NHibernate unique constraints - nhibernate

I've run into some trouble with unique constraints in NHibernate.
I have a User entity that is mapped with a unique constraint for the Username property. What I want to do is to be able to check if a particular username exists before a new user is added and also before an existing user updates it's username.
The first scenario (adding a new user) works just fine. However, when I try to check if the username exists before updating an existing user, I get a constraint violation. Here's what the code for my Save method looks like.
public void Save<T>(T entity) where T : User
{
using (var session = GetSession())
using (var transaction = session.BeginTransaction())
{
try
{
CheckIfUsernameExists(entity);
session.SaveOrUpdate(entity);
session.Flush();
transaction.Commit();
}
catch (HibernateException)
{
transaction.Rollback();
throw;
}
}
}
The constraint is violated in the CheckIfUsernameExists() method and it looks like this:
public void CheckIfUsernameExists<T>(T entity) where T : User
{
var user = GetUserByUsername(entity);
if (user != null)
throw new UsernameExistsException();
}
private T GetUserByUsername<T>(T entity) where T : User
{
var username = entity.Username;
var idToExclude = entity.Id;
var session = GetSession();
var user = session.CreateCriteria<T>()
.Add(Restrictions.Eq("Username", username))
.Add(Restrictions.Not(Restrictions.IdEq(idToExclude)))
.UniqueResult() as T;
return user;
}
It is the session.CreateCriteria() line that causes it to crash resulting in an NHibernateException (SQLiteException) with the message "Abort due to constraint violation. column Username is not unique".
Is it related to the NHibernate cash? The entity that is passed to the save method has been updated with the desired username at the time session.CreateCriteria() is called.
Maybe I'm doing this all wrong (I am an NHibernate beginner) so please feel free to state the obvious and suggest alternatives.
Any help is much appreciated!

Hmm, I'm not sure about the core of the problem, but for the strategy of trying to see whether a user already exists, why do you need the ".UniqueResult()"?
Couldn't you just assume to get a list of users which match that username and which do not have the same id as your current user (obviously). Pseudo-code like I'd do something like this
public bool ExistsUsername(string username, int idToExclude)
{
IList<User> usersFound = someNHibernateCriteria excluding entries that have id = idToExclude
return (usersFound.Count > 0)
}

Two thoughts:
- Why don't you just SaveOrUpdate and see if you succeed. Is that not possible in your scenario?
- I've seen you mentioning SQLite. Is that your real production system, or just something you use for testing. If so, have you checked if it's SQLite that makes the problems, and the query works against a fully featured DBMS? - SQLite frequently makes that kind of problems, because it does not support every kind of constraint...

Are you sure the exception is thrown at CreateCriteria? Because, I don't see how you could get a SQLlite constraint exception from a select statement. I do virtually the same thing...
public bool NameAlreadyExists(string name, int? exclude_id)
{
ICriteria crit = session.CreateCriteria<User>()
.SetProjection(Projections.Constant(1))
.Add(Restrictions.Eq(Projections.Property("name"), name));
if (exclude_id.HasValue)
crit.Add(Restrictions.Not(Restrictions.IdEq(exclude_id.Value)));
return crit.List().Count > 0;
}
I would look at the order of the generated sql to see what's causing it. If that entity was loaded in that session, it could be getting updated before the query.

transaction.Rollback() doesn't remove your entity from session cache, use session.Evict() instead.
See:
- https://www.hibernate.org/hib_docs/nhibernate/html/performance.html#performance-sessioncache

Related

ASP.NET Core Authentication Id?

The tutorials on enabling authentication work all right, but what identifier should be used to store data for a user in the database? The only thing easily available is User.Name, which seems to be my email address.
I see in the database there is an AspNetUsers table with that as the UserName column, and a varchar Id column that appears to be a GUID and is the primary key. It seems like the 'Id' field is the logical value to use, but it's not readily available in my app. I found I can get to it like this:
string ID_TYPE = "http://schemas.xmlsoap.org/ws/2005/05/identity/claims/nameidentifier";
var id = User.Claims.Where(x => x.Type == ID_TYPE).Select(x => x.Value).FirstOrDefault();
But that seems like a weird way to go about it. Is that the proper value to use say if I want to create a 'Posts' table that has a user associated with a post?
I've looked at these pages and it seems that a lot of this might be due to Microsoft integrating the same login process with ActiveDirectory.
Is there a reason to make the id so hard to get to and the name so easy? Should I be using the name instead? Should I be careful not to let the user change their user name then?
The shortest path to UserId is:
User.FindFirst(ClaimTypes.NameIdentifier)?.Value;
Or create extension like so if you need to access UserId a lot:
public static class ClaimsPrincipalExtensions
{
public static string GetUserId(this ClaimsPrincipal principal)
{
if (principal == null)
return null; //throw new ArgumentNullException(nameof(principal));
string ret = "";
try
{
ret = principal.FindFirst(ClaimTypes.NameIdentifier)?.Value;
}
catch (System.Exception)
{
}
return ret;
}
}
Usage:
User.GetUserId()
In your controller use dependency injection to get the user manager:
Create a class MyUser that has your extended properties
public class MyUser : IdentityUser
{
public string MyExendedInfo { get; set; }
public int MyOtherInfo {get;set;}
}
add this property to the database using migration, or manually add it.
In Startup.cs in Configure Services add:
services.AddIdentity<MyUser, IdentityRole>()
Now inject this in your controller class:
private readonly UserManager<MyUser> _userManager;
public HomeController(
UserManager<MyUser> userManager)
{
_userManager = userManager;
}
Now you can access your additional proporties and your Id (if you still need this) in your action methods like this:
var user = await _userManager.GetUserAsync(HttpContext.User);
var id = user.Id;
var myExtendedInfo = user.MyExtendedInfo;
var myOtherInfo = user.MyOtherInfo;
etc
You can also update information about your user:
user.myExtendedInfo = "some string";
user.MyOtherInfo = myDatabase.pointer;
var result = await _userManager.UpdateAsync(user);
if (!result.Succeeded)
{
//handle error
}
So as long as you want only limited additional data stored in the database, you can create a custom user class, and use the Identity system to store it for you. I would not store it myself.
If however, you need to store large information in a separate table and/or reference the user from other tables, the Id is the correct field to use and you can access it as shown above.
I don't know what the best practice is for how much information can be stored in AspNetUsers, versus in claims, versus in your own table, but since the provided table already stores things like user name, phonenumber etc, I think it is Ok to extend it like this.

RavenDB fails with ConcurrencyException when using new transaction

This code always fails with a ConcurrencyException:
[Test]
public void EventOrderingCode_Fails_WithConcurrencyException()
{
Guid id = Guid.NewGuid();
using (var scope1 = new TransactionScope())
using (var session = DataAccess.NewOpenSession)
{
session.Advanced.UseOptimisticConcurrency = true;
session.Advanced.AllowNonAuthoritativeInformation = false;
var ent1 = new CTEntity
{
Id = id,
Name = "George"
};
using (var scope2 = new TransactionScope(TransactionScopeOption.RequiresNew))
{
session.Store(ent1);
session.SaveChanges();
scope2.Complete();
}
var ent2 = session.Load<CTEntity>(id);
ent2.Name = "Gina";
session.SaveChanges();
scope1.Complete();
}
}
It fails at the last session.SaveChanges. Stating that it is using a NonCurrent etag. If I use Required instead of RequiresNew for scope2 - i.e. using the same Transaction. It works.
Now, since I load the entity (ent2) it should be using the newest Etag unless this is some cached value attached to scope1 that I am using (but I have disabled Caching). So I do not understand why this fails.
I really need this setup. In the production code the outer TransactionScope is created by NServiceBus, and the inner is for controlling an aspect of event ordering. It cannot be the same Transaction.
And I need the optimistic concurrency too - if other threads uses the entity at the same time.
BTW: This is using Raven 2.0.3.0
Since no one else have answered, I had better give it a go myself.
It turns out this was a human error. Due to a bad configuration of our IOC container the DataAccess.NewOpenSession gave me the same Session all the time (across other tests). In other words Raven works as expected :)
Before I found out about this I also experimented with using TransactionScopeOption.Suppress instead of RequiresNew. That also worked. Then I just had to make sure that whatever I did in the suppressed scope could not fail. Which was a valid option in my case.

Time out expired while querying through criteria using nhibernate

I am using criteria to query the database based on the unique key. But I am coming through a weird scenario. After two or three queries, it starts giving me timeout expired error.
using (NHibernate.ISession session = m_SessionFactory.OpenSession())
{
using (ITransaction transacion = session.BeginTransaction())
{
if (cashActivity.ActivityState == ApplicationConstants.TaxLotState.Deleted || cashActivity.ActivityState == ApplicationConstants.TaxLotState.Updated)
{
IList<CashActivity> lsCActivity = RetrieveEquals<CashActivity>("UniqueKey",cashActivity.UniqueKey);
if (lsCActivity != null && lsCActivity.Count > 0)
cashActivity.CashActivityID = lsCActivity[0].CashActivityID;
}
if (cashActivity.ActivityState == ApplicationConstants.TaxLotState.Deleted)
{
session.Delete(cashActivity);
}
else
session.SaveOrUpdate(cashActivity);
}
}
}
public IList<T> RetrieveEquals<T>(string propertyName, object propertyValue)
{
using (Isession session = m_SessionFactory.OpenSession())
{
Icriteria criteria = session.CreateCriteria(typeof(T));
criteria.Add(Restrictions.Eq(propertyName, PropertyValue));
IList<T> matchingObjects = criteria.List<T>();
return matchingObjects;
}
}
I made changes in the code and start using StateLess Session but that change only reduces the frequency of timeout error.
After decugging , I found IList matchingObjects = criteria.List(); is cause of the exception. But this is only returning only one value, so it should not result timeout error since table also doesnt contain more than 100 rows as of now. Any Suggestions??
Unless you have wrapped NHibernate's ISessionFactory in something else, each call to OpenSession() will yield a new session. So the above code involves multiple sessions and it isn't clear if this is required.
Theoretically, a query on the session in RetrieveEquals() could block because of locks taken on the connection used in the calling method. But given the code as shown I can't see anything to prove this.
The calling method first updates a property of cashActivity, then in some cases goes on to delete the object. And there is no Commit(). This seems strange - is this really the used code or might there be a copy/paste error?
You also say "after two or three queries"... do you imply that there is a loop somewhere which isn't shown?

How to work around NHibernate caching?

I'm new to NHibernate and was assigned to a task where I have to change a value of an entity property and then compare if this new value (cached) is different from the actual value stored on the DB. However, every attempt to retrieve this value from the DB resulted in the cached value. As I said, I'm new to NHibernate, maybe this is something easy to do and obviously could be done with plain ADO.NET, but the client demands that we use NHibernate for every access to the DB. In order to make things clearer, those were my "successful" attempts (ie, no errors):
1
DetachedCriteria criteria = DetachedCriteria.For<User>()
.SetProjection(Projections.Distinct(Projections.Property(UserField.JobLoad)))
.Add(Expression.Eq(UserField.Id, userid));
return GetByDetachedCriteria(criteria)[0].Id; //this is the value I want
2
var JobLoadId = DetachedCriteria.For<User>()
.SetProjection(Projections.Distinct(Projections.Property(UserField.JobLoad)))
.Add(Expression.Eq(UserField.Id, userid));
ICriteria criteria = JobLoadId.GetExecutableCriteria(NHibernateSession);
var ids = criteria.List();
return ((JobLoad)ids[0]).Id;
Hope I made myself clear, sometimes is hard to explain a problem when even you don't quite understand the underlying framework.
Edit: Of course, this is a method body.
Edit 2: I found out that it doesn't work properly for the method call is inside a transaction context. If I remove the transaction, it works fine, but I need it to be in this context.
I do that opening a new stateless session for geting the actual object in the database:
User databaseuser;
using (IStatelessSession session = SessionFactory.OpenStatelessSession())
{
databaseuser = db.get<User>("id");
}
//do your checks
Within a session, NHibernate will return the same object from its Level-1 Cache (aka Identity Map). If you need to see the current value in the database, you can open a new session and load the object in that session.
I would do it like this:
public class MyObject : Entity
{
private readonly string myField;
public string MyProperty
{
get { return myField; }
set
{
if (value != myField)
{
myField = value;
DoWhateverYouNeedToDoWhenItIsChanged();
}
}
}
}
googles nhforge
http://nhibernate.info/doc/howto/various/finding-dirty-properties-in-nhibernate.html
This may be able to help you.

Refreshing an entity throws an NHibernate.UnresolvableObjectException

The entities and mappings I'm talking about in this question can be found here :)
Here is the context:
I have a parent view-model which helps to manage some entities, and which has its own session.
From this VM, I open another view-model (with its own session too), do some changements to the entity (add and/or remove children), and when I validate the changements, I commit the session and warns the first view-model to refresh the display:
public void Validate()
{
using (var tx = Session.BeginTransaction())
{
try
{
SelectedTeam.ClearRoster();
foreach (var teamPlayer in TeamPlayers)
SelectedTeam.AddPlayer(teamPlayer);
teamsRepository.SaveOrUpdate(SelectedTeam);
tx.Commit();
}
catch (Exception ex)
{
tx.Rollback();
}
finally
{
if (tx.WasCommitted)
ServiceLocator.Current.GetInstance<Mediator>().NotifyColleagues(MediatorMessages.DisplayEntityInfos, SelectedTeam.Id);
}
}
}
Here is the faulted method of the parent VM:
public void RefreshEntitiesListAndDisplayEntityInfos(int selectedEntityId)
{
TEntity entity = entitiesRepository.Load(selectedEntityId);
Session.Refresh(entity);
//...
}
The exception is thrown at the Refresh line:
NHibernate.UnresolvableObjectException
And the message is:
No row with the given identifier exists[Emidee.CommonEntities.PlayerInTeam#3
I can open and change the entity multiple times, but it seems that the exception is thrown when I delete a children, then add another one, and finally delete another one.
After some readings on the web, it seems that's because when I refresh the entity, and because I changed the HasMany relationship (because I have deleted a player for example), NH tries to reload the deleted row.
I've tried to add a NotFound.Ignore statement on the HasMany in my mappings, I've tried to force a new query to the DB instead of a Load, but I still get this exception.
Does someone know how I could fix that?
Thanks in advance
Mike
This is a known behavior when refreshing objects with modified collections.
To force reload, change your method to do session.Evict with the entity as a parameter. This is the code we use in our base model class:
public T ReGet<T>(T entity) where T : IEntity
{
var id = entity.Id;
Session.Evict(entity);
return Session.Get<T>(id);
}
Well, I've just spotted the problem.
To update the players list of the team, I used to clear the list, and add the new players, before updating the entity.
Now, I update the list by removing and adding only the players who have been moved by the user, and I don't have any problems at all now.
That's weird. I don't know what was wrong before, but as least that works now.