C# HtmlDocument类代码示例

 public void CreateElement()
 {
     var doc = new HtmlDocument();
     var a = doc.CreateElement("a");
     Assert.AreEqual("a", a.Name);
     Assert.AreEqual(a.NodeType, HtmlNodeType.Element);
 }

 public void CreateTextNodeWithText()
 {
     var doc = new HtmlDocument();
     var a = doc.CreateTextNode("something");
     Assert.AreEqual("something", a.InnerText);
     Assert.AreEqual(a.NodeType, HtmlNodeType.Text);
 }

 public void Head_SetNonExisting_Null_RemovesElement()
 {
     HtmlDocument document = new HtmlDocument() { Head = null };
     document.Head = null;
     Assert.True(document.IsEmpty);
     Assert.Null(document.Head);
 }

 private static string ExtractCss(HtmlDocument doc)
 {
     var styleTags = doc.FindTagsByName("style", "link").ToArray();
     var css = new StringBuilder();
     foreach (var tag in styleTags)
     {
         if (tag.IsType("style"))
             css.AppendLine(tag.ChildElements.ToSeparatedString(" "));
         else if (tag.IsType("link") && (string.Compare(tag.Attributes["type"], "text/css", true) == 0 || string.Compare(tag.Attributes["rel"], "stylesheet") == 0))
         {
             var src = tag.Attributes["href"];
             if (string.IsNullOrEmpty(src))
                 continue;
             var uri = new Uri(src, UriKind.RelativeOrAbsolute);
             if (!uri.IsAbsoluteUri && HttpContext.Current != null && HttpContext.Current.Request != null)
             {
                 uri = new Uri(HttpContext.Current.Request.Url, uri);
             }
             if (!uri.IsAbsoluteUri)
                 continue;
             var client = new WebClient();
             css.Append(client.DownloadString(uri));
         }
     }
     foreach (var tag in styleTags)
     {
         tag.Remove();
     }
     return css.ToString();
 }

 public static List<string> GetLinksFromWebsite(string htmlSource)
 {
     var doc = new HtmlDocument();
     try
     {
         doc.LoadHtml(htmlSource);
         if (doc.DocumentNode.InnerHtml == null) return null;
         if (doc.DocumentNode.SelectNodes("//a[@ref]") == null)
         {
             return doc
                .DocumentNode
                .SelectNodes("//a[@href]")
                .Select(node => node.Attributes["href"].Value)
                .ToList();
         }
         return doc
             .DocumentNode
             .SelectNodes("//a[@ref]")
             .Select(node => node.Attributes["ref"].Value)
             .ToList();
     }
     catch (Exception)
     {
     }
         return null;
 }

 public void CreateTextNode()
 {
     var doc = new HtmlDocument();
     var a = doc.CreateTextNode();
     Assert.AreEqual(HtmlNode.HtmlNodeTypeNameText, a.Name);
     Assert.AreEqual(a.NodeType, HtmlNodeType.Text);
 }

        public void Mouse_click_will_fail_because_hidden_link_is_matched_instead_of_visible_link()
        {
            // Playback.PlaybackSettings.SmartMatchOptions = SmartMatchOptions.None; // tried Sudhish Mathuria's suggestion

            var example2Path = Path.Combine(TestContext.TestDeploymentDir, "example2.htm");

            var window = BrowserWindow.Launch(example2Path);

            var document = new HtmlDocument(window);
            document.FilterProperties[HtmlDocument.PropertyNames.Title] = "2ad6de55-75f1-403f-8fcb-1d5defac9786";

            var visibleLink = new HtmlHyperlink(document);
            visibleLink.SearchProperties[HtmlHyperlink.PropertyNames.InnerText] = "Hello";

            // ONLY FIND VISIBLE CONTROLS !
            visibleLink.SearchConfigurations.Add(SearchConfiguration.VisibleOnly);

            var allMatches = visibleLink.FindMatchingControls();
            Assert.AreEqual(2, allMatches.Count, "Should be two matching.");

            visibleLink.Find(); // THIS SHOULD BE THE VISIBLE ONE

            Debug.WriteLine("BoundingRectangle: " + visibleLink.BoundingRectangle);
            Assert.IsTrue(visibleLink.BoundingRectangle.Width > 0, "Width should positive.");
            Assert.IsTrue(visibleLink.BoundingRectangle.Height > 0, "Height should positive.");

            Mouse.Click(visibleLink);
        }

 public void CreateAttributeWithText()
 {
     var doc = new HtmlDocument();
     var a = doc.CreateAttribute("href", "http://something.com");
     Assert.AreEqual("href", a.Name);
     Assert.AreEqual("http://something.com", a.Value);
 }

        public void Head_ReturnsHeadObject()
        {
            var doc = new HtmlDocument();
            var head = doc.Head;

            Assert.IsInstanceOf<Head>(head);
        }

 static void Main(string[] args)
 {
     HtmlDocument doc = new HtmlDocument();
     doc.Load(@"..\..\mshome.htm");
     doc.OptionOutputAsXml = true;
     doc.Save("mshome.xml");
 }

	static void Main (string [] args)
	{
		HtmlDocument doc = new HtmlDocument();
		doc.Load(args [0]);
		doc.OptionOutputAsXml = true;
		doc.Save(args [1]);
	}

 public void TestCallingExistingFunction()
 {
     var doc = new HtmlDocument();
     doc.LoadHtml("<html><body class=\"asdfasd\"><p>asdf asdf sdf</p></body></html>");
     dynamic docElement = doc.DocumentNode;
     var item = docElement.Descendants();
     Assert.IsInstanceOf<IEnumerable<HtmlNode>>(item);
 }

        public void Body_ReturnsBodyObject()
        {
            var doc = new HtmlDocument();

            Body body = doc.Body;

            Assert.IsInstanceOf<Body>(body);
        }

 public void Head_SetExisting_NonNull_RemovesElement()
 {
     HtmlElement value = Tag.Head.WithClass("class");
     HtmlDocument document = new HtmlDocument() { Head = Tag.Head };
     document.Head = value;
     Assert.Equal(new HtmlElement[] { value }, document.Elements());
     Assert.Equal(value, document.Head);
 }

        public void Test_HasAttribute()
        {
            var doc = new HtmlDocument("<html><head></head><body><a href=\"http://flo.se\">yo</a></body></html>");

            var link = doc.DocumentNode.SelectSingleNode("//a");

            Assert.IsTrue(link.HasAttribute("href"));
        }