_source/plugins/entities/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         // Base HTML entities.\r
   9         var htmlbase = 'nbsp,gt,lt,quot';\r
  10 \r
  11         var entities =\r
  12                 // Latin-1 Entities\r
  13                 '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 it's entity form from a list of entity names.\r
  50          * @param reverse {Boolean} Whether create a reverse map from the entity string form to 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                         };\r
  66 \r
  67                 entities = entities.replace( /\b(nbsp|shy|gt|lt|amp)(?:,|$)/g, function( match, entity )\r
  68                         {\r
  69                                 var org = reverse ? '&' + entity + ';' : specialTable[ entity ],\r
  70                                         result = reverse ? specialTable[ entity ] : '&' + entity + ';';\r
  71 \r
  72                                 table[ org ] = result;\r
  73                                 regex.push( org );\r
  74                                 return '';\r
  75                         });\r
  76 \r
  77                 if ( !reverse )\r
  78                 {\r
  79                         // Transforms the entities string into an array.\r
  80                         entities = entities.split( ',' );\r
  81 \r
  82                         // Put all entities inside a DOM element, transforming them to their\r
  83                         // final chars.\r
  84                         var div = document.createElement( 'div' ),\r
  85                                 chars;\r
  86                         div.innerHTML = '&' + entities.join( ';&' ) + ';';\r
  87                         chars = div.innerHTML;\r
  88                         div = null;\r
  89 \r
  90                         // Add all chars to the table.\r
  91                         for ( var i = 0 ; i < chars.length ; i++ )\r
  92                         {\r
  93                                 var charAt = chars.charAt( i );\r
  94                                 table[ charAt ] = '&' + entities[ i ] + ';';\r
  95                                 regex.push( charAt );\r
  96                         }\r
  97                 }\r
  98 \r
  99                 table.regex = regex.join( reverse ? '|' : '' );\r
 100 \r
 101                 return table;\r
 102         }\r
 103 \r
 104         CKEDITOR.plugins.add( 'entities',\r
 105         {\r
 106                 afterInit : function( editor )\r
 107                 {\r
 108                         var config = editor.config;\r
 109 \r
 110                         var dataProcessor = editor.dataProcessor,\r
 111                                 htmlFilter = dataProcessor && dataProcessor.htmlFilter;\r
 112 \r
 113                         if ( htmlFilter )\r
 114                         {\r
 115                                 // Mandatory HTML base entities.\r
 116                                 var selectedEntities = htmlbase;\r
 117 \r
 118                                 if ( config.entities )\r
 119                                 {\r
 120                                         selectedEntities += ',' + entities;\r
 121                                         if ( config.entities_latin )\r
 122                                                 selectedEntities += ',' + latin;\r
 123 \r
 124                                         if ( config.entities_greek )\r
 125                                                 selectedEntities += ',' + greek;\r
 126 \r
 127                                         if ( config.entities_additional )\r
 128                                                 selectedEntities += ',' + config.entities_additional;\r
 129                                 }\r
 130 \r
 131                                 var entitiesTable = buildTable( selectedEntities );\r
 132 \r
 133                                 // Create the Regex used to find entities in the text.\r
 134                                 var entitiesRegex = '[' + entitiesTable.regex + ']';\r
 135                                 delete entitiesTable.regex;\r
 136 \r
 137                                 if ( config.entities && config.entities_processNumerical )\r
 138                                         entitiesRegex = '[^ -~]|' + entitiesRegex ;\r
 139 \r
 140                                 entitiesRegex = new RegExp( entitiesRegex, 'g' );\r
 141 \r
 142                                 function getEntity( character )\r
 143                                 {\r
 144                                         return config.entities_processNumerical == 'force' || !entitiesTable[ character ] ?\r
 145                                                    '&#' + character.charCodeAt(0) + ';'\r
 146                                                         : entitiesTable[ character ];\r
 147                                 }\r
 148 \r
 149                                 // Decode entities that the browsers has transformed\r
 150                                 // at first place.\r
 151                                 var baseEntitiesTable = buildTable( [ htmlbase, 'shy' ].join( ',' ) , true ),\r
 152                                         baseEntitiesRegex = new RegExp( baseEntitiesTable.regex, 'g' );\r
 153 \r
 154                                 function getChar( character )\r
 155                                 {\r
 156                                         return baseEntitiesTable[ character ];\r
 157                                 }\r
 158 \r
 159                                 htmlFilter.addRules(\r
 160                                         {\r
 161                                                 text : function( text )\r
 162                                                 {\r
 163                                                         return text.replace( baseEntitiesRegex, getChar )\r
 164                                                                         .replace( entitiesRegex, getEntity );\r
 165                                                 }\r
 166                                         });\r
 167                         }\r
 168                 }\r
 169         });\r
 170 })();\r
 171 \r
 172 /**\r
 173  * Whether to use HTML entities in the output.\r
 174  * @type Boolean\r
 175  * @default true\r
 176  * @example\r
 177  * config.entities = false;\r
 178  */\r
 179 CKEDITOR.config.entities = true;\r
 180 \r
 181 /**\r
 182  * Whether to convert some Latin characters (Latin alphabet No&#46; 1, ISO 8859-1)\r
 183  * to HTML entities. The list of entities can be found at the\r
 184  * <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
 185  * @type Boolean\r
 186  * @default true\r
 187  * @example\r
 188  * config.entities_latin = false;\r
 189  */\r
 190 CKEDITOR.config.entities_latin = true;\r
 191 \r
 192 /**\r
 193  * Whether to convert some symbols, mathematical symbols, and Greek letters to\r
 194  * HTML entities. This may be more relevant for users typing text written in Greek.\r
 195  * The list of entities can be found at the\r
 196  * <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
 197  * @type Boolean\r
 198  * @default true\r
 199  * @example\r
 200  * config.entities_greek = false;\r
 201  */\r
 202 CKEDITOR.config.entities_greek = true;\r
 203 \r
 204 /**\r
 205  * Whether to convert all remaining characters, not comprised in the ASCII\r
 206  * character table, to their relative decimal numeric representation of HTML entity.\r
 207  * When specified as the value 'force', it will simply convert all entities into the above form.\r
 208  * For example, the phrase "This is Chinese: &#27721;&#35821;." is outputted\r
 209  * as "This is Chinese: &amp;#27721;&amp;#35821;."\r
 210  * @type Boolean\r
 211  * @type Boolean|String\r
 212  * @default false\r
 213  * @example\r
 214  * config.entities_processNumerical = true;\r
 215  * config.entities_processNumerical = 'force';          //Convert from "&nbsp;" into "&#160;";\r
 216  */\r
 217 CKEDITOR.config.entities_processNumerical = false;\r
 218 \r
 219 /**\r
 220  * An additional list of entities to be used. It's a string containing each\r
 221  * entry separated by a comma. Entities names or number must be used, exclusing\r
 222  * the "&amp;" preffix and the ";" termination.\r
 223  * @default '#39'  // The single quote (') character.\r
 224  * @type String\r
 225  * @example\r
 226  */\r
 227 CKEDITOR.config.entities_additional = '#39';\r