How to lazy load a property with custom attribute - c#

I want to create a custom attribute using Glass Mapper for getting the Sitecore URL, because it is not possible to lazy load a property with SitecoreInfo(SitecoreInfoType.Url) and we have some performance issues loading URL of mapped items, where the URL will never be used.
Here is what I've got so far:
The Configuration
public class SitecoreUrlConfiguration : AbstractPropertyConfiguration
{
public SitecoreInfoUrlOptions UrlOptions { get; set; }
public bool IsLazy { get; set; }
}
The Attribute
public class SitecoreUrlAttribute : AbstractPropertyAttribute
{
public SitecoreUrlAttribute()
{
this.IsLazy = true;
this.UrlOptions = SitecoreInfoUrlOptions.Default;
}
/// <summary>
/// Gets or sets a value indicating whether is lazy.
/// </summary>
public bool IsLazy { get; set; }
public SitecoreInfoUrlOptions UrlOptions { get; set; }
public override AbstractPropertyConfiguration Configure(PropertyInfo propertyInfo)
{
var config = new SitecoreUrlConfiguration();
this.Configure(propertyInfo, config);
return config;
}
public void Configure(PropertyInfo propertyInfo, SitecoreUrlConfiguration config)
{
config.UrlOptions = this.UrlOptions;
config.IsLazy = this.IsLazy;
base.Configure(propertyInfo, config);
}
}
The Mapper
public class SitecoreUrlMapper : AbstractDataMapper
{
public override object MapToProperty(AbstractDataMappingContext mappingContext)
{
var context = mappingContext as SitecoreDataMappingContext;
if (context == null)
{
throw new MapperException("Mapping Context is null");
}
var item = context.Item;
var scConfig = this.Configuration as SitecoreUrlConfiguration;
if (scConfig == null)
{
throw new MapperException("SitecoreUrlConfiguration is null");
}
var urlOptions = Utilities.CreateUrlOptions(scConfig.UrlOptions);
urlOptions.Language = null;
// now, what?
}
}
So far, so good. But how can I lazy load the URL in the mapper? Does anyone have an idea?

The only way I actually see is to map a Lazy<T> and add a new property to the class which returns the value of this when accessing it. So in you mapper, where you put // now what? I would return the lazy string:
return new Lazy<string>(() => LinkManager.GetItemUrl(item, urlOptions));
Then in your model, put these two properties:
[SitecoreUrl]
public Lazy<string> LazyUrl { private get; set; }
[SitecoreIgnore]
public virtual string Url
{
get
{
return this.LazyUrl.Value;
}
}

You can achieve pretty similar to this with a bit of creativity and the new Delegate functionality
In the fluent configuration map the type like so:
SitecoreType<IWhatever> sitecoreType = new SitecoreType<IWhatever>();
sitecoreType.Delegate(y => y.Url).GetValue(GetLazyUrl);
private LazyString GetLazyUrl(SitecoreDataMappingContext arg)
{
var item = context.Item;
return new LazyString(
() =>
{
// the necessary actions to get the url
});
}
public class LazyString : Lazy<string>
{
public LazyString(Func<string> valueFactory) : base(valueFactory)
{
}
public override string ToString()
{
return Value;
}
public static implicit operator string(LazyString lazyString)
{
return lazyString.Value;
}
}
It's not a string, but for the purposes of many applications, will behave like one.

Related

Dbset as hashset in the Entity Framework

I need to merge data from tables in the database with data based on some logic from third-party sources. I implemented this logic via hashset, for which I overloaded the GetHashCode and Equals methods for entities. Now I don't understand how I can save the result of work in the database via DbSet, with subsequent data loading and subsequent merging (the task of merging/supplementing is periodic)
The directories are quite voluminous, so working through hashset speeds up the process.
class Program
{
private class DummyDbContext { public void SaveChangesAsync() { }}
static void Main(string[] args)
{
var dbContext = new DummyDbContext(); // TODO: Get from DI
// TODO: I don't know how to do it yet with HashSets
var currentFactories = LoadCurrentFactoriesFromDb(dbContext);
var currentProducts = LoadCurrentProductsFromDb(dbContext);
var thirdPartyData = GetThirdPartyData();
foreach (var data in thirdPartyData)
{
/*
In reality, the logic is more complicated, because some data transformation is required.
Some data may be missing. That is why comparing two objects is not quite easy (see the method Product.Equals)
*/
var factory = new Factory(data.otherFactory.Name);
var product = new Product(data.otherProduct.Property1, data.otherProduct.Property2, factory);
if (currentFactories.TryGetValue(factory, out var existedFactory))
factory = existedFactory;
else
currentFactories.Add(factory);
if (currentProducts.TryGetValue(product, out var existedProduct))
{
if (!existedProduct.Factory.Equals(factory))
throw new InvalidOperationException(); // TODO:
product = existedProduct;
factory.Products.Add(product); // TODO:
}
else
currentProducts.Add(product);
}
// **how to implement the saving of combined directories, in hashsets, in the database ?**
dbContext.SaveChangesAsync();
}
private static IEnumerable<(ThirdPartyFactory otherFactory, ThirdPartyProduct otherProduct)> GetThirdPartyData()
{
return new (ThirdPartyFactory otherFactory, ThirdPartyProduct otherProduct)[]
{
( new ThirdPartyFactory () {Name = "SomeFactory"}, new ThirdPartyProduct() {Property1 = "ProductName1"}),
( new ThirdPartyFactory () {Name = "SomeFactory"}, new ThirdPartyProduct() {Property1 = "ProductName2"}),
( new ThirdPartyFactory () {Name = "SomeFactory"}, new ThirdPartyProduct() {Property2 = "Property1"})
};
}
private static HashSet<Factory> LoadCurrentFactoriesFromDb(DummyDbContext context)
{
// DbContext.DbSet<Factory>.GetAll()
return new HashSet<Factory>();
}
private static HashSet<Product> LoadCurrentProductsFromDb(DummyDbContext context)
{
// DbContext.DbSet<Product>.GetAll()
return new HashSet<Product>();
}
}
public class Product
{
public Product(string property1, string property2, Factory factory)
{
Property1 = property1;
Property2 = property2;
Factory = factory;
}
public long Id { get; set; }
public string Property1 { get; }
public string Property2 { get; }
public Factory Factory { get; }
public override bool Equals(object? obj)
{
if (obj == null)
return false;
var product = (Product) obj;
return (string.IsNullOrWhiteSpace(Property1) && string.IsNullOrWhiteSpace(product.Property1)
|| string.CompareOrdinal(this.Property1, product.Property1) == 0)
&& (string.IsNullOrWhiteSpace(Property2) && string.IsNullOrWhiteSpace(product.Property2)
|| string.CompareOrdinal(this.Property2, product.Property2) == 0);
}
public override int GetHashCode()
{
return HashCode.Combine(Property1, Property2).GetHashCode();
}
}
public class Factory
{
public Factory(string name)
{
Name = name;
}
public long Id { get; set; }
public string Name { get; }
public HashSet<Product> Products { get; set; }
}
public class ThirdPartyProduct
{
public string Property1 { get; set; }
public string Property2 { get; set; }
}
public class ThirdPartyFactory
{
public string Name { get; set; }
}
Is it possible to implement this ? Or do I need to convert data from DbSet to HashSet and then back ? But won't I lose information about entities inside the context during such transformations ?

Automapper - Make IncludeMembers() ignore null

IncludeMembers() will always map from first match even if object is null.
Let's say we have the following source models:
public class Item
{
public MovieMetadata MovieMetadata { get; set; }
public BookMetadata BookMetadata { get; set; }
}
public class MovieMetadata
{
public string Title { get; set; }
}
public class BookMetadata
{
public string Title { get; set; }
}
Destination model:
public class ItemDetail
{
public string Title { get; set; }
}
Mapping profile:
public class ItemProfile : Profile
{
public ItemProfile()
{
CreateMap<Item, ItemDetail>()
.IncludeMembers(
src => src.BookMetadata,
src => src.MovieMetadata);
CreateMap<BookMetadata, ItemDetail>();
CreateMap<MovieMetadata, ItemDetail>();
}
}
And Program class with instantiation and testing logic:
public class Program
{
public static void Main()
{
//create mapper
var configuration = new MapperConfiguration(cfg =>
{
cfg.AddMaps(typeof(ItemProfile));
});
var mapper = configuration.CreateMapper();
//check if configuration valid
mapper.ConfigurationProvider.AssertConfigurationIsValid();
Console.WriteLine("Mapper configuration is valid");
//try map book metadata
var bookItem = new Item
{
BookMetadata = new BookMetadata()
{
Title = "book"
},
MovieMetadata = null
};
var bookItemDetail = mapper.Map<ItemDetail>(bookItem);
bool isBookCorrectlyMapped = bookItem.BookMetadata.Title == bookItemDetail.Title;
Console.WriteLine($"Book mapped correctly: {isBookCorrectlyMapped}");
//try map movie metadata
var movieItem = new Item
{
BookMetadata = null,
MovieMetadata = new MovieMetadata()
{
Title = "movie"
}
};
var movieItemDetail = mapper.Map<ItemDetail>(movieItem);
bool isMovieCorrectlyMapped = movieItem.MovieMetadata.Title == movieItemDetail.Title;
Console.WriteLine($"Movie mapped correctly: {isMovieCorrectlyMapped}");
}
}
This is the output we will see:
Mapper configuration is valid
Book mapped correctly: True
Movie mapped correctly: False
We see that mapping for item with BookMetadata succeeded, but for MovieMetadata failed. Need to make updates so this test succeeds.
Suppose the reason it's failing is because of order of items in IncludeMembers():
.IncludeMembers(
src => src.BookMetadata,
src => src.MovieMetadata)
Here we have src.BookMetadata first and src.MovieMetadata second. During mapping it will find Title field in src.BookMetadata and will always use this value even if src.BookMetadata is null.
Is there any way to skip src.BookMetadata if it is null and use next src.MovieMetadata instead? Or probably need to use something else instead of IncludeMembers()?
Here is similar issue: https://github.com/AutoMapper/AutoMapper/issues/3204
Code above you can find here to quickly reproduce issue: https://dotnetfiddle.net/9YLGR6
Found possible solutions.
Option 1. IValueResolver
We can create custom value resolver. This solution is good for small amount of fields that we need to map from child objects. As any field will require specific value resolver.
Let's create TitleResolver:
public class TitleResolver : IValueResolver<Item, ItemDetail, string>
{
public string Resolve(Item source, ItemDetail destination, string destMember, ResolutionContext context)
{
if (source != null)
{
if (source.BookMetadata != null)
{
return source.BookMetadata.Title;
}
if (source.MovieMetadata != null)
{
return source.MovieMetadata.Title;
}
}
return null;
}
}
And update ItemProfile:
public class ItemProfile : Profile
{
public ItemProfile()
{
CreateMap<Item, ItemDetail>()
.ForMember(dest => dest.Title, opt => opt.MapFrom<TitleResolver>());
}
}
Link to whole code sample: https://dotnetfiddle.net/6pfKYh
Option 2. BeforeMap()/AfterMap()
In case if our child objects have more than one field to be mapped to destination object it may be a good idea to use BeforeMap() or AfterMap() methods.
In that case ItemProfile will be updated to:
public class ItemProfile : Profile
{
public ItemProfile()
{
CreateMap<Item, ItemDetail>(MemberList.None)
.AfterMap((src, dest, ctx) =>
{
if (src.BookMetadata != null)
{
ctx.Mapper.Map(src.BookMetadata, dest);
}
else if (src.MovieMetadata != null)
{
ctx.Mapper.Map(src.MovieMetadata, dest);
}
});
CreateMap<BookMetadata, ItemDetail>();
CreateMap<MovieMetadata, ItemDetail>();
}
}
Link to whole code sample: https://dotnetfiddle.net/ny1yRU

ASP.NET Core: Complex Model with comma separated values list

Our request models are growing according to the growing complexity of our APIs and we decided to use complex types instead of using simple types for the parameters of the actions.
One typical type is IEnumerable for comma-separated values, like items=1,2,3,5... and we solved the issue of converting from string to IEnumerable using the workaround provided in https://www.strathweb.com/2017/07/customizing-query-string-parameter-binding-in-asp-net-core-mvc/ where the key point is implementing the IActionModelConvention interface to identify the parameters marked with a specific attribute [CommaSeparated].
Everything worked fine until we moved the simple parameters into a single complex parameter, now we are unable to inspect the complex parameters in the IActionModelConvention implementation. The same happens using IParameterModelConvention. Please, see the code below:
this works fine:
public async Task<IActionResult> GetByIds(
[FromRoute]int day,
[BindRequired][FromQuery][CommaSeparated]IEnumerable<int> ids,
[FromQuery]string order)
{
// do something
}
while this variant does not work
public class GetByIdsRequest
{
[FromRoute(Name = "day")]
public int Day { get; set; }
[BindRequired]
[FromQuery(Name = "ids")]
[CommaSeparated]
public IEnumerable<int> Ids { get; set; }
[FromQuery(Name = "order")]
public string Order { get; set; }
}
public async Task<IActionResult> GetByIds(GetByIdsRequest request)
{
// do something
}
the IActionModelConvention implementation is very simple:
public void Apply(ActionModel action)
{
SeparatedQueryStringAttribute attribute = null;
for (int i = 0; i < action.Parameters.Count; i++)
{
var parameter = action.Parameters[i];
var commaSeparatedAttr = parameter.Attributes.OfType<CommaSeparatedAttribute>().FirstOrDefault();
if (commaSeparatedAttr != null)
{
if (attribute == null)
{
attribute = new SeparatedQueryStringAttribute(",", commaSeparatedAttr.RemoveDuplicatedValues);
parameter.Action.Filters.Add(attribute);
}
attribute.AddKey(parameter.ParameterName);
}
}
}
As you can see, the code is inspecting the parameters marked with CommaSeparatedAttribute...but it doesn't work with complex parameters like the one used in my second variant.
Note: I added some minor changes to the original code provided in the post above mentioned like enabling the CommaSeparatedAttribute to be used not only for parameters but also for properties, but still it doesn't work
Based on itminus's answer I could work out my final solution. The trick was - as itminus pointed out - in the IActionModelConvention implementation. Please, see my implementation which considers other aspects like nested models and also the real name assigned to each property:
public void Apply(ActionModel action)
{
SeparatedQueryStringAttribute attribute = null;
for (int i = 0; i < action.Parameters.Count; i++)
{
var parameter = action.Parameters[i];
var commaSeparatedAttr = parameter.Attributes.OfType<CommaSeparatedAttribute>().FirstOrDefault();
if (commaSeparatedAttr != null)
{
if (attribute == null)
{
attribute = new SeparatedQueryStringAttribute(",", commaSeparatedAttr.RemoveDuplicatedValues);
parameter.Action.Filters.Add(attribute);
}
attribute.AddKey(parameter.ParameterName);
}
else
{
// here the trick to evaluate nested models
var props = parameter.ParameterInfo.ParameterType.GetProperties();
if (props.Length > 0)
{
// start the recursive call
EvaluateProperties(parameter, attribute, props);
}
}
}
}
the EvaluateProperties method:
private void EvaluateProperties(ParameterModel parameter, SeparatedQueryStringAttribute attribute, PropertyInfo[] properties)
{
for (int i = 0; i < properties.Length; i++)
{
var prop = properties[i];
var commaSeparatedAttr = prop.GetCustomAttributes(true).OfType<CommaSeparatedAttribute>().FirstOrDefault();
if (commaSeparatedAttr != null)
{
if (attribute == null)
{
attribute = new SeparatedQueryStringAttribute(",", commaSeparatedAttr.RemoveDuplicatedValues);
parameter.Action.Filters.Add(attribute);
}
// get the binding attribute that implements the model name provider
var nameProvider = prop.GetCustomAttributes(true).OfType<IModelNameProvider>().FirstOrDefault(a => !IsNullOrWhiteSpace(a.Name));
attribute.AddKey(nameProvider?.Name ?? prop.Name);
}
else
{
// nested properties
var props = prop.PropertyType.GetProperties();
if (props.Length > 0)
{
EvaluateProperties(parameter, attribute, props);
}
}
}
}
I also changed the definition of the comma separated attribute
[AttributeUsage(AttributeTargets.Property | AttributeTargets.Parameter, Inherited = true, AllowMultiple = false)]
public class CommaSeparatedAttribute : Attribute
{
public CommaSeparatedAttribute()
: this(true)
{ }
/// <summary>
/// ctor
/// </summary>
/// <param name="removeDuplicatedValues">remove duplicated values</param>
public CommaSeparatedAttribute(bool removeDuplicatedValues)
{
RemoveDuplicatedValues = removeDuplicatedValues;
}
/// <summary>
/// remove duplicated values???
/// </summary>
public bool RemoveDuplicatedValues { get; set; }
}
There are other moving parts I changed too...but this is basically the most important ones. Now, we can use models like this:
public class GetByIdsRequest
{
[FromRoute(Name = "day")]
public int Day { get; set; }
[BindRequired]
[FromQuery(Name = "ids")]
[CommaSeparated]
public IEnumerable<int> Ids { get; set; }
[FromQuery(Name = "include")]
[CommaSeparated]
public IEnumerable<IncludingOption> Include { get; set; }
[FromQuery(Name = "order")]
public string Order { get; set; }
[BindProperty(Name = "")]
public NestedModel NestedModel { get; set; }
}
public class NestedModel
{
[FromQuery(Name = "extra-include")]
[CommaSeparated]
public IEnumerable<IncludingOption> ExtraInclude { get; set; }
[FromQuery(Name = "extra-ids")]
[CommaSeparated]
public IEnumerable<long> ExtraIds { get; set; }
}
// the controller's action
public async Task<IActionResult> GetByIds(GetByIdsRequest request)
{
// do something
}
For a request like this one (not exactly the same as the one defined above but very similar):
http://.../vessels/algo/days/20190101/20190202/hours/1/2?page=2&size=12&filter=eq(a,b)&order=by(asc(a))&include=all,none&ids=12,34,45&extra-include=all,none&extra-ids=12,34,45
If anyone needs the full code, please, let me know. Again, thanks to itminus for his valuable help
The reason
That's because you're trying to detect the existence of the [CommaSeparated] attributes which are decorated on parameter (instead of the parameter's properties):
var commaSeparatedAttr = parameter.Attributes.OfType().FirstOrDefault();
Note your action method looks like the following :
public async Task GetByIds(GetByIdsRequest request)
In other words, the parameter.Attributes.OfType<CommaSeparatedAttribute>() will only get those annotations decorated on the request parameter. However, there's no such a [CommaSeparatedAttribute] at all.
As a result, the SeparatedQueryStringAttribute filter is never added to parameter.Action.Filters.
How to Fix
Looks like you've made a minor tweek in the SeparatedQueryStringAttribute. As we don't get your code, suppose we have such a SeparatedQueryStringAttribute Filter (copied from the blog you mentioned above):
public class SeparatedQueryStringAttribute : Attribute, IResourceFilter
{
private readonly SeparatedQueryStringValueProviderFactory _factory;
public SeparatedQueryStringAttribute() : this(",") { }
public SeparatedQueryStringAttribute(string separator) {
_factory = new SeparatedQueryStringValueProviderFactory(separator);
}
public SeparatedQueryStringAttribute(string key, string separator) {
_factory = new SeparatedQueryStringValueProviderFactory(key, separator);
}
public void OnResourceExecuting(ResourceExecutingContext context) {
context.ValueProviderFactories.Insert(0, _factory);
}
public void OnResourceExecuted(ResourceExecutedContext context) { }
}
Actually, according to your GetByIdsRequest class, we should detect the existence of the [CommaSeparated] attribute that are decorated on parameter's properties:
// CommaSeparatedQueryStringConvention::Apply(action)
public void Apply(ActionModel action)
{
for (int i = 0; i < action.Parameters.Count; i++)
{
var parameter = action.Parameters[i];
var props = parameter.ParameterType.GetProperties()
.Where(pi => pi.GetCustomAttributes<CommaSeparatedAttribute>().Count() > 0)
;
if (props.Count() > 0)
{
var attribute = new SeparatedQueryStringAttribute(",");
parameter.Action.Filters.Add(attribute);
break;
}
}
}
And now it works fine for me.
A Demo

Avoiding usage of Enums for abstraction and dependency injection, struggling at concept

So right now I am trying to design a new hire program that grants access to active directory groups, generates documents with their information and location.
Right now I am doing this with an enumeration, with a switch statement that sets the details on the ViewModel like this:
case CaneRidgeSettings.Departments.SCSC:
Model.ScannerFolder = #"scan1\Supply Chain Service Center\" + Model.UserId;
Model.ExtensionRanges = "list station 8000 to-ext 8349";
Model.AdministrativeAssistant = Loader.SCSCAdminAssistant;
Model.DuoCode = "Franklin TN - 8175";
Model.PrinterSelectedIndex = (int)CaneRidgeSettings.PrinterGroups.Cane_Ridge_5th_Floor_West;
return await find.FindNextComputer("800SCSC");
The problem I have with this design is that if I ever add more departments to this building, I have to manually update this switch. So I tried a few things around this such as a dictionary, but it didn't seem to bind to a combo-box very well (even when implementing my own INotifyCollectionChanged).
So instead I created an interface that contains this information, for simplicity and length lets just say the interface does this:
public interface IDepartmentInfo
{
string DepartmentName { get; }
List<string> ActiveDirectoryGroups { get; }
string AdministrativeAssistant { get; }
string Floor { get; }
}
I then created a new class that implements this interface
public class SCSC : IDepartmentInfo
{
public string DepartmentName { get; } = "Shared Services";
public List<string> ActiveDirectoryGroups { get; } = new List<string>() {"Example_AD_GRP","Domain_Users"};
public string AdministrativeAssistant { get; } = "Lisa_Smith#outlook.com";
public string Floor { get; } = "5th Floor East";
public override string ToString() => DepartmentName;
}
Then, on my main Building Class I have an observable collection that expects an IDepartmentInfo and initializes those departments
public class CaneRidgeBuilding : IBuilding
{
public ObservableCollection<IDepartmentInfo> Departments { get; set; } = new ObservableCollection<IDepartmentInfo>() {new SCSC(), new ARS()};
public override string ToString()
{
return "CaneRidge";
}
}
On my View Model I implemented a few properties, mainly the BuildingSelectedIndex and the DepartmentSelectedIndex.
I also have an IDepartmentInfo property that notifies when it is changed because it is databound to several labels on my UI.
public class MainWindowViewModel : BindableBase
{
public ObservableCollection<IBuilding> Buildings { get; set; } = new ObservableCollection<IBuilding>() { new CaneRidgeBuilding() };
private ObservableCollection<IDepartmentInfo> _departmentInfos = new ObservableCollection<IDepartmentInfo>();
public ObservableCollection<IDepartmentInfo> DepartmentInfos
{
get { return _departmentInfos; }
set { SetProperty(ref _departmentInfos, value); }
}
private int _buildingIndex = -1;
public int BuildingIndex
{
get { return _buildingIndex; }
set
{
SetProperty(ref _buildingIndex, value);
SetDepartments();
}
}
private void SetDepartments()
{
if (BuildingIndex != -1)
DepartmentInfos = Buildings[BuildingIndex].Departments;
}
private int _departmentIndex = -1;
public int DepartmentIndex
{
get { return _departmentIndex; }
set
{
SetProperty(ref _departmentIndex, value);
LoadDepartmentSettings();
}
}
private IDepartmentInfo _departmentInformation;
public IDepartmentInfo DepartmentInformation
{
get { return _departmentInformation; }
set { SetProperty(ref _departmentInformation, value); }
}
private void LoadDepartmentSettings()
{
if (DepartmentIndex != -1)
DepartmentInformation = DepartmentInfos[DepartmentIndex];
}
private string _title = "Prism Application";
public string Title
{
get { return _title; }
set { SetProperty(ref _title, value); }
}
public MainWindowViewModel()
{
}
}
And it works exactly the way I want it to, however to problem I am running into now is how would I handle dependency injection? If I have 10 departments implementing IDepartmentInfo, how exactly could I pass this to an observable collection?
Because the moment I introduce a new building, if I tell Unity to resolve all IDepartmentInfos, what is going to happen is I'll get every single department even if it doesn't belong to CaneRidge.
If I split the departments to each building, then I run into issues where I can't easily load the departments into the ViewModel, because it is expecting an IDepartmentInfo collection. If I limited it to just one type of collection, then it wouldn't work.
Am I over-complicating things?
Here is an idea.
Custom attribute
Introduce a BuilingAttribute so each IDepartmentInfo implementation can declare Type of the building it belongs to (allow multiple if one department can belong to multiple buildings, I got the idea it can't).
[AttributeUsage(AttributeTargets.Class, AllowMultiple = false)]
public class BuildingAttribute : Attribute
{
public Type BuildingType { get; private set; }
public BuildingAttribute(Type buildingType)
{
this.BuildingType = buildingType;
}
}
DepartmentInfo Collection Factory
An interface that knows how to create a collection of DepartmentInfo for each building Type.
public interface IDepartmentInfoCollectionFactory
{
void RegisterDepartment<T>(Func<IDepartmentInfo> departmentCreator) where T : class, IBuilding;
ObservableCollection<IDepartmentInfo> GetDepartments<T>() where T : class, IBuilding;
}
And the implementation (will be registered as singleton).
public class DepartmentInfoCollectionFactory : IDepartmentInfoCollectionFactory
{
private readonly Dictionary<Type, List<Func<IDepartmentInfo>>> departmentCreators =
new Dictionary<Type, List<Func<IDepartmentInfo>>>();
void IDepartmentInfoCollectionFactory.RegisterDepartment<T>(Func<IDepartmentInfo> departmentCreator)
{
Type buildingType = typeof(T);
if (!this.departmentCreators.ContainsKey(buildingType))
this.departmentCreators.Add(buildingType, new List<Func<IDepartmentInfo>>());
if (!this.departmentCreators[buildingType].Contains(departmentCreator))
this.departmentCreators[buildingType].Add(departmentCreator);
}
ObservableCollection<IDepartmentInfo> IDepartmentInfoCollectionFactory.GetDepartments<T>()
{
Type buildingType = typeof(T);
if (!this.departmentCreators.ContainsKey(buildingType))
throw new InvalidOperationException(
string.Format("No departments have been registered for {0}.", buildingType.ToString()));
ObservableCollection<IDepartmentInfo> departmentInfos = new ObservableCollection<IDepartmentInfo>();
foreach(Func<IDepartmentInfo> creator in this.departmentCreators[buildingType])
{
departmentInfos.Add(creator());
}
return departmentInfos;
}
}
Configuring the factory, so it knows how to create IDepartmentInfo collections.
protected override void ConfigureContainer()
{
Container.RegisterType<IDepartmentInfoCollectionFactory, DepartmentInfoCollectionFactory>(
new ContainerControlledLifetimeManager());
this.ConfigureDepartmentInfoCollectionFactory(Container.Resolve<IDepartmentInfoCollectionFactory>());
}
private void ConfigureDepartmentInfoCollectionFactory(IDepartmentInfoCollectionFactory factory)
{
// Types implementing IDepartmentInfo
var deptInfoTypes = AppDomain.CurrentDomain
.GetAssemblies()
.SelectMany(s => s.GetTypes())
.Where(t => typeof(IDepartmentInfo).IsAssignableFrom(t) && !t.IsInterface);
foreach(Type type in deptInfoTypes)
{
// Get collection of BuildingAttribute for the type
var buildingAttributes = type.GetCustomAttributes(typeof(BuildingAttribute), false)
.OfType<BuildingAttribute>();
if (buildingAttributes.Count() < 1)
throw new InvalidOperationException(
string.Format("The type {0} didn't declare BuildingArgument.", type.ToString()));
var buildingType = buildingAttributes.First().BuildingType;
if (buildingType == null || !buildingType.GetInterfaces().Contains(typeof(IBuilding)))
throw new InvalidOperationException(
string.Format("{0}: BuildingType is not an IBuilding.", type.ToString()));
var registerMethod = typeof(IDepartmentInfoCollectionFactory).GetMethod("RegisterDepartment")
.MakeGenericMethod(new Type[] { buildingType });
registerMethod.Invoke(factory, new object[]
{
new Func<IDepartmentInfo>(() => (IDepartmentInfo)Container.Resolve(type))
});
}
}
Inject the factory.
public class FooBuilding : IBuilding
{
private IDepartmentInfoCollectionFactory factory;
private readonly ObservableCollection<IDepartmentInfo> departmentInfos;
public string Name { get; } = "FooBuilding";
public ObservableCollection<IDepartmentInfo> DepartmentInfos
{
get { return this.departmentInfos; }
}
public FooBuilding(IDepartmentInfoCollectionFactory factory)
{
this.factory = factory;
this.departmentInfos = factory.GetDepartments<FooBuilding>();
}
}
Adding new department
It doesn't require any editing, just create new class with the attribute.
[Building(typeof(FooBuilding))]
public class BarDepartment : IDepartmentInfo
{
public string Name { get; } = "Bar department";
}
I was able to figure out how to inject different buildings and departments, probably not the best way
EDIT: Updated it to use reflection to make it less maintenance
protected override void ConfigureContainer()
{
base.ConfigureContainer();
Container.RegisterTypes(AllClasses.FromLoadedAssemblies()
.Where(type => typeof(IDepartment).IsAssignableFrom(type)), WithMappings.FromAllInterfaces, WithName.TypeName, WithLifetime.None);
ObservableCollection<IBuilding> Buildings = new ObservableCollection<IBuilding>()
{
Container.Resolve<Building1>(new ParameterOverride("departments",GetDepartmentCollection("Building1"))),
Container.Resolve<Building2>(new ParameterOverride("departments",GetDepartmentCollection("Building2")))
};
Container.RegisterInstance(typeof(ObservableCollection<IBuilding>), Buildings,
new ExternallyControlledLifetimeManager());
}
private ObservableCollection<IDepartment> GetDepartmentCollection(string buildingName)
{
var departments = new List<IDepartment>();
foreach (var registration in Container.Registrations.Where( s => s.MappedToType.Namespace.Contains(buildingName)))
{
departments.Add((IDepartment)Container.Resolve(registration.MappedToType));
}
return new ObservableCollection<IDepartment>(departments);
}
Now I am able to completely eliminate the enumeration and it can be extended in the future without breaking any code or requiring me to change anything.

What data structure is appropriate for this?

Within code I want to do something like this:
item.Stage = Stage.Values.ONE;
Where Stage.Values.ONE represents some predefined Stage:
public class Stage
{
[Key]
public virtual int StageId { get; set; }
public string Name { get; set; }
public TimeSpan Span { get; set; }
}
I'm dealing with EF CodeFirst... and I have a lot of stages to define. I'm not sure if I should store the data in the database, or in the dbContext, or what, but I'm looking for the simplest implementation.
I've tried this:
I've tried the following (defining two constants):
public class Stage
{
[Key]
public virtual int StageId { get; set; }
public string Name { get; set; }
public TimeSpan Span { get; set; }
public static class Values
{
public static readonly Stage ONE = new Stage()
{
StageId = 0,
Name = "ONE",
Span = new TimeSpan(0, 0, 0)
};
public static readonly Stage TWO = new Stage()
{
StageId = 1,
Name = "TWO",
Span = new TimeSpan(0, 0, 10)
};
}
But whenever I create a new instance of an entity that has a Stage, a new Stage is added to the db. I just need a few constant stages.
Use of Stage:
public class Side
{
public Side()
{
Stage = Stage.Values.ONE; // Adds new Stage to DB, when it should be a reference to the one I defined above
}
public virtual Stage Stage { get; set; }
}
It looks a bit like an enum, and I've used a kind of 'extended enum' patter several times before with some success. Because you're refencing these values in code, it may not make sense to store them in the database as well, but it's possible if needed.
The technique is described in detail here: http://lostechies.com/jimmybogard/2008/08/12/enumeration-classes/
Basically, you create a base class which provides a number of services similar to an enum, and then to create your "enumerated class" you inherit from it and provide a bunch of static instances which call the constructor with however many properties you need to have.
To avoid link rot, here is the base class to use (just put the whole class into your project somewhere), and scroll down for your own code.
public abstract class Enumeration : IComparable
{
private readonly int _value;
private readonly string _displayName;
protected Enumeration()
{
}
protected Enumeration(int value, string displayName)
{
_value = value;
_displayName = displayName;
}
public int Value
{
get { return _value; }
}
public string DisplayName
{
get { return _displayName; }
}
public override string ToString()
{
return DisplayName;
}
public static IEnumerable<T> GetAll<T>() where T : Enumeration, new()
{
var type = typeof(T);
var fields = type.GetFields(BindingFlags.Public | BindingFlags.Static | BindingFlags.DeclaredOnly);
foreach (var info in fields)
{
var instance = new T();
var locatedValue = info.GetValue(instance) as T;
if (locatedValue != null)
{
yield return locatedValue;
}
}
}
public override bool Equals(object obj)
{
var otherValue = obj as Enumeration;
if (otherValue == null)
{
return false;
}
var typeMatches = GetType().Equals(obj.GetType());
var valueMatches = _value.Equals(otherValue.Value);
return typeMatches && valueMatches;
}
public override int GetHashCode()
{
return _value.GetHashCode();
}
public static int AbsoluteDifference(Enumeration firstValue, Enumeration secondValue)
{
var absoluteDifference = Math.Abs(firstValue.Value - secondValue.Value);
return absoluteDifference;
}
public static T FromValue<T>(int value) where T : Enumeration, new()
{
var matchingItem = parse<T, int>(value, "value", item => item.Value == value);
return matchingItem;
}
public static T FromDisplayName<T>(string displayName) where T : Enumeration, new()
{
var matchingItem = parse<T, string>(displayName, "display name", item => item.DisplayName == displayName);
return matchingItem;
}
private static T parse<T, K>(K value, string description, Func<T, bool> predicate) where T : Enumeration, new()
{
var matchingItem = GetAll<T>().FirstOrDefault(predicate);
if (matchingItem == null)
{
var message = string.Format("'{0}' is not a valid {1} in {2}", value, description, typeof(T));
throw new ApplicationException(message);
}
return matchingItem;
}
public int CompareTo(object other)
{
return Value.CompareTo(((Enumeration)other).Value);
}
}
And now your code will look something like this:
public class Stage : Enumeration
{
public TimeSpan TimeSpan { get; private set; }
public static readonly Stage One
= new Stage (1, "Stage one", new TimeSpan(5));
public static readonly Stage Two
= new Stage (2, "Stage two", new TimeSpan(10));
public static readonly Stage Three
= new Stage (3, "Stage three", new TimeSpan(15));
private EmployeeType() { }
private EmployeeType(int value, string displayName, TimeSpan span) : base(value, displayName)
{
TimeSpan = span;
}
}
Once you have that set up, you can just store the .Value in the database. I'm afraid I haven't done it in EF, but in nHibernate it's reasonably straight-forward to tell a property to just store the ".Value" of the property, and you can wire it back up when you load the value by having it call:
Stage.FromValue<Stage>(intValue);
Hold the Stage as a property of your entity, use it the way you're doing and add
Ignore(x => x.Stage)
to your mapping. This will ignore this property when mapping to your database.
Edit: I misinterpreted the question.
If you want just the different stages in your database, you should put the stages in their own table with an ID, and refer to that ID trough a relationship. Every entity will hold an additional reference and you'll have to define relationships for them.
Is this what you were looking for?

Categories