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