/// <summary>
/// Parses the HTML to return the games and results, grouped by league
/// </summary>
/// <param name="html"></param>
/// <returns></returns>
/// <remarks>For a version that handles days in the future (and no results parsing), see <see cref="ScheduleParser.GamesScheduleFromDailyHtml(string)"/></remarks>
public static List <LeagueGames> GamesResultsFromHtml(string html)
{
var results = new List <LeagueGames>();
var svCulture = System.Globalization.CultureInfo.GetCultureInfo("sv-SE");
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(html);
LeagueGames currentLeage = null;
string day = ""; //empty string so compiler will STFU
foreach (var node in doc.DocumentNode.SelectNodes("//table[contains(@class,'tblContent')]/tr"))
{
if (node.ChildNodes.Any(n => n.Name == "th"))
{
if (node.ChildNodes.Any(n => n.HasClass("tdTitleRight")))
{
day = node.ChildNodes.First(n => n.HasClass("tdTitleRight")).InnerText;
}
continue;
}
if (node.ChildNodes.Any(n => n.Attributes["colspan"]?.Value == "5"))
{
if (currentLeage != null)
{
results.Add(currentLeage);
}
currentLeage = new LeagueGames(System.Net.WebUtility.HtmlDecode(node.SelectSingleNode("td/a").InnerText.Trim()));
continue;
}
var strDate = day + " " + node.SelectSingleNode("td[1]").InnerText.Trim();
var resultsNode = node.SelectSingleNode("td[3]/a");
string url = node.SelectSingleNode("td/a")?.Attributes["href"].Value.Replace("
", "");
var game = new Game()
{
Tid = DateTime.ParseExact(strDate, "yyyy-MM-dd HH:mm", svCulture),
Lag = ParserServices.CleanTeams(node.SelectSingleNode("td[2]").InnerText),
Location = System.Net.WebUtility.HtmlDecode(node.SelectSingleNode("td[4]").InnerText)
};
game.End = game.Tid.AddHours(2); //basically just for compatibility reasons, no one should care if it's inaccurate
if (resultsNode != null && !string.IsNullOrEmpty(resultsNode.InnerText))
{
game.Results = resultsNode.InnerText;
}
if (!string.IsNullOrEmpty(url))
{
game.Url = ParserServices.CleanURL(url);
}
else
{
game.Url = "";
}
game.Uid = "swehockey_" + ParserServices.GetGameId(game.Url); //basically just for ical compatibility anyway
game.Series = currentLeage.League; //redundant, but consistent
currentLeage.Games.Add(game);
}
if (currentLeage != null)
{
results.Add(currentLeage);
} //add the last league
return(results);
}
/// <summary>
/// Parse HTML to get a list of all games it contains
/// </summary>
/// <param name="html"></param>
/// <returns></returns>
public static List <Game> GamesScheduleFromHtml(string html, ParserMode mode = ParserMode.Unknown)
{
List <Game> games = new List <Game>();
var svCulture = System.Globalization.CultureInfo.GetCultureInfo("sv-SE");
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(html);
string lastDate = null; //needed to keep track of dates in SDHL/HA/SHL mode, since the date isn't given in every row
foreach (var node in doc.DocumentNode.SelectNodes("//table[contains(@class,'tblContent')]/tr"))
{
if (node.ChildNodes.Any(n => n.Name == "th"))
{
if (mode == ParserMode.Unknown && node.ChildNodes.Any(n => n.HasClass("tdHeader")))
{
//Use this to get the layout of the rest of the table
if (node.ChildNodes.Last().InnerText == "Group")
{
mode = ParserMode.HockeyEttan;
}
else
{
mode = ParserMode.SHL;
}
}
continue;
}
string strDate;
string teams;
string location;
string series = null; //assign null explicitly to tell compiler to STFU
string url;
if ((int)mode == 1)
{
strDate = node.SelectSingleNode("td/div[@class='dateLink']/span").InnerText;
teams = ParserServices.CleanTeams(node.SelectSingleNode("td[3]").InnerText);
location = System.Net.WebUtility.HtmlDecode(node.SelectSingleNode("td[7]").InnerText);
series = System.Net.WebUtility.HtmlDecode(node.SelectSingleNode("td[8]").InnerText);
url = node.SelectSingleNode("td/a")?.Attributes["href"].Value.Replace("
", "");
}
else if ((int)mode == 2)
{
if (!string.IsNullOrEmpty(node.SelectSingleNode("td[2]").InnerText))
{
lastDate = node.SelectSingleNode("td[2]/bold").InnerText + " ";
}
strDate = lastDate + node.SelectSingleNode("td/div[@class='dateLink']/span").InnerText;
teams = ParserServices.CleanTeams(node.SelectSingleNode("td[4]").InnerText);
location = System.Net.WebUtility.HtmlDecode(node.SelectSingleNode("td[8]").InnerText);
url = node.SelectSingleNode("td/a")?.Attributes["href"].Value.Replace("
", "");
}
else
{
throw new Exception("Could not figure out how to parse this HTML");
}
Game game = new Game()
{
Lag = teams,
Location = location,
Tid = DateTime.ParseExact(strDate, "yyyy-MM-dd HH:mm", svCulture)
};
game.End = game.Tid.AddHours(2); //basically just for ical compatibility anyway
if (!string.IsNullOrEmpty(url))
{
game.Url = ParserServices.CleanURL(url);
}
else
{
game.Url = "";
}
game.Uid = "swehockey_" + ParserServices.GetGameId(game.Url); //basically just for ical compatibility anyway
if (!string.IsNullOrEmpty(series))
{
game.Series = series;
}
games.Add(game);
}
return(games);
}