public void LoadRaw(string rawContent)
{
var doc = new HtmlDocument();
doc.LoadHtml(rawContent);
HtmlNodes.Add(doc.DocumentNode);
}
public void LoadFromFile(string filePath)
{
var doc = new HtmlDocument();
doc.Load(filePath);
HtmlNodes.Add(doc.DocumentNode);
}
public IList <string> ToStringList()
{
if (HtmlNodes == null || HtmlNodes.Count == 0)
{
return(new List <string>());
}
return(HtmlNodes.Select(x => x.OuterHtml).ToList());
}
public void RecursionAllChildNodes(Func <HtmlNodes, bool> where, ref HtmlNodes result)
{
foreach (var nd in Child)
{
if (result != null)
{
return;
}
if (where == null || where (nd))
{
result = nd;
return;
}
else
{
nd.RecursionAllChildNodes(where, ref result);
}
}
}
/// <summary>
/// 获得该html页面的指定标签
/// </summary>
/// <param name="html"></param>
/// <param name="selectXPath">选择标签开始路径，null则从根目录开始</param>
/// <param name="where">判断标签是否符合，返回是则符合</param>
/// <returns></returns>
public static HtmlNodes GetHtmlNodesWhere(this string html, string selectXPath = null, Func <HtmlNodes, bool> where = null)
{
var doc = new HtmlDocument();
try
{
doc.LoadHtml(html);
var node = selectXPath == null ? doc.DocumentNode : doc.DocumentNode.SelectSingleNode(selectXPath);
var hd = new HtmlNodes(node);
if (where == null || where (hd))
{
return(hd);
}
HtmlNodes result = null;
hd.RecursionAllChildNodes(where, ref result);
return(result);
}
catch (Exception ex)
{
Console.WriteLine(ex.Message);
}
return(null);
}
public void Clear()
{
HtmlNodes.Clear();
}
