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 TestDropsUnknownTags()
{
String h = "<p><custom foo=true>Test</custom></p>";
String cleanHtml = iText.StyledXmlParser.Jsoup.Jsoup.Clean(h, Whitelist.Relaxed());
NUnit.Framework.Assert.AreEqual("<p>Test</p>", cleanHtml);
}
public virtual void TestDropImageScript()
{
String h = "<IMG SRC=\"javascript:alert('XSS')\">";
String cleanHtml = iText.StyledXmlParser.Jsoup.Jsoup.Clean(h, Whitelist.Relaxed());
NUnit.Framework.Assert.AreEqual("<img>", cleanHtml);
}
public virtual void TestCleanJavascriptHref()
{
String h = "<A HREF=\"javascript:document.location='http://www.google.com/'\">XSS</A>";
String cleanHtml = iText.StyledXmlParser.Jsoup.Jsoup.Clean(h, Whitelist.Relaxed());
NUnit.Framework.Assert.AreEqual("<a>XSS</a>", cleanHtml);
}
public virtual void TestDropXmlProc()
{
String h = "<?import namespace=\"xss\"><p>Hello</p>";
String cleanHtml = iText.StyledXmlParser.Jsoup.Jsoup.Clean(h, Whitelist.Relaxed());
NUnit.Framework.Assert.AreEqual("<p>Hello</p>", cleanHtml);
}
public virtual void TestDropScript()
{
String h = "<SCRIPT SRC=//ha.ckers.org/.j><SCRIPT>alert(/XSS/.source)</SCRIPT>";
String cleanHtml = iText.StyledXmlParser.Jsoup.Jsoup.Clean(h, Whitelist.Relaxed());
NUnit.Framework.Assert.AreEqual("", cleanHtml);
}
public virtual void TestDropComments()
{
String h = "<p>Hello<!-- no --></p>";
String cleanHtml = iText.StyledXmlParser.Jsoup.Jsoup.Clean(h, Whitelist.Relaxed());
NUnit.Framework.Assert.AreEqual("<p>Hello</p>", cleanHtml);
}
public virtual void SimpleBehaviourTest()
{
String h = "<div><p class=foo><a href='http://evil.com'>Hello <b id=bar>there</b>!</a></div>";
String cleanHtml = iText.StyledXmlParser.Jsoup.Jsoup.Clean(h, Whitelist.SimpleText());
NUnit.Framework.Assert.AreEqual("Hello <b>there</b>!", TextUtil.StripNewlines(cleanHtml));
}
public virtual void SimpleBehaviourTest2()
{
String h = "Hello <b>there</b>!";
String cleanHtml = iText.StyledXmlParser.Jsoup.Jsoup.Clean(h, Whitelist.SimpleText());
NUnit.Framework.Assert.AreEqual("Hello <b>there</b>!", TextUtil.StripNewlines(cleanHtml));
}
public virtual void TestHandlesEmptyAttributes()
{
String h = "<img alt=\"\" src= unknown=''>";
String cleanHtml = iText.StyledXmlParser.Jsoup.Jsoup.Clean(h, Whitelist.BasicWithImages());
NUnit.Framework.Assert.AreEqual("<img alt=\"\">", cleanHtml);
}
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 TestRelaxed()
{
String h = "<h1>Head</h1><table><tr><td>One<td>Two</td></tr></table>";
String cleanHtml = iText.StyledXmlParser.Jsoup.Jsoup.Clean(h, Whitelist.Relaxed());
NUnit.Framework.Assert.AreEqual("<h1>Head</h1><table><tbody><tr><td>One</td><td>Two</td></tr></tbody></table>"
, 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 TestScriptTagInWhiteList()
{
Whitelist whitelist = Whitelist.Relaxed();
whitelist.AddTags("script");
NUnit.Framework.Assert.IsTrue(iText.StyledXmlParser.Jsoup.Jsoup.IsValid("Hello<script>alert('Doh')</script>World !"
, whitelist));
}
public virtual void BasicWithImagesTest()
{
String h = "<div><p><img src='http://example.com/' alt=Image></p><p><img src='ftp://ftp.example.com'></p></div>";
String cleanHtml = iText.StyledXmlParser.Jsoup.Jsoup.Clean(h, Whitelist.BasicWithImages());
NUnit.Framework.Assert.AreEqual("<p><img src=\"http://example.com/\" alt=\"Image\"></p><p><img></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 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 AddsTagOnAttributesIfNotSet()
{
String html = "<p class='foo' src='bar'>One</p>";
Whitelist whitelist = new Whitelist().AddAttributes("p", "class");
// ^^ whitelist does not have explicit tag add for p, inferred from add attributes.
String clean = iText.StyledXmlParser.Jsoup.Jsoup.Clean(html, whitelist);
NUnit.Framework.Assert.AreEqual("<p class=\"foo\">One</p>", clean);
}
public virtual void HandlesAllPseudoTag()
{
String html = "<p class='foo' src='bar'><a class='qux'>link</a></p>";
Whitelist whitelist = new Whitelist().AddAttributes(":all", "class").AddAttributes("p", "style").AddTags("p"
, "a");
String clean = iText.StyledXmlParser.Jsoup.Jsoup.Clean(html, whitelist);
NUnit.Framework.Assert.AreEqual("<p class=\"foo\"><a class=\"qux\">link</a></p>", clean);
}
public virtual void HandlesCustomProtocols()
{
String html = "<img src='cid:12345' /> <img src='data:gzzt' />";
String dropped = iText.StyledXmlParser.Jsoup.Jsoup.Clean(html, Whitelist.BasicWithImages());
NUnit.Framework.Assert.AreEqual("<img> \n<img>", dropped);
String preserved = iText.StyledXmlParser.Jsoup.Jsoup.Clean(html, Whitelist.BasicWithImages().AddProtocols(
"img", "src", "cid", "data"));
NUnit.Framework.Assert.AreEqual("<img src=\"cid:12345\"> \n<img src=\"data:gzzt\">", preserved);
}
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 TestCleanAnchorProtocol()
{
String validAnchor = "<a href=\"#valid\">Valid anchor</a>";
String invalidAnchor = "<a href=\"#anchor with spaces\">Invalid anchor</a>";
// A Whitelist that does not allow anchors will strip them out.
String cleanHtml = iText.StyledXmlParser.Jsoup.Jsoup.Clean(validAnchor, Whitelist.Relaxed());
NUnit.Framework.Assert.AreEqual("<a>Valid anchor</a>", cleanHtml);
cleanHtml = iText.StyledXmlParser.Jsoup.Jsoup.Clean(invalidAnchor, Whitelist.Relaxed());
NUnit.Framework.Assert.AreEqual("<a>Invalid anchor</a>", cleanHtml);
// A Whitelist that allows them will keep them.
Whitelist relaxedWithAnchor = Whitelist.Relaxed().AddProtocols("a", "href", "#");
cleanHtml = iText.StyledXmlParser.Jsoup.Jsoup.Clean(validAnchor, relaxedWithAnchor);
NUnit.Framework.Assert.AreEqual(validAnchor, cleanHtml);
// An invalid anchor is never valid.
cleanHtml = iText.StyledXmlParser.Jsoup.Jsoup.Clean(invalidAnchor, relaxedWithAnchor);
NUnit.Framework.Assert.AreEqual("<a>Invalid anchor</a>", cleanHtml);
}
public virtual void ResolvesRelativeLinks()
{
String html = "<a href='/foo'>Link</a><img src='/bar'>";
String clean = iText.StyledXmlParser.Jsoup.Jsoup.Clean(html, "http://example.com/", Whitelist.BasicWithImages
());
NUnit.Framework.Assert.AreEqual("<a href=\"http://example.com/foo\" rel=\"nofollow\">Link</a>\n<img src=\"http://example.com/bar\">"
, clean);
}
public virtual void CleansInternationalText()
{
NUnit.Framework.Assert.AreEqual("привет", iText.StyledXmlParser.Jsoup.Jsoup.Clean("привет", Whitelist.None
()));
}
public virtual void SupplyOutputSettings()
{
// test that one can override the default document output settings
OutputSettings os = new OutputSettings();
os.PrettyPrint(false);
os.EscapeMode(Entities.EscapeMode.extended);
os.Charset("ascii");
String html = "<div><p>ℬ</p></div>";
String customOut = iText.StyledXmlParser.Jsoup.Jsoup.Clean(html, "http://foo.com/", Whitelist.Relaxed(), os
);
String defaultOut = iText.StyledXmlParser.Jsoup.Jsoup.Clean(html, "http://foo.com/", Whitelist.Relaxed());
NUnit.Framework.Assert.AreNotSame(defaultOut, customOut);
NUnit.Framework.Assert.AreEqual("<div><p>ℬ</p></div>", customOut);
NUnit.Framework.Assert.AreEqual("<div>\n" + " <p>ℬ</p>\n" + "</div>", defaultOut);
os.Charset("ASCII");
os.EscapeMode(Entities.EscapeMode.@base);
String customOut2 = iText.StyledXmlParser.Jsoup.Jsoup.Clean(html, "http://foo.com/", Whitelist.Relaxed(),
os);
NUnit.Framework.Assert.AreEqual("<div><p>ℬ</p></div>", customOut2);
}
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()));
}
public virtual void PreservesRelativeLinksIfConfigured()
{
String html = "<a href='/foo'>Link</a><img src='/bar'> <img src='javascript:alert()'>";
String clean = iText.StyledXmlParser.Jsoup.Jsoup.Clean(html, "http://example.com/", Whitelist.BasicWithImages
().PreserveRelativeLinks(true));
NUnit.Framework.Assert.AreEqual("<a href=\"/foo\" rel=\"nofollow\">Link</a>\n<img src=\"/bar\"> \n<img>",
clean);
}
/// <summary>Create a new cleaner, that sanitizes documents using the supplied whitelist.</summary>
/// <param name="whitelist">white-list to clean with</param>
public Cleaner(Whitelist whitelist)
{
Validate.NotNull(whitelist);
this.whitelist = whitelist;
}