_source/core/htmlparser.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 /**\r
   7  * Creates a {@link CKEDITOR.htmlParser} class instance.\r
   8  * @class Provides an "event like" system to parse strings of HTML data.\r
   9  * @example\r
  10  * var parser = new CKEDITOR.htmlParser();\r
  11  * parser.onTagOpen = function( tagName, attributes, selfClosing )\r
  12  *     {\r
  13  *         alert( tagName );\r
  14  *     };\r
  15  * parser.parse( '&lt;p&gt;Some &lt;b&gt;text&lt;/b&gt;.&lt;/p&gt;' );\r
  16  */\r
  17 CKEDITOR.htmlParser = function()\r
  18 {\r
  19         this._ =\r
  20         {\r
  21                 htmlPartsRegex : new RegExp( '<(?:(?:\\/([^>]+)>)|(?:!--([\\S|\\s]*?)-->)|(?:([^\\s>]+)\\s*((?:(?:[^"\'>]+)|(?:"[^"]*")|(?:\'[^\']*\'))*)\\/?>))', 'g' )\r
  22         };\r
  23 };\r
  24 \r
  25 (function()\r
  26 {\r
  27         var attribsRegex        = /([\w\-:.]+)(?:(?:\s*=\s*(?:(?:"([^"]*)")|(?:'([^']*)')|([^\s>]+)))|(?=\s|$))/g,\r
  28                 emptyAttribs    = {checked:1,compact:1,declare:1,defer:1,disabled:1,ismap:1,multiple:1,nohref:1,noresize:1,noshade:1,nowrap:1,readonly:1,selected:1};\r
  29 \r
  30         CKEDITOR.htmlParser.prototype =\r
  31         {\r
  32                 /**\r
  33                  * Function to be fired when a tag opener is found. This function\r
  34                  * should be overriden when using this class.\r
  35                  * @param {String} tagName The tag name. The name is guarantted to be\r
  36                  *              lowercased.\r
  37                  * @param {Object} attributes An object containing all tag attributes. Each\r
  38                  *              property in this object represent and attribute name and its\r
  39                  *              value is the attribute value.\r
  40                  * @param {Boolean} selfClosing true if the tag closes itself, false if the\r
  41                  *              tag doesn't.\r
  42                  * @example\r
  43                  * var parser = new CKEDITOR.htmlParser();\r
  44                  * parser.onTagOpen = function( tagName, attributes, selfClosing )\r
  45                  *     {\r
  46                  *         alert( tagName );  // e.g. "b"\r
  47                  *     });\r
  48                  * parser.parse( "&lt;!-- Example --&gt;&lt;b&gt;Hello&lt;/b&gt;" );\r
  49                  */\r
  50                 onTagOpen       : function() {},\r
  51 \r
  52                 /**\r
  53                  * Function to be fired when a tag closer is found. This function\r
  54                  * should be overriden when using this class.\r
  55                  * @param {String} tagName The tag name. The name is guarantted to be\r
  56                  *              lowercased.\r
  57                  * @example\r
  58                  * var parser = new CKEDITOR.htmlParser();\r
  59                  * parser.onTagClose = function( tagName )\r
  60                  *     {\r
  61                  *         alert( tagName );  // e.g. "b"\r
  62                  *     });\r
  63                  * parser.parse( "&lt;!-- Example --&gt;&lt;b&gt;Hello&lt;/b&gt;" );\r
  64                  */\r
  65                 onTagClose      : function() {},\r
  66 \r
  67                 /**\r
  68                  * Function to be fired when text is found. This function\r
  69                  * should be overriden when using this class.\r
  70                  * @param {String} text The text found.\r
  71                  * @example\r
  72                  * var parser = new CKEDITOR.htmlParser();\r
  73                  * parser.onText = function( text )\r
  74                  *     {\r
  75                  *         alert( text );  // e.g. "Hello"\r
  76                  *     });\r
  77                  * parser.parse( "&lt;!-- Example --&gt;&lt;b&gt;Hello&lt;/b&gt;" );\r
  78                  */\r
  79                 onText          : function() {},\r
  80 \r
  81                 /**\r
  82                  * Function to be fired when CDATA section is found. This function\r
  83                  * should be overriden when using this class.\r
  84                  * @param {String} cdata The CDATA been found.\r
  85                  * @example\r
  86                  * var parser = new CKEDITOR.htmlParser();\r
  87                  * parser.onCDATA = function( cdata )\r
  88                  *     {\r
  89                  *         alert( cdata );  // e.g. "var hello;"\r
  90                  *     });\r
  91                  * parser.parse( "&lt;script&gt;var hello;&lt;/script&gt;" );\r
  92                  */\r
  93                 onCDATA         : function() {},\r
  94 \r
  95                 /**\r
  96                  * Function to be fired when a commend is found. This function\r
  97                  * should be overriden when using this class.\r
  98                  * @param {String} comment The comment text.\r
  99                  * @example\r
 100                  * var parser = new CKEDITOR.htmlParser();\r
 101                  * parser.onComment = function( comment )\r
 102                  *     {\r
 103                  *         alert( comment );  // e.g. " Example "\r
 104                  *     });\r
 105                  * parser.parse( "&lt;!-- Example --&gt;&lt;b&gt;Hello&lt;/b&gt;" );\r
 106                  */\r
 107                 onComment       : function() {},\r
 108 \r
 109                 /**\r
 110                  * Parses text, looking for HTML tokens, like tag openers or closers,\r
 111                  * or comments. This function fires the onTagOpen, onTagClose, onText\r
 112                  * and onComment function during its execution.\r
 113                  * @param {String} html The HTML to be parsed.\r
 114                  * @example\r
 115                  * var parser = new CKEDITOR.htmlParser();\r
 116                  * // The onTagOpen, onTagClose, onText and onComment should be overriden\r
 117                  * // at this point.\r
 118                  * parser.parse( "&lt;!-- Example --&gt;&lt;b&gt;Hello&lt;/b&gt;" );\r
 119                  */\r
 120                 parse : function( html )\r
 121                 {\r
 122                         var parts,\r
 123                                 tagName,\r
 124                                 nextIndex = 0,\r
 125                                 cdata;  // The collected data inside a CDATA section.\r
 126 \r
 127                         while ( ( parts = this._.htmlPartsRegex.exec( html ) ) )\r
 128                         {\r
 129                                 var tagIndex = parts.index;\r
 130                                 if ( tagIndex > nextIndex )\r
 131                                 {\r
 132                                         var text = html.substring( nextIndex, tagIndex );\r
 133 \r
 134                                         if ( cdata )\r
 135                                                 cdata.push( text );\r
 136                                         else\r
 137                                                 this.onText( text );\r
 138                                 }\r
 139 \r
 140                                 nextIndex = this._.htmlPartsRegex.lastIndex;\r
 141 \r
 142                                 /*\r
 143                                  "parts" is an array with the following items:\r
 144                                         0 : The entire match for opening/closing tags and comments.\r
 145                                         1 : Group filled with the tag name for closing tags.\r
 146                                         2 : Group filled with the comment text.\r
 147                                         3 : Group filled with the tag name for opening tags.\r
 148                                         4 : Group filled with the attributes part of opening tags.\r
 149                                  */\r
 150 \r
 151                                 // Closing tag\r
 152                                 if ( ( tagName = parts[ 1 ] ) )\r
 153                                 {\r
 154                                         tagName = tagName.toLowerCase();\r
 155 \r
 156                                         if ( cdata && CKEDITOR.dtd.$cdata[ tagName ] )\r
 157                                         {\r
 158                                                 // Send the CDATA data.\r
 159                                                 this.onCDATA( cdata.join('') );\r
 160                                                 cdata = null;\r
 161                                         }\r
 162 \r
 163                                         if ( !cdata )\r
 164                                         {\r
 165                                                 this.onTagClose( tagName );\r
 166                                                 continue;\r
 167                                         }\r
 168                                 }\r
 169 \r
 170                                 // If CDATA is enabled, just save the raw match.\r
 171                                 if ( cdata )\r
 172                                 {\r
 173                                         cdata.push( parts[ 0 ] );\r
 174                                         continue;\r
 175                                 }\r
 176 \r
 177                                 // Opening tag\r
 178                                 if ( ( tagName = parts[ 3 ] ) )\r
 179                                 {\r
 180                                         tagName = tagName.toLowerCase();\r
 181 \r
 182                                         // There are some tag names that can break things, so let's\r
 183                                         // simply ignore them when parsing. (#5224)\r
 184                                         if ( /="/.test( tagName ) )\r
 185                                                 continue;\r
 186 \r
 187                                         var attribs = {},\r
 188                                                 attribMatch,\r
 189                                                 attribsPart = parts[ 4 ],\r
 190                                                 selfClosing = !!( attribsPart && attribsPart.charAt( attribsPart.length - 1 ) == '/' );\r
 191 \r
 192                                         if ( attribsPart )\r
 193                                         {\r
 194                                                 while ( ( attribMatch = attribsRegex.exec( attribsPart ) ) )\r
 195                                                 {\r
 196                                                         var attName = attribMatch[1].toLowerCase(),\r
 197                                                                 attValue = attribMatch[2] || attribMatch[3] || attribMatch[4] || '';\r
 198 \r
 199                                                         if ( !attValue && emptyAttribs[ attName ] )\r
 200                                                                 attribs[ attName ] = attName;\r
 201                                                         else\r
 202                                                                 attribs[ attName ] = attValue;\r
 203                                                 }\r
 204                                         }\r
 205 \r
 206                                         this.onTagOpen( tagName, attribs, selfClosing );\r
 207 \r
 208                                         // Open CDATA mode when finding the appropriate tags.\r
 209                                         if ( !cdata && CKEDITOR.dtd.$cdata[ tagName ] )\r
 210                                                 cdata = [];\r
 211 \r
 212                                         continue;\r
 213                                 }\r
 214 \r
 215                                 // Comment\r
 216                                 if ( ( tagName = parts[ 2 ] ) )\r
 217                                         this.onComment( tagName );\r
 218                         }\r
 219 \r
 220                         if ( html.length > nextIndex )\r
 221                                 this.onText( html.substring( nextIndex, html.length ) );\r
 222                 }\r
 223         };\r
 224 })();\r