I have implemented the custom contract resolver to remove the some properties from serialized Json. Most of the time below code works but occasionally this code fails. Sometime it loops through the list and skips item I want to exclude which causing to appear that properties which I wanted to exclude.
class TestClass
{
public static void Test()
{
var objectToSerialise = //Coming from some WebAPI calls
string[] stringToSkip = {"skip1", "skip2"};
var settings = new JsonSerializerSettings
{
ContractResolver = new MyContractResolver(stringToSkip)
};
var json = JsonConvert.SerializeObject(objectToSerialise, settings);
}
}
public class MyContractResolver : DefaultContractResolver {
private string[] _skipthis;
public MyContractResolver(string[] skipthis)
{
_skipthis = skipthis;
}
private JsonProperty CreateProperty(MemberInfo member, MemberSerialization memberSerialization){
var property = base.CreateProperty(member, memberSerialization);
property.ShouldSerialize = (prop) =>{
return !(_skipthis.Where(n => property.PropertyName.Contains(n)).Any());
};
return property;
}
}
Can someone please suggest why this code is failing silently intermittently with out throwing any kind of exception?
Note that skipthis array is not modified outside this. I want to skip the property when property name exist or substring of propertyname exist in skipthis array.
Related
I have an .NET Core Web Api 2.1 app in which I only serialize the properties requested by the client.
Example: GET orders/1?select=Id,TotalAmount
Example: GET orders/1?select=Id,CustomerName,DeliveryAddress,Location,ZipCode
For that, the app creates an object in every request (Lifetime Scoped) to get the requested properties.
Then, I created Custom ContractResolver class that is instanciated in every request by a global IResultFilter:
public class SerializationFilter : IResultFilter
{
private readonly IApiQueryParameters _apiQueryParameters;
public SerializationFilter(IApiQueryParameters apiQueryParameters)
{
this._apiQueryParameters = apiQueryParameters;
}
public void OnResultExecuting(ResultExecutingContext context)
{
var objectResult = context.Result as ObjectResult;
if (objectResult != null)
{
var contractResolver = new SelectiveResourceContractResolver(this._apiQueryParameters);
var serializerSettings = new JsonSerializerSettings
{
ContractResolver = contractResolver
};
var jsonFormatter = new JsonOutputFormatter(
serializerSettings,
ArrayPool<char>.Shared);
objectResult.Formatters.Add(jsonFormatter);
}
}
}
Now, this works for the first request made after the app is online.
The second request creates the ContractResolver correctly, but the response is not correct. The returned serialized properties are not the ones requested by the client.
Debugging the code, I noticed that when the SelectiveResourceContractResolver is created, it access the constructor method right,
public SelectiveResourceContractResolver(IApiQueryParameters apiQueryParameters)
{
this._apiQueryParameters = apiQueryParameters;
}
Also, it runs the CreateProperty overriden method:
protected override JsonProperty CreateProperty(MemberInfo member, MemberSerialization memberSerialization)
But in the second request, the CreatePropertyMethod is not run.
protected override JsonProperty CreateProperty(MemberInfo member, MemberSerialization memberSerialization)
{
var property = base.CreateProperty(member, memberSerialization);
property.ShouldSerialize =
instance =>
{
...
};
return property;
}
The second time on, it justs executes the property.ShouldSerialize = instance => part. I think some sort of caching is done by the app, but the wierd thing is that the ContractResolver is created in every request (it´s not set globally in Startup.cs) when the filter is executed. Also, when I debug it in the second request, the property IApiQueryParameters in the ContractResolver has all the values of the first request and not from the second execution of the constructor.
What I´m missing here?
I'm assuming your SelectiveResourceContractResolver inherits from DefaultContractResolver.
DefaultContractResolver.ResolveContract, which is called internally during serialization, caches contracts in a static, thread-safe dictionary, so CreateContract (which in turn calls CreateProperty) will only be called once per type for as long as your WebAPI process runs.
This is because calling CreateContract is expensive, and not caching contracts will seriously impact serialisation performance.
If you only need to cache contracts per response, you can create your own instance cache in SelectiveResourceContractResolver and override ResolveContract to use your cache instead:
public class InstanceCachingContractResolver : DefaultContractResolver
{
private readonly Dictionary<Type, JsonContract> contractCache = new Dictionary<Type, JsonContract>();
public override JsonContract ResolveContract(Type type)
{
if (!contractCache.TryGetValue(type, out JsonContract contract))
{
contract = CreateContract(type);
contractCache.Add(type, contract);
}
return contract;
}
}
I am exactly in the same case that this question:
How do I make JSON.NET ignore object relationships?
I see the proposed solution and I know I must use a Contract Revolver, and I also see the code of the Contract Resolver, but I do not know how to use it.
Should I use it in the WebApiConfig.vb?
Should I modify my Entity Model anyway?
It is a useful question👍 and I hope this help:
A)
If you have created your models manually (without Entity Framework), mark the relation properties as virtual first.
If your models were created by EF, It has already done it for you and each Relation Property is marked as virtual, as seen below:
Sample class:
public class PC
{
public int FileFolderId {get;set;}
public virtual ICollection<string> Libs { get; set; }
public virtual ICollection<string> Books { get; set; }
public virtual ICollection<string> Files { get; set; }
}
B)
Those relation properties can now be ignored by the JSON serializer by using the following ContractResolver for JSON.NET:
CustomResolver:
class CustomResolver : DefaultContractResolver
{
private readonly List<string> _namesOfVirtualPropsToKeep=new List<string>(new String[]{});
public CustomResolver(){}
public CustomResolver(IEnumerable<string> namesOfVirtualPropsToKeep)
{
this._namesOfVirtualPropsToKeep = namesOfVirtualPropsToKeep.Select(x=>x.ToLower()).ToList();
}
protected override JsonProperty CreateProperty(MemberInfo member, MemberSerialization memberSerialization)
{
JsonProperty prop = base.CreateProperty(member, memberSerialization);
var propInfo = member as PropertyInfo;
if (propInfo != null)
{
if (propInfo.GetMethod.IsVirtual && !propInfo.GetMethod.IsFinal
&& !_namesOfVirtualPropsToKeep.Contains(propInfo.Name.ToLower()))
{
prop.ShouldSerialize = obj => false;
}
}
return prop;
}
}
C)
Finally, to serialize your model easily use the above ContractResolver. Set it up like this:
// -------------------------------------------------------------------
// Serializer settings
JsonSerializerSettings settings = new JsonSerializerSettings
{
// ContractResolver = new CustomResolver();
// OR:
ContractResolver = new CustomResolver(new []
{
nameof(PC.Libs), // keep Libs property among virtual properties
nameof(PC.Files) // keep Files property among virtual properties
}),
PreserveReferencesHandling = PreserveReferencesHandling.None,
ReferenceLoopHandling = ReferenceLoopHandling.Ignore,
Formatting = Formatting.Indented
};
// -------------------------------------------------------------------
// Do the serialization and output to the console
var json = JsonConvert.SerializeObject(new PC(), settings);
Console.WriteLine(json);
// -------------------------------------------------------------------
// We can see that "Books" filed is ignored in the output:
// {
// "FileFolderId": 0,
// "Libs": null,
// "Files": null
// }
Now, all the navigation (relation) properties [virtual properties] will be ignored automatically except you keep some of them by determine them in your code.😎
Live DEMO
Thanks from #BrianRogers for his answer here.
If you are using Newtonsoft.Json
Mark field with
Newtonsoft.Json.JsonIgnore
Instead of
System.Text.Json.Serialization.JsonIgnore
I am using Json.net in my MVC 4 program.
I have an object item of class Item.
I did:
string j = JsonConvert.SerializeObject(item);
Now I want to add an extra property, like "feeClass" : "A" into j.
How can I use Json.net to achieve this?
You have a few options.
The easiest way, as #Manvik suggested, is simply to add another property to your class and set its value prior to serializing.
If you don't want to do that, the next easiest way is to load your object into a JObject, append the new property value, then write out the JSON from there. Here is a simple example:
class Item
{
public int ID { get; set; }
public string Name { get; set; }
}
class Program
{
static void Main(string[] args)
{
Item item = new Item { ID = 1234, Name = "FooBar" };
JObject jo = JObject.FromObject(item);
jo.Add("feeClass", "A");
string json = jo.ToString();
Console.WriteLine(json);
}
}
Here is the output of the above:
{
"ID": 1234,
"Name": "FooBar",
"feeClass": "A"
}
Another possibility is to create a custom JsonConverter for your Item class and use that during serialization. A JsonConverter allows you to have complete control over what gets written during the serialization process for a particular class. You can add properties, suppress properties, or even write out a different structure if you want. For this particular situation, I think it is probably overkill, but it is another option.
Following is the cleanest way I could implement this
dynamic obj = JsonConvert.DeserializeObject(jsonstring);
obj.NewProperty = "value";
var payload = JsonConvert.SerializeObject(obj);
You could use ExpandoObject.
Deserialize to that, add your property, and serialize back.
Pseudocode:
Expando obj = JsonConvert.Deserializeobject<Expando>(jsonstring);
obj.AddeProp = "somevalue";
string addedPropString = JsonConvert.Serializeobject(obj);
I think the most efficient way to serialize a property that doesn't exist in the type is to use a custom contract resolver. This avoids littering your class with the property you don't want, and also avoids the performance hit of the extra serialization round trip that most of the other options on this page incur.
public class SpecialItemContractResolver : DefaultContractResolver {
protected override IList<JsonProperty> CreateProperties(Type type, MemberSerialization memberSerialization) {
var list = base.CreateProperties(type, memberSerialization);
if (type.Equals(typeof(Item))) {
var feeClassProperty = CreateFeeClassProperty();
list.Add(feeClassProperty);
}
return list;
}
private JsonProperty CreateFeeClassProperty() {
return new JsonProperty {
PropertyName = "feeClass",
PropertyType = typeof(string),
DeclaringType = typeof(Item),
ValueProvider = new FeeClassValueProvider(),
AttributeProvider = null,
Readable = true,
Writable = false,
ShouldSerialize = _ => true
};
}
private class FeeClassValueProvider : IValueProvider {
public object GetValue(object target) => "A";
public void SetValue(object target, object value) { }
}
}
To use this functionality:
// This could be put in a static readonly place so it's reused
var serializerSettings = new JsonSerializerSettings {
ContractResolver = new SpecialItemContractResolver()
};
// And then to serialize:
var item = new Item();
var json = JsonConvert.Serialize(item, serializerSettings);
I have this first version of a class
public class GenerateAuthorisationWorkflows : IGenerateAuthorisationWorkflows
{
public IList<Guid> FromDtaObjects(IList<DtaObject> dtaObjects, Employee requestingEmployee)
{
foreach (var dtaObject in dtaObjects) { }
return new List<Guid>();
}
public IList<Guid> FromDtaObjects()
{
return new List<Guid>();
}
}
And the MSpec tests for it
public abstract class specification_for_generate_workflows : Specification<GenerateAuthorisationWorkflows>
{
protected static IWorkflowService workflowService;
Establish context = () => { workflowService = DependencyOf<IWorkflowService>(); };
}
[Subject(typeof(GenerateAuthorisationWorkflows))]
public class when_generate_workflows_is_called_with_a_dta_object_list_and_an_employee : specification_for_generate_workflows
{
static IList<Guid> result;
static IList<DtaObject> dtaObjects;
static Employee requestingEmployee;
Establish context = () =>
{
var mocks = new MockRepository();
var stubDtaObject1 = mocks.Stub<DtaObject>();
var stubDtaObject2 = mocks.Stub<DtaObject>();
var dtaObjectEnum = new List<DtaObject>{stubDtaObject1,stubDtaObject2}.GetEnumerator();
dtaObjects = mocks.Stub<IList<DtaObject>>();
dtaObjects.Stub(x => x.GetEnumerator()).Return(dtaObjectEnum).WhenCalled(x => x.ReturnValue = dtaObjectEnum);
requestingEmployee = mocks.Stub<Employee>();
mocks.ReplayAll();
};
Because of = () => result = subject.FromDtaObjects(dtaObjects, requestingEmployee);
It should_enumerate_the_dta_objects = () => dtaObjects.received(x=> x.GetEnumerator());
It should_call_workflow_host_helper = () => workflowService.AssertWasCalled(x => x.StartWorkflow());
}
With this configuration, my first test passes and my second test fails, as expected. I added a constructor to the class to inject the IWorkflowService.
public class GenerateAuthorisationWorkflows : IGenerateAuthorisationWorkflows
{
IWorkflowService _workflowService;
GenerateAuthorisationWorkflows(IWorkflowService workflowService)
{
_workflowService = workflowService;
}
public IList<Guid> FromDtaObjects(IList<DtaObject> dtaObjects, Employee requestingEmployee)
{
foreach (var dtaObject in dtaObjects)
{
Guid workflowKey = _workflowService.StartWorkflow();
}
return new List<Guid>();
}
public IList<Guid> FromDtaObjects()
{
return new List<Guid>();
}
}
Now, when I run the tests, they fail at the Because:
System.InvalidOperationException: Sequence contains no elements
at System.Linq.Enumerable.First(IEnumerable`1 source)
at MSpecTests.EmployeeRequestSystem.Tasks.Workflows.when_generate_workflows_is_called_with_a_dta_object_list_and_an_employee.<.ctor>b__4() in GenerateAuthorisationWorkflowsSpecs.cs: line 76
For clarity, line 76 above is:
Because of = () => result = subject.FromDtaObjects(dtaObjects, requestingEmployee);
I've tried tracing the problem but am having no luck. I have tried setting up a constructor that takes no arguments but it raises the same error. I have similar classes with IoC dependencies that work fine using MSpec/Rhino Mocks, where am I going wrong?
Castle Windsor requires a public constructor to instantiate a class. Adding public to the constructor allows correct operation.
public class GenerateAuthorisationWorkflows : IGenerateAuthorisationWorkflows
{
IWorkflowService _workflowService;
public GenerateAuthorisationWorkflows(IWorkflowService workflowService)
{
_workflowService = workflowService;
}
public IList<Guid> FromDtaObjects(IList<DtaObject> dtaObjects, Employee requestingEmployee)
{
foreach (var dtaObject in dtaObjects)
{
Guid workflowKey = _workflowService.StartWorkflow();
}
return new List<Guid>();
}
public IList<Guid> FromDtaObjects()
{
return new List<Guid>();
}
}
Rowan, looks like you answered your own question. It's good practice to explicitly state the access modifiers! By default, C# chooses private. These kinds of errors are easy to miss!
I can also see that your Establish block is too complicated. You're testing the implementation details and not the behavior. For example, you are
stubbing the GetEnumerator call that's implicitly made inside the foreach loop.
asserting that the workflow service was called only once
mixing MSpec automocking and your own local mocks
You're not actually testing that you got a GUID for every object in the input list. If I were you, I'd test the behavior like this...
public class GenerateAuthorisationWorkflows : IGenerateAuthorisationWorkflows
{
private readonly IWorkflowService _service;
public GenerateAuthorisationWorkflows(IWorkflowService service)
{
_service = service;
}
public List<Guid> FromDtaObjects(List<DtaObject> input, Employee requestor)
{
// I assume that the workflow service generates a new key
// per input object. So, let's pretend the call looks like
// this. Also using some LINQ to avoid the foreach or
// building up a local list.
input.Select(x => _service.StartWorkflow(requestor, x)).ToList();
}
}
[Subject(typeof(GenerateAuthorisationWorkflows))]
public class When_generating_authorisation_keys_for_this_input
: Specification<GenerateAuthorisationWorkflows>
{
private static IWorkflowService _service;
private static Employee _requestor = new Employee();
private static List<DtaObject> _input = new List<DtaObject>()
{
new DtaObject(),
new DtaObject(),
};
private static List<Guid> _expected = new List<Guid>()
{
Guid.NewGuid(),
Guid.NewGuid(),
};
private static List<Guid> _actual = new List<Guid>();
Establish context = () =>
{
// LINQ that takes each item from each list in a pair. So
// the service is stubbed to return a specific GUID per
// input DtaObject.
_input.Zip(_expected, (input, output) =>
{
DependencyOf<IWorkflowService>().Stub(x => x.StartWorkflow(_requestor, input)).Return(output);
});
};
Because of = () => _actual = Subject.FromDtaObjects(_input, _requestor);
// This should be an MSpec collection assertion that
// ensures that the contents of the collections are
// equivalent
It should_get_a_unique_key_per_input = _actual.ShouldEqual(_expected);
}
Having problem reading bytearray of custom objects. Any help is appreciated
public class CustomObject extends Object {
public function CustomObject() {
public var _x:Number = 100
public var _y:Number = 10
public var _z:Number = 60
}
}
var cObj:CustomObject = new CustomObject()
var bytes:ByteArray = new ByteArray()
bytes.writeObject(cObj)
bytes.compress()
//read
try { bytes.uncompress() } catch (e:Error) { }
var obj:CustomObject = bytes.readObject() as CustomObject
trace(obj) // null why?!
trace(obj._z) // Obviously - TypeError: Error #1009: Cannot access a property or method of a null object reference.
What you want to do is use the registerClassAlias method to register type information along with the data. That way Flash will know how to serialize/deserialize your object. Here's some sample code from Adobe's documentation:
registerClassAlias("com.example.eg", ExampleClass);
var eg1:ExampleClass = new ExampleClass();
var ba:ByteArray = new ByteArray();
ba.writeObject(eg1);
ba.position = 0;
var eg2:* = ba.readObject();
trace(eg2 is ExampleClass); // true
It should be noted that all types that should be serialized must be registered for the type information to be saved. So if you have another type that is referenced by your type, it too must be registered.
Your CustomObject class is wrong , it should throw an error actually , it should be this instead
public class CustomObject
{
public var _x:Number = 100
public var _y:Number = 10
public var _z:Number = 60
public function CustomObject()
{
}
}
Edit:
Sounds like macke has a point, because this works...
//read
try { bytes.uncompress() } catch (e:Error) { }
var obj:Object = bytes.readObject();
trace(obj) // [object Object]
trace(obj._z) // 60
Look at object that ByteArray.readObject() returns. You'll probably see that all properties are there, but type information is lost. So, you can solve this by creating some
public static function fromObject(value:Object):CustomObject {
var result:CustomObject = new CustomObject();
result._x = value._x;
//and so on...
return result;
}
To serialize custom classes to the ByteArray, you must put registerClassAlias in the constructor of the class calling the byteArray.writeObject() function.
If you don't, your custom class will be serialized as Object type. I was calling registerClassAlias in the serialize function below and my custom class keeps getting serialized as Object until I moved the registerClassAlias to the constructor.
public class MyClass{
public function MyClass(){
registerClassAlias("com.myclass", MyClass); // Ok, serializes as MyClass
serialize( new MyClass() );
}
private function serialize( _c:MyClass ){
var byteArray:ByteArray = new ByteArray();
byteArray.writeObject( _c );
//registerClassAlias("com.myclass", MyClass); Not ok, serialized as Object
EncryptedLocalStorage.setItem('key', byteArray);
}
}