227                                         if ( !( lastChild && lastChild.name in listItems ) )\r
 228                                                 addElement( ( lastChild = new CKEDITOR.htmlParser.element( 'li' ) ), currentNode );\r
 229 \r
 230                                         returnPoint = currentNode, addPoint = lastChild;\r
 231                                 }\r
 232                                 // If the element name is the same as the current element name,\r
 233                                 // then just close the current one and append the new one to the\r
 234                                 // parent. This situation usually happens with <p>, <li>, <dt> and\r
 235                                 // <dd>, specially in IE. Do not enter in this if block in this case.\r
 236                                 else if ( tagName == currentName )\r
 237                                 {\r
 238                                         addElement( currentNode, currentNode.parent );\r
 239                                 }\r
 240                                 else if ( tagName in CKEDITOR.dtd.$listItem )\r
 241                                 {\r
 242                                         parser.onTagOpen( 'ul', {} );\r
 243                                         addPoint = currentNode;\r
 244                                         reApply = true;\r
 245                                 }\r
 246                                 else\r
 247                                 {\r
 248                                         if ( nonBreakingBlocks[ currentName ] )\r
 249                                         {\r
 250                                                 if ( !returnPoint )\r
 251                                                         returnPoint = currentNode;\r
 252                                         }\r
 253                                         else\r
 254                                         {\r
 255                                                 addElement( currentNode, currentNode.parent, true );\r
 256 \r
 257                                                 if ( !optionalClose[ currentName ] )\r
 258                                                 {\r
 259                                                         // The current element is an inline element, which\r
 260                                                         // cannot hold the new one. Put it in the pending list,\r
 261                                                         // and try adding the new one after it.\r
 262                                                         pendingInline.unshift( currentNode );\r
 263                                                 }\r
 264                                         }\r
 265 \r
 266                                         reApply = true;\r
 267                                 }\r
 268 \r
 269                                 if ( addPoint )\r
 270                                         currentNode = addPoint;\r
 271                                 // Try adding it to the return point, or the parent element.\r
 272                                 else\r
 273                                         currentNode = currentNode.returnPoint || currentNode.parent;\r
 274 \r
 275                                 if ( reApply )\r
 276                                 {\r
 277                                         parser.onTagOpen.apply( this, arguments );\r
 278                                         return;\r
 279                                 }\r
 280                         }\r
 281 \r
 282                         checkPending( tagName );\r
 283                         sendPendingBRs();\r
 284 \r
 285                         element.parent = currentNode;\r
 286                         element.returnPoint = returnPoint;\r
 287                         returnPoint = 0;\r
 288 \r
 289                         if ( element.isEmpty )\r
 290                                 addElement( element );\r
 291                         else\r
 292                                 currentNode = element;\r
 293                 };\r
 294 \r
 295                 parser.onTagClose = function( tagName )\r
 296                 {\r
 297                         // Check if there is any pending tag to be closed.\r
 298                         for ( var i = pendingInline.length - 1 ; i >= 0 ; i-- )\r
 299                         {\r
 300                                 // If found, just remove it from the list.\r
 301                                 if ( tagName == pendingInline[ i ].name )\r
 302                                 {\r
 303                                         pendingInline.splice( i, 1 );\r
 304                                         return;\r
 305                                 }\r
 306                         }\r
 307 \r
 308                         var pendingAdd = [],\r
 309                                 newPendingInline = [],\r
 310                                 candidate = currentNode;\r
 311 \r
 312                         while ( candidate.type && candidate.name != tagName )\r
 313                         {\r
 314                                 // If this is an inline element, add it to the pending list, if we're\r
 315                                 // really closing one of the parents element later, they will continue\r
 316                                 // after it.\r
 317                                 if ( !candidate._.isBlockLike )\r
 318                                         newPendingInline.unshift( candidate );\r
 319 \r
 320                                 // This node should be added to it's parent at this point. But,\r
 321                                 // it should happen only if the closing tag is really closing\r
 322                                 // one of the nodes. So, for now, we just cache it.\r
 323                                 pendingAdd.push( candidate );\r
 324 \r
 325                                 candidate = candidate.parent;\r
 326                         }\r
 327 \r
 328                         if ( candidate.type )\r
 329                         {\r
 330                                 // Add all elements that have been found in the above loop.\r
 331                                 for ( i = 0 ; i < pendingAdd.length ; i++ )\r
 332                                 {\r
 333                                         var node = pendingAdd[ i ];\r
 334                                         addElement( node, node.parent );\r
 335                                 }\r
 336 \r
 337                                 currentNode = candidate;\r
 338 \r
 339                                 if ( currentNode.name == 'pre' )\r
 340                                         inPre = false;\r
 341 \r
 342                                 if ( candidate._.isBlockLike )\r
 343                                         sendPendingBRs();\r
 344 \r
 345                                 addElement( candidate, candidate.parent );\r
 346 \r
 347                                 // The parent should start receiving new nodes now, except if\r
 348                                 // addElement changed the currentNode.\r
 349                                 if ( candidate == currentNode )\r
 350                                         currentNode = currentNode.parent;\r
 351 \r
 352                                 pendingInline = pendingInline.concat( newPendingInline );\r
 353                         }\r
 354 \r
 355                         if ( tagName == 'body' )\r
 356                                 fixForBody = false;\r
 357                 };\r
 358 \r
 359                 parser.onText = function( text )\r
 360                 {\r
 361                         // Trim empty spaces at beginning of element contents except <pre>.\r
 362                         if ( !currentNode._.hasInlineStarted && !inPre )\r
 363                         {\r
 364                                 text = CKEDITOR.tools.ltrim( text );\r
 365 \r
 366                                 if ( text.length === 0 )\r
 367                                         return;\r
 368                         }\r
 369 \r
 370                         sendPendingBRs();\r
 371                         checkPending();\r
 372 \r
 373                         if ( fixForBody\r
 374                                  && ( !currentNode.type || currentNode.name == 'body' )\r
 375                                  && CKEDITOR.tools.trim( text ) )\r
 376                         {\r
 377                                 this.onTagOpen( fixForBody, {} );\r
 378                         }\r
 379 \r
 380                         // Shrinking consequential spaces into one single for all elements\r
 381                         // text contents.\r
 382                         if ( !inPre )\r
 383                                 text = text.replace( /[\t\r\n ]{2,}|[\t\r\n]/g, ' ' );\r
 384 \r
 385                         currentNode.add( new CKEDITOR.htmlParser.text( text ) );\r
 386                 };\r
 387 \r
 388                 parser.onCDATA = function( cdata )\r
 389                 {\r
 390                         currentNode.add( new CKEDITOR.htmlParser.cdata( cdata ) );\r
 391                 };\r
 392 \r
 393                 parser.onComment = function( comment )\r
 394                 {\r
 395                         checkPending();\r
 396                         currentNode.add( new CKEDITOR.htmlParser.comment( comment ) );\r
 397                 };\r
 398 \r
 399                 // Parse it.\r
 400                 parser.parse( fragmentHtml );\r
 401 \r
 402                 // Send all pending BRs except one, which we consider a unwanted bogus. (#5293)\r
 403                 sendPendingBRs( !CKEDITOR.env.ie && 1 );\r
 404 \r
 405                 // Close all pending nodes.\r
 406                 while ( currentNode.type )\r
 407                 {\r
 408                         var parent = currentNode.parent,\r
 409                                 node = currentNode;\r
 410 \r
 411                         if ( fixForBody\r
 412                                  && ( !parent.type || parent.name == 'body' )\r
 413                                  && !CKEDITOR.dtd.$body[ node.name ] )\r
 414                         {\r
 415                                 currentNode = parent;\r
 416                                 parser.onTagOpen( fixForBody, {} );\r
 417                                 parent = currentNode;\r
 418                         }\r
 419 \r
 420                         parent.add( node );\r
 421                         currentNode = parent;\r
 422                 }\r
 423 \r
 424                 return fragment;\r
 425         };\r
 426 \r
 427         CKEDITOR.htmlParser.fragment.prototype =\r
 428         {\r
 429                 /**\r
 430                  * Adds a node to this fragment.\r
 431                  * @param {Object} node The node to be added. It can be any of of the\r
 432                  *              following types: {@link CKEDITOR.htmlParser.element},\r
 433                  *              {@link CKEDITOR.htmlParser.text} and\r
 434                  *              {@link CKEDITOR.htmlParser.comment}.\r
 435                  * @example\r
 436                  */\r
 437                 add : function( node )\r
 438                 {\r
 439                         var len = this.children.length,\r
 440                                 previous = len > 0 && this.children[ len - 1 ] || null;\r
 441 \r
 442                         if ( previous )\r
 443                         {\r
 444                                 // If the block to be appended is following text, trim spaces at\r
 445                                 // the right of it.\r
 446                                 if ( node._.isBlockLike && previous.type == CKEDITOR.NODE_TEXT )\r
 447                                 {\r
 448                                         previous.value = CKEDITOR.tools.rtrim( previous.value );\r
 449 \r
 450                                         // If we have completely cleared the previous node.\r
 451                                         if ( previous.value.length === 0 )\r
 452                                         {\r
 453                                                 // Remove it from the list and add the node again.\r
 454                                                 this.children.pop();\r
 455                                                 this.add( node );\r
 456                                                 return;\r
 457                                         }\r
 458                                 }\r
 459 \r
 460                                 previous.next = node;\r
 461                         }\r
 462 \r
 463                         node.previous = previous;\r
 464                         node.parent = this;\r
 465 \r
 466                         this.children.push( node );\r
 467 \r
 468                         this._.hasInlineStarted = node.type == CKEDITOR.NODE_TEXT || ( node.type == CKEDITOR.NODE_ELEMENT && !node._.isBlockLike );\r
 469                 },\r
 470 \r
 471                 /**\r
 472                  * Writes the fragment HTML to a CKEDITOR.htmlWriter.\r
 473                  * @param {CKEDITOR.htmlWriter} writer The writer to which write the HTML.\r
 474                  * @example\r
 475                  * var writer = new CKEDITOR.htmlWriter();\r
 476                  * var fragment = CKEDITOR.htmlParser.fragment.fromHtml( '&lt;P&gt;&lt;B&gt;Example' );\r
 477                  * fragment.writeHtml( writer )\r
 478                  * alert( writer.getHtml() );  "&lt;p&gt;&lt;b&gt;Example&lt;/b&gt;&lt;/p&gt;"\r
 479                  */\r
 480                 writeHtml : function( writer, filter )\r
 481                 {\r
 482                         var isChildrenFiltered;\r
 483                         this.filterChildren = function()\r
 484                         {\r
 485                                 var writer = new CKEDITOR.htmlParser.basicWriter();\r
 486                                 this.writeChildrenHtml.call( this, writer, filter, true );\r
 487                                 var html = writer.getHtml();\r
 488                                 this.children = new CKEDITOR.htmlParser.fragment.fromHtml( html ).children;\r
 489                                 isChildrenFiltered = 1;\r
 490                         };\r
 491 \r
 492                         // Filtering the root fragment before anything else.\r
 493                         !this.name && filter && filter.onFragment( this );\r
 494 \r
 495                         this.writeChildrenHtml( writer, isChildrenFiltered ? null : filter );\r
 496                 },\r
 497 \r
 498                 writeChildrenHtml : function( writer, filter )\r
 499                 {\r
 500                         for ( var i = 0 ; i < this.children.length ; i++ )\r
 501                                 this.children[i].writeHtml( writer, filter );\r
 502                 }\r
 503         };\r
 504 })();\r