I need to read an JSON object from TypeScript, which has a variable names prop which of two types, either Identifier or Expression to C#. TypeScript can have a variable with multiple types (with its union type feature), e.g., prop is defined as
var property:Identifier | Expression
I am reading the JSON object string from C# with JsonConvert.DeserializeObject, e.g.,
Object facebookFriends = new JavaScriptSerializer().Deserialize<Object>(JSON_Object);
How should I declare a variable of two types in C#? Or should I just use object (i.e., declare as Object property)?
If the property can have either of these two formats:
{ "prop": "identifier" }
{ "prop": { "complex": "object" } }
ie. either a simple string or a more complex object you can solve this by using dynamic to accept the parsed prop value, and extra properties to return either the identifier or the complex object. Here is a LINQPad program that demonstrates:
void Main()
{
var jsonStrings = new[]
{
"{ \"prop\": \"identifier\" }",
"{ \"prop\": { \"complex\": \"object\" } }"
};
jsonStrings.Select(json => JsonConvert.DeserializeObject<Test>(json)).Dump();
}
public class Test
{
public Test()
{
_Complex = new Lazy<Complex>(GetComplex);
}
[JsonProperty("prop")]
public dynamic prop { get; set; }
[JsonIgnore]
public string Identifier => prop as string;
private readonly Lazy<Complex> _Complex;
[JsonIgnore]
public Complex Complex => _Complex.Value;
private Complex GetComplex()
{
if (!(prop is JObject))
return null;
return ((JObject)prop).ToObject<Complex>();
}
}
public class Complex
{
public string complex { get; set; }
}
Output:
A different approach would be to create your own types to handle the multiple outputs and again use a secondary property to evaluate what you actually got from the json:
void Main()
{
var jsonStrings = new[]
{
"{ \"prop\": \"identifier\" }",
"{ \"prop\": { \"complex\": \"object\" } }"
};
jsonStrings.Select(json => JsonConvert.DeserializeObject<Test>(json)).Dump();
}
public class Test
{
[JsonProperty("prop")]
public dynamic prop { get; set; }
[JsonIgnore]
public PropertyValue Property
{
get
{
if (prop is string)
return new IdentifierProperty(prop as string);
return new ExpressionProperty(((JObject)prop).ToObject<Expression>());
}
}
}
public abstract class PropertyValue
{
}
public class IdentifierProperty : PropertyValue
{
public IdentifierProperty(string identifier)
{
Identifier = identifier;
}
public string Identifier { get; }
public override string ToString() => Identifier;
}
public class ExpressionProperty : PropertyValue
{
public ExpressionProperty(Expression expression)
{
Expression = expression;
}
public Expression Expression { get; }
public override string ToString() => Expression.ToString();
}
public class Expression
{
public string complex { get; set; }
public override string ToString() => $"complex: {complex}";
}
Output:
Related
Some of my actions accept models like:
public class PaymentRequest
{
public decimal Amount { get; set; }
public bool? SaveCard { get; set; }
public int? SmsCode { get; set; }
public BankCardDetails Card { get; set; }
}
public class BankCardDetails
{
public string Number { get; set; }
public string HolderName { get; set; }
public string ExpiryDate { get; set; }
public string ValidationCode { get; set; }
}
And the action method looks like:
[HttpPost]
[Route("api/v1/payment/pay")]
public Task<BankCardActionResponse> Pay([FromBody] PaymentRequest request)
{
if (request == null)
throw new HttpResponseException(HttpStatusCode.BadRequest);
return _paymentService.PayAsync(DataUserHelper.PhoneNumber, request);
}
I use Nlog. I think it's clear this is a bad idea to log all this bank data. My log config file contained the following line:
<attribute name="user-requestBody" layout="${aspnet-request-posted-body}"/>
I logged the request. I decided to refactor that and planned the following strategy. Actions that contain sensitive data into their requests I will mark with an attribute like
[RequestMethodFormatter(typeof(PaymentRequest))]
then take a look at my custom renderer:
[LayoutRenderer("http-request")]
public class NLogHttpRequestLayoutRenderer : AspNetRequestPostedBody
{
protected override void DoAppend(StringBuilder builder, LogEventInfo logEvent)
{
base.DoAppend(builder, logEvent);
var body = builder.ToString();
// Get attribute of the called action.
var type = ... // How can I get "PaymentRequest" from the [RequestMethodFormatter(typeof(PaymentRequest))]
var res = MaskHelper.GetMaskedJsonString(body, type);
// ... and so on
}
}
I think you understand the idea. I need the type from the method's RequestMethodFormatter attribute. Is it even possible to get it into the renderer? I need it because I'm going to deserialize request JSON into particular models (it's gonna be into the MaskHelper.GetMaskedJsonString), work with the models masking the data, serialize it back into JSON.
So, did I choose a wrong approach? Or it's possible to get the type from the attribute into the renderer?
After some research, I ended up with the following solution:
namespace ConsoleApp7
{
internal class Program
{
private static void Main()
{
var sourceJson = GetSourceJson();
var userInfo = JsonConvert.DeserializeObject(sourceJson, typeof(User));
Console.WriteLine("----- Serialize without Resolver-----");
Console.WriteLine(JsonConvert.SerializeObject(userInfo));
Console.WriteLine("----- Serialize with Resolver-----");
Console.WriteLine(JsonConvert.SerializeObject(userInfo, new JsonSerializerSettings
{
ContractResolver = new MaskPropertyResolver()
}));
}
private static string GetSourceJson()
{
var guid = Guid.Parse("3e92f0c4-55dc-474b-ae21-8b3dac1a0942");
return JsonConvert.SerializeObject(new User
{
UserId = guid,
Age = 19,
Name = "John",
BirthDate = new DateTime(1990, 5, 12),
Hobbies = new[]
{
new Hobby
{
Name = "Football",
Rating = 5,
DurationYears = 3,
},
new Hobby
{
Name = "Basketball",
Rating = 7,
DurationYears = 4,
}
}
});
}
}
public class User
{
[MaskGuidValue]
public Guid UserId { get; set; }
[MaskStringValue("***")] public string Name { get; set; }
public int Age { get; set; }
[MaskDateTimeValue]
public DateTime BirthDate { get; set; }
public Hobby[] Hobbies { get; set; }
}
public class Hobby
{
[MaskStringValue("----")]
public string Name { get; set; }
[MaskIntValue(replacement: 11111)]
public int Rating { get; set; }
public int DurationYears { get; set; }
}
public class MaskPropertyResolver : DefaultContractResolver
{
protected override IList<JsonProperty> CreateProperties(Type type, MemberSerialization memberSerialization)
{
var props = base.CreateProperties(type, memberSerialization);
var allowedPropertyTypes = new Type[]
{
typeof(Guid),
typeof(DateTime),
typeof(string),
typeof(int),
};
foreach (var prop in props.Where(p => allowedPropertyTypes.Contains(p.PropertyType)))
{
if (prop.UnderlyingName == null)
continue;
var propertyInfo = type.GetProperty(prop.UnderlyingName);
var attribute =
propertyInfo?.GetCustomAttributes().FirstOrDefault(x => x is IMaskAttribute) as IMaskAttribute;
if (attribute == null)
{
continue;
}
if (attribute.Type != propertyInfo.PropertyType)
{
// Log this case, cause somebody used wrong attribute
continue;
}
prop.ValueProvider = new MaskValueProvider(propertyInfo, attribute.Replacement, attribute.Type);
}
return props;
}
private class MaskValueProvider : IValueProvider
{
private readonly PropertyInfo _targetProperty;
private readonly object _replacement;
private readonly Type _type;
public MaskValueProvider(PropertyInfo targetProperty, object replacement, Type type)
{
_targetProperty = targetProperty;
_replacement = replacement;
_type = type;
}
public object GetValue(object target)
{
return _replacement;
}
public void SetValue(object target, object value)
{
_targetProperty.SetValue(target, value);
}
}
}
[AttributeUsage(AttributeTargets.Property)]
public class MaskStringValueAttribute : Attribute, IMaskAttribute
{
public Type Type => typeof(string);
public object Replacement { get; }
public MaskStringValueAttribute(string replacement)
{
Replacement = replacement;
}
}
[AttributeUsage(AttributeTargets.Property)]
public class MaskIntValueAttribute : Attribute, IMaskAttribute
{
public object Replacement { get; }
public Type Type => typeof(int);
public MaskIntValueAttribute(int replacement)
{
Replacement = replacement;
}
}
[AttributeUsage(AttributeTargets.Property)]
public class MaskGuidValueAttribute : Attribute, IMaskAttribute
{
public Type Type => typeof(Guid);
public object Replacement => Guid.Empty;
}
[AttributeUsage(AttributeTargets.Property)]
public class MaskDateTimeValueAttribute : Attribute, IMaskAttribute
{
public Type Type => typeof(DateTime);
public object Replacement => new DateTime(1970, 1, 1);
}
public interface IMaskAttribute
{
Type Type { get; }
object Replacement { get; }
}
}
I hope somebody will find it helpful.
You can try nuget package https://www.nuget.org/packages/Slin.Masking and https://www.nuget.org/packages/Slin.Masking.NLog.
It can easily be integrated with DotNet projects with slight changes, and you can define your rules for it. But the document needs some improvement.
As a suggestion, you can use two files:
masking.json (can be a generic one, that shared across all projects)
masking.custom.json (can be used with particular rules for specific projects)
I have two classes OnlineBoolTag and OnlineDoubleTag. I add these objects to a list and want to get the Value of different types. How to return Value property of double or bool?
public class OnlineDoubleTag : IOnlineTag
{
public string Name { get; set; }
public double Value { get; set; }
}
public class OnlineBoolTag : IOnlineTag
{
public string Name { get; set; }
public bool Value { get; set; }
}
Add objects to a list:
var onlinetags = new List<IOnlineTag>();
onlinetags.Add(new OnlineBoolTag { Name = "Bool1", Value = true });
onlinetags.Add(new OnlineDoubleTag { Name = "Float1", Value = 777.22 });
foreach (var tag in onlinetags)
{
Console.WriteLine(tag.*****Value*****);
}
You can use dynamic instead of object
interface IOnlineTag
{
public dynamic GetValue();
}
public class OnlineDoubleTag : IOnlineTag
{
public string Name { get; set; }
public double Value { get; set; }
public dynamic GetValue()
{
return this.Value;
}
}
public class OnlineBoolTag : IOnlineTag
{
public string Name { get; set; }
public bool Value { get; set; }
public dynamic GetValue()
{
return this.Value;
}
}
public static void Main()
{
var onlinetags = new List<IOnlineTag>();
onlinetags.Add(new OnlineBoolTag { Name = "Bool1", Value = true });
onlinetags.Add(new OnlineDoubleTag { Name = "Float1", Value = 7777.22 });
foreach (var tag in onlinetags)
{
Console.WriteLine($"{tag.GetValue()} {tag.GetValue().GetType()}");
}
// Value: True Type: System.Boolean
// Value: 7777.22 Type: System.Double
}
Using dynamic will help here, as Guy's answer points out. We can simplify this even further though, by defining the property Value to be of type dynamic itself. This removes the need to implement the interface for GetValue method. Instead, we can do this:
public class OnlineTag
{
public string Name {get;set;}
public dynamic Value {get;set;}
}
public class Program
{
public static void Main(string[] args)
{
var onlinetags = new List<OnlineTag>();
onlinetags.Add(new OnlineTag { Name = "Bool1", Value = true });
onlinetags.Add(new OnlineTag { Name = "Float1", Value = 7777.22 });
foreach (var tag in onlinetags)
{
Console.WriteLine($"{tag.Value} {tag.Value.GetType()}");
//prints the below
//True System.Boolean
//7777,22 System.Double
}
}
}
The pitfall with this is that defining Value as dynamic allows you to reassign a value of totally different type later on in your code. For example, the below code will not throw any errors:
Console.WriteLine($"{onlinetags[0].Value} {onlinetags[0].Value is bool}"); //prints True True
onlinetags[0].Value = 123M;
Console.WriteLine($"{onlinetags[0].Value} {onlinetags[0].Value.GetType()}"); // prints 123 System.Decimal
I have the following poco (other properties omitted for simplicity) :
public class Address
{
. . .
public string CountryCode { get; set; }
. . .
}
What do I have to do in the BsonClassMap to enforce Upper Case only for this property.
For Example "us" will be stored in the db as "US"
BsonClassMap.RegisterClassMap<Address>(cm =>
{
// what am I missing here ?
});
Or am I approaching this the wrong way ?
here's a custom serializer attribute you can decorate the country code property with:
[AttributeUsage(AttributeTargets.Property, AllowMultiple = false)]
public class UpperCaseAttribute : BsonSerializerAttribute
{
public UpperCaseAttribute() : base(typeof(UpperCaseSerializer)) { }
private class UpperCaseSerializer : SerializerBase<string>
{
public override void Serialize(BsonSerializationContext ctx, BsonSerializationArgs args, string value)
{
if (value is null)
ctx.Writer.WriteNull();
else
ctx.Writer.WriteString(value.ToUpper());
}
public override string Deserialize(BsonDeserializationContext ctx, BsonDeserializationArgs args)
{
switch (ctx.Reader.CurrentBsonType)
{
case BsonType.String:
return ctx.Reader.ReadString();
case BsonType.Null:
ctx.Reader.ReadNull();
return null;
default:
throw new BsonSerializationException($"'{ctx.Reader.CurrentBsonType}' values are not valid on properties decorated with an [UpperCase] attribute!");
}
}
}
usage:
public class Address
{
[UpperCase]
public string CountryCode { get; set; }
}
Check this:
SetElementName way:
BsonClassMap.RegisterClassMap<Address>(cm =>
{
cm.MapField(e => e.CountryCode).SetElementName("COUNTRY_CODE");
});
var address = new Address();
var bson = address.ToBsonDocument();
// bson: { "COUNTRY_CODE" : null }
BsonElement attribute way:
public class Address
{
[BsonElement("COUNTRY_CODE")]
public string CountryCode { get; set; }
}
Sample class:
public class ClassA
{
public int Id { get; set; }
public string SomeString { get; set; }
public int? SomeInt { get; set; }
}
Default deserializer:
var myObject = JsonConvert.DeserializeObject<ClassA>(str);
Create the same object for two different inputs
{"Id":5}
or
{"Id":5,"SomeString":null,"SomeInt":null}
How can I track properties that were missing during deserialization process and preserve the same behavior? Is there a way to override some of JSON.net serializer methods (e.g. DefaultContractResolver class methods) to achive this. For example:
List<string> missingProps;
var myObject = JsonConvert.DeserializeObject<ClassA>(str, settings, missingProps);
For the first input list should contains the names of the missing properties ("SomeString", "SomeInt") and for second input it should be empty. Deserialized object remains the same.
1. JSON has a property which is missing in your class
Using property JsonSerializerSettings.MissingMemberHandling you can say whether missing properties are handled as errors.
Than you can install the Error delegate which will register errors.
This will detect if there is some "garbage" property in JSON string.
public class ClassA
{
public int Id { get; set; }
public string SomeString { get; set; }
}
internal class Program
{
private static void Main(string[] args)
{
const string str = "{'Id':5, 'FooBar': 42 }";
var myObject = JsonConvert.DeserializeObject<ClassA>(str
, new JsonSerializerSettings
{
Error = OnError,
MissingMemberHandling = MissingMemberHandling.Error
});
Console.ReadKey();
}
private static void OnError(object sender, ErrorEventArgs args)
{
Console.WriteLine(args.ErrorContext.Error.Message);
args.ErrorContext.Handled = true;
}
}
2. Your class has a property which is missing in JSON
Option 1:
Make it a required property:
public class ClassB
{
public int Id { get; set; }
[JsonProperty(Required = Required.Always)]
public string SomeString { get; set; }
}
Option 2:
Use some "special" value as a default value and check afterwards.
public class ClassB
{
public int Id { get; set; }
[DefaultValue("NOTSET")]
public string SomeString { get; set; }
public int? SomeInt { get; set; }
}
internal class Program
{
private static void Main(string[] args)
{
const string str = "{ 'Id':5 }";
var myObject = JsonConvert.DeserializeObject<ClassB>(str
, new JsonSerializerSettings
{
DefaultValueHandling = DefaultValueHandling.Populate
});
if (myObject.SomeString == "NOTSET")
{
Console.WriteLine("no value provided for property SomeString");
}
Console.ReadKey();
}
}
Option 3:
Another good idea would be to encapsulate this check iside the class istself. Create a Verify() method as shown below and call it after deserialization.
public class ClassC
{
public int Id { get; set; }
[DefaultValue("NOTSET")]
public string SomeString { get; set; }
public int? SomeInt { get; set; }
public void Verify()
{
if (SomeInt == null ) throw new JsonSerializationException("SomeInt not set!");
if (SomeString == "NOTSET") throw new JsonSerializationException("SomeString not set!");
}
}
Another way to find null/undefined tokens during De-serialization is to write a custom JsonConverter , Here is an example of custom converter which can report both omitted tokens (e.g. "{ 'Id':5 }") and null tokens (e.g {"Id":5,"SomeString":null,"SomeInt":null})
public class NullReportConverter : JsonConverter
{
private readonly List<PropertyInfo> _nullproperties=new List<PropertyInfo>();
public bool ReportDefinedNullTokens { get; set; }
public IEnumerable<PropertyInfo> NullProperties
{
get { return _nullproperties; }
}
public void Clear()
{
_nullproperties.Clear();
}
public override bool CanConvert(Type objectType)
{
return true;
}
public override object ReadJson(JsonReader reader, Type objectType, object existingValue, JsonSerializer serializer)
{
existingValue = existingValue ?? Activator.CreateInstance(objectType, true);
var jObject = JObject.Load(reader);
var properties =
objectType.GetProperties(BindingFlags.Instance | BindingFlags.Public | BindingFlags.NonPublic);
foreach (var property in properties)
{
var jToken = jObject[property.Name];
if (jToken == null)
{
_nullproperties.Add(property);
continue;
}
var value = jToken.ToObject(property.PropertyType);
if(ReportDefinedNullTokens && value ==null)
_nullproperties.Add(property);
property.SetValue(existingValue, value, null);
}
return existingValue;
}
//NOTE: we can omit writer part if we only want to use the converter for deserializing
public override void WriteJson(JsonWriter writer, object value, JsonSerializer serializer)
{
var objectType = value.GetType();
var properties =
objectType.GetProperties(BindingFlags.Instance | BindingFlags.Public | BindingFlags.NonPublic);
writer.WriteStartObject();
foreach (var property in properties)
{
var propertyValue = property.GetValue(value, null);
writer.WritePropertyName(property.Name);
serializer.Serialize(writer, propertyValue);
}
writer.WriteEndObject();
}
}
Note: we can omit the Writer part if we don't need to use it for serializing objects.
Usage Example:
class Foo
{
public int Id { get; set; }
public string SomeString { get; set; }
public int? SomeInt { get; set; }
}
class Program
{
static void Main(string[] args)
{
var nullConverter=new NullReportConverter();
Console.WriteLine("Pass 1");
var obj0 = JsonConvert.DeserializeObject<Foo>("{\"Id\":5, \"Id\":5}", nullConverter);
foreach(var p in nullConverter.NullProperties)
Console.WriteLine(p);
nullConverter.Clear();
Console.WriteLine("Pass2");
var obj1 = JsonConvert.DeserializeObject<Foo>("{\"Id\":5,\"SomeString\":null,\"SomeInt\":null}" , nullConverter);
foreach (var p in nullConverter.NullProperties)
Console.WriteLine(p);
nullConverter.Clear();
nullConverter.ReportDefinedNullTokens = true;
Console.WriteLine("Pass3");
var obj2 = JsonConvert.DeserializeObject<Foo>("{\"Id\":5,\"SomeString\":null,\"SomeInt\":null}", nullConverter);
foreach (var p in nullConverter.NullProperties)
Console.WriteLine(p);
}
}
I got this problem, but defaultValue was not solution due to POCO object. I think this is simpler approach than NullReportConverter.
There are three unit tests. Root is class that encapsulate whole json. Key is type of the Property. Hope this helps someone.
using Microsoft.VisualStudio.TestTools.UnitTesting;
using Newtonsoft.Json;
namespace SomeNamespace {
[TestClass]
public class NullParseJsonTest {
[TestMethod]
public void TestMethod1()
{
string slice = "{Key:{guid:\"asdf\"}}";
var result = JsonConvert.DeserializeObject<Root>(slice);
Assert.IsTrue(result.OptionalKey.IsSet);
Assert.IsNotNull(result.OptionalKey.Value);
Assert.AreEqual("asdf", result.OptionalKey.Value.Guid);
}
[TestMethod]
public void TestMethod2()
{
string slice = "{Key:null}";
var result = JsonConvert.DeserializeObject<Root>(slice);
Assert.IsTrue(result.OptionalKey.IsSet);
Assert.IsNull(result.OptionalKey.Value);
}
[TestMethod]
public void TestMethod3()
{
string slice = "{}";
var result = JsonConvert.DeserializeObject<Root>(slice);
Assert.IsFalse(result.OptionalKey.IsSet);
Assert.IsNull(result.OptionalKey.Value);
}
}
class Root {
public Key Key {
get {
return OptionalKey.Value;
}
set {
OptionalKey.Value = value;
OptionalKey.IsSet = true; // This does the trick, it is never called by JSON.NET if attribute missing
}
}
[JsonIgnore]
public Optional<Key> OptionalKey = new Optional<Key> { IsSet = false };
};
class Key {
public string Guid { get; set; }
}
class Optional<T> {
public T Value { get; set; }
public bool IsSet { get; set; }
}
}
I have a multilingual database, which returns values based on a key and an enum Language. When I convert a DB object to a model, I want the model to contain the translated value based on the key and the current language.
The key comes from the DB object but how can I pass the current language to the the Mapper.Map() function?
Currently, I am using a [ThreadStatic] attribute to set the culture before calling Mapper.Map<>, and to retrieve it in the TypeConverter.
public enum Language
{
English, French, Italian, Maltese
}
public class MultilingualValue<T>
{
public Dictionary<Language, T> Value { get; set; }
public MultilingualValue()
{
this.Value = new Dictionary<Language, T>();
}
}
public class PersonData
{
public string FirstName { get; set; }
public MultilingualValue<string> City { get; set; }
}
public void MapPerson()
{
PersonData personData = new PersonData();
personData.FirstName = "John";
personData.City = new MultilingualValue<string>();
personData.City.Value[ Language.English] = "The Zurrieq";
personData.City.Value[Language.French] = "Le Zurrieque";
MultilingualValueData.CurrentLanguage = Language.English;
var personModel = Mapper.Map<PersonData, PersonModel>(personData);
}
public class MultilingualValueToBasicDataTypeConverter<T> : ITypeConverter<MultilingualValue<T>, T>
{
public T Convert(ResolutionContext context)
{
var currentLanguage = MultilingualValueData.CurrentLanguage; //THIS IS THE [ThreadStatic] VARIABLE
if (currentLanguage == null) throw new InvalidOperationException("Please make sure to fill in CurrentLanguage");
MultilingualValue<T> sourceMultilingualValue = (MultilingualValue < T > )context.SourceValue;
T destinationValue = default(T);
if (sourceMultilingualValue != null)
{
destinationValue = sourceMultilingualValue.Value[currentLanguage.Value];
}
return destinationValue;
}
}
public static class MultilingualValueData
{
[ThreadStatic]
public static Language? CurrentLanguage;
}
I left out the configurations as I think they're unneccessary for this example. If you need them, I'll post them as well.
While this works, I find this workaround quite ugly. Is there any way to pass data through the ResolutionContext?
Just use the Map overload that takes a Action<IMappingOperationOptions>. You can add configuration elements to the Items property that are then passed to your ITypeConverter
public class CustomConverter : ITypeConverter<string, string>
{
public string Convert(ResolutionContext context)
{
return "translated in " + context.Options.Items["language"];
}
}
internal class Program
{
private static void Main(string[] args)
{
AutoMapper.Mapper.CreateMap<string, string>().ConvertUsing<CustomConverter>();
var result = AutoMapper.Mapper.Map<string, string>("value" , opt => opt.Items["language"] = "english");
Console.Write(result); // prints "translated in english"
Console.ReadLine();
}
}