I have a query that pulls back a user's "feed" which is essentially all of their activity. If the user is logged in the query will be filtered so that the feed not only includes all of the specified user's data, but also any of their friends.
The database structure includes an Actions table that holds the user that created the action and a UserFriends table which holds any pairing of friends using a FrienderId and FriendeeId column which map to UserIds.
I have set up my LINQ query and it works fine to pull back the data I want, however, I noticed that the query gets turned into X number of CASE clauses in profiler where X is the number of total Actions in the database. This will obviously be horrible when the database has a user base larger than just me and 3 test users.
Here's the SQL query I'm trying to achieve:
select * from [Action] a
where a.UserId = 'GUID'
OR a.UserId in
(SELECT FriendeeId from UserFriends uf where uf.FrienderId = 'GUID')
OR a.UserId in
(SELECT FrienderId from UserFriends uf where uf.FriendeeId = 'GUID')
This is what I currently have as my LINQ query.
feed = feed.Where(o => o.User.UserKey == user.UserKey
|| db.Users.Any(u => u.UserFriends.Any(ufr => ufr.Friender.UserKey ==
user.UserKey && ufr.isApproved)
|| db.Users.Any(u2 => u2.UserFriends.Any(ufr => ufr.Friendee.UserKey ==
user.UserKey && ufr.isApproved)
)));
This query creates this:
http://pastebin.com/UQhT90wh
That shows up X times in the profile trace, once for each Action in the table. What am I doing wrong? Is there any way to clean this up?
I would split the query into two queries,
Find all friends for the user, select their user keys and put them into a list.
Filter the feed according to the userkey and the keys of her.
Here is my shot at it without knowing your exact itnerfaces and ojbects, but it shows the concept:
var friends = db.UserFriends
.Where(x => x.isApproved && (
x.Friender.UserKey == user.userKey ||
x.Friendee == user.userKey
)
)
.Select(x => x.userKey)
.Distinct()
.ToList();
feed = feed.Where(x => x.userKey == user.userKey || friends.Contains(x.UserKey));
This should yield a query similar to this
SELECT ...
FROM feed
WHERE userKey == 'userkey1' OR userKey in ('userKey2', 'userKey3', ...)
Related
I have an Asp.net core web API that uses EF Core DB first approach. I have the following Tables in my SQL server database.
Application table - contains a list of Applications.
Role table - contains a list of Roles (user, admin, Super Admin, etc.)
User table - contains a list of Users
User Role table - contains a mapping between User and Role table
Feature tables - contains a list of Features (Home page, user management, etc.)
RoleFeature table - contains the mapping between Feature and Role table
I am trying to get the RoleName and list of features for the given UserId and appId.
Below is the Linq query I have so far:
RoleDto role =
from a in ctx.Application.Where(x => x.ApplicationId == appId)
from r in ctx.Role.Where(x => x.ApplicationId == a.ApplicationId)
from ur in ctx.UserRole.Where(x => x.UserId == userId && x.RoleId == r.RoleId)
from rf in ctx.RoleFeature.Where(x => x.RoleId == ur.RoleId)
from f in ctx.Feature.Where(x => x.FeatureId == rf.FeatureId).Where(x => x.IsActive)
group new { r.RoleName, f.FeatureId } by ur.RoleId into g
select new RoleDto
{
Name = g.Select(x => x.RoleName).FirstOrDefault(),
FeatureIds = g.Select(x => x.FeatureId).ToList()
}.AsNoTracking()
However, I am getting an error saying unable to translate the LINQ expression.
The problem is with the FirstOrDefault() in the select section of the "groupby" query.
You are probably using ef core version higher than 2.1 and there are some changes in handling groupby query. You should take a look at the following link :
https://learn.microsoft.com/en-us/ef/core/what-is-new/ef-core-2.1#linq-groupby-translation
Before version 2.1, in EF Core the GroupBy LINQ operator would always
be evaluated in memory. We now support translating it to the SQL GROUP
BY clause in most common cases.
So the query should be translated to SQL GROUP BY, but methods such as FirstOrDefault() can not be translated. As a fast solution you can change FirstOrDefault() with Max(), Min() or other aggregate functions that are supported in DB.
You don't need to select the Application. You can start by selecting Role directly as -
from r in dbCtx.Role.Where(x => x.ApplicationId == appId)
This will simplify the final SQL generated by EF. So the query will be faster.
If a User has multiple Role then you are trying to take the first one. You should do that when you are selecting Role as -
from r in ctx.Role.Where(x => x.ApplicationId == a.ApplicationId).Take(1)
Finally, you can fetch a list of RoleName and FeatureId, and then do the grouping on client side -
var query =
from r in dbCtx.Role.Where(x => x.ApplicationId == appId).Take(1)
from ur in dbCtx.UserRole.Where(x => x.UserId == userId && x.RoleId == r.RoleId)
from rf in dbCtx.RoleFeature.Where(x => x.RoleId == ur.RoleId && x.Feature.IsActive)
select new
{
RoleName = rf.Role.RoleName,
FeatureId = rf.FeatureId
};
var roleDto = query.AsNoTracking()
.AsEnumerable()
.GroupBy(p => p.RoleName)
.Select(g => new RoleDto
{
Name = g.Key,
FeatureIds = g.Select(p => p.FeatureId).ToList()
})
.FirstOrDefault();
I have a query in which I pass the search term to filter the list of Companies, either by Email or company Title which is stored in another table (TranslationTexts) as Text column for multiple locales.
The query runs fine but it is very heavy and takes time. How can I make it more efficient?
See Table Diagram Image
The query:
gm.ListData = context.Companies.ToList()
.Where(a => a.AspNetUser.Email.NullableContains(searchTerm) ||
a.TitleTranslation.TranslationTexts
.Where(b => b.Text.NullableContains(searchTerm)).Any()
).Select(c => new ListCompany
{
CompanyID = c.CompanyID,
EmailID = c.AspNetUser.Email,
Title = c.TitleTranslation.TranslationTexts.FirstOrDefault(d => d.Locale == Locale).Text
}).ToList();
I want to get the latest record of each group from a SQL Server table using Linq.
Table Example:
I want to get this result:
My Linq query returns one record for each company, but it doesn't return the most recent ones:
var query = from p in db.Payments
where p.Status == false
&& DateTime.Compare(DateTime.Now, p.NextPaymentDate.Value) == 1
group p by p.CompanyID into op
select op.OrderByDescending(nd => nd.NextPaymentDate.Value).FirstOrDefault();
What am i missing here? Why isn't the NextPaymentDate being ordered correctly?
!!UPDATE!!
My query is working as expected. After analysing #Gilang and #JonSkeet comments i ran further tests and found that i wasn't getting the intended results due to a column that wasn't being updated.
var query = from p in db.Payments
where p.Status == false
group p by p.CompanyID into op
select new {
CompanyID = op.Key,
NextPaymentDate = op.Max(x => x.NextPaymentDate),
Status = false
};
The reason your query is not being ordered correctly is that your query does not do proper grouping. You did correctly grouping by CompanyID, but then you have to retrieve the maximum NextPaymentDate by calling aggregate function.
Status can be assigned false because it is already filtered by Where clause in the early clauses.
So I have the following Linq query:
var member = (from mem in
context.Members.Include(m =>
m.MemberProjects.Select(mp => mp.Project))
where mem.MemberId == memberId
select mem).FirstOrDefault();
This returns a Member entity, with a set of MemberProjects that have a Project child. I would like to limit the MemberProjects to only those for which the Project child has a property
ProjectIdParent == null.
One of my failed attempts might make the intent clearer:
var member = (from mem in context.Members
.Include(m => m.MemberProjects
.Where(mp =>
mp.Project.ProjectIdParent == null)
.Select(proj => proj.Project))
where mem.MemberId == memberId
select mem).FirstOrDefault();
This of course complains of an invalid Include expression because of the Where clause.
Any thoughts on how to do this would be great :)
DISCLAIMER: I havent tested this. This is just an idea. If you let me know the results, I will update this accordingly. (Skip to the update part for the tested solutions)
var member = (from mps in context.MemberProjects
.Include(m => m.Members)
.Include(m => m.Projects)
where mps.Project.ProjectIdParent == null
select mps)
.FirstOrDefault(mprojs => mprojs.Member.MemberId == memberId);
I'd also analyze the queries using something like EFProfiler to make sure the generated queries dont leave the realm of sanity.
You can also take a look at this post by Jimmy Bogard on Many to Many relationships with ORMs.
Update
I came up with multiple tested solutions for this with EF 6.1.3. My Edmx looked like below:
The setup data is like below:
I was able to run code below to get the MemberFive correctly
var member = context.Members.FirstOrDefault
(m => m.MemberId == memberId
&& m.Projects.Any(p => p.ProjectParentId == null));
The generated SQL looked like this:
SELECT TOP (1) [Extent1].[MemberId] AS [MemberId],
[Extent1].[MemberName] AS [MemberName]
FROM [dbo].[Members] AS [Extent1]
WHERE ([Extent1].[MemberId] = 1)
AND (EXISTS (SELECT 1 AS [C1]
FROM (SELECT [MemberProjects].[MemberId] AS [MemberId],
[MemberProjects].[ProjectId] AS [ProjectId]
FROM [dbo].[MemberProjects] AS [MemberProjects])
AS [Extent2]
INNER JOIN [dbo].[Projects] AS [Extent3]
ON [Extent3].[ProjectId] = [Extent2].[ProjectId]
WHERE ([Extent1].[MemberId] = [Extent2].[MemberId])
AND ([Extent3].[ProjectParentId] IS NULL)))
If you dont like the generated query you can use this:
var memberQuery = #"Select M.* from Members M
inner join MemberProjects MP on M.MemberId = Mp.ProjectId
inner join Projects P on MP.ProjectId = P.ProjectId
where M.MemberId = #MemberId and P.ProjectParentId is NULL";
var memberParams = new[]
{
new SqlParameter("#MemberId", 1)
};
var member3 = context.Members.SqlQuery(memberQuery, memberParams)
.FirstOrDefault();
The later consistently returned under 20ms vs the other one hovered around 60ms (if that matters to you).
I hope this helps.
I have the following query to start with:
var query = from p in db.Products
from pc in p.NpProductCategories
where pc.CategoryId == categoryId
select p;
I'm applying some more filtering on it and in the end I want to sort the results:
if (orderBy == ProductSortingEnum.Name)
query = query.OrderBy(x => x.Name);
else
query = query.OrderBy(............);
My big problem (coming from not knowing linq too good) is the ELSE here. How can I sort results by a column that is not in the current result set? I would like to somehow link to another linq query in the orderby. The sorting I'm trying to achive is to link to NpProductVariants query using the ProductId to match between NpProductVariant and Products
and sort by the Price of the NpProductVariant
Assuming you have the relationship set up in the dbml...
For one to one (and many to one):
query = query.OrderBy(p => p.NpProductVariant.Price);
For one to many:
query = query.OrderBy(p => p.NpProductVariants.Select(v => v.Price).Max());
Also:
var query =
from p in db.Products
where p.NpProductCategories.Any(pc => pc.CategoryId == categoryId)
select p;
I think you can hook your Join to your query as long as it is returning the same thing. So maybe something like (I'm not 100 % sure since I haven't tried it):
query = from i1 in query
join i2 in query2 on i1.PropertyToJoin equals i2.PropertyToJoin
orderby i1.OrderProp1, i2.OrderProp2
select i1;
But I think it might be a good idea to check the generated sql so it is still effective.