C# HtmlDocument.DetectEncodingAndLoad方法代码示例

 //        
 //                <add key="AppNamePath" value="//div[@class='details']//a[@class='title']" />
 //        <add key="PricePath" value="//div[@class='reason-set']//button[@class='price buy']/span" />
 //        <add key="BundleIDPath" value="//div[@class='details']//a[@class='title']" />
 //        <add key="BundleIDAttr" value="href" />
 //        <add key="BundleIDRegex" value="(?:id=([^&amp;]+))" />
 //        <add key="AppIDPath" value="//div[@class='details']//a[@class='title']" />
 //        <add key="AppIDAttr" value="href" />
 //        <add key="AppIDRegex" value="(?:id=([^&amp;]+))" />
 public static void CheckHtml()
 {
     HtmlDocument doc = new HtmlDocument();
     string path = @"C:\Users\Pat\Documents\SharpDevelop Projects\IdeaTests\googletest.htm";
     doc.DetectEncodingAndLoad(path);
     HtmlNodeCollection containers =  doc.DocumentNode.SelectNodes(@"//*[@class='cluster-container']/*/div[@class='card-list']");
     Console.WriteLine(containers.Count );
     foreach (var node in containers) {
         HtmlNodeCollection appNames = node.SelectNodes(@"//div[@class='details']//a[@class='title']");
         Console.WriteLine("appname: " + appNames.Count);
     }
 }

		private HtmlDocument LoadUrl(Uri uri, string method)
		{
			HtmlDocument doc = new HtmlDocument();
			doc.OptionAutoCloseOnEnd = false;
			doc.OptionFixNestedTags = true;
			_statusCode = Get(uri, method, null, doc);
			if (_statusCode == HttpStatusCode.NotModified)
			{
				// read cached encoding
				doc.DetectEncodingAndLoad(GetCachePath(uri));
			}
			return doc;
		}

		/// <summary>
		/// Loads an HTML document from an Internet resource.
		/// </summary>
		/// <param name="url">The requested URL, such as "http://Myserver/Mypath/Myfile.asp".</param>
		/// <param name="method">The HTTP method used to open the connection, such as GET, POST, PUT, or PROPFIND.</param>
		/// <returns>A new HTML document.</returns>
		public HtmlDocument Load(string url, string method)
		{
			Uri uri = new Uri(url);
			HtmlDocument doc;
			if ((uri.Scheme == Uri.UriSchemeHttps) ||
				(uri.Scheme == Uri.UriSchemeHttp))
			{
				doc = LoadUrl(uri, method);
			}
			else
			{

				if (uri.Scheme == Uri.UriSchemeFile)
				{
					doc = new HtmlDocument();
					doc.OptionAutoCloseOnEnd = false;
					doc.OptionAutoCloseOnEnd = true;
					doc.DetectEncodingAndLoad(url, _autoDetectEncoding);
				}
				else
				{
					throw new HtmlWebException("Unsupported uri scheme: '" + uri.Scheme + "'.");
				}
			}
			if (PreHandleDocument != null)
			{
				PreHandleDocument(doc);
			}
			return doc;
		}

		private HtmlDocument LoadUrl(Uri uri, string method, IWebProxy proxy, ICredentials creds)
		{
		    var doc = new HtmlDocument
		    {
		        OptionAutoCloseOnEnd = false,
		        OptionFixNestedTags = true
		    };
		    StatusCode = Get(uri, method, null, doc, proxy, creds);
			if (StatusCode == HttpStatusCode.NotModified)
			{
				// read cached encoding
				doc.DetectEncodingAndLoad(GetCachePath(uri));
			}
			return doc;
		}

		/// <summary>
		/// Loads an HTML document from an Internet resource.
		/// </summary>
		/// <param name="url">The requested URL, such as "http://Myserver/Mypath/Myfile.asp".</param>
		/// <param name="method">The HTTP method used to open the connection, such as GET, POST, PUT, or PROPFIND.</param>
		/// <param name="proxy">Proxy to use with this request</param>
		/// <param name="credentials">Credentials to use when authenticating</param>
		/// <returns>A new HTML document.</returns>
		public HtmlDocument Load(string url, string method, IWebProxy proxy, NetworkCredential credentials)
		{
			var uri = new Uri(url);
			HtmlDocument doc;
			if ((uri.Scheme.ToLower() == "https") ||
				(uri.Scheme.ToLower() == "http"))
			{
				doc = LoadUrl(uri, method, proxy, credentials);
			}
			else
			{
				if (uri.Scheme.ToLower() == "file")
				{
				    doc = new HtmlDocument
				    {
				        OptionAutoCloseOnEnd = false
				    };
				    doc.OptionAutoCloseOnEnd = true;
					doc.DetectEncodingAndLoad(url, AutoDetectEncoding);
				}
				else
				{
					throw new HtmlWebException("Unsupported uri scheme: '" + uri.Scheme + "'.");
				}
			}
		    PreHandleDocument?.Invoke(doc);
		    return doc;
		}

 private HtmlDocument LoadUrl(Uri uri, string method, WebProxy proxy, NetworkCredential creds)
 {
     HtmlDocument doc = new HtmlDocument();
     doc.OptionAutoCloseOnEnd = false;
     doc.OptionFixNestedTags = true;
     if (OptionMaxNestedChildNodes > 0)
         doc.OptionMaxNestedChildNodes = OptionMaxNestedChildNodes;
     _statusCode = Get(uri, method, null, doc, proxy, creds);
     if (_statusCode == HttpStatusCode.NotModified)
     {
         // read cached encoding
         doc.DetectEncodingAndLoad(GetCachePath(uri));
     }
     return doc;
 }

 /// <summary>
 /// Loads an HTML document from an Internet resource.
 /// </summary>
 /// <param name="url">The requested URL, such as "http://Myserver/Mypath/Myfile.asp".</param>
 /// <param name="method">The HTTP method used to open the connection, such as GET, POST, PUT, or PROPFIND.</param>
 /// <param name="proxy">Proxy to use with this request</param>
 /// <param name="credentials">Credentials to use when authenticating</param>
 /// <returns>A new HTML document.</returns>
 public HtmlDocument Load(string url, string method, WebProxy proxy, NetworkCredential credentials)
 {
     Uri uri = new Uri(url);
     HtmlDocument doc;
     if ((uri.Scheme == Uri.UriSchemeHttps) ||
         (uri.Scheme == Uri.UriSchemeHttp))
     {
         doc = LoadUrl(uri, method, proxy, credentials);
     }
     else
     {
         if (uri.Scheme == Uri.UriSchemeFile)
         {
             doc = new HtmlDocument();
             doc.OptionAutoCloseOnEnd = false;
             doc.OptionAutoCloseOnEnd = true;
             if (OptionMaxNestedChildNodes > 0)
                 doc.OptionMaxNestedChildNodes = OptionMaxNestedChildNodes;
             doc.DetectEncodingAndLoad(url, _autoDetectEncoding);
         }
         else
         {
             throw new HtmlWebException("Unsupported uri scheme: '" + uri.Scheme + "'.");
         }
     }
     if (PreHandleDocument != null)
     {
         PreHandleDocument(doc);
     }
     return doc;
 }

        public static List<TextureInfo> Read(string fileName)
        {
            HtmlAgilityPack.HtmlDocument htmlDoc = new HtmlAgilityPack.HtmlDocument();
            htmlDoc.DetectEncodingAndLoad(fileName, true);
            //htmlDoc.
            HtmlNode Plist = htmlDoc.DocumentNode.Element("plist");
            HtmlNodeCollection dicNodes = Plist.ChildNodes["dict"].ChildNodes[3].ChildNodes;

            int count = dicNodes.Count / 2;
            int i = 0;

            var texInfoes = new List<TextureInfo>();

            string key = null;
            //var value = null;
            var dic = new Dictionary<string, Dictionary<string, string>>();
            foreach (HtmlNode node in dicNodes)
            {
                if (node.Name[0] == '#')
                    continue;

                if (node.Name[0] == 'k')
                {
                    key = node.InnerHtml;
                }
                else
                {
                    var value = new Dictionary<string, string>();
                    string subKey = null;
                    string subValue = null;
                    foreach (HtmlNode infoNode in node.ChildNodes)
                    {
                        if (infoNode.Name[0] == '#')
                            continue;

                        if (infoNode.Name[0] == 'k')
                        {
                            subKey = infoNode.InnerHtml;
                        }
                        else
                        {
                            subValue = infoNode.InnerHtml == string.Empty ? infoNode.Name : infoNode.InnerHtml;

                            value.Add(subKey, subValue);
                            subKey = null;
                            subValue = null;
                        }
                    }
                    dic.Add(key, value);
                    key = null;
                    value = null;
                }

            }

            foreach (string item in dic.Keys)
            {
                TextureInfo texInfo = new TextureInfo();

                texInfo.fileName = item;

                var value = dic[item];

                //获取FrameRect
                texInfo.texRect = GetFrameRect(value[plistKeyInfo.FrameRect]);

                //计算并存储offsetRect
                texInfo.offset = new point(0, 0);
                GetTowNumber(value[plistKeyInfo.Offset], out texInfo.offset.x, out texInfo.offset.y);

                //获取SourceSize
                int width;
                int height;
                GetTowNumber(value[plistKeyInfo.SourceSize], out width, out height);

                texInfo.sourceSize = new size(width, height);
                // 				Console.WriteLine("{0}: {1}, {2} ; {3}, {4}",
                // 					tempinfo.fileName,
                // 					tempinfo.texRect.X, tempinfo.texRect.Y,
                // 					tempinfo.texRect.Width, tempinfo.texRect.Height
                // 					);

                //计算实际的innerRect的值
                texInfo.innerRect = new rect(0, 0, 0, 0);
                texInfo.innerRect.Width = texInfo.texRect.Width;
                texInfo.innerRect.Height = texInfo.texRect.Height;
                texInfo.innerRect.X = (texInfo.sourceSize.Width - texInfo.texRect.Width) / 2 + texInfo.offset.x;
                texInfo.innerRect.Y = (texInfo.sourceSize.Height - texInfo.texRect.Height) / 2 - texInfo.offset.y;

                texInfoes.Add(texInfo);
            }

            return texInfoes;
        }

 private HtmlDocument LoadUrl(Uri uri, string method, WebProxy proxy, NetworkCredential creds)
 {
   HtmlDocument doc = new HtmlDocument();
   doc.OptionAutoCloseOnEnd = false;
   doc.OptionFixNestedTags = true;
   this._statusCode = this.Get(uri, method, (string) null, doc, (IWebProxy) proxy, (ICredentials) creds);
   if (this._statusCode == HttpStatusCode.NotModified)
     doc.DetectEncodingAndLoad(this.GetCachePath(uri));
   return doc;
 }

 /// <summary>
 /// Loads an HTML document from an Internet resource.
 /// 
 /// </summary>
 /// <param name="url">The requested URL, such as "http://Myserver/Mypath/Myfile.asp".</param><param name="method">The HTTP method used to open the connection, such as GET, POST, PUT, or PROPFIND.</param><param name="proxy">Proxy to use with this request</param><param name="credentials">Credentials to use when authenticating</param>
 /// <returns>
 /// A new HTML document.
 /// </returns>
 public HtmlDocument Load(string url, string method, WebProxy proxy, NetworkCredential credentials)
 {
   Uri uri = new Uri(url);
   HtmlDocument document;
   if (uri.Scheme == Uri.UriSchemeHttps || uri.Scheme == Uri.UriSchemeHttp)
   {
     document = this.LoadUrl(uri, method, proxy, credentials);
   }
   else
   {
     if (!(uri.Scheme == Uri.UriSchemeFile))
       throw new HtmlWebException("Unsupported uri scheme: '" + uri.Scheme + "'.");
     document = new HtmlDocument();
     document.OptionAutoCloseOnEnd = false;
     document.OptionAutoCloseOnEnd = true;
     document.DetectEncodingAndLoad(url, this._autoDetectEncoding);
   }
   if (this.PreHandleDocument != null)
     this.PreHandleDocument(document);
   return document;
 }