I have two classes, a ViewModel and a Dto, that are basically identical except that the Dto has a field 'readonly long? Phone;' while the ViewModel has the a property 'string Phone { get; set; }'.
The only way I've found to get AutoMapper to work is by changing the ViewModel Property to a backing property:
public long? Phone { get; set; }
public string PhoneNumberString
{
get
{
var srv = DependencyResolver.Current.GetService<IPhoneNumberService>();
return srv.GetFormattedPhoneNumber(Phone);
}
set
{
var srv = DependencyResolver.Current.GetService<IPhoneNumberService>();
Phone = srv.GetLongPhoneNumber(value);
}
}
And then in AutoMapper, have a gigantic line to call the constructor:
Mapper.Initialize(cfg =>
{
cfg.CreateMap<MyViewModel, MyDto>()
.ConstructUsing(src => new MyDto(
src.Phone
/* ...Some ~30 other parameters here... */))
.ReverseMap();
});
...There must be a better way to do this? I've tried these:
.ForSourceMember(x => x.PhoneNumberString, opt => opt.DoNotValidate())
and
.ForMember(x => x.PhoneNumberString, opt => opt.Ignore())
and
.ForMember(viewModel => viewModel.Phone, options => options.MapFrom<PhoneNumberResolver>());//PhoneNumberResolver implements IValueResolver<ProspectMaintenanceViewModel, ProspectMaintenanceDto, long?>
Which all give 'Core.DTO.MyDto needs to have a constructor with 0 args or only optional args.' when trying to map, and:
.ForMember(dest => dest.Phone, opt => opt.MapFrom(src => 5))
Which gives 'System.ArgumentException: 'Expression must be writeable' when trying to configure AutoMapper.
Is there some way I can make AutoMapper understand that it can entirely ignore PhoneNumberString (or, even better, some way by which I can make it map long? to string so I don't need the backing property) without having to use the dto's constructor?
Is there any special reason that requires your DTO to not have a default constructor?
I have all my fields as readonly so that I can include a constructor that modifies (e.g. 'Description = description?.Trim();') and validates (e.g. 'if (Phone.HasValue && Phone.ToString().Length != 10) throw ...') the parameters. This way I can ensure that the Dto, being a value object, is always in a valid state.
1) Mapping to readonly field
So you have a Dto class:
public class Dto
{
public readonly long? PhoneNumber;
}
And then you are trying to force AutoMapper to do this:
var dto = new Dto();
dto.PhoneNumber = 123; // <== ERROR! A readonly field cannot be assigned to.
AutoMapper cannot write to readonly fields or properties. In matter of fact you neither. Either turn your field into a property with protected or private setter:
public class Dto
{
public long? PhoneNumber { get; private set; }
}
or make it a regular field by removing the readonly keyword:
public class Dto
{
public long? PhoneNumber;
}
2) Custom value resolving
ASP.NET MVC
Use a ValueResolver:
public class StringPhoneNumberResolver : IValueResolver<Dto, ViewModel, string>
{
private readonly IPhoneNumberService _phoneNumberService;
public StringPhoneNumberResolver()
{
_phoneNumberService = DependencyResolver.Current.GetService<IPhoneNumberService>();
}
public string Resolve(Dto source, ViewModel destination, string destMember, ResolutionContext context)
{
return _phoneNumberService.GetFormattedPhoneNumber(source.PhoneNumber);
}
}
You should know that generally it is an anti-pattern to have service injection in a DTO or IValueResolver. AutoMapper should be dumb and all kind of injections and so on should be handled elsewhere. That being said, here is the AutoMapper configuration:
Mapper.Initialize(cfg =>
{
cfg.CreateMap<Dto, ViewModel>()
.ForMember(viewModel => viewModel.PhoneNumber, options =>
options.MapFrom<StringPhoneNumberResolver>());
});
If you want to reverse the process of long ==> string to string ==> long simply add another value resolver:
public class LongPhoneNumberResolver : IValueResolver<ViewModel, Dto, long?>
{
private readonly IPhoneNumberService _phoneNumberService;
public LongPhoneNumberResolver()
{
_phoneNumberService = DependencyResolver.Current.GetService<IPhoneNumberService>();
}
public long? Resolve(ViewModel source, Dto destination, long? destMember, ResolutionContext context)
{
return _phoneNumberService.GetLongPhoneNumber(source.PhoneNumber);
}
}
.NET Core
If you would operate in .NET Core environment, which fully supports IServiceCollection integration, you should add this AutoMapper configuration:
serviceCollection.AddAutoMapper(config =>
{
config.CreateMap<Dto, ViewModel>()
.ForMember(viewModel => viewModel.PhoneNumber, options =>
options.MapFrom<StringPhoneNumberResolver>());
}, typeof(Startup));
and then have IPhoneNumberServce automagically injected into value resolver:
public StringPhoneNumberResolver(IPhoneNumberService phoneNumberService)
{
_phoneNumberService = phoneNumberService;
}
For dependency injection I used automapper.extensions.microsoft.dependencyinjection package.
Well, I found the problem. It has absolutely nothing to do with what I thought it did. Mapping map long? to string works out of the box.
The problem I had was with an entirely different property.
I had the following structure:
public class MyDto
{
public readonly AddressDto BillingAddress;
public readonly AddressDto ShippingAddress;
public readonly long? Phone;
...
}
public class AddressDto
{
public readonly string Country;
public readonly string SubnationalEntity;
...
}
public class MyViewModel
{
public string BillingAddressCountry { get; set; }
public string BillingAddressSubnationalEntity { get; set; }
public string ShippingAddressCountry { get; set; }
public string ShippingAddressSubnationalEntity { get; set; }
public string Phone { get; set; }
...
}
It worked once I changed it to the following:
public class MyDto
{
public readonly AddressDto BillingAddress;
public readonly AddressDto ShippingAddress;
public readonly long? Phone;
...
}
public class AddressDto
{
public readonly string Country;
public readonly string SubnationalEntity;
...
}
public class MyViewModel
{
public string AddressViewModel BillingAddress { get; set; }
public string AddressViewModel ShippingAddress { get; set; }
public string Phone { get; set; }
...
}
public class AddressViewModel
{
public string Country { get; set; }
public string SubnationalEntity { get; set; }
...
}
Please Consider this code:
public class MyClass
{
[CustomAttributes.GridColumn(1)]
public string Code { get; set; }
[CustomAttributes.GridColumn(3)]
public string Name { get; set; }
[CustomAttributes.GridColumn(2)]
public DateTime? ProductionDate { get; set; }
public DateTime? ProductionExpiredDate { get; set; }
[CustomAttributes.GridColumn(4)]
public int ProductOwner { get; set; }
}
I want to get a dictionary for all properties that have CustomAttributes.GridColumn and sort them by the number in GridColumn attribute and type of them like this:
PropertyName Type
---------------------------------
Code string
ProductionDate DateTime?
Name string
ProductOwner int
How I can do this?
Thanks
Something like this ought to work:
private IDictionary<string, Type> GetProperties<T>()
{
var type = typeof(T);
return type.GetProperties(BindingFlags.Instance | BindingFlags.Public)
.Select(p => new { Property = p, Attribute = p.GetCustomAttribute<CustomAttributes.GridColumnAttribute>() })
.Where(p => p.Attribute != null)
.OrderBy(p => p.Attribute.Index)
.ToDictionary(p => p.Property.Name, p => p.Property.PropertyType);
}
It first gets all of the public properties, creates an object which contains the property and the attribute, filters the list to only include properties where the attribute exists, sorts by the attribute index, and finally converts it into a dictionary.
I'm assuming the attribute is defined similar to this:
public class GridColumnAttribute : System.Attribute
{
public GridColumnAttribute(int index)
{
this.Index = index;
}
public int Index { get; set; }
}
P.S. GetCustomAttribute<T>() is an extension method that lives in System.Reflection.CustomAttributeExtensions so make sure you include a using System.Reflection;
Try it online
I am creating a reusable library using .NET Core (targeting .NETStandard 1.4) and I am using Entity Framework Core (and new to both). I have an entity class that looks like:
public class Campaign
{
[Key]
public Guid Id { get; set; }
[Required]
[MaxLength(50)]
public string Name { get; set; }
public JObject ExtendedData { get; set; }
}
and I have a DbContext class that defines the DbSet:
public DbSet<Campaign> Campaigns { get; set; }
(I am also using the Repository pattern with DI, but I don't think that is relevant.)
My unit tests give me this error:
System.InvalidOperationException: Unable to determine the relationship
represented by navigation property 'JToken.Parent' of type
'JContainer'. Either manually configure the relationship, or ignore
this property from the model..
Is there a way to indicate that this is not a relationship but should be stored as a big string?
Going to answer this one differently.
Ideally the domain model should have no idea how data is stored. Adding backing fields and extra [NotMapped] properties is actually coupling your domain model to your infrastructure.
Remember - your domain is king, and not the database. The database is just being used to store parts of your domain.
Instead you can use EF Core's HasConversion() method on the EntityTypeBuilder object to convert between your type and JSON.
Given these 2 domain models:
public class Person
{
public int Id { get; set; }
[Required]
[MaxLength(50)]
public string FirstName { get; set; }
[Required]
[MaxLength(50)]
public string LastName { get; set; }
[Required]
public DateTime DateOfBirth { get; set; }
public IList<Address> Addresses { get; set; }
}
public class Address
{
public string Type { get; set; }
public string Company { get; set; }
public string Number { get; set; }
public string Street { get; set; }
public string City { get; set; }
}
I have only added attributes that the domain is interested in - and not details that the DB would be interested in; I.E there is no [Key].
My DbContext has the following IEntityTypeConfiguration for the Person:
public class PersonsConfiguration : IEntityTypeConfiguration<Person>
{
public void Configure(EntityTypeBuilder<Person> builder)
{
// This Converter will perform the conversion to and from Json to the desired type
builder.Property(e => e.Addresses).HasConversion(
v => JsonConvert.SerializeObject(v, new JsonSerializerSettings { NullValueHandling = NullValueHandling.Ignore }),
v => JsonConvert.DeserializeObject<IList<Address>>(v, new JsonSerializerSettings { NullValueHandling = NullValueHandling.Ignore }));
}
}
With this method you can completely decouple your domain from your infrastructure. No need for all the backing field & extra properties.
The key to making the the Change Tracker function correctly is to implement a ValueComparer as well as a ValueConverter. Below is an extension to implement such:
public static class ValueConversionExtensions
{
public static PropertyBuilder<T> HasJsonConversion<T>(this PropertyBuilder<T> propertyBuilder) where T : class, new()
{
ValueConverter<T, string> converter = new ValueConverter<T, string>
(
v => JsonConvert.SerializeObject(v),
v => JsonConvert.DeserializeObject<T>(v) ?? new T()
);
ValueComparer<T> comparer = new ValueComparer<T>
(
(l, r) => JsonConvert.SerializeObject(l) == JsonConvert.SerializeObject(r),
v => v == null ? 0 : JsonConvert.SerializeObject(v).GetHashCode(),
v => JsonConvert.DeserializeObject<T>(JsonConvert.SerializeObject(v))
);
propertyBuilder.HasConversion(converter);
propertyBuilder.Metadata.SetValueConverter(converter);
propertyBuilder.Metadata.SetValueComparer(comparer);
propertyBuilder.HasColumnType("jsonb");
return propertyBuilder;
}
}
Example of how this works.
public class Person
{
public int Id { get; set; }
[Required]
[MaxLength(50)]
public string FirstName { get; set; }
[Required]
[MaxLength(50)]
public string LastName { get; set; }
[Required]
public DateTime DateOfBirth { get; set; }
public List<Address> Addresses { get; set; }
}
public class Address
{
public string Type { get; set; }
public string Company { get; set; }
public string Number { get; set; }
public string Street { get; set; }
public string City { get; set; }
}
public class PersonsConfiguration : IEntityTypeConfiguration<Person>
{
public void Configure(EntityTypeBuilder<Person> builder)
{
// This Converter will perform the conversion to and from Json to the desired type
builder.Property(e => e.Addresses).HasJsonConversion<IList<Address>>();
}
}
This will make the ChangeTracker function correctly.
#Michael's answer got me on track but I implemented it a little differently. I ended up storing the value as a string in a private property and using it as a "Backing Field". The ExtendedData property then converted JObject to a string on set and vice versa on get:
public class Campaign
{
// https://learn.microsoft.com/en-us/ef/core/modeling/backing-field
private string _extendedData;
[Key]
public Guid Id { get; set; }
[Required]
[MaxLength(50)]
public string Name { get; set; }
[NotMapped]
public JObject ExtendedData
{
get
{
return JsonConvert.DeserializeObject<JObject>(string.IsNullOrEmpty(_extendedData) ? "{}" : _extendedData);
}
set
{
_extendedData = value.ToString();
}
}
}
To set _extendedData as a backing field, I added this to my context:
protected override void OnModelCreating(ModelBuilder modelBuilder)
{
modelBuilder.Entity<Campaign>()
.Property<string>("ExtendedDataStr")
.HasField("_extendedData");
}
Update: Darren's answer to use EF Core Value Conversions (new to EF Core 2.1 - which didn't exist at the time of this answer) seems to be the best way to go at this point.
For those using EF 2.1 there is a nice little NuGet package EfCoreJsonValueConverter that makes it pretty simple.
using Innofactor.EfCoreJsonValueConverter;
using Microsoft.EntityFrameworkCore;
using Microsoft.EntityFrameworkCore.Metadata.Builders;
public class Campaign
{
[Key]
public Guid Id { get; set; }
[Required]
[MaxLength(50)]
public string Name { get; set; }
public JObject ExtendedData { get; set; }
}
public class CampaignConfiguration : IEntityTypeConfiguration<Campaign>
{
public void Configure(EntityTypeBuilder<Campaign> builder)
{
builder
.Property(application => application.ExtendedData)
.HasJsonValueConversion();
}
}
I have made a solution based on Robert Raboud's contribution. The change made by me is that my implementation uses a HasJsonConversion method that depends on the System.Text.Json package rather than Newtonsofts library:
public static PropertyBuilder<T> HasJsonConversion<T>(this PropertyBuilder<T> propertyBuilder) where T : class, new()
{
var options = new JsonSerializerOptions
{
PropertyNamingPolicy = JsonNamingPolicy.CamelCase,
WriteIndented = true,
AllowTrailingCommas = true,
PropertyNameCaseInsensitive = true
};
ValueConverter<T, string> converter = new ValueConverter<T, string>
(
v => JsonSerializer.Serialize(v, options),
v => JsonSerializer.Deserialize<T>(v, options) ?? new T()
);
ValueComparer<T> comparer = new ValueComparer<T>
(
(l, r) => JsonSerializer.Serialize(l, options) == JsonSerializer.Serialize(r, options),
v => v == null ? 0 : JsonSerializer.Serialize(v, options).GetHashCode(),
v => JsonSerializer.Deserialize<T>(JsonSerializer.Serialize(v, options), options)
);
propertyBuilder.HasConversion(converter);
propertyBuilder.Metadata.SetValueConverter(converter);
propertyBuilder.Metadata.SetValueComparer(comparer);
propertyBuilder.HasColumnType("LONGTEXT");
return propertyBuilder;
}
Note also that this implementation expects for the column to be LONGTEXT since I am using a MySQL setup.
Here's something I used
Model
public class FacilityModel
{
public string Name { get; set; }
public JObject Values { get; set; }
}
Entity
[Table("facility", Schema = "public")]
public class Facility
{
public string Name { get; set; }
public Dictionary<string, string> Values { get; set; } = new Dictionary<string, string>();
}
Mapping
this.CreateMap<Facility, FacilityModel>().ReverseMap();
DBContext
base.OnModelCreating(builder);
builder.Entity<Facility>()
.Property(b => b.Values)
.HasColumnType("jsonb")
.HasConversion(
v => JsonConvert.SerializeObject(v),
v => JsonConvert.DeserializeObject<Dictionary<string, string>>(v));
Could you try something like this?
[NotMapped]
private JObject extraData;
[NotMapped]
public JObject ExtraData
{
get { return extraData; }
set { extraData = value; }
}
[Column("ExtraData")]
public string ExtraDataStr
{
get
{
return this.extraData.ToString();
}
set
{
this.extraData = JsonConvert.DeserializeObject<JObject>(value);
}
}
here is the migration output:
ExtraData = table.Column<string>(nullable: true),
For developers, who work with EF Core 3.1 and meet such error ("The entity type 'XXX' requires a primary key to be defined. If you intended to use a keyless entity type call 'HasNoKey()'.") the solution is:
Move .HasConversion() method with it's lambda from:
OrderConfiguration : IEntityTypeConfiguration<T> to
OnModelCreating(ModelBuilder modelBuilder) in your DataContext.
// DbContext
protected override void OnModelCreating(ModelBuilder modelBuilder)
{
var entityTypes = modelBuilder.Model.GetEntityTypes();
foreach (var entityType in entityTypes)
{
foreach (var property in entityType.ClrType.GetProperties().Where(x => x != null && x.GetCustomAttribute<HasJsonConversionAttribute>() != null))
{
modelBuilder.Entity(entityType.ClrType)
.Property(property.PropertyType, property.Name)
.HasJsonConversion();
}
}
base.OnModelCreating(modelBuilder);
}
Create an attribute to handle the properties of the entities.
public class HasJsonConversionAttribute : System.Attribute
{
}
Create extention class to find Josn properties
public static class ValueConversionExtensions
{
public static PropertyBuilder HasJsonConversion(this PropertyBuilder propertyBuilder)
{
ParameterExpression parameter1 = Expression.Parameter(propertyBuilder.Metadata.ClrType, "v");
MethodInfo methodInfo1 = typeof(Newtonsoft.Json.JsonConvert).GetMethod("SerializeObject", types: new Type[] { typeof(object) });
MethodCallExpression expression1 = Expression.Call(methodInfo1 ?? throw new Exception("Method not found"), parameter1);
ParameterExpression parameter2 = Expression.Parameter(typeof(string), "v");
MethodInfo methodInfo2 = typeof(Newtonsoft.Json.JsonConvert).GetMethod("DeserializeObject", 1, BindingFlags.Static | BindingFlags.Public, Type.DefaultBinder, CallingConventions.Any, types: new Type[] { typeof(string) }, null)?.MakeGenericMethod(propertyBuilder.Metadata.ClrType) ?? throw new Exception("Method not found");
MethodCallExpression expression2 = Expression.Call(methodInfo2, parameter2);
var converter = Activator.CreateInstance(typeof(ValueConverter<,>).MakeGenericType(typeof(List<AttributeValue>), typeof(string)), new object[]
{
Expression.Lambda( expression1,parameter1),
Expression.Lambda( expression2,parameter2),
(ConverterMappingHints) null
});
propertyBuilder.HasConversion(converter as ValueConverter);
return propertyBuilder;
}
}
Entity example
public class Attribute
{
[Key, DatabaseGenerated(DatabaseGeneratedOption.Identity)]
public int Id { get; set; }
public string Name { get; set; }
[HasJsonConversion]
public List<AttributeValue> Values { get; set; }
}
public class AttributeValue
{
public string Value { get; set; }
public IList<AttributeValueTranslation> Translations { get; set; }
}
public class AttributeValueTranslation
{
public string Translation { get; set; }
public string CultureName { get; set; }
}
Download Source
For those who are working on entity framework core 5.0 and above. below can work if you are getting error like below
The entity type '{EntityName}' requires a primary key to be defined. If you intended to use a keyless entity type, call 'HasNoKey' in 'OnModelCreating'. The error is asking to define a primary key on the model
Try this
protected override void OnModelCreating(ModelBuilder modelBuilder)
{
modelBuilder
.Entity<Person>(
eb =>
{
eb.Property(p => p.Addresses).HasConversion(
v => JsonConvert.SerializeObject(v, new JsonSerializerSettings { NullValueHandling = NullValueHandling.Ignore }),
v => JsonConvert.DeserializeObject<IList<Address>>(v, new JsonSerializerSettings { NullValueHandling = NullValueHandling.Ignore })
);
});
}
----------------- 2022 Update -----------------
Hey there,
Just sharing an update from Dec/2022.
Recently, EF Core 7.0 was launched, containing one of the most awaited features called JSON Columns.
This new feature allows us for mapping aggregates (written from .NET types) into JSON documents.
Just remember that, In EF Core, aggregate types are defined using Owned Entity Types.
https://learn.microsoft.com/en-us/ef/core/modeling/owned-entities
Let's consider this scenario: A object called "LogDetail", that needs to be stored into a single column as a JSON, inside another object (or table) called "Log".
public class Log : BaseEntity
{
public string TraceID { get; set; } = string.Empty;
public string Code { get; set; } = string.Empty;
public LogDetail LogDetail { get; set; } = null!;
public string IpAddress { get; set; } = string.Empty;
}
public class LogDetail
{
public string InnerException { get; set; } = null!;
public string MemberMap { get; set; } = null!;
public string Message { get; set; } = null!;
public string Source { get; set; } = null!;
public string StackTrace { get; set; } = null!;
public string TypeMap { get; set; } = null!;
public string Path { get; set; } = null!;
}
By the end, all that you need is to configure the behavior on your configuration map class:
public sealed class LogMap : IEntityTypeConfiguration<Log>
{
public override void Configure(EntityTypeBuilder<Log> builder)
{
/* Owned Type Configurations */
builder.OwnsOne(e => e.LogDetail, options =>
{
options.ToJson("LOG_DETAIL");
});
}
}
The comment by #Métoule:
Be careful with this approach: EF Core marks an entity as modified only if the field is assigned to. So if you use person.Addresses.Add, the entity won't be flagged as updated; you'll need to call the property setter person.Addresses = updatedAddresses.
made me take a different approach so that this fact is obvious: use Getter and Setter methods, rather than a property.
public void SetExtendedData(JObject extendedData) {
ExtendedData = JsonConvert.SerializeObject(extendedData);
_deserializedExtendedData = extendedData;
}
//just to prevent deserializing more than once unnecessarily
private JObject _deserializedExtendedData;
public JObject GetExtendedData() {
if (_extendedData != null) return _deserializedExtendedData;
_deserializedExtendedData = string.IsNullOrEmpty(ExtendedData) ? null : JsonConvert.DeserializeObject<JObject>(ExtendedData);
return _deserializedExtendedData;
}
You could theoretically do this:
campaign.GetExtendedData().Add(something);
But it's much more clear that That Doesn't Do What You Think It Does™.
If you're using database-first and using some kind of class auto-generator for EF, then the classes will usually be declared as partial, so you can add this stuff in a separate file that won't get blown away the next time you update your classes from your database.
Say I have this simple entity:
public class Person
{
public int PersonID { get; set; }
public string Name { get; set; }
}
The entity framework can infer by convention that the PersonID field is the primary key. However, if I give the model builder this class:
public class PersonCfg : EntityTypeConfiguration<Person>
{
public PersonCfg()
{
ToTable("Person", "Person");
HasKey(p => p.PersonID);
}
}
Would that improve startup performance? My thought was it might allow EF to do less reflecting and startup the app faster but I don't know how it works behind the scenes to know if it has any impact.
To test this, you can use the DbModelBuilder class to build the model yourself and track the speed of the "Compile" step.
Here's my example code (LinqPad script):
void Main()
{
// Initialize the overall system, but don't count the result.
BuildC();
DateTime startDateA = DateTime.Now;
BuildA();
DateTime.Now.Subtract(startDateA).TotalMilliseconds.Dump("A");
DateTime startDateB = DateTime.Now;
BuildB();
DateTime.Now.Subtract(startDateB).TotalMilliseconds.Dump("B");
}
public class PersonA
{
public int PersonAId { get; set; }
public string Name { get; set; }
}
private void BuildA()
{
var builder = new DbModelBuilder();
builder.Entity<PersonA>();
var model = builder.Build(new DbProviderInfo("System.Data.SqlClient", "2008"));
model.Compile();
}
public class PersonB
{
public int PersonBId { get; set; }
public string Name { get; set; }
}
private void BuildB()
{
var builder = new DbModelBuilder();
builder.Conventions.Remove<IdKeyDiscoveryConvention>();
builder.Entity<PersonB>()
.HasKey(p => p.PersonBId);
var model = builder.Build(new DbProviderInfo("System.Data.SqlClient", "2008"));
model.Compile();
}
public class PersonC
{
public int PersonCId { get; set; }
public string Name { get; set; }
}
private void BuildC()
{
var builder = new DbModelBuilder();
builder.Entity<PersonC>()
.HasKey(p => p.PersonCId);
var model = builder.Build(new DbProviderInfo("System.Data.SqlClient", "2008"));
model.Compile();
}
I get the result of 2.0004ms to 2.0009ms. Curiously, removing conventions made the operation take longer.
This is probably a simple answer, but i'm just starting to get the hand of AutoMapper.
Anyway, i have a domain object like this:
public class User
{
public string Name { get; set; }
public FacebookUser FacebookUser { get; set; }
}
And a ViewModel like this:
public class UserViewModel
{
public string Name { get; set; }
public long FacebookUniqueId { get; set; }
}
Here's what i have in my AutoMapper configuration:
Mapper.CreateMap<User,UserViewModel>()
.ForMember(dest => dest.FacebookUniqueId, opt => opt.MapFrom(src => src.FacebookUser.FacebookUniqueId))
But it throws an exception when the FacebookUser object is null, which is to be expected.
How do i tell AutoMapper:
Map UserViewModel.FacebookUniqueId to User.FacebookUser.FacebookUniqueId, except for when it's null, then use 0.
Any ideas?
opt => opt.MapFrom(src => src.FacebookUser == null ? 0 : src.FacebookUser.FacebookUniqueId)
Yikes, if I had just checked if the first idea worked, thought the MapFrom method only took an expression that pointed to a property to resolve it that way...
The code you provided works on the latest Automapper. I investigated the sources a little and found the following code:
try
{
var result = _method((TSource)source.Value);
return source.New(result, MemberType);
}
catch (NullReferenceException)
{
return source.New(null, MemberType);
}
As you can see now Automapper catches all the Null exceptions while resolving chain mappings. The following test is green:
[TestFixture]
public class AutomapperChainingMappingTest
{
[Test]
public void ChainMapping_NullProperty_DefaultValueSet()
{
AutoMapper.Mapper.CreateMap<User, UserViewModel>()
.ForMember(x => x.FacebookUniqueId, o => o.MapFrom(x => x.FacebookUser.FacebookUniqueId));
var source = new User();
var model = AutoMapper.Mapper.Map<UserViewModel>(source);
Assert.AreEqual(model.FacebookUniqueId, default(long));
}
}
public class User
{
public FacebookUser FacebookUser { get; set; }
}
public class UserViewModel
{
public long FacebookUniqueId { get; set; }
}
public class FacebookUser
{
public long? FacebookUniqueId { get; set; }
}