C# HtmlNode.SelectHtmlNodes方法代码示例

        private static void HandleRow(HtmlNode row)
        {
            var onclick = row.Attributes["onclick"].Value;
            var url = onclick.Split('\'')[1];
            string ftid = url.Split('{')[1].Replace("}", "");

            var db = new DBDataContext();
            var trip = db.CommitteeTrips.SingleOrDefault(_ => _.FTId == ftid);

            //if (trip != null && trip.ActualExpenses.HasValue && trip.ActualExpenses != 0)
            //{
            //    // this trip is prolly completely accounted for
            //    return;
            //}

            var cells = row.SelectHtmlNodes("td");

            var startstring = cells.ElementAt(0).InnerText.Trim();
            var startdate = DateTime.ParseExact(startstring, "dd-MM-yyyy", null);

            var endstring = cells.ElementAt(1).InnerText.Trim();
            var enddate = DateTime.ParseExact(endstring, "dd-MM-yyyy", null);

            if (startdate.Year < 1900 || enddate.Year < 1900)
            {
                return;
            }

            var commname = cells.ElementAt(2).InnerText.Trim();

            var committee = Scrape2009.GetCommitteeId(commname, db);

            var purpose = cells.ElementAt(5).InnerText.Trim();
            var place = cells.ElementAt(3).InnerText.Trim();
            if (place.ToLower().Contains("aflyst") || purpose.ToLower().Contains("aflyst"))
            {
                // give up
                return;
            }

            var doc = Scrape2009.GetDoc("http://www.ft.dk" + url);

            var menudiv = doc.DocumentNode.SelectSingleNode("//div[@id='menuSkip']");
            if (menudiv.InnerText.ToLower().Contains("afbud"))
            {
                return;
            }

            var participantnode = menudiv.SelectHtmlNodes("p/h3").
                SingleOrDefault(_ => _.InnerText.Trim() == "Deltagere");

            if (participantnode == null)
            {
                // no politicians went, discard
                return;
            }

            var participants = participantnode.
                NextSibling.SelectHtmlNodes("li/a").
                Select(_ => _.Attributes["href"].Value);

            var polids = participants.Select(_ => Scrape2009.GetPoliticianByUrl(_, db));

            var otherparticipantnode = menudiv.SelectHtmlNodes("p/h3").
                SingleOrDefault(_ => _.InnerText.Trim() == "Øvrige deltagere");

            var othercount = 0;
            if (otherparticipantnode != null)
            {
                othercount = OtherMemberCount(otherparticipantnode);
            }

            var budgetstring = menudiv.SelectHtmlNodes("p/h3").
                Single(_ => _.InnerText.Trim() == "Budget").
                NextSibling.InnerText.Trim().Split(' ')[0]
                .Replace(".", "").Replace(",", "");

            var spendstring = menudiv.SelectHtmlNodes("p/h3").
                Single(_ => _.InnerText.Trim() == "Regnskab").
                NextSibling.InnerText.Trim().Split(' ')[0]
                .Replace(".", "").Replace(",", "");

            var provider = new CultureInfo("da-dk");
            //var provider = new CultureInfo("en-us");
            var budget = decimal.Parse(budgetstring, provider);
            var spend = decimal.Parse(spendstring, provider);

            lock (dblock)
            {
                if (trip == null)
                {
                    trip = new CommitteeTrip();
                    db.CommitteeTrips.InsertOnSubmit(trip);
                }

                trip.ActualExpenses = spend;
                trip.Budget = budget;
                trip.CommitteeId = committee;
                trip.EndDate = enddate;
                trip.Place = place;
//.........这里部分代码省略.........