DataTable loadDataTableFromSelection(FileHelper fileInput, XmlDocument document)
{
XpathSelector selection = new XpathSelector(XElement.Parse(document.InnerXml), XElement.Load(fileInput.Xpath));
DataTable result = selection.SelectData();
result = result.Rows.Cast <DataRow>()
.Where(row => !row.ItemArray.All(field => field is DBNull || string.IsNullOrWhiteSpace(field as string)))
.CopyToDataTable();
XmlHelper.RemoveDuplicatesFromDataTable(ref result);
return(result);
}
public string ParseHtml(string html, string url = "")
{
var article = XpathSelector.Get(html, "//article[contains(@class, 'b-singlepost-body')]").First();
var innerHtml = article.InnerHtml
.Replace("-Москва", "")
.Replace("ПФК", "")
.Replace("СКА-Хабаровск", "СКА Хабаровск")
;
MatchCollection matches = new Regex(@">\s*(?<number>\d+)[.:]\s*(\w*\s+)?(?<owner>[А-Яа-я]+)\s*(\(\w*\))?\s*[:-]\s*(?<guest>[А-Яа-я]+)").Matches(innerHtml);
string resultMessage = "";
using (IUnitOfWork uow = unitOfWorkFactory.Create())
{
foreach (System.Text.RegularExpressions.Match regexMatch in matches)
{
var number = int.Parse(regexMatch.Groups["number"].Value);
var owners = regexMatch.Groups["owner"].Value;
var guests = regexMatch.Groups["guest"].Value;
var match = queryFactory.GetMatchByNumber(number).Execute();
if (match == null)
{
var guestCommand = queryFactory.GetCommandByName(guests).Execute();
if (guestCommand == null)
{
uow.Save(guestCommand = new Command {
Name = guests
});
}
var ownerCommand = queryFactory.GetCommandByName(owners).Execute();
if (ownerCommand == null)
{
uow.Save(ownerCommand = new Command {
Name = owners
});
}
uow.Save(new Match
{
Number = number,
Owners = ownerCommand,
Guests = guestCommand
});
}
resultMessage += string.Format("<br/>{0}. {1} - {2}", number, owners, guests);
}
uow.Commit();
}
resultMessage += "<br/>";
List <HtmlNode> comments = XpathSelector.Get(html, "//div[contains(@class, 'b-tree-twig')]").ToList();
int forecastCount = 0;
foreach (HtmlNode comment in comments)
{
var nameNode = XpathSelector.Get(comment.OuterHtml, "//a/b/text()").FirstOrDefault();
if (nameNode == null)
{
continue;
}
string name = nameNode.InnerText;
var singleOrDefault = XpathSelector.Get(comment.OuterHtml, "//div[@class='b-leaf-article']").SingleOrDefault();
if (singleOrDefault != null)
{
string content = singleOrDefault.InnerHtml;
using (IUnitOfWork uow = unitOfWorkFactory.Create())
{
Ljuser user = queryFactory.GetLjuserByName(name).Execute() ?? new Ljuser {
Name = name
};
List <Forecast> forecasts = Parse(content, user).ToList();
if (forecasts.Count() != 8)
{
string message = string.Format("Юзер {0} сделал {1} прогнозов", user.Name, forecasts.Count());
resultMessage += message + Environment.NewLine;
log.Warn(message);
log.WarnFormat(content);
}
if (forecasts.Any())
{
int addForecasts = user.AddForecasts(forecasts);
forecastCount += addForecasts;
if (addForecasts != 8)
{
string message = string.Format("Юзеру {0} добавлено {1} прогнозов", user.Name, addForecasts);
resultMessage += message + Environment.NewLine;
log.Warn(message);
}
uow.Save(user);
}
uow.Commit();
}
}
}
if (comments.Count() != 25)
{
log.WarnFormat("на странице {0} комментариев{1}", comments.Count(), Environment.NewLine);
log.WarnFormat(url);
}
resultMessage += string.Format("Распарсено {0} комментариев {1}", comments.Count(), Environment.NewLine);
resultMessage += string.Format("добавлено {0} новых прогнозов {1}", forecastCount, Environment.NewLine);
return(resultMessage);
}
public string ParseResult(string html)
{
string resultMessage = "";
IEnumerable <HtmlNode> matchesTr = XpathSelector.Get(html, "//table[@class='stat-table']/tbody/tr");
using (IUnitOfWork uow = unitOfWorkFactory.Create())
{
IEnumerable <Match> allMatches = queryFactory.FindAll <Match>().Execute().ToList();
foreach (HtmlNode matchTr in matchesTr)
{
var scores = XpathSelector.Get(matchTr.OuterHtml, "//td[@class='score-td']//a").Single().InnerText.Split(new[] { ":" }, StringSplitOptions.RemoveEmptyEntries);
int ownersGoals;
if (int.TryParse(scores.First(), out ownersGoals) == false)
{
continue;
}
var owners = XpathSelector.Get(matchTr.OuterHtml, "//td[@class='owner-td']//a").Single().InnerText.Replace(" Москва", "");
var guests = XpathSelector.Get(matchTr.OuterHtml, "//td[@class='guests-td']//a").Single().InnerText.Replace(" Москва", "");
var dateString = XpathSelector.Get(matchTr.OuterHtml, "//td[@class='name-td alLeft']").Single().InnerText;
Match matchFromDb = allMatches.SingleOrDefault(x => x.Guests.Name == guests && x.Owners.Name == owners);
if (matchFromDb != null)
{
var guestsGoals = int.Parse(scores.Last());
DateTime date = DateTime.Parse(dateString.Replace("|", " "));
if (matchFromDb.OwnersGoals.HasValue && matchFromDb.GuestsGoals.HasValue)
{
if ((matchFromDb.Guests.Name != guests ||
matchFromDb.GuestsGoals != guestsGoals ||
matchFromDb.Owners.Name != owners ||
matchFromDb.OwnersGoals != ownersGoals))
{
throw new ArgumentException(string.Format("Результаты матча {0} не совпадают!", owners + "-" + guests));
}
}
else
{
matchFromDb.GuestsGoals = guestsGoals;
matchFromDb.OwnersGoals = ownersGoals;
matchFromDb.Date = date;
resultMessage += string.Format("{0}. {1} - {2} {3}:{4}{5}", matchFromDb.Number, owners, guests, ownersGoals, guestsGoals, Environment.NewLine);
}
}
// else
// {
// uow.Save(new Match
// {
// Number = number,
// OwnersGoals = ownersGoals,
// GuestsGoals = guestsGoals,
// Guests = queryFactory.GetCommandByName(guests).Execute(),
// Owners = queryFactory.GetCommandByName(owners).Execute(),
// Date = date
// });
// }
}
uow.Commit();
}
return(resultMessage);
}