public virtual void DropsUnresolvableRelativeLinks()
{
String html = "<a href='/foo'>Link</a>";
String clean = iText.StyledXmlParser.Jsoup.Jsoup.Clean(html, Whitelist.Basic());
NUnit.Framework.Assert.AreEqual("<a rel=\"nofollow\">Link</a>", clean);
}
public virtual void BasicBehaviourTest()
{
String h = "<div><p><a href='javascript:sendAllMoney()'>Dodgy</a> <A HREF='HTTP://nice.com/'>Nice</a></p><blockquote>Hello</blockquote>";
String cleanHtml = iText.StyledXmlParser.Jsoup.Jsoup.Clean(h, Whitelist.Basic());
NUnit.Framework.Assert.AreEqual("<p><a rel=\"nofollow\">Dodgy</a> <a href=\"http://nice.com/\" rel=\"nofollow\">Nice</a></p><blockquote>Hello</blockquote>"
, TextUtil.StripNewlines(cleanHtml));
}
public virtual void TestRemoveTags()
{
String h = "<div><p><A HREF='HTTP://nice.com'>Nice</a></p><blockquote>Hello</blockquote>";
String cleanHtml = iText.StyledXmlParser.Jsoup.Jsoup.Clean(h, Whitelist.Basic().RemoveTags("a"));
NUnit.Framework.Assert.AreEqual("<p>Nice</p><blockquote>Hello</blockquote>", TextUtil.StripNewlines(cleanHtml
));
}
public virtual void TestRemoveProtocols()
{
String h = "<p>Contact me <a href='mailto:[email protected]'>here</a></p>";
String cleanHtml = iText.StyledXmlParser.Jsoup.Jsoup.Clean(h, Whitelist.Basic().RemoveProtocols("a", "href"
, "ftp", "mailto"));
NUnit.Framework.Assert.AreEqual("<p>Contact me <a rel=\"nofollow\">here</a></p>", TextUtil.StripNewlines(cleanHtml
));
}
public virtual void TestRemoveAttributes()
{
String h = "<div><p>Nice</p><blockquote cite='http://example.com/quotations'>Hello</blockquote>";
String cleanHtml = iText.StyledXmlParser.Jsoup.Jsoup.Clean(h, Whitelist.Basic().RemoveAttributes("blockquote"
, "cite"));
NUnit.Framework.Assert.AreEqual("<p>Nice</p><blockquote>Hello</blockquote>", TextUtil.StripNewlines(cleanHtml
));
}
public virtual void HandlesFramesets()
{
String dirty = "<html><head><script></script><noscript></noscript></head><frameset><frame src=\"foo\" /><frame src=\"foo\" /></frameset></html>";
String clean = iText.StyledXmlParser.Jsoup.Jsoup.Clean(dirty, Whitelist.Basic());
NUnit.Framework.Assert.AreEqual("", clean);
// nothing good can come out of that
Document dirtyDoc = iText.StyledXmlParser.Jsoup.Jsoup.Parse(dirty);
Document cleanDoc = new Cleaner(Whitelist.Basic()).Clean(dirtyDoc);
NUnit.Framework.Assert.IsFalse(cleanDoc == null);
NUnit.Framework.Assert.AreEqual(0, cleanDoc.Body().ChildNodeSize());
}
public virtual void TestIsValid()
{
String ok = "<p>Test <b><a href='http://example.com/'>OK</a></b></p>";
String nok1 = "<p><script></script>Not <b>OK</b></p>";
String nok2 = "<p align=right>Test Not <b>OK</b></p>";
String nok3 = "<!-- comment --><p>Not OK</p>";
// comments and the like will be cleaned
NUnit.Framework.Assert.IsTrue(iText.StyledXmlParser.Jsoup.Jsoup.IsValid(ok, Whitelist.Basic()));
NUnit.Framework.Assert.IsFalse(iText.StyledXmlParser.Jsoup.Jsoup.IsValid(nok1, Whitelist.Basic()));
NUnit.Framework.Assert.IsFalse(iText.StyledXmlParser.Jsoup.Jsoup.IsValid(nok2, Whitelist.Basic()));
NUnit.Framework.Assert.IsFalse(iText.StyledXmlParser.Jsoup.Jsoup.IsValid(nok3, Whitelist.Basic()));
}
