/// <summary>
/// Converts tag names, attribute names, and style text to lowercase.
/// </summary>
private string CleanupHtml(string html, bool xml)
{
bool needsCleanup;
do
{
needsCleanup = false;
StringBuilder output = new StringBuilder(html.Length);
SimpleHtmlParser htmlParser = new SimpleHtmlParser(html);
for (Element el; null != (el = htmlParser.Next());)
{
if (el is BeginTag)
{
BeginTag bt = (BeginTag)el;
if (RemoveMeaninglessTags(htmlParser, bt))
{
// Since we are removing a tag, we will want to clean up again, since that might mean
// there will be another tag to remove
needsCleanup = true;
continue;
}
output.Append("<");
output.Append(bt.Name.ToLower(CultureInfo.InvariantCulture));
foreach (Attr attr in bt.Attributes)
{
if (attr.NameEquals("contenteditable") || attr.NameEquals("atomicselection") ||
attr.NameEquals("unselectable"))
{
continue;
}
output.Append(" ");
output.Append(attr.Name.ToLower(CultureInfo.InvariantCulture));
if (attr.Value != null)
{
string attrVal = attr.Value;
if (attr.NameEquals("style"))
{
attrVal = LowerCaseCss(attrVal);
}
else if (attr.Name == attr.Value)
{
attrVal = attrVal.ToLower(CultureInfo.InvariantCulture);
}
output.AppendFormat("=\"{0}\"",
xml
? HtmlUtils.EscapeEntitiesForXml(attrVal, true)
: HtmlUtils.EscapeEntities(attrVal));
}
}
if (bt.HasResidue)
{
if (bt.Attributes.Length == 0)
{
output.Append(" ");
}
output.Append(bt.Residue);
}
if (bt.Complete)
{
output.Append(" /");
}
output.Append(">");
}
else if (el is EndTag)
{
output.AppendFormat("</{0}>", ((EndTag)el).Name.ToLower(CultureInfo.InvariantCulture));
}
else if (el is Text)
{
string textHtml = HtmlUtils.TidyNbsps(el.RawText);
if (xml)
{
textHtml =
HtmlUtils.EscapeEntitiesForXml(
HtmlUtils.UnEscapeEntities(textHtml, HtmlUtils.UnEscapeMode.NonMarkupText), false);
}
output.Append(textHtml);
}
else if (el is StyleText)
{
output.Append(el.RawText.ToLower(CultureInfo.InvariantCulture));
}
else
{
output.Append(el.RawText);
}
}
html = output.ToString();
} while (needsCleanup);
return(html);
}
