/*\r
-Copyright (c) 2003-2010, CKSource - Frederico Knabben. All rights reserved.\r
+Copyright (c) 2003-2011, CKSource - Frederico Knabben. All rights reserved.\r
For licensing, see LICENSE.html or http://ckeditor.com/license\r
*/\r
\r
(function()\r
{\r
- var entities =\r
-\r
- // Base HTML entities.\r
- 'nbsp,gt,lt,quot,' +\r
+ // Base HTML entities.\r
+ var htmlbase = 'nbsp,gt,lt,amp';\r
\r
+ var entities =\r
// Latin-1 Entities\r
- 'iexcl,cent,pound,curren,yen,brvbar,sect,uml,copy,ordf,laquo,' +\r
+ 'quot,iexcl,cent,pound,curren,yen,brvbar,sect,uml,copy,ordf,laquo,' +\r
'not,shy,reg,macr,deg,plusmn,sup2,sup3,acute,micro,para,middot,' +\r
'cedil,sup1,ordm,raquo,frac14,frac12,frac34,iquest,times,divide,' +\r
\r
'omicron,pi,rho,sigmaf,sigma,tau,upsilon,phi,chi,psi,omega,thetasym,' +\r
'upsih,piv';\r
\r
- function buildTable( entities )\r
+ /**\r
+ * Create a mapping table between one character and its entity form from a list of entity names.\r
+ * @param reverse {Boolean} Whether to create a reverse map from the entity string form to an actual character.\r
+ */\r
+ function buildTable( entities, reverse )\r
{\r
var table = {},\r
regex = [];\r
nbsp : '\u00A0', // IE | FF\r
shy : '\u00AD', // IE\r
gt : '\u003E', // IE | FF | -- | Opera\r
- lt : '\u003C' // IE | FF | Safari | Opera\r
+ lt : '\u003C', // IE | FF | Safari | Opera\r
+ amp : '\u0026' // ALL\r
};\r
\r
entities = entities.replace( /\b(nbsp|shy|gt|lt|amp)(?:,|$)/g, function( match, entity )\r
{\r
- table[ specialTable[ entity ] ] = '&' + entity + ';';\r
- regex.push( specialTable[ entity ] );\r
+ var org = reverse ? '&' + entity + ';' : specialTable[ entity ],\r
+ result = reverse ? specialTable[ entity ] : '&' + entity + ';';\r
+\r
+ table[ org ] = result;\r
+ regex.push( org );\r
return '';\r
});\r
\r
- // Transforms the entities string into an array.\r
- entities = entities.split( ',' );\r
-\r
- // Put all entities inside a DOM element, transforming them to their\r
- // final chars.\r
- var div = document.createElement( 'div' ),\r
- chars;\r
- div.innerHTML = '&' + entities.join( ';&' ) + ';';\r
- chars = div.innerHTML;\r
- div = null;\r
-\r
- // Add all chars to the table.\r
- for ( var i = 0 ; i < chars.length ; i++ )\r
+ if ( !reverse && entities )\r
{\r
- var charAt = chars.charAt( i );\r
- table[ charAt ] = '&' + entities[ i ] + ';';\r
- regex.push( charAt );\r
+ // Transforms the entities string into an array.\r
+ entities = entities.split( ',' );\r
+\r
+ // Put all entities inside a DOM element, transforming them to their\r
+ // final chars.\r
+ var div = document.createElement( 'div' ),\r
+ chars;\r
+ div.innerHTML = '&' + entities.join( ';&' ) + ';';\r
+ chars = div.innerHTML;\r
+ div = null;\r
+\r
+ // Add all chars to the table.\r
+ for ( var i = 0 ; i < chars.length ; i++ )\r
+ {\r
+ var charAt = chars.charAt( i );\r
+ table[ charAt ] = '&' + entities[ i ] + ';';\r
+ regex.push( charAt );\r
+ }\r
}\r
\r
- table.regex = regex.join( '' );\r
+ table.regex = regex.join( reverse ? '|' : '' );\r
\r
return table;\r
}\r
{\r
var config = editor.config;\r
\r
- if ( !config.entities )\r
- return;\r
-\r
var dataProcessor = editor.dataProcessor,\r
htmlFilter = dataProcessor && dataProcessor.htmlFilter;\r
\r
if ( htmlFilter )\r
{\r
- var selectedEntities = entities;\r
+ // Mandatory HTML base entities.\r
+ var selectedEntities = '';\r
\r
- if ( config.entities_latin )\r
- selectedEntities += ',' + latin;\r
+ if ( config.basicEntities !== false )\r
+ selectedEntities += htmlbase;\r
+\r
+ if ( config.entities )\r
+ {\r
+ selectedEntities += ',' + entities;\r
+ if ( config.entities_latin )\r
+ selectedEntities += ',' + latin;\r
\r
- if ( config.entities_greek )\r
- selectedEntities += ',' + greek;\r
+ if ( config.entities_greek )\r
+ selectedEntities += ',' + greek;\r
\r
- if ( config.entities_additional )\r
- selectedEntities += ',' + config.entities_additional;\r
+ if ( config.entities_additional )\r
+ selectedEntities += ',' + config.entities_additional;\r
+ }\r
\r
var entitiesTable = buildTable( selectedEntities );\r
\r
- // Create the Regex used to find entities in the text.\r
- var entitiesRegex = '[' + entitiesTable.regex + ']';\r
+ // Create the Regex used to find entities in the text, leave it matches nothing if entities are empty.\r
+ var entitiesRegex = entitiesTable.regex ? '[' + entitiesTable.regex + ']' : 'a^';\r
delete entitiesTable.regex;\r
\r
- if ( config.entities_processNumerical )\r
+ if ( config.entities && config.entities_processNumerical )\r
entitiesRegex = '[^ -~]|' + entitiesRegex ;\r
\r
entitiesRegex = new RegExp( entitiesRegex, 'g' );\r
\r
+ function getEntity( character )\r
+ {\r
+ return config.entities_processNumerical == 'force' || !entitiesTable[ character ] ?\r
+ '&#' + character.charCodeAt(0) + ';'\r
+ : entitiesTable[ character ];\r
+ }\r
+\r
+ // Decode entities that the browsers has transformed\r
+ // at first place.\r
+ var baseEntitiesTable = buildTable( [ htmlbase, 'shy' ].join( ',' ) , true ),\r
+ baseEntitiesRegex = new RegExp( baseEntitiesTable.regex, 'g' );\r
+\r
function getChar( character )\r
{\r
- return entitiesTable[ character ] || ( '&#' + character.charCodeAt(0) + ';' );\r
+ return baseEntitiesTable[ character ];\r
}\r
\r
htmlFilter.addRules(\r
{\r
text : function( text )\r
{\r
- return text.replace( entitiesRegex, getChar );\r
+ return text.replace( baseEntitiesRegex, getChar )\r
+ .replace( entitiesRegex, getEntity );\r
}\r
});\r
}\r
})();\r
\r
/**\r
+ * Whether to escape basic HTML entities in the document, including:\r
+ * <ul>\r
+ * <li><code>nbsp</code></li>\r
+ * <li><code>gt</code></li>\r
+ * <li><code>lt</code></li>\r
+ * <li><code>amp</code></li>\r
+ * </ul>\r
+ * <strong>Note:</strong> It should not be subject to change unless when outputting a non-HTML data format like BBCode.\r
+ * @type Boolean\r
+ * @default <code>true</code>\r
+ * @example\r
+ * config.basicEntities = false;\r
+ */\r
+CKEDITOR.config.basicEntities = true;\r
+\r
+/**\r
* Whether to use HTML entities in the output.\r
+ * @name CKEDITOR.config.entities\r
* @type Boolean\r
- * @default true\r
+ * @default <code>true</code>\r
* @example\r
* config.entities = false;\r
*/\r
\r
/**\r
* Whether to convert some Latin characters (Latin alphabet No. 1, ISO 8859-1)\r
- * to HTML entities. The list of entities can be found at the\r
+ * to HTML entities. The list of entities can be found in the\r
* <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
+ * @name CKEDITOR.config.entities_latin\r
* @type Boolean\r
- * @default true\r
+ * @default <code>true</code>\r
* @example\r
* config.entities_latin = false;\r
*/\r
/**\r
* Whether to convert some symbols, mathematical symbols, and Greek letters to\r
* HTML entities. This may be more relevant for users typing text written in Greek.\r
- * The list of entities can be found at the\r
+ * The list of entities can be found in the\r
* <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
+ * @name CKEDITOR.config.entities_greek\r
* @type Boolean\r
- * @default true\r
+ * @default <code>true</code>\r
* @example\r
* config.entities_greek = false;\r
*/\r
CKEDITOR.config.entities_greek = true;\r
\r
/**\r
- * Whether to convert all remaining characters, not comprised in the ASCII\r
- * character table, to their relative numeric representation of HTML entity.\r
- * For example, the phrase "This is Chinese: 汉语." is outputted\r
+ * Whether to convert all remaining characters not included in the ASCII\r
+ * character table to their relative decimal numeric representation of HTML entity.\r
+ * When set to <code>force</code>, it will convert all entities into this format.\r
+ * For example the phrase "This is Chinese: 汉语." is output\r
* as "This is Chinese: &#27721;&#35821;."\r
- * @type Boolean\r
- * @default false\r
+ * @name CKEDITOR.config.entities_processNumerical\r
+ * @type Boolean|String\r
+ * @default <code>false</code>\r
* @example\r
* config.entities_processNumerical = true;\r
+ * config.entities_processNumerical = 'force'; //Converts from " " into " ";\r
*/\r
-CKEDITOR.config.entities_processNumerical = false;\r
\r
/**\r
- * An additional list of entities to be used. It's a string containing each\r
- * entry separated by a comma. Entities names or number must be used, exclusing\r
- * the "&" preffix and the ";" termination.\r
- * @default '#39' // The single quote (') character.\r
+ * A comma separated list of additional entities to be used. Entity names\r
+ * or numbers must be used in a form that excludes the "&" prefix and the ";" ending.\r
+ * @name CKEDITOR.config.entities_additional\r
+ * @default <code>'#39'</code> (The single quote (') character.)\r
* @type String\r
* @example\r
+ * config.entities_additional = '#1049'; // Adds Cyrillic capital letter Short I (Й).\r
*/\r
CKEDITOR.config.entities_additional = '#39';\r
Got questions, comments, patches, etc.? Contact Jason Woofenden