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