Index: /CKEditor/branches/features/pasting/_source/core/htmlparser/element.js =================================================================== --- /CKEditor/branches/features/pasting/_source/core/htmlparser/element.js (revision 4207) +++ /CKEditor/branches/features/pasting/_source/core/htmlparser/element.js (revision 4208) @@ -1,3 +1,3 @@ -/* +/* Copyright (c) 2003-2009, CKSource - Frederico Knabben. All rights reserved. For licensing, see LICENSE.html or http://ckeditor.com/license @@ -100,12 +100,4 @@ var attributes = this.attributes; - // The "_cke_replacedata" indicates that this element is replacing - // a data snippet, which should be outputted as is. - if ( attributes._cke_replacedata ) - { - writer.write( attributes._cke_replacedata ); - return; - } - // Ignore cke: prefixes when writing HTML. var element = this, Index: /CKEditor/branches/features/pasting/_source/core/tools.js =================================================================== --- /CKEditor/branches/features/pasting/_source/core/tools.js (revision 4207) +++ /CKEditor/branches/features/pasting/_source/core/tools.js (revision 4208) @@ -64,15 +64,4 @@ var clone; - // Array. - if ( obj && ( obj instanceof Array ) ) - { - clone = []; - - for ( var i = 0 ; i < obj.length ; i++ ) - clone[ i ] = this.clone( obj[ i ] ); - - return clone; - } - // "Static" types. if ( obj === null @@ -81,5 +70,6 @@ || ( obj instanceof Number ) || ( obj instanceof Boolean ) - || ( obj instanceof Date ) ) + || ( obj instanceof Date ) + || ( obj instanceof RegExp) ) { return obj; Index: /CKEditor/branches/features/pasting/_source/plugins/pagebreak/plugin.js =================================================================== --- /CKEditor/branches/features/pasting/_source/plugins/pagebreak/plugin.js (revision 4207) +++ /CKEditor/branches/features/pasting/_source/plugins/pagebreak/plugin.js (revision 4208) @@ -56,5 +56,6 @@ div : function( element ) { - var style = element.attributes.style, + var attributes = element.attributes + style = attributes && element.attributes.style, child = style && element.children.length == 1 && element.children[ 0 ], childStyle = child && ( child.name == 'span' ) && child.attributes.style; Index: /CKEditor/branches/features/pasting/_source/plugins/pastefromword/plugin.js =================================================================== --- /CKEditor/branches/features/pasting/_source/plugins/pastefromword/plugin.js (revision 4207) +++ /CKEditor/branches/features/pasting/_source/plugins/pastefromword/plugin.js (revision 4208) @@ -20,5 +20,9 @@ } ); - var config = editor.config; + var config = editor.config, + keepHeadingStructure = config.pasteFromWordKeepsStructure, + ignoreFontFace = config.pasteFromWordIgnoreFontFace, + removeStyleAttr = config.pasteFromWordRemoveStyle; + editor.on( 'paste', function( evt ) { @@ -28,10 +32,149 @@ && /(class=\"?Mso|style=\"[^\"]*\bmso\-|w:WordDocument)/.test( mswordHtml ) ) { + var onlyChildOf = function( element ) + { + var children = element.children, + count = children.length, + firstChild = count && children[ 0 ]; + return firstChild; + }; + + var falsyFilter = function() + { + return false; + }, + dropStyles = function( styles ) + { + return function( styleText ) + { + var rules = []; + styleText.replace( /(?:"| |;|^ )\s*([^ :]+?)\s*:\s*([^;"]+?)\s*(?=;|"|$)/g, + function( match, name, value ) + { + name = name.toLowerCase(); + var namePattern, + valuePattern; + for( var i = 0 ; i < styles.length ; i++ ) + { + namePattern = styles[ i ][ 0 ], + valuePattern = styles[ i ][ 1 ]; + + if ( !( name.match( namePattern ) && ( !valuePattern || value.match( valuePattern ) ) ) ) + rules.push( [ name, value ] ); + } + } ); + + for ( var i = 0 ; i < rules.length ; i++ ) + rules[ i ] = rules[ i ].join( ':' ); + return rules && ( rules.join( ';' ).replace( /\s+/g, '' ) + ';' ); + }; + }; + var filter = editor.pasteProcessor.dataFilter; - // TODO: Migrate the 'CKEDITOR.plugins.pastefromword' to filter rules. filter.addRules( { - - } ); + elementNames : + [ + // Remove style, meta and link elements. + [ /style|meta|link/, '' ] + ], + + elements : + { + $ : function( element ) + { + var tagName = element.name, + child = onlyChildOf( element ); + + var match, level; + // Processing headings. + if ( ( match = tagName.match( /h(\d)/i ) ) && ( level = match[ 1 ] ) ) + { + // Remove empty headings. + if( child + && child.type == CKEDITOR.NODE_TEXT + && !CKEDITOR.tools.trim( child.value ) ) + return false; + + // The original tag send from Word is something like this: + delete element.attributes; + + if ( keepHeadingStructure ) + { + // Word likes to insert extra tags, when using MSIE. (Wierd). + if ( child && child.type == CKEDITOR.NODE_ELEMENT + && child.name.match( /em|font/ ) ) + element.children = child.children; + } + else + { + // Transform headings to divs. + element.name = 'div'; + var bold = new CKEDITOR.htmlParser.element( 'b' ), + font = new CKEDITOR.htmlParser.element( 'font', { size : Math.abs( 7 - level ) } ); + font.children = element.children; + bold.children = [ font ]; + element.children = [ bold ]; + } + } + // Remove empty space inline wrapper. + else if( tagName.match( /u|i|strike|/ ) ) + { + if ( child + && child.type == CKEDITOR.NODE_TEXT + && CKEDITOR.tools.trim( child.value ) == ' ' ) + delete element.name; + } + // Remove dummy wrapper span. + else if( tagName.match( /span|font/ ) ) + { + if( !element.attributes ) + delete element.name; + } + // Remove namespaced element while preserving the content. + else if( tagName.indexOf( ':' ) != -1 ) + { + delete element.name; + } + } + }, + + attributeNames : + [ + // Remove onmouseover and onmouseout events (from MS Word comments effect) + [ /^onmouse(:?out|over)/, '' ], + // Remove lang/language attributes. + [ /^lang/, '' ], + removeStyleAttr ? [ 'style', '' ] : + ignoreFontFace ? [ 'face', '' ] : null + ], + + attributes : + { + // Remove mso-xxx styles. + // Remove margin styles. + 'style' : dropStyles( + [ + [ /^mso-/ ], + [ 'margin', /0(?:cm|in) 0(?:cm|in) 0pt/ ], + [ 'text-indent', '0cm' ], + [ 'page-break-before' ], + [ 'tab-stops' ], + [ 'display', 'none' ], + ignoreFontFace ? [ 'font-family' ] : null + ] ), + + 'class' : falsyFilter, + + // Remove align="left" attribute. + 'align' : function( value ) + { + return ! ( value == 'left ' ); + } + }, + // Remove comments [SF BUG-1481861]. + comment : falsyFilter + }, 5 ); + } } ); @@ -42,4 +185,7 @@ CKEDITOR.plugins.pastefromword = { + /** + * @depprecated Leave it here for reference. + */ cleanWord : function( editor, html, ignoreFont, removeStyles ) { @@ -179,5 +325,5 @@ * config.pasteFromWordRemoveStyle = true; */ -CKEDITOR.config.pasteFromWordRemoveStyle = false; +CKEDITOR.config.pasteFromWordRemoveStyle = true; /** @@ -190,3 +336,3 @@ * config.pasteFromWordKeepsStructure = true; */ -CKEDITOR.config.pasteFromWordKeepsStructure = false; +CKEDITOR.config.pasteFromWordKeepsStructure = true; Index: /CKEditor/branches/features/pasting/_source/plugins/styles/plugin.js =================================================================== --- /CKEditor/branches/features/pasting/_source/plugins/styles/plugin.js (revision 4207) +++ /CKEditor/branches/features/pasting/_source/plugins/styles/plugin.js (revision 4208) @@ -199,6 +199,6 @@ continue; - var elementAttr = element.getAttribute( attName ); - if ( attribs[attName] == + var elementAttr = element.getAttribute( attName ) || ''; + if ( attribs[ attName ] == ( attName == 'style' ? normalizeCssText( elementAttr, false ) : elementAttr ) )