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