C# HtmlNode.SelectNodes方法代码示例

        internal static List<ChannelItemInfo> ParseNode(HtmlNode node, bool abroadOnly)
        {
            var items = new List<ChannelItemInfo>();

            if (node == null)
                return items;

            var playableArticles = node.SelectNodes(".//article[contains(@class, 'playJsInfo-Core') or contains(@class, 'slick_item')]");

            if (playableArticles != null)
                foreach (var article in playableArticles)
                {
                    var playable = ParsePlayableArticle(article, abroadOnly);
                    if (playable != null)
                        items.Add(playable);
                }

            var folderArticles = node.SelectNodes(".//article[not(contains(@class, 'playJsInfo-Core') or contains(@class, 'slick_item'))]");

            if (folderArticles != null)
                foreach (var article in folderArticles)
                {
                    var folder = ParseFolderArticle(article);
                    if (folder != null)
                        items.Add(folder);
                }


            return items;

        }

        private List<Movie> ParseMovieListHtml(HtmlNode htmlNode, string xPath)
        {
            try
            {
                var movies = new List<Movie>();
                var hnc = htmlNode.SelectNodes(xPath);
                var hnc2 = htmlNode.SelectNodes("//div/a/img");
                if (hnc.Count < 1)
                    return null;

                for (int i = 0; i < hnc.Count; i++)
                {
                    var node1 = hnc2[i];
                    var node = hnc[i];
                    var movie = new Movie();
                    var hac = node1.Attributes;

                    movie.Grade =node.SelectSingleNode("div[@class='fm-movie-desc']/div/span[@class='fm-rating']").InnerText.Replace("\n", "").RemoveSpace().Trim();
                    movie.Image = hac[0].Value.Replace("-poster100","").Trim();
                    movie.Name = node.SelectNodes("div[@class='fm-movie-desc']/div")[0].InnerText.Replace("\n","").RemoveSpace().Trim();
                    movie.Director =node.SelectNodes("div[@class='fm-movie-desc']/div")[2].InnerText.Replace("\n", "").RemoveSpace().Trim();
                    movie.Story = node.SelectNodes("div[@class='fm-movie-desc']/div")[3].InnerText.Replace("\n", "").RemoveSpace().Trim(); 
                    movie.Actor =node.SelectNodes("div[@class='fm-movie-desc']/div")[4].InnerText.Replace("\n","").RemoveSpace().Trim();
                    movie.DetailUrl = node.SelectSingleNode("div[@class='fm-movie-cover']/a").Attributes["href"].Value.Trim();
                    movies.Add(movie);
                }
                return movies;
            }
            catch (Exception e)
            {
                throw e;
            }
        }

        public YoutubeVideoEntry(HtmlNode node)
        {
            var url_node = node.SelectNodes(".//a[@href]");

            if (url_node != null)
            {
                var url_value = url_node.FirstOrDefault().Attributes["href"].Value;
                var splitIndex = url_value.IndexOf("&");

                if (splitIndex > 0)
                {
                    url = "http://www.youtube.com" + url_value.Substring(0, splitIndex);
                }
            }

            var title_node = node.SelectNodes(".//span[contains(@class, 'video-title')]");

            if (title_node != null)
                title = title_node.FirstOrDefault().InnerText;

            if (!String.IsNullOrEmpty(title))
                title = title.Trim();

            var img_node = node.SelectNodes(".//img[@src]");

            if (img_node != null)
                imageUrl = "http:" + img_node.FirstOrDefault().Attributes["src"].Value;
        }

 protected override string retrieveTitle(HtmlNode node)
 {
     string title = node.SelectNodes("//div[contains(@class, 'entry-content')]//strong"
         + "|//div[contains(@class, 'entry-content')]//b")?.First()?.InnerText ?? "";
     node.SelectNodes("//div[contains(@class, 'entry-content')]//strong"
         + "|//div[contains(@class, 'entry-content')]//b")?.First()?.Remove();
     return title;
 }

        private EuroMillionsResult ParseResultSection(HtmlNode section)
        {
            var date = DateTime.ParseExact(section.SelectSingleNode(".//div[@class = 'floatLeft']/a").InnerText, "dd/MM/yyyy", CultureInfo.InvariantCulture);

            var balls = section.SelectNodes(".//td[@class = 'euro-ball-s']").Select(x => Convert.ToInt32(x.InnerText));
            var bonusBalls = section.SelectNodes(".//td[@class = 'euro-lucky-star-s']").Select(x => Convert.ToInt32(x.InnerText));

            return new EuroMillionsResult(date, 0, balls.ToList(), bonusBalls.ToList());
        }

        /// <summary>
        /// Run xpath from html or node
        /// </summary>
        private List<KeyValuePair<string, object>> run(HtmlNode node)
        {
            Factory.Instance.iInfo(string.Format("Running xpathSingle id : {0}", rule.id));

            if (node == null)
                return new List<KeyValuePair<string, object>>();

            //Get all attriibutes by type and save to List<KeyValuePair<string, object>>
             foreach (Db.xpathSingleAttributes attr in rule.attributes)
            {
                object val = null;

                if (attr.getType == Db.xpathSingleAttributesGetType.nodeCollection)
                    val = node.SelectNodes(attr.xpath);
                else if (attr.getType == Db.xpathSingleAttributesGetType.count)
                {
                    HtmlNodeCollection c = node.SelectNodes(attr.xpath);
                    if (c != null)
                        val = c.Count.ToString();
                }
                else
                {
                    string val2 = string.Empty;
                    HtmlNode n = node.SelectSingleNode(attr.xpath);
                    if (n != null)
                    {
                        if (attr.getType == Db.xpathSingleAttributesGetType.singleNode)
                            val = n;
                        else
                        {
                            if (attr.getType == Db.xpathSingleAttributesGetType.text)
                                val2 = n.InnerText.Trim();

                            if (attr.getType == Db.xpathSingleAttributesGetType.html)
                                val2 = n.InnerHtml.Trim();

                            if (attr.getType == Db.xpathSingleAttributesGetType.attribute)
                            {
                                if (n.Attributes[attr.attributeName] != null)
                                    val2 = n.Attributes[attr.attributeName].Value;
                            }



                            val = postProcessResult(val2, attr);
                            if(attr.getType != Db.xpathSingleAttributesGetType.html && attr.getType != Db.xpathSingleAttributesGetType.nodeCollection && 
                                attr.getType != Db.xpathSingleAttributesGetType.singleNode)
                            Factory.Instance.iInfo(string.Format("{0} = {1}",attr.id,val));
                        }
                    }
                }
                res.Add(new KeyValuePair<string, object>(attr.id, val));
            }
                       
            return res;
        }

 public HtmlNodeCollection GetDishInfoList(HtmlNode dishTypeNode)
 {
     var baseCollectionSite = new BaseCollectionSite(PageUrl);
     var dishNodeList = dishTypeNode.SelectNodes(DishesPath());
     if (dishNodeList == null || dishNodeList.Count <= 0)
     {
         return new HtmlNodeCollection(null);
     }
     var scripNode = dishTypeNode.SelectSingleNode(@"./../../../../..//div[@class='rec-dishes tab-item active']/div[@class='pic-list J_toggle']/ul/script");
     if (scripNode != null && !string.IsNullOrWhiteSpace(scripNode.InnerText))
     {
         var liNodeList = baseCollectionSite.BaseHtmlNodeCollection(scripNode.InnerText);
         if (liNodeList != null)
         {
             var dishLiList = liNodeList.SelectNodes(".//li");
             if (dishLiList != null)
             {
                 foreach (var dishLi in dishLiList)
                 {
                     dishNodeList.Add(dishLi);
                 }
             }
         }
     }
     return dishNodeList;
 }

        public bool SetValue(HtmlNode n, string value)
        {
            if (n is HtmlNode && n.Name == "select")
            {
                foreach (HtmlNode o in n.SelectNodes("option"))
                {
                    o.SetAttributeValue("selected", o.GetAttributeValue("value", "").Equals(value) ? "selected" : "");
                }
                return true;
            }

            if (n is HtmlNode && n.Name == "input")
            {
                switch (n.GetAttributeValue("type", ""))
                {
                    case "radio":
                        n.SetAttributeValue("checked", n.GetAttributeValue("value", "").Equals(value) ? "checked" : "");
                        break;
                    default:
                        n.SetAttributeValue("value", value);
                        break;
                }
                n.SetAttributeValue("value", value);
                return true;
            }

            return false;
        }

        private void ParseRegionElement(HtmlNode region)
        {
            var regionTitle = region.SelectSingleNode("h2").InnerText;

            foreach (var server in region.SelectNodes(".//div[@class=\"server\" or @class=\"server alt\"]"))
            {
                var serverName = server.SelectSingleNode(".//div[@class=\"server-name\"]").InnerText.Trim();

                var pollCategoryValue = new PollCategoryValue();
                var possibleCategoryMatch = Categories.FirstOrDefault(p => string.Compare(p.Region, regionTitle, true) == 0 && string.Compare(p.ServerCategory, serverName) == 0);
                if (possibleCategoryMatch == null)
                    continue;

                pollCategoryValue.CategoryID = possibleCategoryMatch.PollCategoryID;
                pollCategoryValue.Status = PollStatusType.Unknown;
                pollCategoryValue.CreatedTime = DateTime.Now;
                foreach (var div in server.SelectNodes("div"))
                {
                    if (div.OuterHtml.Contains("status-icon"))
                    {
                        pollCategoryValue.Status = div.OuterHtml.Contains("status-icon up") ? PollStatusType.Up : PollStatusType.Down;
                    }
                }
                DB.InsertPollCategoryValue(pollCategoryValue);
            }
        }

 public List<FormElement> GetOptions(HtmlNode htmlNode)
 {
     List<FormElement> options = new List<FormElement>();
     HtmlNodeCollection nodeTags = htmlNode.SelectNodes(@".//option");
     if (nodeTags != null)
     {
         foreach (HtmlNode node in nodeTags)
         {
             string id = node.GetAttributeValue("id", "");
             string type = "option";
             string name = node.GetAttributeValue("name", "");
             string value = node.GetAttributeValue("value", "");
             bool chk = node.Attributes["selected"] != null;
             FormElement el = new FormElement();
             el.Id = id;
             el.Type = type;
             el.Name = node.NextSibling.InnerText;
             el.Value = value;
             el.Type = type;
             el.Checked = chk;
             options.Add(el);
         }
     }
     return options;
 }

        /// <summary>
        /// Run xpath from html or node
        /// </summary>
        public List<List<KeyValuePair<string, object>>> run(HtmlNode node)
        {
            Factory.Instance.iInfo(string.Format("Running xpathCollection id : {0}", rule.id));

            HtmlNodeCollection nodes = new HtmlNodeCollection(node);
            HtmlNodeCollection n2 = node.SelectNodes(rule.xpath);
            if (n2 != null)
            {
                foreach (HtmlNode n in n2)
                    nodes.Add(n);
            }

            //run
            if (node != null)
            {

                foreach (HtmlNode n in nodes)
                {
                    List<KeyValuePair<string, object>> last_val = null;
                    if (rule.xpathSingle != null)
                    {
                        XPathSingle xs = new XPathSingle(rule.xpathSingle, last_val);
                        last_val = (List<KeyValuePair<string, object>>)xs.Run(n);
                        res.Add(last_val);
                    }
                }
            }
            return res;
        }

        private static Nhl_Players_Rtss_Skater MapHtmlRowToModel(HtmlNode row, NhlSeasonType nhlSeasonType, int year)
        {
            HtmlNodeCollection tdNodes = row.SelectNodes(@"./td");

            Nhl_Players_Rtss_Skater model = new Nhl_Players_Rtss_Skater();

            model.NhlSeasonType = nhlSeasonType;
            model.Year = year;

            model.Number = 0;
            model.Name = tdNodes[1].InnerText;
            model.Team = tdNodes[2].InnerText;
            model.Position = tdNodes[3].InnerText;

            model.GamesPlayed = ConvertStringToInt(tdNodes[4].InnerText);
            model.Hits = ConvertStringToInt(tdNodes[5].InnerText);
            model.BlockedShots = ConvertStringToInt(tdNodes[6].InnerText);
            model.MissedShots = ConvertStringToInt(tdNodes[7].InnerText);
            model.Giveaways = ConvertStringToInt(tdNodes[8].InnerText);
            model.Takeaways = ConvertStringToInt(tdNodes[9].InnerText);
            model.FaceoffsWon = ConvertStringToInt(tdNodes[10].InnerText);
            model.FaceoffsLost = ConvertStringToInt(tdNodes[11].InnerText);
            model.FaceoffsTaken = ConvertStringToInt(tdNodes[12].InnerText);
            model.FaceoffWinPercentage = Convert.ToDouble(tdNodes[13].InnerText);
            model.PercentageOfTeamFaceoffsTaken = Convert.ToDouble(tdNodes[14].InnerText);
            model.Shots = ConvertStringToInt(tdNodes[15].InnerText);
            model.Goals = ConvertStringToInt(tdNodes[16].InnerText);
            model.ShootingPercentage = Convert.ToDouble(tdNodes[17].InnerText);

            return model;
        }

        /// <summary>
        /// Identify Macros
        /// </summary>
        /// <param name="skeleton">skeleton</param>
        /// <param name="pageType">page Type</param>
        /// <returns></returns>
        private void IdentifyMacros(HtmlNode skeleton, PageType pageType)
        {
            var menuProperties = new List<PropertyDTO>();
            var doc = skeleton.OwnerDocument;
            var propertyFactory = factory.PropertyFactory;
            foreach (var menuNode in pageType.MacroXpaths
                .SelectMany(xpath => skeleton.SelectNodes(xpath))
                .Where(n => n != null && n.ParentNode != null))
            {
                var property = propertyFactory.GetNew();
                menuProperties.Add(property);

                var propertyNode = doc.CreateTextNode(property.TemplateReference);
                menuNode.ParentNode.ReplaceChild(propertyNode, menuNode);
            }

            var macros = menuProperties.Select(p => new Definition
            {
                Number = p.Number,
                Name = p.Name,
                TemplateReference = p.TemplateReference,
                IsMacro = true
            });
            pageType.Definitions.AddRange(macros);
        }

        private static Nhl_Players_Bio_Goalie MapHtmlRowToModel(HtmlNode row, NhlSeasonType nhlSeasonType, int year)
        {
            HtmlNodeCollection tdNodes = row.SelectNodes(@"./td");

            Nhl_Players_Bio_Goalie model = new Nhl_Players_Bio_Goalie();

            model.NhlSeasonType = nhlSeasonType;
            model.Year = year;

            model.Number = ConvertStringToInt(tdNodes[0].InnerText);
            model.Name = tdNodes[1].InnerText;
            model.Team = tdNodes[2].InnerText;
            model.Position = "G";
            model.DateOfBirth = Convert.ToDateTime(tdNodes[3].InnerText.Replace("'", "/"));
            model.BirthCity = tdNodes[4].InnerText;
            model.StateOrProvince = tdNodes[5].InnerText;
            model.BirthCountry = tdNodes[6].InnerText;
            model.HeightInches = ConvertStringToInt(tdNodes[7].InnerText);
            model.WeightLbs = ConvertStringToInt(tdNodes[8].InnerText);
            model.Catches = tdNodes[9].InnerText;
            model.Rookie = tdNodes[10].InnerText;
            model.DraftYear = ConvertStringToInt(tdNodes[11].InnerText);
            model.DraftRound = ConvertStringToInt(tdNodes[12].InnerText);
            model.DraftOverall = ConvertStringToInt(tdNodes[13].InnerText);

            model.GamesPlayed = ConvertStringToInt(tdNodes[14].InnerText);
            model.Wins = ConvertStringToInt(tdNodes[15].InnerText);
            model.Losses = ConvertStringToInt(tdNodes[16].InnerText);
            model.OTSOLosses = ConvertStringToInt(tdNodes[17].InnerText);
            model.GAA = Convert.ToDouble(tdNodes[18].InnerText);
            model.SavePercentage = Convert.ToDouble(tdNodes[19].InnerText);
            model.Shutouts = ConvertStringToInt(tdNodes[20].InnerText);

            return model;
        }

        private IList<Achievement> ParseSubAchievements(Achievement achievement, HtmlNode subAchievementNode)
        {
            IList<Achievement> subAchievements = new List<Achievement>();
            HtmlNodeCollection achievements = subAchievementNode.SelectNodes("./li");
            if (achievements != null)
            {
                foreach (HtmlNode subNode in achievements)
                {
                    // TODO : Parse out achievement id
                    if ( subNode.Attributes["onmousemove"] != null )
                    {
                        Match match = parseTooltip.Match(subNode.Attributes["onmousemove"].Value);
                        if ( match.Success )
                        {
                            string subAchievementId = match.Groups["achievementid"].Value;
                            int blizzardId = 0;
                            int.TryParse(subAchievementId, out blizzardId);
                            Achievement subAchievement = _service.Find(blizzardId);
                            if (subAchievement == null)
                            {
                                subAchievement = new Achievement() { BlizzardID = blizzardId };
                                subAchievement.Name = GetValueAsString(subNode, ".//h3");
                                subAchievement.Description = GetValueAsString(subNode, ".//div[@class='color-tooltip-yellow']");
                                subAchievement.Points = GetValueAsInt32(subNode, ".//span[@class='points border-3']");
                                _service.Save(subAchievement);
                            }
                            achievement.Points = achievement.Points - subAchievement.Points;
                            subAchievements.Add(subAchievement);
                        }
                    }
                }
            }

            return subAchievements;
        }