Develop Reference

Linq an aggregate with a grouping

I'm translating a query and here's the original:
select top 5 t.usrID, u.FirstName, u.LastName, t.cnt as sCount
from (
select usrID, COUNT(rID) as cnt
from sessions as s where s.sDate > DATEADD(yy, -1, getdate())
group by usrID
) as t
inner join users as u on t.usrID = u.usrID
order by t.cnt desc
Here's what I have so far:
var topUser = (from p in _context.Sessions
where p.SDate > DateTime.Now.AddYears(-1)
join c in _context.Users on p.UsrId equals c.UsrId into j1
from j2 in j1.DefaultIfEmpty()
// group j2 by p.UsrId into grouped
select new
{
p.UsrId,
j2.FirstName,
j2.LastName,
cnt = p.RId
})
//.OrderBy(d => d.cnt)
//.GroupBy(o => o.UsrId)
.Take(5);
I'm having trouble figuring out how to include count() and group by clauses. When I include groupBy my other columns disappear. Thank you.

This is the answer to your answer - not to your original query. I would put it as comment, but without formatting it's hard to explain
Assuming User object has collection of Session your first statement can be drastically simplified:
var topUsers = _context.Sessions
.Where(s => s.SDate > DateTime.Now.AddYears(-1))
.Select(s => new
{
s.UsrId,
cnt = s.User.Sessions.Count(u => u.UsrId == s.UsrId)
})
.OrderByDescending(s => s.cnt)
.Take(5);
You can shape the results to get a ViewModel that also has FirstName and LastName. It all boils down to defining a model with one-to-many relationship

var topUser = (from p in _context.Sessions
where p.SDate > DateTime.Now.AddYears(-1)
join c in _context.Users on p.UsrId equals c.UsrId into j1
from j2 in j1.DefaultIfEmpty()
group j2 by p.UsrId into g
select new
{
UsrId = g.Key,
FirstName = g.Select(x => x.FirstName).FirstOrDefault(),
LastName = g.Select(x => x.LastName).FirstOrDefault(),
sCount = g.Count()
})
.OrderByDescending(d => d.sCount)
.Take(5);

Alright so it's not the most efficient but it works:
var topUsers = _context.Sessions
.Where(s => s.SDate > DateTime.Now.AddYears(-1))
.GroupBy(s => s.UsrId)
.Select(ws => new { ws.Key, cnt = ws.Count() })
.OrderByDescending(s => s.cnt)
.Take(5);
var topNamedUsers = topUsers.Join(_context.Users, ws => ws.Key, ud => ud.UsrId, (ws, ud) => new { ws.Key, ud.FirstName, ud.LastName, ws.cnt });

LINQ 3 Inner Joins with 1 Left Outer Join

Wondering why LINQ doesn't have a Left Join method. I've been trying to figure this out with myriad examples on SO, but no such luck. The other examples show simple examples with one join. If I group the joins then I only get references to the TradeCountries table in the select statement.
Being new to LINQ, I could've had this done 4 hours ago with a simple SELECT statement, but here I'm am trying to figure out why the LeftJoin method was left out of LINQ.
What does the line with "LeftJoin" need to be changed to make this work?
/*
* GetTop5Distributors
#param int array of series IDs
*/
public List<TopDistributors> Get5TopDistributors(IEnumerable<int> seriesIds)
{
_context = new MySQLDatabaseContext();
var result = _context.TradesTrades
.Join(_context.TradesSeries, tt => tt.SeriesId, ts => ts.Id, (tt, ts) => new { tt, ts })
.Join(_context.TradesTradeDistributors, tsd => tsd.tt.Id, ttd => ttd.TradeId,
(tsd, ttd) => new { tsd, ttd })
.Join(_context.TradesOrganisations, tsdto => tsdto.ttd.DistributorId, to => to.Id,
(tsdto, to) => new { tsdto, to })
.LeftJoin(_context.TradesCountries, tsdc => tsdc.to.CountryId, tc => tc.Id,
(tsdc, tc) => new {tsdc, tc})
.Where(x => seriesIds.Contains(x.tsdc.tsdto.tsd.tt.SeriesId))
.Where(x => x.tsdc.tsdto.tsd.tt.FirstPartyId == null)
.Where(x => x.tsdc.tsdto.tsd.tt.Status != "closed")
.Where(x => x.tsdc.tsdto.tsd.tt.Status != "cancelled")
.GroupBy(n => new { n.tsdc.tsdto.tsd.tt.SeriesId, n.tsdc.tsdto.ttd.DistributorId })
.Select(g =>
new TopDistributors
{
SeriesId = g.Key.SeriesId,
DistributorName = g.Select(i => i.tsdc.to.Name).Distinct().First(),
IsinNickname = g.Select(i => i.tsdc.tsdto.tsd.ts.Nickname).Distinct().First(),
CountryName = g.Select(i => i.tc.Name).Distinct().First(),
CommissionTotal = Math.Ceiling(g.Sum(i => i.tsdc.tsdto.ttd.Commission))
}
)
.OrderByDescending(x => x.CommissionTotal)
.Take(5)
.ToList();
return result;
}
Here's the rather simple select statement that is taking orders or magnitude too long to convert to LINQ.
SELECT
trades_trades.series_id,
trades_organisations.`name`,
trades_series.nickname,
trades_countries.name as Country_Name,
SUM(trades_trade_distributors.commission) as Commission_Total
FROM
trades_trades
JOIN trades_series
ON trades_series.id = trades_trades.series_id
JOIN trades_trade_distributors
ON trades_trades.id = trades_trade_distributors.trade_id
JOIN trades_organisations
ON trades_trade_distributors.distributor_id = trades_organisations.id
LEFT JOIN trades_countries
ON trades_organisations.country_id = trades_countries.id
WHERE trades_trades.series_id IN (
17,
18)
AND trades_trades.first_party_id IS NULL
AND trades_trades.status <> 'closed'
AND trades_trades.status <> 'cancelled'
GROUP BY trades_trades.series_id, trades_trade_distributors.distributor_id
ORDER BY Commission_Total DESC

Following my recipe, here is a more or less straightforward translation of the SQL to LINQ. I moved the where to be near what it constrains, and used let to create a convenient name for the Sum, as LINQ doesn't allow you to forward reference anonymous object members.
var ans = from tt in trades_trades
where new[] { 17, 18 }.Contains(tt.series_id) && tt.first_party_id == null &&
tt.status != "closed" && tt.status != "cancelled"
join ts in trades_series on tt.series_id equals ts.id
join ttd in trades_trade_distributors on tt.id equals ttd.trade_id
join to in trades_orginizations on ttd.distributor_id equals to.id
join tc in trades_countries on to.country_id equals tc.id into tcj
from tc in tcj.DefaultIfEmpty() // GroupJoin -> left join
group new { tt, ts, ttd, to, tc } by new { tt.series_id, ttd.distributor_id } into tradeg
let Commission_Total = tradeg.Sum(trade => trade.ttd.commission)
orderby Commission_Total descending
select new {
tradeg.Key.series_id,
tradeg.First().to.name,
tradeg.First().ts.nickname,
Country_Name = tradeg.First().tc == null ? null : tradeg.First().tc.name,
Commission_Total
};

How to join in Entity dotnet core / linq?

I've got the following query.
SELECT *
FROM Movie m INNER JOIN
(
SELECT movie_id, COUNT(movie_id) as amount
FROM Watchhistory
WHERE watch_date >= DATEADD(day,-3000, GETDATE())
GROUP BY movie_id
) as a on m.movie_id = a.movie_id
ORDER BY a.amount DESC
And the following tables:
My attempt right now:
var movies = _context.Movie
.AsNoTracking()
.Join(
_context.Watchhistory.Where(
w => w.WatchDate >= Convert.ToDateTime(DateTime.Now).AddDays(-14))
.GroupBy(w => w.MovieId)
.Select(m => new { amount = m.Count(), mID = m.Key })
.OrderByDescending(a => a.amount)
.Select(a => new { movieid = a.mID }),
m => m.MovieId, w => w.movieid, (m, w) => m);
I cant seem to get it working in c# dotnet entity. How do I use a right outer join in linq?

You rarely use JOIN in LINQ. Instead just write queries navigate the model.
so
SELECT *
FROM Movie m INNER JOIN
(
SELECT movie_id, COUNT(movie_id) as amount
FROM Watchhistory
WHERE watch_date >= DATEADD(day,-3000, GETDATE())
GROUP BY movie_id
) as a on m.movie_id = a.movie_id
ORDER BY a.amount DESC
becomes something like:
var q = from m in db.Movies
select new {m.MovieID,m.Name, amount = m.WatchHistory.Where( w => w.WatchDate >= ...).Count() };
var results = q.Where( m => m.amount > 0).OrderByDesc( m => m.amount).ToList();
David

linq after groupby unable to get column values

I am getting data from multiple tables by joining and i want to group data on particular column value but after group by statement i can access my aliases and their properties. What mistake i am making?
public List<PatientHistory> GetPatientHistory(long prid)
{
using(var db = new bc_limsEntities())
{
List<PatientHistory> result =
(from r in db.dc_tresult
join t in db.dc_tp_test on r.testid equals t.TestId into x
from t in x.DefaultIfEmpty()
join a in db.dc_tp_attributes on r.attributeid equals a.AttributeId into y
from a in y.DefaultIfEmpty()
where r.prid == prid
group new {r,t,a} by new {r.testid} into g
select new PatientHistory
{
resultid = r.resultid,
bookingid = r.bookingid,
testid = r.testid,
prid = r.prid,
attributeid = r.attributeid,
result = r.result,
Test_Name = t.Test_Name,
Attribute_Name = a.Attribute_Name,
enteredon = r.enteredon,
Attribute_Type = a.Attribute_Type
}).ToList();
return result;
}
}

You're doing this wrong way. As been said by Jon after grouping the sequences with aliases r,t,a doesn't exist. After grouping you receive the sequence g with sequances of r,t,a in each element of g. If you want get one object from each group (for example most recent) you should try this:
List<PatientHistory> result =
(from r in db.dc_tresult
join t in db.dc_tp_test on r.testid equals t.TestId into x
from t in x.DefaultIfEmpty()
join a in db.dc_tp_attributes on r.attributeid equals a.AttributeId into y
from a in y.DefaultIfEmpty()
where r.prid == prid
group new {r,t,a} by new {r.testid} into g
select new PatientHistory
{
resultid = g.Select(x => x.r.resultid).Last(), // if you expect single value get it with Single()
// .... here add the rest properties
Attribute_Type = g.Select(x => x.a.Attribute_Type).Last()
}).ToList();

I appreciated this question so I thought I would add another potential usage case. I would like feedback on what the cleanest approach is to getting table information through a group operation so that I can project later in the select operation. I ended up combining what the OP did which is to pass objects into his group clause and then used the g.Select approach suggested by YD1m to get table information out later. I have a LEFT JOIN so I'm defending against nulls :
// SQL Query
//DECLARE #idCamp as Integer = 1
//
//select *,
//(select
//count(idActivityMaster)
//FROM tbActivityMasters
//WHERE dftidActivityCategory = A.idActivityCategory) as masterCount
//FROM tbactivitycategories A
//WHERE idcamp = #idCamp
//ORDER BY CategoryName
int idCamp = 1;
var desiredResult =
(from c in tbActivityCategories
.Where(w => w.idCamp == idCamp)
from m in tbActivityMasters
.Where(m => m.dftidActivityCategory == c.idActivityCategory)
.DefaultIfEmpty() // LEFT OUTER JOIN
where c.idCamp == idCamp
group new {c, m} by new { m.dftidActivityCategory } into g
select new
{
idActivityCategory = g.Select(x => x.m == null ? 0 : x.m.dftidActivityCategory).First(),
idCamp = g.Select(x => x.c.idCamp).First(),
CategoryName = g.Select(x => x.c.CategoryName).First(),
CategoryDescription = g.Select(x => x.c.CategoryDescription).First(),
masterCount = g.Count(x => x.m != null)
}).OrderBy(o=> o.idActivityCategory);
desiredResult.Dump("desiredResult");
If I just use a basic group approach I get the results but not the extra column information. At least I can't find it once I group.
var simpleGroup = (from c in tbActivityCategories
.Where(w => w.idCamp == idCamp)
.OrderBy(o => o.CategoryName)
from m in tbActivityMasters
.Where(m => m.dftidActivityCategory == c.idActivityCategory)
.DefaultIfEmpty() // LEFT OUTER JOIN
where c.idCamp == idCamp
group m by m == null ? 0 : m.dftidActivityCategory into g
select new
{
// How do I best get the extra desired column information from other tables that I had before grouping
// but still have the benefit of the grouping?
// idActivityCategory = g.Select(x => x.m == null ? 0 : x.m.dftidActivityCategory).First(),
// idCamp = g.Select(x => x.c.idCamp).First(),
// CategoryName = g.Select(x => x.c.CategoryName).First(),
// CategoryDescription = g.Select(x => x.c.CategoryDescription).First(),
// masterCount = g.Count(x => x.m != null)
idActivityCategory = g.Key,
masterCount = g.Count(x => x != null)
});
simpleGroup.Dump("simpleGroup");
Please tear this up. I'm trying to learn and it just seems like I'm missing the big picture here. Thanks.
UPDATE : Cleaned up by moving the work into the group and making the select more straight forward. If I had known this yesterday then this would have been my original answer to the OP question.
int idCamp = 1;
var desiredResult =
(from c in tbActivityCategories
.Where(w => w.idCamp == idCamp)
from m in tbActivityMasters
.Where(m => m.dftidActivityCategory == c.idActivityCategory)
.DefaultIfEmpty() // LEFT OUTER JOIN
where c.idCamp == idCamp
group new { c, m } by new
{ idActivityCategory = m == null ? 0 : m.dftidActivityCategory,
idCamp = c.idCamp,
CateGoryName = c.CategoryName,
CategoryDescription = c.CategoryDescription
} into g
select new
{
idActivityCategory = g.Key.idActivityCategory,
idCamp = g.Key.idCamp,
CategoryName = g.Key.CateGoryName,
CategoryDescription = g.Key.CategoryDescription,
masterCount = g.Count(x => x.m != null)
}).OrderBy(o => o.idActivityCategory);
desiredResult.Dump("desiredResult");

Convert aggregate SQL query to LINQ

I decided to try my hand at LINQ and so far its been a miserable failure. I need to convert the following SQL query to LINQ:
SELECT
MAX(A.NEXTPAYMENTDATE) as NextPaymentDate
, SUM(B.PurchasePrice) - SUM(A.Amount) AS BALANCE
, c.FirstName
, c.LastName
, b.[year]
, b.make
, b.model
FROM Payments A
JOIN Vehicles B ON A.VehicleId = B.Id
JOIN Customers C ON b.CustomerId = c.Id
GROUP BY VehicleId, c.FirstName, c.LastName, b.[year], b.make, b.model
HAVING SUM(B.PurchasePrice) - SUM(A.Amount) > 0
This is what I have so far. It seems to work to a certain extent, but I don't know how to progress from here.
var groupedpayments =
from payments in db.Payments
group payments by new { payments.VehicleId }
into paymentGroup
let maxDate = paymentGroup.Max(x => x.NextPaymentDate)
let paid = paymentGroup.Sum(x => x.Amount)
select
new { Payments = paymentGroup.Where(x => x.NextPaymentDate == maxDate)};

I think that is what you need.
var query =
Payments.Join(Vehicles, p => p.VehicleId, v => v.Id, (p, v) => new {p, v})
.Join(Customers, d => d.v.CustomerId, c => c.Id, (d, c) => new {d, c})
.GroupBy(r =>
new {
r.d.p.VehicleId,
r.d.v.year,
r.d.v.make,
r.d.v.model,
r.c.FirstName,
r.c.LastName
},
(g, data) =>
new {
FirstName = g.FirstName,
LastName = g.LastName,
Year = g.year,
Make = g.make,
Model = g.model,
NextPaymentDate = data.Max(dd => dd.d.p.NEXTPAYMENTDATE),
Balance = data.Sum(dd => dd.d.v.PurchasePrice)
- data.Sum(dd => dd.d.p.Amount)})
.Where(r => r.Balance > 0);