C# HtmlNode.Descendants方法代码示例

        protected static MatchInfos ParseLine(HtmlNode node)
        {
            MatchInfos infos = null;

            infos = new MatchInfos();
            var infosAdresse = node.Descendants("div").ElementAt(0).ChildNodes.ElementAt(1).Descendants("#text");
            infos.Nom = infosAdresse.ElementAt(0).InnerText.Trim().Replace(" ", " ");
            infos.Adresse = infosAdresse.ElementAt(1).InnerText.Trim().Replace(" ", " ");
            infos.Ville = infosAdresse.ElementAt(2).InnerText.Trim().Replace(" ", " ");

            if(node.ChildNodes.Where(item => item.Name == "div").Count() > 2)
            {
                var infosArbitres = node.ChildNodes.Where(item => item.Name == "div")
                                    .ElementAt(2)
                                    .ChildNodes
                                    .ElementAt(1)
                                    .Descendants("ul")
                                    .ElementAt(0)
                                    .Descendants("li");
                foreach (var arbitre in infosArbitres)
                {
                    infos.Arbitres.Add(HtmlEntity.DeEntitize(arbitre.InnerText.Trim()));
                }
            }
            return infos;
        }

        public static ForumUserEntity FromPost(HtmlNode postNode)
        {
            var user = new ForumUserEntity
            {
                Username =
                    WebUtility.HtmlDecode(
                        postNode.Descendants("dt")
                            .FirstOrDefault(node => node.GetAttributeValue("class", string.Empty).Contains("author"))
                            .InnerHtml),
                DateJoined =
                   DateTime.Parse(postNode.Descendants("dd")
                        .FirstOrDefault(node => node.GetAttributeValue("class", string.Empty).Contains("registered"))
                        .InnerHtml)
            };
            var avatarTitle = postNode.Descendants("dd").FirstOrDefault(node => node.GetAttributeValue("class", string.Empty).Equals("title"));
            var avatarImage = postNode.Descendants("dd").FirstOrDefault(node => node.GetAttributeValue("class", string.Empty).Contains("title")).Descendants("img").FirstOrDefault();

            if (avatarTitle != null)
            {
                user.AvatarTitle = WebUtility.HtmlDecode(avatarTitle.InnerText).WithoutNewLines().Trim();
            }
            if (avatarImage != null)
            {
                user.AvatarLink = FixPostHtmlImage(avatarImage.OuterHtml);
            }
            user.Id = Convert.ToInt64(postNode.DescendantsAndSelf("td").FirstOrDefault(node => node.GetAttributeValue("class", string.Empty).Contains("userinfo")).GetAttributeValue("class", string.Empty).Split('-')[1]);
            return user;
        }

        private static Activity ExtractActivity(HtmlNode node, int index)
        {
            var name = node.Descendants("div")
                           .Where(div => div.GetAttributeValue("class", null) == "action_prompt")
                           .Select(div => HtmlEntity.DeEntitize(div.InnerText).Trim().Replace("  ", " "))
                           .FirstOrDefault();
            if (name == null)
            {
                throw new InvalidDataException("Unable to find activity name");
            }

            return new Activity
                {
                    Sequence = index,
                    Name = name,
                    Note = node.Descendants("li")
                               .Where(li => li.GetAttributeValue("class", null) == "stream_note")
                               .Select(li => HtmlEntity.DeEntitize(li.InnerText).Trim())
                               .FirstOrDefault(),
                    Sets = node.Descendants("li")
                               .Where(li => li.GetAttributeValue("class", null) != "stream_note")
                               .Select(ExtractSet)
                               .ToList()
                };
        }

        public Team GetTeamInfo(HtmlNode div)
        {
            HtmlNode[] links = div.Descendants("a").ToArray();

            string name = div.Descendants("h1").First().InnerHtml;
            name = name.Substring(name.IndexOf(" ") + 1, name.IndexOf("Roster") - 9);

            string divName = links[3].NextSibling.InnerHtml;
            divName = divName.Substring(1, divName.Length - 12);

            Team team = new Team(name, new Division(divName));

            string coachName = links[5].InnerHtml;
            team.Coach = new Coach(coachName, team);

            HtmlNode[] strongs = div.Descendants("strong").ToArray();

            string arenaName = strongs[strongs.Length - 2].NextSibling.InnerHtml;
            arenaName = arenaName.Substring(1, arenaName.IndexOf("&") - 2);

            string arenaAttendance = strongs[strongs.Length - 1].NextSibling.InnerHtml;
            arenaAttendance = arenaAttendance.Substring(1, arenaAttendance.IndexOf("(") - 2).Replace(",", string.Empty);

            team.Arena = new Arena(arenaName, int.Parse(arenaAttendance));

            return team;
        }

 public void Parse(HtmlNode rowNode)
 {
     Status =
         rowNode.Descendants("td")
             .FirstOrDefault(node => node.GetAttributeValue("class", string.Empty).Equals("status"))
             .Descendants("img")
             .FirstOrDefault()
             .GetAttributeValue("src", string.Empty);
     Icon =
         rowNode.Descendants("td")
             .FirstOrDefault(node => node.GetAttributeValue("class", string.Empty).Equals("icon"))
             .Descendants("img")
             .FirstOrDefault()
             .GetAttributeValue("src", string.Empty);
     Title =
         rowNode.Descendants("td")
             .FirstOrDefault(node => node.GetAttributeValue("class", string.Empty).Equals("title"))
             .InnerText;
     Sender = rowNode.Descendants("td")
         .FirstOrDefault(node => node.GetAttributeValue("class", string.Empty).Equals("sender"))
         .InnerText;
     Date = rowNode.Descendants("td")
         .FirstOrDefault(node => node.GetAttributeValue("class", string.Empty).Equals("date"))
         .InnerText;
 }

 public void ParseMainArticle(HtmlNode articleNode)
 {
     ArticleImage = articleNode.Descendants("img").FirstOrDefault().GetAttributeValue("src", string.Empty);
     Title =
         WebUtility.HtmlDecode(
             articleNode.Descendants("h2").FirstOrDefault().Descendants("a").FirstOrDefault().InnerText);
     ArticleLink =
         articleNode.Descendants("h2")
             .FirstOrDefault()
             .Descendants("a")
             .FirstOrDefault()
             .GetAttributeValue("href", string.Empty);
     Date =
         articleNode.Descendants("span")
             .FirstOrDefault(node => node.GetAttributeValue("class", string.Empty).Contains("date"))
             .InnerHtml;
     Author =
         WebUtility.HtmlDecode(
             articleNode.Descendants("span")
                 .FirstOrDefault(node => node.GetAttributeValue("class", string.Empty).Contains("author"))
                 .Descendants("a")
                 .FirstOrDefault()
                 .InnerText);
     AuthorLink =
         articleNode.Descendants("span")
             .FirstOrDefault(node => node.GetAttributeValue("class", string.Empty).Contains("author"))
             .Descendants("a")
             .FirstOrDefault()
             .GetAttributeValue("href", string.Empty);
     ArticleText = WebUtility.HtmlDecode(articleNode.Descendants("p").FirstOrDefault().InnerText);
 }

 public OverviewResult ConvertSingleResult(HtmlNode node)
 {
     OverviewResult result = new OverviewResult();
     result.Url = node.Descendants("a").SingleOrDefault()?.Attributes["href"]?.Value;
     result.Type = node.Descendants("h4").SingleOrDefault()?.InnerText;
     result.Name = node.Descendants("h2").SingleOrDefault()?.InnerText;
     return result;
 }

 public void ParseThread(HtmlNode popularThreadsNode)
 {
     Tag = popularThreadsNode.Descendants("img").FirstOrDefault().GetAttributeValue("src", string.Empty);
     Title = WebUtility.HtmlDecode(popularThreadsNode.Descendants("a").FirstOrDefault().InnerText);
     Id =
         Convert.ToInt64(
             popularThreadsNode.Descendants("a")
                 .FirstOrDefault()
                 .GetAttributeValue("href", string.Empty)
                 .Split('=')[1]);
 }

 private HtmlNode FindRosterTable(HtmlNode document)
 {
     IEnumerable<HtmlNode> blocks = document.Descendants("th").
         Where(h => h.InnerHtml.IndexOf("2011-12 Roster") != -1);
     if (blocks.Count() == 0)
     {
         blocks = document.Descendants("td").
             Where(h => h.InnerHtml.IndexOf("2011-12 Roster") != -1);
     }
     return blocks.Last().Ancestors("table").First();
 }

        /// <summary>
        ///     Parses a forum post in a thread.
        /// </summary>
        /// <param name="postNode">The post HTML node.</param>
        public void Parse(HtmlNode postNode)
        {
            User = ForumUserEntity.FromPost(postNode);

            HtmlNode postDateNode =
                postNode.Descendants()
                    .FirstOrDefault(node => node.GetAttributeValue("class", string.Empty).Equals("postdate"));
            string postDateString = postDateNode == null ? string.Empty : postDateNode.InnerText;
            if (postDateString != null)
            {
                PostDate = postDateString.WithoutNewLines().Trim();
            }

            PostIndex = ParseInt(postNode.GetAttributeValue("data-idx", string.Empty));

            var postId = postNode.GetAttributeValue("id", string.Empty);
            if (!string.IsNullOrEmpty(postId) && postId.Contains("#"))
            {
                PostId =
                    Int64.Parse(postNode.GetAttributeValue("id", string.Empty)
                        .Replace("post", string.Empty)
                        .Replace("#", string.Empty));
            }
            else if (!string.IsNullOrEmpty(postId) && postId.Contains("post"))
            {
                PostId =
                    Int64.Parse(postNode.GetAttributeValue("id", string.Empty)
                        .Replace("post", string.Empty));
            }
            else
            {
                PostId = 0;
            }
         
            var postBodyNode = postNode.Descendants("td")
                .FirstOrDefault(node => node.GetAttributeValue("class", string.Empty).Equals("postbody"));
            this.FixQuotes(postBodyNode);
            PostHtml = postBodyNode.InnerHtml;
            HtmlNode profileLinksNode =
                    postNode.Descendants("td")
                        .FirstOrDefault(node => node.GetAttributeValue("class", string.Empty).Equals("postlinks"));
            HtmlNode postRow =
                postNode.Descendants("tr").FirstOrDefault();

            if (postRow != null)
            {
                HasSeen = postRow.GetAttributeValue("class", string.Empty).Contains("seen");
            }

            User.IsCurrentUserPost =
                profileLinksNode.Descendants("img")
                    .FirstOrDefault(node => node.GetAttributeValue("alt", string.Empty).Equals("Edit")) != null;
        }

 public void Parse(HtmlNode node)
 {
     try
     {
         Id = Convert.ToInt32(node.Descendants("input").First().GetAttributeValue("value", string.Empty));
         ImageUrl = node.Descendants("img").First().GetAttributeValue("src", string.Empty);
         Title = node.Descendants("img").First().GetAttributeValue("alt", string.Empty);
     }
     catch (Exception)
     {
         // If, for some reason, it fails to get an icon, ignore the error.
         // The list view won't show it.
     }
 }

        public static ForumUserRapSheetEntity FromRapSheet(HtmlNode rapSheetNode)
        {
            var rapSheet = new ForumUserRapSheetEntity();

            List<HtmlNode> rapSheetData = rapSheetNode.Descendants("td").ToList();
            rapSheet.PunishmentType = rapSheetData[0].Descendants("b").FirstOrDefault().InnerText;
            rapSheet.Date = rapSheetData[1].InnerText;

            rapSheet.HorribleJerk = rapSheetData[2].Descendants("a").FirstOrDefault().InnerText;
            rapSheet.HorribleJerkId =
                Convert.ToInt64(
                    rapSheetData[2].Descendants("a").FirstOrDefault().GetAttributeValue("href", string.Empty).Split('=')
                        [3]);

            rapSheet.PunishmentReason = rapSheetData[3].InnerText;

            rapSheet.RequestedBy = rapSheetData[4].Descendants("a").FirstOrDefault().InnerText;
            rapSheet.RequestedById =
                Convert.ToInt64(
                    rapSheetData[4].Descendants("a").FirstOrDefault().GetAttributeValue("href", string.Empty).Split('=')
                        [3]);

            rapSheet.ApprovedBy = rapSheetData[5].Descendants("a").FirstOrDefault().InnerText;
            rapSheet.ApprovedById =
                Convert.ToInt64(
                    rapSheetData[5].Descendants("a").FirstOrDefault().GetAttributeValue("href", string.Empty).Split('=')
                        [3]);

            return rapSheet;
        }

        public bool LoadFromHtml(HtmlNode RootNode, DataLoadedEventArgs loadedEventArgs)
        {
            try
            {
                IEnumerable<HtmlNode> linkNodes = RootNode.Descendants("a");
                foreach (HtmlNode linkNode in linkNodes)
                {
                    MitbbsClubGroupLink clubGroupLink = new MitbbsClubGroupLink();
                    clubGroupLink.ParentUrl = Url;

                    if (clubGroupLink.LoadFromHtml(linkNode))
                    {
                        ClubGroupLinks.Add(clubGroupLink);
                        IsLoaded = true;
                    }
                }
            }
            catch (Exception e)
            {
                IsLoaded = false;
                loadedEventArgs.Error = e;
            }

            return IsLoaded;
        }

        private void SearchFromNode(HtmlNode baseNode)
        {
            var nodes = Enumerable.Empty<HtmlNode>();

            if (!_html.DocumentNode.HasChildNodes)
                ParseHtml();

            if (chkXPath.IsChecked == true)
                nodes = baseNode.SelectNodes(txtSearchTag.Text);
            else
                nodes = baseNode.Descendants(txtSearchTag.Text);

            if (nodes == null) return;

            listResults.Items.Clear();

            foreach (var node in nodes)
            {
                var tr = new NodeTreeView { BaseNode = node };
                var lvi = new ListBoxItem();
                var pnl = new StackPanel();
                pnl.Children.Add(new Label
                                     {
                                         Content =
                                             string.Format("id:{0} name:{1} children{2}", node.Id, node.Name,
                                                           node.ChildNodes.Count),
                                         FontWeight = FontWeights.Bold
                                     });
                pnl.Children.Add(tr);
                lvi.Content = pnl;
                listResults.Items.Add(lvi);
            }
            tabControl1.SelectedItem = tabSearchResults;
        }

 /// <summary>
 /// NOTE: only works for ONE classname. If several TODO: create another method.. 
 /// </summary>
 /// <param name="tagName"></param>
 /// <param name="className"></param>
 /// <returns></returns>
 public HtmlNode GetSingleNodeByClassName(string tagName, string className, HtmlNode node)
 {
     //TODO: fire a warning if any single methods would have produced more than one result, could mean the layout has changed
     return node.Descendants()
             .Where(x => x.Name == tagName && x.Attributes.Contains("class")
             && x.Attributes["class"].Value.Split().Contains(className)).SingleOrDefault();
 }