public void preservesRelatedLinksIfConfigured()
{
string html = "<a href='/foo'>Link</a><img src='/bar'> <img src='javascript:alert()'>";
string clean = NSoupClient.Clean(html, "http://example.com/", Whitelist.BasicWithImages.PreserveRelativeLinks(true));
Assert.AreEqual("<a href=\"/foo\" rel=\"nofollow\">Link</a>\n<img src=\"/bar\" /> \n<img />", clean);
}
public void supplyOutputSettings()
{
// test that one can override the default document output settings
OutputSettings os = new OutputSettings();
os.PrettyPrint(false);
os.SetEscapeMode(Entities.EscapeMode.Extended);
string html = "<div><p>ℬ</p></div>";
string customOut = NSoupClient.Clean(html, "http://foo.com/", Whitelist.Relaxed, os);
string defaultOut = NSoupClient.Clean(html, "http://foo.com/", Whitelist.Relaxed);
Assert.AreNotSame(defaultOut, customOut);
Assert.AreEqual("<div><p>ℬ</p></div>", customOut);
Assert.AreEqual("<div>\n" +
" <p>ℬ</p>\n" +
"</div>", defaultOut);
os.SetEncoding(Encoding.ASCII);
os.SetEscapeMode(Entities.EscapeMode.Base);
String customOut2 = NSoupClient.Clean(html, "http://foo.com/", Whitelist.Relaxed, os);
Assert.AreEqual("<div><p>ℬ</p></div>", customOut2);
}
public void resolvesRelativeLinks()
{
String html = "<a href='/foo'>Link</a><img src='/bar'>";
String clean = NSoupClient.Clean(html, "http://example.com/", Whitelist.BasicWithImages);
Assert.AreEqual("<a href=\"http://example.com/foo\" rel=\"nofollow\">Link</a>\n<img src=\"http://example.com/bar\" />", clean);
}
public void cleansInternationalTextNoEscaping()
{
var outputSettings = new OutputSettings();
outputSettings = (OutputSettings)outputSettings.Clone();
outputSettings.EscapeMode.GetMap().Clear();
var result = NSoupClient.Clean("׀¿ׁ€׀¸׀²׀µׁ‚", "", Whitelist.None, outputSettings);
Assert.AreEqual("׀¿ׁ€׀¸׀²׀µׁ‚", result);
}
public 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 = NSoupClient.Clean(html, whitelist);
Assert.AreEqual("<p class=\"foo\">One</p>", clean);
}
public void handlesCustomProtocols()
{
String html = "<img src='cid:12345' /> <img src='data:gzzt' />";
String dropped = NSoupClient.Clean(html, Whitelist.BasicWithImages);
Assert.AreEqual("<img /> \n<img />", dropped);
String preserved = NSoupClient.Clean(html, Whitelist.BasicWithImages.AddProtocols("img", "src", "cid", "data"));
Assert.AreEqual("<img src=\"cid:12345\" /> \n<img src=\"data:gzzt\" />", preserved);
}
public 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 = NSoupClient.Clean(html, whitelist);
Assert.AreEqual("<p class=\"foo\"><a class=\"qux\">link</a></p>", clean);
}
public void handlesFramesets()
{
String dirty = "<html><head><script></script><noscript></noscript></head><frameset><frame src=\"foo\" /><frame src=\"foo\" /></frameset></html>";
String clean = NSoupClient.Clean(dirty, Whitelist.Basic);
Assert.AreEqual("", clean); // nothing good can come out of that
Document dirtyDoc = NSoupClient.Parse(dirty);
Document cleanDoc = new Cleaner(Whitelist.Basic).Clean(dirtyDoc);
Assert.IsFalse(cleanDoc == null);
Assert.AreEqual(0, cleanDoc.Body.ChildNodes.Count);
}
public void cleansInternationalText()
{
var result = NSoupClient.Clean("׀¿ׁ€׀¸׀²׀µׁ‚", Whitelist.None);
Assert.AreEqual("׀¿ׁ€׀¸׀²׀µׁ‚", result);
}
public void cleansInternationalText()
{
Assert.AreEqual("׀¿ׁ€׀¸׀²׀µׁ‚", NSoupClient.Clean("׀¿ׁ€׀¸׀²׀µׁ‚", Whitelist.None));
}