json2typescript
v1.5.1
Published
Provides TypeScript methods to map a JSON object to a JavaScript object on runtime
Downloads
156,220
Maintainers
Readme
json2typescript
In Angular applications, everyone consumes JSON API's from an external source. Type checking and object mapping is only possible in TypeScript, but not in the JavaScript runtime. As the API may change at any point, it is important for larger projects to verify the consumed data.
json2typescript is a small package containing a helper class that maps JSON objects to an instance of a TypeScript class. After compiling to JavaScript, the result will still be an instance of this class. One big advantage of this approach is, that you can also use methods of this class.
With json2typescript, only a simple function call is necessary, as demonstrated in this TypeScript snippet:
// Assume that you have a class named User defined at some point
// Assume that you get a JSON string from a webservice
let jsonStr: string = ...;
let jsonObj: any = JSON.parse(jsonStr);
// Now you can map the json object to the TypeScript object automatically
let jsonConvert: JsonConvert = new JsonConvert();
let user: User = jsonConvert.deserializeObject(jsonObj, User);
console.log(user); // prints User{ ... } in JavaScript runtime, not Object{ ... }
Tip: All
serialize()
anddeserialize()
methods may throw anError
in case of failure. Make sure you catch errors in production!
Changelog
See the changelog in the separate file for bug fixes, new features and breaking changes: Changelog
Warning: If you are reading this document on GitHub, it might be ahead of the published NPM version. Please refer to the ReadMe on NPM if in doubt.
Warning: We earlier suggested to use the
@JsonObject(classIdentifier)
decorator, but did not enforce it. Since v1.4.0, it is mandatory to use a unique classIdentifier for every class in order to make (de)serialization work properly with class inheritance. In versions above v1.2.0 and below v1.4.0, it is possible to run into issues when not using the decorator.
Getting started
Requirements
We developed json2typescript for Angular 2+ and Ionic 2+. In this document, we only cover this use case. However, you may use our package for pure TypeScript or even JavaScript applications.
Setup a Test Application
We recommend to use the official angular cli tool in order to set up a new Angular project. Then, all you need to do is type the following into your operating system's terminal:
ng new testApplication
cd testApplication
npm install json2typescript
Our package makes use of TypeScript decorators. If not done already, please activate them in your tsconfig.json
under compilerOptions
as follows:
{
"compilerOptions": {
[
...
]
"experimentalDecorators": true,
"emitDecoratorMetadata": true,
[...]
}
Tip: We have tried to make the compiler options of
json2typescript
to be as strict as possible. This enables you to use compiler options such as"strictNullChecks": true
or"noImplicitAny": true
in your own project.
Now you are ready to use the package.
Mapping example
In order to use the json2typescript package, all you need to do is write decorators and import the package. The following things need to be done if you would like to map JSON to existing classes:
- Classes need to be preceded by
@JsonObject(classIdentifier)
where the classIdentifier is unique in the whole project - Properties need to be preceded by
@JsonProperty(jsonProperty, conversionOption, convertingMode)
- Properties need to have a default value (or undefined), otherwise the mapper will not work
See below an example, so you can learn from it how json2typescript works best.
Assuming that you have created the testApplication in the step before and installed json2typescript as suggested, create a class in a new file city.ts with the following content:
import { JsonObject, JsonProperty } from "json2typescript";
@JsonObject("City") // Make sure "City" is a unique identifier for this class
export class City {
// This property has no @JsonProperty.
// It will not be mapped.
id: number = 123;
// This maps the value of the JSON key "name" to the class property "name".
// If the JSON value is not of type string (or missing), there will be an exception.
@JsonProperty("name", String)
name: string = "";
// This maps the JSON key "founded" to the private class property "_founded".
// Note the use of public getter and setter.
// If the JSON value is not of type number (or missing), there will be an exception.
@JsonProperty("founded", Number)
private _founded: number = 0;
get founded() { return this._founded; }
set founded(value: number) { this._founded = value; }
// This maps the JSON key "beautiful" to the class property "beautiful".
// If the JSON value is not of type boolean (or missing), there will be an exception.
@JsonProperty("beautiful", Boolean)
beautiful: boolean = false;
// This maps the JSON key "data" to the class property "data".
// We are not sure about the type, so we omit the second parameter.
// There will be an exception if the JSON value is missing.
@JsonProperty("data") // is the same as @JsonProperty("data", Any)
data: any = undefined;
// This maps the JSON key "keywords" to the class property "keywords".
// This is an example of a string array. Note our syntax "[String]".
// In the further examples at the end of this document, you can see how to nest complex arrays.
@JsonProperty("keywords", [String])
keywords: string[] = []; // or Array<string>
printInfo() {
if (this.beautiful)
console.log(this.name + " was founded in " + this.founded + " and is really beautiful!");
else
console.log(this.name + " was founded in " + this.founded + ".");
}
}
Now create a file country.ts with the following content:
import { City } from "./city";
import { JsonObject, JsonProperty } from "json2typescript";
@JsonObject("Country") // Make sure "Country" is a unique identifier for this class
export class Country {
// This maps the value of the JSON key "countryName" to the class property "name".
// If the JSON value is not of type string (or missing), there will be an exception.
@JsonProperty("countryName", String)
name: string = "";
// This maps the value of the JSON key "cities" to the class property "cities".
// If the JSON value is not of type array object (or missing), there will be an exception.
// There will be an exception too if the objects in the array do not match the class "City".
@JsonProperty("cities", [City])
cities: City[] = [];
}
Then navigate to the file app.component.ts and add the following code:
import { Component, OnInit } from '@angular/core';
import { JsonConvert, OperationMode, ValueCheckingMode } from "json2typescript"
import { Country } from "./country";
@Component({
selector: 'app-root',
templateUrl: './app.component.html',
styleUrls: ['./app.component.css']
})
export class AppComponent implements OnInit {
ngOnInit() {
// Define a JSON object (could come from a HTTP service, parsed with JSON.parse() if necessary)
const jsonObject: any = {
"countryName": "Switzerland",
"cities": [
{
"id": 1,
"name": "Basel",
"founded": -200,
"beautiful": true,
"data": 123,
"keywords": ["Rhine", "River"]
},
{
"id": 1,
"name": "Zurich",
"founded": 0,
"beautiful": false,
"data": "no",
"keywords": ["Limmat", "Lake"]
}
]
};
// Choose your settings
// Check the detailed reference in the chapter "JsonConvert class properties and methods"
let jsonConvert: JsonConvert = new JsonConvert();
jsonConvert.operationMode = OperationMode.LOGGING; // print some debug data
jsonConvert.ignorePrimitiveChecks = false; // don't allow assigning number to string etc.
jsonConvert.valueCheckingMode = ValueCheckingMode.DISALLOW_NULL; // never allow null
// Map to the country class
let country: Country;
try {
country = jsonConvert.deserializeObject(jsonObject, Country);
country.cities[0].printInfo(); // prints: Basel was founded in -200 and is really beautiful!
} catch (e) {
console.log((<Error>e));
}
}
Play around with the JSON to provocate exceptions when deserializing the object.
Important notes
Avoid circular dependencies on the classes that use json2typescript
. Even if you don't have any errors in your
IDE, json2typescript
will not properly work in this case.
Detailed reference
Class and property decorators
Decorators should be used whenever you would like to map JSON with TypeScript data. As of now, you must not use more than one decorator per class or property.
Class decorators
The class decorators are used infront of the class declaration and do support one parameter:
@JsonObject("User")
export class User {}
Warning: The class decorator uses the parameter to identify the class. Please use a unique identifier for each class in your project, for example simply the name of the class.
Tip: Make sure you import
JsonObject
fromjson2typescript
.
First parameter: classIdentifier
The first parameter of @JsonObject
must be a unique class identifier, usually just the class name.
Note: This class identifier may be used for automatic instantiation when enabling the discriminator feature.
Property decorators
Property decorators are a vital part for type checking. It is important that the type in the decorator matches the TypeScript type.
For class properties to be visible to the mapper they must be initialized, otherwise they are ignored.
They can be initialized using any (valid) value or undefined
.
See the example below for better understanding:
@JsonObject("User")
export class User {
// A correct example
@JsonProperty("name", String, false)
name: string = "";
// An incorrect example
@JsonProperty("alias", string, false) // Wrong type: Must use String instead.
alias: string = "";
// An incorrect example
@JsonProperty("expertise", String, false)
expertise: string; // No initialization: Property will be ignored without visible exception
}
Important note: You must assign any (valid) value or
undefined
to your property at initialization, otherwise our mapper does not work and will simply ignore the property. Assigning no value is not the same as assigningundefined
in context ofjson2typescript
. Non-initialized properties will not trigger any exception, as they are invisible to the mapper.
Tip: Make sure you import
JsonObject
andJsonProperty
fromjson2typescript
.
First parameter: jsonProperty
The first parameter of @JsonProperty
is the JSON object property name. It happens that the property names given by the
server are very ugly. Here you can map any json property name to the User
property name
. In our
case, json["jsonPropertyName"]
gets mapped to user.name
.
Second parameter (optional): conversionOption
The second parameter of @JsonProperty
describes what happens when doing the mapping between JSON and TypeScript
objects. This parameter is optional; the default value is Any
(which means no type check is done when the mapping
happens).
Use of expected type
If you would like that json2typescript
performs an automatic type check according to given TypeScript types, you can
pass a type you expect. Follow the following guide when doing that:
- Make sure you pass the class name and not an instance of the class.
- In case of primitive types, you have to use the upper case names.
- In case of
any
type, import fromjson2typescript
the classAny
.
See the following cheat sheet for reference:
| Expected type | TypeScript type | | --- | --- | | String | string | | Number | number | | Boolean | boolean | | User | User | | Any | any | | | | | [String] | string[] | | [Number] | number[] | | [Boolean] | boolean[] | | [User] | User[] | | [Any] | any[] |
At first, our array notation on the left looks odd. But this notation allows you to define even nested arrays. See the examples at the end of this document for more info about nesting arrays.
Tip: It is possible to lazy-load custom classes to prevent circular dependencies. Then, the expected type for custom classes can be written as a string. Please read further below how to implement this feature.
Adding a custom converter
More advanced users may need to use custom converters. If you want
json2typescript
to use your custom converter, you need to follow these steps:
- Write a converter class that implements
JsonCustomConvert<T>
where<T>
is the type resulting in the TypeScript class. - Make sure you add the
@JsonConverter
decorator to this class (see next chapter how). - Pass your converter class as second param in the
@JsonProperty
decorator
Assume you would like to transform 1893-11-15
(string from JSON) to a TypeScript
Date
instance, then you would write a class DateConverter
(see next chapter how)
and pass it as second param in @JsonProperty
as below:
@JsonObject("User")
export class User {
@JsonProperty("birthdate", DateConverter)
birthdate: Date = new Date();
}
Third parameter (optional): convertingMode
The third parameter of @JsonProperty
determines how nullable property types should be serialized and deserialized.
Nullable types are either missing (in JSON), undefined (in TypeScript) or null (both). This parameter is optional; the
default value is PropertyConvertingMode.MAP_NULLABLE
.
See also the property propertyConvertingMode
of the JsonConvert
instance.
The values should be used as follows:
PropertyConvertingMode.MAP_NULLABLE
: the mapper is applied, type is checkedPropertyConvertingMode.IGNORE_NULLABLE
: the mapper is not applied if the property is missing, undefined or null; the property is not added to the resultPropertyConvertingMode.PASS_NULLABLE
: the mapper is not applied if the property is missing, undefined or null; the property is added with its value to the result
Tip: Make sure you import the
ENUM
PropertyConvertingMode
when assigning a value to this property.
Handling null, undefined and absent values
Be careful when handling special values as null
, undefined
and absent
properties.
By default, json2typescript
throws an exception if a decorated class property cannot be found in the given JSON when
deserializing. If you set the third parameter to IGNORE_NULLABLE
or PASS_NULLABLE
, there will be no exception when
it is missing. The type of a property will only be checked if the property is present in the JSON and not undefined
or null
.
The global setting of valueCheckingMode
determines whether you want to allow null
values for objects or properties.
We recommend to use the most strict option and also set your TypeScript
compiler to the strict mode.
The following table explains the difference between the three property converting modes:
| ALLOW_NULL | serialize(null)
| serialize(undefined)
| deserialize(null)
| deserialize(undefined)
|
| --- | --- | --- | --- | --- |
| MAP_NULLABLE
| null
| error | null
| error |
| IGNORE_NULLABLE
| undefined
(missing) | undefined
(missing) | default value | default value |
| PASS_NULLABLE
| null
| undefined
(missing) | null
| undefined
|
| |
| DISALLOW_NULL | serialize(null)
| serialize(undefined)
| deserialize(null)
| deserialize(undefined)
|
| MAP_NULLABLE
| error | error | error | error |
| IGNORE_NULLABLE
| undefined
(missing) | undefined
(missing) | default value | default value |
| PASS_NULLABLE
| null
| undefined
(missing) | null
| undefined
|
As shown in this table, a property with the default setting MAP_NULLABLE
is never allowed to be undefined
(or
missing). The valueCheckingMode
determines, whether null
is allowed.
Tip: If you want
undefined
to be treated in the same way asnull
values, you may set the instancemapUndefinedToNull
property totrue
.
Important notes
- Make sure you define the expected type as accurate as possible, even if you expect primitive types.
- By default, casting primitives into other primitives is not allowed. Check the public properties below in this document to change this behaviour.
- By default, primitives are not allowed to be null. Check the public properties below in this document to change this.
- If you don't know the type, you may use
Any
as expected type. You may also omit the second parameter of@JsonProperty
.
More about the array syntax
- You can allow arrays by using the bracket notation combined with the types as seen above. For example:
[String]
for a string array - Mixing arrays is allowed. For example:
[String, Number]
for an array where the first entry is a string, the second is a number. - If the real array is longer than indicated here, the last type will be forced to the rest of the array entries (
above:
Number
). - This means,
[String, Number]
is equivalent to[String, Number, Number]
and so on. - Nesting arrays and objects are allowed. For example:
[[String, Number], User]
. - This is equivalent to
[[String, Number, Number], User, User]
and so on. - As further example,
[[String]]
is equal to[[String],[String]]
which is equal to[[String,String], [String,String]]
and so on. - If an array has less elements as given in the expected type, no exception is thrown.
- For example, if we define
[String]
or the equivalent[String, String]
no exception is thrown - even if the JSON gives us an empty array.
Tip: See the examples at the end of this document for advanced examples for nesting arrays.
Custom converter decorators
In some cases, you may need to make custom conversion between JSON objects and TypeScript objects. You can define custom converters like this:
@JsonConverter
class DateConverter implements JsonCustomConvert<Date> {
serialize(date: Date): any {
return date.getFullYear() + "-" + (date.getMonth() + 1) + "-" + date.getDate();
}
deserialize(date: any): Date {
return new Date(date);
}
}
Tip: Make sure that you import
JsonConverter
fromjson2typescript
. Also don't forget to use the same type between the brackets<>
, as theserialize()
param anddeserialize()
return value.
Assume that in your JSON you have a date in a standardized format, such as 2017-07-19 10:00:00
. You could use the
custom converter class above to make sure it is stored as a real TypeScript Date
in your class. For your property, you
simply have use the @JsonProperty
decorator as follows:
@JsonObject("User")
export class User {
@JsonProperty("date", DateConverter)
date: Date = new Date();
}
With this approach, you will achieve that your property date
is going to be a real instance of Date
.
JsonConvert class properties and methods
Public properties
Operation mode
(number) JsonConvert.operationMode
Determines how the JsonConvert class instance should operate.
You may assign three different values:
OperationMode.DISABLE
: json2typescript will be disabled, no type checking or mapping is doneOperationMode.ENABLE
: json2typescript is enabled, but only errors are loggedOperationMode.LOGGING
: json2typescript is enabled and detailed information is logged
The default value is OperationMode.ENABLE
. It will only print errors to the console and is suited for production.
In some cases, you might consider disabling json2typescript
in production by setting the OperationMode.DISABLE
flag.
This only works in case you only use plain objects without functionality and no mapping. However,
disabling json2typescript
might give you a performance advantage in heavy projects.
In case you have issues to find bugs, you can enable additional logging by setting the OperationMode.LOGGING
flag.
Please note that every serializing and deserializing is heavily logged to the console and will make your application
slower. Never use this flag in production.
Tip: Make sure you import the
ENUM
OperationMode
when assigning a value to this property.
Value checking mode
(number) JsonConvert.valueCheckingMode
Determines which types are allowed to be null in the deserialization. You may assign three different values:
ValueCheckingMode.ALLOW_NULL
: All given values can be nullValueCheckingMode.ALLOW_OBJECT_NULL
: Objects can be null, but primitive types cannot be nullValueCheckingMode.DISALLOW_NULL
: No null values are tolerated
The default is ValueCheckingMode.ALLOW_OBJECT_NULL
.
Tip: Make sure you import the
ENUM
ValueCheckingMode
when assigning a value to this property.
Tip: The TypeScript documentation suggests to avoid null values. Compile your TypeScript code with
strictNullChecks=true
and set thevalueCheckingMode
to disallow null values. If your API returnsnull
in some cases, simply mark these properties as optional in the correspondingJsonProperty
decorator to avoid errors on runtime.
Map undefined to null
(number) JsonConvert.mapUndefinedToNull
Determines whether a missing or undefined property value should be considered as null or not.
If true, a missing JSON value will be added and set as null before deserialization. For serialization, undefined values will be set to null before serialization.
Note: ValueCheckingMode and PropertyConvertingMode determine whether an error will be thrown during serialization or deserialization.
Ignore primitive checks
(bool) JsonConvert.ignorePrimitiveChecks
Determines whether primitive types should be checked. If true, it will be allowed to assign primitive to other primitive types.
The default is false
.
Property matching rule
(number) JsonConvert.propertyMatchingRule
Determines the rule of how JSON properties shall be matched with class properties during deserialization. You may assign the following two values:
PropertyMatchingRule.CASE_STRICT
: JSON properties need to match exactly the names in the decoratorsPropertyMatchingRule.CASE_INSENSITIVE
: JSON properties need to match names in the decorators, but names they are not case sensitive
The default is PropertyMatchingRule.CASE_STRICT
.
Property converting mode
(number|undefined) JsonConvert.propertyConvertingMode
Determines how nullable property types should be serialized and deserialized. Nullable types are either missing (in JSON), undefined (in TypeScript) or null (both).
If the propertyConvertingMode has a non-undefined value, it overrides the individual settings of every property. See
also the third parameter of the @JsonProperty
decorator.
The values should be used as follows:
PropertyConvertingMode.MAP_NULLABLE
: the mapper is applied, type is checkedPropertyConvertingMode.IGNORE_NULLABLE
: the mapper is not applied if the property is missing, undefined or null; the property is- not added to the result
PropertyConvertingMode.PASS_NULLABLE
: the mapper is not applied if the property is missing, undefined or null; the property is- added with its value to the result
The default is PropertyMatchingRule.MAP_NULLABLE
.
Note: This property is usually only temporarily set and should be used with caution. It replaces the deprecated property
ignoreRequiredCheck
.
Use discriminator
(bool) JsonConvert.useDiscriminator
Determines if discriminators should be used. If this option is set to true, all registered classes will be serialized with an additional discriminator property (default: "$type"), which has the key of the class (given in the @JsonObject decorator) as value. When deserializing an object containing the discriminator property, json2typescript will attempt to automatically instantiate the correct type (by comparing the value of the discriminator property with the registered classes).
The default is false
.
Note: At the end of this document you may find an example on how to use the discriminator feature.
Discriminator property name
(string) JsonConvert.discriminatorPropertyName
Defines the name of the discriminator property.
The default is "$type"
.
Public methods
json2typescript
allows you to map JSON objects (or arrays) to TypeScript objects (or arrays) and vice versa.
Serializing (TypeScript to JSON)
(any|any[]) serialize<T extends object, U extends object = {}>(data: T | T[], classReference?: { new(): U })
Tries to serialize a TypeScript object or array of objects to JSON.
The first parameter must be a TypeScript object or array, the second parameter is the optional class reference.
If you provide only one parameter, the class for serialization is inferred automatically. For example, if you
call jsonConvert.serialize(user)
where user
is an instance of the class User
, json2typescript
will automatically
use this class for serialization.
By providing two parameters, it will override the class for serialization. For example, this allows you to
call jsonConvert.serialize(userObject, User)
where userObject
is just a plain TypeScript any
object.
The returned value will be any
object or an array of any
objects.
Tip: The return value is not a string. In case you need a string as result, use
JSON.stringify()
after calling the serialize method.
You may optionally provide a class constructor to use the @JsonProperty
mappings defined for that class to serialize
the data object(s), instead of mappings defined on the data class. Note that if the data is an array, the mappings from
the constructor class are used for all elements in the array. If no constructor is provided, the mappings from the
data object class are used to serialize the data object(s).
Tip: This feature is helpful if you need to serialize an object that was not created using a class constructor, or if you want to serialize a subclass with only the properties of the superclass.
Deserializing (JSON to TypeScript)
(T | T[]) deserialize<T extends object>(json: any, classReference: { new(): T } | null = null)
Tries to deserialize given JSON to a TypeScript object or array of objects.
The first parameter must be a Typescript object or array, the second parameter is the class reference. If the discriminator feature is used, the class reference is optional.
The returned value will be an instance or an array of instances of the given class reference.
Tip: The param
json
must not be a string, but anobject
or anarray
. UseJSON.parse()
before applying the deserialize method in case you have a json string.
Registering and unregistering classes
void registerClasses(...classReferences: { new(): any }[])
Registers a list of classes to be used with lazy-loading and in the discriminator feature.
void ungisterClasses(...classReferences: { new(): any }[])
Unregisters a list of classes from lazy-loading and the discriminator feature.
void unregisterAllClasses()
Unregisters all classes from lazy-loading and the discriminator feature.
Note: You only need to register and unregister classes if you use lazy-loading or the discriminator feature. Otherwise, these methods are without any effect.
Other methods
The methods serialize()
and deserialize()
will automatically detect the dimension of your param (either object or
array). In case you would like to force json2typescript
to use a specific way, you can use the following methods
instead:
(any) serializeObject<T extends object, U extends object = {}>(data: T, classReference?: { new(): U })
(any[]) serializeArray<T extends object, U extends object = {}>(dataArray: T[], classReference?: { new(): U })
(T) deserializeObject<T extends object>(jsonObject: any, classReference: { new(): T })
(T[]) deserializeArray<T extends object>(jsonArray: any[], classReference: { new(): T })
Advanced strategies
In this section you will find additional examples.
Nesting arrays
It is heavily discouraged to use nested arrays and use different types in a JSON API. If you need them anyway, here is how you have to define the types:
1) Nested arrays with same type
In the following example, jsonKeyOfWeirdKeywords
is a key in the JSON object defined like this:
{
"jsonKeyOfWeirdKeywords": [
[
"Hello",
"World"
],
[
"Bye",
"Earth"
]
]
}
As we have an array of array of strings, you can define the expected type like this:
@JsonObject("User")
export class User {
@JsonProperty("jsonKeyOfWeirdKeywords", [[String, String], [String, String]])
keywords: (string[])[] = [];
}
Tip: In our syntax,
[[String, String], [String, String]]
is equivalent to[[String], [String]]
and[[String]]
. That is because the last type in an array will be automatically repeated as much as needed.
2) Nested array with mixed depth and type
In the following example, JSONKeyOfWeirdKeywords
is a key in the JSON object defined like this:
{
"jsonKeyOfWeirdKeywords": [
[
"FC",
"Basel"
],
1893
]
}
You can define the expected type in your class like this:
@JsonObject("User")
export class User {
@JsonProperty("jsonKeyOfWeirdKeywords", [[String, String], Number])
keywords: (string[] | number)[] = [];
}
Tip: In our syntax,
[[String, String], Number]
is equivalent to[[String], Number]
.
Instantiation with the lazy-loading feature to avoid circular dependencies
In some scenarios, developers run into circular dependencies when declaring objects. A simple example would be two classes importing each other:
import { User } from "./user";
@JsonObject("Team")
export class Team {
@JsonProperty("users", [User])
user: User[] = [];
}
import { Team } from "./team";
@JsonObject("User")
export class User {
@JsonProperty("teams", [Team])
team: Team[] = [];
}
This is generally not possible because of the circular dependency. But since TypeScript 3.8, you may use type imports in order to bypass circular dependencies.
json2typescript
implements a lazy-loading feature so that the above example is possible.
The following steps need to be done:
- Import only the type, for example by writing
import type { User } from "./user";
. Note that we added thetype
keyword. - Do not use the class reference in the
@JsonProperty
decorator, but instead use the classIdentifier (as string) given in the@JsonObject
decorator. For example, write@JsonProperty("users", ["User"])
. Note that we added the"
around the User. - Register all lazy-loading classes before ever using them with
json2typescript
. You may simply calljsonConvert.registerClasses(User, Team)
for the example above.
Now you can use the classes User
and Team
as expected.
Note: Using type imports may cause other problems in your classes because the imports may only be used for type contexts. For example, you will not be able to write
new Team()
in theUser
class at all. However, you still will be able to useTeam
as a type in theUser
class and access all its properties and methods.
Automatic instantiation using the discriminator feature
If your server adds a discriminator property to every JSON object, json2typescript
is able to automatically instantiate objects.
First, set up the discriminator feature for a class, for example
@JsonObject("app.example.User")
export class User {
@JsonProperty("name", String)
name: string = "";
}
Now, set up a JsonConvert
to enable the discriminator feature and activate it for the classes you like:
// Set up json convert
let jsonConvert: JsonConvert = new JsonConvert();
jsonConvert.useDiscriminator = true; // enable the discriminator
jsonConvert.discriminatorPropertyName = "$type"; // this is the property name
jsonConvert.registerClasses(User); // register all classes
// Assume the following JSON object coming from your server
const jsonObject: any = {
"name": "Walter",
"$type": "app.example.User" // the value of $type matches the @JsonObject decorator above
}
// This is how you would traditionally map an object
// But we have enabled the discriminator functionality and registered the User class
// In that case, the second parameter (User) here is ignored
const user1: User = jsonConvert.deserialize(jsonObject, User);
// But now you may automatically map it thanks to the $type property
const user2: User = jsonConvert.deserialize<User>(jsonObject);
Note: This feature is particularly useful when doing dynamic mapping. Otherwise, you just may provide the type yourself (as done above with
user1
) and disable the discriminator feature.
A real-world example for the discriminator feature is the mapping of child classes.
Assume that you might have two classes AdminUser
and NormalUser
that inherit from User
.
The web client sometimes cannot know the type in advance.
If you use the discriminator feature, the server can define the type in the JSON and json2typescript
will properly instantiate the desired class.
This means, your property can be safely declared in TypeScript as union type of AdminUser | NormalUser
instead of User
.
Warning: If you enable the discriminator feature and try to deserialize a JSON object to a registered class instance, the second parameter of the
deserialize
methods is always ignored.
Tools
Class decorator generator
Since version 1.4, json2typescript
requires the @JsonObject("ClassName")
decorator in front of the TypeScript class
definition. GitHub user tlmurphy
created a Python script that automatically generates the decorator with the original
class name as parameter.
More: https://gist.github.com/tlmurphy/71b58c71e594899120da365159d7d40d
Contributors
This NPM package was originally created in 2016 by Andreas Aeschlimann, founder of and software architect at AppVision GmbH.
Special thanks
You are welcome to report issues and discuss enhancements to make this package even more useful. Thanks for the input and all the pull requests from the community!