_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 \r
  15         // Return the last non-space child node of the block (#4344).\r
  16         function lastNoneSpaceChild( block )\r
  17         {\r
  18                 var lastIndex = block.children.length,\r
  19                         last = block.children[ lastIndex - 1 ];\r
  20                 while(  last && last.type == CKEDITOR.NODE_TEXT && !CKEDITOR.tools.trim( last.value ) )\r
  21                         last = block.children[ --lastIndex ];\r
  22                 return last;\r
  23         }\r
  24 \r
  25         function trimFillers( block, fromSource )\r
  26         {\r
  27                 // If the current node is a block, and if we're converting from source or\r
  28                 // we're not in IE then search for and remove any tailing BR node.\r
  29                 //\r
  30                 // Also, any &nbsp; at the end of blocks are fillers, remove them as well.\r
  31                 // (#2886)\r
  32                 var children = block.children, lastChild = lastNoneSpaceChild( block );\r
  33                 if ( lastChild )\r
  34                 {\r
  35                         if ( ( fromSource || !CKEDITOR.env.ie ) && lastChild.type == CKEDITOR.NODE_ELEMENT && lastChild.name == 'br' )\r
  36                                 children.pop();\r
  37                         if ( lastChild.type == CKEDITOR.NODE_TEXT && tailNbspRegex.test( lastChild.value ) )\r
  38                                 children.pop();\r
  39                 }\r
  40         }\r
  41 \r
  42         function blockNeedsExtension( block )\r
  43         {\r
  44                 var lastChild = lastNoneSpaceChild( block );\r
  45                 return !lastChild || lastChild.type == CKEDITOR.NODE_ELEMENT && lastChild.name == 'br';\r
  46         }\r
  47 \r
  48         function extendBlockForDisplay( block )\r
  49         {\r
  50                 trimFillers( block, true );\r
  51 \r
  52                 if ( blockNeedsExtension( block ) )\r
  53                 {\r
  54                         if ( CKEDITOR.env.ie )\r
  55                                 block.add( new CKEDITOR.htmlParser.text( '\xa0' ) );\r
  56                         else\r
  57                                 block.add( new CKEDITOR.htmlParser.element( 'br', {} ) );\r
  58                 }\r
  59         }\r
  60 \r
  61         function extendBlockForOutput( block )\r
  62         {\r
  63                 trimFillers( block );\r
  64 \r
  65                 if ( blockNeedsExtension( block ) )\r
  66                         block.add( new CKEDITOR.htmlParser.text( '\xa0' ) );\r
  67         }\r
  68 \r
  69         var dtd = CKEDITOR.dtd;\r
  70 \r
  71         // Find out the list of block-like tags that can contain <br>.\r
  72         var blockLikeTags = CKEDITOR.tools.extend( {}, dtd.$block, dtd.$listItem, dtd.$tableContent );\r
  73         for ( var i in blockLikeTags )\r
  74         {\r
  75                 if ( ! ( 'br' in dtd[i] ) )\r
  76                         delete blockLikeTags[i];\r
  77         }\r
  78         // We just avoid filler in <pre> right now.\r
  79         // TODO: Support filler for <pre>, line break is also occupy line height.\r
  80         delete blockLikeTags.pre;\r
  81         var defaultDataFilterRules =\r
  82         {\r
  83                 attributeNames :\r
  84                 [\r
  85                         // Event attributes (onXYZ) must not be directly set. They can become\r
  86                         // active in the editing area (IE|WebKit).\r
  87                         [ ( /^on/ ), '_cke_pa_on' ]\r
  88                 ]\r
  89         };\r
  90 \r
  91         var defaultDataBlockFilterRules = { elements : {} };\r
  92 \r
  93         for ( i in blockLikeTags )\r
  94                 defaultDataBlockFilterRules.elements[ i ] = extendBlockForDisplay;\r
  95 \r
  96         var defaultHtmlFilterRules =\r
  97                 {\r
  98                         elementNames :\r
  99                         [\r
 100                                 // Remove the "cke:" namespace prefix.\r
 101                                 [ ( /^cke:/ ), '' ],\r
 102 \r
 103                                 // Ignore <?xml:namespace> tags.\r
 104                                 [ ( /^\?xml:namespace$/ ), '' ]\r
 105                         ],\r
 106 \r
 107                         attributeNames :\r
 108                         [\r
 109                                 // Attributes saved for changes and protected attributes.\r
 110                                 [ ( /^_cke_(saved|pa)_/ ), '' ],\r
 111 \r
 112                                 // All "_cke" attributes are to be ignored.\r
 113                                 [ ( /^_cke.*/ ), '' ]\r
 114                         ],\r
 115 \r
 116                         elements :\r
 117                         {\r
 118                                 $ : function( element )\r
 119                                 {\r
 120                                         // Remove duplicated attributes - #3789.\r
 121                                         var attribs = element.attributes;\r
 122 \r
 123                                         if ( attribs )\r
 124                                         {\r
 125                                                 var attributeNames = [ 'name', 'href', 'src' ],\r
 126                                                         savedAttributeName;\r
 127                                                 for ( var i = 0 ; i < attributeNames.length ; i++ )\r
 128                                                 {\r
 129                                                         savedAttributeName = '_cke_saved_' + attributeNames[ i ];\r
 130                                                         savedAttributeName in attribs && ( delete attribs[ attributeNames[ i ] ] );\r
 131                                                 }\r
 132                                         }\r
 133                                 },\r
 134 \r
 135                                 embed : function( element )\r
 136                                 {\r
 137                                         var parent = element.parent;\r
 138 \r
 139                                         // If the <embed> is child of a <object>, copy the width\r
 140                                         // and height attributes from it.\r
 141                                         if ( parent && parent.name == 'object' )\r
 142                                         {\r
 143                                                 var parentWidth = parent.attributes.width,\r
 144                                                         parentHeight = parent.attributes.height;\r
 145                                                 parentWidth && ( element.attributes.width = parentWidth );\r
 146                                                 parentHeight && ( element.attributes.height = parentHeight );\r
 147                                         }\r
 148                                 },\r
 149                                 // Restore param elements into self-closing.\r
 150                                 param : function( param )\r
 151                                 {\r
 152                                         param.children = [];\r
 153                                         param.isEmpty = true;\r
 154                                         return param;\r
 155                                 },\r
 156 \r
 157                                 // Remove empty link but not empty anchor.(#3829)\r
 158                                 a : function( element )\r
 159                                 {\r
 160                                         if ( !( element.children.length ||\r
 161                                                         element.attributes.name ||\r
 162                                                         element.attributes._cke_saved_name ) )\r
 163                                         {\r
 164                                                 return false;\r
 165                                         }\r
 166                                 }\r
 167                         },\r
 168 \r
 169                         attributes :\r
 170                         {\r
 171                                 'class' : function( value, element )\r
 172                                 {\r
 173                                         // Remove all class names starting with "cke_".\r
 174                                         return CKEDITOR.tools.ltrim( value.replace( /(?:^|\s+)cke_[^\s]*/g, '' ) ) || false;\r
 175                                 }\r
 176                         },\r
 177 \r
 178                         comment : function( contents )\r
 179                         {\r
 180                                 if ( contents.substr( 0, protectedSourceMarker.length ) == protectedSourceMarker )\r
 181                                         return new CKEDITOR.htmlParser.cdata( decodeURIComponent( contents.substr( protectedSourceMarker.length ) ) );\r
 182 \r
 183                                 return contents;\r
 184                         }\r
 185                 };\r
 186 \r
 187         var defaultHtmlBlockFilterRules = { elements : {} };\r
 188 \r
 189         for ( i in blockLikeTags )\r
 190                 defaultHtmlBlockFilterRules.elements[ i ] = extendBlockForOutput;\r
 191 \r
 192         if ( CKEDITOR.env.ie )\r
 193         {\r
 194                 // IE outputs style attribute in capital letters. We should convert\r
 195                 // them back to lower case.\r
 196                 defaultHtmlFilterRules.attributes.style = function( value, element )\r
 197                 {\r
 198                         return value.toLowerCase();\r
 199                 };\r
 200         }\r
 201 \r
 202         var protectAttributeRegex = /<(?:a|area|img|input)[\s\S]*?\s((?:href|src|name)\s*=\s*(?:(?:"[^"]*")|(?:'[^']*')|(?:[^ "'>]+)))/gi;\r
 203 \r
 204         function protectAttributes( html )\r
 205         {\r
 206                 return html.replace( protectAttributeRegex, '$& _cke_saved_$1' );\r
 207         }\r
 208 \r
 209         var protectStyleTagsRegex = /<(style)(?=[ >])[^>]*>[^<]*<\/\1>/gi;\r
 210         var encodedTagsRegex = /<cke:encoded>([^<]*)<\/cke:encoded>/gi;\r
 211         var protectElementNamesRegex = /(<\/?)((?:object|embed|param)[\s\S]*?>)/gi;\r
 212         var protectSelfClosingRegex = /<cke:(param|embed)([\s\S]*?)\/?>/gi;\r
 213 \r
 214         function protectStyleTagsMatch( match )\r
 215         {\r
 216                 return '<cke:encoded>' + encodeURIComponent( match ) + '</cke:encoded>';\r
 217         }\r
 218 \r
 219         function protectStyleTags( html )\r
 220         {\r
 221                 return html.replace( protectStyleTagsRegex, protectStyleTagsMatch );\r
 222         }\r
 223         function protectElementsNames( html )\r
 224         {\r
 225                 return html.replace( protectElementNamesRegex, '$1cke:$2');\r
 226         }\r
 227         function protectSelfClosingElements( html )\r
 228         {\r
 229                 return html.replace( protectSelfClosingRegex, '<cke:$1$2></cke:$1>' );\r
 230         }\r
 231 \r
 232         function unprotectEncodedTagsMatch( match, encoded )\r
 233         {\r
 234                 return decodeURIComponent( encoded );\r
 235         }\r
 236 \r
 237         function unprotectEncodedTags( html )\r
 238         {\r
 239                 return html.replace( encodedTagsRegex, unprotectEncodedTagsMatch );\r
 240         }\r
 241 \r
 242         function protectSource( data, protectRegexes )\r
 243         {\r
 244                 var protectedHtml = [],\r
 245                         tempRegex = /<\!--\{cke_temp\}(\d*?)-->/g;\r
 246                 var regexes =\r
 247                         [\r
 248                                 // First of any other protection, we must protect all comments\r
 249                                 // to avoid loosing them (of course, IE related).\r
 250                                 (/<!--[\s\S]*?-->/g),\r
 251 \r
 252                                 // Script tags will also be forced to be protected, otherwise\r
 253                                 // IE will execute them.\r
 254                                 /<script[\s\S]*?<\/script>/gi,\r
 255 \r
 256                                 // <noscript> tags (get lost in IE and messed up in FF).\r
 257                                 /<noscript[\s\S]*?<\/noscript>/gi\r
 258                         ]\r
 259                         .concat( protectRegexes );\r
 260 \r
 261                 for ( var i = 0 ; i < regexes.length ; i++ )\r
 262                 {\r
 263                         data = data.replace( regexes[i], function( match )\r
 264                                 {\r
 265                                         match = match.replace( tempRegex,               // There could be protected source inside another one. (#3869).\r
 266                                                 function( $, id )\r
 267                                                 {\r
 268                                                         return protectedHtml[ id ];\r
 269                                                 }\r
 270                                         );\r
 271                                         return  '<!--{cke_temp}' + ( protectedHtml.push( match ) - 1 ) + '-->';\r
 272                                 });\r
 273                 }\r
 274                 data = data.replace( tempRegex, function( $, id )\r
 275                         {\r
 276                                 return '<!--' + protectedSourceMarker +\r
 277                                                 encodeURIComponent( protectedHtml[ id ] ).replace( /--/g, '%2D%2D' ) +\r
 278                                                 '-->';\r
 279                         }\r
 280                 );\r
 281                 return data;\r
 282         }\r
 283 \r
 284         CKEDITOR.plugins.add( 'htmldataprocessor',\r
 285         {\r
 286                 requires : [ 'htmlwriter' ],\r
 287 \r
 288                 init : function( editor )\r
 289                 {\r
 290                         var dataProcessor = editor.dataProcessor = new CKEDITOR.htmlDataProcessor( editor );\r
 291 \r
 292                         dataProcessor.writer.forceSimpleAmpersand = editor.config.forceSimpleAmpersand;\r
 293 \r
 294                         dataProcessor.dataFilter.addRules( defaultDataFilterRules );\r
 295                         dataProcessor.dataFilter.addRules( defaultDataBlockFilterRules );\r
 296                         dataProcessor.htmlFilter.addRules( defaultHtmlFilterRules );\r
 297                         dataProcessor.htmlFilter.addRules( defaultHtmlBlockFilterRules );\r
 298                 }\r
 299         });\r
 300 \r
 301         CKEDITOR.htmlDataProcessor = function( editor )\r
 302         {\r
 303                 this.editor = editor;\r
 304 \r
 305                 this.writer = new CKEDITOR.htmlWriter();\r
 306                 this.dataFilter = new CKEDITOR.htmlParser.filter();\r
 307                 this.htmlFilter = new CKEDITOR.htmlParser.filter();\r
 308         };\r
 309 \r
 310         CKEDITOR.htmlDataProcessor.prototype =\r
 311         {\r
 312                 toHtml : function( data, fixForBody )\r
 313                 {\r
 314                         // The source data is already HTML, but we need to clean\r
 315                         // it up and apply the filter.\r
 316 \r
 317                         data = protectSource( data, this.editor.config.protectedSource );\r
 318 \r
 319                         // Before anything, we must protect the URL attributes as the\r
 320                         // browser may changing them when setting the innerHTML later in\r
 321                         // the code.\r
 322                         data = protectAttributes( data );\r
 323 \r
 324                         // IE remvoes style tags from innerHTML. (#3710).\r
 325                         if ( CKEDITOR.env.ie )\r
 326                                 data = protectStyleTags( data );\r
 327 \r
 328                         // Certain elements has problem to go through DOM operation, protect\r
 329                         // them by prefixing 'cke' namespace.(#3591)\r
 330                         data = protectElementsNames( data );\r
 331 \r
 332                         // All none-IE browsers ignore self-closed custom elements,\r
 333                         // protecting them into open-close.(#3591)\r
 334                         data = protectSelfClosingElements( data );\r
 335 \r
 336                         // Call the browser to help us fixing a possibly invalid HTML\r
 337                         // structure.\r
 338                         var div = document.createElement( 'div' );\r
 339                         // Add fake character to workaround IE comments bug. (#3801)\r
 340                         div.innerHTML = 'a' + data;\r
 341                         data = div.innerHTML.substr( 1 );\r
 342 \r
 343                         if ( CKEDITOR.env.ie )\r
 344                                 data = unprotectEncodedTags( data );\r
 345 \r
 346                         // Now use our parser to make further fixes to the structure, as\r
 347                         // well as apply the filter.\r
 348                         var fragment = CKEDITOR.htmlParser.fragment.fromHtml( data, fixForBody ),\r
 349                                 writer = new CKEDITOR.htmlParser.basicWriter();\r
 350 \r
 351                         fragment.writeHtml( writer, this.dataFilter );\r
 352 \r
 353                         return writer.getHtml( true );\r
 354                 },\r
 355 \r
 356                 toDataFormat : function( html, fixForBody )\r
 357                 {\r
 358                         var writer = this.writer,\r
 359                                 fragment = CKEDITOR.htmlParser.fragment.fromHtml( html, fixForBody );\r
 360 \r
 361                         writer.reset();\r
 362 \r
 363                         fragment.writeHtml( writer, this.htmlFilter );\r
 364 \r
 365                         return writer.getHtml( true );\r
 366                 }\r
 367         };\r
 368 })();\r
 369 \r
 370 /**\r
 371  * Whether to force using "&" instead of "&amp;amp;" in elements attributes\r
 372  * values. It's not recommended to change this setting for compliance with the\r
 373  * W3C XHTML 1.0 standards\r
 374  * (<a href="http://www.w3.org/TR/xhtml1/#C_12">C.12, XHTML 1.0</a>).\r
 375  * @type Boolean\r
 376  * @default false\r
 377  * @example\r
 378  * config.forceSimpleAmpersand = false;\r
 379  */\r
 380 CKEDITOR.config.forceSimpleAmpersand = false;\r