I was digging into the documentation of MongoDB http://mongodb.github.io/mongo-csharp-driver/2.7/reference/driver/crud/linq/ and I saw, using .NET drive is possible to make group by on the database.
For 1 ProductId I can have few elements in a database. I want to get whole Last element.
So I am trying to do something like:
var query = _collection.AsQueryable()
.GroupBy(p => p.ProductId, (k, s) =>
new { Name = k, Result = s.First() })
.Select(x => new { Name = x.Name, Result = x.Result }).First();
The problem is that I see an error message like:
System.NotSupportedException: Specified method is not supported.
at
MongoDB.Driver.Linq.Processors.AccumulatorBinder.GetAccumulatorArgument(Expression node)
I know that for now in my example i didnt order by the result.. But this will be my second step. For now i see that i cannot group by. Is it possible to do this kind of group by?
My solution for that is:
var query = _collection.Aggregate(new AggregateOptions { AllowDiskUse = true })
.Match(x => ElementIds.Contains(x.ElementId))
.SortByDescending(x => x.StartDate).ThenByDescending(x => x.CreatedAt)
.Group(x => x.ElementId, x => new
{
StartDate = x.First().StartDate,
Grades = x.First().Grades,
SellingId = x.First().SellingId,
CreatedAt = x.First().CreatedAt,
ModifiedAt = x.First().ModifiedAt,
Id = x.First().Id,
ElementId = x.First().ElementId
})
.ToEnumerable(token);
After that, I parsed it into my model.
AllowDiskUse = true, because in my case MongoDB's memory is not enough to handle this operation.
Related
enter image description hereI have a code. And there you need to make a grouping by name.
//<date,<partid,amount>>
Dictionary<string, Dictionary<int, double>> emSpending = new Dictionary<string, Dictionary<int, double>>();
foreach (Orders order in db.Orders.ToList())
{
foreach (OrderItems orderitem in order.OrderItems.ToList())
{
if (!emSpending.ContainsKey(order.Date.ToString("yyyy-MM"))) emSpending.Add(order.Date.ToString("yyyy-MM"), new Dictionary<int, double>());
if (!emSpending[order.Date.ToString("yyyy-MM")].ContainsKey(Convert.ToInt32(orderitem.PartID))) emSpending[order.Date.ToString("yyyy-MM")].Add(Convert.ToInt32(orderitem.PartID), 0);
emSpending[order.Date.ToString("yyyy-MM")][Convert.ToInt32(orderitem.PartID)] += Convert.ToDouble(orderitem.Amount);
}
}
DataGridViewColumn col1 = new DataGridViewColumn();
col1.CellTemplate = new DataGridViewTextBoxCell();
col1.Name = "Department";
col1.AutoSizeMode = DataGridViewAutoSizeColumnMode.Fill;
col1.HeaderText = "Department";
dgvEMSpending.Columns.Add(col1);
foreach (string date in emSpending.Keys)
{
DataGridViewColumn col = new DataGridViewColumn();
col.Name = date;
col.HeaderText = date;
col.AutoSizeMode = DataGridViewAutoSizeColumnMode.Fill;
col.CellTemplate = new DataGridViewTextBoxCell();
dgvEMSpending.Columns.Add(col);
}
List<string> allKey = emSpending.Keys.ToList();
foreach (string date in allKey)
if (date == "Department") continue;
else
{
dgvEMSpending.Rows.Add();
foreach (int partid in emSpending[date].Keys)
{
dgvEMSpending.Rows[dgvEMSpending.Rows.Count - 1].Cells[0].Value = db.Parts.Where(x => x.ID == partid).SingleOrDefault().Name.GroupBy(Name);
for (int i = 1; i < dgvEMSpending.Columns.Count; i++)
{
if (!emSpending.ContainsKey(dgvEMSpending.Columns[i].Name)) emSpending.Add(dgvEMSpending.Columns[i].Name, new Dictionary<int, double>());
if (!emSpending[dgvEMSpending.Columns[i].Name].ContainsKey(partid)) emSpending[dgvEMSpending.Columns[i].Name].Add(partid, 0);
double val = emSpending[dgvEMSpending.Columns[i].Name][partid];
dgvEMSpending.Rows[dgvEMSpending.RowCount - 1].Cells[i].Value = val;
}
}
}
I tried to use group by myself, but something doesn't work. It just outputs the same names, and I want to group them so that there is a grouping. Pls helped to me.
Ok, a few issues to help you out first. This code:
foreach (Orders order in db.Orders.ToList())
{
foreach (OrderItems orderitem in order.OrderItems.ToList())
{
if (!emSpending.ContainsKey(order.Date.ToString("yyyy-MM"))) emSpending.Add(order.Date.ToString("yyyy-MM"), new Dictionary<int, double>());
if (!emSpending[order.Date.ToString("yyyy-MM")].ContainsKey(Convert.ToInt32(orderitem.PartID))) emSpending[order.Date.ToString("yyyy-MM")].Add(Convert.ToInt32(orderitem.PartID), 0);
emSpending[order.Date.ToString("yyyy-MM")][Convert.ToInt32(orderitem.PartID)] += Convert.ToDouble(orderitem.Amount);
}
}
Right off the bat this is going to trip lazy loading on OrderItems. If you have 10 orders 1-10 you're going to be running 11 queries against the database:
SELECT * FROM Orders;
SELECT * FROM OrderItems WHERE OrderId = 1;
SELECT * FROM OrderItems WHERE OrderId = 2;
// ...
SELECT * FROM OrderItems WHERE OrderId = 10;
Now if you have 100 orders or 1000 orders, you should see the problem. At a minimum ensure that if you are touching a collection or reference on entities you are loading, eager load it with Include:
foreach (Orders order in db.Orders.Include(x => x.OrderItems).ToList())
This will run a single query that fetches the Orders and their OrderItems. However, if you have a LOT of rows this is going to take a while and consume a LOT of memory.
The next tip is "only load what you need". You need 1 field from Order and 2 fields from OrderItem. So why load everything from both tables??
var orderItemDetails = db.Orders
.SelectMany(o => o.OrderItems.Select(oi => new { o.Date, oi.PartId, oi.Amount })
.ToList();
This would give us just the Order date, and each Part ID and Amount. Now that this data is in memory we can group it to populate your desired dictionary structure without having to iterate over it row by row.
var emSpending = orderItemDetails.GroupBy(x => x.Date.ToString("yyyy-MM"))
.ToDictionary(g => g.Key,
g => g.GroupBy(y => y.PartId)
.ToDictionary(g2 => g2.Key, g2 => g2.Sum(z => z.Amount)));
Depending on the Types in your entities you may need to insert casts. This first groups the outer dictionary of the yyyy-MM of the order dates, then it groups the remaining data for each date by part ID, and sums the Amount.
Now relating to your question, from your code example I'm guessing the problem area you are facing is this line:
dgvEMSpending.Rows[dgvEMSpending.Rows.Count - 1].Cells[0].Value = db.Parts
.Where(x => x.ID == partid)
.SingleOrDefault().Name.GroupBy(Name);
Now the question would be to explain what exactly you are expecting from this? You are fetching a single Part by ID. How would you expect this to be "grouped"?
If you want to display the Part name instead of the PartId then I believe you would just want to Select the Part Name:
dgvEMSpending.Rows[dgvEMSpending.Rows.Count - 1].Cells[0].Value = db.Parts
.Where(x => x.ID == partid)
.Select(x => x.Name)
.SingleOrDefault();
We can go one better to fetch the Part names for each used product in one hit using our loaded order details:
var partIds = orderItemDetails
.Select(x=> x.PartId)
.Distinct()
.ToList();
var partDetails = db.Parts
.Where(x => partIds.Contains(x.ID))
.ToDictionary(x => x.ID, x => x.Name);
This fetches us a dictionary set indexed by ID for the part names, it would be done outside of the loop after we had loaded the orderItemDetails. Now we don't have to go to the DB with every row:
dgvEMSpending.Rows[dgvEMSpending.Rows.Count - 1].Cells[0].Value = partDetails[partId];
I have a query where my ThenInclude Where needs to access data in the original data set e.g.
var bikes = context.Bikes
.Include(bike => bikes.Models)
.ThenInclude(model => model.Spec.Where(spec=> spec.SpecYear == bike.YearCreated ))
.ToList();
I currently have this in two separate queries but it would save a lot of time if I completed this request on the database.
Try to use projection instead of Include:
var bikes = context.Bikes
.Select(bike = new Bike
{
Id = bike.Id,
// ... other fields
Models = bike.Models.Select(m => new Model
{
Id = m.Id
// ... other fields
Spec = m.Spec
.Where(spec => spec.SpecYear == bike.YearCreated)
.ToList()
})
})
.ToList();
I have this SQL that I would like to execute in Entity Framework Core 2.1:
Select ItemTypeId, Count(ItemTypeId) as Count from Items i
where i.ParentId = 2
group by ItemTypeId
How do I do that?
This is what I came up with, but it returns zero:
var query = this._context.Items.Where(a => a.ParentId == 2)
.GroupBy(i => new { i.ItemTypeId })
.Select(g => new { g.Key.ItemTypeId, Count = g.Count(i=> i.ItemTypeId == g.Key.ItemTypeId) });
var items = query.ToList();
The only example I could find was here
You don't need Count = g.Count(i=> i.ItemTypeId == g.Key.ItemTypeId), instead use g.Count().
I need to extract some information about support tickets from a database. Each ticket is associated with a medical imaging system, and each system may or may not have service cover associated with it. If it does, there may be multiple service cover entries, but only one that interests us.
I know this is not valid Linq, but what I would really like to do is the following...
var tickets = cxt.SupportTickets
.Select( t => new {
ID = t.ID,
Customer = t.Customer.Name,
var cover = t.System.CoverItems.FirstOrDefault(ci => // some query)
CoverLevel = cover?.Level.Name,
Expiry = cover?.Expiry.ToLongDateString()
});
Is there any way to do this? I know that I could repeat the t.CoverItems.FirstOrDefault(...) bit for every bit of data I want from the cover, but apart from the absolutely awful code mess this would produce, it would be very inefficient, as it would need to do the same subquery multiple times for every ticket.
I thought about breaking it all up into a foreach loop, but then I couldn't see how to create the tickets collection. I can't create an empty collection and then add objects to it, as they are anonymous types, and I wouldn't like to think about how you would specify the generic type!
Anyone any ideas?
You could improve readability:
var tickets = cxt.SupportTickets
.Select(t => new {
Ticket = t,
CoverItem = t.System.CoverItems.FirstOrDefault(ci => // some query)
})
.Select(x => new {
ID = x.Ticket.ID,
Customer = x.Ticket.Customer.Name,
CoverLevel = x.CoverItem?.Level.Name,
Expiry = x.CoverItem?.Expiry.ToLongDateString()
});
You could use query notation instead to use let clause:
var query=from t in cxt.SupportTickets
let cover = t.System.CoverItems.FirstOrDefault(ci => some query)
select new {
ID = t.ID,
Customer = t.Customer.Name,
CoverLevel = cover?.Level.Name,
Expiry = cover?.Expiry//.ToLongDateString()
};
At the end is going to do the same that #TimSchmelter answer, but for things like that you can use let. Another thing, I'm almost sure ToLongDateString() method is not supported in EF.
I've tried this (If you wanted to develop a subQuery separately, because of SoC principle):
var innerQuery = cxt.SupportTickets
.Where(artist => artist.coverId == SomeParameter)
.Select(artist => new {
artistId = artist.artistId,
artistCompleteName = artist.artistName,
artistMasterPiece = artist.CoverName
});
var tickets = cxt.SupportTickets
.Where(
t => innerQuery.Contains(t.coverId)
)
.Select( t => new {
ID = t.ID,
Customer = t.Customer.Name,
var cover = t.System.CoverItems.FirstOrDefault()
CoverLevel = cover?.Level.Name,
Expiry = cover?.Expiry.ToLongDateString()
});
var diaryEntries = (from entry in repository.GetQuery<OnlineDiary.Internal.Model.DiaryEntry>()
.Include("DiaryEntryGradeChangeLog")
.Include("DiaryEntryAction")
join diary in repository.GetQuery<OnlineDiary.Internal.Model.OnlineDiary>()
on entry.DiaryId equals diary.Id
group entry
by diary
into diaryEntriesGroup
select new { Diary = diaryEntriesGroup.Key,
DiaryEntry = diaryEntriesGroup.OrderByDescending(diaryEntry => diaryEntry.DateModified).FirstOrDefault(),
});
This query does not include "DiaryEntryGradeChangeLog" and "DiaryEntryAction" navigation properties, what is wrong in this query?
I have removed join from the query and corrected as per below, and still it populates nothing
var diaryEntries = from entry in repository.GetQuery<OnlineDiary.Internal.Model.DiaryEntry>()
.Include("DiaryEntryGradeChangeLog").Include("DiaryEntryAction")
.Where(e => 1 == 1)
group entry
by entry.OnlineDiary
into diaryEntryGroups
select
new { DiaryEntry = diaryEntryGroups.OrderByDescending(diaryEntry => diaryEntry.DateModified).FirstOrDefault() };
It will not. Include works only if the shape of the query does not change (by design). If you use this query it will work because the shape of the query is still same (OnlineDiary.Internal.Model.DiaryEntry):
var diaryEntries = (from entry in repository.GetQuery<OnlineDiary.Internal.Model.DiaryEntry>()
.Include("DiaryEntryGradeChangeLog")
.Include("DiaryEntryAction");
But once you use manual join, grouping or projection (select new { }) you have changed the shape of the query and all Include calls are skipped.
Edit:
You must use something like this (untested) to get related data:
var diaryEntries = from entry in repository.GetQuery<OnlineDiary.Internal.Model.DiaryEntry>()
group entry by entry.OnlineDiary into diaryEntryGroups
let data = diaryEntryGroups.OrderByDescending(diaryEntry => diaryEntry.DateModified).FirstOrDefault()
select new {
DiaryEntry = data,
GradeChangeLog = data.DiaryEntryGradeChangeLog,
Action = data.DiaryEntryAction
};
or any similar query where you manually populate property for relation in projection to anonymous or unmapped type.