internal static string GetAttributeValue(this Common.Html.Node node, string key, string defaultValue)
{
foreach (var attr in node.Attr)
{
if (attr.Namespace == "" && attr.Key == key)
{
return(attr.Val);
}
}
return(defaultValue);
}
internal static void SetAttributeValue(this Common.Html.Node node, string key, string value)
{
foreach (var attr in node.Attr)
{
if (attr.Namespace == "" && attr.Key == key)
{
attr.Val = value;
return;
}
}
node.Attr.Add(new Common.Html.Attribute
{
Namespace = "",
Key = key,
Val = value,
});
}
internal static IEnumerable <Common.Html.Node> Descendants(this Common.Html.Node root, string tagName)
{
if (root.Type != Common.Html.NodeType.Element)
{
yield break;
}
if (root.Data == tagName)
{
yield return(root);
}
for (var c = root.FirstChild; c != null; c = c.NextSibling)
{
foreach (var d in c.Descendants(tagName))
{
yield return(d);
}
}
}
internal static string GetInnerText(this Common.Html.Node node)
{
switch (node.Type)
{
case Common.Html.NodeType.Text:
return(node.Data);
case Common.Html.NodeType.Element:
var buf = new StringBuilder();
for (var c = node.FirstChild; c != null; c = c.NextSibling)
{
buf.Append(c.GetInnerText());
}
return(buf.ToString());
default:
return("");
}
}
private static IEnumerable <string> GetLinkUrls(Common.Html.Node node)
{
if (node.Type != Common.Html.NodeType.Element)
{
yield break;
}
var href = node.GetAttributeValue("href", null);
if (href != null)
{
yield return(href);
}
for (var c = node.FirstChild; c != null; c = c.NextSibling)
{
foreach (var u in GetLinkUrls(c))
{
yield return(u);
}
}
}
public static string Clean(string html)
{
html = Cleaner.Preprocess(Config, html);
var nodes = Cleaner.Parse(html);
foreach (var img in nodes.Descendants("img"))
{
img.Data = "a";
img.DataAtom = atom.A;
var alt = new Common.Html.Node
{
Type = Common.Html.NodeType.Text,
Data = "[image]",
};
img.AppendChild(alt);
foreach (var attr in img.Attr)
{
if (attr.Namespace != "")
{
continue;
}
if (attr.Key == "src")
{
attr.Key = "href";
}
if (attr.Key == "alt")
{
alt.Data = attr.Val;
}
}
}
foreach (var a in nodes.Descendants("a"))
{
a.SetAttributeValue("rel", "nofollow");
}
nodes = Cleaner.CleanNodes(Config, nodes);
return(Cleaner.Render(nodes));
}