I have a very simple code of what I am trying to achieve here.
Basically the idea here would be to be able to do a simple FindBy(x => x.<the_column_name>, <the value>); therefore I wouldn't have to copy-paste the same query by changing only the column name.
For now I keep getting an exception from LINQ saying LINQ to Entities does not recognize the method 'System.String Invoke(Spunky.Entities.User)' method, and this method cannot be translated into a store expression
Is that something possible ? Or maybe we're not yet there with EF 6.1?
public class UsersRepository
{
private Lazy<IDataContext> context;
public UsersRepository()
: this(() => new DataContext())
{
}
public UsersRepository(Func<IDataContext> context)
{
this.context = new Lazy<IDataContext>(context);
}
public IQueryable<User> Find(int id)
{
return FindBy(u => u.Id, id);
}
public IQueryable<User> Find(string username)
{
return FindBy(u => u.Username, username);
}
public IQueryable<User> FindBy<T>(Func<User, T> property, T value)
{
return context.Value
.Users
.Where(u => property.Invoke(u).Equals(value));
}
}
If you change the signature
IQueryable<User> FindBy(Expression<Func<User, bool>> predicate)
{
return context.Users.Where(predicate);
}
you can call
return FindBy(u => u.Username == username);
The code hardly changes, but you don't need to manufacture expressions.
You have to enter an expression (not a Func) because expressions can be translated into SQL. A Func is just a .Net delegate, for which no SQL equivalent exists.
You need to construct an expression tree linq to entities can translate to sql:
public IQueryable<T> FindBy<TKey>(Expression<Func<T,TKey>> memberExpression, object value)
{
var parameterVisitor = new ParameterVisitor(memberExpression);
var parameter = parameterVisitor.Parameter;
var constant = Expression.Constant(value);
var equal = Expression.Equal(memberExpression,constant);
var predicate = Expression.Lambda(equal, parameter);
return context.Value.Users.Where(predicate);
}
public class ParameterVisitor : ExpressionVisitor
{
public ParameterExpression Parameter { get;set;}
public ParameterVisitor(Expression expr)
{
this.Visit(expr);
}
protected override VisitParameter(ParameterExpression parameter)
{
Parameter = parameter;
return parameter;
}
}
Still untested from the top of my head.
Edit: Code corrected.
Related
My generic repository method is declared like this:
public virtual async Task<List<T>> GetMany(Expression<Func<T, bool>> filterExpression)
I would like to investigate that filterExpression and check if T is of a certain type AND if so, check if there's a filter for a property called ShardKey. Also, if the type checks out but the filter is not present, I would like to add it.
How can I achieve this?
Using an ExpressionVisitor you can search the Expression tree to determine if a reference to the desired property is made. If you wanted to be more sure, you could test to see if it is in a certain type of logical test as well.
If you don't find the desired reference, you can add a parent test to the filterExpression.
Here is an example, assuming ShardKey is an int:
public virtual async Task<List<T>> GetMany<T>(Expression<Func<T, bool>> filterExpression) {
if (typeof(T) == typeof(CertainType) && !filterExpression.ReferencesMember<CertainType>("ShardKey")) {
var skTest = Expression.Equal(Expression.PropertyOrField(filterExpression.Parameters[0], "ShardKey"), Expression.Constant(7));
filterExpression = Expression.Lambda<Func<T,bool>>(Expression.AndAlso(skTest, filterExpression.Body), filterExpression.Parameters);
}
Here is the ExpressionVisitor implementation:
public static class ExpressionExt {
public static bool ReferencesMember<T>(this Expression ex, string memberName) {
var mi = typeof(T).GetMember(memberName);
if (mi.Length != 1)
throw new ArgumentException($"{memberName} ambiguous or not found for {typeof(T).Name}");
var mf = new MemberFinder(mi[0]);
mf.Visit(ex);
return mf.Found;
}
class MemberFinder : ExpressionVisitor {
MemberInfo mi;
public bool Found = false;
public MemberFinder(MemberInfo _mi) => mi = _mi;
[return: NotNullIfNotNull("node")]
public override Expression Visit(Expression node) {
if (Found)
return node;
else
return base.Visit(node);
}
protected override Expression VisitMember(MemberExpression me) {
if (me.Member == mi)
Found = true;
return me;
}
}
}
I'm looking for a way to use reflection in my linq queries to increase code reusability. for example, i want to change this method :
private Orders GetObjectGraph(MyDbContext dbContext, int orderID)
{
return dbContext.Orders.Include(o => o.OrderDetails).AsNoTracking().FirstOrDefault(o => o.OrderID == orderID);
}
To something like this :
private object GetObjectGraph(MyDbContext dbContext, string masterEntityName, string detailsEntityName, string masterFieldName, object masterFieldValue)
{
return dbContext[masterEntityName].Include(o => o[detailsEntityName]).AsNoTracking().FirstOrDefault(o => o.[masterFieldName] == masterFieldValue);
}
Can anybody help me how to accomplish this task ?
Thanks in advance
You can try something like this (if you are not getting strings from somewhere outside in runtime, if you do than this solution can be adapted to build corresponding Expression Trees from strings):
private T GetObjectGraph<T, TDetails, TField>(
MyDbContext dbContext,
Expression<Func<T, TDetails>> details,
Expression<Func<T, TField>> field,
TField value) where T: class
{
var p = field.Parameters[0];
var eq = Expression.Equal(field.Body, Expression.Constant(value));
var expr = Expression.Lambda<Func<T, bool>>(eq, p);
return dbContext
.Set<T>()
.Include(details)
.AsNoTracking()
.FirstOrDefault(expr);
}
And usage:
GetObjectGraph(dbContext, (Order o) => o.OrderDetails, o => o.OrderID, orderID)
If you don't need to parse strings and so on this could be simplified to:
public static class Ext
{
public static T GetObjectGraph<T, TDetails>(
this DbSet<T> set,
Expression<Func<T, TDetails>> details,
Expression<Func<T, bool>> filter) where T : class
{
return set
.Include(details)
.AsNoTracking()
.FirstOrDefault(filter);
}
}
And usage:
dbContext.Orders(o => o.OrderDetails, o => o.OrderID == orderID);
And for more insight how to handle strings you can look at this question.
I have an extension method that lets you generically include data in EF:
public static IQueryable<T> IncludeMultiple<T>(this IQueryable<T> query, params Expression<Func<T, object>>[] includes)
where T : class
{
if (includes != null)
{
query = includes.Aggregate(query, (current, include) => current.Include(include));
}
return query;
}
This allows me to have methods in my repository like this:
public Patient GetById(int id, params Expression<Func<Patient, object>>[] includes)
{
return context.Patients
.IncludeMultiple(includes)
.FirstOrDefault(x => x.PatientId == id);
}
I believe the extension method worked before EF Core, but now including "children" is done like this:
var blogs = context.Blogs
.Include(blog => blog.Posts)
.ThenInclude(post => post.Author);
Is there a way to alter my generic extension method to support EF Core's new ThenInclude() practice?
As said in comments by other, you can take EF6 code to parse your expressions and apply the relevant Include/ThenInclude calls. It does not look that hard after all, but as this was not my idea, I would rather not put an answer with the code for it.
You may instead change your pattern for exposing some interface allowing you to specify your includes from the caller without letting it accessing the underlying queryable.
This would result in something like:
using YourProject.ExtensionNamespace;
// ...
patientRepository.GetById(0, ip => ip
.Include(p => p.Addresses)
.ThenInclude(a=> a.Country));
The using on namespace must match the namespace name containing the extension methods defined in the last code block.
GetById would be now:
public static Patient GetById(int id,
Func<IIncludable<Patient>, IIncludable> includes)
{
return context.Patients
.IncludeMultiple(includes)
.FirstOrDefault(x => x.EndDayID == id);
}
The extension method IncludeMultiple:
public static IQueryable<T> IncludeMultiple<T>(this IQueryable<T> query,
Func<IIncludable<T>, IIncludable> includes)
where T : class
{
if (includes == null)
return query;
var includable = (Includable<T>)includes(new Includable<T>(query));
return includable.Input;
}
Includable classes & interfaces, which are simple "placeholders" on which additional extensions methods will do the work of mimicking EF Include and ThenInclude methods:
public interface IIncludable { }
public interface IIncludable<out TEntity> : IIncludable { }
public interface IIncludable<out TEntity, out TProperty> : IIncludable<TEntity> { }
internal class Includable<TEntity> : IIncludable<TEntity> where TEntity : class
{
internal IQueryable<TEntity> Input { get; }
internal Includable(IQueryable<TEntity> queryable)
{
// C# 7 syntax, just rewrite it "old style" if you do not have Visual Studio 2017
Input = queryable ?? throw new ArgumentNullException(nameof(queryable));
}
}
internal class Includable<TEntity, TProperty> :
Includable<TEntity>, IIncludable<TEntity, TProperty>
where TEntity : class
{
internal IIncludableQueryable<TEntity, TProperty> IncludableInput { get; }
internal Includable(IIncludableQueryable<TEntity, TProperty> queryable) :
base(queryable)
{
IncludableInput = queryable;
}
}
IIncludable extension methods:
using Microsoft.EntityFrameworkCore;
// others using ommitted
namespace YourProject.ExtensionNamespace
{
public static class IncludableExtensions
{
public static IIncludable<TEntity, TProperty> Include<TEntity, TProperty>(
this IIncludable<TEntity> includes,
Expression<Func<TEntity, TProperty>> propertySelector)
where TEntity : class
{
var result = ((Includable<TEntity>)includes).Input
.Include(propertySelector);
return new Includable<TEntity, TProperty>(result);
}
public static IIncludable<TEntity, TOtherProperty>
ThenInclude<TEntity, TOtherProperty, TProperty>(
this IIncludable<TEntity, TProperty> includes,
Expression<Func<TProperty, TOtherProperty>> propertySelector)
where TEntity : class
{
var result = ((Includable<TEntity, TProperty>)includes)
.IncludableInput.ThenInclude(propertySelector);
return new Includable<TEntity, TOtherProperty>(result);
}
public static IIncludable<TEntity, TOtherProperty>
ThenInclude<TEntity, TOtherProperty, TProperty>(
this IIncludable<TEntity, IEnumerable<TProperty>> includes,
Expression<Func<TProperty, TOtherProperty>> propertySelector)
where TEntity : class
{
var result = ((Includable<TEntity, IEnumerable<TProperty>>)includes)
.IncludableInput.ThenInclude(propertySelector);
return new Includable<TEntity, TOtherProperty>(result);
}
}
}
IIncludable<TEntity, TProperty> is almost like IIncludableQueryable<TEntity, TProperty> from EF, but it does not extend IQueryable and does not allow reshaping the query.
Of course if the caller is in the same assembly, it can still cast the IIncludable to Includable and start fiddling with the queryable. But well, if someone wants to get it wrong, there is no way we would prevent him doing so (reflection allows anything). What does matter is the exposed contract.
Now if you do not care about exposing IQueryable to the caller (which I doubt), obviously just change your params argument for a Func<Queryable<T>, Queryable<T>> addIncludes argument, and avoid coding all those things above.
And the best for the end: I have not tested this, I do not use Entity Framework currently!
For posterity, another less eloquent, but simpler solution that makes use of the Include() overload that uses navigationPropertyPath:
public static class BlogIncludes
{
public const string Posts = "Posts";
public const string Author = "Posts.Author";
}
internal static class DataAccessExtensions
{
internal static IQueryable<T> IncludeMultiple<T>(this IQueryable<T> query,
params string[] includes) where T : class
{
if (includes != null)
{
query = includes.Aggregate(query, (current, include) => current.Include(include));
}
return query;
}
}
public Blog GetById(int ID, params string[] includes)
{
var blog = context.Blogs
.Where(x => x.BlogId == id)
.IncludeMultiple(includes)
.FirstOrDefault();
return blog;
}
And the repository call is:
var blog = blogRepository.GetById(id, BlogIncludes.Posts, BlogIncludes.Author);
You can do something like this:
public Patient GetById(int id, Func<IQueryable<Patient>, IIncludableQueryable<Patient, object>> includes = null)
{
IQueryable<Patient> queryable = context.Patients;
if (includes != null)
{
queryable = includes(queryable);
}
return queryable.FirstOrDefault(x => x.PatientId == id);
}
var patient = GetById(1, includes: source => source.Include(x => x.Relationship1).ThenInclude(x => x.Relationship2));
I made this method to do the dynamic includes. This way the "Select" command can be used in lambda to include just as it was in the past.
The call works like this:
repository.IncludeQuery(query, a => a.First.Second.Select(b => b.Third), a => a.Fourth);
private IQueryable<TCall> IncludeQuery<TCall>(
params Expression<Func<TCall, object>>[] includeProperties) where TCall : class
{
IQueryable<TCall> query;
query = context.Set<TCall>();
foreach (var property in includeProperties)
{
if (!(property.Body is MethodCallExpression))
query = query.Include(property);
else
{
var expression = property.Body as MethodCallExpression;
var include = GenerateInclude(expression);
query = query.Include(include);
}
}
return query;
}
private string GenerateInclude(MethodCallExpression expression)
{
var result = default(string);
foreach (var argument in expression.Arguments)
{
if (argument is MethodCallExpression)
result += GenerateInclude(argument as MethodCallExpression) + ".";
else if (argument is MemberExpression)
result += ((MemberExpression)argument).Member.Name + ".";
else if (argument is LambdaExpression)
result += ((MemberExpression)(argument as LambdaExpression).Body).Member.Name + ".";
}
return result.TrimEnd('.');
}
Ofcourse there is,
you could traverse the Expression tree of original params, and any nested includes, add them as
.Include(entity => entity.NavigationProperty)
.ThenInclude(navigationProperty.NestedNavigationProperty)
But its not trivial, but definitely very doable, please share if you do, as it can defintiely be reused!
public Task<List<TEntity>> GetAll()
{
var query = _Db.Set<TEntity>().AsQueryable();
foreach (var property in _Db.Model.FindEntityType(typeof(TEntity)).GetNavigations())
query = query.Include(property.Name);
return query.ToListAsync();
}
I adhere the simpler solution that makes use of the Include() overload that uses string navigationPropertyPath. The simplest that I can write is this extension method below.
using Microsoft.EntityFrameworkCore;
using System.Linq;
namespace MGame.Data.Helpers
{
public static class IncludeBuilder
{
public static IQueryable<TSource> Include<TSource>(this IQueryable<TSource> queryable, params string[] navigations) where TSource : class
{
if (navigations == null || navigations.Length == 0) return queryable;
return navigations.Aggregate(queryable, EntityFrameworkQueryableExtensions.Include); // EntityFrameworkQueryableExtensions.Include method requires the constraint where TSource : class
}
}
}
public TEntity GetByIdLoadFull(string id, List<string> navigatonProoperties)
{
if (id.isNullOrEmpty())
{
return null;
}
IQueryable<TEntity> query = dbSet;
if (navigationProperties != null)
{
foreach (var navigationProperty in navigationProperties)
{
query = query.Include(navigationProperty.Name);
}
}
return query.SingleOrDefault(x => x.Id == id);
}
Here is a much simpler solution, idea is to cast the dbset to iqueryable and then recursively include properties
I handle it with like that;
I have Article entity. It includes ArticleCategory entity. And also ArticleCategory entity includes Category entity.
So: Article -> ArticleCategory -> Category
In My Generic Repository;
public virtual IQueryable<T> GetIncluded(params Func<IQueryable<T>, IIncludableQueryable<T, object>>[] include)
{
IQueryable<T> query = Entities; // <- this equals = protected virtual DbSet<T> Entities => _entities ?? (_entities = _context.Set<T>());
if (include is not null)
{
foreach (var i in include)
{
query = i(query);
}
}
return query;
}
And I can use it like that;
var query = _articleReadRepository.GetIncluded(
i => i.Include(s => s.ArticleCategories).ThenInclude(s => s.Category),
i => i.Include(s => s.User)
);
for access control purposes in a intensive DB use system I had to implement an objectset wrapper, where the AC will be checked.
The main objective is make this change preserving the existing code for database access, that is implemented with linq to entities all over the classes (there is no centralized layer for database).
The ObjectSetWrapper created is like that:
public class ObjectSetWrapper<TEntity> : IQueryable<TEntity> where TEntity : EntityObject
{
private IQueryable<TEntity> QueryableModel;
private ObjectSet<TEntity> ObjectSet;
public ObjectSetWrapper(ObjectSet<TEntity> objectSetModels)
{
this.QueryableModel = objectSetModels;
this.ObjectSet = objectSetModels;
}
public ObjectQuery<TEntity> Include(string path)
{
return this.ObjectSet.Include(path);
}
public void DeleteObject(TEntity #object)
{
this.ObjectSet.DeleteObject(#object);
}
public void AddObject(TEntity #object)
{
this.ObjectSet.AddObject(#object);
}
public IEnumerator<TEntity> GetEnumerator()
{
return QueryableModel.GetEnumerator();
}
public Type ElementType
{
get { return typeof(TEntity); }
}
public System.Linq.Expressions.Expression Expression
{
get { return this.QueryableModel.Expression; }
}
public IQueryProvider Provider
{
get { return this.QueryableModel.Provider; }
}
public void Attach(TEntity entity)
{
this.ObjectSet.Attach(entity);
}
public void Detach(TEntity entity)
{
this.ObjectSet.Detach(entity);
}
IEnumerator IEnumerable.GetEnumerator()
{
return this.QueryableModel.GetEnumerator();
}
}
It's really simple and works for simple queries, like that:
//db.Product is ObjectSetWrapper<Product>
var query = (from item in db.Product where item.Quantity > 0 select new { item.Id, item.Name, item.Value });
var itensList = query.Take(10).ToList();
But when I have subqueries like that:
//db.Product is ObjectSetWrapper<Product>
var query = (from item in db.Product
select new
{
Id = item.Id,
Name = item.Name,
SalesQuantity = (from sale in db.Sale where sale.ProductId == item.Id select sale.Id).Count()
}).OrderByDescending(x => x.SalesQuantity);
var productsList = query.Take(10).ToList();
I get NotSupportedException, saying I can't create a constant value of my inner query entity type:
Unable to create a constant value of type 'MyNamespace.Model.Sale'.
Only primitive types or enumeration types are supported in this
context.
How can I get my queries working? I don't really need to make my wrapper an ObjectSet type, I just need to use it in queries.
Updated
I have changed my class signature. Now it's also implementing IObjectSet<>, but I'm getting the same NotSupportedException:
public class ObjectSetWrapper<TEntity> : IQueryable<TEntity>, IObjectSet<TEntity> where TEntity : EntityObject
EDIT:
The problem is that the following LINQ construction is translated into LINQ expression containing your custom class inside (ObjectSetWrapper).
var query = (from item in db.Product
select new
{
Id = item.Id,
Name = item.Name,
SalesQuantity = (from sale in db.Sale where sale.ProductId == item.Id select sale.Id).Count()
}).OrderByDescending(x => x.SalesQuantity);
LINQ to Entities tries to convert this expression into SQL statement, but it has no idea how to deal with the custom classes (as well as custom methods).
The solution in such cases is to replace IQueryProvider with the custom one, which should intercept the query execution and translate LINQ expression, containing custom classes/methods into valid LINQ to Entities expression (which operates with entities and object sets).
Expression conversion is performed using the class, derived from ExpressionVisitor, which performs expression tree traversal, replacing relevant nodes, to the nodes which can be accepted by LINQ to Entities
Part 1 - IQueryWrapper
// Query wrapper interface - holds and underlying query
interface IQueryWrapper
{
IQueryable UnderlyingQueryable { get; }
}
Part 2 - Abstract QueryWrapperBase (not generic)
abstract class QueryWrapperBase : IQueryProvider, IQueryWrapper
{
public IQueryable UnderlyingQueryable { get; private set; }
class ObjectWrapperReplacer : ExpressionVisitor
{
public override Expression Visit(Expression node)
{
if (node == null || !typeof(IQueryWrapper).IsAssignableFrom(node.Type)) return base.Visit(node);
var wrapper = EvaluateExpression<IQueryWrapper>(node);
return Expression.Constant(wrapper.UnderlyingQueryable);
}
public static Expression FixExpression(Expression expression)
{
var replacer = new ObjectWrapperReplacer();
return replacer.Visit(expression);
}
private T EvaluateExpression<T>(Expression expression)
{
if (expression is ConstantExpression) return (T)((ConstantExpression)expression).Value;
var lambda = Expression.Lambda(expression);
return (T)lambda.Compile().DynamicInvoke();
}
}
protected QueryWrapperBase(IQueryable underlyingQueryable)
{
UnderlyingQueryable = underlyingQueryable;
}
public abstract IQueryable<TElement> CreateQuery<TElement>(Expression expression);
public abstract IQueryable CreateQuery(Expression expression);
public TResult Execute<TResult>(Expression expression)
{
return (TResult)Execute(expression);
}
public object Execute(Expression expression)
{
expression = ObjectWrapperReplacer.FixExpression(expression);
return typeof(IQueryable).IsAssignableFrom(expression.Type)
? ExecuteQueryable(expression)
: ExecuteNonQueryable(expression);
}
protected object ExecuteNonQueryable(Expression expression)
{
return UnderlyingQueryable.Provider.Execute(expression);
}
protected IQueryable ExecuteQueryable(Expression expression)
{
return UnderlyingQueryable.Provider.CreateQuery(expression);
}
}
Part 3 - Generic QueryWrapper<TElement>
class QueryWrapper<TElement> : QueryWrapperBase, IOrderedQueryable<TElement>
{
private static readonly MethodInfo MethodCreateQueryDef = GetMethodDefinition(q => q.CreateQuery<object>(null));
public QueryWrapper(IQueryable<TElement> underlyingQueryable) : this(null, underlyingQueryable)
{
}
protected QueryWrapper(Expression expression, IQueryable underlyingQueryable) : base(underlyingQueryable)
{
Expression = expression ?? Expression.Constant(this);
}
public virtual IEnumerator<TElement> GetEnumerator()
{
return ((IEnumerable<TElement>)Execute<IEnumerable>(Expression)).GetEnumerator();
}
IEnumerator IEnumerable.GetEnumerator()
{
return GetEnumerator();
}
public Expression Expression { get; private set; }
public Type ElementType
{
get { return typeof(TElement); }
}
public IQueryProvider Provider
{
get { return this; }
}
public override IQueryable CreateQuery(Expression expression)
{
var expressionType = expression.Type;
var elementType = expressionType
.GetInterfaces()
.Single(t => t.IsGenericType && t.GetGenericTypeDefinition() == typeof(IEnumerable<>))
.GetGenericArguments()
.Single();
var createQueryMethod = MethodCreateQueryDef.MakeGenericMethod(elementType);
return (IQueryable)createQueryMethod.Invoke(this, new object[] { expression });
}
public override IQueryable<TNewElement> CreateQuery<TNewElement>(Expression expression)
{
return new QueryWrapper<TNewElement>(expression, UnderlyingQueryable);
}
private static MethodInfo GetMethodDefinition(Expression<Action<QueryWrapper<TElement>>> methodSelector)
{
var methodCallExpression = (MethodCallExpression)methodSelector.Body;
return methodCallExpression.Method.GetGenericMethodDefinition();
}
}
Part 4 - finally your ObjectSetWrapper
public class ObjectSetWrapper<TEntity> : IQueryable<TEntity>, IQueryWrapper where TEntity : class
{
private IQueryable<TEntity> QueryableModel;
private ObjectSet<TEntity> ObjectSet;
public ObjectSetWrapper(ObjectSet<TEntity> objectSetModels)
{
this.QueryableModel = new QueryWrapper<TEntity>(objectSetModels);
this.ObjectSet = objectSetModels;
}
public ObjectQuery<TEntity> Include(string path)
{
return this.ObjectSet.Include(path);
}
public void DeleteObject(TEntity #object)
{
this.ObjectSet.DeleteObject(#object);
}
public void AddObject(TEntity #object)
{
this.ObjectSet.AddObject(#object);
}
public IEnumerator<TEntity> GetEnumerator()
{
return QueryableModel.GetEnumerator();
}
public Type ElementType
{
get { return typeof(TEntity); }
}
public System.Linq.Expressions.Expression Expression
{
get { return this.QueryableModel.Expression; }
}
public IQueryProvider Provider
{
get { return this.QueryableModel.Provider; }
}
public void Attach(TEntity entity)
{
this.ObjectSet.Attach(entity);
}
public void Detach(TEntity entity)
{
this.ObjectSet.Detach(entity);
}
IEnumerator IEnumerable.GetEnumerator()
{
return this.QueryableModel.GetEnumerator();
}
IQueryable IQueryWrapper.UnderlyingQueryable
{
get { return this.ObjectSet; }
}
}
Your inner query fails because you are referencing another dataset when you should be traversing foreign keys:
SalesQuantity = item.Sales.Count()
I'm trying to add a LINQ or DbContext extension method to get an element (FirstOrDefault) but if one does not already exist then create a new instance with data (FirstOrCreate) instead of returning null.
is this possible?
i.e.:
public static class LINQExtension
{
public static TSource FirstOrCreate<TSource>(
this IEnumerable<TSource> source,
Func<TSource, bool> predicate)
{
if (source.First(predicate) != null)
{
return source.First(predicate);
}
else
{
return // ???
}
}
}
and a usage could be:
using (var db = new MsBoxContext())
{
var status = db.EntitiesStatus.FirstOrCreate(s => s.Name == "Enabled");
//Here we should get the object if we find one
//and if it doesn't exist create and return a new instance
db.Entities.Add(new Entity()
{
Name = "New Entity",
Status = status
});
}
I hope that you understand my approach.
public static class LINQExtension
{
public static TSource FirstOrCreate<TSource>(
this IQueryable<TSource> source,
Expression<Func<TSource, bool>> predicate,
Func<T> defaultValue)
{
return source.FirstOrDefault(predicate) ?? defaultValue();
}
}
usage
var status = db.EntitiesStatus.FirstOrCreate(s => s.Name == "Enabled",
() => new EntityStatus {Name = "Enabled"});
However you must note that this will not work quite like FirstOrDefault().
If you did the following
var listOfStuff = new List<string>() { "Enabled" };
var statuses = from s in listOfStuff
select db.EntitiesStatus.FirstOrCreate(s => s.Name == "Enabled",
() => new EntityStatus {Name = "Enabled"});
You would get O(n) hits to the database.
However I suspect if you did...
var listOfStuff = new List<string>() { "Enabled" };
var statuses = from s in listOfStuff
select db.EntitiesStatus.FirstOrDefault(s => s.Name == "Enabled")
?? new EntityStatus {Name = "Enabled"};
It is plausible it could work...
conclussion:
instead of implement an extension method the best solution is using ?? operator in that way:
var status = db.EntitiesStatus.FirstOrDefault(s => s.Name == "Enabled") ?? new EntityStatus(){Name = "Enabled"};
I am a self taught programmer and I am really bad at typing so i was looking for the exact same thing. I ended up writing my own. It took a few steps and revisions before it would work with more than 1 property. Of course there are some limitations and I haven't fully tested it but so far it seems to work for my purposes of keeping the records distinct in the DB and shortening the code (typing time).
public static class DataExtensions
{
public static TEntity InsertIfNotExists<TEntity>(this ObjectSet<TEntity> objectSet, Expression<Func<TEntity, bool>> predicate) where TEntity : class, new()
{
TEntity entity;
#region Check DB
entity = objectSet.FirstOrDefault(predicate);
if (entity != null)
return entity;
#endregion
//NOT in the Database... Check Local cotext so we do not enter duplicates
#region Check Local Context
entity = objectSet.Local().AsQueryable().FirstOrDefault(predicate);
if (entity != null)
return entity;
#endregion
///********* Does NOT exist create entity *********\\\
entity = new TEntity();
// Parse Expression Tree and set properties
//Hit a recurrsive function to get all the properties and values
var body = (BinaryExpression)((LambdaExpression)predicate).Body;
var dict = body.GetDictionary();
//Set Values on the new entity
foreach (var item in dict)
{
entity.GetType().GetProperty(item.Key).SetValue(entity, item.Value);
}
return entity;
}
public static Dictionary<string, object> GetDictionary(this BinaryExpression exp)
{
//Recurssive function that creates a dictionary of the properties and values from the lambda expression
var result = new Dictionary<string, object>();
if (exp.NodeType == ExpressionType.AndAlso)
{
result.Merge(GetDictionary((BinaryExpression)exp.Left));
result.Merge(GetDictionary((BinaryExpression)exp.Right));
}
else
{
result[((MemberExpression)exp.Left).Member.Name] = exp.Right.GetExpressionVaule();
}
return result;
}
public static object GetExpressionVaule(this Expression exp)
{
if (exp.NodeType == ExpressionType.Constant)
return ((ConstantExpression)exp).Value;
if (exp.Type.IsValueType)
exp = Expression.Convert(exp, typeof(object));
//Taken From http://stackoverflow.com/questions/238413/lambda-expression-tree-parsing
var accessorExpression = Expression.Lambda<Func<object>>(exp);
Func<object> accessor = accessorExpression.Compile();
return accessor();
}
public static IEnumerable<T> Local<T>(this ObjectSet<T> objectSet) where T : class
{
//Taken From http://blogs.msdn.com/b/dsimmons/archive/2009/02/21/local-queries.aspx?Redirected=true
return from stateEntry in objectSet.Context.ObjectStateManager.GetObjectStateEntries(
EntityState.Added |
EntityState.Modified |
EntityState.Unchanged)
where stateEntry.Entity != null && stateEntry.EntitySet == objectSet.EntitySet
select stateEntry.Entity as T;
}
public static void Merge<TKey, TValue>(this Dictionary<TKey, TValue> me, Dictionary<TKey, TValue> merge)
{
//Taken From http://stackoverflow.com/questions/4015204/c-sharp-merging-2-dictionaries
foreach (var item in merge)
{
me[item.Key] = item.Value;
}
}
}
Usage is as simple as:
var status = db.EntitiesStatus.InsertIfNotExists(s => s.Name == "Enabled");
The extension will check the database first, if is not found it will check the local context (so you do not add it twice), if it is still not found it creates the entity, parses the expression tree to get the properties and values from the lambda expression, sets those values on a new entity, adds the entity to the context and returns the new entity.
A few things to be aware of...
This does not handle all possible uses (assumes all the expressions in the lambda are ==)
The project I did this in is using an ObjectContext as apposed to a DBContext (I have not switched yet so I don't know if this would work with DBContext. I assume it would not be difficult to change)
I am self-taught so there maybe many ways to optimize this. If you have any input please let me know.
What about this extension that also adds the new created entity to the DbSet.
public static class DbSetExtensions
{
public static TEntity FirstOrCreate<TEntity>(
this DbSet<TEntity> dbSet,
Expression<Func<TEntity, bool>> predicate,
Func<TEntity> defaultValue)
where TEntity : class
{
var result = predicate != null
? dbSet.FirstOrDefault(predicate)
: dbSet.FirstOrDefault();
if (result == null)
{
result = defaultValue?.Invoke();
if (result != null)
dbSet.Add(result);
}
return result;
}
public static TEntity FirstOrCreate<TEntity>(
this DbSet<TEntity> dbSet,
Func<TEntity> defaultValue)
where TEntity : class
{
return dbSet.FirstOrCreate(null, defaultValue);
}
}
The usage with predicate:
var adminUser = DbContext.Users.FirstOrCreate(u => u.Name == "Admin", () => new User { Name = "Admin" });
or without predicate:
var adminUser = DbContext.Users.FirstOrCreate(() => new User { Name = "Admin" });