I have an app using Entity Framework. I want to add a tree view listing products, grouped by their categories. I have an old SQL query that will grab all of the products and categories and arrange them into parent nodes and children. I am trying to translate it into LINQ that uses the EF. But the SQL has a WITH sub-query that I am not familiar with using. I have tried using Linqer and LinqPad to sort it out, but they choke on the WITH clause and I am not sure how to fix it. Is this sort of thing possible in LINQ?
Here is the query:
declare #id int
set #id=0
WITH ChildIDs(id,parentid,type,ChildLevel) AS
(
SELECT id,parentid,type,0 AS ChildLevel
FROM dbo.brooks_product
WHERE id = #id
UNION ALL
SELECT e.id,e.parentid,e.type,ChildLevel + 1
FROM dbo.brooks_product AS e
INNER JOIN ChildIDs AS d
ON e.parentid = d.id
WHERE showitem='yes' AND tribflag=1
)
SELECT ID,parentid,type,ChildLevel
FROM ChildIDs
WHERE type in('product','productchild','productgroup','menu')
ORDER BY ChildLevel, type
OPTION (MAXRECURSION 10);
When I run the query, I get data that looks like this (a few thousand rows, truncated here):
ID.....parentid.....type.....ChildLevel
35429..0............menu.....1
49205..0............menu.....1
49206..49205........menu.....2
169999.49206........product..3
160531.169999.......productchild..4
and so on.
The WITH block is a Common Table Expression, and in this case is used to create a recursive query.
This will be VERY difficult in Linq as Linq doesn't play well with recursion. If you need all of the data on one result set that a Stored Procedure would be easier. Another option is to do the recursion in C# (not in Linq but a recursive function) and do multiple round-trips. The performance will not be as good but if you result set is small it may not make much difference (and you will get a better object model).
You may be able to solve this using LINQ to Entities, but it is non-trivial and I suspect it will be very time consuming.
In situations like this, you may prefer to build a SQL View or Table-Valued Function that returns the results for which you're looking. Then import that View or Table-Valued Function into your EF model and you can pull data directly from it using LINQ.
Querying the View in LINQ is no different than querying a table.
To get data from a Table-Valued Function in LINQ, you pass the function's parameters in after the name of the function, like so:
var query = from tvf in _db.MyTableValuedFunction(parameters)
select tvf;
EDIT
As suggested by #thepirat000, Table-Valued Function support is not available in Entity Framework versions prior to version 5. In order to use this functionality, EF must be running with .NET 4.5 or higher.
At the end of the day, I could not get this to work. I ended up writing out a SQL query dynamically and sending that straight to the database. It works fine, and I am not relying on any direct user input so there is no chance of SQL injection. But it seems so old school! For the rest of my program I am using EF and LINQ.
Thanks for the replies!
Related
I have a Linq query that reads from a SQL table and 1 of the fields it returns are from a custom function (in C#).
Something like:
var q = from my in MyTable
select new
{
ID = my.ID,
Amount = GetAmount(ID)
};
If I do a q.Dump() in LinqPad, it shows the results, which tells me that it runs the custom function without trying to send it to SQL.
Now I want to union this to another query, with:
var q1 = (from p in AnotherQuery.Union(q)...
and the I get the error that Method has no supported translation to SQL.
So, my logic tells me that I need to dump q in memory and then try to union to that. I've tried doing that with ToList() and creating a secondary query that populates itself from the List, but that leads to a long list of different errors. Am I on the right track, by trying to get q in memory and union on that, or are there better ways of doing this?
You can't use any custom functions in a LINQ query that gets translated - only the functions supported by the given LINQ provider. If you want your query to happen on the server, you need to stick with the supported functions (even if it sometimes means having to inline code that would otherwise be reused).
The difference between your two queries boils down to when (and where) the projection happens. In your first case, the data from MyTable is returned from the DB - in your sample, just the ID. Then, the projection happens on top of this - the GetAmount method is called in your application for each of ID.
On the other hand, there's no such way for this to happen in your second query, since you're not using GetAmount in the final projection.
You either need to replace the custom function with inlined query the provider understands, or refactor all your queries to use the supported functions in addition with whatever you need to do in-memory. There's no point in giving you any sample code, since it depends entirely on your actual query, and what you're really trying to query for.
I have a table of 200,000 record where I am getting only the top 10 using .Take() but it is taking about 10 seconds to get the data.
My question is: does the .Take() method get all the data from the database and filter the top 10 on the client side?
Here is my code:
mylist = (from mytable in db.spdata().OrderByDescending(f => f.Weight)
group feed by mytable.id into g
select g.FirstOrDefault()).Take(10).ToList();
spdata() is a function Import from stored procedure.
Thanks
The stored procedure probably returns a lot of data to the client which is very slow. You cannot remote a query to an sproc. That would be possible using a view or a table-valued function.
There's no way to use an sproc in a query. You can only execute it by itself.
Your intention probably was to execute the Take(10) on the server. For that to work you need to switch to an inline query, a view or a TVF.
The extension method Take does not fetch all the results from the database. That is not how Takeworks.
However your db.spdata() call probably does fetch all rows.
I'm not 100% sure but as I remember you get an IEnumerable result when you call an SP using EF DataContext...
There are a couple of was to optimize the performance:
Pass the search criteria s as SP params and do the filtering in the stored procedure.
Or if you have a quite simple query in the SP where you are not declaring any variables and where you are just joining some tables then:
Create an indexed view where specify the query that you need and call the Take method on it.
What this will give you? You can map to the created view and EF will now be returning an IQueryable result and not an IEnumerable. This will optimize the sql command and rather the receiving all of the data and then taking the 10 elements that you need, a sql command that just retrieves the 10 elements will be formed.
I also advice you to see what is the deference between IEnumerable vs IQueryable.
It does, because you are sorting the data before grouping, which is not possible to do in SQL.
You should use an aggregate to get the highest weight from each group, then sort the weights to get the ten largest:
mylist = (
from mytable in db.spdata()
group feed by mytable.id into g
select g.Max(f => f.Weight)
).OrderByDescending(w => w).Take(10).ToList();
With Entity Framework 4.3 and Linq, I want to match a search string against certain properties of contained objects and also on the properties of child objects. This turns out to be a rather complex query though, and I'm not sure how to do it. For instance, one property is an integer and I'm not allowed to call .ToString() in a Linq query.
In order to give you an idea of what I'm trying to do, consider this example code:
var dbVersions = from ver in db.Versions
where ver.Name.Contains(search) ||
ver.Children.Any(c=>c.Id.ToString().Contains(search))
select ver;
How should I implement this search? Perhaps via a stored procedure?
The database server is SQL Server 2012.
If you want to use LINQ the framework internally will do the proper optimizations and from my experience the results are quite OK.
If you don't want to use an stored procedure and stick to LINQ keeping it all in your class code use:
SqlFunctions.StringConvert((double)c.Id)
for converting your int to an string. Note that there is no overload for int, so you need to cast it to double or decimal.
In your situation I would suggest using a stored procedure.
If you end up passing one search term which will be used against multiple columns, then your better off writing a stored proc. I once tried something similar to what you're doing, and the end result was really messy LINQ statements that left me feeling dirty :)
Good reference:
x.ToString() is not supported by the entity framework!
Here's an exmaple of how to use stored procs with EF:
http://blogs.msdn.com/b/bindeshv/archive/2008/11/20/using-stored-procedures-in-entity-framework.aspx
Remember to add the stored proc when you "Update Model from database".
Suppose I have a collection (of arbitrary size) of IQueryable<MyEntity> (all for the same MyEntity type). Each individual query has successfully been dynamically built to encapsulate various pieces of business logic into a form that can be evaluated in a single database trip. Is there any way I can now have all these IQueryables executed in a single round-trip to the database?
For example (simplified; my actual queries are more complex!), if I had
ObjectContext context = ...;
var myQueries = new[] {
context.Widgets.Where(w => w.Price > 500),
context.Widgets.Where(w => w.Colour == 5),
context.Widgets.Where(w => w.Supplier.Name.StartsWith("Foo"))
};
I would like to have EF perform the translation of each query (which it can do indivudually), then in one database visit, execute
SELECT * FROM Widget WHERE Price > 500
SELECT * FROM Widget WHERE Colour = 5
SELECT W.* FROM Widget
INNER JOIN SUpplier ON Widget.SupplierId = Supplier.Id
WHERE Supplier.Name LIKE 'Foo%'
then convert each result set into an IEnumerable<Widget>, updating the ObjectContext in the usual way.
I've seen various posts about dealing with multiple result sets from a stored procedure, but this is slightly different (not least because I don't know at compile time how many results sets there are going to be). Is there an easy way, or do I have to use something along the lines of Does the Entity Framework support the ability to have a single stored procedure that returns multiple result sets??
No. EF deosn't have query batching (future queries). One queryable is one database roundtrip. As a workaround you can try to play with it and for example use:
string sql = ((ObjectQuery<Widget>)context.Widgets.Where(...)).ToTraceString();
to get SQL of the query and build your own custom command from all SQLs to be executed. After that you can use similar approach as with stored procedures to translate results.
Unless you really need to have each query executed separately you can also union them to single query:
context.Widgets.Where(...).Union(context.Widgets.Where(...));
This will result in UNION. If you need just UNION ALL you can use Concat method instead.
It might be late answer, hopefully it would help some one else with the same issue.
There is Entity Framework Extended Library on NuGet which provides the future queries feature (among others). I played a bit with it and it looks promising.
You can find more information here.
I'm fairly new to nHibernate having come from an EF background and I'm struggling with the following query :
_patientSearchResultModel = (from patient in _patientRepository.Query(patientSearch.BuildPatientSpecification())
join admission in _admissionRepository.Query(patientSearch.BuildAdmissionSpecification())
on patient.Id equals admission.Patient.Id
orderby admission.AdmissionDate
select new PatientSearchResultModel(patient.Id,
admission.Id,
false,
_phaseTypeMapper.GetPhaseTypeModel(admission.PhaseType),
patient.Last, patient.First,
admission.InPatientLocation,
admission.AdmissionDate,
admission.DischargeDate,
admission.RRI,
null,
admission.CompletionStatus,
admission.FollowupStatus)).ToList();
The intent of this query is to allow users to filter the two queries on parameters built up using the two Build???Specification functions and return the resultset. There could be many admission records and I would only like one PatientSearchResultModel per patient object, with the admission object being the newest one by Admission Date.
These objects are coming from nHibernate and it keeps return a Not Supported exception. There is also an association between Patient and Admissions thus : Patient.Admissions but i couldn't figure out how to then add the query filters return from the function Build???Specifications.
I'd be really grateful if someone could point me in the right direction; am I up against the Linq provider implementation here in nHibernate and need to move to Criteria or is it my Linq query ?
If anyone has any links or suggestions for good books or other learning materials in this area that would also be really helpful too.
I see several potential problems:
If you're using NHibernate 2.x + Linq2NHibernate explicit joins like that are not supported; in other versions they're just considered a smell.
I dont think NHibernate supports calling parameterized constructors in select clauses
I'm very sure NHibernate does not support calling instance methods in the select lambda
I'd suggest using the lambda syntax and SelectMany to alleviate potential join issues. Points #2 & #3 can be solved by projecting into an anonymous type, calling AsEnumerable then projecting into your model type.
Overall I'd suggest restructuring your code like:
var patientSpec = patientSearch.BuildPatientSpecification();
var admissionSpec = patientSearch.BuildAdmissionSpecification();
_patientSearchResultModel = _patientRepository.Where(patientSpec)
.SelectMany(p=>p.Admissions).Where(admissionSpec)
.Select(a=> new {
PatientId = a.Patient.Id,
AdminssionId = a.Id,
a.PhaseType,
a.Patient.Last,
a.Patient.First,
a.InPatientLocation,
a.AdmissionDate,
a.DischargeDate,
a.RRI,
a.CompletionStatus,
a.FollowupStatus
}).AsEnumerable()
.Select(x=> new PatientSearchResultModel(x.PatientId, x.AdmissionId ...))
.ToList();
Divide your query into parts and check which part runs and which doesn't.
My take on this is that select new ... is not supported in Linq to nHibernate.
I would recomend using something else, because it is simply too imature and feature-less to use seriously.
As with most popular LINQ-to-Database query providers, NHibernate will try to translate the whole query into a SQL statement to run against the database. This requires that all elements of your query are possible to express in the SQL flavour you're using.
In your query, the select new statement cannot be expressed in SQL, because you're making a call to the constructor of your PatientSearchResultModel class and are making a call to a GetPhaseTypeModel method.
You should restructure your query to express what you want to execute on the SQL database, then call AsEnumerable() to force the remainder of the query to be evaluated in-memory. After that call, you can call the constructor of your class and any .NET methods, and they will be executed as native code.
This query is too complex to describe it using Linq. It would give wrong result finally (if Patient has more than one admission records, result would have duplicate entries).
I see two steps for solution:
1) At development stage, use in-memory query. So, take Patients using ToList() first (query db at this moment). Some predicates (Patient filter like MRN, First, Last) could be used at this stage.
And then do search in-memory. Not performance, but working solution. Mark it for refactor to optimize later.
2) Finally, use NHibernate IQuery (ISQLQuery) and build sql query manually to make sure it would work as expected and work fast enough on SQL Server side. This is just read-only query and do not require Nhibernate query engine (Linq to Nhibernate) at all.