_source/plugins/htmldataprocessor/plugin.js

   1 /*
   2 Copyright (c) 2003-2011, CKSource - Frederico Knabben. All rights reserved.
   3 For licensing, see LICENSE.html or http://ckeditor.com/license
   4 */
   5
   6 (function()
   7 {
   8         // Regex to scan for &nbsp; at the end of blocks, which are actually placeholders.
   9         // Safari transforms the &nbsp; to \xa0. (#4172)
  10         var tailNbspRegex = /^[\t\r\n ]*(?:&nbsp;|\xa0)$/;
  11
  12         var protectedSourceMarker = '{cke_protected}';
  13
  14         // Return the last non-space child node of the block (#4344).
  15         function lastNoneSpaceChild( block )
  16         {
  17                 var lastIndex = block.children.length,
  18                         last = block.children[ lastIndex - 1 ];
  19                 while (  last && last.type == CKEDITOR.NODE_TEXT && !CKEDITOR.tools.trim( last.value ) )
  20                         last = block.children[ --lastIndex ];
  21                 return last;
  22         }
  23
  24         function trimFillers( block, fromSource )
  25         {
  26                 // If the current node is a block, and if we're converting from source or
  27                 // we're not in IE then search for and remove any tailing BR node.
  28                 //
  29                 // Also, any &nbsp; at the end of blocks are fillers, remove them as well.
  30                 // (#2886)
  31                 var children = block.children, lastChild = lastNoneSpaceChild( block );
  32                 if ( lastChild )
  33                 {
  34                         if ( ( fromSource || !CKEDITOR.env.ie ) && lastChild.type == CKEDITOR.NODE_ELEMENT && lastChild.name == 'br' )
  35                                 children.pop();
  36                         if ( lastChild.type == CKEDITOR.NODE_TEXT && tailNbspRegex.test( lastChild.value ) )
  37                                 children.pop();
  38                 }
  39         }
  40
  41         function blockNeedsExtension( block, fromSource, extendEmptyBlock )
  42         {
  43                 if( !fromSource && ( !extendEmptyBlock ||
  44                         typeof extendEmptyBlock == 'function' && ( extendEmptyBlock( block ) === false ) ) )
  45                         return false;
  46
  47         // 1. For IE version >=8,  empty blocks are displayed correctly themself in wysiwiyg;
  48         // 2. For the rest, at least table cell and list item need no filler space.
  49         // (#6248)
  50         if ( fromSource && CKEDITOR.env.ie &&
  51                 ( document.documentMode > 7
  52                 || block.name in CKEDITOR.dtd.tr
  53                 || block.name in CKEDITOR.dtd.$listItem ) )
  54             return false;
  55
  56                 var lastChild = lastNoneSpaceChild( block );
  57
  58                 return !lastChild || lastChild &&
  59                                 ( lastChild.type == CKEDITOR.NODE_ELEMENT && lastChild.name == 'br'
  60                                 // Some of the controls in form needs extension too,
  61                                 // to move cursor at the end of the form. (#4791)
  62                                 || block.name == 'form' && lastChild.name == 'input' );
  63         }
  64
  65         function getBlockExtension( isOutput, emptyBlockFiller )
  66         {
  67                 return function( node )
  68                 {
  69                         trimFillers( node, !isOutput );
  70
  71                         if ( blockNeedsExtension( node, !isOutput, emptyBlockFiller ) )
  72                         {
  73                                 if ( isOutput || CKEDITOR.env.ie )
  74                                         node.add( new CKEDITOR.htmlParser.text( '\xa0' ) );
  75                                 else
  76                                         node.add( new CKEDITOR.htmlParser.element( 'br', {} ) );
  77                         }
  78                 };
  79         }
  80
  81         var dtd = CKEDITOR.dtd;
  82
  83         // Define orders of table elements.
  84         var tableOrder = [ 'caption', 'colgroup', 'col', 'thead', 'tfoot', 'tbody' ];
  85
  86         // Find out the list of block-like tags that can contain <br>.
  87         var blockLikeTags = CKEDITOR.tools.extend( {}, dtd.$block, dtd.$listItem, dtd.$tableContent );
  88         for ( var i in blockLikeTags )
  89         {
  90                 if ( ! ( 'br' in dtd[i] ) )
  91                         delete blockLikeTags[i];
  92         }
  93         // We just avoid filler in <pre> right now.
  94         // TODO: Support filler for <pre>, line break is also occupy line height.
  95         delete blockLikeTags.pre;
  96         var defaultDataFilterRules =
  97         {
  98                 elements : {},
  99                 attributeNames :
 100                 [
 101                         // Event attributes (onXYZ) must not be directly set. They can become
 102                         // active in the editing area (IE|WebKit).
 103                         [ ( /^on/ ), 'data-cke-pa-on' ]
 104                 ]
 105         };
 106
 107         var defaultDataBlockFilterRules = { elements : {} };
 108
 109         for ( i in blockLikeTags )
 110                 defaultDataBlockFilterRules.elements[ i ] = getBlockExtension();
 111
 112         var defaultHtmlFilterRules =
 113                 {
 114                         elementNames :
 115                         [
 116                                 // Remove the "cke:" namespace prefix.
 117                                 [ ( /^cke:/ ), '' ],
 118
 119                                 // Ignore <?xml:namespace> tags.
 120                                 [ ( /^\?xml:namespace$/ ), '' ]
 121                         ],
 122
 123                         attributeNames :
 124                         [
 125                                 // Attributes saved for changes and protected attributes.
 126                                 [ ( /^data-cke-(saved|pa)-/ ), '' ],
 127
 128                                 // All "data-cke-" attributes are to be ignored.
 129                                 [ ( /^data-cke-.*/ ), '' ],
 130
 131                                 [ 'hidefocus', '' ]
 132                         ],
 133
 134                         elements :
 135                         {
 136                                 $ : function( element )
 137                                 {
 138                                         var attribs = element.attributes;
 139
 140                                         if ( attribs )
 141                                         {
 142                                                 // Elements marked as temporary are to be ignored.
 143                                                 if ( attribs[ 'data-cke-temp' ] )
 144                                                         return false;
 145
 146                                                 // Remove duplicated attributes - #3789.
 147                                                 var attributeNames = [ 'name', 'href', 'src' ],
 148                                                         savedAttributeName;
 149                                                 for ( var i = 0 ; i < attributeNames.length ; i++ )
 150                                                 {
 151                                                         savedAttributeName = 'data-cke-saved-' + attributeNames[ i ];
 152                                                         savedAttributeName in attribs && ( delete attribs[ attributeNames[ i ] ] );
 153                                                 }
 154                                         }
 155
 156                                         return element;
 157                                 },
 158
 159                                 // The contents of table should be in correct order (#4809).
 160                                 table : function( element )
 161                                 {
 162                                         var children = element.children;
 163                                         children.sort( function ( node1, node2 )
 164                                                                    {
 165                                                                            return node1.type == CKEDITOR.NODE_ELEMENT && node2.type == node1.type ?
 166                                                                                         CKEDITOR.tools.indexOf( tableOrder, node1.name )  > CKEDITOR.tools.indexOf( tableOrder, node2.name ) ? 1 : -1 : 0;
 167                                                                    } );
 168                                 },
 169
 170                                 embed : function( element )
 171                                 {
 172                                         var parent = element.parent;
 173
 174                                         // If the <embed> is child of a <object>, copy the width
 175                                         // and height attributes from it.
 176                                         if ( parent && parent.name == 'object' )
 177                                         {
 178                                                 var parentWidth = parent.attributes.width,
 179                                                         parentHeight = parent.attributes.height;
 180                                                 parentWidth && ( element.attributes.width = parentWidth );
 181                                                 parentHeight && ( element.attributes.height = parentHeight );
 182                                         }
 183                                 },
 184                                 // Restore param elements into self-closing.
 185                                 param : function( param )
 186                                 {
 187                                         param.children = [];
 188                                         param.isEmpty = true;
 189                                         return param;
 190                                 },
 191
 192                                 // Remove empty link but not empty anchor.(#3829)
 193                                 a : function( element )
 194                                 {
 195                                         if ( !( element.children.length ||
 196                                                         element.attributes.name ||
 197                                                         element.attributes[ 'data-cke-saved-name' ] ) )
 198                                         {
 199                                                 return false;
 200                                         }
 201                                 },
 202
 203                                 // Remove dummy span in webkit.
 204                                 span: function( element )
 205                                 {
 206                                         if ( element.attributes[ 'class' ] == 'Apple-style-span' )
 207                                                 delete element.name;
 208                                 },
 209
 210                                 // Empty <pre> in IE is reported with filler node (&nbsp;).
 211                                 pre : function( element ) { CKEDITOR.env.ie && trimFillers( element ); },
 212
 213                                 html : function( element )
 214                                 {
 215                                         delete element.attributes.contenteditable;
 216                                         delete element.attributes[ 'class' ];
 217                                 },
 218
 219                                 body : function( element )
 220                                 {
 221                                         delete element.attributes.spellcheck;
 222                                         delete element.attributes.contenteditable;
 223                                 },
 224
 225                                 style : function( element )
 226                                 {
 227                                         var child = element.children[ 0 ];
 228                                         child && child.value && ( child.value = CKEDITOR.tools.trim( child.value ));
 229
 230                                         if ( !element.attributes.type )
 231                                                 element.attributes.type = 'text/css';
 232                                 },
 233
 234                                 title : function( element )
 235                                 {
 236                                         var titleText = element.children[ 0 ];
 237                                         titleText && ( titleText.value = element.attributes[ 'data-cke-title' ] || '' );
 238                                 }
 239                         },
 240
 241                         attributes :
 242                         {
 243                                 'class' : function( value, element )
 244                                 {
 245                                         // Remove all class names starting with "cke_".
 246                                         return CKEDITOR.tools.ltrim( value.replace( /(?:^|\s+)cke_[^\s]*/g, '' ) ) || false;
 247                                 }
 248                         }
 249                 };
 250
 251         if ( CKEDITOR.env.ie )
 252         {
 253                 // IE outputs style attribute in capital letters. We should convert
 254                 // them back to lower case, while not hurting the values (#5930)
 255                 defaultHtmlFilterRules.attributes.style = function( value, element )
 256                 {
 257                         return value.replace( /(^|;)([^\:]+)/g, function( match )
 258                                 {
 259                                         return match.toLowerCase();
 260                                 });
 261                 };
 262         }
 263
 264         function protectReadOnly( element )
 265         {
 266                 var attrs = element.attributes;
 267
 268                 // We should flag that the element was locked by our code so
 269                 // it'll be editable by the editor functions (#6046).
 270                 if ( attrs.contenteditable != "false" )
 271                         attrs[ 'data-cke-editable' ] = attrs.contenteditable ? 'true' : 1;
 272
 273                 attrs.contenteditable = "false";
 274         }
 275         function unprotectReadyOnly( element )
 276         {
 277                 var attrs = element.attributes;
 278                 switch( attrs[ 'data-cke-editable' ] )
 279                 {
 280                         case 'true':    attrs.contenteditable = 'true'; break;
 281                         case '1':               delete attrs.contenteditable;   break;
 282                 }
 283         }
 284         // Disable form elements editing mode provided by some browers. (#5746)
 285         for ( i in { input : 1, textarea : 1 } )
 286         {
 287                 defaultDataFilterRules.elements[ i ] = protectReadOnly;
 288                 defaultHtmlFilterRules.elements[ i ] = unprotectReadyOnly;
 289         }
 290
 291         var protectElementRegex = /<(a|area|img|input)\b([^>]*)>/gi,
 292                 protectAttributeRegex = /\b(href|src|name)\s*=\s*(?:(?:"[^"]*")|(?:'[^']*')|(?:[^ "'>]+))/gi;
 293
 294         var protectElementsRegex = /(?:<style(?=[ >])[^>]*>[\s\S]*<\/style>)|(?:<(:?link|meta|base)[^>]*>)/gi,
 295                 encodedElementsRegex = /<cke:encoded>([^<]*)<\/cke:encoded>/gi;
 296
 297         var protectElementNamesRegex = /(<\/?)((?:object|embed|param|html|body|head|title)[^>]*>)/gi,
 298                 unprotectElementNamesRegex = /(<\/?)cke:((?:html|body|head|title)[^>]*>)/gi;
 299
 300         var protectSelfClosingRegex = /<cke:(param|embed)([^>]*?)\/?>(?!\s*<\/cke:\1)/gi;
 301
 302         function protectAttributes( html )
 303         {
 304                 return html.replace( protectElementRegex, function( element, tag, attributes )
 305                 {
 306                         return '<' +  tag + attributes.replace( protectAttributeRegex, function( fullAttr, attrName )
 307                         {
 308                                 // We should not rewrite the existed protected attributes, e.g. clipboard content from editor. (#5218)
 309                                 if ( attributes.indexOf( 'data-cke-saved-' + attrName ) == -1 )
 310                                         return ' data-cke-saved-' + fullAttr + ' ' + fullAttr;
 311
 312                                 return fullAttr;
 313                         }) + '>';
 314                 });
 315         }
 316
 317         function protectElements( html )
 318         {
 319                 return html.replace( protectElementsRegex, function( match )
 320                         {
 321                                 return '<cke:encoded>' + encodeURIComponent( match ) + '</cke:encoded>';
 322                         });
 323         }
 324
 325         function unprotectElements( html )
 326         {
 327                 return html.replace( encodedElementsRegex, function( match, encoded )
 328                         {
 329                                 return decodeURIComponent( encoded );
 330                         });
 331         }
 332
 333         function protectElementsNames( html )
 334         {
 335                 return html.replace( protectElementNamesRegex, '$1cke:$2');
 336         }
 337
 338         function unprotectElementNames( html )
 339         {
 340                 return html.replace( unprotectElementNamesRegex, '$1$2' );
 341         }
 342
 343         function protectSelfClosingElements( html )
 344         {
 345                 return html.replace( protectSelfClosingRegex, '<cke:$1$2></cke:$1>' );
 346         }
 347
 348         function protectPreFormatted( html )
 349         {
 350                 return html.replace( /(<pre\b[^>]*>)(\r\n|\n)/g, '$1$2$2' );
 351         }
 352
 353         function protectRealComments( html )
 354         {
 355                 return html.replace( /<!--(?!{cke_protected})[\s\S]+?-->/g, function( match )
 356                         {
 357                                 return '<!--' + protectedSourceMarker +
 358                                                 '{C}' +
 359                                                 encodeURIComponent( match ).replace( /--/g, '%2D%2D' ) +
 360                                                 '-->';
 361                         });
 362         }
 363
 364         function unprotectRealComments( html )
 365         {
 366                 return html.replace( /<!--\{cke_protected\}\{C\}([\s\S]+?)-->/g, function( match, data )
 367                         {
 368                                 return decodeURIComponent( data );
 369                         });
 370         }
 371
 372         function unprotectSource( html, editor )
 373         {
 374                 var store = editor._.dataStore;
 375
 376                 return html.replace( /<!--\{cke_protected\}([\s\S]+?)-->/g, function( match, data )
 377                         {
 378                                 return decodeURIComponent( data );
 379                         }).replace( /\{cke_protected_(\d+)\}/g, function( match, id )
 380                         {
 381                                 return store && store[ id ] || '';
 382                         });
 383         }
 384
 385         function protectSource( data, editor )
 386         {
 387                 var protectedHtml = [],
 388                         protectRegexes = editor.config.protectedSource,
 389                         store = editor._.dataStore || ( editor._.dataStore = { id : 1 } ),
 390                         tempRegex = /<\!--\{cke_temp(comment)?\}(\d*?)-->/g;
 391
 392                 var regexes =
 393                         [
 394                                 // Script tags will also be forced to be protected, otherwise
 395                                 // IE will execute them.
 396                                 ( /<script[\s\S]*?<\/script>/gi ),
 397
 398                                 // <noscript> tags (get lost in IE and messed up in FF).
 399                                 /<noscript[\s\S]*?<\/noscript>/gi
 400                         ]
 401                         .concat( protectRegexes );
 402
 403                 // First of any other protection, we must protect all comments
 404                 // to avoid loosing them (of course, IE related).
 405                 // Note that we use a different tag for comments, as we need to
 406                 // transform them when applying filters.
 407                 data = data.replace( (/<!--[\s\S]*?-->/g), function( match )
 408                         {
 409                                 return  '<!--{cke_tempcomment}' + ( protectedHtml.push( match ) - 1 ) + '-->';
 410                         });
 411
 412                 for ( var i = 0 ; i < regexes.length ; i++ )
 413                 {
 414                         data = data.replace( regexes[i], function( match )
 415                                 {
 416                                         match = match.replace( tempRegex,               // There could be protected source inside another one. (#3869).
 417                                                 function( $, isComment, id )
 418                                                 {
 419                                                         return protectedHtml[ id ];
 420                                                 }
 421                                         );
 422
 423                                         // Avoid protecting over protected, e.g. /\{.*?\}/
 424                                         return ( /cke_temp(comment)?/ ).test( match ) ? match
 425                                                 : '<!--{cke_temp}' + ( protectedHtml.push( match ) - 1 ) + '-->';
 426                                 });
 427                 }
 428                 data = data.replace( tempRegex, function( $, isComment, id )
 429                         {
 430                                 return '<!--' + protectedSourceMarker +
 431                                                 ( isComment ? '{C}' : '' ) +
 432                                                 encodeURIComponent( protectedHtml[ id ] ).replace( /--/g, '%2D%2D' ) +
 433                                                 '-->';
 434                         }
 435                 );
 436
 437                 // Different protection pattern is used for those that
 438                 // live in attributes to avoid from being HTML encoded.
 439                 return data.replace( /(['"]).*?\1/g, function ( match )
 440                 {
 441                         return match.replace( /<!--\{cke_protected\}([\s\S]+?)-->/g, function( match, data )
 442                         {
 443                                 store[ store.id ] = decodeURIComponent( data );
 444                                 return '{cke_protected_'+ ( store.id++ )  + '}';
 445                         });
 446                 });
 447         }
 448
 449         CKEDITOR.plugins.add( 'htmldataprocessor',
 450         {
 451                 requires : [ 'htmlwriter' ],
 452
 453                 init : function( editor )
 454                 {
 455                         var dataProcessor = editor.dataProcessor = new CKEDITOR.htmlDataProcessor( editor );
 456
 457                         dataProcessor.writer.forceSimpleAmpersand = editor.config.forceSimpleAmpersand;
 458
 459                         dataProcessor.dataFilter.addRules( defaultDataFilterRules );
 460                         dataProcessor.dataFilter.addRules( defaultDataBlockFilterRules );
 461                         dataProcessor.htmlFilter.addRules( defaultHtmlFilterRules );
 462
 463                         var defaultHtmlBlockFilterRules = { elements : {} };
 464                         for ( i in blockLikeTags )
 465                                 defaultHtmlBlockFilterRules.elements[ i ] = getBlockExtension( true, editor.config.fillEmptyBlocks );
 466
 467                         dataProcessor.htmlFilter.addRules( defaultHtmlBlockFilterRules );
 468                 },
 469
 470                 onLoad : function()
 471                 {
 472                         ! ( 'fillEmptyBlocks' in CKEDITOR.config ) && ( CKEDITOR.config.fillEmptyBlocks = 1 );
 473                 }
 474         });
 475
 476         CKEDITOR.htmlDataProcessor = function( editor )
 477         {
 478                 this.editor = editor;
 479
 480                 this.writer = new CKEDITOR.htmlWriter();
 481                 this.dataFilter = new CKEDITOR.htmlParser.filter();
 482                 this.htmlFilter = new CKEDITOR.htmlParser.filter();
 483         };
 484
 485         CKEDITOR.htmlDataProcessor.prototype =
 486         {
 487                 toHtml : function( data, fixForBody )
 488                 {
 489                         // The source data is already HTML, but we need to clean
 490                         // it up and apply the filter.
 491
 492                         data = protectSource( data, this.editor );
 493
 494                         // Before anything, we must protect the URL attributes as the
 495                         // browser may changing them when setting the innerHTML later in
 496                         // the code.
 497                         data = protectAttributes( data );
 498
 499                         // Protect elements than can't be set inside a DIV. E.g. IE removes
 500                         // style tags from innerHTML. (#3710)
 501                         data = protectElements( data );
 502
 503                         // Certain elements has problem to go through DOM operation, protect
 504                         // them by prefixing 'cke' namespace. (#3591)
 505                         data = protectElementsNames( data );
 506
 507                         // All none-IE browsers ignore self-closed custom elements,
 508                         // protecting them into open-close. (#3591)
 509                         data = protectSelfClosingElements( data );
 510
 511                         // Compensate one leading line break after <pre> open as browsers
 512                         // eat it up. (#5789)
 513                         data = protectPreFormatted( data );
 514
 515                         // Call the browser to help us fixing a possibly invalid HTML
 516                         // structure.
 517                         var div = new CKEDITOR.dom.element( 'div' );
 518                         // Add fake character to workaround IE comments bug. (#3801)
 519                         div.setHtml( 'a' + data );
 520                         data = div.getHtml().substr( 1 );
 521
 522                         // Unprotect "some" of the protected elements at this point.
 523                         data = unprotectElementNames( data );
 524
 525                         data = unprotectElements( data );
 526
 527                         // Restore the comments that have been protected, in this way they
 528                         // can be properly filtered.
 529                         data = unprotectRealComments( data );
 530
 531                         // Now use our parser to make further fixes to the structure, as
 532                         // well as apply the filter.
 533                         var fragment = CKEDITOR.htmlParser.fragment.fromHtml( data, fixForBody ),
 534                                 writer = new CKEDITOR.htmlParser.basicWriter();
 535
 536                         fragment.writeHtml( writer, this.dataFilter );
 537                         data = writer.getHtml( true );
 538
 539                         // Protect the real comments again.
 540                         data = protectRealComments( data );
 541
 542                         return data;
 543                 },
 544
 545                 toDataFormat : function( html, fixForBody )
 546                 {
 547                         var writer = this.writer,
 548                                 fragment = CKEDITOR.htmlParser.fragment.fromHtml( html, fixForBody );
 549
 550                         writer.reset();
 551
 552                         fragment.writeHtml( writer, this.htmlFilter );
 553
 554                         var data = writer.getHtml( true );
 555
 556                         // Restore those non-HTML protected source. (#4475,#4880)
 557                         data = unprotectRealComments( data );
 558                         data = unprotectSource( data, this.editor );
 559
 560                         return data;
 561                 }
 562         };
 563 })();
 564
 565 /**
 566  * Whether to force using "&" instead of "&amp;amp;" in elements attributes
 567  * values, it's not recommended to change this setting for compliance with the
 568  * W3C XHTML 1.0 standards (<a href="http://www.w3.org/TR/xhtml1/#C_12">C.12, XHTML 1.0</a>).
 569  * @name CKEDITOR.config.forceSimpleAmpersand
 570  * @name CKEDITOR.config.forceSimpleAmpersand
 571  * @type Boolean
 572  * @default false
 573  * @example
 574  * config.forceSimpleAmpersand = false;
 575  */
 576
 577 /**
 578  * Whether a filler text (non-breaking space entity - &nbsp;) will be inserted into empty block elements in HTML output,
 579  * this is used to render block elements properly with line-height; When a function is instead specified,
 580  * it'll be passed a {@link CKEDITOR.htmlParser.element} to decide whether adding the filler text
 581  * by expecting a boolean return value.
 582  * @name CKEDITOR.config.fillEmptyBlocks
 583  * @since 3.5
 584  * @type Boolean
 585  * @default true
 586  * @example
 587  * config.fillEmptyBlocks = false;      // Prevent filler nodes in all empty blocks.
 588  *
 589  * // Prevent filler node only in float cleaners.
 590  * config.fillEmptyBlocks = function( element )
 591  * {
 592  *      if ( element.attributes[ 'class' ].indexOf ( 'clear-both' ) != -1 )
 593  *              return false;
 594  * }
 595  */