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