public static string CleanupHtml(string html, string baseUrl, bool preserveImages, bool strip, bool preserveTables)
{
// sterilize the HTML
html = UnsafeHtmlFragmentHelper.SterilizeHtml(html, UnsafeHtmlFragmentHelper.Flag.AllFlags ^ UnsafeHtmlFragmentHelper.Flag.RemoveStyles);
html = StripNamespacedTags(html);
// get the text into a DOM to ensure tags are balanced
IHTMLDocument2 document = HTMLDocumentHelper.StringToHTMLDoc(html, null, false);
if (document.body == null)
{
return(string.Empty);
}
// thin it
if (preserveTables)
{
html = LightWeightHTMLThinner2.Thin(document.body.innerHTML, preserveImages, strip, LightWeightHTMLThinner2.PreserveTables);
}
else
{
html = LightWeightHTMLThinner2.Thin(document.body.innerHTML, preserveImages, strip);
}
html = LightWeightHTMLUrlToAbsolute.ConvertToAbsolute(html, baseUrl, false, true, true);
// balance it
string balancedHtml = HTMLBalancer.Balance(html);
// return
return(balancedHtml);
}
public static string PreserveFormatting(string html, string baseUrl)
{
UnsafeHtmlFragmentHelper.Flag flags = UnsafeHtmlFragmentHelper.Flag.RemoveDocumentTags |
UnsafeHtmlFragmentHelper.Flag.RemoveScriptTags |
UnsafeHtmlFragmentHelper.Flag.RemoveScriptAttributes |
UnsafeHtmlFragmentHelper.Flag.RemoveMarkupDirectives |
UnsafeHtmlFragmentHelper.Flag.RemoveComments;
html = UnsafeHtmlFragmentHelper.SterilizeHtml(html, flags);
html = LightWeightHTMLUrlToAbsolute.ConvertToAbsolute(html, baseUrl, false, true, true);
return(html);
}