Castle ActiveRecord BelongsTo with NotNull=true not-null Exception - c#

I'm having the a problem when trying to persist a many to one relationship using Castle ActiveRecord and I hope someone has a better idea than me with this, the idea is to save a single object with a list of dependant objects in a single Save().
I have these classes:
[ActiveRecord("SomeClass")
public class SomeClass : ActiveRecordValidationBase<SomeClass>
{
[PrimaryKey]
public virtual long Id { get; set; }
[HasMany(Cascade = ManyRelationCascadeEnum.AllDeleteOrphan, Inverse = false)]
public virtual IList<AnotherClass> SomeObjects { get; set; }
}
[ActiveRecord("AnotherClass")
public class AnotherClass : ActiveRecordValidationBase<AnotherClass>
{
[PrimaryKey]
public virtual long Id { get; set; }
[Property(NotNull = true, Unique = true, Length = 70)]
public string Something { get; set; }
[BelongsTo("SomeId", NotNull = true)]
public virtual SomeClass Parent { get; set; }
}
What I want to do is something like this
var someClass = new SomeClass
{
SomeObjects = new List<AnotherClass>
{
new AnotherClass
{
Something = "Hello"
}
}
};
someClass.Save();
But I get this error:
Hibernate.PropertyValueException: not-null property references a null or transient value
Is there a way I could do that without setting a reference to the parent to every object before calling save?
Thanks!

you have to override BeforeSave and/or Save in your class SomeClass
public virtual void Save()
{
using(Transaction t = new Transaction())
{
foreach(AnotherClass a in this.SomeObjects??new AnotherClass[]{})
{
a.Parent = this;
a.Save();
}
base.Save();
}
}
Greetings
Juy Juka

Related

trying to pass generic id's to class constructor

I have below class LibrarySourceTableInput and having structure like as this
public class LibrarySourceTableInput<T> where T: ISourceOfData
{
public LibrarySourceTableInput(List<T> libraries, string mappedLibrarySource)
{
this.LibrarySourceRowInputs = libraries?.Select(l => new LibrarySourceRowInput()
{
LibrarySourceId = l.Id,
SourceOfDataId = l.SourceOfData.Id
}).ToList() ?? new(),
this.MappedLibrarySource = mappedLibrarySource;
}
public List<LibrarySourceRowInput> LibrarySourceRowInputs { get; set; }
public string MappedLibrarySource { get; set; }
}
and then the interface ISourceOfData structure looks like as this below
public interface ISourceOfData : IIdentity
{
public new Guid Id { get; set; }
public CodeStandardGuideline SourceOfData { get; set; }
}
And i am calling above class constructor in other place like as this
var mechanicalData = spaceTypeObject.TargetObject.MechanicalData;
var librarySourceTableInputs = new List<LibrarySourceTableInput<ISourceOfData>>
{
new LibrarySourceTableInput<ISourceOfData>(mechanicalData?.Environments, mappedLibrarySource), // I am getting conversion error here
new LibrarySourceTableInput<ISourceOfData>(mechanicalData?.AirflowsA621 , mappedLibrarySource)
.......
.......
}
and the structure for mechanicaldata.environments is looks like this
public class MechanicalData
{
public List<LibraryEnvironment> Environments { get; set; }
public List<LibraryA621> AirflowsA621 { get; set; }
.......
}
and then last one libraryEnvironment class is looks like this
public class LibraryEnvironment : ISourceOfData
{
public virtual CodeStandardGuideline SourceOfData { get; set; }
.....
......
public Guid Id {get; set;}
}
I have got the below error at this line new LibrarySourceTableInput<ISourceOfData>(mechanicalData?.Environments, mappedLibrarySource)
and the error is Cannot convert from Generic.List<LibraryEnvironment> to Generic.List<IsourceOfData>
Could any one please let me know or any idea on how to send that generic list to that class constructor, Many thanks in advance.
Instead of specifying the interface ISourceOfData (your class already knows the type will have that interface), specify the actual class:
var librarySourceTableInputs = new List<LibrarySourceTableInput<LibraryEnvironment>>
{
new LibrarySourceTableInput<LibraryEnvironment>(mechanicalData?.Environments, mappedLibrarySource),
.......
.......
}

Opinions on mapping fields of nested/hierarchical objects to flat list?

I'm writing a tool which accesses a word document to prefill it with data. The document has a subset of custom document properties, each identified by a name, whose values are used to update fields in the document.
My ViewModel should both be able to initiate/update its instances from data of those document properties, aswell as write its values back and update the fields of the document.
Something like this:
class PersonVM : INotifyPropertyChanged
{
// properties
string Name { get; set; }
string PhoneNumber { get; set; }
// methods to get data or save data of this properties to or from the word document
void saveMyPropertyValuesToWord()
{
// …
}
void updateMyPropertiesFromWord()
{
// …
}
}
class ProjectVM : INotifyPropertyChanged
{
int ProjectNumber { get; set; }
PersonVM Manager { get; set; }
PersonVM Mechanic1 { get; set; }
PersonVM Mechanic2 { get; set; }
void saveMyPropertyValuesToWord()
{
Manager.saveMyPropertyValuesToWord();
Mechanic1.saveMyPropertyValuesToWord();
Mechanic2.saveMyPropertyValuesToWord();
// handle ProjectNumber etc.
}
void updateMyPropertiesFromWord()
{
Manager.updateMyPropertiesFromWord();
Mechanic1.updateMyPropertiesFromWord();
Mechanic2.updateMyPropertiesFromWord();
// handle ProjectNumber etc.
}
class CompanyVM : INotifyPropertyChanged
{
string Name { get; set; }
PersonVM Owner { get; set; }
ProjectVM Project1 { get; set; }
ProjectVM Project2 { get; set; }
// …
}
// …
}
Right now I have a class with static string properties for each document property that might be present in a word document from which I would like to load the data accordingly:
class WordUtils
{
// Company
static string CompanyName = "dp_CompanyName";
// Company.Owner
static string CompanyOwnerName = "dp_CompanyOwnerName";
static string CompanyOwnerPhone = "dp_CompanyOwnerPhone";
// Company.Project1
static string CompanyProject1Number = "dp_CompanyProject1Number";
// Company.Project1.Manager
static string CompanyProject1ManagerName = "dp_CompanyProject1ManagerName";
static string CompanyProject1ManagerPhone = "dp_CompanyProject1ManagerPhone";
// Company.Project1.Mechanic1
// … etc
}
Now back to implementing those PersonVM.saveMyPropertyValuesToWord() - I thought of something like this:
void saveMyPropertyValuesToWord()
{
Name = MyApp.MyWordDocument.GetCustomProperty(WordUtils.OwnerName);
}
but here I need to know on class Level exactly what instance of it this is called from (i.e. what PersonVM am I, Company.Owner or Project1.Manager or ?) in order to decide which WordUtils.Name I need to provide.
I'm not sure how this should be done, maybe make PersonVM abstract and make a new class for each role (which would again only have one instance of itself, not very pretty in my eyes)? I have also taken a short look at Attributes and expect those might be helpfull in this scenario. Maybe I am missing something obvious, but extensive search for a robust way to tackle this problem have been fruitless so far.
How about something like this:
class Property
{
public string Key { get; }
public string Value { get; set; }
public Property(string key) => Key = key;
}
interface IPropertyTree
{
IEnumerable<IPropertyTree> ChildNodes { get; }
IEnumerable<Property> Properties { get; }
}
class PersonVM : IPropertyTree
{
private readonly string prefix;
public PersonVM(string prefix)
{
Name = new Property(prefix + "Name" );
PhoneNumber = new Property(prefix + "PhoneNumber");
}
public Property Name { get; }
public Property PhoneNumber { get; }
public IEnumerable<IPropertyTree> ChildNodes => Enumerable.Empty<IPropertyTree>();
public IEnumerable<Property> Properties => new[] {Name, PhoneNumber};
}
static class PropertyTreeExtensions
{
public static void Update(this IPropertyTree self)
{
foreach (var property in self.Flatten().SelectMany(tree => tree.Properties))
{
property.Value = MyApp.MyWordDocument.GetCustomProperty(property.Key);
}
}
public static IEnumerable<IPropertyTree> Flatten(this IPropertyTree self)
{
var stack = new Stack<IPropertyTree>();
stack.Push(self);
while (stack.Count > 0)
{
var current = stack.Pop();
yield return current;
foreach (var child in current.ChildNodes)
{
stack.Push(child);
}
}
}
}
This should allow each property to have a unique key, and keep the key and property value tightly coupled. It should also allow you to move the save/update logic to a centralized place.
Of course you can implement a concrete class of IPerson for each type and hard code the individual implementations.
Since you know the person type the moment you are creating an instance of PersonVMM, you could add an attribute PersonTypeId and set it from the constructor,
void SomeMethod()
{
var personVm = new PersonVM(WordUtils.OwnerName);
}
class PersonVM : INotifyPropertyChanged
{
// properties
string PersonTypeId { get; set; }
string Name { get; set; }
string PhoneNumber { get; set; }
public PersonVM()
{}
public PersonVM(string personTypeId)
{
PersonTypeId = personTypeId;
}
// methods to get data or save data of this properties to or from the word document
void saveMyPropertyValuesToWord()
{
Name = MyApp.MyWordDocument.GetCustomProperty(PersonTypeId);
}
}

AutoMapper isn't recognizing profile-specific prefixes

I'm trying to use AutoMapper to take data from a class that has prefixes before property names and map it to a second class that doesn't have those prefixes. However, I don't necessarily want it to always strip out that prefix: I just want it to do it for this particular mapping.
My source class looks like this:
public class AdvancedSearchFilterDataModel
{
// ....
public string ServiceMeterNumber { get; set; }
// ....
}
My destination class looks like this:
[DataContract]
public class ServicesAdvancedSearchFilterData : AdvancedSearchFilterData
{
// ....
[DataMember]
public string MeterNumber { get; set; }
// ....
}
When I try to map values like this, it works:
Mapper.Configuration.RecognizePrefixes("Service");
Mapper.CreateMap<AdvancedSearchFilterDataModel, ServicesAdvancedSearchFilterData>();
ServicesAdvancedSearchFilterData servciesFilterData =
Mapper.Map<ServicesAdvancedSearchFilterData>(model);
But I only want "Service" to be recognized as a prefix for certain mappings, since it's also used as a normal part of property names in other mappings. I tried to handle this with a profile, but this didn't work -- no data was mapped:
Mapper.CreateProfile("ServicePrefix").RecognizePrefixes("Service");
Mapper.CreateMap<AdvancedSearchFilterDataModel, ServicesAdvancedSearchFilterData>()
.WithProfile("ServicePrefix");
ServicesAdvancedSearchFilterData servciesFilterData =
Mapper.Map<ServicesAdvancedSearchFilterData>(model);
How can I make it recognize the prefix only when I want it to, either using profiles or some other technique? (I also have other prefixes that I'm going to need it to recognize for other mappings in the same way.)
I achieved this functionality by creating following structure:
I have Person model for my view which is flattened from PersonCombined
public class PersonCombined
{
public Person Person { get; set; }
public Address DefaultAddress { get; set; }
public Contact EmailContact { get; set; }
public Contact PhoneContact { get; set; }
public Contact WebsiteContact { get; set; }
}
public class Person : IWebServiceModel
{
public int ID { get; set; }
public string PersonFirstName { get; set; }
public string PersonSurname { get; set; }
public string PersonDescription { get; set; }
public Nullable<bool> PersonIsActive { get; set; }
}
Then I have separate class for this mapping only that looks like this:
public class PersonCustomMapping : ICustomMapping
{
const string separator = " ";
private static IMappingEngine _MappingEngine;
public IMappingEngine MappingEngine
{
get
{
if (_MappingEngine == null)
{
var configuration = new ConfigurationStore(new TypeMapFactory(), AutoMapper.Mappers.MapperRegistry.Mappers);
configuration.RecognizePrefixes("Person");
configuration.RecognizeDestinationPrefixes("Person");
configuration.CreateMap<Person, MCIACRM.Model.Combine.PersonCombined>();
configuration.CreateMap<MCIACRM.Model.Combine.PersonCombined, Person>();
_MappingEngine = new MappingEngine(configuration);
}
return _MappingEngine;
}
}
}
In my generic view I have mappingEngine property like this:
private IMappingEngine mappingEngine
{
get
{
if (_mappingEngine == null)
{
_mappingEngine = AutoMapper.Mapper.Engine;
}
return _mappingEngine;
}
}
Finally in my generic view constructor i have:
public GenericEntityController(IGenericLogic<S> logic, ICustomMapping customMapping)
: base()
{
this._mappingEngine = customMapping.MappingEngine;
this.logic = logic;
}
And that's how I do mapping:
result = items.Project(mappingEngine).To<R>();
or
logic.Update(mappingEngine.Map<S>(wsItem));
Because I use 1 entity per view I can define custom mapping configuration per entity.
Hope this helps

EF adding duplicate records into lookup/reference table

I have 3 tables,
1. AttributeTypes (Columns: AttributeId (PK), AttributeName, ..)
2. Location (Columns: locationId (PK), LocationName, ...)
3. LocationAttributeType (Columns: locationId (FK), AttributeId (FK))
Whenever I am trying to insert new location record along with its attribute type from GUI, it should create new record for Table- Location and LocationAttributeType. But EF trying to add new record in Table- AttributeTypes as well, which is just used as reference table and should not add new/duplicate records in it. How can I prevent that?
here is my code,
The model which GUI sends is,
public class LocationDataModel
{
[DataMember]
public int Id { get; set; }
[DataMember]
public string Code { get; set; }
[DataMember]
public List<AttributeTypeDataModel> AssignedAttributes = new List<AttributeTypeDataModel>();
}
public class AttributeTypeDataModel
{
protected AttributeTypeDataModel() {}
public AttributeTypeDataModel(int id) { this.Id = id; }
public AttributeTypeDataModel(int id, string name)
: this(id)
{
this.Name = name;
}
[DataMember]
public int Id { get; set; }
[DataMember]
public string Name { get; set; }
[DataMember]
public virtual ICollection<LocationDataModel> Locations { get; set; }
}
The Entities created by EF are,
public partial class Location
{
public Location()
{
this.AttributeTypes = new List<AttributeType>();
}
public Location(int campusId, string code)
: this()
{
CampusId = campusId; Code = code;
}
public int Id { get; set; }
public int CampusId { get; set; }
public string Code { get; set; }
public virtual ICollection<AttributeType> AttributeTypes { get; set; }
}
public partial class AttributeType
{
public AttributeType()
{
this.Locations = new List<Location>();
}
public int AttributeTypeId { get; set; }
public string AttributeTypeName { get; set; }
public virtual ICollection<Location> Locations { get; set; }
}
I have below code to Add these new location to database,
private IEnumerable<TEntity> AddEntities<TModel, TEntity, TIdentityType>
(IEnumerable<TModel> models, Func<TModel, TIdentityType> primaryKey,
IGenericRepository<TEntity, TIdentityType> repository)
{
var results = new List<TEntity>();
foreach (var model in models)
{
var merged = _mapper.Map<TModel, TEntity>(model);
var entity = repository.Upsert(merged);
results.Add(entity);
}
repository.Save();
return results.AsEnumerable();
}
I am using following generic repository to do entity related operations
public TEntity Upsert(TEntity entity)
{
if (Equals(PrimaryKey.Invoke(entity), default(TId)))
{
// New entity
return Context.Set<TEntity>().Add(entity);
}
else
{
// Existing entity
Context.Entry(entity).State = EntityState.Modified;
return entity;
}
}
public void Save()
{
Context.SaveChanges();
}
Whats wrong I am doing here?
The DbSet<T>.Add() method attaches an entire object graph as added. You need to indicate to EF that the 'reference' entity is actually already present. There are two easy ways to do this:
Don't set the navigation property to an object. Instead, just set the corresponding foreign key property to the right value.
You need to ensure that you don't load multiple instances of the same entity into your object context. After creating the context, load the full list of AttributeType entities into the context and create a Dictionary<> to store them. When you want to add an attribute to a Location retrieve the appropriate attribute from the dictionary. Before calling SaveChanges() iterate through the dictionary and mark each AttributeType as unchanged. Something like this:
using (MyContext c = new MyContext())
{
c.AttributeTypes.Add(new AttributeType { AttributeTypeName = "Fish", AttributeTypeId = 1 });
c.AttributeTypes.Add(new AttributeType { AttributeTypeName = "Face", AttributeTypeId = 2 });
c.SaveChanges();
}
using (MyContext c = new MyContext())
{
Dictionary<int, AttributeType> dictionary = new Dictionary<int, AttributeType>();
foreach (var t in c.AttributeTypes)
{
dictionary[t.AttributeTypeId] = t;
}
Location l1 = new Location(1, "Location1") { AttributeTypes = { dictionary[1], dictionary[2] } };
Location l2 = new Location(2, "Location2") { AttributeTypes = { dictionary[1] } };
// Because the LocationType is already attached to the context, it doesn't get re-added.
c.Locations.Add(l1);
c.Locations.Add(l2);
c.SaveChanges();
}
In this specific case you are using a many-to-many relationship, with EF automatically handling the intermediate table. This means that you don't actually have the FK properties exposed in the model, and my first suggestion above won't work.
Therefore, you either need to use the second suggestion, which still ought to work, or you need to forgo the automatic handling of the intermediate table and instead create an entity for it. This would allow you to apply the first suggestion. You would have the following model:
public partial class Location
{
public Location()
{
this.AttributeTypes = new List<LocationAttribute>();
}
public Location(int campusId, string code)
: this()
{
CampusId = campusId; Code = code;
}
public int Id { get; set; }
public int CampusId { get; set; }
public string Code { get; set; }
public virtual ICollection<LocationAttribute> AttributeTypes { get; set; }
}
public partial class LocationAttribute
{
[ForeignKey("LocationId")]
public Location Location { get; set; }
public int LocationId { get; set; }
public int AttributeTypeId { get; set; }
}
public partial class AttributeType
{
public int AttributeTypeId { get; set; }
public string AttributeTypeName { get; set; }
}
With this approach you do lose functionality since you can't navigate from a Location to an AttributeType without making an intermediate lookup. If you really want to do that, you need to control the entity state explicitly instead. (Doing that is not so straightforward when you want to use a generic repository, which is why I've focused on this approach instead.)
Thank you all for your suggestions.
I have to get rid of my generic repository here to save my context changes and do it manually as below,
private IEnumerable<int> AddLocationEntities(IEnumerable<LocationDataModel> locations)
{
var results = new List<int>();
foreach (LocationDataModel l in locations)
{
var entity = _mapper.Map<LocationDataModel, Location>(l);//you can map manually also
var AttributeCode = l.AssignedAttributes.FirstOrDefault().AttributeTypeId;
using (MyContext c = new MyContext())
{
var attr = c.AttributeTypes.Where(a => a.Id == AttributeTypeId ).ToList();
entity.AttributeTypes = attr;
c.Locations.Add(entity);
c.SaveChanges();
var locid = entity.Id;
results.Add(locid);
}
}
return results;
}
In the else statement of yourUpsert you should add
context.TEntity.Attach(entity);

Make generic object instantiation more generic

I've got this piece of code to create new objects in a generic way:
var user = User.Create<User>(c => c.Name = "321X");
What I don't like about it is the fact I need to pass the 'generic notation' <T> for every create call. After all I create an object that I'm already referring to...
The code behind this current functionality is:
public class User : CreateBase
{
public string Name { get; set; }
}
public abstract class CreateBase
{
public DateTime CreateDate { get; set; }
public Guid Guid { get; set; }
public static T Create<T>(Action<T> init) where T : CreateBase, new()
{
T obj = new T();
obj.Guid = Guid.NewGuid();
obj.DateTime = DateTime.Now;
init(obj);
return obj;
}
}
Is it possible (and how) to refactor my code to this, to create an object?
var user = User.Create(c => c.Name = "321X");
Thanks!
Define the generic argument on the class level:
public abstract class CreateBase<T> where T : CreateBase<T> , new()
{
public static T Create(Action<T> init)
{
//...
}
}
public class User : CreateBase<User>
{
public string Name { get; set; }
}
Then you can write var user = User.Create(c => c.Name = "321X");
Otherwise the compiler cannot infer the type for your Create method without specifying the type argument.
You were not very far. Try this modification:
public abstract class CreateBase<T> where T : CreateBase<T> , new()
{
public DateTime CreateDate { get; set; }
public Guid Guid { get; set; }
public static T Create(Action<T> init)
{
T obj = new T();
obj.Guid = Guid.NewGuid();
obj.CreateDate = DateTime.Now;
init(obj);
return obj;
}
}
public class User : CreateBase<User>
{
public string Name { get; set; }
}
EDIT: Updated the code after I tested it on my local environment. It works now.
You are doing it the wrong way. Instead of getting rid of the generic argument, get rid of (needlessly) saying User.. Instead:
CreateBase.Create<User>(...)
No more redundancies.
Besides that, calling a static member of the base class through a derived class is an anti-pattern.
A better approach would be to include this functionality in the constructor of the base class (I call it ModelBase)
public abstract class ModelBase
{
public DateTime CreateDate { get; private set; }
public Guid Guid { get; private set; }
public ModelBase()
{
Guid = Guid.NewGuid();
DateTime = DateTime.Now;
}
}
public User : ModelBase
{
public User()
: base()
{
}
public User(string name)
: base()
{
Name = name
}
public string Name { get; set; }
}
Creating a user the standard way will initialize the Guid and date automatically
var user = new User { Name = "xy };
EDIT
I added a second constructor with a name parameter. I you want to force the initialization of the name, drop the first parameterless constructor.
var user = new User("xy");
If you really uncomfortable with that sintax (I, honestly, don't see much problem here)
you can do the following:
public class User : CreateBase
{
public string Name { get; set; }
public static User Create(Action<User> a)
{
return Create<User>(a); //CALL BASE CLASS GENERIC FUNCTION
}
}
After you can call it in a way you would like to do that :
var user = User.Create(c => c.Name = "321X");

Categories

Resources