_source/plugins/entities/plugin.js

   1 /*\r
   2 Copyright (c) 2003-2011, 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         // Base HTML entities.\r
   9         var htmlbase = 'nbsp,gt,lt,amp';\r
  10 \r
  11         var entities =\r
  12                 // Latin-1 Entities\r
  13                 'quot,iexcl,cent,pound,curren,yen,brvbar,sect,uml,copy,ordf,laquo,' +\r
  14                 'not,shy,reg,macr,deg,plusmn,sup2,sup3,acute,micro,para,middot,' +\r
  15                 'cedil,sup1,ordm,raquo,frac14,frac12,frac34,iquest,times,divide,' +\r
  16 \r
  17                 // Symbols\r
  18                 'fnof,bull,hellip,prime,Prime,oline,frasl,weierp,image,real,trade,' +\r
  19                 'alefsym,larr,uarr,rarr,darr,harr,crarr,lArr,uArr,rArr,dArr,hArr,' +\r
  20                 'forall,part,exist,empty,nabla,isin,notin,ni,prod,sum,minus,lowast,' +\r
  21                 'radic,prop,infin,ang,and,or,cap,cup,int,there4,sim,cong,asymp,ne,' +\r
  22                 'equiv,le,ge,sub,sup,nsub,sube,supe,oplus,otimes,perp,sdot,lceil,' +\r
  23                 'rceil,lfloor,rfloor,lang,rang,loz,spades,clubs,hearts,diams,' +\r
  24 \r
  25                 // Other Special Characters\r
  26                 'circ,tilde,ensp,emsp,thinsp,zwnj,zwj,lrm,rlm,ndash,mdash,lsquo,' +\r
  27                 'rsquo,sbquo,ldquo,rdquo,bdquo,dagger,Dagger,permil,lsaquo,rsaquo,' +\r
  28                 'euro';\r
  29 \r
  30         // Latin Letters Entities\r
  31         var latin =\r
  32                 'Agrave,Aacute,Acirc,Atilde,Auml,Aring,AElig,Ccedil,Egrave,Eacute,' +\r
  33                 'Ecirc,Euml,Igrave,Iacute,Icirc,Iuml,ETH,Ntilde,Ograve,Oacute,Ocirc,' +\r
  34                 'Otilde,Ouml,Oslash,Ugrave,Uacute,Ucirc,Uuml,Yacute,THORN,szlig,' +\r
  35                 'agrave,aacute,acirc,atilde,auml,aring,aelig,ccedil,egrave,eacute,' +\r
  36                 'ecirc,euml,igrave,iacute,icirc,iuml,eth,ntilde,ograve,oacute,ocirc,' +\r
  37                 'otilde,ouml,oslash,ugrave,uacute,ucirc,uuml,yacute,thorn,yuml,' +\r
  38                 'OElig,oelig,Scaron,scaron,Yuml';\r
  39 \r
  40         // Greek Letters Entities.\r
  41         var greek =\r
  42                 'Alpha,Beta,Gamma,Delta,Epsilon,Zeta,Eta,Theta,Iota,Kappa,Lambda,Mu,' +\r
  43                 'Nu,Xi,Omicron,Pi,Rho,Sigma,Tau,Upsilon,Phi,Chi,Psi,Omega,alpha,' +\r
  44                 'beta,gamma,delta,epsilon,zeta,eta,theta,iota,kappa,lambda,mu,nu,xi,' +\r
  45                 'omicron,pi,rho,sigmaf,sigma,tau,upsilon,phi,chi,psi,omega,thetasym,' +\r
  46                 'upsih,piv';\r
  47 \r
  48         /**\r
  49          * Create a mapping table between one character and its entity form from a list of entity names.\r
  50          * @param reverse {Boolean} Whether to create a reverse map from the entity string form to an actual character.\r
  51          */\r
  52         function buildTable( entities, reverse )\r
  53         {\r
  54                 var table = {},\r
  55                         regex = [];\r
  56 \r
  57                 // Entities that the browsers DOM don't transform to the final char\r
  58                 // automatically.\r
  59                 var specialTable =\r
  60                         {\r
  61                                 nbsp    : '\u00A0',             // IE | FF\r
  62                                 shy             : '\u00AD',             // IE\r
  63                                 gt              : '\u003E',             // IE | FF |   --   | Opera\r
  64                                 lt              : '\u003C',             // IE | FF | Safari | Opera\r
  65                                 amp : '\u0026'          // ALL\r
  66                         };\r
  67 \r
  68                 entities = entities.replace( /\b(nbsp|shy|gt|lt|amp)(?:,|$)/g, function( match, entity )\r
  69                         {\r
  70                                 var org = reverse ? '&' + entity + ';' : specialTable[ entity ],\r
  71                                         result = reverse ? specialTable[ entity ] : '&' + entity + ';';\r
  72 \r
  73                                 table[ org ] = result;\r
  74                                 regex.push( org );\r
  75                                 return '';\r
  76                         });\r
  77 \r
  78                 if ( !reverse && entities )\r
  79                 {\r
  80                         // Transforms the entities string into an array.\r
  81                         entities = entities.split( ',' );\r
  82 \r
  83                         // Put all entities inside a DOM element, transforming them to their\r
  84                         // final chars.\r
  85                         var div = document.createElement( 'div' ),\r
  86                                 chars;\r
  87                         div.innerHTML = '&' + entities.join( ';&' ) + ';';\r
  88                         chars = div.innerHTML;\r
  89                         div = null;\r
  90 \r
  91                         // Add all chars to the table.\r
  92                         for ( var i = 0 ; i < chars.length ; i++ )\r
  93                         {\r
  94                                 var charAt = chars.charAt( i );\r
  95                                 table[ charAt ] = '&' + entities[ i ] + ';';\r
  96                                 regex.push( charAt );\r
  97                         }\r
  98                 }\r
  99 \r
 100                 table.regex = regex.join( reverse ? '|' : '' );\r
 101 \r
 102                 return table;\r
 103         }\r
 104 \r
 105         CKEDITOR.plugins.add( 'entities',\r
 106         {\r
 107                 afterInit : function( editor )\r
 108                 {\r
 109                         var config = editor.config;\r
 110 \r
 111                         var dataProcessor = editor.dataProcessor,\r
 112                                 htmlFilter = dataProcessor && dataProcessor.htmlFilter;\r
 113 \r
 114                         if ( htmlFilter )\r
 115                         {\r
 116                                 // Mandatory HTML base entities.\r
 117                                 var selectedEntities = '';\r
 118 \r
 119                                 if ( config.basicEntities !== false )\r
 120                                         selectedEntities += htmlbase;\r
 121 \r
 122                                 if ( config.entities )\r
 123                                 {\r
 124                                         selectedEntities += ',' + entities;\r
 125                                         if ( config.entities_latin )\r
 126                                                 selectedEntities += ',' + latin;\r
 127 \r
 128                                         if ( config.entities_greek )\r
 129                                                 selectedEntities += ',' + greek;\r
 130 \r
 131                                         if ( config.entities_additional )\r
 132                                                 selectedEntities += ',' + config.entities_additional;\r
 133                                 }\r
 134 \r
 135                                 var entitiesTable = buildTable( selectedEntities );\r
 136 \r
 137                                 // Create the Regex used to find entities in the text, leave it matches nothing if entities are empty.\r
 138                                 var entitiesRegex = entitiesTable.regex ? '[' + entitiesTable.regex + ']' : 'a^';\r
 139                                 delete entitiesTable.regex;\r
 140 \r
 141                                 if ( config.entities && config.entities_processNumerical )\r
 142                                         entitiesRegex = '[^ -~]|' + entitiesRegex ;\r
 143 \r
 144                                 entitiesRegex = new RegExp( entitiesRegex, 'g' );\r
 145 \r
 146                                 function getEntity( character )\r
 147                                 {\r
 148                                         return config.entities_processNumerical == 'force' || !entitiesTable[ character ] ?\r
 149                                                    '&#' + character.charCodeAt(0) + ';'\r
 150                                                         : entitiesTable[ character ];\r
 151                                 }\r
 152 \r
 153                                 // Decode entities that the browsers has transformed\r
 154                                 // at first place.\r
 155                                 var baseEntitiesTable = buildTable( [ htmlbase, 'shy' ].join( ',' ) , true ),\r
 156                                         baseEntitiesRegex = new RegExp( baseEntitiesTable.regex, 'g' );\r
 157 \r
 158                                 function getChar( character )\r
 159                                 {\r
 160                                         return baseEntitiesTable[ character ];\r
 161                                 }\r
 162 \r
 163                                 htmlFilter.addRules(\r
 164                                         {\r
 165                                                 text : function( text )\r
 166                                                 {\r
 167                                                         return text.replace( baseEntitiesRegex, getChar )\r
 168                                                                         .replace( entitiesRegex, getEntity );\r
 169                                                 }\r
 170                                         });\r
 171                         }\r
 172                 }\r
 173         });\r
 174 })();\r
 175 \r
 176 /**\r
 177  * Whether to escape basic HTML entities in the document, including:\r
 178  * <ul>\r
 179  * <li><code>nbsp</code></li>\r
 180  * <li><code>gt</code></li>\r
 181  * <li><code>lt</code></li>\r
 182  * <li><code>amp</code></li>\r
 183  * </ul>\r
 184  * <strong>Note:</strong> It should not be subject to change unless when outputting a non-HTML data format like BBCode.\r
 185  * @type Boolean\r
 186  * @default <code>true</code>\r
 187  * @example\r
 188  * config.basicEntities = false;\r
 189  */\r
 190 CKEDITOR.config.basicEntities = true;\r
 191 \r
 192 /**\r
 193  * Whether to use HTML entities in the output.\r
 194  * @name CKEDITOR.config.entities\r
 195  * @type Boolean\r
 196  * @default <code>true</code>\r
 197  * @example\r
 198  * config.entities = false;\r
 199  */\r
 200 CKEDITOR.config.entities = true;\r
 201 \r
 202 /**\r
 203  * Whether to convert some Latin characters (Latin alphabet No&#46; 1, ISO 8859-1)\r
 204  * to HTML entities. The list of entities can be found in the\r
 205  * <a href="http://www.w3.org/TR/html4/sgml/entities.html#h-24.2.1">W3C HTML 4.01 Specification, section 24.2.1</a>.\r
 206  * @name CKEDITOR.config.entities_latin\r
 207  * @type Boolean\r
 208  * @default <code>true</code>\r
 209  * @example\r
 210  * config.entities_latin = false;\r
 211  */\r
 212 CKEDITOR.config.entities_latin = true;\r
 213 \r
 214 /**\r
 215  * Whether to convert some symbols, mathematical symbols, and Greek letters to\r
 216  * HTML entities. This may be more relevant for users typing text written in Greek.\r
 217  * The list of entities can be found in the\r
 218  * <a href="http://www.w3.org/TR/html4/sgml/entities.html#h-24.3.1">W3C HTML 4.01 Specification, section 24.3.1</a>.\r
 219  * @name CKEDITOR.config.entities_greek\r
 220  * @type Boolean\r
 221  * @default <code>true</code>\r
 222  * @example\r
 223  * config.entities_greek = false;\r
 224  */\r
 225 CKEDITOR.config.entities_greek = true;\r
 226 \r
 227 /**\r
 228  * Whether to convert all remaining characters not included in the ASCII\r
 229  * character table to their relative decimal numeric representation of HTML entity.\r
 230  * When set to <code>force</code>, it will convert all entities into this format.\r
 231  * For example the phrase "This is Chinese: &#27721;&#35821;." is output\r
 232  * as "This is Chinese: &amp;#27721;&amp;#35821;."\r
 233  * @name CKEDITOR.config.entities_processNumerical\r
 234  * @type Boolean|String\r
 235  * @default <code>false</code>\r
 236  * @example\r
 237  * config.entities_processNumerical = true;\r
 238  * config.entities_processNumerical = 'force';          //Converts from "&nbsp;" into "&#160;";\r
 239  */\r
 240 \r
 241 /**\r
 242  * A comma separated list of  additional entities to be used. Entity names\r
 243  * or numbers must be used in a form that excludes the "&amp;" prefix and the ";" ending.\r
 244  * @name CKEDITOR.config.entities_additional\r
 245  * @default <code>'#39'</code>  (The single quote (') character.)\r
 246  * @type String\r
 247  * @example\r
 248  * config.entities_additional = '#1049';                // Adds Cyrillic capital letter Short I (&#1049;).\r
 249  */\r
 250 CKEDITOR.config.entities_additional = '#39';\r