Beispiel #1
0
 public void Load()
 {
     if (IsLoaded)
     {
         return;
     }
     if (Url.IsUrl())
     {
         Domain      = GetDomain();
         web         = new HtmlWeb();
         html        = web.Load(Url);
         HtmlContent = html.DocumentNode.OuterHtml;
         TextContent = HtmlContent.HtmlToText();
         var _allLinks = HtmlContent.GetUrls();
         GetHtmlTables();
         //BuildKeywords();
         CssIds     = HtmlContent.GetAllIds().ToList();
         CssClasses = HtmlContent.GetAllClasses()
                      .ToList();
         Nodes = html.GetNodes().Where(t => t.IsCssClass());
         //.Where(t=>t.OuterHtml.Contains("class"));
         foreach (string _link in _allLinks)
         {
             if (_link.StartsWith("/"))
             {
                 string _url = Url.GetBefore(Domain) + Domain + _link;
                 Links.Add(_url);
             }
             else
             {
                 Links.Add(_link);
             }
         }
         if (Links != null)
         {
             InternalLinks = Links.Where(t => t.StartsWith(Url.GetBefore(Domain) + Domain)).ToList();
         }
         GetImages("jpg");
         LoadTitle();
         IsLoaded = true;
     }
 }