/// <summary>
/// Main HTML cleaner for <code>LocalToWeb</code>. It calls other HTML cleaners in a certain order
/// and returns the cleaned HTML.
/// </summary>
/// <param name="content">HTML content.</param>
/// <returns>Cleaned HTML content.</returns>
public string Clean(string content)
{
String uncleanedContent = new CorrectAttributesCleaner().Clean(content);
uncleanedContent = new CorrectTagsClosingCleaner("img").Clean(uncleanedContent);
uncleanedContent = new CorrectTagsClosingCleaner("br").Clean(uncleanedContent);
content = new TidyHTMLCleaner(true).Clean(uncleanedContent);
if (content.Length == 0)
{
content = uncleanedContent;
}
content = new XmlNamespaceDefinitionsReplacer(htmlOpeningTag).Clean(content);
content = new ListCharsCleaner().Clean(content);
content = new EmptyParagraphsCleaner().Clean(content);
content = new NbspBetweenTagsRemover().Clean(content);
content = new OfficeNameSpacesTagsRemover().Clean(content);
content = new NbspReplacer().Clean(content);
content = new CommentsDivCleaner().Clean(content);
return content;
}
/// <summary>
/// Main HTML cleaner for <code>LocalToWeb</code>. It calls other HTML cleaners in a certain order
/// and returns the cleaned HTML.
/// </summary>
/// <param name="content">HTML content.</param>
/// <returns>Cleaned HTML content.</returns>
public string Clean(string content)
{
String uncleanedContent = new CorrectAttributesCleaner().Clean(content);
uncleanedContent = new CorrectTagsClosingCleaner("img").Clean(uncleanedContent);
uncleanedContent = new CorrectTagsClosingCleaner("br").Clean(uncleanedContent);
content = new TidyHTMLCleaner(true).Clean(uncleanedContent);
if (content.Length == 0)
{
content = uncleanedContent;
}
content = new XmlNamespaceDefinitionsReplacer(htmlOpeningTag).Clean(content);
content = new ListCharsCleaner().Clean(content);
content = new EmptyParagraphsCleaner().Clean(content);
content = new NbspBetweenTagsRemover().Clean(content);
content = new OfficeNameSpacesTagsRemover().Clean(content);
content = new NbspReplacer().Clean(content);
content = new CommentsDivCleaner().Clean(content);
return(content);
}
public void TestCleaner()
{
bool canLoadXML = false;
initialHTML = new CorrectAttributesCleaner().Clean(initialHTML);
Assert.AreEqual(expectedHTML, initialHTML);
try
{
new XmlDocument().LoadXml(initialHTML);
canLoadXML = true;
}
catch(Exception e)
{
Console.WriteLine(initialHTML.Replace(">",">" + Environment.NewLine));
Console.WriteLine(e);
canLoadXML = false;
}
Assert.IsTrue(canLoadXML);
}
