+ var re = new RegExp( '(
]*>[\\s\\S]*?)(<\/P>)', 'gi' ) ; // Different because of a IE 5.0 error
+ html = html.replace( re, '' ) ;
+
+ // Remove empty tags (three times, just to be sure).
+ // This also removes any empty anchor
+ html = html.replace( /<([^\s>]+)(\s[^>]*)?>\s*<\/\1>/g, '' ) ;
+ html = html.replace( /<([^\s>]+)(\s[^>]*)?>\s*<\/\1>/g, '' ) ;
+ html = html.replace( /<([^\s>]+)(\s[^>]*)?>\s*<\/\1>/g, '' ) ;
+ }
+
+ return html ;
+ },
+
+ utils :
+ {
+ // Create a which indicate an list item type.
+ createListBulletMarker : function ( type )
+ {
+ var marker = new CKEDITOR.htmlParser.element( 'cke:listbullet' ),
+ // We took the decimal as default because we
+ // use 'ol' as root list.
+ defaultListType = 'decimal';
+
+ // TODO: Support more list type mapping rules.
+ if( !isNaN( type[ 1 ] ) && type[ 2 ] == '.' )
+ type = 'decimal';
+ else if ( type[ 1 ].search( /[l·]/ ) != -1 )
+ type = 'disc';
+ else
+ type = defaultListType;
+
+ // Represent list type as CSS style.
+ if( type != defaultListType )
+ marker.attributes = { style : 'list-style-type:' + type };
+
+ return marker;
+ }
+
+ },
+
+ filters : {
+ /**
+ * A simple filter which always rejecting.
+ */
+ falsyFilter : function( value )
+ {
+ return false;
+ },
+
+ /**
+ * A filter dedicated on the 'style' attribute for dropping/replacing style rules.
+ * @param styles {Array} A triple in form of [ styleNameRegexp, styleValueRegexp, newStyleValue ] where the last two are optional.
+ */
+ stylesFilter : function( styles )
+ {
+ return function( styleText, element )
+ {
+ var rules = [];
+ styleText.replace( /\s*([^ :;]+?)\s*:\s*([^;"]+?)\s*(?=;|$)/g,
+ function( match, name, value )
+ {
+ name = name.toLowerCase();
+ var namePattern,
+ valuePattern,
+ newValue;
+ for( var i = 0 ; i < styles.length && styles[ i ] ; i++ )
+ {
+ namePattern = styles[ i ][ 0 ];
+ valuePattern = styles[ i ][ 1 ];
+ newValue = styles[ i ][ 2 ];
+
+ if ( name.match( namePattern )
+ && ( !valuePattern || value.match( valuePattern ) ) )
+ {
+ if( typeof newValue == 'function' )
+ newValue = newValue( value, element );
+ if( typeof newValue == 'string' )
+ rules.push( [ name, newValue ] );
+ return;
+ }
+ }
+ rules.push( [ name, value ] );
+
+ } );
+
+ for ( var i = 0 ; i < rules.length ; i++ )
+ rules[ i ] = rules[ i ].join( ':' );
+ return rules.length ?
+ ( rules.join( ';' ).replace( /\s+/g, '' ) + ';' )
+ // Remove attribute if there's no styles.
+ : false;
+ };
+ },
+
+ listFilter : function( element )
{
var children = element.children,
- count = children.length,
- firstChild = count && children[ 0 ];
- return firstChild;
- };
-
- var falsyFilter = function()
- {
- return false;
- },
- dropStyles = function( styles )
- {
- return function( styleText )
- {
- var rules = [];
- styleText.replace( /(?:"| |;|^ )\s*([^ :]+?)\s*:\s*([^;"]+?)\s*(?=;|"|$)/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;
- filter.addRules(
- {
- elementNames :
- [
- // Remove style, meta and link elements.
- [ /style|meta|link/, '' ]
- ],
-
- elements :
- {
- $ : function( element )
- {
-
- var tagName = element.name;
-
- var match, level;
- // Processing headings.
- if ( ( match = tagName.match( /h(\d)/i ) ) && ( level = match[ 1 ] ) )
+ length = children.length,
+ child, previousChild;
+ for( var i = 0 ; i < length; i++ )
+ {
+ child = children[ i ];
+
+ // Wrap nested list root with the previous list item.
+ if ( child.name && child.name in { ul : 1, ol : 1 } )
+ {
+ children.splice( i, 1 );
+ previousChild.children.push( child );
+ }
+ previousChild = child;
+ }
+ }
+
+ },
+
+ getRules : function( editor )
+ {
+ var falsyFilter = this.filters.falsyFilter,
+ stylesFilter = this.filters.stylesFilter,
+ listFilter = this.filters.listFilter,
+ createListBulletMarker = this.utils.createListBulletMarker,
+ config = editor.config,
+ ignoreFontFace = config.pasteFromWordIgnoreFontFace,
+ removeStyleAttr = config.pasteFromWordRemoveStyle;
+
+ return {
+
+ elementNames :
+ [
+ // Remove style, meta and link elements.
+ [ /style|meta|link/, '' ]
+ ],
+
+ elements :
+ {
+ $ : function( element )
+ {
+
+ var tagName = element.name || '';
+
+ var match, level;
+ // Processing headings.
+ if ( ( match = tagName.match( /h(\d)/i ) ) && ( level = match[ 1 ] ) )
+ {
+ element.filterChildren();
+ var child = element.onlyChild();
+
+ // Remove empty headings.
+ if( child && child.value
+ && !CKEDITOR.tools.trim( child.value ) )
+ return false;
+ // The original tag send from Word is something like this:
+ delete element.attributes;
+
+ // Word likes to insert extra tags, when using MSIE. (Wierd).
+ if ( child && /em|font/.exec( child.name ) )
+ element.children = child.children;
+ }
+ // Remove inline elements which contain only empty spaces.
+ else if( tagName.match( /^(:?b|u|i|strike|span)$/ ) )
+ {
+ element.filterChildren();
+ var child = element.onlyChild();
+ if ( child && /(:?\s| )+/.exec( child.value ) )
+ delete element.name;
+ }
+ // Remove dummy inline wrappers.
+ else if( tagName.match( /span|font/ ) )
+ {
+ if( !element.attributes )
+ delete element.name;
+ }
+ // Remove namespaced element while preserving the content.
+ else if( tagName.indexOf( ':' ) != -1
+ && tagName.indexOf( 'cke' ) == -1 )
+ {
+ delete element.name;
+ }
+ // Any dtd-valid element which could contain a list.
+ else if( !tagName && element.children
+ || tagName in CKEDITOR.dtd.parentOf( 'ol' ) )
+ {
+ element.filterChildren();
+
+ var children = element.children, child,
+ listItem, // The current processing cke:li element.
+ listItemIndent, // Indent attribute represent the level of it.
+ lastListItem, // The previous one just been added to the list.
+ list, parentList, // Current staging list and it's parent list if any.
+ indent;
+
+ for( var i = 0 ; i < children.length; i++ )
{
- element.filterChildren();
- var child = onlyChildOf( element );
-
- // Remove empty headings.
- if( child && child.value
- && !CKEDITOR.tools.trim( child.value ) )
- return false;
-
- // The original tag send from Word is something like this:
- delete element.attributes;
-
- if ( keepHeadingStructure )
+ child = children[ i ];
+
+ if ( 'cke:li' == child.name )
{
- // Word likes to insert extra tags, when using MSIE. (Wierd).
- if ( child && /em|font/.exec( child.name ) )
- element.children = child.children;
+ child.name = 'li';
+
+ listItem = child;
+ listItemIndent = listItem.attributes.indent,
+ delete listItem.attributes.indent;
+
+ if ( !list )
+ {
+ parentList = list = new CKEDITOR.htmlParser.element( 'ol' );
+ list.add( listItem );
+ children[ i ] = list;
+ }
+ else
+ {
+ if( listItemIndent > indent )
+ {
+ parentList = list;
+ list = new CKEDITOR.htmlParser.element( 'ol' );
+ list.add( listItem );
+ lastListItem.add( list );
+ }
+ else if( listItemIndent < indent )
+ {
+ list = parentList;
+ list.add( listItem );
+ }
+ else
+ list.add( listItem );
+
+ children.splice( i-- , 1 );
+ }
+
+ lastListItem = listItem;
+ indent = listItemIndent;
}
else
+ list = null;
+ }
+
+ }
+ },
+
+ 'ul' : listFilter,
+ 'ol' : listFilter,
+ 'p' : function( element )
+ {
+ element.filterChildren();
+
+ var attrs = element.attributes,
+ children = element.children,
+ firstChild = children && children[ 0 ];
+
+ // been the first child of any paragraph
+ // indicate a list item.
+ if( 'cke:listbullet' == firstChild.name )
+ {
+ element.name = 'cke:li';
+ attrs.style = stylesFilter(
+ [
+ [ 'text-indent' ],
+ [ 'margin-left', null, function( value )
{
- // 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 inline elements which contain only empty spaces.
- else if( tagName.match( /^(:?b|u|i|strike|span)$/ ) )
- {
- element.filterChildren();
- var child = onlyChildOf( element );
- if ( child && /(:?\s| )+/.exec( child.value ) )
- delete element.name;
- }
- // Remove dummy inline wrappers.
- 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;
- }
+ // Resolve indent level from 'margin-left' style.
+ attrs.indent = parseInt( value );
+ } ]
+ ] )( attrs.style, element ) || '' ;
+
+ // Inherit list-type-style from bullet.
+ var listBulletAttrs = firstChild.attributes,
+ listBulletStyle = listBulletAttrs && listBulletAttrs.style;
+ if( listBulletStyle )
+ attrs.style = listBulletStyle;
+ children.splice( 0, 1 );
}
},
-
- attributeNames :
+ 'span' : function( element )
+ {
+ element.filterChildren();
+
+ var attrs = element.attributes,
+ styles = attrs && attrs.style;
+ if( styles )
+ {
+ var marker;
+ stylesFilter(
+ [
+ [ 'mso-list', 'Ignore', function( value, element )
+ {
+
+ var listType = element.firstTextChild().value.match( /([^\s])([.)]?)/ );
+ marker = createListBulletMarker( listType );
+ } ]
+ ] )( styles, element );
+
+ if( marker )
+ return marker;
+ }
+
+ var onlyChild = element.onlyChild();
+ if( onlyChild && 'cke:listbullet' == onlyChild.name )
+ return onlyChild;
+ }
+ },
+
+ attributeNames :
+ [
+ // Remove onmouseover and onmouseout events (from MS Word comments effect)
+ [ /^onmouse(:?out|over)/, '' ],
+ // Remove lang/language attributes.
+ [ /^lang/, '' ],
+ ignoreFontFace ? [ 'face', '' ] : null
+ ],
+
+ attributes :
+ {
+ // Remove mso-xxx styles.
+ // Remove margin styles.
+ 'style' : stylesFilter(
[
- // 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 ' ? false : value;
- }
- },
- // Remove comments [SF BUG-1481861].
- comment : falsyFilter
- }, 5 );
-
- }
- } );
-
+ [ /mso-/ ],
+ [ 'margin', /0(?:cm|in) 0(?:cm|in) 0pt/ ],
+ [ 'text-indent', '0cm' ],
+ [ 'page-break-before' ],
+ [ 'tab-stops' ],
+ [ 'display', 'none' ],
+ [ 'text-align', 'left' ],
+ ignoreFontFace ? [ 'font-family' ] : null,
+ ] ),
+
+ 'class' : falsyFilter,
+
+ // Remove align="left" attribute.
+ 'align' : function( value )
+ {
+ return value == 'left' ? false : value;
+ }
+ },
+ comment : function( value )
+ {
+ var imageSource = value.match( // ),
+ listInfo = value.match( /^\[if !supportLists\]([\s\S]*?)\[endif\]$/ );
+
+ // Image 'src' attribute might be embedded within vml.
+ if( imageSource )
+ {
+ var image = CKEDITOR.htmlParser.fragment.fromHtml( imageSource ).children[ 0 ],
+ attrs = image.attributes;
+ delete attrs[ 'v:shapes'];
+
+ // Try to reveal the real image source from vml elements.
+ var imageData = value.match( /]*o:href=['"](.*?)['"]/ ),
+ imageSrc = imageData && imageData[ 1 ];
+ imageSrc && ( attrs.src = imageSrc );
+ return image;
+ }
+ // Seek for list bullet style indicator.
+ else if ( listInfo )
+ {
+ var listSymbol = listInfo[ 1 ],
+ listType = listSymbol.match( />([^\s])([.)]?)/g, '' ) ;
-
- html = html.replace(/\s*<\/o:p>/g, '') ;
- html = html.replace(/[\s\S]*?<\/o:p>/g, ' ') ;
-
- // Remove mso-xxx styles.
- html = html.replace( /\s*mso-[^:]+:[^;"]+;?/gi, '' ) ;
-
- // Remove margin styles.
- html = html.replace( /\s*MARGIN: 0(?:cm|in) 0(?:cm|in) 0pt\s*;/gi, '' ) ;
- html = html.replace( /\s*MARGIN: 0(?:cm|in) 0(?:cm|in) 0pt\s*"/gi, "\"" ) ;
-
- html = html.replace( /\s*TEXT-INDENT: 0cm\s*;/gi, '' ) ;
- html = html.replace( /\s*TEXT-INDENT: 0cm\s*"/gi, "\"" ) ;
-
- html = html.replace( /\s*TEXT-ALIGN: [^\s;]+;?"/gi, "\"" ) ;
-
- html = html.replace( /\s*PAGE-BREAK-BEFORE: [^\s;]+;?"/gi, "\"" ) ;
-
- html = html.replace( /\s*FONT-VARIANT: [^\s;]+;?"/gi, "\"" ) ;
-
- html = html.replace( /\s*tab-stops:[^;"]*;?/gi, '' ) ;
- html = html.replace( /\s*tab-stops:[^"]*/gi, '' ) ;
-
- // Remove FONT face attributes.
- if ( ignoreFont )
+ var result = {};
+ for( var tag in this )
{
- html = html.replace( /\s*face="[^"]*"/gi, '' ) ;
- html = html.replace( /\s*face=[^ >]*/gi, '' ) ;
-
- html = html.replace( /\s*FONT-FAMILY:[^;"]*;?/gi, '' ) ;
+ if( tag.indexOf( '$' ) == -1 && this[ tag ][ tagName ] )
+ result[ tag ] = 1;
}
-
- // Remove Class attributes
- html = html.replace(/<(\w[^>]*) class=([^ |>]*)([^>]*)/gi, "<$1$3") ;
-
- // Remove styles.
- if ( removeStyles )
- html = html.replace( /<(\w[^>]*) style="([^\"]*)"([^>]*)/gi, "<$1$3" ) ;
-
- // Remove style, meta and link tags
- html = html.replace( /]*>[\s\S]*?<\/STYLE[^>]*>/gi, '' ) ;
- html = html.replace( /<(?:META|LINK)[^>]*>\s*/gi, '' ) ;
-
- // Remove empty styles.
- html = html.replace( /\s*style="\s*"/gi, '' ) ;
-
- html = html.replace( /]*>\s* \s*<\/SPAN>/gi, ' ' ) ;
-
- html = html.replace( /]*><\/SPAN>/gi, '' ) ;
-
- // Remove Lang attributes
- html = html.replace(/<(\w[^>]*) lang=([^ |>]*)([^>]*)/gi, "<$1$3") ;
-
- html = html.replace( /([\s\S]*?)<\/SPAN>/gi, '$1' ) ;
-
- html = html.replace( /([\s\S]*?)<\/FONT>/gi, '$1' ) ;
-
- // Remove XML elements and declarations
- html = html.replace(/<\\?\?xml[^>]*>/gi, '' ) ;
-
- // Remove w: tags with contents.
- html = html.replace( /]*>[\s\S]*?<\/w:[^>]*>/gi, '' ) ;
-
- // Remove Tags with XML namespace declarations: <\/o:p>
- html = html.replace(/<\/?\w+:[^>]*>/gi, '' ) ;
-
- html = html.replace( /<(U|I|STRIKE)> <\/\1>/g, ' ' ) ;
-
- html = html.replace( /\s*<\/H\d>/gi, '' ) ;
-
- // Remove "display:none" tags.
- html = html.replace( /<(\w+)[^>]*\sstyle="[^"]*DISPLAY\s?:\s?none[\s\S]*?<\/\1>/ig, '' ) ;
-
- // Remove language tags
- html = html.replace( /<(\w[^>]*) language=([^ |>]*)([^>]*)/gi, "<$1$3") ;
-
- // Remove onmouseover and onmouseout events (from MS Word comments effect)
- html = html.replace( /<(\w[^>]*) onmouseover="([^\"]*)"([^>]*)/gi, "<$1$3") ;
- html = html.replace( /<(\w[^>]*) onmouseout="([^\"]*)"([^>]*)/gi, "<$1$3") ;
-
- if ( editor.config.pasteFromWordKeepsStructure )
- {
- // The original tag send from Word is something like this:
- html = html.replace( /]*)>/gi, '' ) ;
-
- // Word likes to insert extra tags, when using MSIE. (Wierd).
- html = html.replace( /<(H\d)>]*>([\s\S]*?)<\/FONT><\/\1>/gi, '<$1>$2<\/$1>' );
- html = html.replace( /<(H\d)>([\s\S]*?)<\/EM><\/\1>/gi, '<$1>$2<\/$1>' );
- }
- else
- {
- html = html.replace( /]*)>/gi, '' ) ;
- html = html.replace( /]*)>/gi, '' ) ;
- html = html.replace( /]*)>/gi, '' ) ;
- html = html.replace( /]*)>/gi, '' ) ;
- html = html.replace( /]*)>/gi, '' ) ;
- html = html.replace( /]*)>/gi, '' ) ;
-
- html = html.replace( /<\/H\d>/gi, '<\/font><\/b><\/div>' ) ;
-
- // Transform to
- var re = new RegExp( '(
]*>[\\s\\S]*?)(<\/P>)', 'gi' ) ; // Different because of a IE 5.0 error
- html = html.replace( re, '' ) ;
-
- // Remove empty tags (three times, just to be sure).
- // This also removes any empty anchor
- html = html.replace( /<([^\s>]+)(\s[^>]*)?>\s*<\/\1>/g, '' ) ;
- html = html.replace( /<([^\s>]+)(\s[^>]*)?>\s*<\/\1>/g, '' ) ;
- html = html.replace( /<([^\s>]+)(\s[^>]*)?>\s*<\/\1>/g, '' ) ;
- }
-
- return html ;
- }
-}
+ return result;
+ };
+
+} )();
+
/**