Squire/source/Clean.js

/*jshint strict:false, undef:false, unused:false */

var allowedBlock = /^(?:A(?:DDRESS|RTICLE|SIDE|UDIO)|BLOCKQUOTE|CAPTION|D(?:[DLT]|IV)|F(?:IGURE|OOTER)|H[1-6]|HEADER|L(?:ABEL|EGEND|I)|O(?:L|UTPUT)|P(?:RE)?|SECTION|T(?:ABLE|BODY|D|FOOT|H|HEAD|R)|UL)$/;

var fontSizes = {
    1: 10,
    2: 13,
    3: 16,
    4: 18,
    5: 24,
    6: 32,
    7: 48
};

var spanToSemantic = {
    backgroundColor: {
        regexp: notWS,
        replace: function ( doc, colour ) {
            return createElement( doc, 'SPAN', {
                'class': 'highlight',
                style: 'background-color: ' + colour
            });
        }
    },
    color: {
        regexp: notWS,
        replace: function ( doc, colour ) {
            return createElement( doc, 'SPAN', {
                'class': 'colour',
                style: 'color:' + colour
            });
        }
    },
    fontWeight: {
        regexp: /^bold/i,
        replace: function ( doc ) {
            return createElement( doc, 'B' );
        }
    },
    fontStyle: {
        regexp: /^italic/i,
        replace: function ( doc ) {
            return createElement( doc, 'I' );
        }
    },
    fontFamily: {
        regexp: notWS,
        replace: function ( doc, family ) {
            return createElement( doc, 'SPAN', {
                'class': 'font',
                style: 'font-family:' + family
            });
        }
    },
    fontSize: {
        regexp: notWS,
        replace: function ( doc, size ) {
            return createElement( doc, 'SPAN', {
                'class': 'size',
                style: 'font-size:' + size
            });
        }
    }
};

var replaceWithTag = function ( tag ) {
    return function ( node, parent ) {
        var el = createElement( node.ownerDocument, tag );
        parent.replaceChild( el, node );
        el.appendChild( empty( node ) );
        return el;
    };
};

var stylesRewriters = {
    SPAN: function ( span, parent ) {
        var style = span.style,
            doc = span.ownerDocument,
            attr, converter, css, newTreeBottom, newTreeTop, el;

        for ( attr in spanToSemantic ) {
            converter = spanToSemantic[ attr ];
            css = style[ attr ];
            if ( css && converter.regexp.test( css ) ) {
                el = converter.replace( doc, css );
                if ( newTreeBottom ) {
                    newTreeBottom.appendChild( el );
                }
                newTreeBottom = el;
                if ( !newTreeTop ) {
                    newTreeTop = el;
                }
            }
        }

        if ( newTreeTop ) {
            newTreeBottom.appendChild( empty( span ) );
            parent.replaceChild( newTreeTop, span );
        }

        return newTreeBottom || span;
    },
    STRONG: replaceWithTag( 'B' ),
    EM: replaceWithTag( 'I' ),
    STRIKE: replaceWithTag( 'S' ),
    FONT: function ( node, parent ) {
        var face = node.face,
            size = node.size,
            colour = node.color,
            doc = node.ownerDocument,
            fontSpan, sizeSpan, colourSpan,
            newTreeBottom, newTreeTop;
        if ( face ) {
            fontSpan = createElement( doc, 'SPAN', {
                'class': 'font',
                style: 'font-family:' + face
            });
            newTreeTop = fontSpan;
            newTreeBottom = fontSpan;
        }
        if ( size ) {
            sizeSpan = createElement( doc, 'SPAN', {
                'class': 'size',
                style: 'font-size:' + fontSizes[ size ] + 'px'
            });
            if ( !newTreeTop ) {
                newTreeTop = sizeSpan;
            }
            if ( newTreeBottom ) {
                newTreeBottom.appendChild( sizeSpan );
            }
            newTreeBottom = sizeSpan;
        }
        if ( colour && /^#?([\dA-F]{3}){1,2}$/i.test( colour ) ) {
            if ( colour.charAt( 0 ) !== '#' ) {
                colour = '#' + colour;
            }
            colourSpan = createElement( doc, 'SPAN', {
                'class': 'colour',
                style: 'color:' + colour
            });
            if ( !newTreeTop ) {
                newTreeTop = colourSpan;
            }
            if ( newTreeBottom ) {
                newTreeBottom.appendChild( colourSpan );
            }
            newTreeBottom = colourSpan;
        }
        if ( !newTreeTop ) {
            newTreeTop = newTreeBottom = createElement( doc, 'SPAN' );
        }
        parent.replaceChild( newTreeTop, node );
        newTreeBottom.appendChild( empty( node ) );
        return newTreeBottom;
    },
    TT: function ( node, parent ) {
        var el = createElement( node.ownerDocument, 'SPAN', {
            'class': 'font',
            style: 'font-family:menlo,consolas,"courier new",monospace'
        });
        parent.replaceChild( el, node );
        el.appendChild( empty( node ) );
        return el;
    }
};

/*
    Two purposes:

    1. Remove nodes we don't want, such as weird <o:p> tags, comment nodes
       and whitespace nodes.
    2. Convert inline tags into our preferred format.
*/
var cleanTree = function ( node, allowStyles ) {
    var children = node.childNodes,
        i, l, child, nodeName, nodeType, rewriter, childLength,
        data, j, ll;
    for ( i = 0, l = children.length; i < l; i += 1 ) {
        child = children[i];
        nodeName = child.nodeName;
        nodeType = child.nodeType;
        rewriter = stylesRewriters[ nodeName ];
        if ( nodeType === ELEMENT_NODE ) {
            childLength = child.childNodes.length;
            if ( rewriter ) {
                child = rewriter( child, node );
            } else if ( !allowedBlock.test( nodeName ) &&
                    !isInline( child ) ) {
                i -= 1;
                l += childLength - 1;
                node.replaceChild( empty( child ), child );
                continue;
            } else if ( !allowStyles && child.style.cssText ) {
                child.removeAttribute( 'style' );
            }
            if ( childLength ) {
                cleanTree( child, allowStyles );
            }
        } else {
            if ( nodeType === TEXT_NODE ) {
                data = child.data;
                // Use \S instead of notWS, because we want to remove nodes
                // which are just nbsp, in order to cleanup <div>nbsp<br></div>
                // construct.
                if ( /\S/.test( data ) ) {
                    // If the parent node is inline, don't trim this node as
                    // it probably isn't at the end of the block.
                    if ( isInline( node ) ) {
                        continue;
                    }
                    j = 0;
                    ll = data.length;
                    if ( !i || !isInline( children[ i - 1 ] ) ) {
                        while ( j < ll && !notWS.test( data.charAt( j ) ) ) {
                            j += 1;
                        }
                        if ( j ) {
                            child.data = data = data.slice( j );
                            ll -= j;
                        }
                    }
                    if ( i + 1 === l || !isInline( children[ i + 1 ] ) ) {
                        j = ll;
                        while ( j > 0 && !notWS.test( data.charAt( j - 1 ) ) ) {
                            j -= 1;
                        }
                        if ( j < ll ) {
                            child.data = data.slice( 0, j );
                        }
                    }
                    continue;
                }
                // If we have just white space, it may still be important if it
                // separates two inline nodes, e.g. "<a>link</a> <a>link</a>".
                else if ( i && i + 1 < l &&
                        isInline( children[ i - 1 ] ) &&
                        isInline( children[ i + 1 ] ) ) {
                    child.data = ' ';
                    continue;
                }
            }
            node.removeChild( child );
            i -= 1;
            l -= 1;
        }
    }
    return node;
};

// ---

var removeEmptyInlines = function removeEmptyInlines ( root ) {
    var children = root.childNodes,
        l = children.length,
        child;
    while ( l-- ) {
        child = children[l];
        if ( child.nodeType === ELEMENT_NODE && !isLeaf( child ) ) {
            removeEmptyInlines( child );
            if ( isInline( child ) && !child.firstChild ) {
                root.removeChild( child );
            }
        } else if ( child.nodeType === TEXT_NODE && !child.data ) {
            root.removeChild( child );
        }
    }
};

// ---

var notWSTextNode = function ( node ) {
    return node.nodeType === ELEMENT_NODE ?
        node.nodeName === 'BR' :
        notWS.test( node.data );
};
var isLineBreak = function ( br ) {
    var block = br.parentNode,
        walker;
    while ( isInline( block ) ) {
        block = block.parentNode;
    }
    walker = new TreeWalker(
        block, SHOW_ELEMENT|SHOW_TEXT, notWSTextNode );
    walker.currentNode = br;
    return !!walker.nextNode();
};

// <br> elements are treated specially, and differently depending on the
// browser, when in rich text editor mode. When adding HTML from external
// sources, we must remove them, replacing the ones that actually affect
// line breaks by wrapping the inline text in a <div>. Browsers that want <br>
// elements at the end of each block will then have them added back in a later
// fixCursor method call.
var cleanupBRs = function ( root ) {
    var brs = root.querySelectorAll( 'BR' ),
        brBreaksLine = [],
        l = brs.length,
        i, br, parent;

    // Must calculate whether the <br> breaks a line first, because if we
    // have two <br>s next to each other, after the first one is converted
    // to a block split, the second will be at the end of a block and
    // therefore seem to not be a line break. But in its original context it
    // was, so we should also convert it to a block split.
    for ( i = 0; i < l; i += 1 ) {
        brBreaksLine[i] = isLineBreak( brs[i] );
    }
    while ( l-- ) {
        br = brs[l];
        // Cleanup may have removed it
        parent = br.parentNode;
        if ( !parent ) { continue; }
        // If it doesn't break a line, just remove it; it's not doing
        // anything useful. We'll add it back later if required by the
        // browser. If it breaks a line, wrap the content in div tags
        // and replace the brs.
        if ( !brBreaksLine[l] ) {
            detach( br );
        } else if ( !isInline( parent ) ) {
            fixContainer( parent );
        }
    }
};