--- /dev/null
+/*\r
+Copyright (c) 2003-2009, CKSource - Frederico Knabben. All rights reserved.\r
+For licensing, see LICENSE.html or http://ckeditor.com/license\r
+*/\r
+\r
+/**\r
+ * HTML text parser.\r
+ * @constructor\r
+ * @example\r
+ */\r
+CKEDITOR.htmlParser = function()\r
+{\r
+ this._ =\r
+ {\r
+ htmlPartsRegex : new RegExp( '<(?:(?:\\/([^>]+)>)|(?:!--([\\S|\\s]*?)-->)|(?:([^\\s>]+)\\s*((?:(?:[^"\'>]+)|(?:"[^"]*")|(?:\'[^\']*\'))*)\\/?>))', 'g' )\r
+ };\r
+};\r
+\r
+(function()\r
+{\r
+ var attribsRegex = /([\w:]+)(?:(?:\s*=\s*(?:(?:"([^"]*)")|(?:'([^']*)')|([^\s>]+)))|(?=\s|$))/g,\r
+ 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
+\r
+ CKEDITOR.htmlParser.prototype =\r
+ {\r
+ /**\r
+ * Function to be fired when a tag opener is found. This function\r
+ * should be overriden when using this class.\r
+ * @param {String} tagName The tag name. The name is guarantted to be\r
+ * lowercased.\r
+ * @param {Object} attributes An object containing all tag attributes. Each\r
+ * property in this object represent and attribute name and its\r
+ * value is the attribute value.\r
+ * @param {Boolean} selfClosing true if the tag closes itself, false if the\r
+ * tag doesn't.\r
+ * @example\r
+ * var parser = new CKEDITOR.htmlParser();\r
+ * parser.onTagOpen = function( tagName, attributes, selfClosing )\r
+ * {\r
+ * alert( tagName ); // e.g. "b"\r
+ * });\r
+ * parser.parse( "<!-- Example --><b>Hello</b>" );\r
+ */\r
+ onTagOpen : function() {},\r
+\r
+ /**\r
+ * Function to be fired when a tag closer is found. This function\r
+ * should be overriden when using this class.\r
+ * @param {String} tagName The tag name. The name is guarantted to be\r
+ * lowercased.\r
+ * @example\r
+ * var parser = new CKEDITOR.htmlParser();\r
+ * parser.onTagClose = function( tagName )\r
+ * {\r
+ * alert( tagName ); // e.g. "b"\r
+ * });\r
+ * parser.parse( "<!-- Example --><b>Hello</b>" );\r
+ */\r
+ onTagClose : function() {},\r
+\r
+ /**\r
+ * Function to be fired when text is found. This function\r
+ * should be overriden when using this class.\r
+ * @param {String} text The text found.\r
+ * @example\r
+ * var parser = new CKEDITOR.htmlParser();\r
+ * parser.onText = function( text )\r
+ * {\r
+ * alert( text ); // e.g. "Hello"\r
+ * });\r
+ * parser.parse( "<!-- Example --><b>Hello</b>" );\r
+ */\r
+ onText : function() {},\r
+\r
+ /**\r
+ * Function to be fired when CDATA section is found. This function\r
+ * should be overriden when using this class.\r
+ * @param {String} cdata The CDATA been found.\r
+ * @example\r
+ * var parser = new CKEDITOR.htmlParser();\r
+ * parser.onCDATA = function( cdata )\r
+ * {\r
+ * alert( cdata ); // e.g. "var hello;"\r
+ * });\r
+ * parser.parse( "<script>var hello;</script>" );\r
+ */\r
+ onCDATA : function() {},\r
+\r
+ /**\r
+ * Function to be fired when a commend is found. This function\r
+ * should be overriden when using this class.\r
+ * @param {String} comment The comment text.\r
+ * @example\r
+ * var parser = new CKEDITOR.htmlParser();\r
+ * parser.onText = function( comment )\r
+ * {\r
+ * alert( comment ); // e.g. " Example "\r
+ * });\r
+ * parser.parse( "<!-- Example --><b>Hello</b>" );\r
+ */\r
+ onComment : function() {},\r
+\r
+ /**\r
+ * Parses text, looking for HTML tokens, like tag openers or closers,\r
+ * or comments. This function fires the onTagOpen, onTagClose, onText\r
+ * and onComment function during its execution.\r
+ * @param {String} html The HTML to be parsed.\r
+ * @example\r
+ * var parser = new CKEDITOR.htmlParser();\r
+ * // The onTagOpen, onTagClose, onText and onComment should be overriden\r
+ * // at this point.\r
+ * parser.parse( "<!-- Example --><b>Hello</b>" );\r
+ */\r
+ parse : function( html )\r
+ {\r
+ var parts,\r
+ tagName,\r
+ nextIndex = 0,\r
+ cdata; // The collected data inside a CDATA section.\r
+\r
+ while ( ( parts = this._.htmlPartsRegex.exec( html ) ) )\r
+ {\r
+ var tagIndex = parts.index;\r
+ if ( tagIndex > nextIndex )\r
+ {\r
+ var text = html.substring( nextIndex, tagIndex );\r
+\r
+ if ( cdata )\r
+ cdata.push( text );\r
+ else\r
+ this.onText( text );\r
+ }\r
+\r
+ nextIndex = this._.htmlPartsRegex.lastIndex;\r
+\r
+ /*\r
+ "parts" is an array with the following items:\r
+ 0 : The entire match for opening/closing tags and comments.\r
+ 1 : Group filled with the tag name for closing tags.\r
+ 2 : Group filled with the comment text.\r
+ 3 : Group filled with the tag name for opening tags.\r
+ 4 : Group filled with the attributes part of opening tags.\r
+ */\r
+\r
+ // Closing tag\r
+ if ( ( tagName = parts[ 1 ] ) )\r
+ {\r
+ tagName = tagName.toLowerCase();\r
+\r
+ if ( cdata && CKEDITOR.dtd.$cdata[ tagName ] )\r
+ {\r
+ // Send the CDATA data.\r
+ this.onCDATA( cdata.join('') );\r
+ cdata = null;\r
+ }\r
+\r
+ if ( !cdata )\r
+ {\r
+ this.onTagClose( tagName );\r
+ continue;\r
+ }\r
+ }\r
+\r
+ // If CDATA is enabled, just save the raw match.\r
+ if ( cdata )\r
+ {\r
+ cdata.push( parts[ 0 ] );\r
+ continue;\r
+ }\r
+\r
+ // Opening tag\r
+ if ( ( tagName = parts[ 3 ] ) )\r
+ {\r
+ tagName = tagName.toLowerCase();\r
+ var attribs = {},\r
+ attribMatch,\r
+ attribsPart = parts[ 4 ],\r
+ selfClosing = !!( attribsPart && attribsPart.charAt( attribsPart.length - 1 ) == '/' );\r
+\r
+ if ( attribsPart )\r
+ {\r
+ while ( ( attribMatch = attribsRegex.exec( attribsPart ) ) )\r
+ {\r
+ var attName = attribMatch[1].toLowerCase(),\r
+ attValue = attribMatch[2] || attribMatch[3] || attribMatch[4] || '';\r
+\r
+ if ( !attValue && emptyAttribs[ attName ] )\r
+ attribs[ attName ] = attName;\r
+ else\r
+ attribs[ attName ] = attValue;\r
+ }\r
+ }\r
+\r
+ this.onTagOpen( tagName, attribs, selfClosing );\r
+\r
+ // Open CDATA mode when finding the appropriate tags.\r
+ if ( !cdata && CKEDITOR.dtd.$cdata[ tagName ] )\r
+ cdata = [];\r
+\r
+ continue;\r
+ }\r
+\r
+ // Comment\r
+ if( ( tagName = parts[ 2 ] ) )\r
+ this.onComment( tagName );\r
+ }\r
+\r
+ if ( html.length > nextIndex )\r
+ this.onText( html.substring( nextIndex, html.length ) );\r
+ }\r
+ };\r
+})();\r
Got questions, comments, patches, etc.? Contact Jason Woofenden