it("filter will run only once", function () {
var count = 0;
var rules = {
tags: {
p: function (el) {
el.filterChildren();
},
span: function () {
count++;
}
}
};
var filter = new HtmlParser.Filter();
filter.addRules(rules);
var writer = new HtmlParser.BasicWriter();
var before = "<p><span></span></p>";
var n = new HtmlParser.Parser(before).parse();
n.writeHtml(writer, filter);
expect(writer.getHtml()).toBe("<p><span></span></p>");
expect(count).toBe(1);
});
it("can filter attributeNames", function () {
var rules = {
attributeNames: [
// 把保存的作为真正的属性，替换掉原来的
// replace(/^_keSaved_/,"")
// _keSavedHref -> href
[ ( /^_keSaved_/ ), '' ],
[ ( /^ke_on/ ), 'on' ],
[ ( /^_ke.*/ ), '' ],
//!TODO 不知道怎么回事会引入
[ ( /^_ks.*/ ), '' ],
[ ( /^ke:.*$/ ), '' ]
]
};
var filter = new HtmlParser.Filter();
filter.addRules(rules);
var writer = new HtmlParser.BasicWriter();
var before = "<p _keSaved_src='y'>x</p>";
var n = new HtmlParser.Parser(before).parse();
n.writeHtml(writer, filter);
expect(writer.getHtml()).toBe("<p src=\"y\">x</p>");
});
it('can filter elementNames', function () {
var dataFilterRules = {
tagNames: [
[ /^script$/i , '' ],
[ /^iframe$/i , '' ],
[ /^style$/i , '' ],
[ /^link$/i , '' ],
[ /^meta$/i , '' ],
[/^\?xml.*$/i, ''],
[/^.*namespace.*$/i, '']
]
};
var filter = new HtmlParser.Filter();
filter.addRules(dataFilterRules);
var writer = new HtmlParser.BasicWriter(),
before = '<script>alert(1);</script>x<link/>' +
'<?xml:namespace prefix = v ns = "urn:schemas-microsoft-com:vml" />' +
'y';
var n = new HtmlParser.Parser(before).parse();
n.writeHtml(writer, filter);
expect(writer.getHtml()).toBe("xy");
});
it("filter children works while modify html", function () {
var rules = {
tags: {
p: function (el) {
el.filterChildren();
}
}
};
var rules2 = {
tags: {
p: function (el) {
el.appendChild(new HtmlParser.Text(" "));
}
}
};
var filter = new HtmlParser.Filter();
filter.addRules(rules);
filter.addRules(rules2);
var writer = new HtmlParser.BasicWriter();
var before = "<p></p>";
var n = new HtmlParser.Parser(before).parse();
n.writeHtml(writer, filter);
expect(writer.getHtml()).toBe("<p> </p>");
});
it("filterChildren should works", function () {
var html = "<div class='ul'><div class='li'>1</div><div class='li'>2</div></div>",
parser = new Parser(html),
node = parser.parse(),
writer = new HtmlParser.BasicWriter(),
filter = new HtmlParser.Filter();
filter.addRules({
tags: {
$: function (el) {
if (el.getAttribute("class") === "li") {
el.nodeName = el.tagName = "li";
el.removeAttribute("class");
} else if (el.getAttribute("class") === 'ul') {
// filter its children first, root node need children info after filtering
el.filterChildren();
var childNodes = el.childNodes;
for (var i = 0, c = childNodes[i]; i < childNodes.length; i++) {
if (c.nodeType === 1 && c.tagName !== "li") {
return;
}
}
el.nodeName = el.tagName = 'ul';
el.removeAttribute("class");
}
}
}
});
node.writeHtml(writer, filter);
expect(writer.getHtml()).toBe("<ul><li>1</li><li>2</li></ul>");
});
it('can replace text', function () {
var html = "<li>12</li><li>21</li>",
parser = new Parser(html),
node = parser.parse(),
writer = new HtmlParser.BasicWriter(),
filter = new HtmlParser.Filter();
filter.addRules({
text: function (value) {
return value.replace(/2/g, '3');
}
});
node.writeHtml(writer, filter);
expect(writer.getHtml()).toBe("<li>13</li><li>31</li>");
});
it("can filter imagedata in vml@ie", function () {
var dataFilterRules = {
tagNames: [
[/^\?xml.*$/i, '']
],
tags: {
$: function (el) {
var tagName = el.tagName || "";
if (tagName.indexOf(':') !== -1 && !/^ke/.test(tagName)) {
if (tagName === 'v:imagedata') {
var href = el.getAttribute('o:href');
if (href) {
el.setAttribute('src', href);
el.removeAttribute('o:href');
}
var title = el.getAttribute('o:title');
if (title) {
el.setAttribute('title', title);
el.removeAttribute("o:title");
}
el.setTagName("img");
} else {
el.setTagName("");
}
}
}
}
};
var filter = new HtmlParser.Filter();
filter.addRules(dataFilterRules);
var writer = new HtmlParser.BasicWriter();
var before = '';
getTextSync('/kissy/src/html-parser/tests/others/editor/vml_img.html', function (d) {
before = d;
});
var n = new HtmlParser.Parser(before).parse();
n.writeHtml(writer, filter);
// S.log(writer.getHtml());
expect(writer.getHtml().indexOf('<img src="xx.jpg" title="me" />') !== -1).toBe(true);
});
it("can modify html structure on fly", function () {
var tailNbspRegex = /^[\t\r\n ]*(?: |\xa0)$/;
// Return the last non-space child node of the block (#4344).
function lastNoneSpaceChild(block) {
var childNodes = block.childNodes,
lastIndex = childNodes.length,
last = childNodes[ lastIndex - 1 ];
while (last && last.nodeType === 3 && !util.trim(last.nodeValue)) {
last = childNodes[ --lastIndex ];
}
return last;
}
function trimFillers(block, fromSource) {
// If the current node is a block, and if we're converting from source or
// we're not in IE then search for and remove any tailing BR node.
// Also, any at the end of blocks are fillers, remove them as well.
// (#2886)
var lastChild = lastNoneSpaceChild(block);
if (lastChild) {
if (( fromSource || !isIE() ) &&
lastChild.nodeType === 1 &&
lastChild.nodeName === 'br') {
block.removeChild(lastChild);
}
else if (lastChild.nodeType === 3 &&
tailNbspRegex.test(lastChild.nodeValue)) {
block.removeChild(lastChild);
}
}
}
function blockNeedsExtension(block) {
var lastChild = lastNoneSpaceChild(block);
return !lastChild || lastChild.nodeType === 1 &&
lastChild.nodeName === 'br' ||
// Some of the controls in form needs extension too,
// to move cursor at the end of the form. (#4791)
block.nodeName === 'form' &&
lastChild.nodeName === 'input';
}
function extendBlockForDisplay(block) {
trimFillers(block, true);
if (blockNeedsExtension(block)) {
//任何浏览器都要加空格！，否则空表格可能间隙太小，不能容下光标
if (isIE()) {
block.appendChild(new HtmlParser.Text('\xa0'));
} else {
//其他浏览器需要加空格??
block.appendChild(new HtmlParser.Text(' '));
block.appendChild(new HtmlParser.Tag('br'));
}
}
}
var dataFilterRules = {
tags: {
p: extendBlockForDisplay
}
};
var filter = new HtmlParser.Filter();
filter.addRules(dataFilterRules);
var writer = new HtmlParser.BasicWriter();
var before = "<p></p><p>1</p>";
var n = new HtmlParser.Parser(before).parse();
n.writeHtml(writer, filter);
if (isIE()) {
expect(writer.getHtml()).toBe("<p>\xa0</p><p>1</p>");
} else {
expect(writer.getHtml()).toBe("<p> <br /></p><p>1</p>");
}
});
it('can filter attribute', function () {
function stylesFilter(styles, whitelist) {
return function (styleText, element) {
var rules = [];
// html-encoded quote might be introduced by 'font-family'
// from MS-Word which confused the following regexp. e.g.
//'font-family: "Lucida, Console"'
String(styleText)
.replace(/"/g, '"')
.replace(/\s*([^ :;]+)\s*:\s*([^;]+)\s*(?=;|$)/g,
function (match, name, value) {
name = name.toLowerCase();
if (name === 'font-family') {
( value = value.replace(/["']/g, ''));
}
var namePattern,
valuePattern,
newValue,
newName;
for (var i = 0; i < styles.length; i++) {
if (styles[ i ]) {
namePattern = styles[ i ][ 0 ];
valuePattern = styles[ i ][ 1 ];
newValue = styles[ i ][ 2 ];
newName = styles[ i ][ 3 ];
if (name.match(namePattern) &&
( !valuePattern || value.match(valuePattern) )) {
name = newName || name;
if (whitelist) {
( newValue = newValue || value );
}
if (typeof newValue === 'function')
newValue = newValue(value, element, name);
// Return an couple indicate both name and value
// changed.
if (newValue && newValue.push) {
name = newValue[ 0 ];
newValue = newValue[ 1 ];
}
if (typeof newValue === 'string')
rules.push([ name, newValue ]);
return;
}
}
}
if (!whitelist) {
rules.push([ name, value ]);
}
});
for (var i = 0; i < rules.length; i++) {
rules[ i ] = rules[ i ].join(':');
}
return rules.length ?
( rules.join(';') + ';' ) : false;
};
}
var filterStyle = stylesFilter([
// word 自有属性名去除
[/mso/i],
[/w:WordDocument/i],
// ie 自有属性名[/mso/i],
[/^-ms/i],
// firefox 自有属性名
[/^-moz/i],
// webkit 自有属性名
[/^-webkit/i]//
]);
var dataFilterRules = {
attributes: {
// word
"class": function (value) {
if (
!value ||
/(^|\s+)Mso/.test(value)
) {
return false;
}
return value;
},
'style': function (value) {
//去除<i style="mso-bidi-font-style: normal">微软垃圾
var re = filterStyle(value);
if (!re) {
return false;
}
return re;
}
}
};
var filter = new HtmlParser.Filter();
filter.addRules(dataFilterRules);
var writer = new HtmlParser.BasicWriter(),
before = '<div class="Mso-list">1</div>' +
'<b class="">2</b>' +
'<span style="mso-bidi-font-style: normal;-ms-k:1;">3</span>';
var n = new HtmlParser.Parser(before).parse();
n.writeHtml(writer, filter);
expect(writer.getHtml()).toBe("<div>1</div><" + "b>2</b><span>3</span>");
});
(function () {
var blockLike = util.merge(dtd.$block, dtd.$listItem, dtd.$tableContent),
falsyFilter = filters.falsyFilter,
stylesFilter = filters.stylesFilter,
createListBulletMarker = utils.createListBulletMarker,
flattenList = filters.flattenList,
assembleList = filters.assembleList,
isListBulletIndicator = utils.isListBulletIndicator,
containsNothingButSpaces = utils.isContainingOnlySpaces,
resolveListItem = utils.resolveList,
convertToPxStr = function (value) {
value = convertToPx(value);
return isNaN(value) ? value : value + 'px';
},
getStyleComponents = utils.getStyleComponents,
listDtdParents = utils.listDtdParents;
wordFilter.addRules({
tagNames: [
// Remove script, meta and link elements.
[ ( /meta|link|script/ ), '' ]
],
root: function (element) {
element.filterChildren();
assembleList(element);
},
tags: {
'^': function (element) {
// Transform CSS style declaration to inline style.
var applyStyleFilter;
if (UA.gecko && ( applyStyleFilter = filters.applyStyleFilter )) {
applyStyleFilter(element);
}
},
$: function (element) {
var tagName = element.nodeName || '';
// Convert length unit of width/height on blocks to
// a more editor-friendly way (px).
if (tagName in blockLike && element.getAttribute('style')) {
setStyle(element, stylesFilter(
[
[ ( /^(:?width|height)$/ ), null, convertToPxStr ]
])(element.getAttribute('style')));
}
// Processing headings.
if (tagName.match(/h\d/)) {
element.filterChildren();
// Is the heading actually a list item?
if (resolveListItem(element)) {
return;
}
}
// Remove inline elements which contain only empty spaces.
else if (tagName in dtd.$inline) {
element.filterChildren();
if (containsNothingButSpaces(element)) {
element.setTagName(null);
}
}
// Remove element with ms-office namespace,
// with it's content preserved, e.g. 'o:p'.
else if (tagName.indexOf(':') !== -1 && tagName.indexOf('ke') === -1) {
element.filterChildren();
// Restore image real link from vml.
if (tagName === 'v:imagedata') {
var href = element.getAttribute('o:href');
if (href) {
element.setAttribute('src', href);
}
element.setTagName('img');
return;
}
element.setTagName(null);
}
// Assembling list items into a whole list.
if (tagName in listDtdParents) {
element.filterChildren();
assembleList(element);
}
},
// We'll drop any style sheet, but Firefox conclude
// certain styles in a single style element, which are
// required to be changed into inline ones.
'style': function (element) {
if (UA.gecko) {
// Grab only the style definition section.
var styleDefSection = onlyChild(element).nodeValue
.match(/\/\* Style Definitions \*\/([\s\S]*?)\/\*/),
styleDefText = styleDefSection && styleDefSection[ 1 ],
rules = {}; // Storing the parsed result.
if (styleDefText) {
styleDefText
// Remove line-breaks.
.replace(/[\n\r]/g, '')
// Extract selectors and style properties.
.replace(/(.+?)\{(.+?)\}/g,
function (rule, selectors, styleBlock) {
selectors = selectors.split(',');
var length = selectors.length;
for (var i = 0; i < length; i++) {
// Assume MS-Word mostly generate only simple
// selector( [Type selector][Class selector]).
util.trim(selectors[ i ])
.replace(/^(\w+)(\.[\w-]+)?$/g,
function (match, tagName, className) {
tagName = tagName || '*';
className = className.substring(1, className.length);
// Reject MS-Word Normal styles.
if (className.match(/MsoNormal/)) {
return;
}
if (!rules[ tagName ]) {
rules[ tagName ] = {};
}
if (className) {
rules[ tagName ][ className ] = styleBlock;
} else {
rules[ tagName ] = styleBlock;
}
});
}
});
filters.applyStyleFilter = function (element) {
var name = rules[ '*' ] ? '*' : element.nodeName,
className = element.getAttribute('class'),
style;
if (name in rules) {
style = rules[ name ];
if (typeof style === 'object') {
style = style[ className ];
}
// Maintain style rules priorities.
if (style) {
addStyle(element, style, true);
}
}
};
}
}
return false;
},
'p': function (element) {
// This's a fall-back approach to recognize list item in FF3.6,
// as it's not perfect as not all list style (e.g. 'heading list') is shipped
// with this pattern. (#6662)
if (/MsoListParagraph/.exec(element.getAttribute('class'))) {
var bulletText = firstChild(element, function (node) {
return node.nodeType === 3 && !containsNothingButSpaces(node.parentNode);
});
var bullet = bulletText && bulletText.parentNode;
if (bullet && !bullet.getAttribute('style')) {
bullet.setAttribute('style', 'mso-list: Ignore;');
}
}
element.filterChildren();
// Is the paragraph actually a list item?
resolveListItem(element);
},
'div': function (element) {
// Aligned table with no text surrounded is represented by a wrapper div, from which
// table cells inherit as text-align styles, which is wrong.
// Instead we use a clear-float div after the table to properly achieve the same layout.
var singleChild = onlyChild(element);
if (singleChild && singleChild.nodeName === 'table') {
var attrs = element.attributes;
util.each(attrs, function (attr) {
singleChild.setAttribute(attr.name, attr.value);
});
if (element.getAttribute('style')) {
addStyle(singleChild, element.getAttribute('style'));
}
var clearFloatDiv = new HtmlParser.Tag('div');
addStyle(clearFloatDiv, 'clear', 'both');
element.appendChild(clearFloatDiv);
element.setTagName(null);
}
},
'td': function (element) {
// 'td' in 'thead' is actually <th>.
if (getAncestor(element, 'thead')) {
element.setTagName('th');
}
},
// MS-Word sometimes present list as a mixing of normal list
// and pseudo-list, normalize the previous ones into pseudo form.
'ol': flattenList,
'ul': flattenList,
'dl': flattenList,
'font': function (element) {
// Drop the font tag if it comes from list bullet text.
if (isListBulletIndicator(element.parentNode)) {
element.setTagName(null);
return;
}
element.filterChildren();
var styleText = element.getAttribute('style'),
parent = element.parentNode;
if ('font' === parent.name) // Merge nested <font> tags.
{
util.each(element.attributes, function (attr) {
parent.setAttribute(attr.name, attr.value);
});
if (styleText) {
addStyle(parent, styleText);
}
element.setTagName(null);
}
// Convert the merged into a span with all attributes preserved.
else {
styleText = styleText || '';
// IE's having those deprecated attributes, normalize them.
if (element.getAttribute('color')) {
if (element.getAttribute('color') !== '#000000') {
styleText += 'color:' + element.getAttribute('color') + ';';
}
element.removeAttribute('color');
}
if (element.getAttribute('face')) {
styleText += 'font-family:' + element.getAttribute('face') + ';';
element.removeAttribute('face');
}
var size = element.getAttribute('size');
// TODO: Mapping size in ranges of xx-small,
// x-small, small, medium, large, x-large, xx-large.
if (size) {
styleText += 'font-size:' +
(size > 3 ? 'large'
: ( size < 3 ? 'small' : 'medium' ) ) + ';';
element.removeAttribute('size');
}
element.setTagName('span');
addStyle(element, styleText);
}
},
'span': function (element) {
// Remove the span if it comes from list bullet text.
if (isListBulletIndicator(element.parentNode)) {
return false;
}
element.filterChildren();
if (containsNothingButSpaces(element)) {
element.setTagName(null);
return null;
}
// List item bullet type is supposed to be indicated by
// the text of a span with style 'mso-list : Ignore' or an image.
if (isListBulletIndicator(element)) {
var listSymbolNode = firstChild(element, function (node) {
return node.nodeValue || node.nodeName === 'img';
});
var listSymbol = listSymbolNode && ( listSymbolNode.nodeValue || 'l.' ),
listType = listSymbol && listSymbol.match(/^(?:[(]?)([^\s]+?)([.)]?)$/);
if (listType) {
var marker = createListBulletMarker(listType, listSymbol);
// Some non-existed list items might be carried by an inconsequential list,
// indicate by 'mso-hide:all/display:none',
// those are to be removed later, now mark it with 'ke:ignored'.
var ancestor = getAncestor(element, 'span');
if (ancestor && (/ mso-hide:\s*all|display:\s*none /).
test(ancestor.getAttribute('style'))) {
marker.setAttribute('ke:ignored', 1);
}
return marker;
}
}
// Update the src attribute of image element with href.
var styleText = element.getAttribute('style');
// Assume MS-Word mostly carry font related styles on <span>,
// adapting them to editor's convention.
if (styleText) {
setStyle(element, stylesFilter(
[
// Drop 'inline-height' style which make lines overlapping.
[ /^line-height$/ ],
[ /^font-family$/ ] ,
[ /^font-size$/ ] ,
[ /^color$/ ] ,
[ /^background-color$/ ]
]
)(styleText, element));
}
},
// Editor doesn't support anchor with content currently (#3582),
// drop such anchors with content preserved.
'a': function (element) {
var href;
if (!(href = element.getAttribute('href')) && element.getAttribute('name')) {
element.setTagName(null);
} else if (UA.webkit && href && href.match(/file:\/\/\/[\S]+#/i)) {
element.setAttribute('href', href.replace(/file:\/\/\/[^#]+/i, ''));
}
},
'ke:listbullet': function (element) {
if (getAncestor(element, /h\d/)) {
element.setTagName(null);
}
}
},
attributeNames: [
// Remove onmouseover and onmouseout events (from MS Word comments effect)
[ ( /^onmouse(:?out|over)/ ), '' ],
// Onload on image element.
[ ( /^onload$/ ), '' ],
// Remove office and vml attribute from elements.
[ ( /(?:v|o):\w+/ ), '' ],
// Remove lang/language attributes.
[ ( /^lang/ ), '' ]
],
attributes: {
// Provide a white-list of styles that we preserve, those should
// be the ones that could later be altered with editor tools.
'style': stylesFilter([
[ ( /^list-style-type$/ ) ],
// Preserve margin-left/right which used as default indent style in the editor.
[ ( /^margin$|^margin-(?!bottom|top)/ ), null, function (value, element, name) {
if (element.nodeName in { p: 1, div: 1 }) {
var indentStyleName = 'margin-left';
// Extract component value from 'margin' shorthand.
if (name === 'margin') {
value = getStyleComponents(name, value,
[ indentStyleName ])[ indentStyleName ];
} else if (name !== indentStyleName) {
return null;
}
if (value && !emptyMarginRegex.test(value)) {
return [ indentStyleName, value ];
}
}
return null;
} ],
// Preserve clear float style.
[ ( /^clear$/ ) ],
[ ( /^border.*|margin.*|vertical-align|float$/ ), null,
function (value, element) {
if (element.nodeName === 'img') {
return value;
}
} ],
[ (/^width|height$/ ), null,
function (value, element) {
if (element.nodeName in { table: 1, td: 1, th: 1, img: 1 }) {
return value;
}
} ]
],
1),
// Prefer width styles over 'width' attributes.
'width': function (value, element) {
if (element.nodeName in dtd.$tableContent) {
return false;
}
},
// Prefer border styles over table 'border' attributes.
'border': function (value, element) {
if (element.nodeName in dtd.$tableContent) {
return false;
}
},
// Only Firefox carry style sheet from MS-Word, which
// will be applied by us manually. For other browsers
// the css className is useless.
'class': falsyFilter,
// MS-Word always generate 'background-color' along with 'bgcolor',
// simply drop the deprecated attributes.
'bgcolor': falsyFilter,
// Deprecate 'valign' attribute in favor of 'vertical-align'.
'valign': function (value, element) {
addStyle(element, 'vertical-align', value);
return false;
}
},
// Fore none-IE, some useful data might be buried under these IE-conditional
// comments where RegExp were the right approach to dig them out where usual approach
// is transform it into a fake element node which hold the desired data.
comment: UA.ie ?
function (value, node) {
var imageInfo = value.match(/<img.*?>/),
listInfo = value.match(/^\[if !supportLists\]([\s\S]*?)\[endif\]$/);
// Seek for list bullet indicator.
if (listInfo) {
// Bullet symbol could be either text or an image.
var listSymbol = listInfo[ 1 ] || ( imageInfo && 'l.' ),
listType = listSymbol && listSymbol.match(/>(?:[(]?)([^\s]+?)([.)]?)</);
return createListBulletMarker(listType, listSymbol);
}
// Reveal the <img> element in conditional comments for Firefox.
if (UA.gecko && imageInfo) {
var img = new HtmlParser.Parser(imageInfo[0]).parse().childNodes[ 0 ],
previousComment = node.previousSibling,
// Try to dig the real image link from vml markup from previous comment text.
imgSrcInfo = previousComment && previousComment.toHtml().match(/<v:imagedata[^>]*o:href=[''](.*?)['']/),
imgSrc = imgSrcInfo && imgSrcInfo[ 1 ];
// Is there a real 'src' url to be used?
if (imgSrc) {
( img.setAttribute('src', imgSrc) );
}
return img;
}
return false;
}
: falsyFilter
});
})();
define('kg/editor-plugins/1.1.3/word-filter',["html-parser","util","node","ua"],function(require, exports, module) {
/**
* @ignore
* clean html pasted from word. modified from ckeditor.
* @author yiminghe@gmail.com
*/
/*jshint loopfunc:true*/
var HtmlParser = require('html-parser');
var util = require('util');
var $ = require('node'),
UA = require('ua'),
dtd = HtmlParser.DTD,
wordFilter = new HtmlParser.Filter(),
cssLengthRelativeUnit = /^([.\d]*)+(em|ex|px|gd|rem|vw|vh|vm|ch|mm|cm|in|pt|pc|deg|rad|ms|s|hz|khz){1}?/i,
// e.g. 0px 0pt 0px
emptyMarginRegex = /^(?:\b0[^\s]*\s*){1,4}$/,
romanLiteralPattern = '^m{0,4}(cm|cd|d?c{0,3})(xc|xl|l?x{0,3})(ix|iv|v?i{0,3})$',
lowerRomanLiteralRegex = new RegExp(romanLiteralPattern),
upperRomanLiteralRegex = new RegExp(romanLiteralPattern.toUpperCase()),
orderedPatterns = {
'decimal': /\d+/,
'lower-roman': lowerRomanLiteralRegex,
'upper-roman': upperRomanLiteralRegex,
'lower-alpha': /^[a-z]+$/,
'upper-alpha': /^[A-Z]+$/
},
unorderedPatterns = {
'disc': /[l\u00B7\u2002]/,
'circle': /[\u006F\u00D8]/,
'square': /[\u006E\u25C6]/
},
listMarkerPatterns = {
'ol': orderedPatterns,
'ul': unorderedPatterns
},
romans = [
[1000, 'M'],
[900, 'CM'],
[500, 'D'],
[400, 'CD'],
[100, 'C'],
[90, 'XC'],
[50, 'L'],
[40, 'XL'],
[10, 'X'],
[9, 'IX'],
[5, 'V'],
[4, 'IV'],
[1, 'I']
],
alphabets = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ';
// Convert roman numbering back to decimal.
function fromRoman(str) {
str = str.toUpperCase();
var l = romans.length, retVal = 0;
for (var i = 0; i < l; ++i) {
for (var j = romans[i], k = j[1].length; str.substr(0, k) === j[1]; str = str.substr(k)) {
retVal += j[ 0 ];
}
}
return retVal;
}
// Convert alphabet numbering back to decimal.
function fromAlphabet(str) {
str = str.toUpperCase();
var l = alphabets.length, retVal = 1;
for (var x = 1; str.length > 0; x *= l) {
retVal += alphabets.indexOf(str.charAt(str.length - 1)) * x;
str = str.substr(0, str.length - 1);
}
return retVal;
}
function setStyle(element, str) {
if (str) {
element.setAttribute('style', str);
} else {
element.removeAttribute('style');
}
}
/*
Convert the specified CSS length value to the calculated pixel length inside this page.
<strong>Note:</strong> Percentage based value is left intact.
*/
var convertToPx = (function () {
var calculator;
return function (cssLength) {
if (!calculator) {
calculator = $('<div style="position:absolute;left:-9999px;' +
'top:-9999px;margin:0px;padding:0px;border:0px;"' + '></div>')
.prependTo('body');
}
if (!(/%$/).test(cssLength)) {
calculator.css('width', cssLength);
return calculator[0].clientWidth;
}
return cssLength;
};
})();
var listBaseIndent = 0,
previousListItemMargin = null,
previousListId;
function onlyChild(elem) {
var childNodes = elem.childNodes || [],
count = childNodes.length,
firstChild = (count === 1) && childNodes[0];
return firstChild || null;
}
function removeAnyChildWithName(elem, tagName) {
var children = elem.childNodes || [],
ret = [],
child;
for (var i = 0; i < children.length; i++) {
child = children[ i ];
if (!child.nodeName) {
continue;
}
if (child.nodeName === tagName) {
ret.push(child);
children.splice(i--, 1);
}
ret = ret.concat(removeAnyChildWithName(child, tagName));
}
return ret;
}
function getAncestor(elem, tagNameRegex) {
var parent = elem.parentNode;
while (parent && !( parent.nodeName && parent.nodeName.match(tagNameRegex) )) {
parent = parent.parentNode;
}
return parent;
}
function firstChild(elem, evaluator) {
var child,
i,
children = elem.childNodes || [];
for (i = 0; i < children.length; i++) {
child = children[ i ];
if (evaluator(child)) {
return child;
} else if (child.nodeName) {
child = firstChild(child, evaluator);
if (child) {
return child;
}
}
}
return null;
}
function addStyle(elem, name, value, isPrepend) {
var styleText, addingStyleText = '', style;
// name/value pair.
if (typeof value === 'string') {
addingStyleText += name + ':' + value + ';';
} else {
// style literal.
if (typeof name === 'object') {
for (style in name) {
addingStyleText += style + ':' + name[ style ] + ';';
}
}
// raw style text form.
else {
addingStyleText += name;
}
isPrepend = value;
}
styleText = elem.getAttribute('style');
styleText = ( isPrepend ?
[ addingStyleText, styleText ]
: [ styleText, addingStyleText ] ).join(';');
setStyle(elem, styleText.replace(/^;|;(?=;)/, ''));
}
function parentOf(tagName) {
var result = {},
tag;
for (tag in dtd) {
if (tag.indexOf('$') === -1 && dtd[ tag ][ tagName ]) {
result[ tag ] = 1;
}
}
return result;
}
var filters = {
// Transform a normal list into flat list items only presentation.
// E.g. <ul><li>level1<ol><li>level2</li></ol></li> =>
// <ke:li ke:listtype='ul' ke:indent='1'>level1</ke:li>
// <ke:li ke:listtype='ol' ke:indent='2'>level2</ke:li>
flattenList: function (element, level) {
level = typeof level === 'number' ? level : 1;
var listStyleType;
// All list items are of the same type.
switch (element.getAttribute('type')) {
case 'a' :
listStyleType = 'lower-alpha';
break;
case '1' :
listStyleType = 'decimal';
break;
// TODO: Support more list style type from MS-Word.
}
var children = element.childNodes || [],
child;
for (var i = 0; i < children.length; i++) {
child = children[ i ];
if (child.nodeName in dtd.$listItem) {
var listItemChildren = child.childNodes || [],
count = listItemChildren.length,
last = listItemChildren[ count - 1 ];
// Move out nested list.
if (last.nodeName in dtd.$list) {
element.insertAfter(child);
// Remove the parent list item if it's just a holder.
if (!--listItemChildren.length) {
element.removeChild(children[i--]);
}
}
child.setTagName('ke:li');
// Inherit numbering from list root on the first list item.
if (element.getAttribute('start') && !i) {
element.setAttribute('value', element.getAttribute('start'));
}
filters.stylesFilter(
[
['tab-stops', null, function (val) {
var margin = val.split(' ')[ 1 ].match(cssLengthRelativeUnit);
if (margin) {
( previousListItemMargin = convertToPx(margin[ 0 ]) );
}
}
],
( level === 1 ? [ 'mso-list', null, function (val) {
val = val.split(' ');
var listId = Number(val[ 0 ].match(/\d+/));
if (listId !== previousListId) {
child.setAttribute('ke:reset', 1);
}
previousListId = listId;
} ] : null )
])(child.getAttribute('style'));
child.setAttribute('ke:indent', level);
child.setAttribute('ke:listtype', element.nodeName);
child.setAttribute('ke:list-style-type', listStyleType);
}
// Flatten sub list.
else if (child.nodeName in dtd.$list) {
/*jshint noarg:false*/
// Absorb sub list children.
arguments.callee.apply(this, [ child, level + 1 ]);
children = children.slice(0, i).concat(child.childNodes).concat(children.slice(i + 1));
element.empty();
for (var j = 0, num = children.length; j < num; j++) {
element.appendChild(children[j]);
}
}
}
element.nodeName = element.tagName = null;
// We're loosing tag name here, signalize this element as a list.
element.setAttribute('ke:list', 1);
},
/*
Try to collect all list items among the children and establish one
or more HTML list structures for them.
*/
assembleList: function (element) {
var children = element.childNodes || [],
child,
listItem, // The current processing ke:li element.
listItemIndent, // Indent level of current list item.
lastIndent,
lastListItem, // The previous one just been added to the list.
list, // Current staging list and it's parent list if any.
openedLists = [],
previousListStyleType,
previousListType;
// Properties of the list item are to be resolved from the list bullet.
var bullet,
listType,
listStyleType,
itemNumeric;
for (var i = 0; i < children.length; i++) {
child = children[ i ];
if ('ke:li' === child.nodeName) {
child.setTagName('li');
listItem = child;
bullet = listItem.getAttribute('ke:listsymbol');
bullet = bullet && bullet.match(/^(?:[(]?)([^\s]+?)([.)]?)$/);
listType = listStyleType = itemNumeric = null;
if (listItem.getAttribute('ke:ignored')) {
children.splice(i--, 1);
continue;
}
// This's from a new list root.
if (listItem.getAttribute('ke:reset')) {
( list = lastIndent = lastListItem = null );
}
// List item indent level might come from a real list indentation or
// been resolved from a pseudo list item's margin value, even get
// no indentation at all.
listItemIndent = Number(listItem.getAttribute('ke:indent'));
// We're moving out of the current list, cleaning up.
if (listItemIndent !== lastIndent) {
previousListType = previousListStyleType = null;
}
// List type and item style are already resolved.
if (!bullet) {
listType = listItem.getAttribute('ke:listtype') || 'ol';
listStyleType = listItem.getAttribute('ke:list-style-type');
}
else {
// Probably share the same list style type with previous list item,
// give it priority to avoid ambiguous between C(Alpha) and C.(Roman).
if (previousListType &&
listMarkerPatterns[ previousListType ] [ previousListStyleType ].test(bullet[ 1 ])) {
listType = previousListType;
listStyleType = previousListStyleType;
}
else {
for (var type in listMarkerPatterns) {
for (var style in listMarkerPatterns[ type ]) {
if (listMarkerPatterns[ type ][ style ].test(bullet[ 1 ])) {
// Small numbering has higher priority, when dealing with ambiguous
// between C(Alpha) and C.(Roman).
if (type === 'ol' && ( /alpha|roman/ ).test(style)) {
var num = /roman/.test(style) ? fromRoman(bullet[ 1 ]) : fromAlphabet(bullet[ 1 ]);
if (!itemNumeric || num < itemNumeric) {
itemNumeric = num;
listType = type;
listStyleType = style;
}
}
else {
listType = type;
listStyleType = style;
break;
}
}
}
}
}
// Simply use decimal/disc for the rest forms of unrepresentable
// numerals, e.g. Chinese..., but as long as there a second part
// included, it has a bigger chance of being a order list ;)
if (!listType) {
( listType = bullet[ 2 ] ? 'ol' : 'ul' );
}
}
previousListType = listType;
previousListStyleType = listStyleType || ( listType === 'ol' ? 'decimal' : 'disc' );
if (listStyleType && listStyleType !== ( listType === 'ol' ? 'decimal' : 'disc' )) {
addStyle(listItem, 'list-style-type', listStyleType);
}
// Figure out start numbering.
if (listType === 'ol' && bullet) {
switch (listStyleType) {
case 'decimal' :
itemNumeric = Number(bullet[ 1 ]);
break;
case 'lower-roman':
case 'upper-roman':
itemNumeric = fromRoman(bullet[ 1 ]);
break;
case 'lower-alpha':
case 'upper-alpha':
itemNumeric = fromAlphabet(bullet[ 1 ]);
break;
}
// Always create the numbering, swipe out unnecessary ones later.
listItem.setAttribute('value', itemNumeric);
}
// Start the list construction.
if (!list) {
openedLists.push(list = new HtmlParser.Tag(listType));
list.appendChild(listItem);
element.replaceChild(list, children[i]);
} else {
if (listItemIndent > lastIndent) {
openedLists.push(list = new HtmlParser.Tag(listType));
list.appendChild(listItem);
lastListItem.appendChild(list);
}
else if (listItemIndent < lastIndent) {
// There might be a negative gap between two list levels. (#4944)
var diff = lastIndent - listItemIndent,
parent;
while (diff-- && ( parent = list.parentNode )) {
list = parent.parentNode;
}
list.appendChild(listItem);
}
else {
list.appendChild(listItem);
}
children.splice(i--, 1);
}
lastListItem = listItem;
lastIndent = listItemIndent;
}
else if (list && !(child.nodeType === 3 && !util.trim(child.nodeValue))) {
list = lastIndent = lastListItem = null;
}
}
for (i = 0; i < openedLists.length; i++) {
postProcessList(openedLists[ i ]);
}
},
/*
A simple filter which always rejecting.
*/
falsyFilter: function () {
return false;
},
/*
A filter dedicated on the 'style' attribute filtering, e.g. dropping/replacing style properties.
@param styles {Array} in form of [ styleNameRegexp, styleValueRegexp,
newStyleValue/newStyleGenerator, newStyleName ] where only the first
parameter is mandatory.
@param [whitelist] {Boolean} Whether the {@param styles} will be considered as a white-list.
*/
stylesFilter: function (styles, whitelist) {
return function (styleText, element) {
var rules = [];
// html-encoded quote might be introduced by 'font-family'
// from MS-Word which confused the following regexp. e.g.
//'font-family: "Lucida, Console"'
( styleText || '' )
.replace(/"/g, '"')
.replace(/\s*([^ :;]+)\s*:\s*([^;]+)\s*(?=;|$)/g,
function (match, name, value) {
name = name.toLowerCase();
if (name === 'font-family') {
( value = value.replace(/['']/g, '') );
}
var namePattern,
valuePattern,
newValue,
newName;
for (var i = 0; i < styles.length; i++) {
if (styles[ i ]) {
namePattern = styles[ i ][ 0 ];
valuePattern = styles[ i ][ 1 ];
newValue = styles[ i ][ 2 ];
newName = styles[ i ][ 3 ];
if (name.match(namePattern) && ( !valuePattern || value.match(valuePattern) )) {
name = newName || name;
if (whitelist) {
( newValue = newValue || value );
}
if (typeof newValue === 'function') {
newValue = newValue(value, element, name);
}
// Return an couple indicate both name and value
// changed.
if (newValue && newValue.push) {
name = newValue[ 0 ];
newValue = newValue[ 1 ];
}
if (typeof newValue === 'string') {
rules.push([ name, newValue ]);
}
return;
}
}
}
if (!whitelist) {
rules.push([ name, value ]);
}
});
for (var i = 0; i < rules.length; i++) {
rules[ i ] = rules[ i ].join(':');
}
return rules.length ? ( rules.join(';') + ';' ) : false;
};
},
/*
A filter which will be used to apply inline css style according the stylesheet
definition rules, is generated lazily when filtering.
*/
applyStyleFilter: null
};
// 1. move consistent list item styles up to list root.
// 2. clear out unnecessary list item numbering.
function postProcessList(list) {
var children = list.childNodes || [],
child,
count = children.length,
match,
mergeStyle,
styleTypeRegexp = /list-style-type:(.*?)(?:;|$)/,
stylesFilter = filters.stylesFilter;
if (styleTypeRegexp.exec(list.getAttribute('style'))) {
return;
}
for (var i = 0; i < count; i++) {
child = children[ i ];
if (child.getAttribute('value') && Number(child.getAttribute('value')) === i + 1) {
child.removeAttribute('value');
}
match = styleTypeRegexp.exec(child.getAttribute('style'));
if (match) {
if (match[ 1 ] === mergeStyle || !mergeStyle) {
mergeStyle = match[ 1 ];
}
else {
mergeStyle = null;
break;
}
}
}
if (mergeStyle) {
for (i = 0; i < count; i++) {
var style = children[ i ].getAttribute('style');
if (style) {
style = stylesFilter([
[ 'list-style-type']
])(style);
setStyle(children[ i ], style);
}
}
addStyle(list, 'list-style-type', mergeStyle);
}
}
var utils = {
// Create a <ke:listbullet> which indicate an list item type.
createListBulletMarker: function (bullet, bulletText) {
var marker = new HtmlParser.Tag('ke:listbullet');
marker.setAttribute('ke:listsymbol', bullet[ 0 ]);
marker.appendChild(new HtmlParser.Text(bulletText));
return marker;
},
isListBulletIndicator: function (element) {
var styleText = element.getAttribute('style');
if (/mso-list\s*:\s*Ignore/i.test(styleText)) {
return true;
}
},
isContainingOnlySpaces: function (element) {
var text;
return ( ( text = onlyChild(element) ) && ( /^(:?\s| )+$/ ).test(text.nodeValue) );
},
resolveList: function (element) {
// <ke:listbullet> indicate a list item.
var listMarker;
if (( listMarker = removeAnyChildWithName(element, 'ke:listbullet') ) &&
listMarker.length && ( listMarker = listMarker[ 0 ] )) {
element.setTagName('ke:li');
if (element.getAttribute('style')) {
var styleStr = filters.stylesFilter(
[
// Text-indent is not representing list item level any more.
[ 'text-indent' ],
[ 'line-height' ],
// First attempt is to resolve indent level from on a constant margin increment.
[ ( /^margin(:?-left)?$/ ), null, function (margin) {
// Deal with component/short-hand form.
var values = margin.split(' ');
margin = convertToPx(values[ 3 ] || values[ 1 ] || values [ 0 ]);
// Figure out the indent unit by checking the first time of incrementation.
if (!listBaseIndent && previousListItemMargin !== null &&
margin > previousListItemMargin) {
listBaseIndent = margin - previousListItemMargin;
}
previousListItemMargin = margin;
if (listBaseIndent) {
element.setAttribute('ke:indent', listBaseIndent &&
( Math.ceil(margin / listBaseIndent) + 1 ) || 1);
}
} ],
// The best situation: 'mso-list:l0 level1 lfo2' tells the belonged list root, list item indentation, etc.
[ ( /^mso-list$/ ), null, function (val) {
val = val.split(' ');
var listId = Number(val[ 0 ].match(/\d+/)),
indent = Number(val[ 1 ].match(/\d+/));
if (indent === 1) {
if (listId !== previousListId) {
( element.setAttribute('ke:reset', 1) );
}
previousListId = listId;
}
element.setAttribute('ke:indent', indent);
} ]
])(element.getAttribute('style'), element);
setStyle(element, styleStr);
}
// First level list item might be presented without a margin.
// In case all above doesn't apply.
if (!element.getAttribute('ke:indent')) {
previousListItemMargin = 0;
element.setAttribute('ke:indent', 1);
}
util.each(listMarker.attributes, function (a) {
element.setAttribute(a.name, a.value);
});
return true;
}
// Current list disconnected.
else {
previousListId = previousListItemMargin = listBaseIndent = null;
}
return false;
},
// Providing a shorthand style then retrieve one or more style component values.
getStyleComponents: (function () {
var calculator = $('<div style="position:absolute;left:-9999px;top:-9999px;"></div>').prependTo('body');
return function (name, styleValue, fetchList) {
calculator.css(name, styleValue);
var styles = {},
count = fetchList.length;
for (var i = 0; i < count; i++) {
styles[ fetchList[ i ] ] = calculator.css(fetchList[ i ]);
}
return styles;
};
})(),
listDtdParents: parentOf('ol')
};
(function () {
var blockLike = util.merge(dtd.$block, dtd.$listItem, dtd.$tableContent),
falsyFilter = filters.falsyFilter,
stylesFilter = filters.stylesFilter,
createListBulletMarker = utils.createListBulletMarker,
flattenList = filters.flattenList,
assembleList = filters.assembleList,
isListBulletIndicator = utils.isListBulletIndicator,
containsNothingButSpaces = utils.isContainingOnlySpaces,
resolveListItem = utils.resolveList,
convertToPxStr = function (value) {
value = convertToPx(value);
return isNaN(value) ? value : value + 'px';
},
getStyleComponents = utils.getStyleComponents,
listDtdParents = utils.listDtdParents;
wordFilter.addRules({
tagNames: [
// Remove script, meta and link elements.
[ ( /meta|link|script/ ), '' ]
],
root: function (element) {
element.filterChildren();
assembleList(element);
},
tags: {
'^': function (element) {
// Transform CSS style declaration to inline style.
var applyStyleFilter;
if (UA.gecko && ( applyStyleFilter = filters.applyStyleFilter )) {
applyStyleFilter(element);
}
},
$: function (element) {
var tagName = element.nodeName || '';
// Convert length unit of width/height on blocks to
// a more editor-friendly way (px).
if (tagName in blockLike && element.getAttribute('style')) {
setStyle(element, stylesFilter(
[
[ ( /^(:?width|height)$/ ), null, convertToPxStr ]
])(element.getAttribute('style')));
}
// Processing headings.
if (tagName.match(/h\d/)) {
element.filterChildren();
// Is the heading actually a list item?
if (resolveListItem(element)) {
return;
}
}
// Remove inline elements which contain only empty spaces.
else if (tagName in dtd.$inline) {
element.filterChildren();
if (containsNothingButSpaces(element)) {
element.setTagName(null);
}
}
// Remove element with ms-office namespace,
// with it's content preserved, e.g. 'o:p'.
else if (tagName.indexOf(':') !== -1 && tagName.indexOf('ke') === -1) {
element.filterChildren();
// Restore image real link from vml.
if (tagName === 'v:imagedata') {
var href = element.getAttribute('o:href');
if (href) {
element.setAttribute('src', href);
}
element.setTagName('img');
return;
}
element.setTagName(null);
}
// Assembling list items into a whole list.
if (tagName in listDtdParents) {
element.filterChildren();
assembleList(element);
}
},
// We'll drop any style sheet, but Firefox conclude
// certain styles in a single style element, which are
// required to be changed into inline ones.
'style': function (element) {
if (UA.gecko) {
// Grab only the style definition section.
var styleDefSection = onlyChild(element).nodeValue
.match(/\/\* Style Definitions \*\/([\s\S]*?)\/\*/),
styleDefText = styleDefSection && styleDefSection[ 1 ],
rules = {}; // Storing the parsed result.
if (styleDefText) {
styleDefText
// Remove line-breaks.
.replace(/[\n\r]/g, '')
// Extract selectors and style properties.
.replace(/(.+?)\{(.+?)\}/g,
function (rule, selectors, styleBlock) {
selectors = selectors.split(',');
var length = selectors.length;
for (var i = 0; i < length; i++) {
// Assume MS-Word mostly generate only simple
// selector( [Type selector][Class selector]).
util.trim(selectors[ i ])
.replace(/^(\w+)(\.[\w-]+)?$/g,
function (match, tagName, className) {
tagName = tagName || '*';
className = className.substring(1, className.length);
// Reject MS-Word Normal styles.
if (className.match(/MsoNormal/)) {
return;
}
if (!rules[ tagName ]) {
rules[ tagName ] = {};
}
if (className) {
rules[ tagName ][ className ] = styleBlock;
} else {
rules[ tagName ] = styleBlock;
}
});
}
});
filters.applyStyleFilter = function (element) {
var name = rules[ '*' ] ? '*' : element.nodeName,
className = element.getAttribute('class'),
style;
if (name in rules) {
style = rules[ name ];
if (typeof style === 'object') {
style = style[ className ];
}
// Maintain style rules priorities.
if (style) {
addStyle(element, style, true);
}
}
};
}
}
return false;
},
'p': function (element) {
// This's a fall-back approach to recognize list item in FF3.6,
// as it's not perfect as not all list style (e.g. 'heading list') is shipped
// with this pattern. (#6662)
if (/MsoListParagraph/.exec(element.getAttribute('class'))) {
var bulletText = firstChild(element, function (node) {
return node.nodeType === 3 && !containsNothingButSpaces(node.parentNode);
});
var bullet = bulletText && bulletText.parentNode;
if (bullet && !bullet.getAttribute('style')) {
bullet.setAttribute('style', 'mso-list: Ignore;');
}
}
element.filterChildren();
// Is the paragraph actually a list item?
resolveListItem(element);
},
'div': function (element) {
// Aligned table with no text surrounded is represented by a wrapper div, from which
// table cells inherit as text-align styles, which is wrong.
// Instead we use a clear-float div after the table to properly achieve the same layout.
var singleChild = onlyChild(element);
if (singleChild && singleChild.nodeName === 'table') {
var attrs = element.attributes;
util.each(attrs, function (attr) {
singleChild.setAttribute(attr.name, attr.value);
});
if (element.getAttribute('style')) {
addStyle(singleChild, element.getAttribute('style'));
}
var clearFloatDiv = new HtmlParser.Tag('div');
addStyle(clearFloatDiv, 'clear', 'both');
element.appendChild(clearFloatDiv);
element.setTagName(null);
}
},
'td': function (element) {
// 'td' in 'thead' is actually <th>.
if (getAncestor(element, 'thead')) {
element.setTagName('th');
}
},
// MS-Word sometimes present list as a mixing of normal list
// and pseudo-list, normalize the previous ones into pseudo form.
'ol': flattenList,
'ul': flattenList,
'dl': flattenList,
'font': function (element) {
// Drop the font tag if it comes from list bullet text.
if (isListBulletIndicator(element.parentNode)) {
element.setTagName(null);
return;
}
element.filterChildren();
var styleText = element.getAttribute('style'),
parent = element.parentNode;
if ('font' === parent.name) // Merge nested <font> tags.
{
util.each(element.attributes, function (attr) {
parent.setAttribute(attr.name, attr.value);
});
if (styleText) {
addStyle(parent, styleText);
}
element.setTagName(null);
}
// Convert the merged into a span with all attributes preserved.
else {
styleText = styleText || '';
// IE's having those deprecated attributes, normalize them.
if (element.getAttribute('color')) {
if (element.getAttribute('color') !== '#000000') {
styleText += 'color:' + element.getAttribute('color') + ';';
}
element.removeAttribute('color');
}
if (element.getAttribute('face')) {
styleText += 'font-family:' + element.getAttribute('face') + ';';
element.removeAttribute('face');
}
var size = element.getAttribute('size');
// TODO: Mapping size in ranges of xx-small,
// x-small, small, medium, large, x-large, xx-large.
if (size) {
styleText += 'font-size:' +
(size > 3 ? 'large'
: ( size < 3 ? 'small' : 'medium' ) ) + ';';
element.removeAttribute('size');
}
element.setTagName('span');
addStyle(element, styleText);
}
},
'span': function (element) {
// Remove the span if it comes from list bullet text.
if (isListBulletIndicator(element.parentNode)) {
return false;
}
element.filterChildren();
if (containsNothingButSpaces(element)) {
element.setTagName(null);
return null;
}
// List item bullet type is supposed to be indicated by
// the text of a span with style 'mso-list : Ignore' or an image.
if (isListBulletIndicator(element)) {
var listSymbolNode = firstChild(element, function (node) {
return node.nodeValue || node.nodeName === 'img';
});
var listSymbol = listSymbolNode && ( listSymbolNode.nodeValue || 'l.' ),
listType = listSymbol && listSymbol.match(/^(?:[(]?)([^\s]+?)([.)]?)$/);
if (listType) {
var marker = createListBulletMarker(listType, listSymbol);
// Some non-existed list items might be carried by an inconsequential list,
// indicate by 'mso-hide:all/display:none',
// those are to be removed later, now mark it with 'ke:ignored'.
var ancestor = getAncestor(element, 'span');
if (ancestor && (/ mso-hide:\s*all|display:\s*none /).
test(ancestor.getAttribute('style'))) {
marker.setAttribute('ke:ignored', 1);
}
return marker;
}
}
// Update the src attribute of image element with href.
var styleText = element.getAttribute('style');
// Assume MS-Word mostly carry font related styles on <span>,
// adapting them to editor's convention.
if (styleText) {
setStyle(element, stylesFilter(
[
// Drop 'inline-height' style which make lines overlapping.
[ /^line-height$/ ],
[ /^font-family$/ ] ,
[ /^font-size$/ ] ,
[ /^color$/ ] ,
[ /^background-color$/ ]
]
)(styleText, element));
}
},
// Editor doesn't support anchor with content currently (#3582),
// drop such anchors with content preserved.
'a': function (element) {
var href;
if (!(href = element.getAttribute('href')) && element.getAttribute('name')) {
element.setTagName(null);
} else if (UA.webkit && href && href.match(/file:\/\/\/[\S]+#/i)) {
element.setAttribute('href', href.replace(/file:\/\/\/[^#]+/i, ''));
}
},
'ke:listbullet': function (element) {
if (getAncestor(element, /h\d/)) {
element.setTagName(null);
}
}
},
attributeNames: [
// Remove onmouseover and onmouseout events (from MS Word comments effect)
[ ( /^onmouse(:?out|over)/ ), '' ],
// Onload on image element.
[ ( /^onload$/ ), '' ],
// Remove office and vml attribute from elements.
[ ( /(?:v|o):\w+/ ), '' ],
// Remove lang/language attributes.
[ ( /^lang/ ), '' ]
],
attributes: {
// Provide a white-list of styles that we preserve, those should
// be the ones that could later be altered with editor tools.
'style': stylesFilter([
[ ( /^list-style-type$/ ) ],
// Preserve margin-left/right which used as default indent style in the editor.
[ ( /^margin$|^margin-(?!bottom|top)/ ), null, function (value, element, name) {
if (element.nodeName in { p: 1, div: 1 }) {
var indentStyleName = 'margin-left';
// Extract component value from 'margin' shorthand.
if (name === 'margin') {
value = getStyleComponents(name, value,
[ indentStyleName ])[ indentStyleName ];
} else if (name !== indentStyleName) {
return null;
}
if (value && !emptyMarginRegex.test(value)) {
return [ indentStyleName, value ];
}
}
return null;
} ],
// Preserve clear float style.
[ ( /^clear$/ ) ],
[ ( /^border.*|margin.*|vertical-align|float$/ ), null,
function (value, element) {
if (element.nodeName === 'img') {
return value;
}
} ],
[ (/^width|height$/ ), null,
function (value, element) {
if (element.nodeName in { table: 1, td: 1, th: 1, img: 1 }) {
return value;
}
} ]
],
1),
// Prefer width styles over 'width' attributes.
'width': function (value, element) {
if (element.nodeName in dtd.$tableContent) {
return false;
}
},
// Prefer border styles over table 'border' attributes.
'border': function (value, element) {
if (element.nodeName in dtd.$tableContent) {
return false;
}
},
// Only Firefox carry style sheet from MS-Word, which
// will be applied by us manually. For other browsers
// the css className is useless.
'class': falsyFilter,
// MS-Word always generate 'background-color' along with 'bgcolor',
// simply drop the deprecated attributes.
'bgcolor': falsyFilter,
// Deprecate 'valign' attribute in favor of 'vertical-align'.
'valign': function (value, element) {
addStyle(element, 'vertical-align', value);
return false;
}
},
// Fore none-IE, some useful data might be buried under these IE-conditional
// comments where RegExp were the right approach to dig them out where usual approach
// is transform it into a fake element node which hold the desired data.
comment: UA.ie ?
function (value, node) {
var imageInfo = value.match(/<img.*?>/),
listInfo = value.match(/^\[if !supportLists\]([\s\S]*?)\[endif\]$/);
// Seek for list bullet indicator.
if (listInfo) {
// Bullet symbol could be either text or an image.
var listSymbol = listInfo[ 1 ] || ( imageInfo && 'l.' ),
listType = listSymbol && listSymbol.match(/>(?:[(]?)([^\s]+?)([.)]?)</);
return createListBulletMarker(listType, listSymbol);
}
// Reveal the <img> element in conditional comments for Firefox.
if (UA.gecko && imageInfo) {
var img = new HtmlParser.Parser(imageInfo[0]).parse().childNodes[ 0 ],
previousComment = node.previousSibling,
// Try to dig the real image link from vml markup from previous comment text.
imgSrcInfo = previousComment && previousComment.toHtml().match(/<v:imagedata[^>]*o:href=[''](.*?)['']/),
imgSrc = imgSrcInfo && imgSrcInfo[ 1 ];
// Is there a real 'src' url to be used?
if (imgSrc) {
( img.setAttribute('src', imgSrc) );
}
return img;
}
return false;
}
: falsyFilter
});
})();
module.exports = {
toDataFormat: function (html, editor) {
// Firefox will be confused by those downlevel-revealed IE conditional
// comments, fixing them first( convert it to upperlevel-revealed one ).
// e.g. <![if !vml]>...<![endif]>
//<!--[if !supportLists]-->
// <span style=\'font-family: Wingdings;\' lang=\'EN-US\'>
// <span style=\'\'>l<span style=\'font: 7pt "Times New Roman";\'>
// </span></span></span>
// <!--[endif]-->
//变成：
//<!--[if !supportLists]
// <span style=\'font-family: Wingdings;\' lang=\'EN-US\'>
// <span style=\'\'>l<span style=\'font: 7pt "Times New Roman";\'>
// </span></span></span>
// [endif]-->
if (UA.gecko) {
html = html.replace(/(<!--\[if[^<]*?\])-->([\S\s]*?)<!--(\[endif\]-->)/gi,
'$1$2$3');
}
// 针对 word 一次
html = editor.htmlDataProcessor.toDataFormat(html, wordFilter);
return html;
}
};
});
(function() {
var blockLike = S.merge(dtd.$block, dtd.$listItem, dtd.$tableContent), falsyFilter = filters.falsyFilter, stylesFilter = filters.stylesFilter, createListBulletMarker = utils.createListBulletMarker, flattenList = filters.flattenList, assembleList = filters.assembleList, isListBulletIndicator = utils.isListBulletIndicator, containsNothingButSpaces = utils.isContainingOnlySpaces, resolveListItem = utils.resolveList, convertToPxStr = function(value) {
value = convertToPx(value);
return isNaN(value) ? value : value + "px"
}, getStyleComponents = utils.getStyleComponents, listDtdParents = utils.listDtdParents;
wordFilter.addRules({tagNames:[[/meta|link|script/, ""]], root:function(element) {
element.filterChildren();
assembleList(element)
}, tags:{"^":function(element) {
var applyStyleFilter;
if(UA.gecko && (applyStyleFilter = filters.applyStyleFilter)) {
applyStyleFilter(element)
}
}, $:function(element) {
var tagName = element.nodeName || "";
if(tagName in blockLike && element.getAttribute("style")) {
setStyle(element, stylesFilter([[/^(:?width|height)$/, null, convertToPxStr]])(element.getAttribute("style")))
}
if(tagName.match(/h\d/)) {
element.filterChildren();
if(resolveListItem(element)) {
return
}
}else {
if(tagName in dtd.$inline) {
element.filterChildren();
if(containsNothingButSpaces(element)) {
element.setTagName(null)
}
}else {
if(tagName.indexOf(":") !== -1 && tagName.indexOf("ke") === -1) {
element.filterChildren();
if(tagName === "v:imagedata") {
var href = element.getAttribute("o:href");
if(href) {
element.setAttribute("src", href)
}
element.setTagName("img");
return
}
element.setTagName(null)
}
}
}
if(tagName in listDtdParents) {
element.filterChildren();
assembleList(element)
}
}, style:function(element) {
if(UA.gecko) {
var styleDefSection = onlyChild(element).nodeValue.match(/\/\* Style Definitions \*\/([\s\S]*?)\/\*/), styleDefText = styleDefSection && styleDefSection[1], rules = {};
if(styleDefText) {
styleDefText.replace(/[\n\r]/g, "").replace(/(.+?)\{(.+?)\}/g, function(rule, selectors, styleBlock) {
selectors = selectors.split(",");
var length = selectors.length;
for(var i = 0;i < length;i++) {
S.trim(selectors[i]).replace(/^(\w+)(\.[\w-]+)?$/g, function(match, tagName, className) {
tagName = tagName || "*";
className = className.substring(1, className.length);
if(className.match(/MsoNormal/)) {
return
}
if(!rules[tagName]) {
rules[tagName] = {}
}
if(className) {
rules[tagName][className] = styleBlock
}else {
rules[tagName] = styleBlock
}
})
}
});
filters.applyStyleFilter = function(element) {
var name = rules["*"] ? "*" : element.nodeName, className = element.getAttribute("class"), style;
if(name in rules) {
style = rules[name];
if(typeof style === "object") {
style = style[className]
}
if(style) {
addStyle(element, style, true)
}
}
}
}
}
return false
}, p:function(element) {
if(/MsoListParagraph/.exec(element.getAttribute("class"))) {
var bulletText = firstChild(element, function(node) {
return node.nodeType === 3 && !containsNothingButSpaces(node.parentNode)
});
var bullet = bulletText && bulletText.parentNode;
if(bullet && !bullet.getAttribute("style")) {
bullet.setAttribute("style", "mso-list: Ignore;")
}
}
element.filterChildren();
resolveListItem(element)
}, div:function(element) {
var singleChild = onlyChild(element);
if(singleChild && singleChild.nodeName === "table") {
var attrs = element.attributes;
S.each(attrs, function(attr) {
singleChild.setAttribute(attr.name, attr.value)
});
if(element.getAttribute("style")) {
addStyle(singleChild, element.getAttribute("style"))
}
var clearFloatDiv = new HtmlParser.Tag("div");
addStyle(clearFloatDiv, "clear", "both");
element.appendChild(clearFloatDiv);
element.setTagName(null)
}
}, td:function(element) {
if(getAncestor(element, "thead")) {
element.setTagName("th")
}
}, ol:flattenList, ul:flattenList, dl:flattenList, font:function(element) {
if(isListBulletIndicator(element.parentNode)) {
element.setTagName(null);
return
}
element.filterChildren();
var styleText = element.getAttribute("style"), parent = element.parentNode;
if("font" === parent.name) {
S.each(element.attributes, function(attr) {
parent.setAttribute(attr.name, attr.value)
});
if(styleText) {
addStyle(parent, styleText)
}
element.setTagName(null)
}else {
styleText = styleText || "";
if(element.getAttribute("color")) {
if(element.getAttribute("color") !== "#000000") {
styleText += "color:" + element.getAttribute("color") + ";"
}
element.removeAttribute("color")
}
if(element.getAttribute("face")) {
styleText += "font-family:" + element.getAttribute("face") + ";";
element.removeAttribute("face")
}
var size = element.getAttribute("size");
if(size) {
styleText += "font-size:" + (size > 3 ? "large" : size < 3 ? "small" : "medium") + ";";
element.removeAttribute("size")
}
element.setTagName("span");
addStyle(element, styleText)
}
}, span:function(element) {
if(isListBulletIndicator(element.parentNode)) {
return false
}
element.filterChildren();
if(containsNothingButSpaces(element)) {
element.setTagName(null);
return null
}
if(isListBulletIndicator(element)) {
var listSymbolNode = firstChild(element, function(node) {
return node.nodeValue || node.nodeName === "img"
});
var listSymbol = listSymbolNode && (listSymbolNode.nodeValue || "l."), listType = listSymbol && listSymbol.match(/^(?:[(]?)([^\s]+?)([.)]?)$/);
if(listType) {
var marker = createListBulletMarker(listType, listSymbol);
var ancestor = getAncestor(element, "span");
if(ancestor && / mso-hide:\s*all|display:\s*none /.test(ancestor.getAttribute("style"))) {
marker.setAttribute("ke:ignored", 1)
}
return marker
}
}
var styleText = element.getAttribute("style");
if(styleText) {
setStyle(element, stylesFilter([[/^line-height$/], [/^font-family$/], [/^font-size$/], [/^color$/], [/^background-color$/]])(styleText, element))
}
}, a:function(element) {
var href;
if(!(href = element.getAttribute("href")) && element.getAttribute("name")) {
element.setTagName(null)
}else {
if(UA.webkit && href && href.match(/file:\/\/\/[\S]+#/i)) {
element.setAttribute("href", href.replace(/file:\/\/\/[^#]+/i, ""))
}
}
}, "ke:listbullet":function(element) {
if(getAncestor(element, /h\d/)) {
element.setTagName(null)
}
}}, attributeNames:[[/^onmouse(:?out|over)/, ""], [/^onload$/, ""], [/(?:v|o):\w+/, ""], [/^lang/, ""]], attributes:{style:stylesFilter([[/^list-style-type$/], [/^margin$|^margin-(?!bottom|top)/, null, function(value, element, name) {
if(element.nodeName in {p:1, div:1}) {
var indentStyleName = "margin-left";
if(name === "margin") {
value = getStyleComponents(name, value, [indentStyleName])[indentStyleName]
}else {
if(name !== indentStyleName) {
return null
}
}
if(value && !emptyMarginRegex.test(value)) {
return[indentStyleName, value]
}
}
return null
}], [/^clear$/], [/^border.*|margin.*|vertical-align|float$/, null, function(value, element) {
if(element.nodeName === "img") {
return value
}
}], [/^width|height$/, null, function(value, element) {
if(element.nodeName in {table:1, td:1, th:1, img:1}) {
return value
}
}]], 1), width:function(value, element) {
if(element.nodeName in dtd.$tableContent) {
return false
}
}, border:function(value, element) {
if(element.nodeName in dtd.$tableContent) {
return false
}
}, "class":falsyFilter, bgcolor:falsyFilter, valign:function(value, element) {
addStyle(element, "vertical-align", value);
return false
}}, comment:UA.ie ? function(value, node) {
var imageInfo = value.match(/<img.*?>/), listInfo = value.match(/^\[if !supportLists\]([\s\S]*?)\[endif\]$/);
if(listInfo) {
var listSymbol = listInfo[1] || imageInfo && "l.", listType = listSymbol && listSymbol.match(/>(?:[(]?)([^\s]+?)([.)]?)</);
return createListBulletMarker(listType, listSymbol)
}
if(UA.gecko && imageInfo) {
var img = (new HtmlParser.Parser(imageInfo[0])).parse().childNodes[0], previousComment = node.previousSibling, imgSrcInfo = previousComment && previousComment.toHtml().match(/<v:imagedata[^>]*o:href=[''](.*?)['']/), imgSrc = imgSrcInfo && imgSrcInfo[1];
if(imgSrc) {
img.setAttribute("src", imgSrc)
}
return img
}
return false
} : falsyFilter})
})();
KISSY.add("editor/plugin/word-filter", ["html-parser"], function(S, require) {
var HtmlParser = require("html-parser");
var $ = S.all, UA = S.UA, dtd = HtmlParser.DTD, wordFilter = new HtmlParser.Filter, cssLengthRelativeUnit = /^([.\d]*)+(em|ex|px|gd|rem|vw|vh|vm|ch|mm|cm|in|pt|pc|deg|rad|ms|s|hz|khz){1}?/i, emptyMarginRegex = /^(?:\b0[^\s]*\s*){1,4}$/, romanLiteralPattern = "^m{0,4}(cm|cd|d?c{0,3})(xc|xl|l?x{0,3})(ix|iv|v?i{0,3})$", lowerRomanLiteralRegex = new RegExp(romanLiteralPattern), upperRomanLiteralRegex = new RegExp(romanLiteralPattern.toUpperCase()), orderedPatterns = {decimal:/\d+/, "lower-roman":lowerRomanLiteralRegex,
"upper-roman":upperRomanLiteralRegex, "lower-alpha":/^[a-z]+$/, "upper-alpha":/^[A-Z]+$/}, unorderedPatterns = {disc:/[l\u00B7\u2002]/, circle:/[\u006F\u00D8]/, square:/[\u006E\u25C6]/}, listMarkerPatterns = {ol:orderedPatterns, ul:unorderedPatterns}, romans = [[1E3, "M"], [900, "CM"], [500, "D"], [400, "CD"], [100, "C"], [90, "XC"], [50, "L"], [40, "XL"], [10, "X"], [9, "IX"], [5, "V"], [4, "IV"], [1, "I"]], alphabets = "ABCDEFGHIJKLMNOPQRSTUVWXYZ";
function fromRoman(str) {
str = str.toUpperCase();
var l = romans.length, retVal = 0;
for(var i = 0;i < l;++i) {
for(var j = romans[i], k = j[1].length;str.substr(0, k) === j[1];str = str.substr(k)) {
retVal += j[0]
}
}
return retVal
}
function fromAlphabet(str) {
str = str.toUpperCase();
var l = alphabets.length, retVal = 1;
for(var x = 1;str.length > 0;x *= l) {
retVal += alphabets.indexOf(str.charAt(str.length - 1)) * x;
str = str.substr(0, str.length - 1)
}
return retVal
}
function setStyle(element, str) {
if(str) {
element.setAttribute("style", str)
}else {
element.removeAttribute("style")
}
}
var convertToPx = function() {
var calculator;
return function(cssLength) {
if(!calculator) {
calculator = $('<div style="position:absolute;left:-9999px;' + 'top:-9999px;margin:0px;padding:0px;border:0px;"' + "></div>").prependTo("body")
}
if(!/%$/.test(cssLength)) {
calculator.css("width", cssLength);
return calculator[0].clientWidth
}
return cssLength
}
}();
var listBaseIndent = 0, previousListItemMargin = null, previousListId;
function onlyChild(elem) {
var childNodes = elem.childNodes || [], count = childNodes.length, firstChild = count === 1 && childNodes[0];
return firstChild || null
}
function removeAnyChildWithName(elem, tagName) {
var children = elem.childNodes || [], ret = [], child;
for(var i = 0;i < children.length;i++) {
child = children[i];
if(!child.nodeName) {
continue
}
if(child.nodeName === tagName) {
ret.push(child);
children.splice(i--, 1)
}
ret = ret.concat(removeAnyChildWithName(child, tagName))
}
return ret
}
function getAncestor(elem, tagNameRegex) {
var parent = elem.parentNode;
while(parent && !(parent.nodeName && parent.nodeName.match(tagNameRegex))) {
parent = parent.parentNode
}
return parent
}
function firstChild(elem, evaluator) {
var child, i, children = elem.childNodes || [];
for(i = 0;i < children.length;i++) {
child = children[i];
if(evaluator(child)) {
return child
}else {
if(child.nodeName) {
child = firstChild(child, evaluator);
if(child) {
return child
}
}
}
}
return null
}
function addStyle(elem, name, value, isPrepend) {
var styleText, addingStyleText = "", style;
if(typeof value === "string") {
addingStyleText += name + ":" + value + ";"
}else {
if(typeof name === "object") {
for(style in name) {
addingStyleText += style + ":" + name[style] + ";"
}
}else {
addingStyleText += name
}
isPrepend = value
}
styleText = elem.getAttribute("style");
styleText = (isPrepend ? [addingStyleText, styleText] : [styleText, addingStyleText]).join(";");
setStyle(elem, styleText.replace(/^;|;(?=;)/, ""))
}
function parentOf(tagName) {
var result = {}, tag;
for(tag in dtd) {
if(tag.indexOf("$") === -1 && dtd[tag][tagName]) {
result[tag] = 1
}
}
return result
}
var filters = {flattenList:function(element, level) {
level = typeof level === "number" ? level : 1;
var listStyleType;
switch(element.getAttribute("type")) {
case "a":
listStyleType = "lower-alpha";
break;
case "1":
listStyleType = "decimal";
break
}
var children = element.childNodes || [], child;
for(var i = 0;i < children.length;i++) {
child = children[i];
if(child.nodeName in dtd.$listItem) {
var listItemChildren = child.childNodes || [], count = listItemChildren.length, last = listItemChildren[count - 1];
if(last.nodeName in dtd.$list) {
element.insertAfter(child);
if(!--listItemChildren.length) {
element.removeChild(children[i--])
}
}
child.setTagName("ke:li");
if(element.getAttribute("start") && !i) {
element.setAttribute("value", element.getAttribute("start"))
}
filters.stylesFilter([["tab-stops", null, function(val) {
var margin = val.split(" ")[1].match(cssLengthRelativeUnit);
if(margin) {
previousListItemMargin = convertToPx(margin[0])
}
}], level === 1 ? ["mso-list", null, function(val) {
val = val.split(" ");
var listId = Number(val[0].match(/\d+/));
if(listId !== previousListId) {
child.setAttribute("ke:reset", 1)
}
previousListId = listId
}] : null])(child.getAttribute("style"));
child.setAttribute("ke:indent", level);
child.setAttribute("ke:listtype", element.nodeName);
child.setAttribute("ke:list-style-type", listStyleType)
}else {
if(child.nodeName in dtd.$list) {
arguments.callee.apply(this, [child, level + 1]);
children = children.slice(0, i).concat(child.childNodes).concat(children.slice(i + 1));
element.empty();
for(var j = 0, num = children.length;j < num;j++) {
element.appendChild(children[j])
}
}
}
}
element.nodeName = element.tagName = null;
element.setAttribute("ke:list", 1)
}, assembleList:function(element) {
var children = element.childNodes || [], child, listItem, listItemIndent, lastIndent, lastListItem, list, openedLists = [], previousListStyleType, previousListType;
var bullet, listType, listStyleType, itemNumeric;
for(var i = 0;i < children.length;i++) {
child = children[i];
if("ke:li" === child.nodeName) {
child.setTagName("li");
listItem = child;
bullet = listItem.getAttribute("ke:listsymbol");
bullet = bullet && bullet.match(/^(?:[(]?)([^\s]+?)([.)]?)$/);
listType = listStyleType = itemNumeric = null;
if(listItem.getAttribute("ke:ignored")) {
children.splice(i--, 1);
continue
}
if(listItem.getAttribute("ke:reset")) {
list = lastIndent = lastListItem = null
}
listItemIndent = Number(listItem.getAttribute("ke:indent"));
if(listItemIndent !== lastIndent) {
previousListType = previousListStyleType = null
}
if(!bullet) {
listType = listItem.getAttribute("ke:listtype") || "ol";
listStyleType = listItem.getAttribute("ke:list-style-type")
}else {
if(previousListType && listMarkerPatterns[previousListType][previousListStyleType].test(bullet[1])) {
listType = previousListType;
listStyleType = previousListStyleType
}else {
for(var type in listMarkerPatterns) {
for(var style in listMarkerPatterns[type]) {
if(listMarkerPatterns[type][style].test(bullet[1])) {
if(type === "ol" && /alpha|roman/.test(style)) {
var num = /roman/.test(style) ? fromRoman(bullet[1]) : fromAlphabet(bullet[1]);
if(!itemNumeric || num < itemNumeric) {
itemNumeric = num;
listType = type;
listStyleType = style
}
}else {
listType = type;
listStyleType = style;
break
}
}
}
}
}
if(!listType) {
listType = bullet[2] ? "ol" : "ul"
}
}
previousListType = listType;
previousListStyleType = listStyleType || (listType === "ol" ? "decimal" : "disc");
if(listStyleType && listStyleType !== (listType === "ol" ? "decimal" : "disc")) {
addStyle(listItem, "list-style-type", listStyleType)
}
if(listType === "ol" && bullet) {
switch(listStyleType) {
case "decimal":
itemNumeric = Number(bullet[1]);
break;
case "lower-roman":
;
case "upper-roman":
itemNumeric = fromRoman(bullet[1]);
break;
case "lower-alpha":
;
case "upper-alpha":
itemNumeric = fromAlphabet(bullet[1]);
break
}
listItem.setAttribute("value", itemNumeric)
}
if(!list) {
openedLists.push(list = new HtmlParser.Tag(listType));
list.appendChild(listItem);
element.replaceChild(list, children[i])
}else {
if(listItemIndent > lastIndent) {
openedLists.push(list = new HtmlParser.Tag(listType));
list.appendChild(listItem);
lastListItem.appendChild(list)
}else {
if(listItemIndent < lastIndent) {
var diff = lastIndent - listItemIndent, parent;
while(diff-- && (parent = list.parentNode)) {
list = parent.parentNode
}
list.appendChild(listItem)
}else {
list.appendChild(listItem)
}
}
children.splice(i--, 1)
}
lastListItem = listItem;
lastIndent = listItemIndent
}else {
if(list && !(child.nodeType === 3 && !S.trim(child.nodeValue))) {
list = lastIndent = lastListItem = null
}
}
}
for(i = 0;i < openedLists.length;i++) {
postProcessList(openedLists[i])
}
}, falsyFilter:function() {
return false
}, stylesFilter:function(styles, whitelist) {
return function(styleText, element) {
var rules = [];
(styleText || "").replace(/"/g, '"').replace(/\s*([^ :;]+)\s*:\s*([^;]+)\s*(?=;|$)/g, function(match, name, value) {
name = name.toLowerCase();
if(name === "font-family") {
value = value.replace(/['']/g, "")
}
var namePattern, valuePattern, newValue, newName;
for(var i = 0;i < styles.length;i++) {
if(styles[i]) {
namePattern = styles[i][0];
valuePattern = styles[i][1];
newValue = styles[i][2];
newName = styles[i][3];
if(name.match(namePattern) && (!valuePattern || value.match(valuePattern))) {
name = newName || name;
if(whitelist) {
newValue = newValue || value
}
if(typeof newValue === "function") {
newValue = newValue(value, element, name)
}
if(newValue && newValue.push) {
name = newValue[0];
newValue = newValue[1]
}
if(typeof newValue === "string") {
rules.push([name, newValue])
}
return
}
}
}
if(!whitelist) {
rules.push([name, value])
}
});
for(var i = 0;i < rules.length;i++) {
rules[i] = rules[i].join(":")
}
return rules.length ? rules.join(";") + ";" : false
}
}, applyStyleFilter:null};
function postProcessList(list) {
var children = list.childNodes || [], child, count = children.length, match, mergeStyle, styleTypeRegexp = /list-style-type:(.*?)(?:;|$)/, stylesFilter = filters.stylesFilter;
if(styleTypeRegexp.exec(list.getAttribute("style"))) {
return
}
for(var i = 0;i < count;i++) {
child = children[i];
if(child.getAttribute("value") && Number(child.getAttribute("value")) === i + 1) {
child.removeAttribute("value")
}
match = styleTypeRegexp.exec(child.getAttribute("style"));
if(match) {
if(match[1] === mergeStyle || !mergeStyle) {
mergeStyle = match[1]
}else {
mergeStyle = null;
break
}
}
}
if(mergeStyle) {
for(i = 0;i < count;i++) {
var style = children[i].getAttribute("style");
if(style) {
style = stylesFilter([["list-style-type"]])(style);
setStyle(children[i], style)
}
}
addStyle(list, "list-style-type", mergeStyle)
}
}
var utils = {createListBulletMarker:function(bullet, bulletText) {
var marker = new HtmlParser.Tag("ke:listbullet");
marker.setAttribute("ke:listsymbol", bullet[0]);
marker.appendChild(new HtmlParser.Text(bulletText));
return marker
}, isListBulletIndicator:function(element) {
var styleText = element.getAttribute("style");
if(/mso-list\s*:\s*Ignore/i.test(styleText)) {
return true
}
}, isContainingOnlySpaces:function(element) {
var text;
return(text = onlyChild(element)) && /^(:?\s| )+$/.test(text.nodeValue)
}, resolveList:function(element) {
var listMarker;
if((listMarker = removeAnyChildWithName(element, "ke:listbullet")) && listMarker.length && (listMarker = listMarker[0])) {
element.setTagName("ke:li");
if(element.getAttribute("style")) {
var styleStr = filters.stylesFilter([["text-indent"], ["line-height"], [/^margin(:?-left)?$/, null, function(margin) {
var values = margin.split(" ");
margin = convertToPx(values[3] || values[1] || values[0]);
if(!listBaseIndent && previousListItemMargin !== null && margin > previousListItemMargin) {
listBaseIndent = margin - previousListItemMargin
}
previousListItemMargin = margin;
if(listBaseIndent) {
element.setAttribute("ke:indent", listBaseIndent && Math.ceil(margin / listBaseIndent) + 1 || 1)
}
}], [/^mso-list$/, null, function(val) {
val = val.split(" ");
var listId = Number(val[0].match(/\d+/)), indent = Number(val[1].match(/\d+/));
if(indent === 1) {
if(listId !== previousListId) {
element.setAttribute("ke:reset", 1)
}
previousListId = listId
}
element.setAttribute("ke:indent", indent)
}]])(element.getAttribute("style"), element);
setStyle(element, styleStr)
}
if(!element.getAttribute("ke:indent")) {
previousListItemMargin = 0;
element.setAttribute("ke:indent", 1)
}
S.each(listMarker.attributes, function(a) {
element.setAttribute(a.name, a.value)
});
return true
}else {
previousListId = previousListItemMargin = listBaseIndent = null
}
return false
}, getStyleComponents:function() {
var calculator = $('<div style="position:absolute;left:-9999px;top:-9999px;"></div>').prependTo("body");
return function(name, styleValue, fetchList) {
calculator.css(name, styleValue);
var styles = {}, count = fetchList.length;
for(var i = 0;i < count;i++) {
styles[fetchList[i]] = calculator.css(fetchList[i])
}
return styles
}
}(), listDtdParents:parentOf("ol")};
(function() {
var blockLike = S.merge(dtd.$block, dtd.$listItem, dtd.$tableContent), falsyFilter = filters.falsyFilter, stylesFilter = filters.stylesFilter, createListBulletMarker = utils.createListBulletMarker, flattenList = filters.flattenList, assembleList = filters.assembleList, isListBulletIndicator = utils.isListBulletIndicator, containsNothingButSpaces = utils.isContainingOnlySpaces, resolveListItem = utils.resolveList, convertToPxStr = function(value) {
value = convertToPx(value);
return isNaN(value) ? value : value + "px"
}, getStyleComponents = utils.getStyleComponents, listDtdParents = utils.listDtdParents;
wordFilter.addRules({tagNames:[[/meta|link|script/, ""]], root:function(element) {
element.filterChildren();
assembleList(element)
}, tags:{"^":function(element) {
var applyStyleFilter;
if(UA.gecko && (applyStyleFilter = filters.applyStyleFilter)) {
applyStyleFilter(element)
}
}, $:function(element) {
var tagName = element.nodeName || "";
if(tagName in blockLike && element.getAttribute("style")) {
setStyle(element, stylesFilter([[/^(:?width|height)$/, null, convertToPxStr]])(element.getAttribute("style")))
}
if(tagName.match(/h\d/)) {
element.filterChildren();
if(resolveListItem(element)) {
return
}
}else {
if(tagName in dtd.$inline) {
element.filterChildren();
if(containsNothingButSpaces(element)) {
element.setTagName(null)
}
}else {
if(tagName.indexOf(":") !== -1 && tagName.indexOf("ke") === -1) {
element.filterChildren();
if(tagName === "v:imagedata") {
var href = element.getAttribute("o:href");
if(href) {
element.setAttribute("src", href)
}
element.setTagName("img");
return
}
element.setTagName(null)
}
}
}
if(tagName in listDtdParents) {
element.filterChildren();
assembleList(element)
}
}, style:function(element) {
if(UA.gecko) {
var styleDefSection = onlyChild(element).nodeValue.match(/\/\* Style Definitions \*\/([\s\S]*?)\/\*/), styleDefText = styleDefSection && styleDefSection[1], rules = {};
if(styleDefText) {
styleDefText.replace(/[\n\r]/g, "").replace(/(.+?)\{(.+?)\}/g, function(rule, selectors, styleBlock) {
selectors = selectors.split(",");
var length = selectors.length;
for(var i = 0;i < length;i++) {
S.trim(selectors[i]).replace(/^(\w+)(\.[\w-]+)?$/g, function(match, tagName, className) {
tagName = tagName || "*";
className = className.substring(1, className.length);
if(className.match(/MsoNormal/)) {
return
}
if(!rules[tagName]) {
rules[tagName] = {}
}
if(className) {
rules[tagName][className] = styleBlock
}else {
rules[tagName] = styleBlock
}
})
}
});
filters.applyStyleFilter = function(element) {
var name = rules["*"] ? "*" : element.nodeName, className = element.getAttribute("class"), style;
if(name in rules) {
style = rules[name];
if(typeof style === "object") {
style = style[className]
}
if(style) {
addStyle(element, style, true)
}
}
}
}
}
return false
}, p:function(element) {
if(/MsoListParagraph/.exec(element.getAttribute("class"))) {
var bulletText = firstChild(element, function(node) {
return node.nodeType === 3 && !containsNothingButSpaces(node.parentNode)
});
var bullet = bulletText && bulletText.parentNode;
if(bullet && !bullet.getAttribute("style")) {
bullet.setAttribute("style", "mso-list: Ignore;")
}
}
element.filterChildren();
resolveListItem(element)
}, div:function(element) {
var singleChild = onlyChild(element);
if(singleChild && singleChild.nodeName === "table") {
var attrs = element.attributes;
S.each(attrs, function(attr) {
singleChild.setAttribute(attr.name, attr.value)
});
if(element.getAttribute("style")) {
addStyle(singleChild, element.getAttribute("style"))
}
var clearFloatDiv = new HtmlParser.Tag("div");
addStyle(clearFloatDiv, "clear", "both");
element.appendChild(clearFloatDiv);
element.setTagName(null)
}
}, td:function(element) {
if(getAncestor(element, "thead")) {
element.setTagName("th")
}
}, ol:flattenList, ul:flattenList, dl:flattenList, font:function(element) {
if(isListBulletIndicator(element.parentNode)) {
element.setTagName(null);
return
}
element.filterChildren();
var styleText = element.getAttribute("style"), parent = element.parentNode;
if("font" === parent.name) {
S.each(element.attributes, function(attr) {
parent.setAttribute(attr.name, attr.value)
});
if(styleText) {
addStyle(parent, styleText)
}
element.setTagName(null)
}else {
styleText = styleText || "";
if(element.getAttribute("color")) {
if(element.getAttribute("color") !== "#000000") {
styleText += "color:" + element.getAttribute("color") + ";"
}
element.removeAttribute("color")
}
if(element.getAttribute("face")) {
styleText += "font-family:" + element.getAttribute("face") + ";";
element.removeAttribute("face")
}
var size = element.getAttribute("size");
if(size) {
styleText += "font-size:" + (size > 3 ? "large" : size < 3 ? "small" : "medium") + ";";
element.removeAttribute("size")
}
element.setTagName("span");
addStyle(element, styleText)
}
}, span:function(element) {
if(isListBulletIndicator(element.parentNode)) {
return false
}
element.filterChildren();
if(containsNothingButSpaces(element)) {
element.setTagName(null);
return null
}
if(isListBulletIndicator(element)) {
var listSymbolNode = firstChild(element, function(node) {
return node.nodeValue || node.nodeName === "img"
});
var listSymbol = listSymbolNode && (listSymbolNode.nodeValue || "l."), listType = listSymbol && listSymbol.match(/^(?:[(]?)([^\s]+?)([.)]?)$/);
if(listType) {
var marker = createListBulletMarker(listType, listSymbol);
var ancestor = getAncestor(element, "span");
if(ancestor && / mso-hide:\s*all|display:\s*none /.test(ancestor.getAttribute("style"))) {
marker.setAttribute("ke:ignored", 1)
}
return marker
}
}
var styleText = element.getAttribute("style");
if(styleText) {
setStyle(element, stylesFilter([[/^line-height$/], [/^font-family$/], [/^font-size$/], [/^color$/], [/^background-color$/]])(styleText, element))
}
}, a:function(element) {
var href;
if(!(href = element.getAttribute("href")) && element.getAttribute("name")) {
element.setTagName(null)
}else {
if(UA.webkit && href && href.match(/file:\/\/\/[\S]+#/i)) {
element.setAttribute("href", href.replace(/file:\/\/\/[^#]+/i, ""))
}
}
}, "ke:listbullet":function(element) {
if(getAncestor(element, /h\d/)) {
element.setTagName(null)
}
}}, attributeNames:[[/^onmouse(:?out|over)/, ""], [/^onload$/, ""], [/(?:v|o):\w+/, ""], [/^lang/, ""]], attributes:{style:stylesFilter([[/^list-style-type$/], [/^margin$|^margin-(?!bottom|top)/, null, function(value, element, name) {
if(element.nodeName in {p:1, div:1}) {
var indentStyleName = "margin-left";
if(name === "margin") {
value = getStyleComponents(name, value, [indentStyleName])[indentStyleName]
}else {
if(name !== indentStyleName) {
return null
}
}
if(value && !emptyMarginRegex.test(value)) {
return[indentStyleName, value]
}
}
return null
}], [/^clear$/], [/^border.*|margin.*|vertical-align|float$/, null, function(value, element) {
if(element.nodeName === "img") {
return value
}
}], [/^width|height$/, null, function(value, element) {
if(element.nodeName in {table:1, td:1, th:1, img:1}) {
return value
}
}]], 1), width:function(value, element) {
if(element.nodeName in dtd.$tableContent) {
return false
}
}, border:function(value, element) {
if(element.nodeName in dtd.$tableContent) {
return false
}
}, "class":falsyFilter, bgcolor:falsyFilter, valign:function(value, element) {
addStyle(element, "vertical-align", value);
return false
}}, comment:UA.ie ? function(value, node) {
var imageInfo = value.match(/<img.*?>/), listInfo = value.match(/^\[if !supportLists\]([\s\S]*?)\[endif\]$/);
if(listInfo) {
var listSymbol = listInfo[1] || imageInfo && "l.", listType = listSymbol && listSymbol.match(/>(?:[(]?)([^\s]+?)([.)]?)</);
return createListBulletMarker(listType, listSymbol)
}
if(UA.gecko && imageInfo) {
var img = (new HtmlParser.Parser(imageInfo[0])).parse().childNodes[0], previousComment = node.previousSibling, imgSrcInfo = previousComment && previousComment.toHtml().match(/<v:imagedata[^>]*o:href=[''](.*?)['']/), imgSrc = imgSrcInfo && imgSrcInfo[1];
if(imgSrc) {
img.setAttribute("src", imgSrc)
}
return img
}
return false
} : falsyFilter})
})();
return{toDataFormat:function(html, editor) {
if(UA.gecko) {
html = html.replace(/(<!--\[if[^<]*?\])--\>([\S\s]*?)<!--(\[endif\]--\>)/gi, "$1$2$3")
}
html = editor.htmlDataProcessor.toDataFormat(html, wordFilter);
return html
}}
});
(function () {
function wrapAsComment(element) {
var html = HtmlParser.serialize(element);
return new HtmlParser.Comment(protectedSourceMarker +
encodeURIComponent(html).replace(/--/g,'%2D%2D'));
}
// 过滤外边来的 html
var defaultDataFilterRules = {
tagNames: [
[/^\?xml.*$/i, ''],
[/^.*namespace.*$/i, '']
],
attributeNames: [
// Event attributes (onXYZ) must not be directly set. They can become
// active in the editing area (IE|WebKit).
[/^on/, 'ke_on'],
[/^lang$/, '']
],
tags: {
script: wrapAsComment,
noscript: wrapAsComment,
span: filterInline
}
};
// 将编辑区生成 html 最终化
var defaultHTMLFilterRules = {
tagNames: [
// Remove the "ke:" namespace prefix.
[ ( /^ke:/ ), '' ],
// Ignore <?xml:namespace> tags.
[ ( /^\?xml:namespace$/ ), '' ]
],
tags: {
$: function (element) {
var attributes = element.attributes;
if (attributes.length) {
// 先把真正属性去掉，后面会把 _ke_saved 后缀去掉的！
// Remove duplicated attributes - #3789.
var attributeNames = [ 'name', 'href', 'src' ],
savedAttributeName;
for (var i = 0; i < attributeNames.length; i++) {
savedAttributeName = '_keSaved_' + attributeNames[ i ];
if (element.getAttribute(savedAttributeName)) {
element.removeAttribute(attributeNames[i]);
}
}
}
return element;
},
embed: function (element) {
var parent = element.parentNode;
// If the <embed> is child of a <object>, copy the width
// and height attributes from it.
if (parent && parent.nodeName === 'object') {
var parentWidth = parent.getAttribute('width'),
parentHeight = parent.getAttribute('height');
if (parentWidth) {
element.setAttribute('width', parentWidth);
}
if (parentHeight) {
element.setAttribute('width', parentHeight);
}
}
},
// Remove empty link but not empty anchor.(#3829)
a: function (element) {
if (!(element.childNodes.length) && !(element.attributes.length)) {
return false;
}
return undefined;
},
span: filterInline,
strong: filterInline,
em: filterInline,
del: filterInline,
u: filterInline
},
attributes: {
// 清除空style
style: function (v) {
if (!S.trim(v)) {
return false;
}
return undefined;
}
},
attributeNames: [
// 把保存的作为真正的属性，替换掉原来的
// replace(/^_keSaved_/,"")
// _keSavedHref -> href
[ ( /^_keSaved_/ ), '' ],
[ ( /^ke_on/ ), 'on' ],
[ ( /^_ke.*/ ), '' ],
[ ( /^ke:.*$/ ), '' ],
// kissy 相关
[ ( /^_ks.*/ ), '' ]
],
comment: function (contents) {
// If this is a comment for protected source.
if (contents.substr(0, protectedSourceMarker.length) === protectedSourceMarker) {
contents = S.trim(S.urlDecode(contents.substr(protectedSourceMarker.length)));
return HtmlParser.parse(contents).childNodes[0];
}
return undefined;
}
};
if (OLD_IE) {
// IE outputs style attribute in capital letters. We should convert
// them back to lower case.
// bug: style='background:url(www.G.cn)' => style='background:url(www.g.cn)'
// 只对 propertyName 小写
defaultHTMLFilterRules.attributes.style = function (value // , element
) {
return value.replace(/(^|;)([^:]+)/g, function (match) {
return match.toLowerCase();
});
};
}
htmlFilter.addRules(defaultHTMLFilterRules);
dataFilter.addRules(defaultDataFilterRules);
})();
(function () {
// Regex to scan for at the end of blocks,
// which are actually placeholders.
// Safari transforms the to \xa0. (#4172)
// html will auto indent by kissy html-parser to add \r \n at the end of line
var tailNbspRegex = /^[\t\r\n ]*(?: |\xa0)[\t\r\n ]*$/;
// Return the last non-space child node of the block (#4344).
function lastNoneSpaceChild(block) {
var childNodes = block.childNodes,
lastIndex = childNodes.length,
last = childNodes[ lastIndex - 1 ];
while (last && last.nodeType === 3 && !S.trim(last.nodeValue)) {
last = childNodes[ --lastIndex ];
}
return last;
}
function trimFillers(block) {
var lastChild = lastNoneSpaceChild(block);
if (lastChild) {
if (lastChild.nodeType === 1 && lastChild.nodeName === 'br') {
block.removeChild(lastChild);
}
else if (lastChild.nodeType === 3 && tailNbspRegex.test(lastChild.nodeValue)) {
block.removeChild(lastChild);
}
}
}
function blockNeedsExtension(block) {
var lastChild = lastNoneSpaceChild(block);
// empty block <p></p> <td></td>
return !lastChild ||
// Some of the controls in form needs extension too,
// to move cursor at the end of the form. (#4791)
block.nodeName === 'form' &&
lastChild.nodeName === 'input';
}
// 外部 html 到编辑器 html
function extendBlockForDisplay(block) {
trimFillers(block);
if (blockNeedsExtension(block)) {
// non-ie need br for cursor and height
// ie does not need!
if (!OLD_IE) {
block.appendChild(new HtmlParser.Tag('br'));
}
}
}
// 编辑器 html 到外部 html
function extendBlockForOutput(block) {
trimFillers(block);
if (blockNeedsExtension(block)) {
// allow browser need!
// <p></p> does not has height!
block.appendChild(new HtmlParser.Text('\xa0'));
}
}
// Find out the list of block-like tags that can contain <br>.
var dtd = Editor.XHTML_DTD;
var blockLikeTags = S.merge(
dtd.$block,
dtd.$listItem,
dtd.$tableContent), i;
for (i in blockLikeTags) {
if (!( 'br' in dtd[i] )) {
delete blockLikeTags[i];
}
}
// We just avoid filler in <pre> right now.
// TODO: Support filler for <pre>, line break is also occupy line height.
delete blockLikeTags.pre;
var defaultDataBlockFilterRules = { tags: {} };
var defaultHTMLBlockFilterRules = { tags: {} };
for (i in blockLikeTags) {
defaultDataBlockFilterRules.tags[ i ] = extendBlockForDisplay;
defaultHTMLBlockFilterRules.tags[ i ] = extendBlockForOutput;
}
dataFilter.addRules(defaultDataBlockFilterRules);
htmlFilter.addRules(defaultHTMLBlockFilterRules);
})();
init: function (editor) {
var Node = S.Node,
UA = S.UA,
htmlFilter = new HtmlParser.Filter(),
dataFilter = new HtmlParser.Filter();
// remove empty inline element
function filterInline(element) {
var childNodes = element.childNodes,
i,
child,
allEmpty,
l = childNodes.length;
if (l) {
allEmpty = 1;
for (i = 0; i < l; i++) {
child = childNodes[i];
if (!(child.nodeType === S.DOM.NodeType.TEXT_NODE && !child.nodeValue)) {
allEmpty = 0;
break;
}
}
return allEmpty ? false : undefined;
} else {
return false;
}
}
(function () {
function wrapAsComment(element) {
var html = HtmlParser.serialize(element);
return new HtmlParser.Comment(protectedSourceMarker +
encodeURIComponent(html).replace(/--/g,'%2D%2D'));
}
// 过滤外边来的 html
var defaultDataFilterRules = {
tagNames: [
[/^\?xml.*$/i, ''],
[/^.*namespace.*$/i, '']
],
attributeNames: [
// Event attributes (onXYZ) must not be directly set. They can become
// active in the editing area (IE|WebKit).
[/^on/, 'ke_on'],
[/^lang$/, '']
],
tags: {
script: wrapAsComment,
noscript: wrapAsComment,
span: filterInline
}
};
// 将编辑区生成 html 最终化
var defaultHTMLFilterRules = {
tagNames: [
// Remove the "ke:" namespace prefix.
[ ( /^ke:/ ), '' ],
// Ignore <?xml:namespace> tags.
[ ( /^\?xml:namespace$/ ), '' ]
],
tags: {
$: function (element) {
var attributes = element.attributes;
if (attributes.length) {
// 先把真正属性去掉，后面会把 _ke_saved 后缀去掉的！
// Remove duplicated attributes - #3789.
var attributeNames = [ 'name', 'href', 'src' ],
savedAttributeName;
for (var i = 0; i < attributeNames.length; i++) {
savedAttributeName = '_keSaved_' + attributeNames[ i ];
if (element.getAttribute(savedAttributeName)) {
element.removeAttribute(attributeNames[i]);
}
}
}
return element;
},
embed: function (element) {
var parent = element.parentNode;
// If the <embed> is child of a <object>, copy the width
// and height attributes from it.
if (parent && parent.nodeName === 'object') {
var parentWidth = parent.getAttribute('width'),
parentHeight = parent.getAttribute('height');
if (parentWidth) {
element.setAttribute('width', parentWidth);
}
if (parentHeight) {
element.setAttribute('width', parentHeight);
}
}
},
// Remove empty link but not empty anchor.(#3829)
a: function (element) {
if (!(element.childNodes.length) && !(element.attributes.length)) {
return false;
}
return undefined;
},
span: filterInline,
strong: filterInline,
em: filterInline,
del: filterInline,
u: filterInline
},
attributes: {
// 清除空style
style: function (v) {
if (!S.trim(v)) {
return false;
}
return undefined;
}
},
attributeNames: [
// 把保存的作为真正的属性，替换掉原来的
// replace(/^_keSaved_/,"")
// _keSavedHref -> href
[ ( /^_keSaved_/ ), '' ],
[ ( /^ke_on/ ), 'on' ],
[ ( /^_ke.*/ ), '' ],
[ ( /^ke:.*$/ ), '' ],
// kissy 相关
[ ( /^_ks.*/ ), '' ]
],
comment: function (contents) {
// If this is a comment for protected source.
if (contents.substr(0, protectedSourceMarker.length) === protectedSourceMarker) {
contents = S.trim(S.urlDecode(contents.substr(protectedSourceMarker.length)));
return HtmlParser.parse(contents).childNodes[0];
}
return undefined;
}
};
if (OLD_IE) {
// IE outputs style attribute in capital letters. We should convert
// them back to lower case.
// bug: style='background:url(www.G.cn)' => style='background:url(www.g.cn)'
// 只对 propertyName 小写
defaultHTMLFilterRules.attributes.style = function (value // , element
) {
return value.replace(/(^|;)([^:]+)/g, function (match) {
return match.toLowerCase();
});
};
}
htmlFilter.addRules(defaultHTMLFilterRules);
dataFilter.addRules(defaultDataFilterRules);
})();
/*
去除firefox代码末尾自动添加的 <br/>
以及ie下自动添加的
以及其他浏览器段落末尾添加的占位符
*/
(function () {
// Regex to scan for at the end of blocks,
// which are actually placeholders.
// Safari transforms the to \xa0. (#4172)
// html will auto indent by kissy html-parser to add \r \n at the end of line
var tailNbspRegex = /^[\t\r\n ]*(?: |\xa0)[\t\r\n ]*$/;
// Return the last non-space child node of the block (#4344).
function lastNoneSpaceChild(block) {
var childNodes = block.childNodes,
lastIndex = childNodes.length,
last = childNodes[ lastIndex - 1 ];
while (last && last.nodeType === 3 && !S.trim(last.nodeValue)) {
last = childNodes[ --lastIndex ];
}
return last;
}
function trimFillers(block) {
var lastChild = lastNoneSpaceChild(block);
if (lastChild) {
if (lastChild.nodeType === 1 && lastChild.nodeName === 'br') {
block.removeChild(lastChild);
}
else if (lastChild.nodeType === 3 && tailNbspRegex.test(lastChild.nodeValue)) {
block.removeChild(lastChild);
}
}
}
function blockNeedsExtension(block) {
var lastChild = lastNoneSpaceChild(block);
// empty block <p></p> <td></td>
return !lastChild ||
// Some of the controls in form needs extension too,
// to move cursor at the end of the form. (#4791)
block.nodeName === 'form' &&
lastChild.nodeName === 'input';
}
// 外部 html 到编辑器 html
function extendBlockForDisplay(block) {
trimFillers(block);
if (blockNeedsExtension(block)) {
// non-ie need br for cursor and height
// ie does not need!
if (!OLD_IE) {
block.appendChild(new HtmlParser.Tag('br'));
}
}
}
// 编辑器 html 到外部 html
function extendBlockForOutput(block) {
trimFillers(block);
if (blockNeedsExtension(block)) {
// allow browser need!
// <p></p> does not has height!
block.appendChild(new HtmlParser.Text('\xa0'));
}
}
// Find out the list of block-like tags that can contain <br>.
var dtd = Editor.XHTML_DTD;
var blockLikeTags = S.merge(
dtd.$block,
dtd.$listItem,
dtd.$tableContent), i;
for (i in blockLikeTags) {
if (!( 'br' in dtd[i] )) {
delete blockLikeTags[i];
}
}
// We just avoid filler in <pre> right now.
// TODO: Support filler for <pre>, line break is also occupy line height.
delete blockLikeTags.pre;
var defaultDataBlockFilterRules = { tags: {} };
var defaultHTMLBlockFilterRules = { tags: {} };
for (i in blockLikeTags) {
defaultDataBlockFilterRules.tags[ i ] = extendBlockForDisplay;
defaultHTMLBlockFilterRules.tags[ i ] = extendBlockForOutput;
}
dataFilter.addRules(defaultDataBlockFilterRules);
htmlFilter.addRules(defaultHTMLBlockFilterRules);
})();
// html-parser fragment 中的 entities 处理
// el.innerHTML=" "
// http://yiminghe.javaeye.com/blog/788929
htmlFilter.addRules({
text: function (text) {
return text
//.replace(/ /g, "\xa0")
.replace(/\xa0/g, ' ');
}
});
var protectElementRegex = /<(a|area|img|input)\b([^>]*)>/gi,
protectAttributeRegex = /\b(href|src|name)\s*=\s*(?:(?:"[^"]*")|(?:'[^']*')|(?:[^ "'>]+))/gi;
// ie 6-7 会将 关于 url 的 content value 替换为 dom value
// #a -> http://xxx/#a
// ../x.html -> http://xx/x.html
function protectAttributes(html) {
return html.replace(protectElementRegex, function (element, tag, attributes) {
return '<' + tag + attributes.replace(protectAttributeRegex, function (fullAttr, attrName) {
// We should not rewrite the existed protected attributes,
// e.g. clipboard content from editor. (#5218)
if (attributes.indexOf('_keSaved_' + attrName) === -1) {
return ' _keSaved_' + fullAttr + ' ' + fullAttr;
}
return fullAttr;
}) + '>';
});
}
var protectedSourceMarker = '{ke_protected}';
var protectElementsRegex = /(?:<textarea[^>]*>[\s\S]*<\/textarea>)|(?:<style[^>]*>[\s\S]*<\/style>)|(?:<script[^>]*>[\s\S]*<\/script>)|(?:<(:?link|meta|base)[^>]*>)/gi,
encodedElementsRegex = /<ke:encoded>([^<]*)<\/ke:encoded>/gi;
var protectElementNamesRegex = /(<\/?)((?:object|embed|param|html|body|head|title|noscript)[^>]*>)/gi,
unprotectElementNamesRegex = /(<\/?)ke:((?:object|embed|param|html|body|head|title|noscript)[^>]*>)/gi;
var protectSelfClosingRegex = /<ke:(param|embed)([^>]*?)\/?>(?!\s*<\/ke:\1)/gi;
function protectSelfClosingElements(html) {
return html.replace(protectSelfClosingRegex, '<ke:$1$2></ke:$1>');
}
function protectElements(html) {
return html.replace(protectElementsRegex, function (match) {
return '<ke:encoded>' + encodeURIComponent(match) + '</ke:encoded>';
});
}
function unprotectElements(html) {
return html.replace(encodedElementsRegex, function (match, encoded) {
return S.urlDecode(encoded);
});
}
function protectElementsNames(html) {
return html.replace(protectElementNamesRegex, '$1ke:$2');
}
function unprotectElementNames(html) {
return html.replace(unprotectElementNamesRegex, '$1$2');
}
editor.htmlDataProcessor = {
dataFilter: dataFilter,
htmlFilter: htmlFilter,
// 编辑器 html 到外部 html
// fixForBody, <body>t</body> => <body><p>t</p></body>
toHtml: function (html) {
if (UA.webkit) {
// remove filling char for webkit
html = html.replace(/\u200b/g, '');
}
// fixForBody = fixForBody || 'p';
// Now use our parser to make further fixes to the structure, as
// well as apply the filter.
//使用 htmlWriter 界面美观，加入额外文字节点\n,\t空白等
var writer = new HtmlParser.BeautifyWriter(),
n = new HtmlParser.Parser(html).parse();
n.writeHtml(writer, htmlFilter);
html = writer.getHtml();
return html;
},
// 外部html进入编辑器
toDataFormat: function (html, _dataFilter) {
//可以传 wordFilter 或 dataFilter
_dataFilter = _dataFilter || dataFilter;
// Protect elements than can't be set inside a DIV. E.g. IE removes
// style tags from innerHTML. (#3710)
// and protect textarea, in case textarea has un-encoded html
// protect script too, in case script has un-encoded html
// https://github.com/kissyteam/kissy/issues/420
html = protectElements(html);
html = protectAttributes(html);
// Certain elements has problem to go through Dom operation, protect
// them by prefixing 'ke' namespace. (#3591)
html = protectElementsNames(html);
// All none-IE browsers ignore self-closed custom elements,
// protecting them into open-close. (#3591)
html = protectSelfClosingElements(html);
// 标签不合法可能 parser 出错，这里先用浏览器帮我们建立棵合法的 dom 树的 html
// Call the browser to help us fixing a possibly invalid HTML
// structure.
var div = new Node('<div>');
// Add fake character to workaround IE comments bug. (#3801)
div.html('a' + html);
html = div.html().substr(1);
// Unprotect "some" of the protected elements at this point.
html = unprotectElementNames(html);
html = unprotectElements(html);
// fixForBody = fixForBody || 'p';
// bug:qc #3710:使用 basicWriter ，去除无用的文字节点，标签间连续\n空白等
var writer = new HtmlParser.BasicWriter(),
n = new HtmlParser.Parser(html).parse();
n.writeHtml(writer, _dataFilter);
html = writer.getHtml();
return html;
},
/*
最精简html传送到server
*/
toServer: function (html) {
var writer = new HtmlParser.MinifyWriter(),
n = new HtmlParser.Parser(html).parse();
n.writeHtml(writer, htmlFilter);
return writer.getHtml();
}
};
}