I don't know Linq2Sql so well yet and I was wondering if there is a trick for this probably common MVVM scenario. I have Linq2Sql data context containing Domain models, but I am fetching data for my customized ViewModel object from it.
var query = from ord in ctx.Table_Orders
select new OrderViewModel()
{
OrderId = ord.OrderId,
OrderSum = ord.OrderSum,
OrderCurrencyId = ord.OrderCurrencyId,
OrderCurrencyView = ord.Currency.CurrencyText
};
So i want my ViewModel to inculde both CurrencyId from domain object and the CurrencyText from related table to show it nicely in the View.
This code works great. It generates one DB call with join to fetch the CurrencyText. But the model is simplified, real one has many more fields. I want to make the code reusable because I have many different queries, that returns the same ViewModel. Now every minor change to OrderViewModel requires lots of maintainance.
So I moved the code to OrderViewModel itself as a constructor.
public OrderViewModel(Table_Order ord)
{
OrderId = ord.OrderId,
OrderSum = ord.OrderSum,
OrderCurrencyId = ord.OrderCurrencyId,
OrderCurrencyView = ord.Currency.CurrencyText
}
And call it like this.
var query = from ord in ctx.Table_Orders
select new OrderViewModel(ord);
The Problem: The join is gone DB query is no more optimised. Now I get 1+N calls to database to fetch CurrencyText for every line.
Any comments are welcome. Maybe I have missed different great approach.
This is how far i could get on my own, to get the code reusability. I created a function that does the job and has multiple parameters. Then I need to explicitly pass it everything that has crossed the line of entity.
var query = ctx.Table_Orders.Select(m =>
newOrderViewModel(m, m.Currency.CurrencyText));
The DB call is again optimized. But it still does not feel like I am there yet! What tricks do You know for this case?
EDIT : The final solution
Thanks to a hint by #Muhammad Adeel Zahid I arrived at this solution.
I created an extension for IQueryable
public static class Mappers
{
public static IEnumerable<OrderViewModel> OrderViewModels(this IQueryable<Table_Order> q)
{
return from ord in q
select new OrderViewModel()
{
OrderId = ord.OrderId,
OrderSum = ord.OrderSum,
OrderCurrencyId = ord.OrderCurrencyId,
OrderCurrencyView = ord.Currency.CurrencyText
};
}
}
Now i can do this to get all list
var orders = ctx.Table_Order.OrderViewModels().ToList();
or this to get a single item, or anything in between with Where(x => ..)
var order = ctx.Table_Order
.Where(x => x.OrderId == id).OrderViewModels().SingleOrDefault();
And that completely solves this question. The SQL generated is perfect and the code to translate objects is reusable. Approach like this should work with both LINQ to SQL and LINQ to Entities. (Not tested with the latter) Thank You again #Muhammad Adeel Zahid
Whenever we query the database, we mostly require either enumeration of objects (more than one records in db) or we want a single entity (one record in db). you can write your mapping code in method that returns enumeration for whole table like
public IEnumerable<OrderViewModel> GetAllOrders()
{
return from ord in ctx.Table_Orders
select new OrderViewModel()
{
OrderId = ord.OrderId,
OrderSum = ord.OrderSum,
OrderCurrencyId = ord.OrderCurrencyId,
OrderCurrencyView = ord.Currency.CurrencyText
};
}
Now you may want to filter these records and return another enumeration for example on currencyID
public IEnumerable<OrderViewModel> GetOrdersByCurrency(int CurrencyID)
{
return GetAllOrders().Where(x=>x.CurrencyId == CurrencyID);
}
Now you may also want to find single record out of all these view models
public OrderViewModel GetOrder(int OrderID)
{
return GetAllOrders().SingleOrDefault(x=>x.OrderId == OrderID);
}
The beauty of IEnumerable is that it keeps adding conditions to query and does not execute it until it is needed. so your whole table will not be loaded unless you really want it and you have kept your code in single place. Now if there are any changes in ViewModel Mapping or in query itself, it has to be done in GetAllOrders() method, rest of code will stay unchanged
You can avoid the N+1 queries problem by having Linq2SQL eagerly load the referenced entites you need to construct your viewmodels. This way you can build one list of objects (and some referenced objects) and use it to construct everything. Have a look at this blog post.
One word of warning though: This technique (setting LoadOptions for the Linq2SQL data context) can only be done once per data context. If you need to perform a second query with a different eager loading configuration, you must re-initalize your data context. I automated this with a simple wrapper class around my context.
Related
I'm using Sqlite database and System.Data.SQLite 1.0.92
There is 2 table here:
Table Person:
PersonId
PersonName
Table Student:
StudentId
PersonId(reference table Person FK)
StudentNo
Now every time I get the Persons Collection in EF5:
using (var ctx = new myEntities)
{
AllPersons = ctx.Persons.ToList();
}
There is also has AllPersons.student collection will include in the result;
But I don't need it. Of course that's just an example, There is a lot of big table has so many references, it always has performance problems here because of that.
So I'm trying to do not let it in my result. So I change it:
using (var ctx = new myEntities)
{
ctx.Configuration.ProxyCreationEnabled = false;
ctx.Configuration.LazyLoadingEnabled = false;
AllPersons= ctx.Persons.ToList();
}
Now fine, because AllPersons.student collection will always be null
But now I found: If I get Person and Student together:
using (var ctx = new myEntities)
{
ctx.Configuration.ProxyCreationEnabled = false;
ctx.Configuration.LazyLoadingEnabled = false;
AllPersons= ctx.Persons.ToList();
AllStudents = ctx.Student.ToList();
}
Now the reference still include in.
So Is there anyway to don't let the reference include in any time in this situation?
Thank you.
Update
For some friends request, I explain why I need it:
1: When I convert it to json it will be a dead loop. even I already use Json.net ReferenceLoopHandling, the json size very big to crash the server.(if no references, it's just a very small json)
2:Every time I get the client data and need to save, it will display exception about model state, until I set it to null.
Example:
using (myEntities ctx = new myEntities())
{
ctx.Configuration.LazyLoadingEnabled = false;
ctx.Configuration.ProxyCreationEnabled = false;
Person model= ThisIsAModel();
model.students = null; // This is a key, I need set the students collection references to null , otherwise it will throw exception
ctx.Entry(model).State = EntityState.Modified;
ctx.SaveChanges();
}
3: This is More important problem. I already get all data and cache on the server. But It will let the loading time very long when server start. (because the data and references are so many, that is the main problem), I don't know I'll meet what kind of problem again....
public List<Person> PersonsCache; // global cache
public List<Student> StudentsCache; // global cache
using (myEntities ctx = new myEntities())
{
ctx.Configuration.LazyLoadingEnabled = false;
ctx.Configuration.ProxyCreationEnabled = false;
// There is so many references and data, will let it very slow , when I first time get the all cache. even I only get the Person model, not other , just because some Collection has some references problem. It will very slow....
PersonsCache = ctx.Persons.ToList();
StudentsCache= ctx.Student.ToList();
}
The Problem
As you said, when you load both of Parent and Child lists even when LazyLoading is disabled, and then look in parent.Childs you see child items has been loaded too.
var db = new YourDbContext();
db.Configuration.LazyLoadingEnabled = false;
var parentList= db.YourParentSet.ToList();
var childList= db.YourChildSet.ToList();
What happened? Why childs are included in a parent?
The childs under a parent entity, are those you loaded using db.YourChildSet.ToList(); Exactly themselves; In fact Entity Framework never loads childs for a parent again but because of relation between parent and child in edmx, they are listed there.
Is that affect Perforemance?
According to the fact that childs only load once, It has no impact on perforemance because of loading data.
But for serialization or something else's sake, How can I get rid of it?
you can use these solutions:
Solution 1:
Use 2 different instance of YourDbContext:
var db1 = new YourDbContext();
db1.Configuration.LazyLoadingEnabled = false;
var parentList= db.YourParentSet.ToList();
var db2 = new YourDbContext();
db2.Configuration.LazyLoadingEnabled = false;
var childList= db.YourChildSet.ToList();
Now when you look in parent.Childs there is no Child in it.
Solution 2:
use Projection and shape your output to your will and use them.
var db1 = new YourDbContext();
db1.Configuration.LazyLoadingEnabled = false;
var parentList= db.YourParentSet
.Select(x=>new /*Model()*/{
Property1=x.Property1,
Property2=x.Property2, ...
}).ToList();
This way when serialization there is nothing annoying there.
Using a custom Model class is optional and in some cases is recommended.
Additional Resources
As a developer who use Entity Framework reading these resources is strongly recommended:
Performance Considerations for Entity Framework 4, 5, and 6
Connection Management
I'll focus on your third problem because that seems to be your most urgent problem. Then I'll try to give some hints on the other two problems.
There are two Entity Framework features you should be aware of:
When you load data into a context, Entity Framework will try to connect the objects wherever they're associated. This is called relationship fixup. You can't stop EF from doing that. So if you load Persons and Students separately, a Person's Students collection will contain students, even though you didn't Include() them.
By default, a context caches all data it fetches from the database. Moreover, it stores meta data about the objects in its change tracker: copies of their individual properties and all associations. So by loading many objects the internal cache grows, but also the size of the meta data. And the ever-running relationship fixup process gets slower and slower (although it may help to postpone it by turning off automatic change detection). All in all, the context gets bloated and slow like a flabby rhino.
I understand you want to cache data in separate collections for each entity. Two simple modifications will make this much quicker:
Evade the inevitable relationship fixup by loading each collection by a separate context
Stop caching (in the context) and change tracking by getting the data with AsNoTracking.
Doing this, your code will look like this:
public List<Person> PersonsCache;
public List<Student> StudentsCache;
using (myEntities ctx = new myEntities())
{
ctx.Configuration.ProxyCreationEnabled = false;
PersonsCache = ctx.Persons
.AsNoTracking()
.ToList();
}
using (myEntities ctx = new myEntities())
{
ctx.Configuration.ProxyCreationEnabled = false;
StudentsCache= ctx.Student
.AsNoTracking()
.ToList();
}
The reason for turning off ProxyCreationEnabled is that you'll get light objects and that you'll never inadvertently trigger lazy loading afterwards (throwing an exception that the context is no longer available).
Now you'll have cached objects that are not inter-related and that get fetched as fast as it gets with EF. If this isn't fast enough you'll have to resort to other tools, like Dapper.
By the way, your very first code snippet and problem description...
using (var ctx = new myEntities)
{
AllPersons = ctx.Persons.ToList();
}
There is also has AllPersons.student collection will include in the result;
...suggest that Entity Framework spontaneously performs eager loading (of students) without you Include-ing them. I have to assume that your code snippet is not complete. EF never, ever automatically executes eager loading. (Unless, maybe, you have some outlandish and buggy query provider).
As for the first problem, the serialization. You should be able to tackle that in a similar way as shown above. Just load the data you want to serialize in isolation and disable proxy creation. Or, as suggested by others, serialize view models or anonymous types exactly containing what you need there.
As for the second problem, the validation exception. I can only imagine this to happen if you initialize a students collection by default, empty, Student objects. These are bound to be invalid. If this is not the case, I suggest you ask a new question about this specific problem, showing ample detail about the involved classes and mappings. That shouldn't be dealt with in this question.
Explicitly select what you want to return from the Database.
Use Select new. With the select new clause, you can create new objects of an anonymous type as the result of a query and don't let the reference include in. This syntax allows you to construct anonymous data structures. These are created as they are evaluated (lazily). Like this:
using (var ctx = new myEntities())
{
var AllPersons = ctx.People.Select(c => new {c.PersonId, c.PersonName}).ToList();
}
And even you don't need to disable lazy loading anymore.
After running query above:
This query currently allocates an anonymous type using select new { }, which requires you to use var. If you want allocate a known type, add it to your select clause:
private IEnumerable<MyClass> AllPersons;//global variable
using (var ctx = new myEntities())
{
AllPersons = ctx.People
.Select(c => new MyClass { PersonId = c.PersonId, PersonName = c.PersonName }).ToList();
}
And:
public class MyClass
{
public string PersonId { get; set; }
public string PersonName { get; set; }
}
If entities are auto generated, then copy paste it to own code and remove the relation generated like child collection and Foreign key. Or you don't need all this kind of the functionality might be can user lightweight framework like dapper
In normally your student collection doesn't fill from database. it's fill when you reach to property. In addition if you use ToList() method so Entity Framework read data from data to fill your collection.
Pls check this.
https://msdn.microsoft.com/en-us/data/jj574232.aspx#lazy
https://msdn.microsoft.com/en-us/library/vstudio/dd456846(v=vs.100).aspx
Is there anyway to don't let the reference include in any time in this situation?
The solution to this seems to be very simple: don't map the association. Remove the Student collection. Not much more I can say about it.
Decorate any properties with [IgnoreDataMember] if you are using 4.5+
https://msdn.microsoft.com/en-us/library/system.runtime.serialization.ignoredatamemberattribute(v=vs.110).aspx
Also sounds like you are trying to do table inheritance which is a different problem with EF
http://www.asp.net/mvc/overview/getting-started/getting-started-with-ef-using-mvc/implementing-inheritance-with-the-entity-framework-in-an-asp-net-mvc-application
http://www.entityframeworktutorial.net/code-first/inheritance-strategy-in-code-first.aspx
If I understand you correctly, you're just trying to make sure you only get what you specifically ask for right?
This was mentioned a little above, but to do this correctly you just want to select an anonymous type.
var students = from s in _context.Students
select new{
StudentId,
StudentNo};
Then, when you want to update this collection/object, I'd recommend use GraphDiff. GraphDiff really helps with the problems of disconnected entities and updates (https://github.com/refactorthis/GraphDiff)
So your method would look similar to this:
void UpdateStudent(Student student){
_context.UpdateGraph(student, map =>
map
.AssociatedEntity(c => c.Person));
_context.SaveChanges();
}
This way, you're able to update whatever properties on an object, disconnected or not, and not worry about the association.
This is assuming that you correctly mapped your entities, and honestly, I find it easier to declare the object as a property, not just the ID, and use a mapping file to map it correctly.
So:
class Person{
int Id{get;set;}
string Name{get;set}
}
class Student{
int Id{get;set;}
string StudentNo{get;set;}
Person Person{get;set;}
public class StudentMap : EntityTypeConfiguration<Student>
{
public StudentMap()
{
// Primary Key
HasKey(t => t.Id);
// Table & Column Mappings
ToTable("Students");
Property(t => t.Id).HasColumnName("StudentId");
// Relationships
HasRequired(t => t.Person)
.HasForeignKey(d => d.PersonId);
}
}
Hopefully that makes sense. You don't need to create a view model, but you definitely can. This way does make it easier to map disconnected items back to the database though.
I had exact same situation.
All I did to solve it was ask for the Student.ToList() before I asked for Persons.ToList()
I didn't have to disable lazy loading. Just need to load the table that has reference to other table first after that you can load the other table and first table results are already in memory and don't get "fixed" with all the references.
They are automatically linked in the ObjectContext by there EntityKey. Depending on what you want to do with your Persons and Students, you can Detach them from the ObjectContext :
using (var ctx = new myEntities)
{
ctx.Configuration.ProxyCreationEnabled = false;
ctx.Configuration.LazyLoadingEnabled = false;
AllPersons= ctx.Persons.ToList();
foreach(var c in AllPersons)
{
ctx.Detach(c);
}
AllStudents = ctx.Student.ToList();
foreach(var c in AllStudents )
{
ctx.Detach(c);
}
}
UPDATE 18 Sep 2013
It looks like there isn't an easy way to do this. I'm holding out for a solution that involves some extension to Entity Framework.
If you'd like to see these features in Entity Framework, vote for them on the user voice site, perhaps here and here
There are several similar questions on SO but I can't find a question new and similar enough to have the answer I'm looking for.
If this looks like information overload, jump down to In Summary.
Background
I'm writing a WebApi REST service to expose some pre-existing data through an OData end point. I'm using the EntitySetContoller<TEntity, TKey> to do all the grunt work for me. As well as the standard OData parameters, that are routed and translated by the base class, I've added some custom parameters, to allow specific functionality for my controller.
My database server is MS SQL Server with a full text index on the [BigText] NVarChar[4000] column of the [SomeEntity] table.
I have one limitation, I must use a Code First model.
// Model POCO
public class SomeEntity
{
public int Id { get; set; }
public string BigText { get; set; }
}
// Simple Controller
public class SomeEntityController : EntitySetController<SomeEntity, int>
{
private readonly SomeDbContext context = new SomeDbContext();
public override IQueryable<SomeEntity> Get()
{
var parameters = Request.GetQueryNameValuePairs()
.ToDictionary(p => p.Key, p => p.Value);
if (parameters.ContainsKey("BigTextContains")
(
var searchTerms = parameters["BigTextContains"];
// return something special ...
)
return this.context.SomeEntities;
}
// ... The rest is omitted for brevity.
}
The Problem
How to implement the // return something special ... part of my example?
Obviously, the niave
return this.context.SomeEntities.Where(e =>
e.BigText.Contains(searchTerm));
is completely wrong, it composes to a WHERE clause like
[BigText] LIKE '%' + #searchTerm + '%'
This doesn't use Full Text Searching so, doesn't support complex search terms and otherwise, performs terribley.
This approach,
return this.context.SomeEntities.SqlQuery(
"SELECT E.* FROM [dbo].[SomeEntity] E " +
"JOIN CONTAINSTABLE([SomeEntity], [BigText], #searchTerm) FTS " +
" ON FTS.[Key] = E.[Id]",
new object[] { new SqlParameter("#searchTerm", searchTerm) })
.AsQueryable();
Looks promising, it actually uses Full Text Searching, and is quite functional. However, you'll note that DbSqlQuery, the type returned from the SqlQuery function does not implement IQueryable. Here, it is coerced to the right return type with the AsQueryable() extension but, this breaks the "chain of composition". The only statement that will be performed on the server is the one specified in the code above. Any additional clauses, specified on the OData URL will be serviced on the API hosting web server, without benefitting from the indices and specialised set based functionality of the database engine.
In Summary
What is the most expedient way of accessing MS SQL Server's Full Text Search CONTAINSTABLE function with an Entity Framework 5 Code First model and acquiring a "composable" result?
Do I need to write my own IQueryProvider? Can I extend EF in some way?
I don't want to use Lucene.Net, I don't want to use a Database Generated Model. Perhaps I could add extra packages or wait for EF6, would that help?
It is not perfect, but you can accomplish what you are after with 2 calls to the database.
The first call would retrieve a list of matching key's from CONTAINSTABLE and then the second call would be your composable query utilizing the IDs that you returned from the first call.
//Get the Keys from the FTS
var ids = context.Database.SqlQuery<int>(
"Select [KEY] from CONTAINSTABLE([SomeEntity], [BigText], #searchTerm)",
new object[] { new SqlParameter("#searchTerm", searchTerm) });
//Use the IDs as an initial filter on the query
var composablequery = context.SomeEntities.Where(d => ids.Contains(d.Id));
//add on whatever other parameters were captured to the 'composablequery' variable
composablequery = composablequery.Where(.....)
I had this same issue recently:
EF 5 Code First FTS Queriable
Let me extend that post.
Your first option was mine first as well - using SqlQuery
I also needed to do more filtering, so instead of always writing full sql I used QueryBuilder, to which I made some changes and added more functions to fit my needs(I could upload it somewhere if needed):
QueryBuilder
After I have found another idea which I implemented.
Someone already mention it here, and that is to use SqlQuery that will return HashSet of Ids and that you can use it in EF queries with Contains.
This is better but not most optimal since you need 2 queries and Id list in memory.
Example:
public IQueryable<Company> FullTextSearchCompaniesByName(int limit, int offset, string input, Guid accountingBureauId, string orderByColumn)
{
FtsQueryBuilder ftsQueryBuilder = new FtsQueryBuilder();
ftsQueryBuilder.Input = FtsQueryBuilder.FormatQuery(input);
ftsQueryBuilder.TableName = FtsQueryBuilder.GetTableName<Company>();
ftsQueryBuilder.OrderByTable = ftsQueryBuilder.TableName;
ftsQueryBuilder.OrderByColumn = orderByColumn;
ftsQueryBuilder.Columns.Add("CompanyId");
if (accountingBureauId != null && accountingBureauId != Guid.Empty)
ftsQueryBuilder.AddConditionQuery<Guid>(Condition.And, "" , #"dbo.""Company"".""AccountingBureauId""", Operator.Equals, accountingBureauId, "AccountingBureauId", "");
ftsQueryBuilder.AddConditionQuery<bool>(Condition.And, "", #"dbo.""Company"".""Deleted""", Operator.Equals, false, "Deleted", "");
var companiesQuery = ftsQueryBuilder.BuildAndExecuteFtsQuery<Guid>(Context, limit, offset, "Name");
TotalCountQuery = ftsQueryBuilder.Total;
HashSet<Guid> companiesIdSet = new HashSet<Guid>(companiesQuery);
var q = Query().Where(a => companiesIdSet.Contains(a.CompanyId));
return q;
}
However EF 6 now has something called Interceptors that can be used to implement queriable FTS, and it is pretty simple and generic(last post):
EF 6 Interceptors for FTS.
I have tested this and it works fine.
!! REMARK: EF Code First, even with version 6, does not support Custom Stored Procedures.
There are only some for predefined CUD operations if I understood it well:
Code First Insert/Update/Delete Stored Procedure Mapping, so it can't be done with it.
Conclusion: if you can use EF 6 go for third options, is gives all you need.
If you are stucked with EF 5 or less, second option is better then first but not most optimal.
Let's say I have a User Entity, and created partial User class so I can add some methods (like with NHibernate). I added GetByID to make getting user easier:
public static User GetByID(int userID)
{
using (var context = new MyEntities())
{
return context.Users.Where(qq => qq.UserID == userID).Single();
}
}
Now, somewhere in business logic I'd like to do something like this:
var user = User.GetByID(userID);
var posts = user.GetAllPostsForThisMonth();
foreach(var post in posts)
{
Console.WriteLine(post.Answers.Count);
}
GetAllPostsForThisMonth() is similar to GetByID - has context and is disposing it right after execution.
Normally I can't do this because context is disposed when I call post.Answers.Count. This, I think, renders my methods useless... Or am I missing something? Can I anyhow use my entities like this? Or should I create method for every single query I use (like post.GetAnswersCount())? Thanks in advance!
The behavior you're lamenting is actually good, because it keeps you from shooting yourself in the foot. If you had been allowed to do this, it would have cause n round-trips to the database (where n is the number of posts), and each one of those round-trips would have pulled all the data for all the Answers, when all you wanted was the Count. This could have an enormous performance impact.
What you want to do is construct an object that represents all the information you expect to need from the database, and then construct a LINQ query that will actually load in all the information you expect to use.
public class PostSummary
{
public Post Post {get;set;}
public int AnswerCount {get;set;}
}
public IEnumerable<PostSummary> GetPostSummariesByUserAndDateRange(
int userId, DateTime start, DateTime end)
{
using (var context = new MyEntities())
{
return context.Posts
.Where(p => p.UserId == userId)
.Where(p => p.TimeStamp < start && p.TimeStamp > end)
.Select(new PostSummary{Post = p, AnswerCount = p.Answers.Count()})
.ToList();
}
}
This produces a single SQL query and, in a single round-trip, produces exactly the information you wanted without loading in a ton of information you didn't want.
Update
If NHibernate works anything like Java's Hibernate, it won't do lazy loading after the context is disposed, either. Entity Framework does give you a lot of options along these lines: which one works best for you will depend on your particular situation. For example:
Rather than keeping your context scoped inside the data-access method, you can make the context last longer (like once per request in a web application), so that Lazy loading of properties will continue to work beyond your data-access methods.
You can eagerly load whatever entity associations you know you're going to need.
Here's an example of eager loading:
public GetAllPostsAndAnswersForThisMonth()
{
using (var context = new MyEntities())
{
return context.Posts.Include("Answers")
.Where(p => p.UserID == UserID)
.ToList();
}
}
However, since Entity Framework basically constitutes your "Data Access" tier, I would still argue that the best practice will be to create a class or set of classes that accurately model what your business layer actually wants out of the data tier, and then have the data access method produce objects of those types.
One method is to explicitly load related objects that you know you will need before you dispose the context. This will make the related data available, with the downside that if you don't need the related info it is wasted time and memory to retrieve. Of course, you can also handle this with flags:
... GetAllPostsForThisMonth(bool includeAnswers)
{
using (var context = new MyEntities())
{
context.ContextOptions.LazyLoadingEnabled = false;
// code to get all posts for this month here
var posts = ...;
foreach (var post in posts)
if (!post.Answers.IsLoaded)
post.Answers.Load();
return posts;
}
}
This is a pretty vague/subjective question. I want to know if this is the best way to send/retrieve data to/from the browser using ajax calls. On the back end webservice, I want to use the entity framework. Below are two example functions.
The criteria for "best" is speed of writing code, readable code, and robust architecture.
Thanks for any feedback and suggestions and comments.
Get Function
[WebMethod]
public AjaxEmployee EmployeeGetById(int employeeID, bool getTimeOff)
{
using (Time_TrackerEntities ctx = new Time_TrackerEntities())
{
var results = from item in ctx.Employees
where item.ID == employeeID
orderby item.Last_Name
select new AjaxEmployee
{
ID = item.ID,
Employee_ID = item.Employee_ID,
First_Name = item.First_Name,
Middle_Name = item.Middle_Name,
Last_Name = item.Last_Name,
Supervisor_ID = item.Supervisor_ID,
Active = item.Active,
Is_Supervisor = item.Is_Supervisor
};
var emp = results.FirstOrDefault();
if (getTimeOff)
{
var results2 = from item2 in ctx.Time_Off
where item2.Employee_ID == emp.Employee_ID
select new AjaxTime_Off
{
ID = item2.ID,
Employee_ID = item2.Employee_ID,
Date_Off = item2.Date_Off,
Hours = item2.Hours
};
emp.Time_Off = results2.ToList<AjaxTime_Off>();
}
return emp;
}
}
Save Function
[WebMethod]
public bool EmployeeSave(AjaxEmployee emp)
{
using (Time_TrackerEntities ctx = new Time_TrackerEntities())
{
var results = from item in ctx.Employees
where item.ID == emp.ID
select item;
var myEmp = results.FirstOrDefault();
if (myEmp == null)
{
myEmp = new Employee();
ctx.Employees.AddObject(myEmp);
}
myEmp.Employee_ID = emp.Employee_ID;
myEmp.First_Name = emp.First_Name;
myEmp.Middle_Name = emp.Middle_Name;
myEmp.Last_Name = emp.Last_Name;
myEmp.Supervisor_ID = emp.Supervisor_ID;
myEmp.Active = emp.Active;
myEmp.Is_Supervisor = emp.Is_Supervisor;
return ctx.SaveChanges() > 0;
}
}
There are a few improvements to be made.
Save() Method - Don't left-to-right copy, use EF built in logic
Instead of this:
myEmp.Employee_ID = emp.Employee_ID;
myEmp.First_Name = emp.First_Name;
myEmp.Middle_Name = emp.Middle_Name;
myEmp.Last_Name = emp.Last_Name;
myEmp.Supervisor_ID = emp.Supervisor_ID;
myEmp.Active = emp.Active;
myEmp.Is_Supervisor = emp.Is_Supervisor;
You can do this:
ctx.Employees.ApplyCurrentValues(emp).
What this does, is look for an entity with the same key in the graph (which there is, since you have just retrieved it with FirstOrDefault()), and override the scalar values with the entity you pass in - which is exactly what your doing.
So your 7 lines becomes 1, plus if you add any extra scalar properties - you won't have to refactor your code. Just remember - only works for scalar properties, not navigational properties.
Why build query for primary key retrieval? Just use predicate to SingleOrDefault()
Instead of this:
var results = from item in ctx.Employees
where item.ID == emp.ID
select item;
var myEmp = results.FirstOrDefault();
Do this:
var myEmp = ctx.Employees.SingleOrDefault(x => x.ID == emp.Id);
Or even better, use a pipe/filter technique:
var myEmp = ctx.Employees.WithId(emp.Id).SingleOrDefault();
Where WithId is an IQueryable<Employee> extension method, which filters the query based on the supplied employee ID. This allows de-coupling of filtering/business logic from your repository/DAL. It should go in your domain model, so you can have a nice fluent API for query your domain entities via your ORM.
When your retrieving an entity via the primary key, you should always use SingleOrDefault() or Single(), never FirstOrDefault() or First(). If it's a primary key - there should only be one of them, so you should throw an exception if more than one exists, which is what SingleOrDefault() does. And as #Shiraz mentions - your FirstOrDefault() will crash the query below. You always need null checking when you use <First/Single>OrDefault().
The same improvements can be made to your Get method.
Overall, there is nothing functionally wrong with your code - it just needs subtle improvements, null checking and exception handling.
The only functional improvement i highly recommend is refacting your web service code into a Generic Repository. As the code is very trivial and can be re-used across any entity. The web service shouldn't be concerned with transactions, primary key's or EF logic whatsoever. It shouldn't even have a reference to the EF DLL. Encapsulate this logic behind a repository and delegate the persistence logic to there (via an interface of course).
After making the changes i've mentioned above, your web service methods should have no more than 5-7 lines of code each.
You have far too much intelligence in your web service - it should be dumb and persistent ignorant.
I find that it's usually a pretty bad idea to try and use my entities directly on the data contract. It's possible, and works fine in certain cases, but anytime my object model gets even a little complex I start having to worry about the object graph in ways that I don't want to have to.
Instead, and this is regardless of the client, but it's applicable to a JS client just as much, I try and think of the data contract classes as pure data trucks (DTOs) with no mapping in EF at all. Those classes are just he documents I'm passing back and forth, the message body if you will. They might translate into commands on my model or they might be used to populate a query, or whatever, but they're not the entities themselves.
This, I find, simplifies things a great deal. It may feel like more code when you first write a simple service, but over the lifetime it makes things a great deal more maintainable.
And just as a side note, you should also consider separating your responsibilities a bit better. The web service class should not have the responsibility of directly creating adn disposing of the data context, it should depend on a DAO or repository interface (or domain service) which handles all that stuff for you (and applies transactions as needed, etc).
Your get method can crash.
If this line returns null:
var emp = results.FirstOrDefault();
Then this line will crash with a null reference exception:
where item2.Employee_ID == emp.Employee_ID
I would also put in some try catch blocks in with logging of errors.
I'm running into a common need in my project to return collections of my model objects, plus a count of certain types of children within each, but I don't know if it is possible or how to model a "TotalCount" property in a Model class and populate it as part of on single Entity Framework query, preferably using LINQ queries. Is it possible to do this whilst being able to use the Entity Framework .Include("Object") and .Skip() and .Take()? I'm new to the Entity Framework so I may be missing tons of obvious stuff that can allow this...
I would like to be able to paginate on the dynamically constructed count properties as well. I'm thinking that the most scalable approach would be to store the counts as separate database properties and then simply query the count properties. But for cases where there are small row counts that I'm dealing with, I'd rather do the counts dynamically.
In a model like this:
Table: Class
Table: Professor
Table: Attendee
Table: ClassComment
I'd like to return a list of Class objects in the form of List, but I would also like the counts of Attendees and Class comments to be determined in a single query (LINQ preferred) and set in two Class properties called AttendeeCount and ClassCommentCount.
I have this thus far:
var query = from u in context.Classes
orderby tl.Name
select u;
List<Class> topics = ((ObjectQuery<Class>)query)
.Include("ClassComments")
.Skip(startRecord).Take(recordsToReturn).ToList();
Any suggestions or alternative query approaches that can still allow the use of .Include() and pagination would be much much appreciated, in order to produce a single database query, if at all possible. Thank you for any suggestions!
Try this:
public class ClassViewModel {
public Class Class { get; set; }
public int AttendeeCount { get; set; }
public int ClassCommentCount { get; set; }
}
var viewModel = context.Classes.Select(clas =>
new ClassViewModel {
Class = clas,
AttendeeCount = clas.ClassAttendes.Count,
ClassCommentCount = clas.ClassComments.Count}
).OrderBy(model => model.ClassCommentCount).Skip(startRecord).Take(recordsToReturn).ToList();
You don't have to include comments to get count.
It will not work this way. The easiest approach is to use projection into anonymous (or custom) non entity type. I would try something like this:
var query = context.Classes
.Include("ClassComments") // Only add this if you want eager loading of all realted comments
.OrderBy(c => c.Name)
.Skip(startRecord)
.Take(recordsToReturn)
.Select(c => new
{
Class = c,
AttendeeCount = c.Attendees.Count(),
ClassCommentCount = c.ClassComments.Count() // Not needed because you are loading all Class comments so you can call Count on loaded collection
});
The problem in your requirement are AttendeeCount and ClassCommentCount properties. You can't easily add them to your model because there is no corresponding column in database (unless you define one and in such case you don't need to manually count records). You can define them in partial Class implementation but in such case you can't use them in Linq-to-entities query.
The only way to map this in EF is to use DB view and create special read only entity to represent it in your applicaiton or to use DefiningQuery which is custom SQL command defined in SSDL instead of DB table or view.