public static string ToString(string html, HtmlNode node)
{
StringBuilder builder = new StringBuilder(128);
if (node.NodeType == HtmlNodeType.Element)
{
string tagName = HtmlTags.GetTagName(node.TagId);
builder.Append("<");
builder.Append(tagName);
if (node.Attributes != null)
{
foreach (AttributeItem item in node.Attributes)
{
if (item.Value.Contains("\""))
{
builder.AppendFormat(" {0}='{1}'", item.Name, item.Value);
}
else
{
builder.AppendFormat(" {0}=\"{1}\"", item.Name, item.Value);
}
}
}
builder.Append(">");
}
else if (node.NodeType == HtmlNodeType.EndTag)
{
string tagName = HtmlTags.GetTagName(node.TagId);
builder.AppendFormat("</{0}>", tagName);
}
else if (node.NodeType == HtmlNodeType.Text)
{
string text = HtmlUtility.NormalizeText(html, node.Index, node.Length);
if (text != null)
{
builder.Append(text);
}
}
else if (node.NodeType == HtmlNodeType.Comment)
{
string text = html.Substring(node.Index, node.Length);
builder.Append(text);
}
else
{
Debug.Assert(false);
}
return(builder.ToString());
}
private string GetNodeStack(int startTag, int endTag)
{
string nodeStack = "";
if (endTag < parsedHtml.Nodes.Count - 1)
{
endTag++;
}
if (startTag > 0)
{
startTag--;
}
while (startTag <= endTag)
{
HtmlNode node = GetNode(startTag);
switch (node.NodeType)
{
case HtmlNodeType.Element:
if (node.GetAttributeValue(HtmlAttributeId.Class) != null)
{
nodeStack += "<" + HtmlTags.GetTagName(node.TagId) + " class=" + node.GetAttributeValue(HtmlAttributeId.Class) + ">";
}
else
{
nodeStack += "<" + HtmlTags.GetTagName(node.TagId) + ">";
}
break;
case HtmlNodeType.EndTag:
nodeStack += "</" + HtmlTags.GetTagName(node.TagId) + ">";
break;
case HtmlNodeType.Text:
nodeStack += "_";
break;
}
startTag++;
}
return(nodeStack);
}
public static void RootRightpath(string html, List <HtmlNode> nodes)
{
string L1_path = @"D:\265.com\L1\RightL1Categories.txt";
StreamWriter sw = new StreamWriter(L1_path);
int start = 0;
string url = "", category = "";
for (int i = 0; i < nodes.Count; i++)
{
if (nodes[i].NodeType == HtmlNodeType.Element)
{
string tagName = HtmlTags.GetTagName(nodes[i].TagId);
if (tagName.ToLower() == "li" && nodes[i].Attributes != null && nodes[i].Attributes.Count > 0)
{
for (int k = 0; k < nodes[i].Attributes.Count; k++)
{
if (nodes[i].Attributes[k].Name == "class" && nodes[i].Attributes[k].Value == "title")
{
start = 1;
continue;
}
}
}
if (start == 1 && tagName.ToLower() == "a" && nodes[i].Attributes != null && nodes[i].Attributes.Count > 0)
{
string site = @"http://www.265.com";
for (int k = 0; k < nodes[i].Attributes.Count; k++)
{
if (nodes[i].Attributes[k].Name == "href")
{
url = site + nodes[i].Attributes[k].Value;
}
}
}
}
else if (nodes[i].NodeType == HtmlNodeType.Text)
{
if (start == 1)
{
category = html.Substring(nodes[i].Index, nodes[i].Length);
//Console.WriteLine(category);
}
}
else if (nodes[i].NodeType == HtmlNodeType.EndTag)
{
string tagName = HtmlTags.GetTagName(nodes[i].TagId);
if (tagName.ToLower() == "li" && start > 0)
{
start = 0;
sw.WriteLine("{0}\t{1}", category, url);
sw.Flush();
}
}
else
{
continue;
}
}
sw.Close();
}
public static void Subpath(string html, List <HtmlNode> nodes, string filepath)
{
StreamWriter sw = new StreamWriter(filepath);
int start = 0;
string category = "", url = "";
for (int i = 0; i < nodes.Count; i++)
{
if (nodes[i].NodeType == HtmlNodeType.Element)
{
string tagName = HtmlTags.GetTagName(nodes[i].TagId);
if (tagName.ToLower() == "div" && nodes[i].Attributes != null && nodes[i].Attributes.Count > 0)
{
for (int k = 0; k < nodes[i].Attributes.Count; k++)
{
if (nodes[i].Attributes[k].Name == "class" && nodes[i].Attributes[k].Value == "titleCS")
{
start = 1;
}
}
}
if (tagName.ToLower() == "ul" && nodes[i].Attributes != null && nodes[i].Attributes.Count > 0)
{
for (int k = 0; k < nodes[i].Attributes.Count; k++)
{
if (nodes[i].Attributes[k].Name == "class" && nodes[i].Attributes[k].Value == "listUrl")
{
start = 2;
}
}
}
if (tagName.ToLower() == "a" && nodes[i].Attributes != null && nodes[i].Attributes.Count > 0 && start == 2)
{
for (int k = 0; k < nodes[i].Attributes.Count; k++)
{
if (nodes[i].Attributes[k].Name == "href")
{
url = nodes[i].Attributes[k].Value;
}
}
}
}
else if (nodes[i].NodeType == HtmlNodeType.Text)
{
if (start == 1)
{
category = html.Substring(nodes[i].Index, nodes[i].Length).ToString();
start = 2;
}
}
else if (nodes[i].NodeType == HtmlNodeType.EndTag)
{
string tagName = HtmlTags.GetTagName(nodes[i].TagId);
if (tagName.ToLower() == "li" && start == 2)
{
sw.WriteLine("{0}\t{1}", category, url);
sw.Flush();
}
if (tagName.ToLower() == "ul" && start == 2)
{
start = 0;
}
}
}
sw.Close();
}
