C# SgmlReader.Read方法代碼示例

        // Creates XmlDocument from html content and return it with rootitem "<root>".
        public static XmlDocument ParseHtml(string sContent)
        {
            StringReader sr = new StringReader("<root>" + sContent + "</root>");
            SgmlReader reader = new SgmlReader();
            reader.WhitespaceHandling = WhitespaceHandling.All;
            reader.CaseFolding = Sgml.CaseFolding.ToLower;
            reader.InputStream = sr;

            StringWriter sw = new StringWriter();
            XmlTextWriter w = new XmlTextWriter(sw);
            w.Formatting = Formatting.Indented;
            w.WriteStartDocument();
            reader.Read();
            while (!reader.EOF)
            {
                w.WriteNode(reader, true);
            }
            w.Flush();
            w.Close();

            sw.Flush();

            // create document
            XmlDocument doc = new XmlDocument();
            doc.PreserveWhitespace = true;
            doc.XmlResolver = null;
            doc.LoadXml(sw.ToString());

            reader.Close();

            return doc;
        }

 public static string GetWellFormedHTML(string html, string xpathNavPath)
 {
     // StreamReader sReader = null;
     StringWriter sw = null;
     SgmlReader reader = null;
     XmlTextWriter writer = null;
     try
     {
         //  if (uri == String.Empty) uri = "http://www.XMLforASP.NET";
         // HttpWebRequest req = (HttpWebRequest)WebRequest.Create(uri);
         //  HttpWebResponse res = (HttpWebResponse)req.GetResponse();
         //  sReader = new StreamReader(res.GetResponseStream());
         reader = new SgmlReader();
         reader.DocType = "HTML";
         reader.InputStream = new StringReader(html);
         sw = new StringWriter();
         writer = new XmlTextWriter(sw);
         writer.Formatting = Formatting.Indented;
         //writer.WriteStartElement("Test");
         while (reader.Read())
         {
             if (reader.NodeType != XmlNodeType.Whitespace)
             {
                 writer.WriteNode(reader, true);
             }
         }
         //writer.WriteEndElement();
         if (xpathNavPath == null)
         {
             string sr = sw.ToString();
             sr = sr.Replace("\r", "\n");
             sr = sr.Replace("\n\n", "\n");
             return sr;
         }
         else
         { //Filter out nodes from HTML
             StringBuilder sb = new StringBuilder();
             XPathDocument doc = new XPathDocument(new StringReader(sw.ToString()));
             XPathNavigator nav = doc.CreateNavigator();
             XPathNodeIterator nodes = nav.Select(xpathNavPath);
             while (nodes.MoveNext())
             {
                 sb.Append(nodes.Current.Value + "\n");
             }
             string sr = sb.ToString();
             sr = sr.Replace("\r", "\n");
             sr = sr.Replace("\n\n", "\n");
             return sr;
         }
     }
     catch (Exception exp)
     {
         writer.Close();
         reader.Close();
         sw.Close();
         // sReader.Close();
         return exp.Message;
     }
 }

        internal static ImageInfo[] FindImgs(
            string htmlCode)
        {
            var r =
                new SgmlReader
                    {
                        DocType = @"HTML",
                        InputStream = new StringReader(htmlCode)
                    };
            var al = new List<ImageInfo>();

            //find <img src=""
            while (r.Read())
            {
                if (r.NodeType == XmlNodeType.Element)
                {
                    if (string.Compare(r.Name, @"img", StringComparison.OrdinalIgnoreCase) == 0)
                    {
                        if (r.HasAttributes)
                        {
                            var ii = new ImageInfo();

                            while (r.MoveToNextAttribute())
                            {
                                switch (r.Name.ToLowerInvariant())
                                {
                                    case @"src":
                                        ii.Source = r.Value;
                                        break;
                                    case @"width":
                                        ii.Width = ConvertHelper.ToInt32(r.Value);
                                        break;
                                    case @"height":
                                        ii.Height = ConvertHelper.ToInt32(r.Value);
                                        break;
                                }
                            }

                            // --

                            if (!string.IsNullOrEmpty(ii.Source))
                            {
                                al.Add(ii);
                            }
                        }
                    }
                }
            }

            return al.ToArray();
        }

        void Debug(SgmlReader sr)
        {
            NodeTypeFlags[] AllowedContentMap = new NodeTypeFlags[19] {
                                                                          NodeTypeFlags.None, // none
                                                                          NodeTypeFlags.Element | NodeTypeFlags.Attribute | NodeTypeFlags.Text | NodeTypeFlags.CDATA | NodeTypeFlags.EntityReference | NodeTypeFlags.ProcessingInstruction | NodeTypeFlags.Comment | NodeTypeFlags.Whitespace | NodeTypeFlags.SignificantWhitespace | NodeTypeFlags.EndElement, // element
                                                                          NodeTypeFlags.Text | NodeTypeFlags.EntityReference, // attribute
                                                                          NodeTypeFlags.None, // text
                                                                          NodeTypeFlags.None, // cdata
                                                                          NodeTypeFlags.None, // entity reference
                                                                          NodeTypeFlags.None, // entity
                                                                          NodeTypeFlags.None, // processing instruction
                                                                          NodeTypeFlags.None, // comment
                                                                          NodeTypeFlags.Comment | NodeTypeFlags.DocumentType | NodeTypeFlags.Element | NodeTypeFlags.EndElement | NodeTypeFlags.ProcessingInstruction | NodeTypeFlags.Whitespace | NodeTypeFlags.SignificantWhitespace | NodeTypeFlags.XmlDeclaration, // document
                                                                          NodeTypeFlags.None, // document type
                                                                          NodeTypeFlags.None, // document fragment (not expecting these)
                                                                          NodeTypeFlags.None, // notation
                                                                          NodeTypeFlags.None, // whitespace
                                                                          NodeTypeFlags.None, // signification whitespace
                                                                          NodeTypeFlags.None, // end element
                                                                          NodeTypeFlags.None, // end entity
                                                                          NodeTypeFlags.None, // filler
                                                                          NodeTypeFlags.None, // xml declaration.
            };

            Stack s = new Stack();

            while (sr.Read()) {
                if (sr.NodeType == XmlNodeType.EndElement) {
                    s.Pop();
                }
                if (s.Count > 0) {
                    XmlNodeType pt = (XmlNodeType)s.Peek();
                    NodeTypeFlags p = NodeTypeMap[(int)pt];
                    NodeTypeFlags f = NodeTypeMap[(int)sr.NodeType];
                    if ((AllowedContentMap[(int)pt]& f) != f) {
                        Console.WriteLine("Invalid content!!");
                    }
                }
                if (s.Count != sr.Depth-1) {
                    Console.WriteLine("Depth is wrong!");
                }
                if ( (sr.NodeType == XmlNodeType.Element && !sr.IsEmptyElement) ||
                    sr.NodeType == XmlNodeType.Document) {
                    s.Push(sr.NodeType);
                }

                for (int i = 1; i < sr.Depth; i++)
                    Console.Write("  ");
                Console.Write(sr.NodeType.ToString() + " " + sr.Name);
                if (sr.NodeType == XmlNodeType.Element && sr.AttributeCount > 0) {
                    sr.MoveToAttribute(0);
                    Console.Write(" (" + sr.Name+"="+sr.Value + ")");
                    sr.MoveToElement();
                }
                if (sr.Value != null) {
                    Console.Write(" " + sr.Value.Replace("\n"," ").Replace("\r",""));
                }
                Console.WriteLine();
            }
        }

        /// <summary>
        /// Converts the specified html into XHTML compliant text.
        /// </summary>
        /// <param name="reader">sgml reader.</param>
        /// <param name="html">html to convert.</param>
        /// <param name="converter">The converter.</param>
        /// <returns></returns>
        /// ///
        private static string ConvertHtmlToXHtml(SgmlReader reader, string html, Converter<string, string> converter)
        {
            reader.DocType = "html";
            reader.WhitespaceHandling = WhitespaceHandling.All;
            // Hack to fix SF bug #1678030
            html = RemoveNewLineBeforeCDATA(html);
            reader.InputStream = new StringReader("<html>" + html + "</html>");
            reader.CaseFolding = CaseFolding.ToLower;
            StringWriter writer = new StringWriter();
            XmlWriter xmlWriter = null;
            try
            {
                xmlWriter = new XmlTextWriter(writer);

                bool insideAnchor = false;
                bool skipRead = false;
                while ((skipRead || reader.Read()) && !reader.EOF)
                {
                    skipRead = false;
                    switch(reader.NodeType)
                    {
                        case XmlNodeType.Element:
                            //Special case for anchor tags for the time being.
                            //We need some way to communicate which elements the current node is nested within
                            if (reader.IsEmptyElement)
                            {
                                xmlWriter.WriteStartElement(reader.LocalName);
                                xmlWriter.WriteAttributes(reader, true);
                                if (reader.LocalName == "a" || reader.LocalName == "script")
                                    xmlWriter.WriteFullEndElement();
                                else
                                    xmlWriter.WriteEndElement();
                            }
                            else
                            {
                                if (reader.LocalName == "a")
                                    insideAnchor = true;
                                xmlWriter.WriteStartElement(reader.LocalName);
                                xmlWriter.WriteAttributes(reader, true);
                            }
                            break;

                        case XmlNodeType.Text:
                            string text = reader.Value;

                            if (converter != null && !insideAnchor)
                                xmlWriter.WriteRaw(converter(text));
                            else
                                xmlWriter.WriteString(text);
                            break;

                        case XmlNodeType.EndElement:
                            if (reader.LocalName == "a")
                                insideAnchor = false;

                            if (reader.LocalName == "a" || reader.LocalName == "script")
                                xmlWriter.WriteFullEndElement();
                            else
                                xmlWriter.WriteEndElement();
                            break;

                        default:
                            xmlWriter.WriteNode(reader, true);
                            skipRead = true;
                            break;
                    }
                }
            }
            finally
            {
                if (xmlWriter != null)
                {
                    xmlWriter.Close();
                }
            }

            string xml = writer.ToString();
            return xml.Substring("<html>".Length, xml.Length - "<html></html>".Length);
        }

        public static IEnumerable<string> GetAttributeValues(this string html, string tagName, string attributeName)
        {
            var reader = new SgmlReader
            {
                DocType = "html",
                WhitespaceHandling = WhitespaceHandling.All,
                InputStream = new StringReader(string.Format("<html>{0}</html>", html))
            };

            while (reader.Read() && !reader.EOF)
            {
                if (reader.NodeType == XmlNodeType.Element && reader.LocalName == tagName)
                {
                    yield return reader.GetAttribute(attributeName);
                }
            }
        }

		private static string ConvertCommentToMarkdown(string body)
		{
			var sb = new StringBuilder();

			var sgmlReader = new SgmlReader
				{
					InputStream = new StringReader(body),
					DocType = "HTML",
					WhitespaceHandling = WhitespaceHandling.Significant,
					CaseFolding = CaseFolding.ToLower
				};

			bool outputEndElement = false;
			int indentLevel = 0;
			while (sgmlReader.Read())
			{
				switch (sgmlReader.NodeType)
				{
					case XmlNodeType.Text:
						if (indentLevel > 0)
							sb.Append("\t");
						sb.AppendLine(sgmlReader.Value);
						break;
					case XmlNodeType.Element:
						switch (sgmlReader.LocalName)
						{
							case "h1":
								sb.Append("## ");
								break;
							case "br":
								sb.AppendLine("  ");
								break;
							case "a":
								if (sgmlReader.MoveToAttribute("href"))
								{
									string url = sgmlReader.Value;
									sgmlReader.Read();

									sb.AppendFormat("[{0}]({1})", sgmlReader.Value, url);
								}
								break;
							case "html":
								break;
							case "strong":
							case "b":
								sb.AppendFormat("**{0}**", sgmlReader.Value);
								break;
							case "i":
							case "em":
								sb.AppendFormat("_{0}_", sgmlReader.Value);
								break;
							case "li":
								sb.AppendFormat("- {0}", sgmlReader.Value);
								break;
							case "pre":
							case "code":
							case "quote":
								indentLevel = 1;
								break;
							case "ul":
							case "ol":
							case "img":
								break;
							default:
								outputEndElement = true;
								sb.Append("<").Append(sgmlReader.LocalName);
								break;
						}
						break;
					case XmlNodeType.SignificantWhitespace:
					case XmlNodeType.Whitespace:
					case XmlNodeType.CDATA:
						break;
					case XmlNodeType.EndElement:
						indentLevel = 0;
						if (outputEndElement)
							sb.Append(">");
						outputEndElement = false;
						break;
					default:
						throw new ArgumentOutOfRangeException();
				}
			}

			return sb.ToString();
		}

        public static string TransformHtmlToXHTML(string inputHtml)
        {
            var sgmlReader = new SgmlReader { DocType = "HTML",  };
            var stringReader = new StringReader(inputHtml);
            sgmlReader.InputStream = stringReader;

            var stringWriter = new StringWriter();

            using (var xmlWriter = new XmlTextWriter(stringWriter))
            {
                sgmlReader.Read();

                while (!sgmlReader.EOF)
                {
                    xmlWriter.WriteNode(sgmlReader, true);
                }
            }
            return RemoveCopyOfImage(stringWriter.ToString());
        }

        void Process(SgmlReader reader, string uri)
        {
            if (uri == null) {
                reader.InputStream = Console.In;
            } else {
                reader.Href = uri;
            }

            if (this.encoding == null) {
                this.encoding = reader.GetEncoding();
            }

            XmlTextWriter w = null;
            if (output != null) {
                w = new XmlTextWriter(output, this.encoding);
            }
            else {
                w = new XmlTextWriter(Console.Out);
            }
            if (formatted) w.Formatting = Formatting.Indented;
            if (!noxmldecl) {
                w.WriteStartDocument();
            }
            reader.Read();
            while (!reader.EOF) {
                w.WriteNode(reader, true);
            }
            w.Flush();
            w.Close();
        }

        void RunTest(SgmlReader reader, int line, string args, string input, string expectedOutput)
        {
            try {
                bool testdoc = false;
                bool testclone = false;
                bool format = true;
                bool ignore = false;
                foreach(string arg in args.Split(' ')) {
                    string sarg = arg.Trim();
                    if(sarg.Length == 0)
                        continue;
                    if(sarg[0] == '-') {
                        switch(sarg.Substring(1)) {
                        case "html":
                            reader.DocType = "html";
                            break;
                        case "lower":
                            reader.CaseFolding = CaseFolding.ToLower;
                            break;
                        case "upper":
                            reader.CaseFolding = CaseFolding.ToUpper;
                            break;
                        case "testdoc":
                            testdoc = true;
                            break;
                        case "testclone":
                            testclone = true;
                            break;
                        case "noformat":
                            format = false;
                            break;
                        case "ignore":
                            ignore = true;
                            break;
                        }
                    }
                }
                this.tests++;
                if(ignore) {
                    this.ignored++;
                    return;
                }
                reader.InputStream = new StringReader(input);
                if(format) {
                    reader.WhitespaceHandling = WhitespaceHandling.None;
                } else {
                    reader.WhitespaceHandling = WhitespaceHandling.All;
                }
                StringWriter output = new StringWriter();
                XmlTextWriter w = new XmlTextWriter(output);
                if(format) {
                    w.Formatting = Formatting.Indented;
                }
                if(testdoc) {
                    XmlDocument doc = new XmlDocument();
                    doc.Load(reader);
                    doc.WriteTo(w);
                } else if(testclone) {
                    XmlDocument doc = new XmlDocument();
                    doc.Load(reader);
                    XmlNode clone = doc.Clone();
                    clone.WriteTo(w);
                } else {
                    reader.Read();
                    while(!reader.EOF) {
                        w.WriteNode(reader, true);
                    }
                }
                w.Close();
                string actualOutput = output.ToString();

                // validate output
                using(StringReader source = new StringReader(actualOutput)) {
                    XmlDocument doc = new XmlDocument();
                    doc.Load(source);
                }

                // compare output
                if(actualOutput.Trim().Replace("\r", "") != expectedOutput.Trim().Replace("\r", "")) {
                    Console.WriteLine();
                    Console.WriteLine("ERROR: Test failed on line {0}", line);
                    Console.WriteLine("---- Expected output");
                    Console.Write(expectedOutput);
                    Console.WriteLine("---- Actual output");
                    Console.WriteLine(actualOutput);
                } else {
                    this.passed++;
                }
            } catch(Exception e) {
                Console.WriteLine("FATAL ERROR: Test threw an exception on line {0}", line);
                Console.WriteLine(e.Message);
                Console.WriteLine(e.ToString());
                Console.WriteLine("---- Input");
                Console.Write(input);
            }
        }

 void RunTest(SgmlReader reader, int line, string args, string input, string expectedOutput)
 {
     bool testdoc = false;
     foreach (string arg in args.Split(' ')){
         string sarg = arg.Trim();
         if (sarg.Length==0) continue;
         if (sarg[0] == '-'){
             switch (sarg.Substring(1)){
                 case "html":
                     reader.DocType = "html";
                     break;
                 case "lower":
                     reader.CaseFolding = CaseFolding.ToLower;
                     break;
                 case "upper":
                     reader.CaseFolding = CaseFolding.ToUpper;
                     break;
                 case "testdoc":
                     testdoc = true;
                     break;
             }
         }
     }
     this.tests++;
     reader.InputStream = new StringReader(input);
     reader.WhitespaceHandling = WhitespaceHandling.None;
     StringWriter output = new StringWriter();
     XmlTextWriter w = new XmlTextWriter(output);
     w.Formatting = Formatting.Indented;
     if (testdoc) {
         XmlDocument doc = new XmlDocument();
         doc.Load(reader);
         doc.WriteTo(w);
     } else {
         reader.Read();
         while (!reader.EOF) {
             w.WriteNode(reader, true);
         }
     }
     w.Close();
     string actualOutput = output.ToString();
     if (actualOutput.Trim() != expectedOutput.Trim()) {
         Console.WriteLine("ERROR: Test failed on line {0}", line);
         Console.WriteLine("---- Expected output");
         Console.WriteLine(expectedOutput);
         Console.WriteLine("---- Actual output");
         Console.WriteLine(actualOutput);
     } else {
         this.passed++;
     }
 }

		public void LoadHtml(string rawHtml)
		{
			if (rawHtml.StartsWith("<dsi:html"))
			{
				//parse and remove dsi:html tag...
				string tag = rawHtml.Substring(0, rawHtml.IndexOf('>') + 1) + "</dsi:html>";
				rawHtml = rawHtml.Substring(rawHtml.IndexOf('>') + 1);
				rawHtml = rawHtml.Substring(0, rawHtml.Length - 11);

				/*
				<dsi:html 
					formatting = [true | false] // do we convert line-breaks to br tags?
					container = [true | false]  // do we render the html in a container div?
					></dsi:html>
				*/

				SgmlReader sgml = new SgmlReader();
				sgml.InputStream = new StringReader(tag);
				sgml.DocType = "HTML";
				sgml.Read();

				if (sgml.GetAttribute("formatting") != null)
					Formatting = bool.Parse(sgml.GetAttribute("formatting"));

				if (sgml.GetAttribute("container") != null)
					Container = bool.Parse(sgml.GetAttribute("container"));

			}

			this.rawHtml = rawHtml;
		}

		public string dsiTagReplacement(Match m)
		{
			string tagName = "dsi";
			try
			{
				//string[] arrParts = m.Groups[1].Value.Split[" "];
				//Dictionary<string, string> parts = new Dictionary<string, string>();
				SgmlReader sgml = new SgmlReader();
				string inStr = m.Groups[0].Value;
				if (inStr.StartsWith("<dsi:link"))
					inStr += "</dsi:link>";
				sgml.InputStream = new StringReader(inStr);
				sgml.DocType = "HTML";
				sgml.Read();

				tagName = sgml.Name;
				string uniqueId = Guid.NewGuid().ToString("N");

				#region Parse attributes
				Dictionary<string, string> attributes = new Dictionary<string, string>();
				while (sgml.MoveToNextAttribute())
				{
					attributes.Add(sgml.Name.ToLower(), sgml.Value);
				}
				#endregion

				string typeAtt = attributes.ContainsKey("type") ? attributes["type"] : null;
				string refAtt = attributes.ContainsKey("ref") ? attributes["ref"] : null;

				#region Parse styles
				Dictionary<string, string> style = new Dictionary<string, string>();
				if (attributes.ContainsKey("style"))
				{
					foreach (string s in attributes["style"].Split(';'))
					{
						try
						{
							if (s.Contains(":"))
								style[s.Split(':')[0].Trim()] = s.Split(':')[1].Trim();
						}
						catch
						{
						}
					}
				}
				#endregion

				#region Parse class
				List<string> classes = new List<string>();
				if (attributes.ContainsKey("class"))
				{
					foreach (string s in attributes["class"].Split(' '))
					{
						try
						{
							classes.Add(s);
						}
						catch
						{
						}
					}
				}
				#endregion

				if (tagName == "dsi:video")
				{
					#region dsi:video
					/*
					<dsi:video 
						type = [dsi | flv | youtube | google | metacafe | myspace | break | collegehumor | redtube | ebaumsworld | dailymotion] 
						ref = [dsi-photo-k | site-ref]
						src = [flv-url]
						width = [width] (optional)
						height = [height] (optional)
						nsfw = [true | false] (optional)
						/> 
					*/
					bool nsfw = attributes.ContainsKey("nsfw") ? bool.Parse(attributes["nsfw"].ToLower()) : false;
					string draw = attributes.ContainsKey("draw") ? attributes["draw"].ToLower() : "auto";
					if (typeAtt == "youtube")
					{
						#region youtube
						int width = attributes.ContainsKey("width") ? int.Parse(attributes["width"]) : 425;
						int height = attributes.ContainsKey("height") ? int.Parse(attributes["height"]) : 355;

						//<object width="425" height="355"><param name="movie" value="http://www.youtube.com/v/8VtWo8tFdPQ&rel=1"></param><param name="wmode" value="transparent"></param><embed src="http://www.youtube.com/v/8VtWo8tFdPQ&rel=1" type="application/x-shockwave-flash" wmode="transparent" width="425" height="355"></embed></object>
						return GetFlash(uniqueId, height, width, nsfw, draw, "http://www.youtube.com/v/" + refAtt + "&rel=1");
						#endregion
					}
					else if (typeAtt == "metacafe")
					{
						#region metacafe
						int width = attributes.ContainsKey("width") ? int.Parse(attributes["width"]) : 400;
						int height = attributes.ContainsKey("height") ? int.Parse(attributes["height"]) : 345;

						//<embed src="http://www.metacafe.com/fplayer/1029494/how_to_make_fire_balls.swf" width="400" height="345" wmode="transparent" pluginspage="http://www.macromedia.com/go/getflashplayer" type="application/x-shockwave-flash"> </embed><br><font size = 1><a href="http://www.metacafe.com/watch/1029494/how_to_make_fire_balls/">How To Make Fire Balls</a> - <a href="http://www.metacafe.com/">The funniest videos clips are here</a></font>
						return GetFlash(uniqueId, height, width, nsfw, draw, "http://www.metacafe.com/fplayer/" + refAtt + ".swf");
						#endregion
					}
					else if (typeAtt == "google")
//.........這裏部分代碼省略.........

        /// <summary>
        /// Detects URLs in styles.
        /// </summary>
        /// <param name="baseUri">The base URI.</param>
        /// <param name="attributeName">Name of the attribute.</param>
        /// <param name="attributeValue">The attribute value.</param>
        /// <returns></returns>
        //private List<UriResourceInformation> ExtractStyleUrls(
        //    Uri baseUri,
        //    string attributeName,
        //    string attributeValue)
        //{
        //    List<UriResourceInformation> result =
        //        new List<UriResourceInformation>();
        //    if (string.Compare(attributeName, @"style", true) == 0)
        //    {
        //        if (attributeValue != null &&
        //            attributeValue.Trim().Length > 0)
        //        {
        //            MatchCollection matchs = Regex.Matches(
        //                attributeValue,
        //                @"url\s*\(\s*([^\)\s]+)\s*\)",
        //                RegexOptions.Singleline | RegexOptions.IgnoreCase);
        //            if (matchs.Count > 0)
        //            {
        //                foreach (Match match in matchs)
        //                {
        //                    if (match != null && match.Success)
        //                    {
        //                        string url = match.Groups[1].Value;
        //                        UriResourceInformation ui =
        //                            new UriResourceInformation(
        //                            _settings.Options,
        //                            url,
        //                            new Uri(url, UriKind.RelativeOrAbsolute),
        //                            baseUri,
        //                            UriType.Resource,
        //                            _uriInfo.AbsoluteUri,
        //                            );
        //                        bool isOnSameSite =
        //                            ui.IsOnSameSite(baseUri);
        //                        if ((isOnSameSite ||
        //                            !_settings.Options.StayOnSite) &&
        //                            ui.IsProcessableUri)
        //                        {
        //                            result.Add(ui);
        //                        }
        //                    }
        //                }
        //            }
        //        }
        //    }
        //    return result;
        //}
        /// <summary>
        /// Gets the doc reader.
        /// </summary>
        /// <param name="html">The HTML.</param>
        /// <param name="baseUri">The base URI.</param>
        /// <returns></returns>
        private static XmlReader GetDocReader(
            string html,
            Uri baseUri)
        {
            SgmlReader r = new SgmlReader();

            if (baseUri != null &&
                !string.IsNullOrEmpty(baseUri.ToString()))
                r.SetBaseUri(baseUri.ToString());
            r.DocType = @"HTML";
            r.WhitespaceHandling = WhitespaceHandling.All;
            r.CaseFolding = CaseFolding.None;
            StringReader sr = new StringReader(html);
            r.InputStream = sr;
            r.Read();

            return r;
        }

        void Process(SgmlReader reader, string uri, bool loadAsStream)
        {
            if (uri == null) {
                reader.InputStream = Console.In;
            }
            else if (loadAsStream) {
                Uri location = new Uri(uri);
                if (location.IsFile) {
                    reader.InputStream = new StreamReader(uri);
                } else {
                    WebRequest wr = WebRequest.Create(location);
                    reader.InputStream = new StreamReader(wr.GetResponse().GetResponseStream());
                }
            } else {
                reader.Href = uri;
            }

            if (debug) {
                Debug(reader);
                reader.Close();
                return;
            }
            if (crawl) {
                StartCrawl(reader, uri, basify);
                return;
            }

            if (this.encoding == null) {
                this.encoding = reader.GetEncoding();
            }

            XmlTextWriter w = null;
            if (output != null) {
                w = new XmlTextWriter(output, this.encoding);
            }
            else {
                w = new XmlTextWriter(Console.Out);
            }
            if (formatted) w.Formatting = Formatting.Indented;
            if (!noxmldecl) {
                w.WriteStartDocument();
            }
            if (testdoc) {
                XmlDocument doc = new XmlDocument();
                try {
                    doc.Load(reader);
                    doc.WriteTo(w);
                } catch (XmlException e) {
                    Console.WriteLine("Error:" + e.Message);
                    Console.WriteLine("at line " + e.LineNumber + " column " + e.LinePosition);
                }
            } else {
                reader.Read();
                while (!reader.EOF) {
                    w.WriteNode(reader, true);
                }
            }
            w.Flush();
            w.Close();
        }