We Keep Coding

Linq EF group by to get latest entries into list of objects

I am trying to move from simple SQL to EF.
But there are some complex queries(joins) that it seems to hard to generate the linq for.
At first I tried to use sqltolinq tool to generate the linq but it gives error as some of the things are not supported in the query.
here is the linq:
var entryPoint = (from ep in dbContext.tbl_EntryPoint
join e in dbContext.tbl_Entry on ep.EID equals e.EID
join t in dbContext.tbl_Title on e.TID equals t.TID
where e.OwnerID == user.UID
select new {
UID = e.OwnerID,
TID = e.TID,
Title = t.Title,
EID = e.EID
});
The table entry has many entries that I would like to group and get the latest for each group. But then I would need to select into a view model object which will be bind to gridview.
I dont know where I can implement the logic to group by and get the latest from each and be able to get values from join table into viewModel object.
somewhere I need to add
group entry by new
{
entry.aID,
entry.bCode,
entry.Date,
entry.FCode
}
into groups
select groups.OrderByDescending(p => p.ID).First()
in the above linq to retrieve latest from each group.

You can insert group by right after the joins:
var query =
from ep in dbContext.tbl_EntryPoint
join e in dbContext.tbl_Entry on ep.EID equals e.EID
join t in dbContext.tbl_Title on e.TID equals t.TID
where e.OwnerID == user.UID
group new { ep, e, t } by new { e.aID, e.bCode, e.Date, e.FCode } into g
let r = g.OrderByDescending(x => x.e.ID).FirstOrDefault()
select new
{
UID = r.e.OwnerID,
TID = r.e.TID,
Title = r.t.Title,
EID = r.e.EID
};
The trick here is to include what you need after the grouping between group and by.
However, the above will be translated to CROSS APPLY with all joins included twice. If the grouping key contains fields from just one table, it could be better to perform the grouping/selecting the last grouping element first, and then join the result with the rest:
var query =
from e in (from e in dbContext.tbl_Entry
where e.OwnerID == user.UID
group e by new { e.aID, e.bCode, e.Date, e.FCode } into g
select g.OrderByDescending(e => e.ID).FirstOrDefault())
join ep in dbContext.tbl_EntryPoint on e.EID equals ep.EID
join t in dbContext.tbl_Title on e.TID equals t.TID
select new
{
UID = e.OwnerID,
TID = e.TID,
Title = t.Title,
EID = e.EID
};

How do I convert this tSQL statement to LINQ using group by in a sub query

I have the following MSSQL query I am trying to convert to LINQ. I am using entity framework with the following syntax to get at the data.
var rv = (from i in DC.TableA select i).ToList();
This is the sql I want to write a C# LINQ query for but I cannot figure it out. Can someone help?
select BTO.*
from TableA BTO
join
(
select eqnum, max(testdate) as testdate
from TableA BTO1
where
BTO1.eqnum in ('M0435', 'Z0843') and
BTO1.testdate <= '2008-06-01'
group by eqnum
) T1
on
T1.eqnum = BTO.eqnum and
T1.testdate = BTO.testdate
order by EqNum;

I think there is opportunity to rewrite your query, but for information purposes I rewrote your sql into linq verbatim.
If you explain what you are trying to achieve we can provide alternative sql / linq
var eqnums = new[] { "M0435", "Z0843" };
var testdate = "2008-06-01";
var query = from bto in DC.TableA
join t1 in (
from bto1 in DC.TableA
where eqnums.Contains(bto1.eqnum) &&
bto1.testdate.CompareTo(testdate) <= 0
group bto1 by bto1.eqnum into g
select new
{
eqnum = g.Key,
testdate = g.Max(x => x.testdate)
}
) on new { bto.eqnum, bto.testdate } equals new { t1.eqnum, t1.testdate }
orderby bto.eqnum
select bto;

left join in Linq query

I'm trying to do a left join, not an inner join in a linq query. I have found answers related to using DefaultIfEmpty() however I can't seem to make it work. The following is the linq query:
from a in dc.Table1
join e in dc.Table2 on a.Table1_id equals e.Table2_id
where a.Table1_id == id
orderby a.sort descending
group e by new
{
a.Field1,
a.Field2
} into ga
select new MyObject
{
field1= ga.Key.Field1,
field2= ga.Key.Field2,
manySubObjects = (from g in ga select new SubObject{
fielda= g.fielda,
fieldb= g.fieldb
}).ToList()
}).ToList();
The query only gives me the rows from table 1 that have a corresponding record in table 2. I would like every record in table 1 populated into MyObject and a list of 0-n corresponding records listed in manySubObjects for each MyObject.
UPDATE:
I tried the answer to the question that is a "possible duplicate", mentioned below. I now have the following code that does give me one record for each item in Table1 even if there is no Table2 record.
from a in dc.Table1
join e in dc.Table2 on a.Table1_id equals e.Table2_id into j1
from j2 in j1.DefaultIfEmpty()
where a.Table1_id == id
orderby a.sort descending
group j2 by new
{
a.Field1,
a.Field2
} into ga
select new MyObject
{
field1= ga.Key.Field1,
field2= ga.Key.Field2,
manySubObjects = (from g in ga select new SubObject{
fielda= g.fielda,
fieldb= g.fieldb
}).ToList()
}).ToList();
However, with this code, when there is no record in table2 I get "manySubObject" as a list with one "SubObject" in it with all null values for the properties of "SubObject". What I really want is "manySubObjects" to be null if there is no values in table2.

In reply to your update, to create the null listing, you can do a ternary in your assignment of manySubObjects.
select new MyObject
{
field1= ga.Key.Field1,
field2= ga.Key.Field2,
manySubObjects =
(from g in ga select g).FirstOrDefaut() == null ? null :
(from g in ga select new SubObject {
fielda= g.fielda,
fieldb= g.fieldb
}).ToList()
}).ToList();
Here is a dotnetfiddle that tries to do what you're attempting. https://dotnetfiddle.net/kGJVjE
Here is a subsequent dotnetfiddle based on your comments. https://dotnetfiddle.net/h2xd9O
In reply to your comments, the above works with Linq to Objects but NOT with Linq to SQL. Linq to SQL will complain that it, "Could not translate expression ... into SQL and could not treat as a local expression." That's because Linq cannot translate the custom new SubObject constructor into SQL. To do that, you have to write more code to support translation into SQL. See Custom Method in LINQ to SQL query and this article.
I think we've sufficiently answered your original question about left joins. Consider asking a new question about using custom methods/constructors in Linq to SQL queries.

I think the desired Result that you want can be given by using GroupJoin()
The code Below will produce a structure like so
Field1, Field2, List < SubObject > null if empty
Sample code
var query = dc.Table1.Where(x => Table1_id == id).OrderBy(x => x.sort)
.GroupJoin(dc.Table2, (table1 => table1.Table1_id), (table2 => table2.Table2_id),
(table1, table2) => new MyObject
{
field1 = table1.Field1,
field2 = table1.Field2,
manySubObjects = (table2.Count() > 0)
? (from t in table2 select new SubObject { fielda = t.fielda, fieldb = t.fieldb}).ToList()
: null
}).ToList();
Dotnetfiddle link
UPDATE
From your comment I saw this
ga.Select(g = > new SubObject(){fielda = g.fielda, fieldb = g.fieldb})
I think it should be (depends on how "ga" is built)
ga.Select(g => new SubObject {fielda = g.fielda, fieldb = g.fieldb})
Please update your question with the whole query, it will help solve the issue.
** UPDATE BIS **
sentEmails = //ga.Count() < 1 ? null :
//(from g in ga select g).FirstOrDefault() == null ? null :
(from g in ga select new Email{
email_to = g.email_to,
email_from = g.email_from,
email_cc = g.email_cc,
email_bcc = g.email_bcc,
email_subject = g.email_subject,
email_body = g.email_body }).ToList()
Should be:
sentEmails = //ga.Count() < 1 ? null :
((from g in ga select g).FirstOrDefault() == null) ? null :
(from g in ga select new Email{
email_to = g.email_to,
email_from = g.email_from,
email_cc = g.email_cc,
email_bcc = g.email_bcc,
email_subject = g.email_subject,
email_body = g.email_body }).ToList()
Checks if the group has a First, if it doesn't the group doesn't have any records so the Action.Name for a Time Stamp has no emails to send. If the First isn't null the loop throw the group elements and create a list of Email,

var results =
(
// Use from, from like so for the left join:
from a in dc.Table1
from e in dc.Table2
// Join condition goes here
.Where(a.Id == e.Id)
// This is for the left join
.DefaultIfEmpty()
// Non-join conditions here
where a.Id == id
// Then group
group by new
{
a.Field1,
a.Field2
}
).Select(g =>
// Sort items within groups
g.OrderBy(item => item.sortField)
// Project required data only from each item
.Select(item => new
{
item.FieldA,
item.FieldB
}))
// Bring into memory
.ToList();
Then project in-memory to your non-EF-model type.

Linq to SQL: Sum with multiple columns select

I want to convert the following SQL code into linq to sql but can't seem to find a way
select holder_name,agent_code,sum(total)
from agent_commission
group by agent_code
Can anyone help me? Am kinda stuck with this for quite a while.
Thanks in advance
UPDATE:
I tried the following
var query = (from p in context.Agent_Commissions
group p by new
{
p.agent_code
}
into s
select new
{
amount = s.Sum(q => q.total),
}
);
How do I select the other two columns? What am I missing?

In fact your SQL query works only when the corresponding relationship between holder_name and agent_code is 1-1, otherwise the Group by agent_code won't work. So your linq query should be like this:
var query = from p in context.Agent_Commissions
group p by p.agent_code into s
select new {
holder_name = s.FirstOrDefault().holder_name,
agent_code = s.Key,
amount = s.Sum(q => q.total)
};

Here is your linq query
from a in ctx.agent_code
group a by a.holder_name, a.code into totals
select { holder_name = a.holder_name,
code = a.code,
total = totals.Sum(t=>t.total)}
Given that you have linq2sql context in ctx variable and it has your table in it.

how to convert this T-SQL statment to linq

i'm starter in linq, i have write this T-SQL Query
select * from DOCUMENT_TYPES where document_id in(
select document_id from Clearance_Document where Clearance_id=(select clearance_id from clearance_id from request where request_id=3))
i want convert this T-SQL Query to linq, please help me, thanks

Well, I would start first by refactoring your SQL into something other than a chain of nested sub-queries. I think this ought to do the same thing, and it's much more readable:
SELECT
*
FROM
DOCUMENT_TYPES dt
JOIN
Clearance_Document cd
ON
dt.document_id = cd.document_id
JOIN
Request r
ON
cd.clearance_id = r.clearance_id
WHERE
r.request_id = 3
(I'm assuming that from clearance_id from request was a typo.)
Then you can easily refactor into a LINQ statement:
var result = from dt in DOCUMENT_TYPES
join cd in Clearance_Document on dt.document_id equals cd.document_id
join r in Request on cd.clearance_id equals r.clearance_id
where r.request_id = 3
select new {
property1 = dt.something,
property2 = cd.somethingElse,
...
};

var result =
from a in DOCUMENT_TYPES
let list =
(
from b in Clearance_Document
where b.Clearance_id == (from c in clearance_id where request_id == 3).First<string>())
select b
).ToList()
where list.Contains(a.document_id)
select a;

Something like that should do (i guessed you're using EF, but you can easyly adapt to other LinQ-Types):
context.Document_Types.Where(doc =>
conext.Clearance_Document.Where(cd =>
cd.Clearance_Id == context.Request.Single(r => r.Request_Id == 3)
).Contains(doc.Document_Id)
).ToList();

How about
var result = c.Id context.Request.Single(r => r.Id == 3)
.Clearances.SelectMany(c => x.DocumentTypes);
In effect, get the one and only Request with an Id equal to 3, then get all the DocumentTypes of all its Clearances.
If your database is set up with the appropriate foreign keys these relationships will be automatically generated as part of your model.