I am hoping to create a base Entity Class that includes a validation rule that checks if a field called "Title" is unique (which of course requires a db scan). I want the inherited models to run the validation rule in the repo (or service) layer and send a ValidationResult to the (MVC) client-layer.
The problem is one of inheritance.
public interface IUniqueTitle
{
int Id { get; set; }
string Title { get; set; }
// This is a "multi-client, one database" solution.
// Data is isolated using SiteId
int SiteId { get; set; }
}
// Models such as "MemberClub" and "Assessment" will inherit from this
public class EntityUniqueTitle : IUniqueTitle
{
public int Id { get; set; }
public int SiteId { get; set; }
public string Title { get; set; }
}
// This class will be used in production
public class MemberClub : EntityUniqueTitle
{
}
I wrote an extension method that to check to see if the Title field is unique based on the SiteId
public static bool IsUniqueTitle<T>(this IQueryable<T> items, T currentEntity) where T : IUniqueTitle
{
return items.Where(
item => item.Id != currentEntity.Id // INCASE UPDATING OBJECT
& item.SiteId == currentEntity.SiteId
& item.Title == currentEntity.Title)
.Count() == 0;
}
Here is where I get stuck. Where should I put the validation?
I can put in the Repo but can't figure out how to fire the ValidationResult upon Save
public class RepoUniqueTitle<T> : IRepoUniqueTitle<T> where T : EntityUniqueTitle, new()
{
protected readonly DbContext c;
public Repo(IDbContextFactory f) { c = f.GetContext(); }
public void Insert(T o)
{
if (!c.Set<T>().IsUniqueTitle(o))
{
// ***********************
// PROBLEM HERE, HOW DO I STOP AND SEND A VALIDATIONRESULT TO THE CLIENT?
// IF POSSIBLE, AUOTMATIC WHEN MODEL.ISVALID IS CALLED
// code from base repo class for reference
// if (o is IUniqueTitleForSite)
// IoC.Resolve<IRepoUniqueTitle<T>>().Validate(o);
}
else
c.Set<T>().Add(o);
}
}
I am hoping there is a validation solution that:
Models can inherit from a base Entity
Can make db calls to the inherited entity's collection
Works with ValidationResult so it can be cleanly integrated into MVC Tier
Called during Model.isValid if possible
Note: I am using ProDinner as a basis for an "n-tier code-first EF mvc/wf" solution.
Sorry, a lot of this is new to me. Any help you can provide would be greatly appreciated!
If you have separate business logic layer you should place the validation to that layer. Otherwise why to have that layer if you don't use it to execute business rules?
Anyway unique check is tricky because there is a delay between your query and actual saving of data and another thread can insert the item with the same title during that delay. You should place unique index on Title and SiteId to enforce uniqueness in the database. In such case former problem will result in the exception which you must handle somehow but it is probably better then data duplicity.
Related
I have an application based on ASP.Net MVC 6 / ASP.Net Core 1.0, which exposes CRUD methods through a Web API.
I am using Entity Framework 7.0.0-rc1, I plan to migrate everything to EF Core 1.0 / ASP.Net Core 1.0 but haven't done this yet.
Here's a snippet of the type of model which causes me an design issue :
public class ParentEntity
{
public int Id { get; set; }
public virtual ICollection<ChildEntity> ChildCollection { get; set; } = new ChildCollection();
}
public class Child
{
public int Id { get; set; }
public virtual ParentEntity ParentEntity { get; set; }
public int? ParentEntityId { get; set; }
}
All relationships are properly ensured in my ApplicationContext class, and all CRUD operations work as expected, including Adds and Updates of the Child entities when a ParentEntity is saved.
For example, here's the kind of Save method I expose through the API :
// On some web API Save method :
[HttpPost]
public JsonResult Save(ParentEntity entity)
{
if (entity.Id > 0) _context.Update(entity);
else _context.Add(entity);
var result = _context.SaveChanges();
return Json(result);
}
The entity passed to the save method is coming from my AngularJS backend. As long as child entities have been added or modified, everything works as expected.
Let's now imagine the back end has received a ParentEntity loaded with two ChildEntity named A and B.
Now on the backend I remove the B child entity and post it to the save method.
The backend saves the changes of the data it receives : all updates are persisted on ParentEntity and on ChildEntity A.
But the B ChildEntity remains in the datastore, which seem logical since there was no indication on what to do with it.
So my question is, since it seems to me it is a very common case : how do I deal with that ?
I have imagined storing somewhere the child collection, save the changes, reload the root entity, compare the collection, determine which should be removed, then remove it, but it seems pretty hard and dirty work to me. Is there a better way ?
The answer to your question is here
https://msdn.microsoft.com/da-dk/magazine/mt694083
According to some tips explained in the MSDN blog post above, I have implemented a Client state field on every entity, through it's base class :
public enum ObjectState
{
Unchanged = 0,
Added = 1,
Modified = 2,
Deleted = 3
}
public interface IObjectWithState
{
ObjectState ClientState { get; set; }
}
public abstract class BaseEntity : IBaseEntity, IPrimaryKey
{
public int Id { get; set; }
public string Name { get; set; }
[NotMapped]
public bool IsNew => Id <= 0;
[NotMapped]
public ObjectState ClientState { get; set; } = ObjectState.Unchanged;
}
Whenever a child entity is removed from a collection on the client side, instead of being physically removed and never posted back, I will keep it on the collection, and mark it as deleted.
Then I just have to deal with that on server side :
_context.ChildEntityCollection.RemoveRange(
entity.ChildCollection.Where(c => c.ClientState == ObjectState.Deleted).ToList()
);
_context.SaveChanges();
I implemented the whole enum on every entity, but currently I only use the Deleted value, and only through a relation. Up till now, no need for other values, or any value on root entities.
I'm implementing a POCO in my project that represents a row in my database table. I'd like to modify one of the values in the constructor.
Unfortunately, it seems that the values are populated only after the constructor is run, so there's no way for me to perform my required logic. Is this a bug or by design?
I should probably mention that I'm using Code First.
public partial class CheckpointValue
{
[Key, DatabaseGenerated(DatabaseGeneratedOption.Identity)]
public int Id { get; set; }
[Column("saljare")]
public int SalesAgentId { get; set; }
[Column("volym")]
public int Value { get; set; }
[Column("datum")]
public DateTime Date { get; set; }
[Column("typ")]
public string Type { get; set; }
public CheckpointValue()
{
// Values empty... Why haven't they been populated when the constructor is run?
}
}
Unfortunately, it seems that the values are populated only after the
constructor is run, so there's no way for me to perform my required
logic. Is this a bug or by design?
This is by design. BTW, how you would be able to get these properties already populated during construction-time without providing constructor arguments?.
Maybe you're trying to implement your logic in the wrong place. If you're looking for implementing business rules, domain validation or who knows what, it should be done in another class. For example, your repository would be a good place to do things before returning the requested object.
public CheckpointValue GetById(Guid id)
{
CheckpointValue checkpointValue = YourDbContext.CheckpointValues.Find(id);
// Implement here what you wanted in your class constructor...
return checkpointValue;
}
I have an application with 4 layers:
-Core (Models)
-Repository (EF DbContext actions)
-Service (Business logic)
-Web (MVC)
I'm trying to update an object with a 1:1 relationship with EF using the following method:
public async Task<bool> UpdateProductTicketing(ProductTicketing ticketing)
{
var product = await GetProductByIdAsync(ticketing.ProductId);
// Validation removed for simplicity
// 'ticketing' passed validation so let's
// just replace it with the existing record.
product.Ticketing = ticketing;
_repo.ProductRepository.Update(product);
return await _repo.SaveAsync();
}
This works for an initial insert, but it doesn't work as I'd expect when I'm updating the record:
A first chance exception of type 'System.Data.Entity.Infrastructure.DbUpdateException' occurred...
The actual error message is:
Violation of PRIMARY KEY constraint 'PK_dbo.ProductTicketing'. Cannot insert duplicate key in object 'dbo.ProductTicketing'. The statement has been terminated.
Obviously the PK and FK "ProductId" doesn't change - so why does EF try to drop and insert my record instead of just updating it, and why does it fail?
But more importantly - how can I prevent this. I know I can manually map the object values and then update it - that works but it's tedious mapping two identical objects together and doesn't feel correct.
My repository for retrieving the Product object is in my Repository layer, while the method above is in my Service layer.
This is how I'm currently resolving this - and it looks as dirty as it feels:
public async Task<bool> UpdateProductTicketing(ProductTicketing ticketing)
{
var product = await GetProductByIdAsync(ticketing.ProductId);
// Validation removed for simplicity
if (product.Ticketing == null)
{
product.Ticketing = ticketing;
}
else
{
product.Ticketing.AllowEventBooking = ticketing.AllowEventBooking;
// Doing the same for all other properties etc
// etc
// etc
}
_repo.ProductRepository.Update(product);
return await _repo.SaveAsync();
}
How can I achieve this without doing all this horrible mapping an object to an identical object?
Edit
Here are the two models referred to above:
[Table(#"Products")]
public class Product
{
[Key]
public int Id { get; set; }
public virtual ProductTicketing Ticketing { get; set; }
// Removed others for clarity
[Timestamp]
public byte[] RowVersion { get; set; }
}
[Table(#"ProductTicketing")]
public class ProductTicketing
{
[Key, ForeignKey("Product")]
public int ProductId { get; set; }
public bool AllowEventBooking { get; set; }
// Removed others for clarity
public virtual Product Product { get; set; }
}
It's also probably worth noting that the "ProductTicketing" object I'm passing into the UpdateProductTicketing method is a new object created from values in my controller - but the ID is the same so I assume it should work.
I think I see the problem now - when you do product.Ticketing = ticketing;, EF treats this as a new insert.
To avoid this, you can do one of these things:
Continue using the workaround (which is not a wokaround actually but just the way EF expects you to tell when to insert vs. when to update).
Now this depends on rest of your code and design, but instead of fetching the product, you can fetch the ticket and update its properties. Of course, this means that if the ticketing is not found, you need to insert it which then kinda looks like what you're already doing with UpdateProductTicketing.
Use the InsertOrUpdate pattern (I made some assumptions about your code but hopefully it gives you the idea - the main thing here is the InsertOrUpdate method):
public class ProductRepository : IRepository
{
private SomeContext context;
public void InsertOrUpdate(ProductTicketing ticketing)
{
context.Entry(ticketing).State = ticketing.ProductId == 0 ?
EntityState.Added :
EntityState.Modified;
context.SaveChanges();
}
}
// And a generic version
public void InsertOrUpdate<T>(T entity) where T : class
{
if (context.Entry(entity).State == EntityState.Detached)
context.Set<T>().Add(entity);
context.SaveChanges();
}
You are getting that error because ef thinks that the ProductTicket is a new entity and is trying to insert the entity into the db. I don't know about the _repo.ProductRepository.Update(product) call but how about you attach the ProductTicket to the context and set the entity state to modified
My domain class:
public class Address
{
[Key]
public virtual string AddressId { get; set; }
public virtual string Address { get; set; }
}
In my MVC controller I want to check the given Address exist, before I insert.
public ActionResult Create(Address address)
{
if (ModelState.IsValid)
{
if (db.Addresses.Any(a => a.AddressId == address.AddressId)) // how I do it now
{
ModelState.AddModelError(string.Empty, "Address Id already exists!");
}
else
{
db.Addresses.Add(address);
db.SaveChanges();
return RedirectToAction("Index");
}
}
}
But there are lot of other domain classes in my project and I want to do the same check again and again.
My question is I want to write a generic method in my Db context class to perform this check. (looks like below or similar)
public bool Exists(object) {
// return true if exist
}
i.e. a method which I can call like this:
db.Exists(address)
Thanks!
You could use generics and do something like the following:
public class YourDbContext : DbContext
{
...
public bool Exists<TEntity>(object id)
where TEntity : class
{
var dbSet = Set<TEntity>();
var entity = dbSet.Find(id);
return entity != null;
}
Which you'd then use like:
db.Exists<Address>(address.AddressId);
Using Find isn't the most efficient way to handle this, but it has the key benefit that you're not required to know what the actual primary key property on the class is, which would greatly complicate this method. For example, Address has AddressId, but Foo might have FooId.
UPDATE
Since ultimately this just uses Find under the hood, you just have to modify the method slightly to be able to take multiple parameters. Find handles composite keys by allowing one more parameters to be passed to it. But bear in mind, the the order matters and must align with the key order you specified when configuring your entity.
public bool Exists<TEntity>(params object[] keys)
where TEntity : class
{
var dbSet = Set<TEntity>();
var entity = dbSet.Find(keys);
return entity != null;
}
I have a POCO domain model which is wired up to the entity framework using the new ObjectContext class.
public class Product
{
private ICollection<Photo> _photos;
public Product()
{
_photos = new Collection<Photo>();
}
public int Id { get; set; }
public string Name { get; set; }
public virtual IEnumerable<Photo> Photos
{
get
{
return _photos;
}
}
public void AddPhoto(Photo photo)
{
//Some biz logic
//...
_photos.Add(photo);
}
}
In the above example i have set the Photos collection type to IEnumerable as this will make it read only. The only way to add/remove photos is through the public methods.
The problem with this is that the Entity Framework cannot load the Photo entities into the IEnumerable collection as it's not of type ICollection.
By changing the type to ICollection will allow callers to call the Add mentod on the collection itself which is not good.
What are my options?
Edit:
I could refactor the code so it does not expose a public property for Photos:
public class Product
{
public Product()
{
Photos = new Collection<Photo>();
}
public int Id { get; set; }
public string Name { get; set; }
private Collection<Photo> Photos {get; set; }
public IEnumerable<Photo> GetPhotos()
{
return Photos;
}
public void AddPhoto(Photo photo)
{
//Some biz logic
//...
Photos.Add(photo);
}
}
And use the GetPhotos() to return the collection. The other problem with the approach is that I will loose the change tracking abilities as I cannot mark the collection as Virtual - It is not possible to mark a property as private virtual.
In NHibernate I believe it's possible to map the proxy class to the private collection via configuration. I hope that this will become a feature of EF4. Currently i don't like the inability to have any control over the collection!
The way to do this is to have a protected virtual property which is mapped in your model and a public property that returns an IEnumerable.
public class Product
{
public Product()
{
PhotoCollection = new Collcation<Photo>();
}
public int Id { get; set; }
public string Name { get; set; }
protected virtual ICollection<Photo> PhotoCollection {get; set; }
public IEnumerable<Photo> Photos
{
get { return PhotoCollection ; }
}
public void AddPhoto(Photo photo)
{
//Some biz logic
//...
PhotoCollection .Add(photo);
}
}
Anton, it would help me understand your problem more if you can explain why is it that you do not want developers to access the Add method of your collection. Is this because the list is strictly read-only, or is it because you want to run some custom business logic when a new entity is added?
Anyway... I am going to assume that you are trying to do the latter (i.e. run custom business logic when the collection is modified). I have done a similar solution on a project of mine, and the idea is as follows:
The TT template that produces POCOs in EF4 creates all collections as TrackableCollection lists. This class has an event called 'CollectionChanged' which you can subscribe to and listen to any changes to your collection.
So you can do something as follows:
public class Product
{
public Product()
{
Photos.CollectionChanged += ListCollectionChanged;
}
public int Id { get; set; }
public string Name { get; set; }
public TrackableCollection<Photo> Photos
{
get
{
// default code generated by EF4 TT
}
set
{
// default code generated by EF4 TT
}
}
private void ListCollectionChanged(object sender, NotifyCollectionChangedEventArgs e)
{
switch (e.Action)
{
// A new item has been added to collection
case NotifyCollectionChangedAction.Add:
{
T newItem = (T) e.NewItems[0];
// Run custom business logic
}
break;
// An existing item has been removed
case NotifyCollectionChangedAction.Remove:
{
T oldItem = (T) e.OldItems[0];
// Run custom business logic
}
break;
}
}
}
The nice thing about the above solution is that you still use your Product entity in an 'EF' manner... were any developer in your team can simply access a property of the entity directory and need run an explicit hard typed function.
Bit late to the party but this is what Observable objects are for. Allow the data structure to do what it does best. Use ObservableCollection as your field type if you don't want to build your own collection that does what you need and expose the regular ICollection type from your property. You can run any logic in the parent entity you need when the related entities in the collection change via the CollectionChanged event. If you need to selectively enable or disable modifications it's easy enough to extend an existing collection type or write a proxy collection that allows a call to a method to toggle the mutability of the collection (ISupportInitialize can be used to good effect for representing this ability BTW).
(Apologies for my initial post brevity - I was answering from my phone)
You can construct your collection through a LINQ query over an EF entity set. However, you keep the resulting collection as internal data member to your business class and expose the IEnumerable<Photo> returned by calling AsEnumerable() on the entity set as a result of the public photo.
You could cache the IEnumerable<Photos> internally as well, so that you don't call AsEnumerable() every time your caller asks for the collection. Of course, that means that if the user needs to update the collection through your public methods, you might have to refresh the cached IEnumerable. This might pose small issue if the caller has also cached the pointer to the previous IEnumerable.
Alternatively, if your caller will always work with the full entity set, the EntitySet class (of which all your EF sets will inherit) implements IEnumerable<TEntity>, so you can directly return the entity set to your caller.
Note that if you want the loading of the collection from an EF entity set to happen outside of the scope of your business class, you can make a constructor on your class that takes an ICollection. This way, once you create your object, the collection is sealed in it, and exposed only as an IEnumerable.
Why not try the following and leave use properties?
private ICollection<Photo> photos{get; set;}
public IEnumerable<Photo> Photos
{
get {return (IEnumberable<Photo>)photos;}
}
Alternatively you could use the decorator pattern to encapsulate the class into one which the collection can't be directly modified.