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