public void HandlesBaseUri()
{
Tag tag = Tag.ValueOf("a");
Attributes attribs = new Attributes();
attribs["relHref"] = "/foo";
attribs["absHref"] = "http://bar/qux";
Element noBase = new Element(tag, "", attribs);
Assert.AreEqual("", noBase.AbsUrl("relHref")); // with no base, should NOT fallback to href attrib, whatever it is
Assert.AreEqual("http://bar/qux", noBase.AbsUrl("absHref")); // no base but valid attrib, return attrib
Element withBase = new Element(tag, "http://foo/", attribs);
Assert.AreEqual("http://foo/foo", withBase.AbsUrl("relHref")); // construct abs from base + rel
Assert.AreEqual("http://bar/qux", withBase.AbsUrl("absHref")); // href is abs, so returns that
Assert.AreEqual("", withBase.AbsUrl("noval"));
Element dodgyBase = new Element(tag, "wtf://no-such-protocol/", attribs);
Assert.AreEqual("http://bar/qux", dodgyBase.AbsUrl("absHref")); // base fails, but href good, so get that
Assert.AreEqual("", dodgyBase.AbsUrl("relHref")); // base fails, only rel href, so return nothing
}
private bool TestValidProtocol(Element el, Nodes.Attribute attr, ICollection<Whitelist.Protocol> protocols)
{
// try to resolve relative urls to abs, and optionally update the attribute so output html has abs.
// rels without a baseuri get removed
string value = el.AbsUrl(attr.Key);
if (value.Length == 0)
{
value = attr.Value;
}
// if it could not be made abs, run as-is to allow custom unknown protocols
if (!preserveRelativeLinks)
{
attr.Value = value;
}
foreach (Whitelist.Protocol protocol in protocols)
{
string prot = protocol.ToString() + ":";
if (value.ToLower().StartsWith(prot, StringComparison.Ordinal))
{
return true;
}
}
return false;
}
internal void MaybeSetBaseUri(Element @base)
{
if (baseUriSetFromDoc)
{
// only listen to the first <base href> in parse
return;
}
string href = @base.AbsUrl("href");
if (href.Length != 0)
{
// ignore <base target> etc
baseUri = href;
baseUriSetFromDoc = true;
doc.BaseUri = href;
}
}
