C# HtmlDocument.CreateElement方法代码示例

        /// <summary>
        /// Decodes the HTML.
        /// </summary>
        /// <param name="htmlText">The html text.</param>
        /// <returns>String without any html tags.</returns>
        public static string DecodeHtml(string htmlText)
        {
            htmlText = htmlText.Replace("<p>", "").Replace("</p>", "\r\n\r\n");

            string decoded = String.Empty;
            if (htmlText.IndexOf('<') > -1 || htmlText.IndexOf('>') > -1 || htmlText.IndexOf('&') > -1)
            {
                try
                {
                    HtmlDocument document = new HtmlDocument();

                    var decode = document.CreateElement("div");
                    htmlText = htmlText.Replace(".<", ". <").Replace("?<", "? <").Replace("!<", "! <").Replace("&#039;", "'");
                    decode.InnerHtml = htmlText;

                    decoded = WebUtility.HtmlDecode(decode.InnerText);
                    decoded = Regex.Replace(decoded, "<!--.*?-->", string.Empty, RegexOptions.Singleline);
                }
                catch { }
            }
            else
            {
                decoded = htmlText;
            }
            return decoded;
        }

        private static HtmlNode GetExternalReferencesListNode(
            HtmlDocument document,
            IEnumerable<ExternalReference> externalReferences)
        {
            var list = document.CreateElement("ol");
            foreach (var reference in externalReferences)
            {
                var item = document.CreateElement("li");

                string referenceId = GetExternalReferenceId(reference.Index);

                var backLink = document.CreateElement("a");
                backLink.InnerHtml = "<strong>^</strong>";
                backLink.SetAttributeValue("href", "#" + referenceId + BackLinkReferenceIdSuffix);

                var externalLink = document.CreateElement("a");
                externalLink.InnerHtml = reference.Url;
                externalLink.SetAttributeValue("href", reference.Url);
                externalLink.SetAttributeValue("name", referenceId);

                item.AppendChild(backLink);
                item.AppendChild(document.CreateTextNode(" "));
                item.AppendChild(externalLink);

                list.AppendChild(item);
            }

            return list;
        }

        public static string DecodeHtml(string htmltext)
        {
            htmltext = htmltext.Replace("<p>", "").Replace("</p>", "\r\n\r\n");

            string decoded = String.Empty;
            if (htmltext.IndexOf('<') > -1 || htmltext.IndexOf('>') > -1 || htmltext.IndexOf('&') > -1)
            {
                try
                {
                    HtmlDocument document = new HtmlDocument();

                    var decode = document.CreateElement("div");
                    htmltext = htmltext.Replace(".<", ". <").Replace("?<", "? <").Replace("!<", "! <").Replace("&#039;", "'");
                    decode.InnerHtml = htmltext;

                    var allElements = decode.Descendants().ToArray();
                    for (int n = allElements.Length - 1; n >= 0; n--)
                    {
                        if (allElements[n].NodeType == HtmlNodeType.Comment || allElements[n].Name.EqualNoCase("style") || allElements[n].Name.EqualNoCase("script"))
                        {
                            allElements[n].Remove();
                        }
                    }
                    decoded = WebUtility.HtmlDecode(decode.InnerText);
                }
                catch { }
            }
            else
            {
                decoded = htmltext;
            }
            return decoded;
        }

        public void Parse(string input, string[] args = null)
        {
            var xxr = new XamlXmlReader(new StringReader(input), new XamlSchemaContext());
            var graphReader = new XamlObjectWriter(xxr.SchemaContext);

            while (xxr.Read())
                graphReader.WriteNode(xxr);

            var page = (Page)graphReader.Result;

            // Map our generators
            var g = new Generator();
            g.Map<Page, PageGeneratorBlock>();
            g.Map<Button, ButtonGeneratorBlock>();
            g.Map<StackPanel, StackPanelGeneratorBlock>();

            var doc = new HtmlDocument();
            var html = doc.CreateElement("html");

            g.Generate(html, page);
            // HTML5 Doc type
            doc.DocumentNode.AppendChild(doc.CreateComment("<!DOCTYPE html>"));
            doc.DocumentNode.AppendChild(html);

            doc.Save("test.htm");

            var cssContents = g.GenerateStyles(page);
            File.WriteAllText("XamlCore.css", cssContents);
        }

        private void WriteContent(HtmlDocument doc)
        {
            HtmlNode bodyContent = doc.GetElementbyId("bodyContent");

            var heading = doc.CreateElement("h1");
            heading.InnerHtml = "Heading";
            bodyContent.AppendChild(heading);

            HtmlNode table = bodyContent.CreateElement("table");
            table.CreateAttributeWithValue("class", "table table-striped table-bordered");
            var tableHead = table.CreateElement("thead");
            var headerRow = tableHead.CreateElement("tr");
            headerRow.CreateElementWithHtml("td", "First");
            headerRow.CreateElementWithHtml("td", "Second");
            headerRow.CreateElementWithHtml("td", "Third");
            headerRow.CreateElementWithHtml("td", "Fourth");
            headerRow.CreateElementWithHtml("td", "Fifth");

            HtmlNode tableBody = table.CreateElement("tbody");
            const string text = "Fi fa fo fum fi fa fo fum fi fa fo fum fi fa fo fum";
            for (int i = 0; i < 10; i++)
            {
                HtmlNode bodyRow = tableBody.CreateElement("tr");
                bodyRow.CreateElementWithHtml("td", i + " first " + text);
                bodyRow.CreateElementWithHtml("td", i + " second " + text);
                bodyRow.CreateElementWithHtml("td", i + " third " + text);
                bodyRow.CreateElementWithHtml("td", i + " fourth " + text);
                bodyRow.CreateElementWithHtml("td", i + " fifth " + text);
            }
        }

        static Article GrabArticle(HtmlDocument doc)
        {
            Article article = new Article();

            HtmlNode articleContent = doc.CreateElement("div");

            ParagraphParentCollection paragraphParents = new ParagraphParentCollection(new ParagraphCollection(doc.DocumentNode.SelectNodes("//p")));

            // Replace br tags with paragraph tags.
            doc.DocumentNode.InnerHtml = Regex.Replace(doc.DocumentNode.InnerHtml, @"<br/?>[ \r\n\s]*<br/?>", @"</p><p>");

            // TODO handle title.
            article.Title = doc.DocumentNode.SelectSingleNode("//title") == null ? null : doc.DocumentNode.SelectSingleNode("//title").InnerText;

            foreach (ParagraphParent parent in paragraphParents)
            {
                foreach (HtmlAttribute att in parent.Node.Attributes.AttributesWithName("class"))
                {
                    if (Regex.IsMatch(att.Name, @"/(comment|meta|footer|footnote)/"))
                        parent.Score -= 50;
                    else if (Regex.IsMatch(att.Name, @"/((^|\\s)(post|hentry|entry[-]?(content|text|body)?|article[-]?(content|text|body)?)(\\s|$))/"))
                        parent.Score += 25;
                    break;
                }

                foreach (HtmlAttribute att in parent.Node.Attributes.AttributesWithName("id"))
                {
                    if (Regex.IsMatch(att.Name, @"/(comment|meta|footer|footnote)/"))
                        parent.Score -= 50;
                    else if (Regex.IsMatch(att.Name, @"/^(post|hentry|entry[-]?(content|text|body)?|article[-]?(content|text|body)?)$/"))
                        parent.Score += 25;
                }

                foreach (Paragraph paragraph in parent.Paragraphs)
                {
                    if (paragraph.Node.InnerText.Length > 10)
                        parent.Score++;

                    parent.Score += GetCharCount(paragraph.Node);
                }
            }

            ParagraphParent winner = paragraphParents.OrderByDescending(a => a.Score).FirstOrDefault();

            // TODO cleanup.
            winner.Clean("style");
            winner.KillDivs();
            winner.KillBreaks();
            winner.Clean("form");
            winner.Clean("object");
            winner.Clean("table", 250);
            winner.Clean("h1");
            winner.Clean("h2");
            winner.Clean("iframe");
            winner.Clean("script");

            article.Content.DocumentNode.AppendChild(winner.Node);
            return article;
        }

        public void when_html_document_doesnt_contain_rss_feed()
        {
            var document = new HtmlDocument();
            document.CreateElement("html");

            var parser = new Scraper(document);
            var actual = parser.GetRssFeedUrl();

            Assert.IsNull(actual);
        }

        public override void Execute(IEmailItem emailItem = null, int? lastExitCode = null)
        {
            if (AppliesTo(emailItem, lastExitCode))
            {
                if (BodyFormat.Text == emailItem.Message.Body.BodyFormat)
                {
                    StringBuilder sb = new StringBuilder();
                    sb.Append(emailItem.Message.GetBody());
                    sb.AppendLine();
                    sb.AppendLine();
                    sb.Append(Text);
                    emailItem.Message.SetBody(sb.ToString());
                }

                if (BodyFormat.Rtf == emailItem.Message.Body.BodyFormat)
                {
                    var messageRtf = new RtfDocument(emailItem.Message.GetBody());
                    var mergeRtf = new RtfDocument(Rtf);

                    if (messageRtf.Merge(mergeRtf))
                    {
                        emailItem.Message.SetBody(messageRtf.Content);
                    }
                }

                if (BodyFormat.Html == emailItem.Message.Body.BodyFormat)
                {
                    var messageDoc = new HtmlDocument();
                    messageDoc.LoadHtml(emailItem.Message.GetBody());

                    var disclaimerDoc = new HtmlDocument();
                    disclaimerDoc.LoadHtml(Html);

                    var messageBodyNode = messageDoc.DocumentNode.SelectSingleNode("//body");
                    var disclaimerBodyNode = disclaimerDoc.DocumentNode.SelectSingleNode("//body");

                    var brNode = messageDoc.CreateElement("br");

                    messageBodyNode.AppendChild(brNode);

                    messageBodyNode.AppendChildren(disclaimerBodyNode.ChildNodes);

                    emailItem.Message.SetBody(messageDoc.DocumentNode.InnerHtml);
                }

                if (null != Handlers && Handlers.Count > 0)
                {
                    foreach (IHandler handler in Handlers)
                    {
                        handler.Execute(emailItem, lastExitCode);
                    }
                }
            }
        }

        public string TransformImgToPicture(string content)
        {
            try
            {
                HtmlDocument doc = new HtmlDocument();
                doc.LoadHtml(content);

                // - activating"do no harm"-mode
                doc.OptionFixNestedTags = false;
                doc.OptionAutoCloseOnEnd = false;
                doc.OptionCheckSyntax = false;
                HtmlNodeCollection collection = doc.DocumentNode.SelectNodes("//img[@src]");
                if (collection == null)
                    return content;
                var changed = false;
                foreach (HtmlNode img in collection)
                {
                    HtmlAttribute src = img.Attributes["src"];
                    HtmlAttribute cls = img.Attributes["class"];

                    if ((src != null && src.Value.IndexOf("slimmage=true", StringComparison.OrdinalIgnoreCase) > -1)
                        || (cls != null && cls.Value.IndexOf("slimmage",  StringComparison.OrdinalIgnoreCase) > -1))
                    {

                        // - append fallback image
                        HtmlNode container = doc.CreateElement("noscript");
                        container.SetAttributeValue("data-slimmage", "true");
                        //copy attributes for IE6/7/8 support
                        foreach (var a in img.Attributes)
                        {
                            container.SetAttributeValue("data-img-" + a.Name, a.Value);
                        }
                        //Place 'img' inside 'noscript'
                        img.ParentNode.InsertBefore(container, img);
                        img.Remove();
                        container.AppendChild(img);
                        changed = true;
                    }
                }
                //Don't modify the DOM unless you actually edited the HTML
                return changed ? doc.DocumentNode.OuterHtml : content;
            }
            catch(Exception ex)
            {
                Trace.TraceWarning("SlimResponse failed to parse HTML: " + ex.ToString() + ex.StackTrace);
                // - better that nothing(tm) ...
                //return content;
                return ex.ToString();
            }
        }

        public string ReplaceDeletedImageBySpanCoupons(string htmlContent)
        {
            HtmlDocument doc = new HtmlDocument();
            doc.LoadHtml(htmlContent);

            foreach (HtmlNode img in doc.DocumentNode.SelectNodes("//img[@class='" + hdnImageClass.Value + "']"))
            {
                string value = img.Attributes.Contains("value") ? img.Attributes["value"].Value : " ";
                HtmlNode lbl = doc.CreateElement("span");
                lbl.Attributes.Add("class", hdnImageClass.Value);
                lbl.InnerHtml = value;

                img.ParentNode.ReplaceChild(lbl, img);
            }
            return doc.DocumentNode.OuterHtml;
        }

        public string ReplaceDeletedIframeBySpan(string htmlContent)
        {
            HtmlDocument doc = new HtmlDocument();
            doc.LoadHtml(htmlContent);

            foreach (HtmlNode img in doc.DocumentNode.SelectNodes("//iframe"))
            {

                HtmlNode lbl = doc.CreateElement("span");
                lbl.Attributes.Add("class", "IF");
                lbl.Attributes.Add("height", "187");
                lbl.Attributes.Add("width", "240");

                img.ParentNode.ReplaceChild(lbl, img);
            }
            return doc.DocumentNode.OuterHtml;
        }

        public override TaskExecutionDetails Execute(string Value)
        {
            TaskExecutionDetails d = new TaskExecutionDetails();


            HtmlDocument doc = new HtmlDocument();
            doc.Load(IO.IOHelper.MapPath(SystemDirectories.Masterpages) + "/" + TargetFile);

            //if (doc.DocumentNode.SelectSingleNode(string.Format("//link [@href = '{0}']", Value)) == null)
            //{

                HtmlNode target = doc.DocumentNode.SelectSingleNode(string.IsNullOrEmpty(TargetSelector) ? "//head" : TargetSelector.ToLower());

                if (target != null)
                {

                    HtmlNode s = doc.CreateElement("link");
                    //s.Name = "link";
                    s.Attributes.Append("rel", "stylesheet");
                    s.Attributes.Append("type", "text/css");


                    s.Attributes.Append("href", Value);


                    if (!string.IsNullOrEmpty(Media))
                        s.Attributes.Append("media", Media);

                    target.AppendChild(s);

                    doc.Save(IO.IOHelper.MapPath(SystemDirectories.Masterpages) + "/" + TargetFile);

                    d.TaskExecutionStatus = TaskExecutionStatus.Completed;
                    d.NewValue = Value;
                }
                else
                    d.TaskExecutionStatus = TaskExecutionStatus.Cancelled;   
            //}
            //else
            //    d.TaskExecutionStatus = TaskExecutionStatus.Cancelled;

            return d;
        }

 public static string GetPreview(string html, string replacmentHtml, Guid productID)
 {
     var doc = new HtmlDocument();
     doc.LoadHtml(string.Format("<html>{0}</html>", htmlTags.Replace(html, string.Empty)));
     var nodes = doc.DocumentNode.SelectNodes("//div[translate(@class,'ABCDEFGHIJKLMNOPQRSTUVWXYZ','abcdefghijklmnopqrstuvwxyz')='asccut']");
     if (nodes != null)
     {
         foreach (var node in nodes)
         {
             var newNode = doc.CreateElement("div");
             var styleAttr = doc.CreateAttribute("style");
             styleAttr.Value = "display:inline;";
             newNode.Attributes.Append(styleAttr);
             newNode.InnerHtml = replacmentHtml ?? string.Empty;
             node.ParentNode.ReplaceChild(newNode, node);
         }
     }
     ProcessCustomTags(doc, productID);
     return htmlTags.Replace(doc.DocumentNode.InnerHtml, string.Empty);
 }

        public static HtmlNode CreateNode(
            HtmlDocument document,
            string name,
            string className = null,
            HtmlNode parentNode = null)
        {
            var node = document.CreateElement(name);

            if (!string.IsNullOrWhiteSpace(className))
            {
                node.SetAttributeValue("class", className);
            }

            if (parentNode != null)
            {
                parentNode.AppendChild(node);
            }

            return node;
        }

        public void Download(HtmlDocument document, string path)
        {
            Directory.CreateDirectory(ResourcePath);

            //Descargar todos los recursos
            _downloadImages(document);
            _downloadResources(document, "link", "href", true);
            _downloadResources(document, "script", "src");

            //Añadir cabeceras meta para indificar codificación UTF8
            HtmlNode meta = document.CreateElement("meta");
            meta.Attributes.Add("charset", "utf-8");
            document.DocumentNode.SelectSingleNode("//head").ChildNodes.Add(meta);

            /*HtmlNode meta2 = document.CreateElement("meta");
            meta2.Attributes.Add("http-equiv", "Content-Type");
            meta2.Attributes.Add("content", "Type=text/html; charset=utf-8");
            document.DocumentNode.SelectSingleNode("//head").ChildNodes.Add(meta2);*/

            File.WriteAllText(path, document.DocumentNode.WriteTo(), new UTF8Encoding(true));
        }