_source/plugins/htmldataprocessor/plugin.js

   1 /*\r
   2 Copyright (c) 2003-2009, 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         function trimFillers( block, fromSource )\r
  15         {\r
  16                 // If the current node is a block, and if we're converting from source or\r
  17                 // we're not in IE then search for and remove any tailing BR node.\r
  18                 //\r
  19                 // Also, any &nbsp; at the end of blocks are fillers, remove them as well.\r
  20                 // (#2886)\r
  21                 var children = block.children;\r
  22                 var lastChild = children[ children.length - 1 ];\r
  23                 if ( lastChild )\r
  24                 {\r
  25                         if ( ( fromSource || !CKEDITOR.env.ie ) && lastChild.type == CKEDITOR.NODE_ELEMENT && lastChild.name == 'br' )\r
  26                                 children.pop();\r
  27                         if ( lastChild.type == CKEDITOR.NODE_TEXT && tailNbspRegex.test( lastChild.value ) )\r
  28                                 children.pop();\r
  29                 }\r
  30         }\r
  31 \r
  32         function blockNeedsExtension( block )\r
  33         {\r
  34                 if ( block.children.length < 1 )\r
  35                         return true;\r
  36 \r
  37                 var lastChild = block.children[ block.children.length - 1 ];\r
  38                 return lastChild.type == CKEDITOR.NODE_ELEMENT && lastChild.name == 'br';\r
  39         }\r
  40 \r
  41         function extendBlockForDisplay( block )\r
  42         {\r
  43                 trimFillers( block, true );\r
  44 \r
  45                 if ( blockNeedsExtension( block ) )\r
  46                 {\r
  47                         if ( CKEDITOR.env.ie )\r
  48                                 block.add( new CKEDITOR.htmlParser.text( '\xa0' ) );\r
  49                         else\r
  50                                 block.add( new CKEDITOR.htmlParser.element( 'br', {} ) );\r
  51                 }\r
  52         }\r
  53 \r
  54         function extendBlockForOutput( block )\r
  55         {\r
  56                 trimFillers( block );\r
  57 \r
  58                 if ( blockNeedsExtension( block ) )\r
  59                         block.add( new CKEDITOR.htmlParser.text( '\xa0' ) );\r
  60         }\r
  61 \r
  62         var dtd = CKEDITOR.dtd;\r
  63 \r
  64         // Find out the list of block-like tags that can contain <br>.\r
  65         var blockLikeTags = CKEDITOR.tools.extend( {}, dtd.$block, dtd.$listItem, dtd.$tableContent );\r
  66         for ( var i in blockLikeTags )\r
  67         {\r
  68                 if ( ! ( 'br' in dtd[i] ) )\r
  69                         delete blockLikeTags[i];\r
  70         }\r
  71         // We just avoid filler in <pre> right now.\r
  72         // TODO: Support filler for <pre>, line break is also occupy line height.\r
  73         delete blockLikeTags.pre;\r
  74         var defaultDataFilterRules =\r
  75         {\r
  76                 attributeNames :\r
  77                 [\r
  78                         // Event attributes (onXYZ) must not be directly set. They can become\r
  79                         // active in the editing area (IE|WebKit).\r
  80                         [ ( /^on/ ), '_cke_pa_on' ]\r
  81                 ]\r
  82         };\r
  83 \r
  84         var defaultDataBlockFilterRules = { elements : {} };\r
  85 \r
  86         for ( i in blockLikeTags )\r
  87                 defaultDataBlockFilterRules.elements[ i ] = extendBlockForDisplay;\r
  88 \r
  89         var defaultHtmlFilterRules =\r
  90                 {\r
  91                         elementNames :\r
  92                         [\r
  93                                 // Remove the "cke:" namespace prefix.\r
  94                                 [ ( /^cke:/ ), '' ],\r
  95 \r
  96                                 // Ignore <?xml:namespace> tags.\r
  97                                 [ ( /^\?xml:namespace$/ ), '' ]\r
  98                         ],\r
  99 \r
 100                         attributeNames :\r
 101                         [\r
 102                                 // Attributes saved for changes and protected attributes.\r
 103                                 [ ( /^_cke_(saved|pa)_/ ), '' ],\r
 104 \r
 105                                 // All "_cke" attributes are to be ignored.\r
 106                                 [ ( /^_cke.*/ ), '' ]\r
 107                         ],\r
 108 \r
 109                         elements :\r
 110                         {\r
 111                                 $ : function( element )\r
 112                                 {\r
 113                                         // Remove duplicated attributes - #3789.\r
 114                                         var attribs = element.attributes;\r
 115 \r
 116                                         if ( attribs )\r
 117                                         {\r
 118                                                 var attributeNames = [ 'name', 'href', 'src' ],\r
 119                                                         savedAttributeName;\r
 120                                                 for ( var i = 0 ; i < attributeNames.length ; i++ )\r
 121                                                 {\r
 122                                                         savedAttributeName = '_cke_saved_' + attributeNames[ i ];\r
 123                                                         savedAttributeName in attribs && ( delete attribs[ attributeNames[ i ] ] );\r
 124                                                 }\r
 125                                         }\r
 126                                 },\r
 127 \r
 128                                 embed : function( element )\r
 129                                 {\r
 130                                         var parent = element.parent;\r
 131 \r
 132                                         // If the <embed> is child of a <object>, copy the width\r
 133                                         // and height attributes from it.\r
 134                                         if ( parent && parent.name == 'object' )\r
 135                                         {\r
 136                                                 var parentWidth = parent.attributes.width,\r
 137                                                         parentHeight = parent.attributes.height;\r
 138                                                 parentWidth && ( element.attributes.width = parentWidth );\r
 139                                                 parentHeight && ( element.attributes.height = parentHeight );\r
 140                                         }\r
 141                                 },\r
 142                                 // Restore param elements into self-closing.\r
 143                                 param : function( param )\r
 144                                 {\r
 145                                         param.children = [];\r
 146                                         param.isEmpty = true;\r
 147                                         return param;\r
 148                                 },\r
 149 \r
 150                                 // Remove empty link but not empty anchor.(#3829)\r
 151                                 a : function( element )\r
 152                                 {\r
 153                                         if ( !( element.children.length ||\r
 154                                                         element.attributes.name ||\r
 155                                                         element.attributes._cke_saved_name ) )\r
 156                                         {\r
 157                                                 return false;\r
 158                                         }\r
 159                                 }\r
 160                         },\r
 161 \r
 162                         attributes :\r
 163                         {\r
 164                                 'class' : function( value, element )\r
 165                                 {\r
 166                                         // Remove all class names starting with "cke_".\r
 167                                         return CKEDITOR.tools.ltrim( value.replace( /(?:^|\s+)cke_[^\s]*/g, '' ) ) || false;\r
 168                                 }\r
 169                         },\r
 170 \r
 171                         comment : function( contents )\r
 172                         {\r
 173                                 if ( contents.substr( 0, protectedSourceMarker.length ) == protectedSourceMarker )\r
 174                                         return new CKEDITOR.htmlParser.cdata( decodeURIComponent( contents.substr( protectedSourceMarker.length ) ) );\r
 175 \r
 176                                 return contents;\r
 177                         }\r
 178                 };\r
 179 \r
 180         var defaultHtmlBlockFilterRules = { elements : {} };\r
 181 \r
 182         for ( i in blockLikeTags )\r
 183                 defaultHtmlBlockFilterRules.elements[ i ] = extendBlockForOutput;\r
 184 \r
 185         if ( CKEDITOR.env.ie )\r
 186         {\r
 187                 // IE outputs style attribute in capital letters. We should convert\r
 188                 // them back to lower case.\r
 189                 defaultHtmlFilterRules.attributes.style = function( value, element )\r
 190                 {\r
 191                         return value.toLowerCase();\r
 192                 };\r
 193         }\r
 194 \r
 195         var protectAttributeRegex = /<(?:a|area|img|input).*?\s((?:href|src|name)\s*=\s*(?:(?:"[^"]*")|(?:'[^']*')|(?:[^ "'>]+)))/gi;\r
 196 \r
 197         function protectAttributes( html )\r
 198         {\r
 199                 return html.replace( protectAttributeRegex, '$& _cke_saved_$1' );\r
 200         }\r
 201 \r
 202         var protectStyleTagsRegex = /<(style)(?=[ >])[^>]*>[^<]*<\/\1>/gi;\r
 203         var encodedTagsRegex = /<cke:encoded>([^<]*)<\/cke:encoded>/gi;\r
 204         var protectElementNamesRegex = /(<\/?)((?:object|embed|param).*?>)/gi;\r
 205         var protectSelfClosingRegex = /<cke:param(.*?)\/>/gi;\r
 206 \r
 207         function protectStyleTagsMatch( match )\r
 208         {\r
 209                 return '<cke:encoded>' + encodeURIComponent( match ) + '</cke:encoded>';\r
 210         }\r
 211 \r
 212         function protectStyleTags( html )\r
 213         {\r
 214                 return html.replace( protectStyleTagsRegex, protectStyleTagsMatch );\r
 215         }\r
 216         function protectElementsNames( html )\r
 217         {\r
 218                 return html.replace( protectElementNamesRegex, '$1cke:$2');\r
 219         }\r
 220         function protectSelfClosingElements( html )\r
 221         {\r
 222                 return html.replace( protectSelfClosingRegex, '<cke:param$1></cke:param>' );\r
 223         }\r
 224 \r
 225         function unprotectEncodedTagsMatch( match, encoded )\r
 226         {\r
 227                 return decodeURIComponent( encoded );\r
 228         }\r
 229 \r
 230         function unprotectEncodedTags( html )\r
 231         {\r
 232                 return html.replace( encodedTagsRegex, unprotectEncodedTagsMatch );\r
 233         }\r
 234 \r
 235         function protectSource( data, protectRegexes )\r
 236         {\r
 237                 var protectedHtml = [],\r
 238                         tempRegex = /<\!--\{cke_temp\}(\d*?)-->/g;\r
 239                 var regexes =\r
 240                         [\r
 241                                 // First of any other protection, we must protect all comments\r
 242                                 // to avoid loosing them (of course, IE related).\r
 243                                 (/<!--[\s\S]*?-->/g),\r
 244 \r
 245                                 // Script tags will also be forced to be protected, otherwise\r
 246                                 // IE will execute them.\r
 247                                 /<script[\s\S]*?<\/script>/gi,\r
 248 \r
 249                                 // <noscript> tags (get lost in IE and messed up in FF).\r
 250                                 /<noscript[\s\S]*?<\/noscript>/gi\r
 251                         ]\r
 252                         .concat( protectRegexes );\r
 253 \r
 254                 for ( var i = 0 ; i < regexes.length ; i++ )\r
 255                 {\r
 256                         data = data.replace( regexes[i], function( match )\r
 257                                 {\r
 258                                         match = match.replace( tempRegex,               // There could be protected source inside another one. (#3869).\r
 259                                                 function( $, id )\r
 260                                                 {\r
 261                                                         return protectedHtml[ id ];\r
 262                                                 }\r
 263                                         );\r
 264                                         return  '<!--{cke_temp}' + ( protectedHtml.push( match ) - 1 ) + '-->';\r
 265                                 });\r
 266                 }\r
 267                 data = data.replace( tempRegex, function( $, id )\r
 268                         {\r
 269                                 return '<!--' + protectedSourceMarker +\r
 270                                                 encodeURIComponent( protectedHtml[ id ] ).replace( /--/g, '%2D%2D' ) +\r
 271                                                 '-->';\r
 272                         }\r
 273                 );\r
 274                 return data;\r
 275         }\r
 276 \r
 277         CKEDITOR.plugins.add( 'htmldataprocessor',\r
 278         {\r
 279                 requires : [ 'htmlwriter' ],\r
 280 \r
 281                 init : function( editor )\r
 282                 {\r
 283                         var dataProcessor = editor.dataProcessor = new CKEDITOR.htmlDataProcessor( editor );\r
 284 \r
 285                         dataProcessor.writer.forceSimpleAmpersand = editor.config.forceSimpleAmpersand;\r
 286 \r
 287                         dataProcessor.dataFilter.addRules( defaultDataFilterRules );\r
 288                         dataProcessor.dataFilter.addRules( defaultDataBlockFilterRules );\r
 289                         dataProcessor.htmlFilter.addRules( defaultHtmlFilterRules );\r
 290                         dataProcessor.htmlFilter.addRules( defaultHtmlBlockFilterRules );\r
 291                 }\r
 292         });\r
 293 \r
 294         CKEDITOR.htmlDataProcessor = function( editor )\r
 295         {\r
 296                 this.editor = editor;\r
 297 \r
 298                 this.writer = new CKEDITOR.htmlWriter();\r
 299                 this.dataFilter = new CKEDITOR.htmlParser.filter();\r
 300                 this.htmlFilter = new CKEDITOR.htmlParser.filter();\r
 301         };\r
 302 \r
 303         CKEDITOR.htmlDataProcessor.prototype =\r
 304         {\r
 305                 toHtml : function( data, fixForBody )\r
 306                 {\r
 307                         // The source data is already HTML, but we need to clean\r
 308                         // it up and apply the filter.\r
 309 \r
 310                         data = protectSource( data, this.editor.config.protectedSource );\r
 311 \r
 312                         // Before anything, we must protect the URL attributes as the\r
 313                         // browser may changing them when setting the innerHTML later in\r
 314                         // the code.\r
 315                         data = protectAttributes( data );\r
 316 \r
 317                         // IE remvoes style tags from innerHTML. (#3710).\r
 318                         if ( CKEDITOR.env.ie )\r
 319                                 data = protectStyleTags( data );\r
 320 \r
 321                         // Certain elements has problem to go through DOM operation, protect\r
 322                         // them by prefixing 'cke' namespace.(#3591)\r
 323                         data = protectElementsNames( data );\r
 324 \r
 325                         // All none-IE browsers ignore self-closed custom elements,\r
 326                         // protecting them into open-close.(#3591)\r
 327                         data = protectSelfClosingElements( data );\r
 328 \r
 329                         // Call the browser to help us fixing a possibly invalid HTML\r
 330                         // structure.\r
 331                         var div = document.createElement( 'div' );\r
 332                         // Add fake character to workaround IE comments bug. (#3801)\r
 333                         div.innerHTML = 'a' + data;\r
 334                         data = div.innerHTML.substr( 1 );\r
 335 \r
 336                         if ( CKEDITOR.env.ie )\r
 337                                 data = unprotectEncodedTags( data );\r
 338 \r
 339                         // Now use our parser to make further fixes to the structure, as\r
 340                         // well as apply the filter.\r
 341                         var fragment = CKEDITOR.htmlParser.fragment.fromHtml( data, fixForBody ),\r
 342                                 writer = new CKEDITOR.htmlParser.basicWriter();\r
 343 \r
 344                         fragment.writeHtml( writer, this.dataFilter );\r
 345 \r
 346                         return writer.getHtml( true );\r
 347                 },\r
 348 \r
 349                 toDataFormat : function( html, fixForBody )\r
 350                 {\r
 351                         var writer = this.writer,\r
 352                                 fragment = CKEDITOR.htmlParser.fragment.fromHtml( html, fixForBody );\r
 353 \r
 354                         writer.reset();\r
 355 \r
 356                         fragment.writeHtml( writer, this.htmlFilter );\r
 357 \r
 358                         return writer.getHtml( true );\r
 359                 }\r
 360         };\r
 361 })();\r
 362 \r
 363 /**\r
 364  * Whether to force using "&" instead of "&amp;amp;" in elements attributes\r
 365  * values. It's not recommended to change this setting for compliance with the\r
 366  * W3C XHTML 1.0 standards\r
 367  * (<a href="http://www.w3.org/TR/xhtml1/#C_12">C.12, XHTML 1.0</a>).\r
 368  * @type Boolean\r
 369  * @default false\r
 370  * @example\r
 371  * config.forceSimpleAmpersand = false;\r
 372  */\r
 373 CKEDITOR.config.forceSimpleAmpersand = false;\r