I had a class called Document, which I split into two entities, in order to separate an expensive binary field:
[Table("Document")]
public class Document
{
[Key]
public int Id { get; set; }
... other fields ...
[Required]
public virtual DocumentBinary DocumentBinary { get; set; }
}
[Table("Document")]
public class DocumentBinary
{
[Key, ForeignKey("Document")]
public int DocumentId { get; set; }
public Document Document { get; set; }
public byte[] DocumentData { get; set; }
}
So, everything works fine, both entities share the same database table and DocumentData is only loaded when it's needed.
However, when it comes to updating the Document entity, I get an error stating that 'DocumentBinary is required'.
When I remove the [Required] attribute from DocumentBinary virtual property, I get the following error:
The entity types 'Document' and 'DocumentBinary' cannot share table 'Documents' because they are not in the same type hierarchy or do not have a valid one to one foreign key relationship with matching primary keys between them.
I can obviously do something like:
var test = document.DocumentBinary;
before updating the document object:
documentRepository.Update(document);
This will then load the binary data on my request and save the changes without any issues, but the whole point is that I shouldn't need to do that.
This can be achieved using the fluent API. If you remove the data annotations and in your OnModelCreating add this, it should work.
protected override void OnModelCreating(DbModelBuilder modelBuilder)
{
modelBuilder.Entity<Document>().HasRequired(d => d.DocumentBinary).
WithRequiredDependent(db => db.Document);
}
I managed to resolve it by overriding my Update method in DocumentRepository:
public override void Update(Document document)
{
try
{
DataContext.Entry(document.DocumentBinary).State = EntityState.Modified; // added this line
DataContext.Entry(document).State = EntityState.Modified;
}
catch (System.Exception exception)
{
throw new EntityException("Failed to update document");
}
}
I know it probably does the same thing as me evaluating DocumentBinary by assigning it to 'test' variable, but it looks like a much cleaner solution.
Related
I am working on .NET CORE 6 along with EF CORE 7. I need to seed data in joining table but unable to do so and get error.
I am seed FileTypeId but not sure why EF core migration throwing error...
error
The seed entity for entity type 'JobFileType' cannot be added because it has the navigation 'FileType' set. To seed relationships, add the entity seed to 'JobFileType' and specify the foreign key values {'FileTypeId'}. Consider using 'DbContextOptionsBuilder.EnableSensitiveDataLogging' to see the involved property values.
ClassA
public class JobProfile
{
public JobProfile()
{
this.JobFileTypes = new HashSet<JobFileType>();
}
public Guid JobProfileId { get; set; }
public string Name { get; set; }
public ICollection<JobFileType>? JobFileTypes { get; set; }
}
ClassB
public class FileType
{
public FileType()
{
this.JobFileTypes = new HashSet<JobFileType>();
}
public Guid FileTypeId { get; set; }
public string Extension { get; set; } = string.Empty;
public ICollection<JobFileType>? JobFileTypes { get; set; }
}
Joing Table
public class JobFileType
{
public Guid JobFileTypeId { get; set; }
public Guid JobProfileId { get; set; }
public JobProfile JobProfile { get; set; } = new JobProfile();
public Guid FileTypeId { get; set; }
public FileType FileType { get; set; } = new FileType();
}
Seed Extension
public static class JobFileTypeSeed
{
public static void Seed(this ModelBuilder modelBuilder)
{
modelBuilder.Entity<JobFileType>()
.HasData(
new JobFileType {JobFileTypeId = Guid.Parse("aaa"), JobProfileId = Guid.Parse("ccc"), FileTypeId = Guid.Parse("yyy") },
new JobFileType { JobFileTypeId = Guid.Parse("bbb"), JobProfileId = Guid.Parse("ccc"), FileTypeId = Guid.Parse("zzz") }
);
}
}
config
internal class JobFileTypeConfiguration : IEntityTypeConfiguration<JobFileType>
{
public void Configure(EntityTypeBuilder<JobFileType> builder)
{
builder.ToTable("JobFileType", "dbo");
builder.HasKey(column => column.JobFileTypeId);
builder
.HasOne(jobFileType => jobFileType.JobProfile)
.WithMany(jobProfile => jobProfile.JobFileTypes)
.HasForeignKey(jobFileType => jobFileType.JobProfileId);
builder
.HasOne(jobFileType => jobFileType.FileType)
.WithMany(fileType => fileType.JobFileTypes)
.HasForeignKey(jobFileType => jobFileType.FileTypeId);
}
}
There is not much to say about the concrete issue (which btw is not specific to joining entity, but any entity model seeding):
I am seed FileTypeId but not sure why EF core migration throwing error...
as the cause of the issue is included at the beginning of the error message:
because it has the navigation 'FileType' set.
And your entity has
public FileType FileType { get; set; } = new FileType();
// ^ ^ ^
// the problem
and the same for
public JobProfile JobProfile { get; set; } = new JobProfile();
which will be the next error if you resolve the original.
Remove both navigation property initializers (= new ...) and the problem will be gone.
As a general rule, you should never initialize reference navigation properties because it causes many side effects and/or improper behaviors (not only for seeding, but also eager/lazy/explicit data loading). Initializing collection navigation properties is arbitrary, but ok. Only reference navigation property initialization must be avoided. For more info, see EF codefirst : Should I initialize navigation properties? - quite old EF topic, but still applies.
If you are trying to resolve NRT warnings (as I guess), initializing with new is definitely not a proper way. One reason I don't like NRT is because it is forcing people to use "workarounds" for preventing compiler warnings, which in fact break the primary functionality. Specifically in EF Core, enabling NRT also changes the optional/required attribute of some properties, hence database column types (most noticeable for string properties/columns and reference navigations). You could read more about this in the Working with Nullable Reference Types topic in the official EF Core documentation, but in general I would just disable NRT for EF entity model classes.
The proper order is to set the "master data" first and then try to set the join table, as you would expect.
The defaulting
{get;set;} = new Something();
Could be the offending declaration, since any instance upon creation will have the relation JobFileType already set
I'm trying to fetch (in disconnected way) an entity with its all related entities and then trying to update the entity. But I'm getting the following error:
Attaching an entity of type 'Feature' failed because another entity of the same type already has the same primary key value.
public class Person
{
public int PersonId { get; set; }
public string Personname { get; set }
public ICollection Addresses { get; set; }
}
public class Address
{
public int AddressId { get; set; }
public int PersonId { get; set; }
public string Line1 { get; set; }
public string City { get; set; }
public string State { get; set; }
public Person Person { get; set; }
public ICollection<Feature> Features { get; set; }
}
// Many to Many: Represented in database as AddressFeature (e.g Air Conditioning, Central Heating; User could select multiple features of a single address)
public class Feature
{
public int FeatureId { get; set; }
public string Featurename { get; set; }
public ICollection<Address> Addresses { get; set; } // Many-To-Many with Addresses
}
public Person GetCandidate(int id)
{
using (MyDbContext dbContext = new MyDbContext())
{
var person = dbContext.People.AsNoTracking().Where(x => x.PersonId == id);
person = person.Include(prop => prop.Addresses.Select(x => x.Country)).Include(prop => prop.Addresses.Select(x => x.Features));
return person.FirstOrDefault();
}
}
public void UpdateCandidate(Person newPerson)
{
Person existingPerson = GetPerson(person.Id); // Loading the existing candidate from database with ASNOTRACKING
dbContext.People.Attach(existingPerson); // This line is giving error
.....
.....
.....
}
Error:
Additional information: Attaching an entity of type 'Feature' failed because another entity of the same type already has the same primary key value.
It seems like (I may be wrong) GetCandidate is assigning every Feature within Person.Addresses a new instance. So, how could I modify the GetCandidate to make sure that the same instance (for same values) is bing assisgned to Person.Addresses --> Features.
Kindly suggest.
It seems like (I may be wrong) GetCandidate is assigning every Feature within Person.Addresses a new instance. So, how could I modify the GetCandidate to make sure that the same instance (for same values) is bing assisgned to Person.Addresses --> Features.
Since you are using a short lived DbContext for retrieving the data, all you need is to remove AsNoTracking(), thus allowing EF to use the context cache and consolidate the Feature entities. EF tracking serves different purposes. One is to allow consolidating the entity instances with the same PK which you are interested in this case, and the second is to detect the modifications in case you modify the entities and call SaveChanges(), which apparently you are not interested when using the context simply to retrieve the data. When you disable the tracking for a query, EF cannot use the cache, thus generates separate object instances.
What you really not want is to let EF create proxies which hold reference to the context used to obtain them and will cause issues when trying to attach to another context. I don't see virtual navigation properties in your models, so most likely EF will not create proxies, but in order to be absolutely sure, I would turn ProxyCreationEnabled off:
public Person GetCandidate(int id)
{
using (MyDbContext dbContext = new MyDbContext())
{
dbContext.Configuration.ProxyCreationEnabled = false;
var person = dbContext.People.Where(x => x.PersonId == id);
person = person.Include(prop => prop.Addresses.Select(x => x.Country)).Include(prop => prop.Addresses.Select(x => x.Features));
return person.FirstOrDefault();
}
}
How to configure a EF6 migration with a model class having?
A collection o items
A navigation property to one particular item
public class MyModel
{
[Key]
public int Id { get; set; }
// My collection of elements
public virtual ICollection<MyCollectionElement> MyCollection { get; set; }
// Optional navigation to a particular element from the collection
[ForeignKey("CurrentElement")]
public int? CurrentElementId { get; set; }
public virtual MyCollectionElement CurrentElement { get; set; }
}
public class MyCollectionElement
{
[Key]
public int Id { get; set; }
// Required navigation to MyClass
[ForeignKey("MyModel")]
public int MyModelID { get; set; }
public virtual MyModel Model { get; set; }
}
Configuration
modelBuilder.Entity<MyModel>()
.HasMany(x => x.MyCollection)
.WithRequired(x => x.Model)
.HasForeignKey(x => x.MyModelID)
.WillCascadadeOnDelete(false);
Throws several errors on Update-Database, like
Unable to determine a valid ordering for dependent operations.
I would like a solution which doesn't involve a boolean IsCurrent in MyCollectionElement to make another query later and find which element is the current; instead, I would like to store the current element's id with my model, like exposed.
Also, I don't mind making int CurrentElementId non nullable (required) if it's easier.
Thanks.
This chicken-and-egg problem always looms when there are circular relationships. The error...
Unable to determine a valid ordering for dependent operations.
...is not thrown when the database is created. The database can be created just fine. It occurs when you try to insert a MyModel record and a MyCollectionElement referring to one another in the same unit of work. In the Seed method you probably have something like
var element = new MyCollectionElement();
var model = new MyModel();
model.MyCollection.Add(element);
model.CurrentElement = element;
The statement model.MyCollection.Add(element); requires model to be inserted first, so element can refer to it in its foreign key. But model.CurrentElement = element; requires element to be inserted first.
You can only avoid this situation by calling SaveChanges twice, and wrapping everything in a TransactionScope if you want the assignments to be transactional:
using(var ts = new TransactionScope())
{
using(var db = new MyContext()
{
var element = new MyCollectionElement();
var model = new MyModel();
model.MyCollection.Add(element);
db.MyModels.Add(model);
db.SaveChanges();
model.CurrentElement = element;
db.SaveChanges();
}
ts.Complete();
}
This also means that int CurrentElementId should remain nullable.
I have the following Entity class definition:
[Table("Users")]
public class WebUser
{
public virtual int Id { get; set; }
public virtual ICollection<Client> Clients { get; set; }
// more properties...
}
Notice that table name is different than the class name. I also have a ClientUsers table which is a many-to-many mapping for clients and users. Problem is, when I try to access the webUser.Clients property I get the following exception:
"Invalid object name 'dbo.ClientWebUsers'."
Looks like Entity Framework is trying to guess the name of the third table, but it apparently was not smart enough to take into account the table attribute that I have there. How can I tell EF that it is ClientUsers and not ClientWebUsers? Also what rule does it follow to know which table name comes first and which one comes second in the new table name? I think it's not alphabetical order.
I'm using EF 5.0. Thanks!
From the looks of things you're using Code First, so I'll answer accordingly. If this is incorrect, please let me know.
I believe the convention being used to determine the name of the many-to-many table is determined by the order in which they occur as DbSet properties in your SomeContext : DbContext class.
As for forcing EntityFramework to name your table whatever you like, you can use the Fluent API in the OnModelCreating method of your SomeContext : DbContext class as follows:
public class DatabaseContext : DbContext
{
public DatabaseContext()
: base("SomeDB")
{
}
public DbSet<WebUser> Users { get; set; }
public DbSet<Client> Clients { get; set; }
protected override void OnModelCreating(DbModelBuilder modelBuilder)
{
modelBuilder.Entity<WebUser>().HasMany(c => c.Clients)
.WithMany(p => p.WebUsers).Map(
m =>
{
m.MapLeftKey("ClientId");
m.MapRightKey("UserId");
m.ToTable("ClientUsers");
});
}
}
This assumes your classes are something like the following:
[Table("Users")]
public class WebUser
{
public virtual int Id { get; set; }
public virtual ICollection<Client> Clients { get; set; }
// more properties...
}
public class Client
{
public int Id { get; set; }
public ICollection<WebUser> WebUsers { get; set; }
// more properties
}
Finally, here's an integration test (NUnit) demonstrating the functionality working. You may need to drop your database before running it as Code First should want to update/migrate/recreate it.
[TestFixture]
public class Test
{
[Test]
public void UseDB()
{
var db = new DatabaseContext();
db.Users.Add(new WebUser { Clients = new List<Client> { new Client() } });
db.SaveChanges();
var webUser = db.Users.First();
var client = webUser.Clients.FirstOrDefault();
Assert.NotNull(client);
}
}
Edit: Link to relevant documentation for the Fluent API
Rowan's answer (adding here for reference):
Here is the information on how to configure a many-to-many table (including specifying the table name). The code you are after is something like:
modelBuilder.Entity<WebUser>()
.HasMany(u => u.Clients)
.WithMany(c => c.WebUsers)
.Map(m => m.ToTable("ClientUsers");
~Rowan
I have not run into this before, because I usually handled collections by them selves instead of modifying them directly on the entity.
public class Schedule: BaseEntity
{
public Guid Id {get;set;}
public virtual int? DayOfTheWeekTypeId { get; set; }
public virtual DayOfTheWeekType DayOfTheWeekType { get; set; }
public virtual ICollection<Instructor> Instructors { get; set; }
public DateTime? StartDateTime { get; set; }
public DateTime? EndDateTime { get; set; }
public string SpecialInstructions { get; set; }
}
Mapping class:
public ScheduleMapping()
{
HasMany(c => c.Instructors).WithMany().Map(m => { m.MapLeftKey("ScheduleId");
m.MapRightKey("InstructorId");
m.ToTable("Schedule_Instructors");
});
HasOptional(s => s.DayOfTheWeekType).WithMany().HasForeignKey(s => s.DayOfTheWeekTypeId).WillCascadeOnDelete(false);
Property(s => s.SpecialInstructions).IsMaxLength();
}
This is my update method:
public virtual void Update(TEntity entity)
{
if (entity == null)
throw new ArgumentNullException("entity");
//this is the original persisted entity
var persistedEntity = _repository.GetById(entity.Id);
if(originalEntity != null)
{
entity.Id = persistedEntity.Id;
UnitOfWork.ApplyCurrentValues<TEntity>(originalEntity,entity);
UnitOfWork.Commit();
}
}
This is the Method that handled the "merge"
public void ApplyCurrentValues<TEntity>(TEntity original, TEntity current) where TEntity : class
{
base.Entry<TEntity>(original).CurrentValues.SetValues(current);
}
If I modify the Instructors collection then try to apply the update, it seems to keep my original values. I have tried loading the the Schedule entity prior to the update and make my changes, but sometimes that causes a PK error (on the Instructors collection) in entity framework. As if it is trying to add an entity with the same key. So, instead I am rebuilding the Schedule entity (including the ID) manually and then updating it. When I do that I do not get any more errors, however, the Instructors collections doesn't change. I am thinking because CurrentValues. SetValues is being applied based on the persisted entity and not my updated version. the Should I handle my updates differently or do I need to manully
SetValues never updates navigation properties. When you execute your code it only knows about changes in simple / complex properties of the entity passed to your Update method. EF even don't know about related entities of the entity passed to your Update method.
You must manually tell EF about each change in your object graph - EF doesn't have any resolution mechanism for object graphs.