public void TestParseHtml()
{
string path = Path.GetFullPath(TestDataSample.GetHtmlPath("mshome.html"));
ParserContext context = new ParserContext(path);
IDomParser parser = (IDomParser)ParserFactory.CreateDom(context);
ToxyDom toxyDom = parser.Parse();
List <ToxyNode> metaNodeList = toxyDom.Root.SelectNodes("//meta");
Assert.AreEqual(7, metaNodeList.Count);
ToxyNode aNode = toxyDom.Root.SingleSelect("//a");
Assert.AreEqual(1, aNode.Attributes.Count);
Assert.AreEqual("href", aNode.Attributes[0].Name);
Assert.AreEqual("http://www.microsoft.com/en/us/default.aspx?redir=true", aNode.Attributes[0].Value);
ToxyNode titleNode = toxyDom.Root.ChildrenNodes[0].ChildrenNodes[0].ChildrenNodes[0];
Assert.AreEqual("title", titleNode.Name);
Assert.AreEqual("Microsoft Corporation", titleNode.ChildrenNodes[0].InnerText);
ToxyNode metaNode = toxyDom.Root.ChildrenNodes[0].ChildrenNodes[0].ChildrenNodes[7];
Assert.AreEqual("meta", metaNode.Name);
Assert.AreEqual(3, metaNode.Attributes.Count);
Assert.AreEqual("name", metaNode.Attributes[0].Name);
Assert.AreEqual("SearchDescription", metaNode.Attributes[0].Value);
Assert.AreEqual("scheme", metaNode.Attributes[2].Name);
Assert.AreEqual(string.Empty, metaNode.Attributes[2].Value);
}
void AppendChildren(ToxyNode tnode, XmlNode ele)
{
if (ele.ChildNodes.Count == 0)
return;
foreach (XmlNode child in ele.ChildNodes)
{
ToxyNode x = ConvertToToxyNode(child);
tnode.ChildrenNodes.Add(x);
AppendChildren(x, child);
}
}
void AppendChildren(ToxyNode tnode, XmlNode ele)
{
if (ele.ChildNodes.Count == 0)
{
return;
}
foreach (XmlNode child in ele.ChildNodes)
{
ToxyNode x = ConvertToToxyNode(child);
tnode.ChildrenNodes.Add(x);
AppendChildren(x, child);
}
}
ToxyNode ConvertToToxyNode(XmlNode ele)
{
ToxyNode tnode = new ToxyNode();
tnode.Name = ele.Name;
if (ele.Name == "#text")
{
tnode.Text = ele.InnerText;
return tnode;
}
if (ele.Attributes != null)
{
foreach (XmlAttribute attr in ele.Attributes)
tnode.Attributes.Add(new ToxyAttribute(attr.Name, attr.Value));
}
return tnode;
}
public ToxyDom Parse()
{
if (!File.Exists(Context.Path))
{
throw new FileNotFoundException("File " + Context.Path + " is not found");
}
XmlDocument doc = new XmlDocument();
doc.Load(Context.Path);
ToxyNode rootNode = ConvertToToxyNode(doc.DocumentElement);
ToxyDom dom = new ToxyDom();
dom.Root = rootNode;
AppendChildren(rootNode, doc.DocumentElement);
return(dom);
}
ToxyNode ConvertToToxyNode(XmlNode ele)
{
ToxyNode tnode = new ToxyNode();
tnode.Name = ele.Name;
if (ele.Name == "#text")
{
tnode.Text = ele.InnerText;
return(tnode);
}
if (ele.Attributes != null)
{
foreach (XmlAttribute attr in ele.Attributes)
{
tnode.Attributes.Add(new ToxyAttribute(attr.Name, attr.Value));
}
}
return(tnode);
}
void AppendTree(TreeNode node, ToxyNode tnode)
{
if (tnode.ChildrenNodes == null || tnode.ChildrenNodes.Count == 0)
{
return;
}
foreach (var child in tnode.ChildrenNodes)
{
TreeNode childNode;
if (child.Name == "#text")
{
childNode = node.Nodes.Add(child.Text);
}
else
{
childNode = node.Nodes.Add(child.NodeString);
}
AppendTree(childNode, child);
}
}
/// <summary>
/// Parse HTML document
/// Note:Context.Path must be absolute path,not relative path
/// </summary>
/// <returns></returns>
public ToxyDom Parse()
{
if (!File.Exists(Context.Path))
{
throw new FileNotFoundException("File " + Context.Path + " is not found");
}
HtmlWeb hw = new HtmlWeb();
HtmlDocument htmlDoc = hw.Load(Context.Path);
HtmlNode docNode = htmlDoc.DocumentNode;
ToxyNode root = ToxyNode.TransformHtmlNodeToToxyNode(docNode);
Queue <KeyValuePair <HtmlNode, ToxyNode> > nodeQueue = new Queue <KeyValuePair <HtmlNode, ToxyNode> >();
nodeQueue.Enqueue(new KeyValuePair <HtmlNode, ToxyNode>(docNode, root));
while (nodeQueue.Count > 0)
{
KeyValuePair <HtmlNode, ToxyNode> pair = nodeQueue.Dequeue();
HtmlNode htmlParentNode = pair.Key;
ToxyNode toxyParentNode = pair.Value;
foreach (HtmlNode htmlChildNode in htmlParentNode.ChildNodes)
{
ToxyNode toxyChildNode = ToxyNode.TransformHtmlNodeToToxyNode(htmlChildNode);
if (htmlChildNode.Name == "#text")
{
toxyChildNode.Text = htmlChildNode.InnerText;
}
toxyParentNode.ChildrenNodes.Add(toxyChildNode);
nodeQueue.Enqueue(new KeyValuePair <HtmlNode, ToxyNode>(htmlChildNode, toxyChildNode));
}
}
return(new ToxyDom()
{
Root = root
});
}
