internal virtual void MaybeSetBaseUri(iText.StyledXmlParser.Jsoup.Nodes.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.SetBaseUri(href);
}
}
public virtual void TestBaidu()
{
// tests <meta http-equiv="Content-Type" content="text/html;charset=gb2312">
FileInfo @in = iText.StyledXmlParser.Jsoup.PortTestUtil.GetFile("/htmltests/baidu-cn-home.html");
Document doc = iText.StyledXmlParser.Jsoup.Jsoup.Parse(@in, null, "http://www.baidu.com");
// http charset is gb2312, but NOT specifying it, to test http-equiv parse
iText.StyledXmlParser.Jsoup.Nodes.Element submit = doc.Select("#su").First();
NUnit.Framework.Assert.AreEqual("百度一下", submit.Attr("value"));
// test from attribute match
submit = doc.Select("input[value=百度一下]").First();
NUnit.Framework.Assert.AreEqual("su", submit.Id());
iText.StyledXmlParser.Jsoup.Nodes.Element newsLink = doc.Select("a:contains(新)").First();
NUnit.Framework.Assert.AreEqual(newsHref, newsLink.AbsUrl("href"));
// check auto-detect from meta
NUnit.Framework.Assert.AreEqual("GB2312", doc.OutputSettings().Charset().DisplayName());
NUnit.Framework.Assert.AreEqual("<title>百度一下，你就知道 </title>", doc.Select("title").OuterHtml());
doc.OutputSettings().Charset("ascii");
NUnit.Framework.Assert.AreEqual("<title>百度一下，你就知道 </title>"
, doc.Select("title").OuterHtml());
}
private bool TestValidProtocol(iText.StyledXmlParser.Jsoup.Nodes.Element el, iText.StyledXmlParser.Jsoup.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.SetValue(value);
}
foreach (Whitelist.Protocol protocol in protocols)
{
String prot = protocol.ToString();
if (prot.Equals("#"))
{
// allows anchor links
if (IsValidAnchor(value))
{
return(true);
}
else
{
continue;
}
}
prot += ":";
if (value.ToLowerInvariant().StartsWith(prot))
{
return(true);
}
}
return(false);
}