Develop Reference

How to make LINQ query with Unions more efficient

I inherited the LINQ query below and I feel that the query can be refactored for efficiency. The query currently takes about 6-8 seconds of processing time to return one record to the user on the front-end of the application. LINQ is not my strong suite, so any help would be greatly appreciated.
The query should ultimately produce a distinct list of CA_TASK_VW objects that are tied to a list of distinct CA_OBJECT_ID's obtained from the CA_OBJECT, CA_PEOPLE, and CA_CONTRACTOR tables.
var data = (from a in _db.CA_TASK_VW
where a.TASK_TYPE == "INSPECTION" && a.TASK_AVAILABLE_FLAG == "Y" && a.TARGET_END_DATE == null
select a).AsQueryable();
data = data.Join(_db.CA_OBJECT.Where(o => o.ENTERED_BY == _userId),
o => o.CA_OBJECT_ID, p => p.CA_OBJECT_ID,
(t, p) => t)
.Union(data.Join(_db.CA_PEOPLE.Where(p => p.EMAIL == _email),
t => t.CA_OBJECT_ID, p => p.CA_OBJECT_ID,
(t, p) => t))
.Union(data.Join(_db.CA_CONTRACTOR.Where(c => c.CONTRACTOR.EMAIL == _email),
t => t.CA_OBJECT_ID, c => c.CA_OBJECT_ID,
(t, c) => t));

The code seems to be using Join/Union to execute basically a where predicate on the list of CA_TASK_VW, filtering it step by step to the final result, so what happens if you just specify the where condition directly?
var data = from a in _db.CA_TASK_VW
where a.TASK_TYPE == "INSPECTION" && a.TASK_AVAILABLE_FLAG == "Y" && a.TARGET_END_DATE == null
select a;
data = data.Where(t => _db.CA_OBJECT.Where(o => o.ENTERED_BY == _userId).Select(o => o.CA_OBJECT_ID).Contains(t.CA_OBJECT_ID) ||
_db.CA_PEOPLE.Where(p => p.EMAIL == _email).Select(p => p.CA_OBJECT_ID).Contains(t.CA_OBJECT_ID) ||
_db.CA_CONTRACTOR.Where(c => c.CONTRACTOR.EMAIL == _email).Select(c => c.CA_OBJECT_ID).Contains(t.CA_OBJECT_ID));

You could try using UNION ALL if you don`t really care about duplicates in your query results as it works much faster than UNION

Using two Linq query in a single method

As shown in the below code, the API will hit the database two times to perform two Linq Query. Can't I perform the action which I shown below by hitting the database only once?
var IsMailIdAlreadyExist = _Context.UserProfile.Any(e => e.Email == myModelUserProfile.Email);
var IsUserNameAlreadyExist = _Context.UserProfile.Any(x => x.Username == myModelUserProfile.Username);

In order to make one request to database you could first filter for only relevant values and then check again for specific values in the query result:
var selection = _Context.UserProfile
.Where(e => e.Email == myModelUserProfile.Email || e.Username == myModelUserProfile.Username)
.ToList();
var IsMailIdAlreadyExist = selection.Any(x => x.Email == myModelUserProfile.Email);
var IsUserNameAlreadyExist = selection.Any(x => x.Username == myModelUserProfile.Username);
The .ToList() call here will execute the query on database once and return relevant values

Start with
var matches = _Context
.UserProfile
.Where(e => e.Email == myModelUserProfile.Email)
.Select(e => false)
.Take(1)
.Concat(
_Context
.UserProfile
.Where(x => x.Username == myModelUserProfile.Username)
.Select(e => true)
.Take(1)
).ToList();
This gets enough information to distinguish between the four possibilities (no match, email match, username match, both match) with a single query that doesn't return more than two rows at most, and doesn't retrieve unused information. Hence about as small as such a query can be.
With this done:
bool isMailIdAlreadyExist = matches.Any(m => !m);
bool isUserNameAlreadyExist = matches.LastOrDefault();

It's possible with a little hack, which is grouping by a constant:
var presenceData = _Context.UserProfile.GroupBy(x => 0)
.Select(g => new
{
IsMailIdAlreadyExist = g.Any(x => x.Email == myModelUserProfile.Email),
IsUserNameAlreadyExist = g.Any(x => x.Username == myModelUserProfile.Username),
}).First();
The grouping gives you access to 1 group containing all UserProfiles that you can access as often as you want in one query.
Not that I would recommend it just like that. The code is not self-explanatory and to me it seems a premature optimization.

You can do it all in one line, using ValueTuple and LINQ's .Aggregate() method:
(IsMailIdAlreadyExist, IsUserNameAlreadyExist) = _context.UserProfile.Aggregate((Email:false, Username:false), (n, o) => (n.Email || (o.Email == myModelUserProfile.Email ? true : false), n.Username || (o.Username == myModelUserProfile.Username ? true : false)));

Need to optimize LINQ code using Nhibernate

I 'm new to NHibernate & LINQ. I have a piece of code which I think can be optimized. Please help me to do so.
foreach (var geography in geographyList.OrderBy(x => x.Name))
{
var introductionDateDetail = environment.IntroductionDateInfo
.IntroductionDateDetails
.OrderByDescending(x => x.ApplicationDate)
.FirstOrDefault(x => x.Geography.Id == geography.Id &&
x.VaccineDetail.Id == vaccineDetail.Id &&
x.MasterForecastInfo.Id == masterforecast.Id &&
x.ViewInfo.Id == viewInfoDetail.ViewInfo.Id);
}
The for loop may iterate to about thousand records.And hence the LINQ statement is also executed that many times. Can we write a piece of code where we can execute the LINQ statement just once?

You can try something like this:
var geographiesId = geographyList.Select(g => g.Id);
var introductionDetails = environment.IntroductionDateInfo
.IntroductionDateDetails
.OrderByDescending(x => x.ApplicationDate)
.FirstOrDefault(x => geographiesId.Contains(x.Geography.Id) &&
x.VaccineDetail.Id == vaccineDetail.Id &&
x.MasterForecastInfo.Id == masterforecast.Id &&
x.ViewInfo.Id == viewInfoDetail

How to implement SkipWhile with Linq to Sql without first loading the whole list into memory?

I need to order the articles stored in a database by descending publication date and then take the first 20 records after the article with Id == 100.
This is what I would like to do with Linq:
IQueryable<Article> articles =
db.Articles
.OrderByDescending(a => a.PublicationDate)
.SkipWhile(a => a.Id != 100)
.Take(20);
However, this generates a NotSupportedException because SkipWhile is not supported in Linq to Sql (see here).
A possible solution is to execute the query and then apply SkipWhile using Linq to Object:
IEnumerable<ArticleDescriptor> articles =
db.Articles
.OrderByDescending(a => a.PublicationDate)
.ToList()
.SkipWhile(a => a.Article.Id != 100)
.Take(20);
But this means I need to load the whole ordered list into memory first and then take 20 articles after the one with Id == 100.
Is there a way to avoid this huge memory consumption?
More in general, what is the best way to achieve this in SQL?

If, as I'm guessing from the column name, PublicationDate doesn't change, you can do this in two separate queries:
Establish the PublicationDate of the Article with Id == 100
Retrieve the 20 articles from that date onwards
Something like:
var thresholdDate = db.Articles.Single(a => a.Id == 100).PublicationDate;
var articles =
db.Articles
.Where(a => a.PublicationDate <= thresholdDate)
.OrderByDescending(a => a.PublicationDate)
.Take(20);
It might even be that LINQ to SQL can translate this:
var articles =
db.Articles
.Where(a => a.PublicationDate
<= db.Articles.Single(aa => aa.Id == 100).PublicationDate)
.OrderByDescending(a => a.PublicationDate)
.Take(20);
but that may be too complex for it. Try it and see.

You can try like this
var articles =
db.Articles
.Where(a => a.PublicationDate < db.Articles
.Where(aa => aa.Id==100)
.Select(aa => aa.PublicationDate)
.SingleOrDefault())
.OrderByDescending(a => a.PublicationDate)
.Take(20);

Isnt the solution to just add a where statement?
IQueryable<Article> articles = db.Articles.Where(a => a.id != 100).OrderByDescending(a => a.PublicationDate).Take(20);

Dealing with null values in chained linq-to-sql query expressions

I have a L2S repository query which I'm stuggling to write in a nice way. It looks something like...
_orderRepository
.GetAllByFilter(o => o.CustomerId == id)
.Select(o =>
new CustomerOrderRecord
(
o.Id,
o.PartNumber,
o.Date
// ... etc, more order details
/* Here I need the last DateTime? the customer placed
an order for this item, which might be null.
So I end up with the following horrible part of
the query */
o.Customer.CustomerOrderRecords
.Where(x => x.PartNumber == o.PartNumber)
.OrderByDescending(x => x.Date).FirstOrDefault()
== null ? null :
o.Customer.CustomerOrderRecords
.Where(x => x.PartNumber == o.PartNumber)
.OrderByDescending(x => x.Date).First().Date;
)).ToList();
So hopefully you can see the problem that I'm having to write the whole query chain twice just to do the null check when receiving the LastOrdered value.
This needs to be written in-line (I think) because GetAllByFilter returns an IQueryable.
I tried to use an intermediate variable within the select statement, so I'd have something a bit like the following, but I couldn't get anything like that to compile.
.Select(o =>
new CustomerOrderRecord
(
o.Id,
o.PartNumber,
o.Date
// ... etc, more order details
var last = o.Customer.CustomerOrderRecords
.Where(x => x.PartNumber == o.PartNumber)
.OrderByDescending(x => x.Date).FirstOrDefault()
== null ? null : last.Date;
)).ToList();
Is there a syntax trick available which solves this problem?

Try using Select to fetch the Date member:
o.Customer.CustomerOrderRecords
.Where(x => x.PartNumber == o.PartNumber)
.OrderByDescending(x => x.Date)
.Select(x => (DateTime?)x.Date)
.FirstOrDefault()