public void RemoveNodes(RemoveNode[] nodesToRemove)
{
foreach (RemoveNode nodeToRemove in nodesToRemove)
{
removeNodes(doc.DocumentNode, nodeToRemove.nodeName, nodeToRemove.attrib, nodeToRemove.attribVal, true);
}
}
public static void TestRemoveNodeFromUrl()
{
String url = "http://www.bbc.co.uk/news/";
HtmlSanitiser sn = new HtmlSanitiser(url);
RemoveNode rn1 = new RemoveNode("script", null, null);
sn.RemoveNodes(new RemoveNode[] { rn1 });
string htmlout = sn.Html;
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(htmlout);
bool foundNode = false;
if (doc.DocumentNode.Name == "script")
foundNode = true;
if(doc.DocumentNode.HasChildNodes)
{
foreach(HtmlNode cNode in doc.DocumentNode.ChildNodes)
if (cNode.Name == "script")
{
foundNode = true;
}
}
if (htmlout.Contains("<script"))
{
foundNode = true;
}
Assert.AreEqual(foundNode, false);
}
private void Button_Click(object sender, RoutedEventArgs e)
{
string path = txtOutputDir.Text;
List<RemoveNode> removeNodes = new List<RemoveNode>();
RemoveNode rn = new RemoveNode("script", null, null);
removeNodes.Add(rn);
rn = new RemoveNode("div", "class", "portal");
removeNodes.Add(rn);
Downloader dn = new Downloader(path,txtUrl.Text, removeNodes, txtOutputFile.Text + ".html");
dn.Download();
string str = "\"" + path + dn.FileName + "\"";
ProcessStartInfo psi = new ProcessStartInfo(kindleGenExe, str);
psi.UseShellExecute = false;
Process p = new Process();
p.StartInfo = psi;
p.Start();
p.WaitForExit();
}
